Full text of DOI:10.1007/s005429900037

Microsystem Technologies 6 ꢀ2000) 179±183 Ó Springer-Verlag 2000
Dimensional measurement of high aspect ratio micro structures
with a resonating micro cantilever probe
M. Yamamoto, H. Takeuchi, S. Aoki
179
Abstract In non-destructive dimensional measurement of EDM and LIGA are beginning to be used in practice. The
high aspect ratio micro structures ꢀHARMS), optical
evaluation of tooth surface roughness and wear of such
methods cannot offer full three dimensional information gears call for a new measurement method.
due to the lack of observation light. Again, conventional
HARMS measurement has inherent dif®culties because
mechanical measurement, such as a surface pro®ler or a high walls stand with extremely narrow spacing for
coordinate measurement machine, cannot be applied be- probing means. Because the vertical surface doesn't re¯ect
cause their stylus is too large. Furthermore, the AFM,
light back, optical methods are not suitable for three-di-
though popular among the semiconductor industry, is also mensional evaluation. Thus, measurement with physical
limited in terms of dimensional measurement, because its contact probes seems to be the promising solution for
system is usually designed for planar samples. Thus, we
have developed a new sensor-integrated micro resonating
cantilever probe and a new dimensional measurement
HARMS nondestructive evaluation.
The ®rst approach for HARMS measurement is con-
ventional mechanical probing measurement such as co-
machine, which allows the probe's vertical access to mic- ordinate measurement machine ꢀCMM). The smallest
rostructures in a sample. The new probe is made of
tungsten carbide super hard alloy and possesses design
probe readily available as of present is /0.2 mm. Because
the probing force decides the minimum probe size, low
¯exibility according to its intended application. Validity of contact force systems have been investigated ꢀfor example,
the system is con®rmed through the measurement exper- [Pril et al. ꢀ1997)] and [Schepperie et al. ꢀ1998)]). Al-
iment of EDM drilled and chemically etched micro holes. though these approaches have succeeded in realizing 1 lN
probing force and /100 micron ®ber probes respectively,
1
the dif®culty of probe assembly limits further downsizing
Introduction
of the CMM probe.
Recent development in the ®eld of high aspect ratio micro
structure technologies ꢀHARMST) accelerates the need for
nondestructive measurement of such structures.
Among various HARMS measurement, micro hole
measurement is recognized as one of the most important
industrial applications. For example, contemporary micro
nozzle tests have been done by actually spraying liquid
from the holes. The drawback; the test fails to give the
exact location of the defect, thus stirring a desire for the
measurement of hole pro®le and internal surface rough-
ness. Again in the micromachine ®eld, micro gears by
Another promising approach for HARMS measurement
is the AFM. Because most AFM systems are designed for
planer samples like Si wafers, they employ a light leverage
method to detect the probe's strain. As a result, when the
probe is inserted into microstructures, the detective light
can easily be obstructed and the system fails. One excep-
tion is critical dimension AFM ꢀCD-AFM) [Vachet and
Young ꢀ1995)]. Equipped with a ¯are-shape tip, this device
can measure a trench of maximum 8 lm depth and min-
imum 0.5 lm width, which is used for devise isolation in
LSI process. Thanks to the recent progress in AFM probe
technology, some AFM probes have their own integrated
strain sensor ꢀfor example, [Indermuhle et al. ꢀ1997) and
Itoh and Suga ꢀ1993)]). These probes can be vertically
inserted into micro holes and can measure the surface,
once the measurement setup is prepared. Although the
AFM approach seems to be a good candidate for HARMS
measurement, its rigid probe dimension can be an obsta-
cle. In other words, requirements for probe dimension can
M. Yamamoto ꢀ&), H. Takeuchi, S. Aoki
Matsushita Research Institute Tokyo, Inc.,
3-10-1 Higashimita, Tama-ku,
Kawasaki 214-8501, Japan
E-mail: tyama@mrit.mei.co.jp
A part of this work was performed under the management of the
Micromachine Center as the Industrial Science and Technology
Frontier Program, ``Research and Development of Micromachine change according to the measurement sample because the
Technology'', of MITI supported by the New Energy and
Industrial Technology Development Organization. The authors
gratefully acknowledge suggestions and encouragement by Prof.
Masuzawa of the University of Tokyo.
probe and its tip dimension decide the maximum depth,
maximum protrusion and minimum gap it can handle.
Lithographically manufactured AFM probes cannot
change shape once the mask pattern and the process
conditions are ®xed.
This paper was presented at the Third International Workshop on
High Aspect Ratio Microstructure Technology HARMST '99 in
June 1999.
The third approach for HARMS measurement is to
utilize a ¯exible manufacturing technology for custom-

180
Fig. 1. Conventional micro hole measurement technique ꢀVS)
Fig. 2. Structure of RVS probe
made probes. Such an example is the vibroscanning ꢀVS)
method [Masuzawa et al. ꢀ1993)] that uses a probe ma-
chined by WEDG technology [Masaki et al. ꢀ1990)].
WEDG is a variation of EDM and works as a NC lathe for
micro parts. The VS measurement principle is shown in
Fig. 1. The probe is vibrated at a low frequency, 100 Hz,
with amplitude of 2 lm. When the probe hits against a
surface, an electric current is detected. By measuring the
contact time per vibration period, the probe's position can
be controlled at a constant distance from the surface. As is
clear from its principle, the VS method cannot be applied
to nonconductive samples. Measurement error can occur
even in cases where the sample is covered by a thick oxide
layer. Nevertheless, the VS method is a robust measure-
ment tool for on-the-machine evaluation of micro EDMed
sample [Yamamoto et al. ꢀ1996)]. To overcome the
shortcoming of the VS method, the twin probe vibro-
scanning method was proposed [Masuzawa et al. ꢀ1997)].
By utilizing the twin cantilever and detecting the con-
ductivity between them, contact to the surface can be de-
tected irrespective of the sample's electric conductivity.
Drawbacks which still remain include the fact that the
electric contacts are open to the air and vulnerable to dust
and oxidation. Moreover, the total thickness of the probe
cannot be minimized, as it requires two cantilevers.
Based on the above-mentioned survey, we have con-
cluded that a combination of AFM technology and VS
technology gives the best methodology for HARMS mea-
surement. In other words, the new probe should accom-
pany an AFM-like integrated strain sensor while
Fig. 3a, b. RVS probe properties. a Frequency response;
b contact detection characteristics
vibroscanning method'' ꢀRVS) with respect to the fact that
it employs the same cantilever technology as in VS. Fig-
ure 3a shows the frequency response of a probe possessing
a cantilever 900 lm in length and 40 lm in thickness. The
Q of resonance is 400. Figure 3b illustrates the change of
vibration signal when the contact occurs. The tip ampli-
tude is set to 0.1 lm. These results indicate the probe's
detecting ability of over 0.1 lm. Calculating from probe
stiffness of 500 N/m, the contact force is predicted to be
less than 50 lN.
maintaining VS's ¯exible probe shape control.
2
A new probe and measurement principle
The fabrication process of the RVS probe is shown in
Fig. 4. First, the PZT thin plate is glued to the WC sub-
In a new probe, the key is how to form a strain sensor on a strate. The PZT plate is patterned by sand-blast machin-
cylindrical cantilever machined by WEDG. Our solution
ing, forming PZT islands. The WC substrate is separated
is to divide the cantilever and strain sensing part as shown into pieces by wire cut EDM. Then the piece is placed onto
in Fig. 2. The main WC substrate is made of tungsten
an alumina substrate with the driving PZT and electric
carbide super hard alloy, which is a popular material as a wiring, forming a probe cartridge. At the last step, the
mechanical probe for its high hardness and Young's
probe cartridge is mounted on the WEDG machine, and
modulus. From the substrate, the cantilever part is shaped the cantilever and strain sensing part are machined.
by WEDG. The strain sensing part is a WEDG machined Forming the cantilever after cartridge assembly enables it
thin ¯at part on which a detecting PZT plate is glued.
to be accurately aligned to the cartridge, thus guaranteeing
When the cantilever is resonated by the driving PZT, its high mounting accuracy on the measurement machine.
strain is measured by the detecting PZT. The contact of the Finally, each manufactured probe cartridge can have a
probe can be detected by observing the amplitude of the custom-made probe shape for each application thanks to
strain signal. We call the technology ``resonance mode
WEDG's NC machining.

181
Fig. 5. Con®guration of measurement machine
Fig. 4. Fabrication process of RVS probe
3
Measurement machine
In conventional CMM, the measurement probe can easily
be observed by the naked eye. In HARMS measurement,
on the other hand, even inserting the probe into the region
of interest is a painful process. To handle HARMS's spe-
ci®c dif®culties, we have also developed a dedicated
measurement machine.
As in Fig. 5, the measurement system has a teaching
microscope, measurement probe, and XYZ stage. As the
positional relationship between the microscope focal
point and probe tip is ®xed, once a operator instructs
points in the microscope image, the probe can be auto-
matically inserted into the sample and perform the
measurement automatically. The sample is positioned by
an XYZ air slide system. The XY air slides are actuated by
two linear motors with feedback of a 10 nm resolution
Fig. 6. Micro hole cross section measurement
linear scale. The Z stage is guided by air slide and driven micropositioner so that the probe tip faces the measure-
by ball screw. The Z stage position is monitored by a
20 nm linear scale but employs no feedback control. The
smooth motion of the air slide guarantees submicron
measurement accuracy. The micropositioner, under
which the probe is vertically held as shown in Fig. 6,
consists of a PZT actuator, displacement amplifying
ment surface at the right angle.
4
Experiment
Here we demonstrate our system through some measured
results. Figure 7 shows measured results of a /200 lm
mechanism and a LVDT sensor. By the micropositioner, EDM hole in the tool steel plate of 1000 lm thickness. The
the tip contact condition is kept constant and its position inset photo is an optical microscope observation. Vertical
is monitored by LVDT. Since the micropositioner's servo and horizontal cross sections of the hole are illustrated in
control is limited to one direction, a rotary air spindle
the chart respectively to the left and right. The arrows
with an accuracy of over 0.1 lm is equipped. Using the from right to left show the corresponding layer at each
operator's registered instructions, the spindle rotates the depth. Although the optical image doesn't show any
Fig. 7. Vertical and horizontal cross sec-
tions of EDM micro hole

182
Fig. 8. Three dimensional rendering of etched micro hole
information of the inner surface, the vertical cross section
measurement clearly shows a barrel shape cross section.
This shape can be explained as electro discharge gap
variation along the axis; the gap is in¯uenced by the ma-
chined debris ejection condition. These results suggest that
such measurements can be used to optimize the machining
parameters.
Fig. 9. RVS probe shape variations
Figure 8 is the three dimensional display of a double-
side etched hole. An optical microscope image and a
vertical cross section are shown in the upper left and right
hand corners respectively. The measurement was done by
consecutive cross sectioning at 1.8-degree interval and all
cross sections are smoothly reconstructed in three-di-
mensional space later. The ®gure helps us to intuitively
perceive the hole shape and surface roughness. As this
measurement can be done totally nondestructively, it can
be a powerful tool for etching process control.
5
Discussion
Figure 9 shows several variations of the probe tip. Probe
ꢀa) is a commonly used shape. The axis is /40 lm and tip
length is 40 lm. Probe ꢀb) is a ball shaped probe, which is
an analogy from a CMM probe. Probe ꢀc) is an ultra high
aspect ratio probe, /12 lm in diameter and 1000 lm in
length. The minimum possible diameter as of present is
10 lm, attempts for smaller diameter resulted in the bend
of the probe caused by residual stress.
RVS application for HARMS measurement is basically
limited by the available probe shape. Figure 10 is a rough
sketch of the RVS application area, taking micro holes as
examples.
Fig. 10. RVS Application area in HARMS measurement
One drawback of the RVS probe is its relatively un-
certain tip radius. Because EDM uses electro discharge
heat to remove material, EDM craters on the workpiece are
inevitable. The tip curvature radius is directly affected by
the craters, and varies from 0.5 to 2 lm in our SEM ob-
servations. In order to make the tip shape constant, we are
now preparing a debarring process for the tips.
In the chart, CD-AFM and CMM application areas are
also drawn for comparison. The area of RVS ranges from 6
20 to 500 lm in diameter and 10±2000 lm in depth. The Conclusion
smallest diameter is de®ned by the thinnest possible
probe, and the maximum comes from the competition
with CMM. The maximum depth is decided by the WC
Based on a survey of the current technology for HARMS
measurement, we have proposed a new approach: the
resonance mode vibroscanning method ꢀRVS). The RVS
substrate size we prepare. As we limit the probe's aspect probe inherits favorable characteristics of both VS and
ratio to less than 100, the maximum aspect ratio of the
AFM methods. It is made of tungsten carbide super hard
hole should be less than 30. As is clear from the chart, the alloy and has ¯exibility in its shape and size ꢀVS proper-
RVS method ®lls the vacancy between CD-AFM and CMM. ties). It also has an integrated strain sensor to detect

contact at a low contact force ꢀAFM properties). The probe
is attached to a newly proposed HARMS dimensional
measurement machine. The internal pro®le of EDMed and
chemically etched holes were examined. These results
illustrate the validity of RVS for its intended application
as a method for HARMS measurement.
References
Indermuhle P-F et al. ꢀ1997) Atomic force microscopy using
cantilevers with integrated tips and piezoelectric layers for actu-
ation and detection. J Micromech Microeng 7: 218±220
Itoh T; Suga T ꢀ1993) Development of a force sensor for atomic
force microscopy using piezoelectric thin ®lms. Nanotechynology
4: 218±224
183
Masaki T et al. ꢀ1990) Micro electo-discharge machining and its
applications. Proc IEEE MEMS, 21±26
Masuzawa T et al. ꢀ1993) Vibroscanning method for nonde-
structive measurement of small holes. Annals of the CIRP, 42:
589±592
Masuzawa T et al. ꢀ1997) Twin-probe vibroscanning method for
dimensional measurement of microholes, Annals of the CIRP 46:
437±440
Pril WO et al. ꢀ1997) Development of a 2D probing system with
nanometer resolution. American Soc Prec Eng 16: 438±442
Schepperie K et al. ꢀ1998) Coordinate measuring apparatus
having a probe in the form of solid-state oscillator. USP5782004
Vachet G; Young M ꢀ1995) Critical dimension atomic force mi-
croscopy for 0.25-micrometer process development. Solid State
Technology 38: 57±62
Yamamoto M et al. ꢀ1996) Application of vibroscanning method
to on-the-machine measurement of micro electro discharge ma-
chining. Int J Japan Soc Prec Eng 30: 49±50