Page 1
1
Abstract— The report discusses procedures and methods of
fabrication of a designed micro cantilever beam. The cantilever
was fabricated using several techniques and which were carried
out in five main steps. These steps consist of patterning the
sacrificial layer, copper deposition, beam cantilever patterning,
dry etching, and wet etching for releasing the cantilever. Only a
small number of cantilevers were released at the end of the
process due to a number of factors discussed in the report.
Index Terms— Copper evaporation, micro cantilever, wet
etching, dry etching.
I. INTRODUCTION
ICRO and nanotechnology miniaturization has allowed
great achievements in fields such as biomedical,
aerospace, maritime, chemistry, and others. Micro
electromechanical systems (MEMS) are widely used in the
fields mentioned before for the use of sensors, accelerometers
and resonators among others. One of the most common
devices in MEMS is micro cantilevers. These microcantilevers
have several advantages such as ultra sensitivity, ease of
production and low cost [1].
Micro cantilever can be used as the sense element for
atomic force microscopy (AFM), high resolution chemical
analysis [2], and biosensors.
Due to the importance of micro cantilevers, the following
report shows the procedure for its fabrication.
II. MATERIAL AND METHODS
The materials used for the fabrication of the micro
cantilevers were:
Wyko NT 900 optical profilometer (Veeco, USA)
Wyko NT9100 optical interferometer (Veeco, USA)
Veeco’s Vision® analysis software (Veeco, USA)
Acetone (Merck Chemicals Ltd, UK) [3]
Copper beds 2.8 mm 99.99% purity
Thermal evaporator Pcod 1001133651 (Aldrich, USA)
Cas: 7440 – 508 Cu Usa Aldrich,
MicropositTM S1828 G2 positive (Rohmand Haas
Electronic Material, UK) [3]
MF-319 Developer (MicroChem Corp., Norway) [3]
Isopropanol (Merck Chemicals Ltd, UK) [4]
S-1828 G2 Positive Photoresist (Rohmand Haas
Electronic Materials, UK) [5]
UV exposure(Karl Suss)
Long Metal needle (for removing the bubbles)
Photoresist Spinner, Silicon wafer (4”)
Hotplate / Oven
Tweezers
Mask aligner
Mask (master pattern)
Large petri dish/dust proof storage box
Metal Plate( for Carrying the silicon wafer)
Rubber gloves
Safety Glasses
The experiments were performed in clean room where all
the materials and microscopes used are operated. The
fabrication process composed of five main steps: patterning
the sacrificial layer, copper deposition, beam cantilever
patterning, dry etching, and wet etching for releasing the
cantilever. The process can be seen Figure 1.
A. Patterning of the Sacrificial Layer
The silicon wafer was cleaned using nitrogen gas to remove
any contaminants present on its surface. The wafer was then
subjected to a dehydration process. In this process, the wafer
was baked for 2 minutes at 115oC. The wafer was then
allowed to cool down to the room temperature. The wafer was
then placed in a spinner and S-1818 Photoresist was poured
onto the wafer directly from the bottle. The spinner was then
set at the parameters shown in Table 1: Parameters of Spinner.
Table 1: Parameters of Spinner
Sequence Speed
(RPM)
Time (s) Acceleration
(RPM/s)
1 1000 50 100
2 700 10 500
This routine results in a photoresist thickness of roughly 5
µm. After the coating, the wafer was cleaned using Nitrogen
gas to remove any stray dust particles. Once the spin coating is
complete, the wafer was subjected to a soft baking session. In
this process, the wafer was baked for 2 minutes at 115oC.
Fabrication of a Designed Microcantilever
Beam
Salman Mahmood [email protected] , Antonio Martellota
[email protected] , Ronish Patel [email protected] , Amar Romi
[email protected] , Veralia Sánchez [email protected]
M
Page 2
2
Figure 1: Fabrication Process of Micro cantilevers
Once the post spin baking step was completed and the wafer
was cooled down to the room temperature and was exposed to
UV light using mask aligner for 40 seconds. The mask pattern
used was previously designed by the group using L-Edit
software. Figure 2 shows the mask for patterning sacrificial
layer.
Figure 2: Mask 1
Once the exposure process was complete, the wafer was
removed from the mask aligner and immersed into a developer
solution for 2 minutes at room temperature to remove the
unexposed resist from the wafer. This process also involves
mechanical agitation in which the wafer is continuously shaken
clock wise and counter clockwise in order to increase the
contact area between the wafer and the developer solution.
Once the excess photoresist was removed, the wafer was rinsed
with IPA solution and dried with nitrogen gas. The wafer was
then examined under the profilometer to review the structure.
The thickness of photoresist structures was observed to be
4.6µm.
B. Copper Deposition
The second step of the micro fabrication process involved
the deposition of a thin copper layer through thermal
evaporation. The equipment used for the deposition was a
thermal evaporator by Aldrich.
The thermal evaporation technique for deposition of thin
metal layer relies on the application of high AC current to a
filament of tungsten (high melting temperature) for the heating
and evaporation of metal beds. The target sample of the
deposition is positioned inside the evaporation chamber and
when the evaporation occurs, it is hit by the metal particles
which are propagating in all the directions. A metal layer will
thus be deposited on top of the sample. The growth rate, the
uniformity of the layer and its thickness strictly depend on
parameters such as the current amplitude and frequency, the
exposal time, and the amount of metal used.
A thickness sensor positioned inside the chamber, close to
the sample assists the monitoring of the process. The sensor
does not measure the real thickness of the layer deposited on
the sample but it estimates its actual thickness detecting the
increase of the metal mass deposited. Hence, to ensure proper
achievement of the desired thickness further analysis is
recommended.
In order to avoid the interaction between the vaporized
particles and the molecules of the atmosphere, the entire
procedure has to be conducted in high vacuum condition.
The substrate was positioned inside the chamber of the
machine, on the top side, facing down. A copper bed was
Page 3
3
hence placed on the dedicated indentation of the tungsten
beam at the bottom of the chamber. The process was
conducted for 53:50 minutes and the parameters shown in
Table 2: Parameters for Copper Thermal were used.
Table 2: Parameters for Copper Thermal Evaporation
Current amplitude 130 A
Initial current increasing rate 4.6 A/s
Density 8.93
Tooling 1000
Z-Factor 0.437
Life (%) 39.3
Figure 3 shows the screen of the thermal evaporator
software controlling system. The final achieved thickness
measured by the sensor was of 4.793 kA
Figure 3: Thermal Evaporation Controlling System
C. Beam Cantilever Patterning
The procedure of the beam cantilever patterning can be
split into cleaning of silicon substrate, S-1828 spin coating,
soft baking, UV Exposure, development, hard baking, wet
etching and thickness measurements using profilometer. All of
these steps are discussed in detail as follows:
1) Cleaning of silicon substrate
Cleaning of silicon wafer was performed by just flushing
the wafer with nitrogen gun since one mask was already
fabricated previously; hence, cleaning with water was not
required.
2) S-1828 spin coating
S-1828 spin coating was performed once the wafer
appeared to be clean and dry. To achieve a uniform and
desired thickness of photoresist on wafer it was necessary to
program the spin coater at different steps having different
acceleration. The wafer is then carried in metal plate from hot
plate to spin coater and carefully placed on small disc of spin
coater exactly at the center so that the vacuum under the small
disc of spin coater can firmly hold the wafer. After that, S1828
positive photoresist was poured according to the requirement
but it was recommended to pour a slightly extra quantity than
the required because it was observed that less than required
amount of S1828 significantly affects the thickness of
photoresist. Since S1828 is very viscous liquid, it was poured
at once all and the pouring was started exactly from the center.
The spin coated machine was programmed as shown in Table
3 .
Table 3: Spin Coating Parameters of the S1828 on the silicon
wafer
Steps Spin Speed
(rpm)
Accelerate Time(s) Decelerate
1 2500 500 5 -
2 2500 500 60 -
3 2500 - 5 500
3) Soft Baking
After the photoresist was applied to the substrate, it was soft
baked to evaporate the solvent and increases density of the
film. S1828 is normally baked on hot plate although
convection oven may be used; bake times was optimized since
solvent evaporation rate is influenced by heat transfer. To
obtained best results wafer was heated for 2 minutes at 1000 0C.
4) UV Exposure
S1828 is optimized for near UV (350-450 nm) exposure.
S1828 is virtually transparent and insensitive above 400 nm
but includes high actinic absorption below 350 nm. Over
exposure of the top portion of the resist film may result in
exaggerated negative sidewall profiles or T-topping. The
optimal exposure dose depends on film thickness (thicker
films require higher dosage) and process parameters. Figure 4
shows the mask used for cantilever patterning.
Figure 4: Mask 2
5) Development
The development of S-1828 resists was performed, which
was optimized for use with MF-319 Developer. Immersion,
spray or spray- puddle processes can be used. Other solvent
based developers such as ethyl lactate and diacetone alcohol
may also be used. Strong agitation was recommended for high
aspect ratio and/or thick film structures development process
was carried out for 45 seconds. After development, the wafer
was washed either with isopropanol and then diluted water.
Page 4
4
Table 4: Thicknesses before photoresist etching
Wafer
Anchor
Thickness(µm)
Cantilever
Thickness(µm)
1 2.74 6.75
Figure 5: Width and Length of Cantilever
6) Hard Baking
Post exposure baking was performed to selectively cross
link the exposed areas of the film. S1828 photoresist was
readily cross linked and can result in highly stressed film.
Therefore, to minimize stress heating of the wafer, post
exposure baking was performed in at 1000 C for 1 minute.
7) Wet Etching
Wet etching of the wafer was performed to remove copper
from wafer using alkaline solution of acetone. Alkaline
solution is compatible with the most of the metal resist,
relatively high etch rate typically 2-2.5 mils/min and is
environmental friendly and less undercut and sideways etch.
The etching was performed until etched copper was optically
visible on the wafer.
8) Thickness Measurements using Profilometer
The different wafers were coated by photoresist and the
thickness of photoresist on the wafers was measured using
profilometer. Figure 6 and Figure 7Error! Reference source
not found.Error! Reference source not found. shows the
thickness obtained from the wafer. Table 4 shows the
thickness measurement with photoresist before etching.
Error! Reference source not found. shows the width and
length of cantilever.
Figure 6: Thickness of Anchor profile from cantilever
Figure 7: Thickness of cantilever profile
D. Dry Etching
The dry etching process was performed using reactive ions
etch. The wafer used was directly dry etched to remove the
sacrificial layer. The wafer was exposed under O2 (48sccm)
plasma in RIE chamber for dry etching. Other etching
parameters were high RF power of 300W and low pressure of
100mTorr. The exposure time was 20 min. In this step, high
power atoms are bombarded at low pressure in plasma state.
The wafer was inspected under the profilometer after the
first etching exposure. It was observed that although plasma
etched the photoresist from top and inside of the cantilevers a
significant amount of resist was still present under the
cantilever beams.
The wafers were subjected to a second exposure of plasma
for 20 min. After second exposure of plasma, it was observed
that the resist was still visible under the beam; in addition,
some of the beam structures were damaged due to high power
plasma particles.
Figure 8 shows the profilometer image of cantilever after
second exposure with small quantity of photo-resist. It was
possible to observe that high power bombardment of plasma
atoms damaged the metal layer on the wafer. Figure 9 shows
the cantilever structure damaged due to RIE plasma during the
release process.
It was shown that power and pressure were the main factors
for the optimization of dry release process, variation of power
and pressure from high to low and vice versa can change the
plasma behavior inside the chamber.
A third exposure of plasma was given again to the wafers with
different parameters to optimize the RIE release process. In
this case, the wafer was exposed under O2 (48sccm) in plasma
in RIE chamber, RF power of 100W and high pressure of
500mTorr. The exposure time was 10 min. The power and
Page 5
5
pressure parameters of the RIE tool were changed from high
power to low pressure to high pressure as it created an
isotropic behavior of the plasma instead of anisotropic
behavior, which was observed after all the exposures. Table 5
summarizes the parameters of RIE.
The low RF power generated a weak bombardment of
plasma atom with high pressure. The sample was then studied
under the profilometer and it was observed that the photoresist
was cleaned from top and inside the beam, but there was still
some photoresist under the beam.
Figure 8: Cantilever profile after second RIE exposure
Figure 9: Cantilever profile after third RIE exposure
Table 5: Parameters of RIE
Anisotropic Isotropic
1st
exposure
2nd
exposure
3th
exposure
Pressure
(mTorr)
100 100 500
Power
(W)
300 300 100
O2
(Sccm)
48 48 48
Time
(min)
20 20 10
E. Wet Etching for Cantilever Release
Firstly, some measurements of the cantilever were
performed. An interferometer was used on which the wafer
was placed and the Vision computer program was started for
the analysis of the surface. The intensity window was used to
take the required measurements, obtain proper focus and
adjust light intensity. The fringe with the strongest intensity
was selected according to the appearance of red pixels, which
are at the maximum, though the measurements were taken
without the red pixels as shown in Figure 10.
Figure 10: Fringes in the Interferometer
Figure 11: measurement of the cantilever before wet-etching
The resulting thickness of the cantilever was ranging
between 2.5 and 5 µm, whereas the resulting thickness of the
anchor was 220.2 nm (Figure 11). Error! Reference source
not found. illustrates the measured readings of the cantilever
for the x and y profile. Figure 13 shows a capture of the wafer
using the vertical-scanning interferometry (VSI) that uses
multiple wavelengths of light and enables step measurements
up to eight millimeters [6]. Finally a 3-D profile was obtained
using the 3-D plot and interactive display as depicted in Figure
14Error! Reference source not found..
After all the measurements were taken, it was observed that
the release of the cantilevers was not achieved yet. Therefore,
a final step of wet etching was performed. The etchant used
was Acetone. Given the high difficulties in controlling the
Page 6
6
process and the short times of exposure required, the wafer
was divided into four pieces and different wet etching times
were adopted (see Table 6). Short etching times could lead to
an incomplete release, whereas long etching times could over
etch the structures and destroy them. Nevertheless, the main
risk of a wet etching of these structures was the collapse and
stiction on the substrate after the etching.
Table 6: Wet Etching Time
Wafer Piece Time (s)
1 60
2 40
3 25
4 7
Figure 12: the measuring of the cantilever for the x and y profile
After the wet etching was performed, the pieces of wafer
were immediately dried with nitrogen gas. The nitrogen gas
was not blown straight onto the wafer, but a soft towel paper
was put as a mask to avoid damage to the wafer. Figure 15
shows the release of one cantilever.
Following the release of the cantilever, measurements were
taken again. The optical profilometer was used first to perform
the measurements and Figure 16 was obtained. Then, the
interferometer was used to obtain more measurements as
shown in Figure 17 where the x-profile shows a resulting
measurement of 0.6051mm and a z-profile (anchor thickness)
of 0.2481 µm. Finally, Figure 18 shows the release of the
cantilever with the VSI mode.
Figure 13: VSI Image of Wafer
Figure 14: 3D profile
Figure 15: Release of Cantilever
Page 7
7
Figure 16: Measurement after Wet Etching with Profilometer
Figure 17: Measurement after Wet Etching with Interferometer
Figure 18: Release of the Cantilever VSI mode
III. RESULTS AND DISCUSSION
Figure 15 shows the final structure of the designed
cantilever. Among few hundreds of cantilevers designed
above the silicon substrate, just a couple of structures were
properly released and produced. The development time for the
wafer that produced the released cantilever was 7 seconds.
Figure 15 shows on the right a cantilever that was released and
on the left a cantilever that was not. It can also be seen that the
released cantilever is characterized by residual stress coming
from the various processes that have been conducted. This
stress caused the bending of the structure of roughly 45°.
The reason why only few cantilever were properly produced
relies on the difficulties that occurred during the releasing of
the structure. The decision of using dry etching first, rather
than wet etching was taken to reduce the risk for the structure
to collapse after releasing. This collapse would have occurred
due the surface tension of the liquid used. However, even
though the dry etching was performed three times, it did not
bring to a proper release of the structures and wet etching was
hence performed.
IV. CONCLUSION
A micro cantilever beam on based silicon wafer was
successfully produced. Several steps were performed for this
fabrication: a sacrificial photoresist layer was coated and
patterned through photolithography above the silicon wafer; a
thin copper layer was deposited with thermal evaporation
technique, a further photoresist layer was coated and patterned
for the wet etching of the copper layer; a dry etching process
was performed three times for trying an optimal release of the
structure avoiding stiction; a final wet etching process was
hence performed after having observed that the cantilevers
were not properly released after dry etching.
However, in the initial design few hundreds of cantilevers
were present. The most of the structures were damaged or
collapsed due to the stiction occurring during the final wet
etching process. Suggestions for a future fabrication would be:
design the cantilever with bigger holes and increasing the time
of the dry etching and use oxidation layer as a sacrificial layer.
V. REFERENCES
[1] A. Vasudev, "Microcantilever-Based Sensor," NNIN REU
2, Research Accomplishments, pp. 98-99, NNIN REU
2006 .
[2] B. Hubert, A. Nishimoto, M. Ottensmeyer, D. Robinson
and B. Romo, "Process Development for the Fabrication of
a Piezeresistive Microcantilever," MIT Media Lab, 1999.
[3] CHEMCUT, "Process Guideliness for Alkaline Etching,"
2002.
[4] MERCK, "Safety Data Sheet for Isopropanol," Merck
Chemicals, Nottingham, 2006.
[5] "Safety Data Sheet S1828," Rohmand Haas Electronic
Materials, Coventry, 2006.
[6] M. Zecchino, "Wyko Optical Profilers: Most Repeatable
Step Measurements," Veeco Instruments Inc.