Bonding Quartz Glass at Room Temperature
Tetsuya Hamaguchi, Natsuki Yamamoto, Takeshi Ooi, Kensuke Tsuchiya, Masayuki Nakao
The University of Tokyo, Graduate School of Engineering
7-3-1 Hongo, Bunkyo-ku, 113-8656 Tokyo, Japan
1. Introduction
The authors are developing a method for directly bonding two pieces of quartz glass at
room temperature. The conventional way of bonding glass with additives, like soda glass or
borosilicate glass, is anodic bonding. For bonding pure quartz glass, ways exist by softening the
surface with high temperature, melting the surface with hydrogen fluoride acid (HF)[1], and using
glue. Applying these methods, however, for bonding micro Total Analysis System (µ-TAS ) chips
(Fig. 1), causes surface deformation or glue entering the micro grooves and destroys the
micro-features on the chip surfaces. Also, optical parts that pass ultraviolet ray are made of
quartz glass and gluing them together causes drops in transmissivity of the ultraviolet ray.
We thus want to develop a technology to directly bond two pieces of quartz glass together.
For direct bonding of metal, a known technique[2] bonds metal surfaces activated by inert gas ion
beam or fast atom beam (FAB) irradiation cleaning under vacuum. We applied FAB irradiation
and succeeded in bonding quartz glass. This paper reports the relations of bonding strength
versus quartz glass surface roughness, FAB irradiation time, FAB source gas type and bonding
pressure.
2. Test equipment and procedure
Fig. 2 sketches our test equipment, with a vacuum chamber, 2 FAB guns, 2 samples
fixtures, and a pressing mechanism. The FAB guns are oriented to irradiate FAB on two samples
both at angles of 45 degrees. The pressing mechanism has a load cell to measure the bonding
pressure.
FAB
FAB
Vacuum pump
bonding
Samples
hundreds
of nano
meter order
Load cell
Fig. 1 Example of µ –TAS Chip
Fig. 2 Test equipment
The test proceeds as follows: First the sample surfaces are cleaned with organic solvent,
a mixture of H2SO4+H2O2 and then with NH4F. The cleaning process then takes the samples
through boiling in a H2SO4+H2O2 mixture, and ends with pure water rinsing and drying. The two
samples fixtures hold the sample to thoroughly dry them. The fixtures are then set inside the
vacuum chamber which we vacuum down to 7.5 x 10-4Pa. We then fill the chamber with the
source gas at a pressure of 1.3 x 10-2Pa and irradiate FAB on the sample surface. After the set
irradiation time, with the FAB irradiation running, the pressing mechanism presses the two
samples together to bond them.
We then take the bonded sample out from the vacuum chamber and set it on the tensile
test machine with the fixtures holding it in place. The load cell measures the force when the
bonded surface separate to give the bonding strength value.
The samples were quartz glass plates of 6 x 6mm with a thickness of 0.2mm. We
prepared two types of sample with different surface roughness values of 0.17 and 3.2nmRa. We
used an atomic force microscope (AFM) for measuring the surface roughness. For FAB source gas,
we readied Ar gas, an inert gas commonly used for diffusion bonding, and CHF3 gas expecting
some chemical reaction with quartz glass.
3. Test results and discussion
3-1. FAB irradiation time and surface roughness
We expected that the sample surface roughness should strongly affect the bonding
strength. Cleaning the sample surface requires FAB irradiation for its activation and we were
concerned that the irradiation may degrade the surface roughness. We thus tested how the FAB
irradiation time affects the surface roughness. We applied CHF3 gas FAB irradiation to the
sample with an original surface roughness of 3.2nmRa, and both the CHF3 gas and Ar gas FAB
irradiation to ones with 0.17nmRa. Fig. 3 shows the test results, and Fig. 4, the AFM images of
the sample surface before and after 300 seconds of CHF3 gas FAB irradiation. The figures show
Surface Roughness (nm Ra)
6
5
CHF3
4
3
CHF3
2
1
Ar
0
0
200
400
600
FAB irradiation time (sec)
800
Fig.3 Relation between FAB irradiation
time and surface roughness
6 (nm)
4.0
0
(nm)
0
5
5
(µm)
(µm)
2.0 0.0
10
6 (nm)
(a) Before FAB irradiation (surface roughness = 0.17 nmRa)
4.0
0
(nm)
5
(µm)
5
10
0
(µm)
2.0 0.0
(b) After 300 seconds of FAB irradiation (surface roughness = 0.31 nmRa)
Source gas: CHF3
Fig.4 Change of surface roughness with FAB irradiation
that FAB irradiation degrades surface roughness.
Longer irradiation worsens the surface roughness with both Ar and CHF3 gas. When
neutral atoms hit the sample surface, they clean the surface by removing molecules and
contamination, but at the same time, roughs the glass surface by hitting it.
3-2. Bonding strength and FAB source gas
Fig. 5 shows the test results. The horizontal axis gives the surface roughness after FAB
irradiation, and the vertical axis the bonding strength. Ar gas failed bonding the glass plates
with both surface roughness values. CHF3 gas, on the other hand, succeeded in bonding the two
pieces. A component of the CHF3 gas must have caused some chemical reaction on the quartz
glass, however, the detail of this process has not been clarified.
With CHF3 gas, the bonding strength maximized when the surface roughness was about
0.25nmRa. Samples with smoother surfaces had shorter FAB irradiation time. The results
indicate that smoother surfaces produce larger contact area and thus stronger bonding strength,
however, if the FAB irradiation time is too short for smoothness, the process loses the cleaning
effect and the bonding strength drops.
3-3. Bonding strength and bonding pressure
Once FAB irradiation is complete, the two samples are pushed together to make the
bonding and Fig. 6 shows the bonding pressure in the horizontal axis and the bonding strength in
the vertical. We compared the results with samples made of gold and quartz glass. The quartz
glass samples had surface roughness of 0.31nmRa after FAB irradiation with CHF3 gas. Larger
bonding pressure generated stronger bonding strength for gold. The smaller Young’s modulus of
gold produced greater contact area with larger bonding pressure. In contrast, quartz glass has a
large Young’s modulus and increasing the bonding pressure does not affect the contact area. The
bonding strength did not increase with the bonding pressure.
16
Bonding Strength (MPa)
Bonding Strength (MPa)
1
0.8
CHF3
0.6
0.4
0.2
Ar
0
0
0.2
0.4
0.6
0.8
1
1.2
14
12
Au
10
8
6
4
Quartz glass
2
0
0
Surface Roughness (nm Ra)
Fig.5 Relation between surface
roughness and bonding strength
2
4
Bonding Pressure (Mpa)
6
Fig.6 Relation between bonding pressure and
bonding strength (CHF3, 0.31nmRa)
4. Conclusions
We confirmed that we can directly bond two pieces of quartz glass at room temperature by
using FAB. The following conditions are important for stronger bonding strength:
-
CHF3 gas is effective for the FAB source gas.
-
Surface roughness of 0.25nmRa after FAB irradiation produces the best results .
-
The bonding pressure, when it is 3MPa or less, does not affect the bonding strength.
-
The bonding strength will improve if we can clean the surface without affecting the surface
roughness.
References
[1]Akihide Hibara, et al.: Nanochannels on a Fused-Silica Microchip and Liquid Properties
Investigation by Time-Resolved Fluorescence Measurements, ANALYTICAL CHEMISTRY,
Vol.74, pp.6170-6176, 2002.
[2]Yuzo Matsuzawa, Toshihiro Itoh, and Tadatomo Suga: Room-temperture Interconnection of
Electroplated Au Microbump by
means of Surface Activated Bonding Method, Electronic
Components and Technology Conference, pp.384-387, 2001.