You are on page 1of 6

Anal. Chem.

2004, 76, 5597-5602

Bonding of Glass Microfluidic Chips at Room

Zhi-Jian Jia, Qun Fang,* and Zhao-Lun Fang

Institute of Microanalytical Systems, Chemistry Department, Zhejiang University, Hangzhou, 310028, China

A simple, room-temperature bonding process was developed for the fabrication of glass microfluidic chips. Highquality bonding with high yields (>95%) was achieved
without the requirement of clean room facilities, programmed high-temperature furnaces, pressurized water
sources, adhesives, or pressurizing weights. The plates
to be bonded were sequentially prewashed with acetone,
detergent, high-flow-rate (10-20 m/s) tap water, and
absolute ethyl alcohol and were soaked in concentrated
sulfuric acid for 8-12 h. The plates were again washed
in high-flow-rate tap water for 5 min and, finally, with
demineralized water. The plates were bonded by bringing
the cleaned surfaces into close contact under a continuous
flow of demineralized water and air-dried at room temperature for more than 3 h. This bonding process features
simple operation, good smoothness of the plate surface,
and high bonding yield. The procedures can be readily
applied in any routine laboratory. The bonding strength
of glass chips thus produced, measured using a shear
force testing procedure, was higher than 6 kg/cm2. The
mechanism for the strong bonding strength is presumably
related to the formation of a hydrolyzed layer on the plate
surfaces after soaking the substrates in acid or water for
extended periods. Microfluidic chips bonded by the above
procedure were tested in the CE separation of fluorescein
isothiocyanate-labeled amino acids.
Chip-based microfluidic systems have attracted broad interest
in recent years as a major form for realizing the lab-on-a-chip, or
TAS, concept.1-3 Although microfluidic chips may be produced
from glass, quartz, silicon, and various polymeric materials, those
fabricated on glass substrates have been employed most frequently owing to their favorable optical properties and support of
reproducible electroosmotic flow.1-3 Fabrication of glass microfluidic chips usually involves micro channel/structure fabrication
using a photolithographic and wet-etching procedure, followed by
the bonding of the etched plate with a flat coverplate. Currently,
bonding processes involving high temperatures are employed
most frequently. Glass plates are brought into close contact with
* To whom correspondence should be addressed. Phone: +86-571-88273496.
Fax: +86-571-88273496. E-mail:
(1) Manz, A.; Fettinger, J. C.; Verpoorte, E.; Ludi, H.; Widmer, H. M.; Harrison,
D. J. Trends Anal. Chem. 1991, 10, 144.
(2) Harrison, D. J.; Manz, A.; Fan, Z.; Lu
di, H.; Widmer, H. M. Anal. Chem.
1992, 64, 1926.
(3) Jacobson, S. C.; Hergenro
der, R.; Koutny, L. B.; Warmack, R. J.; Ramsey,
J. M. Anal. Chem. 1994, 66, 1107.
10.1021/ac0494477 CCC: $27.50
Published on Web 08/11/2004

2004 American Chemical Society

each other under clean room conditions and are usually subjected
to sophisticated temperature programming in a high-temperature
furnace at 500-650 C.2-4 High-temperature bonding processes
usually involve complicated manipulations requiring considerable
expertise for achieving high yields and stringent laboratory
conditions. Weights made from quartz, graphite, or refractory
materials have often been imposed on the plates to increase the
bonding pressure. Although such measures may be effective in
improving the bonding quality and yield, our experiences have
shown that smoothness of the plate surface was often deteriorated,
which reduced the optical quality of the chips. Recently, Yin et
al.4 reported the high-temperature bonding of glass plates under
routine laboratory conditions involving mating the plates under a
continuous flow of demineralized water.
In contrast to high-temperature bonding processes, lowtemperature bonding of glass microfluidic chips usually involves
the use of various adhesives below 100 C. Such procedures are,
therefore, suitable for the fabrication of microfluidic devices
containing structural features that are nonresistant to high
temperatures, such as electrodes and waveguides. Bonding with
adhesives is also suitable for materials with significantly different
thermal expansion coefficients. Wang et al. bonded glass plates
at 90 C employing sodium silicate as an adhesive layer.5
Nakanishi et al. reported a room-temperature pressure-assisted
procedure for bonding quartz wafers by treating the wafer surface
with dilute hydrofluoric acid.6,7 Sayah et al. described two methods
for direct pressure-assisted low-temperature bonding of glass
substrates,8 involving epoxy gluing at 90 C in the first method
and exposure of the glass stack to high pressure in the 100-200
C temperature range in the second. Chiem et al. proposed a roomtemperature bonding procedure featuring stringent washing of
the plate surfaces under clean-room conditions.9 A UV-curable glue
was used by Huang et al.10 for bonding glass chips at room
temperatures. The glue was filled into guide channels on the
microchip architecture, followed by exposure to UV light.
(4) Yin, X. F.; Shen, H.; Fang, Z. L. Fenxi Huaxue (Chinese J. Anal. Chem.)
2003, 31, 116.
(5) Wang, H. Y.; Foote, R. S.; Jacobson, S. C.; Schneibel, J. H.; Ramsey, J. M.
Sens. Actuators, B 1997, 45, 199.
(6) Nakanishi, H.; Nishimoto, T.; Nakamura, N.; Nagamachi, S.; Arai, A.; Iwata,
Y.; Mito, Y. IEEE. 1997, 299.
(7) Nakanishi, H.; Nishimoto, T.; Kanai, M.; Saitoh, T.; Nakamura, K.; Shoji,
S. Proc. Transducers 99, Sendai, Japan, 1999, 1332.
(8) Sayah, A.; Solignac, D.; Cueni, T.; Gijs, M. A. M. Sens. Actuators, A 2000,
84, 103.
(9) Chiem, N.; Lockyear-Shultz, L.; Andersson, P.; Skinner, C.; Harrison, D. J.
Sens. Actuators, B 2000, 63, 147.

Analytical Chemistry, Vol. 76, No. 18, September 15, 2004 5597

In this work, we report a simple low-temperature bonding

process with high bonding quality and yield and without the need
for clean-room conditions, adhesives, pressurizing equipment, and
temperature programmable furnaces.
Apparatus and Equipment. UV lithography was performed
using a model JKG-2A lithographic device (Photomechanical Co.
Shanghai). Thermal bonding for glass chips was performed in a
programmable furnace (model SX2-4-10, Shenyang Electrical
Furnace Co., Shenyang). An electronic testing machine (model
WDS-100, KN, Shijin Group Co., Jinan) was used for measurement
of bonding strength of the glass chips. A model 9323-HVPS HV
power supply (Institute of Applied New Technology, Beijing),
variable in the range 0-30 kV, was used for endurance testing of
the bonded chips under high voltage.
A homemade high-voltage power unit,11 variable in the range
0-1500 V was used for on-chip sample injection and CE separation. A LIF detection device, as detailed in a previous report,11
with an argon ion laser (488 nm, 4 mW, model 367, Nanjing
Electronic Equipment Co. Nanjing) was employed for detection.
The signal output from the detector was recorded using a model
XWTD-164 chart recorder (Dahua Instruments. Shanghai).
Reagents. All reagents used were of analytical grade, and
demineralized water was used throughout. Sodium tertraborate
buffer (5 mM, pH 9.2) was used as the working electrolyte for
capillary electrophoresis (CE) separation. Stock solutions of 10
mM amino acids were prepared from L-arginine, D,L--phenylalanine, and glycine purchased from Kangda Amino Acid Works
(Shanghai, China). Fluorescein isothiocyanate (36 mM, FITC)
labeling reagent was prepared by dissolving 8.5 mg of FITC
obtained from Acros Organics (Geel, Belgium) in 0.6 mL of
acetone (with 0.03 mL of pyridine added) before use.11
FITC-labeled amino acid solutions containing 2.5 mM of each
amino acid were prepared by mixing 0.5 mL of the stock solutions
of each of the amino acids with 1.3 mL of 10 mM sodium
tetraborate buffer (pH 9.2) and 0.2 mL of FITC reagent. The
mixture was allowed to stand overnight in the dark. A working
solution containing a mixture of 20 M of each amino acid was
prepared by mixing 0.4 mL of each of the labeled amino acid
solutions and diluting to 50 mL with 5 mM borate buffer.
Procedures. Fabrication of Glass Chips. Photolithographic and
wet chemical etching techniques were used for fabricating
microchannels onto a 1.6-mm-thick 20 60-mm glass substrate.
Type SG2506 glass substrate with chromium and AZ1805 photoresist coating and glass cover plates of the same dimensions were
obtained from Shaoguang Microeletronics Corp. (Changsha,
China). Production of the photomask and procedures for fabrication of microchannels were previously described.4 Briefly, a design
on a photomask with microchannels was transferred onto the glass
plate following a UV exposure. The microchannels were etched
into the plate in a well-stirred bath containing dilute HF/NH4F.
Four 1.2-mm-diameter access holes were drilled on the etched
plate at channel terminals using an emery drill-bit, forming four
reservoirs. Microchips with a crossed-channel design were used
for testing the bonding strength as well as their performance in
(10) Huang, Z. L.; Sanders, J. C.; Dunsmor, C.; Ahmadzadeh, H.; Lander, J. P.
Electrophoresis 2001, 11, 382.
(11) Fang, Q.; Xu, G. M.; Fang, Z. L. Anal. Chem. 2002, 74, 1223.

5598 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004

Figure 1. Processes of the room-temperature bonding approach:

(a) plates washed with tap water at high flow rate (10-20 m/s); (b)
plates soaked in concentrated H2SO4 or water; (c) plates brought into
close contact under a continuously flowing stream of demineralized
water; (d) combined plates stand at room temperature for more than
3 h.

CE separations. Channel dimensions were 30 m deep and 85

m wide, with a 10-mm-long sampling channel, and a 40-mm
separation channel.
The bonding process of the glass microchips was as follows:
Etched substrates and cover plates were washed sequentially with
acetone, household dishwashing detergent, tap water at high flowrate (10-20 m/s), and ethanol to remove solid particles and
organic contaminants from the glass surface as well as the micro
features. The plates were dried with a hair drier before being
soaked in concentrated sulfuric acid for 8-12 h (as shown in
Figure 1a). Subsequently, the plates were aligned vertically; held
with a space of 1-2 mm between the surfaces, the channel
structures of the etched plate facing inside; and washed again
under a low flow of tap water (1.3-2.3 m/s) for 5 min. Finally,
the aligned plates were brought into close contact under a
continuous stream of demineralized water flowing between the
two vertically held plates (Figure 1b). The combined plates were

Figure 2. Schematic diagram of chip structure for endurable shear

force testing. The device was composed of two 30 14 1.2 mm
glass plates with 3-mm-i.d. holes, bonded with an overlapping area
of 15 14 mm.

then allowed to stand at room temperature for more than 3 h (as

shown in Figure 1c) after removing the residual water film from
the exposed chip surface using filter paper. The bonding was weak
during the first few hours, and care was taken not to disturb the
position of the aligned plates during this period. This period could
be shortened to 15 min by drying the chips with a hair drier (80100 C). When bonding quality is unsatisfactory owing to insufficient treatment, which could be observed by the appearance of
interference fringes, the bonded plates may be separated for
recleaning immediately after bonding. This could be achieved by
soaking the chip in 50-60 C water for a few minutes and then
carefully forcing a sharp blade into the bonding interface.
Testing of Bonding Quality for Chips. Tests were carried out
using the bonded chips produced from type SG2506 blank glass
substrate with a bonding area of 15 15 mm2. The maximum
shear force endurable by the chips (before the bonding failed)
was measured using an electronic testing machine. A schematic
diagram of the testing arrangement is shown in Figure 2. Holes
(3-mm) were drilled on the two bonded plates at the center of
the nonbonded regions. Nylon strings were inserted through the
holes and fixed on the testing machine. The plates were then
pulled with the force increasing until a bonding failure occurred.
The endurance to high voltage of the room-temperature
bonded chips with crossed-channel design was tested by continuously applying a voltage drop of 1 kV at two reservoirs connected
by a 4-cm-long channel (field strength, 250 V/cm), filled with 5
mM borate buffer, for 50-100 h. The current was digitally
displayed, and recorded at 15-min intervals. Potentials of 6-15
kV were also applied under otherwise identical conditions for
identifying an electrical breakdown value.
The endurance of the room-temperature-bonded chips to
hydrodynamic pressure was tested using two 20 60-mm chips
on which were fabricated 85 m (w) 30 m (d) 40 mm (l)
straight channels, with 1.5-mm diameter holes drilled at the
channel terminals. MicroLine tubings (0.5 mm i.d., 1.5 mm o.d.;
Thermoplastics, Sterling, NY) were inserted in the inlet holes of
the chips and secured with epoxy. The distal ends of the tubings
were push-fitted over the outlet needle of a syringe micropump
(model WZS-50F, ZJU Medical Instruments, Hangzhou) and
secured by applying epoxy at the connection point. Demineralized
water was pumped through the microchannels at varying flow
rates while they were monitored for leakages at the bonding
surfaces. A further test was conducted by directly connecting a
plastic syringe to a conical pipet tip epoxied to the chip inlet and
applying pressure to the water-charged syringe using the electronic testing equipment employed for bonding strength testing.
The hydrodynamic pressures exerted on the walls of the chip
channels were estimated from the meter readings and the plunger
cross-sectional area of the syringe.

CE Separation of Amino Acids. Chip-based CE separation with

LIF detection employing the room-temperature-bonded chips was
performed using a multicontact high-voltage power supply. Before
use, the microchannels on chips were treated with 0.1 M NaOH
or concentrated H2SO4 for 1 h and rinsed with demineralized water
and 5 mM sodium tetraborate buffer (pH 9.2). Buffer solution was
filled into the buffer and waste reservoirs, and a sample with FITClabeled amino acids was filled into the sample reservoir. Platinum
wire electrodes were inserted into the reservoirs. The sample was
injected into the CE separation channel by pinched injection mode.
A field strength of 220 V/cm was applied for CE separation, with
an effective separation length of 18 mm.
Safety Considerations. The soaking of glass plates in concentrated sulfuric acid should be handled with care using PTFE forceps,
while wearing protective gloves and goggles. Plates should not be
immersed in the acid immediately after washing, but should be first
dried as indicated in the procedures. Immersing of wet plates into
the acid can cause disastrous results. The vessel containing the acid
should be covered and labeled during treatment.
Cleaning and Bonding of Glass Substrates. One of the most
important factors affecting successful bonding of planar glass chips
is the cleanliness of the bonding surfaces of glass substrates. This
was reported by Chiem et al.9 and has now become a broadly
accepted reason for the requirement of a clean-room environment
during chip fabrication. Other factors often reported included
surface flatness of the glass plates, bonding temperature, and
pressure. Our experiences also strongly support the importance
of cleanliness, having shown that the most unsuccessful bonding
events are often associated with solid particles or organic material
remaining on the glass surfaces, for example, dust, residual
photoresist or glass particles, and grease, etc., prior to bonding.
The washing and cleaning of glass substrates are often
performed using detergents, water-soluble organic solvents (e.g.,
ethanol, acetone), concentrated sulfuric acid, and high-flow-rate
tap water. High-flow-rate tap water showed better effects for
removing minute solid particles on chip surfaces.9 In this work,
most surface contaminants were removed by such treatment, and
those remaining in the microstructures were removed using a
cotton swab soaked with absolute ethyl alcohol. To remove further
residual organic contaminants, the plates were immersed in
concentrated sulfuric acid and soaked for 8-12 h. This soaking
procedure proved to be most effective for eliminating such
contaminants and also significantly contributed to the formation
of a hydrolyzed layer on the glass surface, which is important for
improving bonding quality and yield at room temperature12 (see
the Evaluation of Bonding Quality section for details). We found
that glass substrates that are free from organic contamination do
not require such rigorous treatment. Instead, soaking in demineralized water or dilute sulfuric acid solution for the same or
shorter period was in most cases sufficient for achieving satisfactory bonding.
Another important practice before bonding is preventing
exposure of the cleaned and hydrolyzed glass surfaces to an
unclean environment, such as that in a routine laboratory. This
(12) Fang, Q.; Jia, Z. J.; Fang, Z. L. Chinese Patent, Appl. no. 200410016216.3,

Analytical Chemistry, Vol. 76, No. 18, September 15, 2004


is the reason most published processes required operations under

clean-room conditions. In an earlier report from our group,
employing a high-temperature bonding process, glass substrates
and cover plates were brought into contact under a continuous
stream of demineralized water flow4 to avoid contamination from
the environment, immediately before bonding (see Figure 1b).
This approach was adopted in the room-temperature bonding
process to avoid the need for clean-room facilities. In contrast to
our previous approach, which still required a high-temperature
treatment, in this method, reliable bonding was achieved at room
temperature following a period of standing. This is due to the more
stringent prebonding treatment discussed in the preceding
paragraph. A more than 95% high bonding yield was achieved in
our routine laboratory during the fabrication of 150 pieces of glass
microchips with various dimensions.
Surface flatness of the glass plates is another factor that affects
close contact of surfaces and, therefore, the bonding quality. These
effects were studied using planar substrates and cover plates,
described in the Experimental Section (0-2 m flatness), as well
as untreated microscope slides (3-5 m flatness) with identical
chip areas (25 25 mm). Stronger bondings were obtained for
the former compared with the latter, with bonded areas of 100%
(20 60 mm2 chip area) and 30-60% (15 25 mm2 chip area),
respectively. The bonding yield for microscope slides was strongly
dependent on the chip area, with the yield increasing for smaller
chip areas, apparently because extreme flatness is more difficult
to ensure over larger areas. Thus, for a 13 15 mm2 chip, the
bonded area was increased to over 90%.
Although the room-temperature bonding procedure is sufficient
for most chip-based applications, a stronger permanent bonding
may be achieved by a subsequent high-temperature treatment if
required. In that case, the present approach may be used as a
pretreatment before the furnace heating process, during which
no pressurizing weights (which could affect glass surface smoothness) on the chip surfaces are then required, and almost 100%
bonding yield could be obtained. Typical high-temperature-treated
glass chips produced using pressurized weights during heating
and one without weights following room-temperature bonding are
shown in Figure 3a and b, respectively.
Mechanism of Room-Temperature Bonding for Glass
Chips. Our work has shown that considerably strong bonding of
glass chips, sufficient for most microfluidic operations, could be
obtained at room temperature by employing the approach described in the Experimental section. In a low-temperature bonding
approach for glass chips employing sodium silicate as an adhesive,
Wang et al.5 proposed a bonding mechanism involving siloxane
bond formation between the sodium silicate and glass surfaces.
Our observations and results support the assumption that a
hydrolyzed gel layer is formed on the glass substrate surface after
soaking the plates in acid or water for relatively extended periods
of at least a few hours, during which most of the Si-ONa groups
near the surface of the chips are transformed into Si-OH groups,
as shown in Figure 4a. After intimate contact of the hydrolyzed
surface and standing at room temperature, the Si-OH groups
gradually dehydrate, forming siloxane bonds, and terminating with
a condensation-polymerization reaction, as shown in Figure 4b.
The natural dehydration process is slow, apparently being
determined by the speed of water evaporation from the bonded
5600 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004

Figure 3. Comparison of surface smoothness of glass chips (a)

using pressurizing weights in high-temperature bonding process and
(b) with the room-temperature bonding method, both after being
subjected to the same high-temperature treatment.

surface, and in this work, usually 30 days was required to reach

an equilibrium state, when maximum bonding strength was
achieved. Such observations also support the presumed bonding
Evaluation of Bonding Quality. The bonding quality of glass
chips using the present approach was compared with those using
a conventional high-temperature bonding method employing both
SEM images and bonding-strength testing. The cross sections of
a bonded chip using the different bonding approaches are shown
in Figure 5. No obvious differences at the bonding interface were
In this study, the bonding strengths of room-temperature (1525 C)-bonded chips following different standing periods were
evaluated using a shear force testing technique.
A total number of 100 separate chips were subjected to the
tests, and successfully bonded chips (98% bonding yield) were
used for testing. The effects of postbonding standing time on
bonding strength for chips soaked in water or sulfuric acid were
studied, and the results are shown in Figure 6a and b, respectively.
The first measurements were made 3 h after initial bonding, and
tests were conducted until a 1-month standing period. Since each
chip was destroyed during the test, the data for varying standing
times were collected from different chips. For either mode of
pretreatment, bonding strengths increased on standing, and
gradually stabilized at a maximum value 20-30 days after bonding.
A maximum of 72.2 ( 4.1 N/cm2 was reached for plates treated
with acid, as compared to 61.8 ( 2.6 N/cm2 for those soaked in
water. It should be noted that 80% of maximum bonding strength
could be achieved 3-4 days after bonding, when the strength was
sufficient for most applications. It should also be noted that
although the bonding strengths of chips soaked in water or acid
did not differ significantly, the yield for successful bonding of
water-treated chips was significantly lower, by 20%. This might
be explained by the stronger cleaning power of the acid in
removing minute particles from the glass surface.
Endurances of the room-temperature-bonded chips to high
voltage and hydrodynamic pressure were tested as described in
the Experimental Section. Bonding quality was monitored by

Figure 4. Presumed mechanism for room-temperature bonding process: (a) the formation of hydrolyzed layer on the substrate surface and
(b) dehydration and condensation/polymerization of Si-OH groups.

Figure 5. SEM images of channel cross-sectional profiles of glass

chips fabricated using different boding procedures: (a) high-temperature bonding and (b) room-temperature bonding.

observing current fluctuations under different applied voltages.

With an applied voltage of 1 kV (250 V/cm field strength), the
observed variation in current during a total testing time of 50100 h for 3 chips showed an average reproducibility of 2.4-4.7%
RSD. The largest RSD for a single chip was 4.7% (n ) 24). The
longest continuously operated period during this study was 12 h,
during which the current RSD was 4.3% (n ) 30). Under an applied
voltage of 1.5 kV (375 V/cm field strength), results similar to those
obtained under 250 V/cm field strength were obtained. Under
applied voltages of 6-8 kV (1.5-2.0 kV/cm field strength),
constant currents were maintained for at least 3 h. When higher
voltages (>12 kV, 3.0 kV/cm field strength) were applied,
excessive current fluctuations occurred owing to partial bonding
failure. Endurance of the room-temperature-bonded chips to
hydrodynamic pressure was tested by increasing the flow rate of
demineralized water through the chip channel described in the
Experimental Section to a maximum of 100 L/min (corresponding to a linear flow rate of 65 cm/s), leaving a residence time of
only 60 ms for any fluid element flowing through the channel.
Above this flow rate, the pump stopped working owing to actuation

Figure 6. Effect of standing time vs bonding strength of glass chips

using room-temperature bonding approach: (a) plates soaked in
concentrated H2SO4 and (b) plates soaked in water, before bonding.

of the protective circuitry when an upper pressure limit was

exceeded. A further test was conducted using an electronic testing
device, as described in the Experimental Section. When a pressure
of 2.8 MPa (30 atm) was reached within the channel, with the
flow-rate achieving 700 L/min, the epoxied connection between
the syringe and chip broke away, No leakages or interference
fringes were observed at the bonding surfaces during both tests,
indicating satisfactory performance of the chip, even under the
most demanding hydrodynamic conditions encountered in microfluidic applications.
The reliability and performance of glass microfluidic chips
bonded at room temperature were tested in the CE separation of
a mixture of FITC-labeled amino acids under an applied voltage
of 800 V. Electropherograms obtained are shown in Figure 7b,
with a comparison to those obtained using a high-temperatureAnalytical Chemistry, Vol. 76, No. 18, September 15, 2004


Figure 7. Typical LIF electropherograms of repetitively introduced samples of a mixture of 20 M FITC-labeled arginine, phenylalanine, and
glycine employing glass microchips fabricated by (a) conventional high-temperature bonding method and (b) room-temperature bonding method.
Working electrolyte, 5 mM borax buffer; applied field strengths for filling of sampling channel and CE separation were 500 and 200 V/cm,
respectively; effective separation length, 18 mm.

bonded chip fabricated with conventional procedures (Figure 7a).

RSD values of the fluorescence intensity ranged from 2.1% for
arginine to 3.0% for glycine (n ) 10) using conventional hightemperature-bonded chips (plate numbers 7000-11000), and
ranged from 1.9% for arginine to 1.5% for glycine (n ) 21) using
a room-temperature-bonded chip (plate numbers 7600-13000).
Precision of the residence times for different amino acids using
chips fabricated by the high-temperature and room-temperature
bonding approaches were both better than 1% RSD. The results
show that the CE performance of room-temperature-bonded chips
are comparable to those obtained using chips bonded at high
Effective bonding of glass chips has been achieved at room
temperature without the requirement of clean-room facilities,
adhesives, pressurized water sources, high-temperature furnaces,
or pressurizing weights. Bonding the plates within a continuous
stream of water avoided the requirement of a clean room facility.
The method has been employed in our routine laboratory for more
than a year with high yields, during which time glass microfluidic
chips with various dimensions and designs were fabricated,
demonstrating performance comparable to high-temperature-


Analytical Chemistry, Vol. 76, No. 18, September 15, 2004

bonded chips. Three to four days is required for stabilizing the

bonding quality, which is sufficiently short for most applications.
However, for urgent use, the room-temperature bonded chips may
be subjected to a high-temperature treatment at 550 C immediately after forced drying. No weights were required during
the latter treatment, and chips may be ready for use on the very
day of designing the channel structures. The labor involved in
the microfabrication of glass devices is reduced by eliminating
the thermal bonding process. Avoiding the use of clean room
facilities and programmed furnaces allows the microfabrication
process to be easily mastered by students and technicians in
almost every analytical lab and with very limited investment.
Proliferation of the technique could significantly accelerate
research in microfluidic systems, currently often limited by lack
of on-site microfabrication capabilities for optimizing chip designs.
This work was supported by the National Natural Science
Foundation of China (Grant 20299030).
Received for review April 10, 2004. Accepted July 2, 2004.