Microfluid Nanofluid (2006) 2: 185–193
DOI 10.1007/s10404-005-0068-1
R EV IE W
Jeffrey L. Perry Æ Satish G. Kandlikar
Review of fabrication of nanochannels for single phase liquid flow
Received: 1 August 2005 / Accepted: 15 November 2005 / Published online: 9 December 2005
Springer-Verlag 2005
Abstract The topic of single phase liquid flow in submicron or nanochannels is a nascent field. There have
only been a couple papers that have dealt with this area
directly. The most probable reason for this is that currently most research in fluid mechanics or heat transfer
is being focused on micron size channels. To help facilitate researchers to focus on this undeveloped area, this
paper serves as a review for some of the micro-fabrication processes that will make it possible for engineers
and scientists to study this field in greater detail.
Keywords Fabrication Æ Nanochannels Æ Review
List of symbols
a
b
d
E
h
L
R
t
c
Trench width of NIL mold (m)
Ridge width of NIL mold (m)
Trench depth of NIL mold (m)
Young’s modulus (Pa)
Trench or channel height (m)
Wafer thickness (m)
Channel width (m)
Initial thickness of polymer layer (m)
Surface energy (J/m2)
J. L. Perry
Microsystems Engineering,
Rochester Institute of Technology,
Rochester, NY 14623, USA
S. G. Kandlikar (&)
Department of Mechanical Engineering,
Rochester Institute of Technology,
Rochester, NY, USA
E-mail: [email protected]
Tel.: +1-585-4756728
Fax: +1-585-4757710
1 Introduction
The nanometer length scale will allow discovery of a new
range of phenomena, where the channel height is on the
order of the size of atoms or molecules comprising the
fluid or dissolved/dispersed material in it. First, there is a
need to study flow in nanochannels because the tremendous potential of nanofluidics is yet to be explored.
Secondly, nanofluidics may evolve to be a key technology just as microfluidics has come to be a part of the
many technological advances of the modern era.
Nanofabrication and nanometer-scale fluidic structures have in recent years provided new tools for the
study of molecular behavior at the single-molecule level.
Nanofluidics is expected to find significant applications
in biotechnology and medicine (Petersen et al. 1998).
Therefore, the study of fluid dynamics and bimolecular
transport on the nanometer-scale is relevant.
A major upcoming application of nanochannels is in
the analysis of DNA. Researchers in this field have
found that qualitatively the degree of DNA stretching is
inversely proportional to the channel dimensions due to
confinement effects (Guo et al. 2004). In biological
applications, the interaction of biopolymers, such as
DNA molecules in nanochannels with dimensions close
to the persistence length (length to which a molecule can
be laid out in a straight manner) allows for a whole new
way of detecting, analyzing and separating these biomolecules (Guo et al. 2004). Typically a DNA molecule
will form a compact arrangement in its natural state.
However, when a DNA molecule flows through a
nanochannel with a cross section comparable to the
persistence length of the molecule (50 nm), it will be
thermodynamically more favorable for the DNA molecule to be in a stretched state (Guo et al. 2004). This
DNA stretching can lead to important biological
applications such as: (a) quick mapping of restrictioncut genomic DNA segments in very short times (minutes
vs. hours or days), (b) reduction in required DNA
sample to that of the genomic material in a single cell, (c)
186
to localize transcription factors for protein synthesis to a
specific gene or even a specific binding site, (d) parallel
analysis and (e) more sensitive detection with high signal-to-noise ratios and minimized multiple occupancies
(Li et al. 2003; Foquet et al. 2002; Tegenfeldt et al.
2004).
Another use of nanochannels is in the area of drug
delivery. There is presently a need for high precision
nanoengineered devices to yield long term zero-order
release of drugs for therapeutic applications. Previously,
various technologies have been developed to achieve this
goal. However, they have a number of shortcomings
which are related to (a) degradable polymer implants
which have initial burst effects prior to sustained release
of a drug and poor control of release rates of small
molecules, and (b) osmotic pumps which lack the
capability of electronics integration for achieving higher
levels of functionality (Sinha et al. 2004). Nanochannels
fabricated in silicon can allow for the creation of drug
delivery that possesses a combination of structural and
integrated electronic features that may overcome these
challenges.
Nanochannels have been studied recently for use in a
battery which takes advantage of the electrokinetic
phenomena of streaming current and potential with
flowing salt solutions (Daiguji et al. 2004). If the Debye
length (characteristic shielding distance of a point
charge) is on the order of or larger than the height of a
nanofluidic channel with a charged surface, a unipolar
solution of counterions will be generated to maintain
electrical neutrality. Under conditions of a pressure
driven flow, ions will separate giving rise to an electrochemical–mechanical energy conversion or a streaming
current and streaming potential which can be used to
create a battery. This concept was first realized in microfluidic channels (Yang et al. 2003). However, when
employed with nanochannels the efficiency increased by
more than two orders of magnitude. This is because the
streaming current which is generated within the electrical double layer of the ions is confined to within the
Debye length, which characterizes the size of the layer.
For dilute salt solutions of monovalent ions with a
concentration between 0.1 M and 0.1 lM the Debye
length varies from 1 to 1,000 nm, respectively. Therefore, this effect is confined very close to the channel
walls. This underscores why this technology is only
capable of having reasonable efficiencies when employed
with nanochannels.
In a similar vein, nanochannels have been used to
create a nanofluidic transistor based on a metal-oxidesolution (MOSol) system that is similar to a metal-oxidesemiconductor field-effect transistor (MOSFET) (Daiguji
et al. 2005). It has been demonstrated that gate voltages
are able to modulate the concentration of ions and molecules in the channel and control ion conductance. This
again is because the height of the nanochannel is on the
order of or smaller than the Debye length for the ionic
solutions. Therefore, the electric field created by the gate
can penetrate the entire nanofluidic channel to precisely
control ion flow. This technology could have broad
implications on integrated nanofluidic circuits for
manipulation of ions and biomolecules in sub-femtoliter
volumes.
Additional uses of nanochannels have been in scanning nanolithography (Hong et al. 2000), chemical
experiments on a quartz-chip laboratory (Matsumoto
et al. 1998), capillary electrophoresis for chemical and
biochemical analysis (Becker et al. 1998), and use in
chemical sensors (Stern et al. 1997).
With the many possibilities availed to this technology, the objective of this paper is to introduce four
fundamental methods by which nanochannels may be
fabricated. Each method uses standard semiconductor
processing techniques that are very effective, reproducible, have high volume potential, and have decades of
processing technology to facilitate employing them.
These methods are: (1) bulk nanomachining and wafer
bonding, (2) surface nanomachining, (3) buried channel
technology, and (4) nanoimprint lithography (NIL).
2 Challenges and issues
The main underlying issue related to nanochannel fabrication where at least one of the dimensions is less than
several hundred nanometers is related to particle contamination. Fabrication must be done in clean rooms
with much attention given to particle counts in wafer
processing (Kern 1993). Frequent and effective cleaning
chemistries must regularly be employed to render
nanochannels free of particulates which can disrupt flow
and associated experimental data and even bonding
during the fabrication process. This also necessitates
good handling practices not usually found in typical
research environments.
Another issue is with channel collapsing. For bulk
nanomachining and wafer bonding this occurs if proper
channel aspect ratios are not maintained. With surface
nanomachining it is required to carefully consider the
thin film stresses to ensure good channel dimensions and
to prevent their collapse. Moreover, in NIL the mold
lifetime can be an issue since the process is more physical
than chemical in nature. If the mold wears more quickly
there is a greater likelihood of defects to arise in the final
product.
3 Bulk nanomachining and wafer bonding
In bulk nanomachining and wafer bonding, features are
created out of the bulk of a silicon wafer. This can be
done by reactive ion etching (RIE) (Williams and Muller
1996) or by a wet anisotropic etchant with aqueous
KOH or ethylenediamine based solutions (Seidel et al.
1990; Kendall and de Guel 1985). Creation of features
with RIE normally roughens the surface, and the sidewalls of the trench may be tapered. This is especially the
case when the width of the trench is on the same order of
187
magnitude as the depth (Haneveld et al. 2003). Both the
roughness and the shape of the sidewall will have a
major influence on the flow characteristics of the nanochannels. However, if wet anisotropic etching is performed with good crystal alignment, the side walls will
have a mirror like finish and be vertical.
The next step in the process is to bond another wafer
or clear Pyrex cover plate on top of the nanochannels to
allow for fluid visualization. Substrate bonding techniques such as thermal or anodic bonding have been
popular for sealing nanochannels. However, these
techniques are sensitive to particles which can disrupt
bonding. However, if polymer adhesives can be coated
thin enough they too are excellent alternatives for
channel sealing.
Figure 1 depicts this process where wet anisotropic
etching of 1-D nanochannels was performed by
(Haneveld et al. 2003). A (110) silicon wafer is used
which has a thin native oxide. The wafer is then
lithography patterned and the oxide mask is etched with
an HF solution. The silicon is then anisotropically
etched with a developer solution at an elevated temperature which is essentially a water-dilute solution of
tetra methyl ammonium hydroxide (TMAH). Next, the
oxide mask is stripped and bonded to a borofloat glass
wafer. Figure 2 shows a cross section of their structure.
This technique, however, is subject to collapsing of
channels during wafer bonding. This occurrence can be
prevented by understanding that the collapse of the
trenches is a function of wafer thickness, stiffness, surface adhesion energy and of the geometry of the channels. Figure 3 shows a configuration of a trench formed
from a bonded wafer pair. Kim et al. (2003) presented
criteria for channel collapse when two substrates have
the same thickness.
When the channel width, R, is greater than the wafer
thickness, L, (R>2L), trench collapsing occurs for
Fig. 1 A fabrication process for bulk nanomachining with wafer
bonding. This process was used by Haneveld et al. (2003)
Fig. 2 Cross section of a silicon wafer with 50 nm deep channel
bonded to a borofloat wafer (Haneveld et al. 2003)
R
h\ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ;
1:2EL3 =c
ð1Þ
where h is the trench height, c is the surface energy which
typically has a value around 100 mJ/m2 for hydrophilic
surfaces and 20 mJ/m2 for hydrophobic surfaces and, E
is the Young’s modulus. When R<2L which is relevant
for nanochannel fabrication, the trenches between a
wafer pair will collapse if
1=2
Rc
h\2:6
:
ð2Þ
E
Kim et al. calculated the threshold for collapse as a
function of h versus R using Eq. 2 for silicon substrates
(E=165 GPa, c=0.1 J/m2). When channel collapse occurs the two substrates undergo a low temperature
bonding process which essentially ‘‘fills in’’ the nanochannel. This is due to Van der Waal forces which can
have a strong effect because the wafers are in intimate
contact with each other. Equation 2 illustrates that
trench collapse (substrate bonding) is more likely to
occur if (a) the channel has a low aspect ratio (height to
width), (b) the surface energy of attraction between the
two substrates is large and (c) the Young’s modulus is
low so that plastic deformation of the materials can
Fig. 3 Two wafers bonded together forming a trench
188
occur. Figure 4 plots the calculated values of channel
height (2h) versus channel width (2R). Trenches collapse
below the line but are able to survive above it. Table 1
lists some specified values which clearly show that as the
trench width decreased so does the allowable trench
depth. For substrates of different thickness and/or
elastic properties analogous formulae are available from
Cha et al. (1993).
4 Surface nanomachining
Enclosed nanochannels can also be fabricated by surface
nanomachining. It consists of embedding the structures
in a layer of appropriate sacrificial material on the surface of the substrate. The sacrificial material is dissolved
which leaves a complete nanochannel. The dimensions
of these channels are generally restricted by the maximum sacrificial layer thickness that can be deposited
within an acceptable time period (several microns).
Figure 5 is a schematic cross section of an amorphous-Si (a-Si) nanochannel array made by Stern et al.
(1997). First, a thick layer of thermal oxide is grown for
the electrical isolation of electronic devices. Fifty nanometer each of LPCVD tetra ethyl ortho silicate (TEOS)
and LPCVD Si3N4 are then put down to form the lower
channel dielectric layer. The alternating layers of TEOS
and Si3N4 help maintain the channel dimensions because
each imposes an opposite thin film stress. The TEOS
layer is compressive while the nitride layer is tensile.
Typical values for the materials are 150 and
1,000 MPa, respectively. Afterwards a thin a-Si film of
nanometer thickness is grown in an LPCVD furnace.
The a-Si is patterned lithographically to define the
nanochannels. The nitride film below the a-Si is used as
an etch stop during chemical wet etching. The top
channel dielectric layers are then deposited over the
patterned a-Si layer and capped with a thick phosphosilicate glass (PSG) to protect the structure during
channel etching. In addition, to the basic structure,
Fig. 4 Trench collapse threshold values using Eq. 2 for silicon with
c=0.1 J/m2 and E=165 GPa
Table 1 Specified values of allowable trench depth given certain
trench widths for silicon with c=0.1 J/m2 and E=165 GPa
Trench width
(2R) (lm)
Allowable trench
depth (2h) (nm)
0
5
10
30
50
100
200
300
400
500
600
0
2.98
4.22
7.31
9.43
13.34
18.86
23.10
26.68
29.82
32.67
reservoir regions are created at the ends of the channels.
They form a large basin into which the channels open
and a wet etchant selective to a-Si can form conduits.
The removal of the sacrificial layer requires a long
immersion time in a chemical solution such as aqueous
TMAH and special irrigation etching holes may be
required to dissolve the sacrificial layer in a reasonable
time. The experiments of Stern et al. showed that this
method has an upper limit of channel lengths of about
3–5 mm. Moreover, it can take up to 80 h of etching
time for a 2-mm long, 10-lm wide and 50-nm high
channel. This data is shown in Fig. 6. An example of this
type of channel formation is depicted in Fig. 7.
5 Buried channel technology
As an alternative to conventional bulk and surface
nanomachining, a newer approach called buried channel
technology is one of the more elegant methods. Figure 8
shows an example of these channels that can be fabricated using this method. Important features of this
method are large freedom of design and the absence of
assembly of wafer-to-wafer alignment steps because
Fig. 5 Schematic cross-section of an a-Si nanochannel array
fabricated by Stern et al. (1997). Figure is not drawn to scale
189
Fig. 6 Etched channel length versus time for 1-, 5- and 10-lm wide
channels. Heights are 50 nm with etch times up to 80 h. Graph
demonstrates that etch rates decrease with time (Stern et al. 1997)
processing only occurs on one side of the silicon wafer.
Moreover, since the structures are formed below the
surface of the wafer, in principle, the surface is available
for integration of electronic circuits or fluidic devices.
This leads to a more efficient use of the substrate surface
and to further overall device miniaturization. Additionally, by varying the etch processes of the channels
different shapes can be made such as pear-shaped,
circular and v-grove (de Boer et al. 2000).
This technology which is published in more detail by
de Boer et al. (2000) consists of eight basic steps which
are depicted in Fig. 9. To start, a bare substrate is covered with a suitable masking material and lithographically patterned (step 1). An isotropic etchant is then used
to make a rounded out feature (step 2). A trench is
etched in the substrate by deep RIE (step 3) and conformally coated with a material (step 4) to prevent lateral etching of the sidewalls in the sixth step. The coating
is removed only at the bottom of the trench (step 5) and
the structure is etched in the bulk of the substrate again
with an isotropic etchant (step 6). After stripping the
coating (step 7), the structure is closed by filling the
trench with a suitable material (step 8). Moreover,
Table 2 provides two different process schemes for
employing the buried channel technology with wet and
dry etching techniques.
Fig. 7 Picture of surface nanomachined channels. a 0.5 lm wide,
100 nm high and b 1 lm wide, 100 nm high (Stern et al. 1997).
Channels are directly enclosed with silicon nitride and TEOS
Fig. 8 Picture of microchannels formed with buried channel
technology (de Boer et al. 2000). Channels are closed with silicon
nitride which is buried underneath the silicon surface
6 Nanoimprint lithography
Nanoimprint lithography starts with a mold that is
formed usually with interferometric lithography (a lowcost process) in conjunction with anisotropic etching of
Fig. 9 Fabrication sequence for conduit using buried channel
technology
190
Table 2 Two feasible process schemes for buried channel technology
Process step
Scheme 1 (isotropic-dry)
RIE
Scheme 2 (isotropic-wet)
HF/HNO3/H2O
1.
2.
3.
4.
5.
6.
7.
8.
Thermal SiO2
Isotropic SF6 Plasma
DRIE-SF6 plasma
Thermal SiO2
RIE-SF6
Isotropic SF6 Plasma
SiO2 in Buffered HF
LPCVD of poly-Si, SiO2 or Si3N4
LPCVD Si3N4
HF/HNO3/H2O
DRIE-SF6 plasma
LPCVD Si3N4
RIE-SF6
HF/HNO3/H2O
HF (50%)
LPCVD of poly-Si, SiO2 or Si3N4
Initial pattern formation (masking material)
First isotropic etch
Deep reactive ion etch to form trench
Coat trench with protective material
Etch coating at bottom of trench
Second isotropic etch to round out bottom
Strip coating
Close channel by trench filling
the patterned features to form high density arrays of
nanofluidic channels (Zaidi and Brueck 1999; O’Brien
et al. 2003). Once a mold is formed the process of NIL
has two basic steps as shown in Fig. 10. The first step is
the imprint step in which a mold with nanostructures on
its surface is pressed into a thin polymer on a substrate.
This step duplicates the nanostructures on the mold in
the polymer film. The second step is the pattern transfer
where an anisotropic etching process, such as RIE is
used to remove the residual polymer in the compressed
area. This step transfers the thickness contrast pattern
into the entire polymer.
During the imprint step, the polymer is heated to a
temperature above its glass transition temperature. At
this temperature the material will become a viscous
liquid and flow. This allows for it to deform into the
shape of the mold. NIL is a physical process more than a
chemical one. Typically a silicon mold is used in this
process and used in conjunction with poly-methylmethacrylate (PMMA) which is a common polymer
used in NIL. PMMA is favored because it has excellent
properties for imprint lithography with a small thermal
expansion and pressure shrinkage coefficients of
5·105 1/C and of 5.5·108 1/kPa, respectively (Rubin
1973). Chou et al. (1996a) has reported that when using
PMMA which has a glass transition temperature of
105C the imprint temperature used in their experiments
was between 140 and 180C, and the pressure varied
from 4.14 to 6.21 MPa (600–1,900 psi). Additionally,
the imprint process should be done in vacuum to reduce
the formation of air bubbles and mold release agents
used to reduce the polymer adhesion to the mold.
Nanoimprint lithography is a parallel high throughput technique that makes it possible to create nanometric-scale features over a large substrate surface area
at low cost (Chou et al. 1996b). The process is capable of
creating smooth, vertical sidewalls with nearly 90 corners. Cao et al. (2002) have used this technique to create
millions of enclosed nanofluidic channels with dimensions smaller than 10 nm on a 100 mm wafer. In order
to use a larger area it is necessary to have a substrate
with good planarity, and the particle count must be kept
minimal over the entire surface. In addition, if larger
features are desired, optical lithography can be used in
conjunction with interferometric lithography to print
bigger features. This will allow feature sizes to range
from nanometers to millimeters.
The last important aspect necessary to fabricate a
working nanofluidics system is to enclose the channels.
The sealing technique to close up nanochannels is not as
easy as one would first believe. Cao et al. (2002) has used
a shadowing technique by sputtering silicon dioxide over
the nanochannels at a wide distribution of angles. This
leads to a non-uniform deposition that can reduce the
original size of the channels and seal them off on the top
as shown in Fig. 11. Sealed nanochannels using this
process are depicted in Fig. 12.
Fig. 10 Schematic of nanoimprint lithography process: 1 imprinting using a mold to create a thickness contrast in a polymer, 2 mold
removal, and 3 pattern transfer using anisotropic etching to remove
residue polymer in the compressed areas
Fig. 11 A schematic illustration of the sputtering deposition
process that relies on local shadowing of the NIL features to
enclose the channels. a and b represent wide distribution angles
during sputtering
191
Fig. 12 Nanofluidic channels with trench widths of 85 nm were
sealed with SiO2 sputtering. The sealed PMMA channels widths
were reduced to nearly 55 nm after the sealing process (Cao et al.
2002). The scale bar is 500 nm
Guo et al. (2004) has developed a more practical
solution to address this issue of enclosing the nanochannels. The technique is to simply imprint a channel
template into a thin polymer film while on a glass substrate in a single step. Using their technique it is easy to
control the nanochannel dimensions by a simple relationship involving the initial polymer layer thickness and
the mold pattern configuration. The modified NIL
process can be compared and contrasted with the typical
process by looking at Fig. 13. As shown in Fig. 13b, if a
very thin polymer layer is used during imprinting, the
displaced polymer will not be able to completely fill the
trenches on the mold. This results in creating enclosed
nanochannel features. In this process, the mold serves a
channel template, which itself is fabricated by using NIL
and RIE.
This fabrication process can be well controlled to give
predictable channel heights. Figure 14a shows a layout
of a periodic array of channel templates. A simple geometrical argument shows that the height of an enclosed
nanochannel can be determined by the depth of the
etched channel template as well as the initial thickness of
Fig. 14 Nanochannel dimension control by varying initial polymer
layer thickness and mold pattern configuration (Guo et al. 2004)
the polymer layer, which follows a simple linear relationship (Fig. 14b). As shown in Fig. 14b, the height of
the channels can also be controlled by adjusting the ratio
of the ridge width to the trench width on the channel
template.
Figure 14a illustrates the key dimensional parameters
for an arrayed channel template: a trench width, b ridge
width, d trench depth, t initial thickness of the polymer
layer, and h nanochannel height after NIL. Figure 14b
shows the simple relationship of the height of the enclosed nanochannels with the initial polymer thickness
and the mold pattern sizes, h=d(1+b/a)t (obtained by
considering the polymer displacement during the imprint
process, assuming the polymer material to be incompressible). Figure 15 shows the channels made by this
method.
7 Conclusions
The fabrication methods reviewed for making nanochannels for use in single phase liquid flow will aid
researchers to break open this field. Little work has been
published in this area and it is ready to be explored. Since
the manufacturing of nanochannels in principle is no
more difficult that creating microchannels there should
be no major technical hurdles preventing this research to
Fig. 13 Schematics of a the conventional NIL process of using a
mold with surface protrusion patterns to imprint into a polymer
layer and b the nanofluidic channel fabrication by using a template
mold to imprint into a thin polymer layer to leave unfilled and selfenclosed channels
Fig. 15 Pictures of nanoimprinted fluidic channels with template
used to enclose the nanochannels (Guo et al. 2004). Template is
made of a thermal oxide layer on a silicon substrate
192
Table 3 Comparison of nanochannel fabrication methods
Method
Advantages
Disadvantages
Bulk nanomachining
and wafer bonding
Simple concept and processing methodology
Allows for easy fluid visualization when using
an optically clear cover plate or substrate
Possible to achieve stacked structures with
one or more bonded substrates
Trench depth is limited by its width to
prevent trench collapsing. Requires bonding
to realize device (need an additional substrate
to enclose channels). Difficulties with bonding
Surface
nanomachining
Simple concept. Fluid visualization is possible
with transparent surface layers
Buried channel
technology
Large freedom of design
Absence of assembly of wafer-to-wafer
alignment steps or bonding
Surface is available for integration of electronic
circuits or fluidic devices
which leads to more efficient use
of the substrate surface
and to further overall device miniaturization
Channel shapes may be varied
(pear-shaped, circular and v-grove)
Easy to fabricate nanosized channels in 2-D
Long etch times of sacrificial layer. Upper limit
of channel lengths is about 3–5 mm
Need to consider thin film stresses
when fabricating channels
Fluid visualization is not possible
Need to develop processing technology to
exploit ability to build sensors/electronics on
top of nanochannels for overall
device miniaturization
Nanoimprint
lithography
Low-cost process which is capable of high throughput
Mold can easily be adjusted to make large and small
lateral features (nm to mm size)
Easy to fabricate nanosized channels in 2-D
Fluid visualization is possible if mold
is fabricated from glass
Difficulty in accommodating wide ranges
of feature sizes into a single mold
Lifetime of mold may be an issue
Table 4 Demonstrated nanochannels by various investigators
Author
Year
Height
(nm)
Width
Upper channel
length (mm)
Application
Manufacturing
Technique
Stern et al.
Becker et al.
Matsumoto
et al.
Cao et al.
O’Brien et al.
Kim et al.
Haneveld et al.
Sinha et al.
Guo et al.
1997
1998
1998
20–100
1,000
360
0.5–20 lm
50–300 nm
50 nm
5
NA
NA
Chemical sensors
Capillary electrophoresis
Lab on chip
Surface nanomachining
Bulk nanomachining
Bulk nanomachining
2002
2003
2003
2003
2004
2004
‡33
£500
‡30
50–500
60
‡75
‡10 nm
100 nm
2–50 lm
4 lm
4.2 lm
‡120 nm
100
30–40
NA
100
<3 mm
NA
DNA analysis
Fluid mechanics
Basic fluid transfer
Basic fluid transfer
Drug delivery
DNA analysis
Nanoimprint lithography
Nanoimprint lithography
Bulk nanomachining
Bulk nanomachining
Bulk nanomachining
Nanoimprint lithography
go forward. Table 3 summarizes the advantages and
disadvantages of each method. This will allow researchers working in fluid mechanics or heat transfer to determine which method is best suited for their individual
needs and available resources. In addition, Table 4 lists
dimensions of nanochannels fabricated by several authors using various fabrication techniques.
Bulk nanomachining and wafer bonding is by far the
simplest method. It requires the least processing time,
with the lowest capital and operating costs. Many
university cleanrooms should be able to use this process.
Similarly, NIL is a fairly low cost technology. However,
most research cleanrooms are not equipped with an
interferometric lithography system. Conversely, surface
nanomachining takes a much longer time to employ and
will be more costly because it is CVD intensive. As for
buried channel technology it is at the intermediate level
for the number of processing steps and operational costs
but in principle has the largest freedom of design among
all the technologies.
Acknowledgements The authors are thankful for the support of the
Microsystems program and Thermal Analysis and Microfluidics
Laboratory at the Rochester Institute of Technology.
References
Becker H, Lowack K, Manz A (1998) Planar quartz chips with
submicron channels for two-dimensional capillary electrophoresis applications. J Micromech Microeng 8:24–28
de Boer MJ et al (2000) Micromachining of buried micro channels
in silicon. J Microelectromech Syst 9:94–103
Cao H et al (2002) Fabrication of 10 nm enclosed nanofluidic
channels. Appl Phys Lett 81:174–176
193
Cha G, Gafiteanu R, Tong QY, Gösele U (1993) Design considerations for wafer bonding of dissimilar materials. In: Second
international symposium on semiconductor wafer bonding:
science, technology and applications, vol 93–29. Electrochem
Soc Proc, pp 257–266
Chou SY et al (1996a) Nanoimprint lithography. J Vac Sci Technol
B 14:4129–4133
Chou SY et al (1996b) Imprint lithography with 25-nanometer
resolution. Science 272:85–87
Daiguji H et al (2004) Electrochemomechanical energy conversion
in nanofluidic channels. Nano Lett 4:2315–2321
Daiguji H et al (2005) Electrostatic control of ions and molecules in
nanofluidic transistors. Nano Lett 5:943–948
Foquet M et al (2002) DNA fragment sizing by single molecule
detection in submicrometer-sized closed fluidic channels. Anal
Chem 74:1415–1422
Guo LJ, Cheng X, Chou C (2004) Fabrication of size-controllable
nanofluidics channels by nanoimprinting and its applications
for DNA stretching. Nano Lett 4:69–73
Haneveld J, Jansen H, Berenschot E, Tas N, Elwenspoek M (2003)
Wet anisotropic etching for fluidic 1D nanochannels. J Micromech Microeng 13:S62–S66
Hong M, Kim KH, Bae J, Jhe W (2000) Scanning nanolithography
using a material-filled nanopipette. Appl Phys Lett 77:2604–
2606
Kendall DL, de Guel GR (1985) Orientations of the third kind: the
coming of age of (110) silicon. Stud Electr Electron Eng
20:107–124
Kern W (ed) (1993) Handbook of semiconductor wafer cleaning
technology. Noyes Publications, Park Ridge, pp 497–516
Kim WS, Lee J, Ruoff R (2003) Nanofludic channel fabrication
and characterization by micromachining. In: Proceedings of
IMECE’03, Washington D.C., pp 1–6
Li W et al (2003) Sacrificial polymers for nanofluidic channels in
biological applications. Nanotechnology 14:578–583
Matsumoto K et al (1998) Nano-channel on quartz-chip laboratory
using single molecular detectable thermal lens microscope. In:
Proceedings of the IEEE micro electro mechanical systems, pp
127–130
O’Brien MJ II et al (2003) Fabrication of an integrated nanofluidic
chip using interferometric lithography. J Vac Sci Technol B
21:2941–2945
Petersen K et al. (1998) Promise of miniaturized clinical diagnostic
systems. IVD Technol 4:43–49
Rubin I (1973) Injection molding: theory and practice. Wiley, New
York
Seidel H et al (1990) Anisotropic etching of crystalline silicon in
alkaline solutions. J Electrochem Soc 137:3612–3629
Sinha P et al (2004) Nanoengineered device for drug delivery
application. Nanotechnology 15:S585–S589
Stern MB, Geis MW, Curtin JE (1997), Nanochannel fabrication
for chemical sensors. J Vac Sci Technol B 15:2887–2891
Tegenfeldt J et al (2004) Micro- and nanofluidics for DNA analysis.
Anal Bioanal Chem 378:1678–1692
Williams KR, Muller RS (1996) Etch rates for micromachining
processing. J Microelectromech Syst 5:256–269
Yang J et al (2003) Electrokinetic microchannel battery by means
of electrokinetic and microfluidic phenomena. J Micromech
Microeng 13:963–970
Zaidi SH, Brueck SRJ (1999) Interferometric lithography for
nanoscale fabrication. In: Proceedings of SPIE, San Jose, vol
3618, pp 2–8