As microfluidics matures as a technology, scientists are beginning to understand the fundamental principles that govern system behavior.

Microfluidics is the study of interactions between flowing fluids and microscale structures. Though the field of microfluidics is still in its infancy, microfluidic devices are already widely used in aerospace, computer, automotive, and biomedical applications.

At its most basic level, microfluidics requires instrumentation capable of accurately measuring both the structures and the flows. Recent developments in confocal microscopy are playing an important role in both types of measurement. For structural metrology, confocal microscopy permits fast, accurate 3-D modeling of top surface and subsurface features in operating devices. For flow measurement, it promises to refine the capabilities of existing techniques, such as particle image velocimetry, by resolving data in depth as well as laterally.

Active research in microfluidics runs the gamut from fundamental research in micro scale fluid dynamics to engineering development of functional devices. In our laboratory at Purdue Univ., West Lafayette, Ind., we are actively investigating a wide range of topics that include gaseous flows in micro channels, particle image velocimetry, an in silico cell physiology lab based on measurements of calcium ion concentrations and currents around single cells, and a biochip that provides for the interrogation and impedance spectroscopy of biological species.

Structural metrology
Advanced confocal microscopy is an ideal technique for the structural metrology needs of microfluidics applications. The 3Dmap system, from Hyphenated Systems, Burlingame, Calif., can acquire and display high-resolution, 3-D models in seconds and extract critical dimensions under automated or interactive control. Like all confocal microscopes, resolution in the third dimension is provided by the presence of an aperture in the imaging plane that excludes light from regions in the object space above or below a selected plane. Unlike scanning laser confocal systems that must scan the entire plane sequentially, one point at a time, the advanced confocal optics of the 3Dmap incorporate a spinning disk that contains multiple apertures allowing simultaneous parallel data collection from multiple points, dramatically reducing image acquisition time. Sophisticated software manages the acquisition of a sequence of images through adjacent planes and assembles them into a 3-D model that can be rotated and cross sectioned for detailed examination.

Flow measurement
A junction between three microchannels was fabricated using microlithography. The channel is approximately 60 µm wide by 40 µm deep. The image was generated with a Hyphenated Systems advanced white light confocal microscope system. Click to enlarge.
Physical characterization of a microfluidic structure is only useful to the extent that we can determine its effects on the fluid flowing around it. A number of techniques are available. Many use time-based imaging of some visible local marker, which may be a bleach or fluorescent additive, or small particles.

Particle image velocimetry (PIV), a well-developed technique for macro-scale fluid flow visualization, correlates particle movements between pairs of images acquired over a known interval of time. Micro-PIV techniques apply the same algorithms to microscopic images to characterize microfluidic flow. PIV, as typically implemented, is a 2-D technique—it looks down through the full depth of the flow and is not capable of distinguishing the motions of a particle near the bottom of a channel, where flow velocity is reduced by the proximity of the boundary from one near the surface. Confocal PIV, a current research topic in the Purdue laboratory, offers the capability to resolve flow in all three dimensions.

Flow in micro channels is quite different from flow observed at the macroscopic scale. One observation often made is that microscale flow tends to be laminar, without the turbulence and mixing that occurs in larger channels. Another early observation was that flow rates through micro channels exceeds those that would be predicted by conventional fluid dynamic analysis. A closer look at the dynamics of micro flow reveals the reason. Conventional analysis includes a “continuum” assumption: that the number of collisions between fluid molecules is much greater than the number of collisions between fluid molecules and the channel boundaries. Under this assumption, the fluid can be treated as a continuous substance. In order to avoid discontinuities in the velocity profile, molecules in contact with the channel boundary are assumed to be stationary and velocity increases smoothly to its maximum in the center of the channel. However, when the dimensions of the channel approach the mean free path that a molecule travels between collisions in the fluid, the continuum assumption begins to break down as interactions with the channel surface become more significant.

Microfluidic flow often occurs under non-continuum conditions known as slip flow. Here the mean free path of fluid molecules is only about ten times the dimension of the channel and velocity slip exists at the fluid-solid interfaces. A substantial analytical method has been developed for this flow primarily because it is analogous (in terms of ratios of mean free path and characteristic dimensions) to the flows encountered in the aerodynamics of flight structures.

Slip flow analysis accurately accounts for the increased flow observed in micro channels. It also describes more accurately the nonlinear decrease in pressure observed for gas flowing through a uniform micro channel. At the macro scale, the pressure would decrease linearly. We have fabricated a microfluidic device with integrated pressure sensors that allow us to characterize the pressure differential along the channel length. Using this device, we have been able to accurately describe the pressure drop and derive a dimensionless location for the point of maximum deviation from linearity.

Particle image velocimetry
There is great need for flow characterization techniques with high spatial resolution. A number of techniques exist with spatial resolutions between 10 and 100 µm, however, continuing development of advanced microfluidic devices demands resolution at least an order of magnitude better, approaching 1 µm. A number of challenges exist, including the optical resolution of the microscope, which can never improve much beyond the micrometer regime because of the wavelength of visible light. Another challenge is the speed of image acquisition, though significance of this limitation is reduced for low speed flows.

Among the available flow characterization techniques, micro-PIV has the greatest potential to achieve high resolution. We have developed advanced processing algorithms that address issues such as background noise removal, and have demonstrated micro-PIV resolution of approximately 1µm.

However, room exists for further improvement. In particular, the use of smaller seed particles could improve resolution to perhaps 250 nm. Higher resolution still could be achieved by adding a particle tracking step after the correlation-based PIV. A spatial resolution an order of magnitude smaller could then reasonably be reached.

Devices in development
Turning away from the fundamental issues of fluid dynamics and flow measurement, we are also developing microfluidic devices with immediate practical applications. There has been great interest in recent years in the miniaturization of bio sensing systems. These have come to be known as micro total analysis systems (µTAS) or lab-on-a-chip.

One of the most important applications of “biochips” is the detection of small amounts of pathogenic bacteria or toxic substances in food, bodily fluids, tissue samples, soil, etc. The challenge in these applications is the extreme requirement for sensitivity and specificity. In the case of microorganisms, we may also need to know whether they are alive, since dead bacteria may not be a threat. One of the main drawbacks of current detection methods is their incorporation of a growth step, in which the organism is allowed to multiply to ease detection and confirm viability, but which adds days to the detection process. We would like a chip that reduces the detection time to an hour or two and is fully closed, permitting the incorporation of sample preprocessing steps like filtering and chromatography.

Impedimetric detection of biological binding events and amperometric detection of enzymatic reactions are techniques that may meet these requirements. Impedimetric detection works by measuring impedance changes produced by the binding of target molecules to receptors immobilized on the surface of microelectrodes. Amperometric devices measure the current generated by electrochemical reactions at the surface of microelectrodes, typically coated with enzymes.

Fluidic self-assembly (FSA) is an exciting new method for manufacturing microscopic assemblies, including integration of electronic, mechanical, and optical devices on silicon, or of silicon electronic chips onto plastic or other substrates. The process works automatically, using random fluidic transport, allowing the placement of very large numbers of devices in minutes. Particles suspended in a fluid will automatically self-assemble into mating cavities in the substrate. It is noteworthy that the process is in many ways analogous to protein mediated biological processes such as antigen-antibody interactions. The prospect of FSA on a nanometer scale is yet another example of the convergence between nanotechnology and biotechnology. As device dimensions shrink, the biochemical principles of diffusion limited molecular interactions will be as important as the engineering concepts and mechanical assembly.

Recently, industry has exploited FSA to dramatically lower the cost of radio frequency identification (RFID) tags—an emerging technology for automatically managing the flow of products, supplies and equipment. One of the limiting factors for widespread adoption of RFID tags is their cost, which is currently measured at dollars per tag, due in large part to the material costs associated with the size of the device, and manufacturing costs from robotic assembly. FSA can help solve both of the problems by simultaneously eliminating the robotic assembly and allowing precision assembly of much smaller tags.

Looking to the future
Microfluidics is still in the very early stages of its development but its potential value in a wide range of applications is already apparent. As with many new fields of development, scientists are still working on understanding the fundamental principles that govern system behavior, while at the same time, engineers are hard at work producing useful functional devices notwithstanding a lack of fundamental knowledge of exactly how or why they work. We have described several examples of work that we have been involved in that spans the spectrum from the fundamental to the practical. Microfluidics has often been compared to microelectronics in its potential to influence the way we live our lives. If this proves to be true then we are witnessing now the birth of an industry.

—Steven Wereley,
Purdue University
—Edward Robinson and Terence Lundy,
Hyphenated Systems