C&EN London

The notion of putting a conventional, general-purpose chemistry laboratory onto a single microchip is possibly fanciful. But the miniaturization of chemical and physical processes and their integration onto such a chip for a specific application are a definite reality.

The development of microscale devices that can process and analyze minuscule amounts of samples and reagents is exciting the interest of increasing numbers of chemists. According to some, it could revolutionize chemical analysis and synthesis in the same way that microchips have revolutionized computers and electronics.

"The field is one that is growing very rapidly," says George M. Whitesides, a chemistry professor at Harvard University. "The first real applications will very probably be in analytical systems."

Many companies are racing to develop and commercialize this "lab-on-a-chip" technology. Last May, for example, Hewlett-Packard and bioinformation company Caliper Technologies, both in Palo Alto, Calif., announced that they had signed an agreement to jointly develop lab-on-a-chip technology for analytical instrumentation and information systems. The two companies are investing $20 million in the first year of the project and say they plan to invest an additional $80 million over the next four years to further develop and commercialize the technology. They add that the worldwide market for these systems is expected to be $1 billion early in the next century.

"I believe that these microanalytical systems, once they are fully developed, will be much less expensive than big ones," Whitesides notes. "If correct, that fact has significant implications for instrument companies, which will have to go from a business of a few expensive sales to bulk sales of inexpensive systems."

According to Richard D. Kniss, vice president of Hewlett-Packard and general manager of its Chemical Analysis Group, the emerging lab-on-a-chip technology will revolutionize the drug discovery process in the same way that the processes of miniaturization and integration have recast the microelectronics industry.

The pharmaceutical industry is the main driver for developing this technology right now, observes J. Michael Ramsey, a corporate fellow and group leader at Oak Ridge National Laboratory, Oak Ridge, Tenn.

"The primary interest of the industry in this technology is in high-throughput screening of combinatorial libraries," he says, pointing out that throughput is a function of speed and parallelism; that is, carrying out a number of analyses in parallel on the same chip. The prospect of using small quantities of materials in assays is also attractive, Ramsey adds.

Richard A. Mathies, a chemistry professor at the University of California, Berkeley, agrees. "The real power of microfabrication is the ability to produce vast numbers of chemical analysis systems on a single substrate so that you have huge gains in throughput," he says. "This high performance will be particularly valuable for DNA-based diagnostics and genotyping in pharmaceutical and health care applications."

High throughput and the use of very small amounts of materials are perhaps the two major advantages of microscale chemical systems compared with conventional bench-scale systems. But they are not the only ones. The report "Lab-on-a-Chip: The Revolution in Portable Instrumentation" (2nd Ed., 1997) lists low manufacturing, operating, and maintenance costs, and low power consumption, among the advantages. The report was published by Technical Insights, Englewood, N.J., a unit of John Wiley & Sons.

To the list of potential benefits, Ramsey adds automation, reduced waste streams, increased precision and accuracy, and disposability.

In the long term, microfabricated systems could be used not only for combinatorial synthesis, but also for low-inventory process synthesis, environmental maintenance, monitoring sophisticated biological analyses, and other applications, according to Whitesides.

He adds that much of the current research activity in this field is concentrated in five areas: analytical systems for DNA sequencing for the human and other genomic projects, systems for high-throughput drug screening, analytical systems for use in defense against biological and chemical weapons, devices for point-of-care clinical analyses, and microreactor systems that permit the large-scale economic synthesis of toxic compounds such as hydrogen cyanide and phosgene while maintaining only small inventories.

The Technical Insights report also notes that microscale systems offer the potential to integrate chemistry with mechanics, electronics, and optics, and to integrate several analytical systems into very small areas. The report foresees pocket-sized analyzers that can run multiple operations in parallel or in nonlaboratory settings such as the field, factory, hospital, clinic, or home.

Andreas Manz, professor of analytical chemistry at Imperial College, London, prefers not to use the term "lab-on-a-chip." He argues that the concept is surrounded by hype and, furthermore, that the term is confusing and nonscientific.

The term brings to mind "the idea of shrinking everything down to a small scale, but this is only part of the truth," he says. "The main advantage of miniaturization is the dramatically increased performance of the system, regardless of the size of surrounding instrumentation--for example, a laser or mass spectrometer. Only the pathway the chemicals take has to be shrunk. It would definitely be interesting to shrink, for example, existing analytical techniques to a smaller scale, but with chemical and physical processes, scaling laws come into play."

As a result of these scaling laws, some of the characteristics of microsystems are different from those of macrosystems.

Manz points out that microchips are already used commercially, not only for the microelectronic integrated circuits in computers and other electronic devices, but for many mechanical and chemical applications. Examples include the mechanical microchips that trigger the inflation of air bags in automobiles, the mechanical microchips that are used in ink-jet print heads, and microchip-based chemical sensing devices used, for example, in portable blood analyzers.

"There's a fair amount of misunderstanding about the term lab-on-a-chip," Ramsey comments. "Some people call conventional chemical sensing lab-on-a-chip. To my mind, a lab-on-a-chip is a device where materials are moved around either to mix them together for chemical reactions or to deliver them to analysis functionalities to generate information."

D. Jed Harrison, a chemistry professor at the University of Alberta, Edmonton, has a slightly different definition. "To me a lab-on-a-chip is a planar device on which or in which a number of chemical processes are being performed in order to go from reactants to products or from sample to analysis."

Devices on which or in which fluids flow along microchannels are known as microfluidic devices.

Both Harrison and Ramsey stress the distinction between microfluidic devices and microarray devices used for DNA analysis. "DNA arrays are a separate element that have largely been coined 'biochips.' I would not refer to these as lab-on-a-chip systems although they certainly could be included as a functional element in a lab-on-a-chip device," Ramsey says.

DNA microchip arrays contain from several hundred to several hundred thousand immobilized DNA fragments. They provide a systematic way to survey DNA and RNA variation and may well provide a standard tool for molecular biology research and clinical diagnostics [Nat. Genet., 21, 3 (1999) and Science, 282, 396 (1998)]. Such chips are produced commercially by companies such as Affymetrix of Santa Clara, Calif.

According to Manz, there are no chip-based microfluidic chemical analysis instruments commercially available at present. And no matter how small in size and how highly efficient such devices are, they will still have to interface with the rest of the world, he says.

"The rest of the world will be a normal lab as we know it today," Manz continues. "If you do a job with a mass spectrometer today, this does not necessarily mean that your mass spectrometer will need to be the size of a fingernail tomorrow in order to interface with a microchip."

Harrison agrees. "Shrinking the device on which all the chemical processing is done doesn't mean that all the peripherals needed to run the system need to be small," he says. "If, for example, we need a syringe pump, we use a conventional syringe pump right now. And if we need, say, fluorescence detection, then we use conventional optics, which are then directed to make the measurement on the chip. The peripherals sit off-chip and are typically benchtop-instrument sized."

At present, the typical microchip used for miniaturized chemical systems consists of a 2- or 3-cm square sliver of silicon, glass, quartz, or plastic that is etched or molded with chambers and channels with cross-sections as low as 50 m. The chip is covered with a plate to contain the samples and reagents.

Each material has its own advantages and disadvantages. Quartz is suitable for electrophoresis because it is a good electrical insulator and it is transparent to the UV light required for absorbance and fluorescence detection of the separated products. Polymers have the potential of being mass produced inexpensively. Silicon benefits from the large variety of microfabrication techniques already available. It is a semiconductor, however, so the application of high voltages to silicon microchips is virtually impossible.

"In terms of size, the chips must be much bigger than microelectronic chips because molecules need more space than electrons," Manz points out. "Going below a millimeter square would not make much sense at this stage."

At present, it is possible to inject volumes as low as 1 picoliter onto chips with miniaturized chemical systems. But in electrophoresis applications, for example, such volumes require separation channels of between 10 and 50 mm in length to show the full efficiency of the system, according to Manz.

For synthesis applications, one of the critical factors is the cross-section of the channels, because this determines the production output per unit time. For analysis, detection limits determine the smallest amount of material that can be used.

"The key concept is the use of miniaturization to optimize mass transport, optimize thermal transport, and possibly to optimize the integration density of the required chemical and physical processes," Manz says.

Manz points out that silicon and glass devices usually are fabricated by using standard photolithographic methods developed for microelectronics. Polymer-based chips can be patterned by using techniques such as hot embossing, injection molding, and laser ablation.

Using such fabrication techniques, it is now possible to make miniaturized components--such as chromatography and electrophoresis separation columns, polymerase chain reaction vessels, pumps, and valves--for use on centimeter-sized microchips.

"This field started by extending the technologies used for silicon manufacturing to the fabrication of small analytical systems," Whitesides observes. "As the field has developed, this technology is being rapidly abandoned for technologies that are based in plastics and molding. These are much less expensive, easier to work with, more rugged, and have obvious advantages such as transparency. Glass may remain the material of choice for some applications--for example, analyses in organic solvents or of low molecular weight organic materials--but for biologicals, I believe that the technology will end in using almost entirely organic polymers."

In microfluidic devices, fluids normally flow through channels without turbulence. As a result, layers of fluids containing different components are able to flow along together and mix by diffusion only. The mixing resulting from this laminar flow is very rapid. It is one of the main advantages of microfluidic systems.

"Mass transport by diffusion is 100 times faster when a system is 10-fold smaller," Manz says. "So in chemical separations such as electrophoresis, the timescale of separation can also be performed 100 times faster. You get the same type of relationship with heat transport."

One of the key parameters that determines these properties in a fluid system is the Reynolds number. This is a dimensionless number calculated by multiplying the fluid density, the fluid velocity, and the internal diameter of the pipe or tube, and then dividing by the fluid viscosity.

"Reynolds numbers are system-specific," explains Andrew de Mello, Zeneca Lecturer of Analytical Sciences at Imperial College. "Generally, if you have a Reynolds number of more than 2,000, there is turbulence. Below this number, you just have laminar flow and this means reactions are mediated simply by molecular diffusion."

According to Whitesides, this aspect of microfluidics provides wonderful opportunities for basic science. "Most fluid flow in small channels occurs at low Reynolds number, so there is no turbulent mixing," he explains. "Fluid behavior at low Reynolds number is not intuitive to many chemists, and it offers the opportunity to carry out experiments that cannot be carried out at high Reynolds number and therefore to find new phenomena and, one hopes, new types of applications.

"So one should not think of microfluidic systems as ones in which there is a compromise between small size and high performance," he continues. "Rather, one is getting small size and high performance."

For its operation, a microfluidic chemical device requires a number of microcomponents, such as microfilters, microreactors, and microseparation columns, depending on the application. It also needs some means of driving the fluids around the chip.

"You can use external pumps, such as syringe pumps, or microfabricated pumps to provide pressure-driven flows," Harrison explains. "However, micropumps tend to suffer from relatively poor valves so they have poor back pressures and a reasonable amount of leakage."

The development of microfabricated pumps has struggled in the 1990s, he suggests, although in the past year or so there has been some success in the development of plastic micropumps with low dead volumes. Dead volumes are a kind of stagnant backwater that can occur outside of the mainstream of a fluid system.

There has been more progress with electrokinetic devices, which rely on electroosmotic flow to provide the pumping and valving action required to move bulk fluid through microchannels. Electroosmosis is a macroscopic phenomenon that results from an electrical double layer formed by ions in the fluid and by surface electrical charges immobilized on the capillary walls. When an electric field is applied, the bulk solution moves toward one of the electrodes of the device.

"Electrodes sit in the reservoirs that connect to the ends of the various channels," Ramsey explains. "Electrode potentials are applied to the various reservoirs in a time-dependent fashion to move the fluid in the direction you desire it to go."

According to Manz, electroosmotic flow has several advantages over pressure-driven flow for miniaturized analysis systems. For example, separation efficiencies of the systems are improved. And multiple channels on a chip can be readily controlled with a few electrodes, which means no mechanically moving parts are required.

The electrodes used for electrokinetic devices are microfabricated on the chip. "They are connected by platinum, chromium, or gold wires deposited onto the microchip substrate," de Mello explains. "This type of microchip is a hybrid of electronic circuitry and chemical circuitry. In a simple application you might only need a small voltage, so you can imagine using a small battery as the power source."

In another approach, chemical engineering professor Nicholas L. Abbot and coworkers at the University of California, Davis, have demonstrated that electroactive surfactants can be used to move and position aqueous and organic liquids on surfaces on submillimeter scales [Science, 283, 57 (1999)].

"These surfactants are water-soluble molecules that can be switched electrochemically between surface-active states-- that is, states in which the molecules adsorb to surfaces--and surface-inactive states," he explains.

Switching between "on" and "off" at rates set by the magnitude of the electrical potential applied to an electrode creates mechanical stresses that act on the surfaces of the liquids. The liquids can therefore be set into controlled motion and directed to specified regions of a surface and mixed within small volumes.

Abbot suggests that the capability of manipulating liquids in this way has the potential to impact lab-on-a-chip technologies. "The majority of schemes for pumping liquids within fluidic networks rely on electrokinetic phenomena that require potentials of kilovolts or more delivered from a substantial transformer," he says. "Our approach based on electroactive surfactants, in contrast, requires a few 10ths of a volt to move and position liquids."

Abbott points out that if a lab-on-a-chip device is to be part of a portable analytical instrument, then the capability to drive fluid motion from a standard off-the-supermarket-shelf battery will be useful.

"Our results demonstrate principles for containerless lab-on-a-chip technologies," he continues. "The capability to vectorially transport materials across an unconfined surface suggests approaches to the fabrication of transport systems for lab-on-a-chip technologies that are flexible, easily reconfigured, and yet do not require networks of tubes and valves."

Liquid microchannels can also be created by patterning a surface with hydrophobic and hydrophilic regions. A group at the Max Planck Institute of Colloids & Interfaces in Berlin-Adlershof and in Teltow-Seehof, Germany, has shown that above a certain volume, these liquid microchannels unexpectedly develop a single bulge rather than forming a row of single droplets [Science,283, 46 (1999)].

"We have discovered a new wetting phenomenon at surfaces that are laterally structured on the micrometer scale," says group member Peter Lenz, a postdoctoral researcher at the institute. "If the hydrophilic surface domains have the shape of long stripes, the liquids form channels along these stripes. As one increases the volume of the liquid, these channels undergo a shape instability from a homogeneous channel state to another state with a single bulge. This instability is a new general phenomenon for a liquid confined to a hydrophilic region in a hydrophobic surrounding."

This new type of instability could be applied to the design of microreactors, Lenz tells C&EN. "Our results show how the position of the contact line of the liquid channels can be controlled by the structure of the substrate and by the amount of adsorbed volume. Thus, the bulges of two neighboring channels can be localized close to each other."

Sidebar: British lab-on-a-chip consortium established

This volume-induced coalescence of bulges can be used to build microbridges between neighboring microchannels, Lenz suggests. If the channels are filled with different reactants, such bridges lead to a well-mixed state without any stirring, simply by volume control.

"More generally, this new nonmechanical fluid control could be a first step toward the design of low-cost microfluidic devices for microchemical analysis of complex fluids or devices for sorting proteins and cells," he concludes.

Whitesides and colleagues at Harvard are developing techniques based on polymers and lithography to make prototype microfluidic systems very rapidly. "These are intended initially to accelerate research in this area by shortening the time between the idea and the experimental device," he tells C&EN.

The group has recently reported a method for creating microscopic channels in an elastomeric material, poly(dimethylsiloxane) or PDMS, that makes it possible to rapidly carry out a complete cycle of design, fabrication, and testing of microfluidic systems [ Anal. Chem., 70, 4974 (1998)]. The method relies on the oxidation of PDMS in oxygen plasma. Oxidized PDMS seals irreversibly to other materials used in microfluidic systems, such as glass, silicon, silicon oxide, and oxidized polystyrene. A number of substrates for devices are, therefore, practical options with this method, the group suggests.

Whitesides' team is also developing fabrication strategies for making three-dimensional and complex microchannel systems to serve as the basis for new types of experiments. "There are currently no general strategies for making complex 3-D systems, and the methods that we have developed will open this part of this field for study," he says.

A total-analysis system, Manz explains, periodically transforms chemical information into an electronic or optical signal. A typical system, for example, converts analyte concentration to fluorescence intensity and then to voltage in a photocell. With conventional laboratory instruments, such systems, although reliable, are bulky.

A miniaturized total-analysis system carries out such an analysis on a micrometer scale using centimeter-sized glass, silicon, or plastic chips. These systems are also known as micro-total-analysis systems, or TAS for short.

According to Albert van den Berg, research coordinator of TAS at MESA Research Institute at the University of Twente, in Enschede, the Netherlands, TAS research is principally directed toward the control of microfluidics and nanofluidics, the development of high-throughput screening systems, and the incorporation of microreactors into analysis systems. "Major application areas are clinical diagnostics, DNA analysis, and drug discovery and screening," van den Berg says. "These all have enormous market potential.

"For many applications, people now are looking at using plastic materials," he continues. "However, we have demonstrated that very special microchannel structures can be fabricated using silicon. In our view, there is still a large variety of new microfluidic phenomena to be discovered and investigated using such microdevices."

De Mello at Imperial College points out that the TAS approach relies on the design of microfluidic circuitry for a number of analytical steps on a single microchip. The steps include sample introduction, sample pretreatment, mixing, chemical reactions, product separation, detection, and product isolation.

The development of microcomponents that can carry out one or more of these steps is one of the major aims of current research on such systems. At Imperial College, for example, Manz's team has recently demonstrated the use of a microreactor for precolumn derivatization of samples to be separated by high-performance liquid chromatography (HPLC). The team has used the device for preparing fluorescent derivatives of amino acids prior to HPLC separation and fluorescence detection [J. Chromatogr. A, 815, 265 (1998)]

Pre- or postcolumn derivatization is generally carried out using a heated reactor that requires a thermostatically controlled reservoir, typically a large and slowly reacting water bath, the team notes. To serve the same function, the group uses a micromachined, heated chemical reactor consisting of a sandwich of glass and silicon chips with a "meandering" reaction channel. The microreactor has on-chip heating resistors that provide rapid heating and also on-chip resistors that enable feedback control of the temperature.

"The device lends itself to future integration with suitable components for chip-based fluid handling, separation, and detection in a micro-total-analysis system," the group states.

Manz, de Mello, and coworker Martin U. Kopp at Imperial College also have developed a microfabricated chemical amplifier for carrying out the polymerase chain reaction (PCR) in a continuous flow at high speed (C&EN, May 18, 1998, page 35). The device is a micromachined glass chip with a single capillary channel that repeatedly passes through three well-defined temperature zones. PCR doubles the concentration of a specific DNA fragment in a sample in each cycle through the three zones.

"Analytical performance in terms of separations and reactions is better on a smaller scale in terms of speed and efficiency," de Mello says. "Once you've integrated the analytical steps onto a chip, you can automate the whole analysis. There is no manual sample transfer and very little human input, and so the analysis is more reliable and easy to use."

But microfluidic applications also have real-world problems. "If you take a sample from a crime scene, a blood sample for example, the sample is likely to have particulates such as dust in it," de Mello says. "Because we are working with very small dimensions, dust particles can block the chip. So the development of microfiltration devices that can be integrated on a chip is very important if these devices are to be used in the real world."

Sidebar: Diagnostic dielectrophoresis-on-a-chip

A TAS component that addresses such complications has been developed by Micronics in Redmond, Wash. Known as a T-Sensor, it combines separation and detection functions [Science, 283, 346 (1999)].

"The sensor allows, for example, the detection of diffusing analytes directly in particle-laden samples such as whole blood," says Bernhard H. Weigl, manager of business development and senior scientist at Micronics. He developed the device with Paul Yager, a professor of bioengineering at University of Washington, Seattle.

Weigl observes that many microfluidic technologies work very well for highly predictable and homogeneous samples common in genetic testing and drug discovery processes. "One of the biggest challenges for current labs-on-a-chip, however, is to perform analysis in the presence of the complexity and heterogeneity of actual samples such as whole blood or contaminated environmental samples," he says.

"Separation of particles from soluble components is usually done by filtering or centrifugation," he adds. "None of these methods are suitable for integrated microsystems."

The Micronics T-Sensor is a microfluidics system that has three separate inlets and channels for the sample, a detection solution, and a reference solution. The three inlet streams then interdiffuse during parallel flow in interaction zones. Species such as hydrogen ions, sodium ions, and small molecules can diffuse rapidly between the streams, whereas large molecules diffuse more slowly and equilibrate further along the device's detection channel.

The analytes are usually detected through a change in indicator absorption or fluorescence intensity in the diffusion zone between the streams. Single-frame video or charge-coupled device images of the detection channel for each sample are digitized and processed to provide information about the analyte concentration.

"The reference solution provides real- time calibration and control," Weigl explains. "In addition to separation and analyte detection in complex sample solutions, T-Sensors also enable novel immunoassay and kinetic assay formats re-quiring only minute amounts of sample."

To date, T-Sensor assay feasibility has been demonstrated for a variety of clinical parameters such as blood pH and oxygen, electrolytes, proteins, enzymes, and drugs through the use of detection methods ranging from fluorescence and light absorption to voltammetry, Weigl and Yager state.

In as-yet-unpublished work, Manz and Imperial College coworker H. John Crabtree have developed what they believe to be a revolutionary method for improving the detectability of analytes separated by chromatography or electrophoresis.

"Typically in chromatography and electrophoresis you inject a sample plug, that is a small amount of sample with a well-defined volume, and you then expose it to either an extraction, which is chromatography, or to an electric field, which is electrophoresis," Manz explains. "You have a detector at the end which gives you a trace on a time scale. The trace of the compound that comes off is roughly a speed measurement of the species.

"Our method uses slit arrays covering the length of the separation capillary," he notes. "A sample is thus seen multiple times and the recorded detector signal is a convolution of the slit function with the electropherogram."

Fourier transformation is used to convert these time-domain analytical outputs into frequency-domain plots, which record the constituent analytes in terms of their "blinking" frequency.

"With multiple detection at regularly spaced intervals, we can sum the signals for a single component moving along the column," Manz continues. "The method is not sensitive to injection patterns or volumes and the signal-to-noise ratio of the result is altogether better."

The group has tested the method using a micromachined glass structure to perform capillary electrophoresis separations with laser-induced fluorescence detection on two-component fluorescent samples.

The researchers state that the method can easily be integrated into a miniaturized total-analysis system and, in principle, is equally applicable to detection in full-sized analytical instrumentation.

Much current research on TAS focuses on development of microfluidic devices that can be used for high-resolution, high-throughput electrophoresis separations.

"Electrophoresis is a standard way of separating molecules on the basis of their charge-to-mass ratios," de Mello explains. "An electric potential is applied across a column containing a sample in a fluid medium. Positive molecules migrate to the cathode and negatives ones to the anode at different speeds depending on the charge-to-mass ratios."

The use of micromachining technology to prepare chemical-analysis systems on glass chips that rely on electroosmotic pumping to drive fluid flow and electrophoretic separation to distinguish sample components was first demonstrated by Harrison, Manz, and coworkers in 1993 [Science, 261, 895 (1993)].

Since then, the speed and sensitivity of electrophoresis on microplates has increased dramatically. In recent work, for example, Ramsey and coworkers at Oak Ridge have developed microchip devices that can separate and count single chromophore molecules [ Anal. Chem., 70, 431 (1998)]. In this experiment, molecules of two fluorescent rhodamine dyes are separated in micromachined electrophoresis channels and detected by counting fluorescence bursts from individual molecules.

"These are the lowest detection limits reported for microchip separation devices and the first example of single-chromophore molecular counting for detection of chemical separations," the team states.

Ramsey's team also has designed microchips that can carry out electrophoretic separations in less than a millisecond [ Anal. Chem., 70, 3476 (1998)].

Integration of the various components of electrophoresis microsytems is one of the key themes of current research in this field. Last year, for example, Harrison and graduate student Nghia H. Chiem at the University of Alberta reported a microchip system that integrates an immunoreactor with capillary electrophoretic separation for the competitive immunoassay of the drug theophylline in human serum [Clin. Chem., 44, 591 (1998)].

A diluted serum sample containing the drug is mixed on-chip with a fluorescein-labeled theophylline tracer and an antitheophylline antibody. The free theophylline competes with tracer-bound theophylline for a limited amount of the antibody.

Electroosmotic pumping is used to control the mixing and reactions of the reagents and serum sample. After reaction, the solution is injected into an electrophoresis separation channel integrated within the same chip. The separated free and bound tracers are measured by fluorescence detection, yielding data from which the concentration of the drug in the serum can be calculated.

"This demonstrates the complete micro-total-analysis system," Harrison says. "We've taken a serum sample and we've got a number out at the other end in terms of the drug content in a serum sample."

Harrison and graduate student Shakuntala D. Mangru at the University of Alberta also have carried out immunoassays using chemiluminescence detection in a postseparation reactor integrated on a microchip-based capillary electrophoresis device [Electrophoresis, 19, 2301 (1998)].

Ramsey's group at Oak Ridge has described a microchip design and procedure for the complete analysis of as many as four DNA samples on a single microchip [ Anal. Chem., 70, 5172 (1998)]. The chip was tested on fragments of Escherichia coli genomic and plasmid DNAs. Thermal lysis of the bacteria's cells were integrated into a PCR cycle for amplifying the fragments. The PCR products were marked with a fluorescent dye and analyzed on the same device by microchip gel electrophoresis using fluorescence detection.

The researchers state that the microchip offers a rapid and potentially inexpensive means of manipulating and analyzing DNA in a way that minimizes sample handling and the potential for contamination. They suggest that numerous applications of this microchip design can be envisioned; for example, genetic mapping, forensic analysis, and clinical diagnoses of inheritable diseases.

According to Mathies, his group at UC Berkeley has shown that virtually all types of electrophoretic genetic analy-ses can be successfully performed on microfabricated capillary electrophoresis systems.

"We demonstrated the first fragment sizing and PCR fragment separations in 1994, the first sequencing separations in 1995, and the best sequencing separations on chips in 1998," Mathies tells C&EN. "These separations extend to 600 bases in only 20 minutes, demonstrating that microfabricated systems can be used for genomic sequencing."

His group has recently designed, built, and tested microfabricated capillary array electrophoresis systems that can perform up to 96 parallel genetic analyses on a single 10-cm-diameter circular wafer.

"Each chip is a sandwich of two glass plates," Mathies says. "One of the plates is chemically etched through a mask. The other plate is thermally bonded to the etched plate to form the channels. The microplates are reusable, although it is feasible to make these disposable devices from plastic. We use either hydroxyethyl cellulose or linear polyacrylamide as the separation matrix."

A sample throughput for sizing DNA fragments as fast as one sample per second is possible on these microplates, notes Mathies. "With appropriate color and size multiplexing, this should rise to 10 samples per second," he says. "This throughput is 100 times that of typical slab gel electrophoresis systems."

The Berkeley group has developed a rotary confocal fluorescence detection system to scan the chips. The system interrogates the chips from below with a rotating objective without reducing separation performance.

Mathies' group is also working on the integration of separation and other detection methods on a single chip. "We were the first to fabricate and utilize integrated electrochemical detectors to analyze on-chip electrophoretic analyses," Mathies says.

"Electrochemical detection is very sensitive, as good as fluorescence, and it can be used to detect all types of analytes including DNA," he continues. "These integrated electrochemical detectors have allowed us to make the first fully miniaturized micro-total-analysis systems. These detectors have also enabled us to make the first fully portable DNA analysis systems."

These capillary electrophoresis systems are microfabricated on glass substrates with photolithographic placement of working electrodes just outside the exit of the electrophoresis channels [ Anal. Chem., 70, 684 (1998)]. The chips provide high-sensitivity electrochemical detection of neurotransmitters, for example, with minimal interference from the separation electric field, the UC Berkeley team states.

The researchers suggest that by improving layout and packaging of such systems it may even be feasible to make chemical analysis microprocessors where the detection and computer circuitry are also integrated on-chip. "With these improvements, integrated microanalysis devices should be capable of probing for biological signatures in a variety of challenging locations, including even extraterrestrial environments," the Berkeley group predicts.

Mathies points out that capillary array electrophoresis microplates have evolved rapidly over recent years. It is now possible to analyze 96 DNA samples in parallel in as little as 30 seconds, whereas in 1994 it took 120 seconds to analyze just one DNA sample. There has been approximately a 10-fold increase in both the number of samples per microplate and the throughput per microplate every two years. "It will be interesting to see what magnitude of performance increase will be achieved in the next two years," he says.

Chemical & Engineering News
Copyright © 1999 American Chemical Society