Collins

April 1, 1998

19 Min Read
Packaging Technology for Miniature IVD Instrumentation

Medical Device & Diagnostic Industry Magazine
MDDI Article Index

An MD&DI April 1998 Column

IN VITRO DIAGNOSTICS

MEMS technologies are poised to take future analytical systems—and especially IVD systems—to new heights.

To reduce the cost of in vitro diagnostics (IVDs), medical device designers are searching for new technologies that will enable them to develop high-quality instruments at a fraction of the cost of current laboratory systems. For many IVD manufacturers the best hope comes from recent successes with miniaturized devices, which enable clinicians to perform sophisticated diagnostic techniques in the field or at a patient's bedside, thereby removing routine testing from hospital settings and further reducing the costs of diagnostic testing.

The technologies associated with microelectromechanical systems (MEMS) are making a major contribution to the miniaturization of diagnostic instrumentation. MEMS technologies are already being applied for performing immunological analysis, polymerase chain reaction (PCR), in vitro fertilization, glucose sensing, and semen analysis. Continued development of MEMS technologies promises to produce inexpensive, self-contained microfluidic devices with performance and accuracy superior to their larger and more-expensive laboratory counterparts. In time, these instruments may replace many other time-consuming and less-sensitive laboratory instruments used to identify chemical compounds, biological species, and pathogens in forensic, environmental, clinical, and industrial samples.1,2 By 2000, it is estimated that MEMS will become a $14-billion industry, in part because of its expected impact on analytical instrumentation.

THE CHALLENGE OF MEMS PACKAGING

MEMS are produced using the microfabrication techniques pioneered by the integrated circuit (IC) manufacturing industry. These fabrication processes enable manufacturers to micromachine miniature three-dimensional devices that incorporate a variety of electrical, mechanical, optical, thermal, magnetic, or chemical components.3,4 Such components, or modules, include pressure, chemical, flow, temperature, and humidity sensors; microfluidic pumps, valves, spray nozzles, and ink-jet heads; optical switches, couplers, and micropositioners; accelerometers; and miniature bioreaction chambers. With such a range of possibilities, the components necessary to assemble a miniature monolithic analytical system already exist as separate entities.

Despite recent progress in the MEMS field, however, full commercialization has so far been achieved for only a small number of applications (e.g., pressure sensors and accelerometers).5 The primary weakness that has prevented more MEMS products from reaching the marketplace is the difficulty of packaging or assembling such devices.6 Unlike modern integrated circuitry, in which off-the-shelf components are conveniently packaged for use on a variety of platforms, MEMS packaging is commonly application-specific. In fact, packaging-related expenses can account for up to 80% of the cost of a MEMS device.

Before MEMS modules can be routinely assembled into functional microsystems, it will be necessary to develop a technology for interconnecting and interfacing them—that is, a MEMS packaging technology. This is a challenging issue, because it entails the assembly of two or more individual MEMS modules (e.g., microfluidic pumps and valves, miniature reaction chambers, optical sources and detectors, CMOS control circuitry) of different domains (e.g., for handling electrical, optical, or fluidic signals), all in a package that fits in the palm of a hand. Creation of such devices will push MEMS technologies to their limits but will also demonstrate their potential for creating complex miniature instruments that can perform a variety of tasks better, faster, and more economically than the laboratory-scale instruments they emulate.

Two opposing approaches have been proposed for the creation of such monolithic microinstrument systems: integration and hybrid assembly. An integrated approach would attempt to fabricate all the necessary microcomponents on a single substrate. It is thought that total integration of the microfabrication processes would be the most efficient method of producing a MEMS instrument, and that it would result in systems with better performance. However, the technology for integrating electrical, mechanical, fluidic, and optical components on the same substrate does not yet exist. Differing types of modules require different substrates, and the processes needed to fabricate them are not compatible with one another. Even if it were possible to integrate the manufacturing processes, the likelihood is that doing so would compromise the integrity of one or more modules, thereby lowering system performance. In short, in the current state of MEMS technology, total integration is unrealistic.

At present, a hybrid approach offers the most practical means of producing self-contained miniature MEMS instruments for diagnostic applications. Using a hybrid approach to the challenge of system integration, each module is microfabricated separately by means of proven techniques and processes best suited for its operation, then assembled and packaged into a completed microsystem. Like the integrated circuits that form the building blocks of modern computers, off-the-shelf MEMS components could be assembled to create novel diagnostic instruments.

INTERCONNECT DESIGN

The ultimate goal of current research in IVD MEMS is to miniaturize entire instruments and reduce the amounts of sample and reagent fluids they handle. But a more readily achievable goal would be to integrate existing MEMS components into larger, more traditional instruments. Pursuing this strategy could provide added sensitivity, faster operation, more-precise control, and fault detection to existing devices. Moreover, for companies involved in the design of miniaturized devices, this strategy could open more market opportunities at an earlier stage of development than if they waited for a fully integrated product. In either case, it will be necessary to develop a reliable means of interconnecting miniature components to one another or to the macrocomponents of a large-scale system.

Even for designers of traditional, macro-sized fluid-handling equipment, the development of fluid interconnects is among the most difficult of design challenges. Moving fluid from one device to another can be as simple as connecting two pieces of tubing with an injection-molded barbed fitting or as complicated as using a robotic pipettor to aspirate fluid from one location and dispense it to another. For their devices to gain acceptance in the market, manufacturers of miniaturized components and devices must offer simple, robust, and high-performance interconnects for their products.

For large-scale fluid-handling systems that use miniaturized components, the design of interconnects is complicated by two factors. First, the interconnects must bridge large transitions in scale, and, second, as dimensions shrink, surface effects become dominant forces. Capillary forces can match or exceed the pressures that can be generated by pumps, and small imperfections in a fluid connection can result in leaks that are quite large in relation to the amount of fluid handled by a miniature device.

Evaluation Criteria. Four factors influence the design of interconnects in a fluid-handling instrument:

  • Intended application of the instrument.

  • Instrument format (e.g., robotic versus fluidic).

  • Types of fluid manipulations to be performed.

  • Performance criteria.

The ideal interconnect design is one that has the least possible effect on fluid flow. Table I lists five criteria that can be used to evaluate the performance of a fluid interconnect. These criteria should be evaluated in relation to the amount of fluid that the device handles. Thus, a nanoliter of dead volume is insignificant in devices handling milliliters of fluid but can be catastrophic in a miniaturized device handling only nanoliters of fluid. Similarly, it is difficult to avoid the need for cross section changes for the interconnects of miniature components having channel dimensions on the order of 100 µm.

Evaluation criterion

Reasons for importance

Dead volume

Influences carryover and metering precision

Cross section change

Influences degassing due to sudden pressure drops and carryover

Leak rate/maximum pressure

Loss of fluid and entrance for bubbles

Material compatibility

Influences reliability and carryover

Reversibility

Influences cost of servicing and suitability forconnecting disposables



Table I. Interconnect performance criteria.

Analytical microsystems challenge the performance of fluid interconnects more severely than most applications. The high pressures, extreme temperature range, and wide variety of fluids being handled all act to accentuate any small imperfections in a fluid connection. Nevertheless, there are many companies pursuing this market. The subsections that follow describe some of the approaches currently being used to create interconnects for macro-, mini-, and microinstrumentation.

Macrointerconnect Technologies. Macro-scale interconnect strategies fall into two categories depending on whether the system is robotic or fluidic. Robotic systems move fluid between elements by aspirating fluid into a probe with the use of some type of pump, moving the probe to another location, and then dispensing the fluid into the target element. Fluidic systems are interconnected with the use of gaskets and mechanical fasteners. In the case of a barbed fitting, the gasket is the elastic tubing that seals around the fitting, and mechanical fastening is accomplished by frictional forces against the barbs.

Figure 1. A fluidic breadboard fabricated using 4-in. silicon/glass wafer pairs. Inset shows schematic of the interlocking finger joints and the rubber gasket held under compression by the two halves of the breadboard.

Figure 1 shows the structure of a connector offered by Lee Co. (Westbrook, CT), in which fluid sealing is accomplished through the use of a ferrule and tapered features in both the tubing and the connector boss, and mechanical fastening is accomplished by a threaded fitting. This connector and others like it have the benefit of low dead volume, minimal cross section changes, all inert exposed materials, and reasonably high pressure ratings. Because they are threaded connectors, they are also reusable.

Mini-interconnect Technologies. At Abbott Laboratories (Abbott Park, IL), fluid circuit technology was developed around three basic elements, all manufactured by a common process: channels formed in acrylic, pneumatically actuated membrane valves, and optical fluid detectors.7 Combinations of these three elements are capable of performing all basic fluid-handling operations. Fluid circuit technology is a fluidics-based technology that is best suited for handling volumes of fluid between 1 µl and 1 ml.

Since Abbott's fluid circuit technology is an integrated fluidics technology, many of the problems related to interconnections were eliminated. Because it was impractical and expensive to fabricate every fluid-handling instrument as a single piece, however, the company needed a good method for connecting fluid circuits together. Similarly, the company needed a method to connect fluid circuit tubing to non—fluid circuit components such as syringe pumps and reagent bottles. It was determined that these needs could be met by using O-ring seals between elements held together by bolts, and tubing adapters made by Lee Co. These interconnect technologies provided low dead volume, few cross section changes, and sufficient high-pressure capability for use in the fluid circuit— based instruments. To reduce cost and increase manufacturability, the company also designed and tested tubing adapters machined directly into the acrylic fluid circuit.

Microinterconnect Technologies. Designers of microinterconnects face two challenges. First, many microfabrication technologies are very planar, making it difficult to manufacture out-of-plane structures such as screws and clips. Second, because leakage and dead volume are always referenced to the volume of fluid handled by a device, microinstruments have much lower tolerances to leakage, dead volume, or cross-sectional change.

MEMS researchers have recently reported success in the use of several novel approaches to the miniaturization of fluid interconnections. In the case of robotic systems, for instance, piezoelectric dispensing of picoliter- to nanoliter-sized droplets has been combined with positive displacement pump aspiration to yield a pipetting station capable of moving nanoliter amounts of fluid.

Meanwhile, a group of researchers at the MicroInstruments and Systems Laboratory (MISL) of the University of California, Davis (UCD), has developed a microfabrication approach to interconnection and assembly that is applicable to both robotic and fluidic microsystems. The UCD approach uses traditional micromachining and wafer die-sawing techniques to create structures with an inherently strong mechanical interlock, and ultraviolet light—cured polysiloxane films to accomplish fluid sealing.8 The group has also fabricated silicon structures to be used as tubing connectors. The following section describes the work of the UCD group, with examples of the types of microcomponents they have fabricated.

THE UCD APPROACH

The hybrid approach adopted by UCD researchers addresses many of the most serious issues involved in the interconnection, assembly, and packaging of MEMS modules of different signal types into functional microsystems for in vitro diagnostic applications. Making use of established MEMS fabrication techniques, this approach bridges the technological gap between the current state of MEMS and future single-chip microsystems. And in the meantime, it provides a means for interconnecting the multitude of sophisticated devices and components already recorded in the literature.

MAKING A MEMS INTERCONNECT

In the MicroInstruments and Systems Laboratory at the University of California, Davis, MEMS researchers have successfully created a wide variety of fluidic system components useful for miniaturized IVDs (see accompanying article). All of these components were created using similar MEMS fabrication steps based on processes developed by the integrated circuit industry. Described below are the fabrication steps for creating microinterconnects for a miniaturized fluidic system (Figure 1).

Figure 1. Schematic of a fluid connector designed to fit into a standard device interface to provide connectivity to tubing.

As a first step, a set of interlocking and aligning fingers were fabricated using a diamond-tipped circular saw (Figure 2). In order that parts made on the same wafer could fit into one another after dicing, a set of periodic grooves were cut, with the period set at twice the width of the blade.

Fluidic interconnects require a fluid seal. To meet this need, an O-ring or gasket was made using a patternable ultraviolet light (UV)—cured polysiloxane film 10—100 µm thick. With the appropriate choice of finger length, surface recess, and O-ring thickness, the O-ring is held in compression when the two interlocking surfaces of the component are pressed together.

To produce thin silicon membranes where fluid vias would later be formed, silicon wafers with (100) orientation were etched in potassium hydroxide using a silicon nitride mask. Using either anodic or fusion bonding, these substrates were then attached to channels containing substrates of either glass or silicon. UV-curable polysiloxane, which forms the O-ring or gasket for sealing, was then spin coated on the wafer at 2000 rpm, resulting in a film 10—20 µm thick.

To pattern the O-ring, the film was then exposed to UV radiation. This caused the exposed regions to cross-link and cure on the surface, leaving the unexposed regions to be washed away in the developer (xylene). A shadow mask was used to reactive ion etch (sulfur hexafluoride and oxygen) open the thin silicon membrane inside the O-ring, thereby opening the via.

In space left at the perimeter of each module, a wafer saw was used to machine the interlocking fins that hold the halves of the interconnect together. To produce interlocking modules, the surface of one module must be recessed. This was achieved by potassium hydroxide etching before forming the membranes.

Figure 2. Schematic showing the bonding and assembly of a fluidic input-output device.

Figure 2 shows a cross section of the two mating parts used in the assembly of the fluidic interconnect. The inset shows a set of interlocking fins, where the potassium hydroxide—etched recess is recognizable by the 54.7° angle formed with the surface.

The fluidic interconnection technology developed at UCD is based on plug-in, interlocking structures and patternable gaskets. The interlocking structures are formed from two separately fabricated substrates, each including the elements of mechanical fasteners that hold them together and maintain compression on a gasket that effectively seals the space between them.

When two complementary structures are pushed together, the fastening fins of each structure open slightly to accommodate the incoming fins of its opposite. Friction between the side walls, and the restoring force of the displaced fins, create a locking mechanism that holds both structures in place. The structures can be separated by applying a force greater than the sum of the forces. The very nature of these structures makes them self-aligning and reattachable. Such structures can be readily formed by using a wafer saw to machine the substrates, making this approach applicable to many different types of substrates (see box below).

Fluidic Input/Output Devices. The UCD approach successfully addresses the need for a modular fluidic input/output (I/O) device to serve as a bridge between macro-sized instrumentation and the fluid channels of microfabricated devices. The I/O devices designed at UCD were fabricated by etching channels partway through each of two (100) silicon layers, with each layer having a via etched through one end of the channel and angled dado joints on either side of the channel.9 The two layers were then aligned and fusion-bonded together, so that the interlocked channels formed a silicon tube (Figure 2).

Figure 2. Cross-sectional photo of two mating parts used in the fabrication of a fluidic interconnect. Inset is a close-up view of the mating fins of the module.

Standard tubing was then selected to match the outer diameter of the silicon tube. The tubing was slipped over the silicon tube, and heat (and in some cases sealant) was applied to shrink the tubing and seal the connection (Figure 3).

Figure 3. Close-up of Tygon tubing slipped over the silicon tube of an input-output device.

Finally, the batch-assembled devices were diced apart, leaving a small section of silicon on either side of the silicon tube as a guide and structural support. It was determined that a length of 5 mm was enough for the silicon tube to remain structurally sound while providing sufficient overlap of the tubing to properly seal the connection.

Fluidic System Quick-Connects. By fabricating a structure complementary to its I/O device, the UCD group has also developed a self-aligning and interlocking fluidic quick-connection scheme. The complementary structure was fabricated with etched channels and vias corresponding to those of the I/O device, and with angled dado joints forming silicon beams to match the voids in the I/O device. The two structures were then able to slide into one another (Figure 4).

Figure 4. Fluidic quick-connect, showing end views of the mating tube and gripping alignment guides (top and middle) and quick-connect with fluidic channel interconnect (bottom).

Sealing of the device was accomplished by sandwiching an O-ring between the butted fluid channels. Mechanical fastening of the structures relies on the clamping force created when the silicon tube and guides of the I/O device are pushed together with the silicon beams of its mate. This force was increased by making the base of the silicon tube larger, thereby exerting more pressure on the silicon beams and locking them in place. The devices can be quickly released by applying a force greater than the clamping force of the structures.

Figure 5. Close-up of an assembled specimen holder with three different-sized compartments. Inset shows the layered structure of the specimen holder.

Miniature Biosample Holders. The UCD group has also demonstrated the fabrication and assembly of miniature sample holders formed by using sliding dovetail stages (Figures 5 and 6).10 Since the compartments are crystallographically etched in (100) silicon wafers, they have highly reflective, inwardly sloped walls ((111) silicon planes at 54.7° with respect to the surface), which can be used to view specimens simultaneously on all four sides, in addition to viewing through the top and bottom layers of 7740 Pyrex.

Figure 6. A completed specimen-holder wafer (4-in.) anodically bonded to Pyrex. On the bottom left is an assembled holder partly opened; to the left is a pair of tweezers.

The compartments of the holder can be sealed for either short- or long-term storage. The quality of the seal depends on the materials and sealing technique used, which in turn depends on the temperature and materials compatibility of the specimens. For example, it is possible to anodically bond the Pyrex tops to the silicon drawers, but that can only be performed on dry samples that can withstand temperatures of 100°—150°C. Room-temperature sealing techniques that are compatible with wet specimens include organic sealants such as silicons and waxes.

CONCLUSION

The designers of micro fluid-handling systems for IVD applications have an ideal tool to meet the challenge—MEMS. The MISL group at UCD has developed a microfabricated approach for interconnecting and assembling fluidic and robotic microsystems applicable to miniature diagnostic instrumentation. This technology employs a generic methodology for packaging and assembling MEMS microsystems and offers a number of advantages over other approaches. Features of the UCD technology include:

  • Compatibility with IC fabrication technologies, thereby taking advantage of the established processing infrastructure.

  • Integration with current MEMS technologies, making use of sophisticated MEMS devices already designed and tested, and eliminating the need to introduce a new design platform.

  • Batch fabrication, thereby reducing the cost per unit.

  • Applicability to different types of substrates (e.g., silicon, GaAs, quartz, plastics), facilitating the assembly of multidomain modules fabricated using their optimal technology without compromising their performance.

  • Reversibility, so failed modules can be replaced.

  • Flexibility, making it possible to assemble microsystems from off-the-shelf modules or to interchange modules in order to alter the operation or function of the entire system.

  • Interconnectability, facilitating the transfer of fluidic, electrical, mechanical, thermal, and optical signals from one module to another.

This packaging technology solves many problems that have plagued MEMS, including how to reliably and reversibly interconnect separate MEMS components into functional microsystems, and how to interface MEMS components and macrocomponents to create systems that partake of the benefits of both. With continued research and development along these lines, MEMS is poised to take future analytical systems—and IVD systems in particular—to new heights.

REFERENCES

1. Northrup MA, Chang MT, White RW, et al., "DNA Amplification with a Microfabricated Reaction Chamber," in Technical Proceedings of the 7th International Conference on Solid-State Sensors and Actuators, Yokohama, Japan, Institute of Electrical Engineers of Japan, pp 924—926, 1993.

2. González C, Smith RL, and Collins SD, "Mesoscopic Instrumentation: Preliminary Report on a Miniature Fourier Transform Spectrometer," submitted to Photonics Letters.

3. Allan R, "Silicon MEMS Technology Is Coming of Age Commercially," Electronic Design, 45(2):75—83, 1997.

4. Petersen KE, "Silicon as a Mechanical Material," Proceedings of the IEEE, 70(5): 420—457, 1982.

5. Peeters E, "Challenges in Commercializing MEMS," IEEE Computational Science and Engineering, January-March, pp 44—48, 1997.

6. Markus KW, "Developing Infrastructure to Mass-Produce MEMS," IEEE Computational Science and Engineering, January-March, pp 49—54, 1997.

7. VerLee D, Alcock A, Clark G, et al., "Fluid Circuit Technology: Integrated Interconnect Technology for Miniature Fluidic Devices," in Proceedings of Solid State Sensor and Actuator Workshop, Hilton Head Island, SC, pp 9-14, 1996.

8. González C, Smith RL, and Collins SD, "Fluidic Interconnects for Modular Assembly of Chemical Microsystems," in Proceedings of 1997 International Conference on Solid-State Sensors and Actuators, Chicago, Institute of Electrical and Electronics Engineers, pp 527—530, 1997.

9. González C, "A Modular Technology for Packaging and Assembling Microelectromechanical Systems (MEMS)," PhD dissertation, Davis, CA, University of California, Davis, 1997.

10. González C, and Collins SD, "Mesoscopic Optical Instrumentation," Laser Focus World, forthcoming May 1998.

Carlos González is a research instrument scientist and Jeffery Y. Pan is a group leader, automation engineering, at Abbott Laboratories Pharmaceutical Products Div. (Abbott Park, IL); Scott D. Collins and Rosemary L. Smith are professors of electrical engineering in the MicroInstruments and Systems Laboratory (MISL) of the Department of Electrical and Computer Engineering, University of California, Davis.

Copyright ©1998 Medical Device & Diagnostic Industry

About the Author(s)

Sign up for the QMED & MD+DI Daily newsletter.

You May Also Like