Originally Published January 2000
Technological advances are every day bringing changes in diagnostic products, enabling IVD manufacturers to address ever-more-complex medical and marketplace challenges. As manufacturers press to resolve such challenges, however, they also make increasing demands on the design teams responsible for developing the tools they use and the products they sell. Such a combination of constantly evolving tools, technologies, and market demands places a similar requirement on design teams and their management. In short, today's rapid pace of technological advance is making it imperative that design engineers develop equally advanced methods for managing product R&D.
As an example, one of the key technologies driving changes in the design and management arena is that of microelectromechanical systems (MEMS), products that combine electronic and mechanical functions in a single microminiature device. Among diagnostic manufacturers, there is a significant effort under way to employ MEMS technologies in an expanding array of products. One vision for the application of MEMS technologies is the production of lab-on-a-chip devices that can perform both the preanalytical and analytical phases of a diagnostic assay. Numerous technical hurdles remain to be overcome before the necessary electronic, mechanical, and fluid-handling features will be able to be combined on a single or hybrid substrate. So long as there is market pressure for smaller, more-versatile, and cheaper diagnostics, however, it is a certainty that such hurdles will be overcome.
One advantage of advanced MEMS devices is the fact that they combine the products of many engineering disciplines into a single unit that can be manufactured all at once. Attaining the desired electronic, mechanical, microfluidic, or optical functions, in other words, requires no further assembly. However, considerable difficulties can arise in the design and manufacturing of a MEMS device for a commercial product, and such difficulties are multiplied when the device is intended for a medical product.
Square Pegs, Round Holes
The fundamental manufacturing processes required to produce MEMS were developed by the semiconductor industry. Scaled up to meet the world's insatiable demand for computer microchips, such processes constitute a cookie-cutter approach to manufacturing: a design that is assumed to be reliable is replicated on a massive scale, and is indirectly tested during manufacturing to verify its functionality and performance. So far, most of the companies that have sought to employ MEMS technologies in IVD applications have also adopted the semiconductor industry's model for the design and manufacturing of such products.
Because the benefits of such a model derive from ever-greater production volumes and lower costs, however, the model has some distinct limitations when applied to the manufacture of medical products. For instance, it forces manufacturers to negotiate the difficult trade-off between having faster manufacturing cycles or devoting time to product testing. In the semiconductor industry, this dilemma is almost universally resolved in favor of faster cycles. Direct testing and verification of a product's performance is shunned in favor of simple tests for known failure modes (e.g., wire bonding failures, die cracking). Full functional testing is rarely performed and, more importantly for the types of MEMS suitable for IVD applications, multidisciplinary testing is almost nonexistent.
Such an attitude toward product testing falls far short of FDA's requirements for verification and validation, which apply to virtually all electronic medical devices and their components. The semiconductor industry has mostly avoided such regulatory liabilities simply by stating that their devices should not be used in medical products. But, obviously such a sentiment can never be considered acceptable if MEMS devices are to be incorporated into any type of medical product.
For product designers and engineers, the implications of adopting MEMS technologies can be understood by considering the array of engineering disciplines involved in producing a single device, and the scale at which they are applied. Advanced MEMS technologies will provide a means to intimately intertwine electronics, mechanisms, optics, and chemistry in a way not easily separable for test and verification. By the same token, the design efforts will not be so easily partitioned as with today's familiar macroscale devices.
In turn, the development of such intertwined microdevices will place a demand on the design team for extraordinary multidisciplinary capabilities and understanding. Because the scale of such devices is so small, every functional unit is either combined into the same physical device or is close enough to affect other functions. For designers who lack a thorough understanding of the technologies and fabrication processes involved in the device, dealing with the effects of such cross-functional interactions will be a tremendous challenge.
One solution to the problem of systems interactions is the use of cross-technology simulation tools that can enable engineers to predict the combined performance of a device's electronic, mechanical, fluidic, and optical systems. However, such simulators are only as good as the models they are based on. They will not predict the performance of systems whose parameters are not well understood and characterized, nor will they reveal parasitic parameters not included in the model. In short, no matter how good a designer's simulation tools may be, true verification of the system must come through testing.
The problems with simulation tools are especially apparent when dealing with microscale devices. Simulating the performance of such a device cannot be done exclusively on one scale. Within the same device, some design problems will require molecular-level modeling, while others must be viewed using a systems-level model that can predict interactions among higher-level functions. To be useful, any simulation tool will therefore have to function on a wide range of scales.
Evaluating such a range of information requires designers and engineers to have an organized, coherent, and insightful understanding of the technologies employed in the device. Moreover, they must be able to make use of this understanding at every scale, from molecular physics through testability for manufacturing.
Eliminating the need for researchers to understand both systems-level and molecular interactions can be achieved by developing advanced modeling and simulation tools that can be used at all levels of design. Such tools would assist engineers to create bulk property models and, using them, to develop complete functional models. Accomplishing such a task, however, would be akin to modeling an integrated circuit from the level of semiconductor physics up through a full circuit simulation. It would almost certainly require a supercomputer—and a lot of time.
For many years, analog circuit designers faced a similar problem. The increasing complexity of analog circuitry made it necessary for designers to understand semiconductor physics in order to successfully predict and achieve product performance parameters. Eventually, the advent of standardized digital interfaces eliminated this need; circuit designers no longer need to know calculus or concepts of charge, they only need to know what functions a device must perform and what companies make components that fulfill those functions. Until the MEMS field reaches such a level of sophistication, the design and design verification of MEMS devices will likely continue to require the direct involvement of scientists and engineers with significant grounding in chemistry and physics.
To reduce costs, companies typically urge their engineers to use off-the-shelf components wherever possible, and to minimize the amount of new design work required for a project. Such instructions obviously make no sense in an environment where every component is new. Developing a strong MEMS-based approach to a diagnostic issue will almost certainly require more design work, not less. And, to meet FDA requirements, manufacturers should anticipate the need for significantly expanding their staff expertise in the area of designing for reliability.
For MEMS devices, reliability is a critical issue that defies understanding based upon the interaction of macroscale objects. MEMS can exhibit puzzling behaviors that are not common to everyday engineering experience. In microscale devices, normally weak electrostatic and van der Waals forces can become dominant influences, dramatically altering the performance of a design that would be reliable in a larger scale. Molecular-level imperfections become significant factors of performance. Small MEMS parts seem to fuse together at the slightest touch, and basic design margins shrink to almost nothing, thereby increasing the likelihood of errors or failures. Further, products that incorporate microscale devices tend to have a greater number of components, and as the number of components in any system increases, so does its failure rate. Due to reduced design margin, the least of all possible consequences related to microscale designs is that the system will become less tolerant of variations arising from unknown or unpredictable sources.
Reliable designs will evolve from improved knowledge about the materials in use and stronger understanding of the theoretical bases for component design. In the present state of the field, however, the available theories and associated mathematics are often unwieldy or suffer from uncertainties due to an absence of detailed knowledge of parameters or initial conditions.
In such cases, designers and engineers must adopt an empirical approach, using experience gained from previous projects to guide their own work. With regard to the specific issue of reliability, for example, historical data indicate that the most unreliable components of any product are those that include interconnections of any kind, including electrical connectors.1 Hence, interconnections should be an area of specific focus. MEMS devices present engineers with opportunities to overcome such obstacles by using such innovative technologies as fiber optics for interconnections.
The Design Team: Science (Still) Spoken Here
From a design perspective at least, the field of MEMS is still in its infancy. Few fully integrated tools are available to support the testing activities necessary for design. In fact, much of the activity that goes under the name of testing is still truly basic research, focused more on confirming the properties of materials and processes than on product-level functional testing.
This condition will change as the field matures. But for now, it places manufacturers in the awkward situation of having to recruit and maintain a design team that comprises both basic scientists and product engineers. Assembling and managing such a team is a difficult task for most market-driven companies. Scientists are generally driven by the quest for pure knowledge, while engineers are driven by the need to apply that knowledge productively. Getting engineers to thrive in a market-driven environment is a challenge with which most R&D managers are familiar. Getting scientists to thrive in that same environment is a task that few such managers would dare to attempt. Getting both to thrive simultaneously is beyond the skill of most managers.
An alternative to involving such a team in every phase of development is to identify a generalist who can act as coordinator for the product development program. Such individuals may not need much technical depth, but they must understand the key mix of concepts and be capable of explaining functional requirements in terms of specifications that technology specialists can use. The remaining team members then become more akin to subcontractors—specialists in a particular technology who are responsible for development efforts only in their areas. This apparently common approach becomes atypical when the need for a creative and responsive mix of science and engineering is demanded.
For companies that are seeking to implement advanced MEMS technologies, deciding what type of R&D management model to follow presents a real dilemma. On one hand, the complexity of the field requires experts with significant scientific depth in a variety of disciplines. On the other hand, the technical range of the field requires experts whose capabilities span the full development process at every scale, from molecular dynamics and systems engineering to process engineering and product testing. Making such a decision can be fraught with peril: aside from the high cost of purchasing advanced-technology products, poor R&D management is probably the greatest obstacle to the commercial success of such technologies.
In today's diagnostics industry, building a team of people that blends such wide-ranging capabilities is a state-of-the-art management task. Maintaining such a team is time-consuming and costly. Ultimately, company management must determine what kind of team it can afford to recruit, recognizing that every choice implies trade-offs, and that even the best of managers may be unable to keep a diverse and high-spirited team together for long.
Companies involved in early-stage R&D often face such management dilemmas. But for the IVD sector, the business environment has shifted in such a way that investors are less and less willing or able to tolerate the effects of poor R&D management. Competitive market pressures, investor demands, and rapidly changing technologies are forcing scientists and engineers to work against ever-more-aggressive schedules and cost limits. Where managers used to press for faster product engineering, they now press for faster basic research.
Moreover, the pressure is on for companies to conduct more and more of their engineering and basic research concurrently. Although many speak of concurrent engineering as though it were commonly a successful process, it usually is not. In the IVD arena, it can be a monumental task to get marketing and engineering to cooperate toward achieving a projected time to market. Adding basic science into that mix makes for an enormous task, and one for which there are no pat or easy answers.
A great deal of work remains to be done to establish an infrastructure for the development of MEMS devices. Methodologies in science, engineering, and manufacturing will be driven to respond to this challenge. But it will also be necessary for companies to change the way that they deal with the business and management aspects of their work. At a minimum, the following goals should be undertaken by industry.
Until product development companies show that they are able to deal effectively with the management dilemmas of early-stage R&D, the investment community will almost certainly continue seeking other, easier sources for gaining a reasonable return on investment. If industry responds as it must to address these issues, the future may hold even more changes for state-of-the-art management of people and businesses than it does for technologies.
1. Reliability Prediction of Electronic Equipment, MIL-HDBK-217F (Washington, DC: U.S. Department of Defense, 1991).
Robert L. Kay is president and CEO of Elite Engineering (Westlake Village, CA).