Lab-on-chip inches toward economy

Anyone who thinks that "smaller is better" probably hasn't tried to prototype a microelectromechanical system (MEMS). Developing MEMS designs often requires time and money commitments as big as the systems' micron-sized features are small. The engineers at HandyLab know firsthand about the challenges of prototyping on a MEMS scale. Over the past two years, they've been working to develop a mobile DNA tester that promises to shrink a roomful of laboratory equipment into a package that fits in the palm of the hand and returns results in about 20 minutes. Expected uses for this handheld device include clinical diagnostics, defense against biological attacks, forensics, and even food and beverage testing. At the heart of the tester is a lab-on-a-chip that uses heated air to convey infinitesimal liquid samples to the chip's test sites. HandyLab microsystems engineer Gene Parunak describes this feat of miniaturization in terms of pocket change: "The entire chip is smaller than a dime and has fluid handling channels smaller than the text on a dime," he says.

Because the tester won't hit the market for another couple of years, Parunak won't say much about the chip's proprietary DNA testing technology. At the recent SME Rapid Prototyping Conference, however, he did talk about an enabling technology that is speeding the tester's development. Working with the aptly named FineLine Prototyping (Raleigh, NC), HandyLab has been able to create stereolithography (SLA) prototypes of the chip's microfluidic handing components. With features as small as 50mu (0.0020 inches), "this MEMS application gets very close to the limits of what stereolithography can do," says FineLine president Rob Connelly.

The importance of HandyLab's application goes beyond simply reproducing tiny features. Parunak argues that the ability to prototype MEMS designs quickly and cost effectively could enable a transition away from expensive, time-consuming photolithography (see DN 04.17.02, p. 72). This manufacturing method has served MEMS well, so far, because of its ability to make features down to 10mu (0.0004 inches). But he points out that photo-lithography requires an expensive manufacturing infrastructure, limits material choices, and does a poor job with high-aspect ratio features needed in microfluidics applications. "We think the future will be injection molding because it offers greater design flexibility and lower costs," he says. "But how do you prototype such small molded parts?" High-resolution stereolithography has provided an answer.

Small spot. Also known as "small-spot SLA," this process works much like traditional stereolithography, which builds parts from layers of laser-cured photopolymer. But small-spot SLA features a more finely focused laser as well as other hardware refinements that drastically improve the resolution of the stereolithography process. With its high-resolution Viper machine from 3D Systems (Valencia, CA), FineLine can now cure lines of photopolymer as small as 0.0035 inches wide by 0.002 inches thick-down from 0.010 inches wide by 0.005 inches thick with normal resolution SLA. At the same time, the Viper machine improves the flatness of upward facing surfaces by laying down thinner layers of photopolymer, as little as 0.001 inches in some cases. Enhanced control software, meanwhile, helps compensate for unintended curing of nearby resin-effects known as overcure and print-through. Connelly points out that these resolution numbers represent the ideal case. "Getting the most out of the system still requires an operator with experience in high-resolution applications. There's still some art involved," he says.

With a small-spot SLA machine in skilled hands, the resolution improvements can translate to some truly tiny features in the finished prototypes. Connelly reports that he has produced holes down to 0.006 in. across at aspect ratios of about 2:1. "The minimum size is determined by the ability to get the uncured resin out of the hole," he explains, noting that more flexibility in aspect ratios is possible as hole sizes increase. "At 0.030 inches, virtually any aspect ratio is possible," he says. Miniscule channels, meanwhile, are a bit less challenging than holes since the laser energy doesn't bleed into the channel from all directions as it does with a hole. "We can get down to channels of 0.003 to 0.004 inches wide at a 1:1 aspect ratio," Connelly says. "At about 0.012 inches wide virtually any depth is possible." And while the shape of the laser beam leaves rounded corners on features, with smaller beam size even these radii have been reduced to 0.0015 inches, he adds.

HandyLab took advantage of these capabilities to prototype the fluid-handling functions for its lab-on-a-chip concepts. And it did so in a matter of days rather than the four-plus weeks required to get photolithography parts. Parunak reports that small-spot processes successfully create fluid-handling plates with channels and other features as small as 50mu (0.0020 inches). What's more, the SLA parts have the hydrophilic surface properties that help convey tiny liquid samples around the chip. And he goes on to rave about the quick turnaround for multiple parts, the consistent geometry, and the design flexibility to create through holes and complex shapes.

Despite the benefits, though, the SLA materials themselves do pose problems in this lab-on-a-chip application. As Parunak explains, the light-curable epoxies can produce fluid-handling features accurately, but they don't offer the physical properties needed for the device to perform its actual DNA testing functions. This tough-to-meet balance of properties includes optical transmission above 90%, low autofluorescence, temperature resistance up to 100C, and biological compatibility. "The material has to be non-binding to DNA or bacteria in order for our testing methods to work," Parunak explains.

Tiny tooling. To overcome the light-cure epoxy's inherent shortcomings, Handy-Lab engineers decided to injection mold prototypes of their fluid handling plates in a cyclic olefin. This thermoplastic exhibits all the optical, thermal, and biological properties that allow true functional testing of the chip-sized DNA lab, Parunak reports. "It's really a remarkable material," he says. Even with the molded prototypes, stereolithography played a crucial role in keeping lead times and cost low. Rather than building short-run mold inserts in metal, HandyLab engineers opted to make the inserts right on FineLine's SLA machine. So far, these SLA tools have proved durable enough to withstand about 20 shots when running cyclic olefin. FineLine calls this quick tooling process Direct Inject, and it gave HandyLab access to molded prototypes in less than a week.

According to Parunak, Direct Inject has done a good job picking up where the epoxy prototypes left off. It gave them the ability to mold fluid handling plates with channels down to 0.008 inches. The use of the cyclic olefin provided the material properties that will be needed by the final device. Yet, like the small spot SLA parts, Direct Inject had some shortcomings of its own. The relatively soft tooling material has a limited ability to reproduce detail, resulting in a loss of the finest features. It also has limitations on the kinds of geometry it can produce. The Direct Injection process has not, for example, been able to produce the through holes and other features that might prove useful on the final chip.

With a nod toward the limitations of both small-spot SLA and Direct Inject, Parunak argues that combining the two methods represents the best approach to microfluidics prototyping. Small spot SLA quickly provided the most detailed parts to test fluid-handling concepts, while Direct Inject allowed functional testing of the device's lab.