Rapid prototyping of microfluidic devices with PLT

Rapid prototyping offers other benefits than just a quick turnaround time.

For example, RP—defined as the ability to make 3-D objects from computer-aided designs without tooling—facilitates modifying or customizing earlier-generation prototypes.

Common RP methods include stereolithography, laminated object manufacturing, selective laser sintering and 3-D inkjet printing. These techniques continue to evolve and produce visual and tactile prototypes suitable for mechanical testing. They are invaluable for converting computer-generated ideas into physical reality, reducing development costs and facilitating communication of product performance requirements.

The PLT process can produce complex fluid channels with dimensions of 25µm deep × 100µm wide. All images courtesy ALine.

Some RP techniques have evolved into CAD manufacturing methods. One commercial example of a CAD-manufactured product is custom-formed hearing-aid shells. Another is the new Invisalign dental brace.

Often, though, the prototypes produced by RP processes are not made with materials relevant to the final product. This can result in the prototype and final product performing differently.

A novel fabrication process called polymer laminate technology prevents such performance differences. PLT fabricates microfluidic devices from commercially available sheet and film stock that’s suitable for use in the final product. The process allows solid sample-handling devices that require fluid circuitry, for example, to be created without tubing or without being formed in a diffusion-bonded manifold.

No tooling required

Traditional methods for creating microfluidic devices include glass etching, injection molding and die cutting. These approaches require expensive tooling and commitment to a design that is often poorly understood.

In many cases, a fluid circuit’s requisite feature density and complexity of shape can’t be achieved with traditional injection molding or die cutting. While computer modeling of fluid flow can aid in the initial design work, the large surface-area-to-volume ratio makes laboratory testing of designs and optimization through empirical testing a must.

PLT addresses the unique requirements of microfluidic, lab-on-a-chip devices, where surface-area-to-volume is high; fluid movement is typically laminar; functional elements such as pumps, filters and onboard valves are desirable; and the need for biocompatibility is paramount.

A hybrid RP platform, PLT is a CAD-based method for creating complex fluid circuits that are cut into individual layers, typically with a laser. The layers are then stacked, aligned and bonded using pressure-sensitive or low-temperature adhesives in a batch-fabrication process. Individual devices are then diced from the sheet. Figure 1 shows the main steps in the process. (For more information on the manufacturing process, see "Details for using the PLT strategy for rapid prototyping" below.)

Figure 1: The PLT process for creating a fluid card for a cell culture.

PLT allows the user to test microfluidic circuits made from different CAD files and different materials. The batch-fabrication process is scalable to hundreds of thousands of parts per year, akin to manufacturing printed circuit boards.

PLT also supports the development of microfluidics-based, sample-to-answer products by facilitating interfaces with other components, such as printed circuit boards; silicon, glass and metal-sensing elements; and injection-molded components for reagent containment or sample introduction.

During the PLT process, a laser automatically cuts features in each layer of the CAD design. Laser-cut layers are of two types: Those with cut channel features and intermediary layers that separate channel features and contain vias that connect layers. The bonding adhesive is cut with the same pattern as the channel layer to minimize exposure of the liquid to the adhesive layer.

Product engineering

Figure 2 illustrates the essential steps typically followed in the engineering/RP phase of the product-development process. In the product-concept phase, some initial prototyping that demonstrates proof of concept is done to capture the visual look and feel of the product. Design stringencies related to function are not a concern at this stage.

Figure 2: Essential steps in the engineering process for a microfluidic device.

When design documents and functional requirements have been established, RP streamlines the design/build/test phase, and the project moves more quickly to establish the analysis-instrument-to-fluid-cartridge interface. Once the test bed is developed, prototyping and design testing begin to pick up speed.

In this phase, the ability to iteratively design, build and test accelerates product development. Performance knowledge grows rapidly and significant design changes can occur quickly. Materials and geometry change, functional elements are incorporated and instrument components are established.

With the required instrument components well understood, an alpha prototype ready for testing actual samples is validated.

The PLT process permits ready transition to low-volume production of parts to establish the validity of the test platform. Further along in the development cycle, data is collected to support regulatory submissions and, typically, thousands of parts are required. PLT meets this demand because no time is lost waiting for new tooling when minor changes are required to produce the highest-quality data. Design changes do not impact the fabrication process and can be as simple as changing a CAD file.

Production costs for PLT versus, for example, die cutting vary according to volume. Tooling and process development costs for a large die-cutting production line can reach $250,000. These costs are partially paid up front and partially amortized over several years of contracted part production.

Once a design is set, the PLT production process requires little modification to support high-volume production. And, to satisfy the need for supplier conformance, parts can be 100-percent-inspected with a PLT-fabbed QC tool.

PLT applications

PLT has been successfully proved out in industry.

For example, ALine Inc.’s PLT and volume-production capability supported Osmetech Technology’s eSensor XT8 multiplexed molecular diagnostic platform, from early prototyping through clinical trials to product launch. (The eSensor XT-8 is a microfluidic device for testing sensitivity to warfarin, a blood-clotting drug.)

The PLT process was used to fabricate the award-winning eSensor XT8 microfluidic device, used for warfarin-sensitivity testing.

The device consists of a PCB chip, a plastic cover and a microfluidic component composed of a plate and a laser-cut multilayer laminate. The microfluidic component includes a diaphragm pump and check valves in line with a serpentine channel that forms the hybridization chamber above an array of electrodes.

Osmetech needed a 24-hour turnaround for design and material modifications to the laminate components. The PLT process made it easy to change materials and designs, and optimize performance to meet critical deadlines.

(Editor’s note: Osmetech, a unit of GenMark Diagnostics Inc., Carlsbad, Calif., won the 2009 Medical Design Excellence Award for its eSensor XT8.)

ALine’s PLT platform also supported Rockville, Md.-based TetraCore, a biotechnology company that makes field-portable assays and instruments for veterinary, domestic preparedness and clinical applications. TetraCore needed a quick way to test and develop a low-density microarray intended to function as both an IVD (in vitro diagnostic), point-of-care device and as a lab instrument for detecting hazardous materials.

The microarray device initially consisted of a plastic chamber with a silicone gasket that could be compressed on a microscope slide to create the channel. The design was difficult to change because the gaskets were molded and other parts were machined from plastic. When new gaskets were needed, the machine shop had to make an entirely new chamber for each iteration. During production of a chamber, the microfluidic channels were sealed—insufficiently—with tape.

Using ALine’s PLT process, TetraCore manufactured flow cells containing the microarray that attached to a glass surface. The process allowed TetraCore to quickly test different configurations for the fluidic channel and the chamber, enabling production of new devices with parameters close to the finished product within a few days of receiving a CAD drawing.

TetraCore’s president, William Nelson, said this flexibility will shorten time to market, permit production to scale up quickly and allow the company to mass-produce large numbers of devices in a very short period of time. µ

About the author: Dr. Leanna M. Levine is president and CEO of ALine Inc., Redondo Beach, Calif. Phone: (877) 707- 8575. E-mail: info@alineinc.com. Web: www.alineinc.com.

Details for using the PLT strategy for rapid prototyping

Laser cutting tools

The lasers used for cutting can be of two types: those that ablate material through photochemical bond breaking of the polymer in the plastic material with a UV laser (193nm for the excimer laser or with a frequency-quadrupled Nd:YAG laser emitting at 248nm) or through thermal ablation using a CO₂ gas laser (wavelength 10.6µm), or an Nd:YAG emitting at 1.064µm. Each method has its drawbacks and advantages. The laser cutting process, whether chemical or thermal ablation, interacts with the material based on its molecular properties. UV photochemical ablation relies on the fact that all plastics strongly absorb light below 220nm. While effective at removing material with precision and high quality, it takes a long time to produce a channel with depths greater than 25µm because of the strong absorption coefficient of all plastics at these wavelengths. Typically, UV laser systems use a mask that is placed over the material and the UV laser-beam raster across the surface, only ablating material that is exposed to the beam. UV cutting systems are also quite expensive to purchase and maintain. The footprint for fabrication is limited, making volume production of prototypes impractical.

Figure 3: Polymer Laminate Technology provides the capability to produce devices with fluid channels as small as 25µm deep and 100µm wide with complex channel shapes as shown in the manifold on the top left.

CO₂ lasers ablate material using thermal ablation, heating the material in the infrared and then vaporizing and discharging the material as a plume. The shorter the pulse of laser energy used, the less thermal damage that occurs to surrounding material, making solid-state lasers with nanosecond pulses preferred over the microsecond pulses of a typical CO₂ laser. However, Nd:YAG lasers emitting at 1.06µm, do not interact as strongly with most polymeric materials used for prototyping. For practical application, CO₂ lasers are inexpensive and robust, and are therefore the ideal choice for a wide range of features that are greater than 100µm wide and 25µm in height. (See Figure 3.)

Commonly available CO₂ laser systems cost around $25,000 and are workhorses. In these systems the laser energy is brought to the material using lenses mounted on an X-Y translation arm. The cutting surface remains stationary. The laser moves across the surface in a vector cutting mode, much like a pen plotter, or can be rastered with the pattern etched into the surface being determined by the CAD file loaded into the machine. The typical corner-to-corner repeatability of these systems is 50µm for a 12" x 24" cutting bed. With special focusing optics, the beam size can be as small as about 25µm to 30µm. The small spot size is an advantage in that less power is required for cutting because the power density is higher, while thermal damage is diminished.

Material choices

CO₂ lasers optimally cut acrylic, ABS and Delrin. The laser systems that are used for PLT can cut acrylic up to 3mm thick. The kerf, or the amount of material removed by the cut, is a function of the thickness of the material, and is typically about 50µm for materials up to 125µm thick. With acrylic thicker than 1.5mm, the sloped edge to the cut reflecting the Gaussian beam profile is about 6 percent from the vertical. The edge quality of the acrylic material cut with the CO₂ laser is similar to a machined surface.

Acrylic materials are available in thickness from 100µm (impact modified material) up to the practical limit of 3mm for a typical 25- to 30-watt CO₂ laser tube.

In addition to acrylic, the CO₂ laser can cut almost any thin-film material up to 250µm thick, including polycarbonate, polystyrene, the cyclic olefins (COP and COC), polyester, polyimide, silicone rubber, urethanes, polyethylene, polypropylene and, with some coaxing, PEEK.

Most sheet and film stock materials have a thickness variability that is plus or minus 10 percent of the overall material thickness. All of these films are thermoplastics and have residual thermal stress from manufacture. Since most of the bonding of the devices occurs at or near room temperature, and far below the melt temperature of the materials, they are typically not thermally relaxed before use, except when being used for PCR or when carbon-based electrodes are stenciled onto the plastic and cured at 80 C to 100 C.

Bonding methods

There are an enormous variety of pressure sensitive bonding adhesives (PSAs) available on the market. The performance characteristics that are of interest to PLT include optical clarity, low out-gassing, no leaching of low molecular weight components, alcohol tolerance, temperature tolerance from -40 C to 150 C, high bond strength to low surface energy materials, low initial tack, good structural integrity, thicknesses from 25µm to no more than 125µm.

Silicone-based bonding adhesives seem best suited to meeting these criteria, and have been found to have excellent biocompatibility and are even used for PCR. When used properly they do not ooze into channels, and can be used to create complex channels that are 125µm wide and 50µm tall. As supplied, the thickness variability of the adhesive is a nominal ±5 percent.

Narrow, 125µm channels created as cuts in the adhesive will not provide the same repeatable performance as channels created using a die or master to emboss the features, assuming careful attention to uniform channel depth is paid. Channels formed in these adhesives would not be suitable as timing circuits, for example. But in combination with other control features in the device, such as membranes or valves, these small networks of channels can be used to distribute fluids in manifolds. An example is shown in Figure 4.

Design Rules, Tolerances and Metrology

Design Strategies

Channels are cut from a layer that has adhesive on the top and bottom to minimize exposure of the fluid to adhesive. The capping layers have vias or through holes that lead to the next set of channels in the next channel layer or out through the device to connect to a fluid or pneumatic source or to a sample inlet, or waste reservoir. On-board pneumatic valve layers are usually in a top assembly layer which mates to the channel layer.

Figure 4: Manifolds with networks of fluid channels are successfully made using Polymer Laminate Technology with demonstrated repeatability. These manifolds were made to interface to a pneumatically driven manifold using gaskets.

In considering the various functions of the card, channels typically reside in one or two layers. Reservoirs can be cut to include all the layers and increase the volume containment without increasing the footprint; via layers can be plasma treated to promote capillary flow and even wetting of the device. Porous membranes can be seated into an opening in a thin PET layer with the same thickness as the membrane and bonded onto an area of PSA exposed below, or in the absence of PSA, can be glued onto the layer below using a UV cure adhesive.

Capping layers for the channels can be something other than another plastic layer. Microscope slides, silicon wafers, or microscope cover slips with various modifications are popular alternatives.

Dimensional Tolerances

Channels with dimensions of 125µm width by 50µm tall fabricated in just the pressure sensitive layer have a variability of ±10 percent of the channel width. The height varies by less than 5 percent of the channel height.

Channels with dimensions greater than 125µm wide and 75µm to 500µm tall vary by ±5 percent. Channel shape can also impact the ability to hold tolerances. For example, long stretches of serpentine channel are difficult geometries to hold.

Through holes can be as small as about 25µm.

Metrology

Measurement of channel width variation and feature dimension variation is best done using an optical comparator with photo capture capability.

There are three levels of QC inspection. The first level is visual inspection without magnification to see large debris and gross misalignment. The second level is measurement and averaging of a chosen feature using an optical comparator. The third level of inspection is non-destructive testing of the device. Level one testing is done routinely, and level two and three are done upon request and for a few select pieces from the batch up to 100 percent of the parts produced.

Feature Density

Stacked alignment tolerances are 75µm. Devices with multiple layers should be designed with over sized overlaps between via and channel layers to permit alignment tolerances of 125µm. Features should be no closer than 1mm to ensure that bonding between features is sufficient to prevent fluid from short circuiting through the side of one feature into a neighboring feature.

Surface Treatment

The surface energy of most plastics is low, meaning that the surface is hydrophobic. The hydrophobicity in some cases is an advantage, especially when wicking is not desired. But for many applications, plasma treating the surface will increase its surface energy, making it hydrophilic. Hydrophilic surfaces permit capillary flow and reduce the potential for trapped air bubbles. Common methods for plasma treatment include vacuum plasma with an Ar–O₂ mixed gas, or atmospheric Ar plasma.