Paul Yager Research Group Bioengineering Department, Box 355061, University of Washington, Seattle, WA 98195, USA

---

Progress in

Microfluidic Chemical Analytical Systems

(µFCAS)

Sources of Financial Support

• The Washington Technology Center (1993-present)
• DARPA Defense Sciences Office (1994-1996)
• Senmed Medical Ventures/Micronics (1994-present)
• DARPA MEMS/MicroFlumes Program (1997-present)

Microfluidic chemical analytical systems (µFCAS) in Bioengineering at UW

Motivation for producing such instruments are:

• desire to lower chemical analysis costs by eliminating trained technical personnel,
• the increasing costs of chemical reagents (particularly biochemical reagents),
• the skyrocketing costs of disposal of chemical and biological waste,
• increasing need for rapid and cost-effective chemical analysis

However, development of such tools requires broadly interdisciplinary research and development, which, in turn, requires very diverse training. We therefore assembled an interdisciplinary team from several departments to carry out this project.

The original project funded by DARPA and Senmed was referred to as the "Portable Stat Lab", the idea of which is diagrammatically presented below. The idea was to create a portable device for use in triage in far-forward military medicine.

• First planned for DARPA-inspired combat casualty care
• Now configured for point-of-care, emergency room, operating room, and physicians office laboratory instrumentation

Early PSL Working Design

PSL Preliminary Analyte List

Analyte	Detection Method
pH	Fluorescence
PO2	Fluorescence
Bicarbonate/PCO2	Fluorescence
Na+	Fluorescence
K+	Absorption
Cl-	Absorption
Ca2+ (ionized)	Fluorescence
Hemoglobins, SaO2	Absorption
Bilirubin	Absorption
Hematocrit	Flow Cytometry
Leucocyte Count	Flow Cytometry
Lactate	Absorption
Glucose	Fluorescence
Albumin	Fluorescence

What We've Learned in 3 Years

Miniaturization can buy you more than just the anticipated:

• Smaller sample sizes
• Less use of costly reagents
• Less waste for disposal
• Reduced instrument cost
• Greater instrument ruggedness

However:

• You cannot shrink a device and expect it to work the same way as it did at full size.
• The low Reynolds number flow characteristics of microfabricated devices allows new types of instruments to be designed.

The Reynolds Number

Re is a measure of the ratio of inertial forces to viscous forces

, where r = density u = a characteristic fluid speed, l = length (for example, a channel diameter) h = viscosity

We live in a world in which most of our experience reflect high Re (Re >>1). However, in small channels Re is generally much less than 1, so you get the following conditions:

• no turbulence
• laminar flow
• flow only in the presence of a driving force
• transport by diffusional mixing

Why you don't need valves on a chip

Microfabricated valves have been a major disappointment for use with particle-laden fluids. However, at low Reynolds numbers microfabricated James Brody has shown that valves are not really necessary. Changes in hydrostatic pressure alone (controlled by external valves) can control flow of fluids without appreciable leakage.

Shown are fluorescent beads in a 4-way valve. By controlling hydrostatic pressure or flow from off-chip, the amount of mixing from the non-driven channels is kept to an absolute minimum.

Access to the Microfabricated Devices

From the beginning of the Portable Stat Lab project certain design goals were built into the system. These have influenced our future work . The first was the use of silicon, since we have ready access to Si microfabrication at the Washington Technology Center's Microfabrication Laboratory at Fluke Hall. Optical access is on one side of the modules. All wet connections are made from the back side of the module. The modules are sandwiched between the permanent optical devices and the wet delivery systems. Very precise alignment required for fluorescence and flow cytometer modules. Face-to-face connection with flat gaskets allows far simpler optical and fluid connection technology

Extraction of Plasma Constituents from Blood Cells

• Some assays must be performed in cell-free plasma, so a quick way of removing blood cells had to be developed.
• Centrifugation was considered and discarded as impractical for an instrument designed for field use in which multiple samples would have to be processed between servicing.
• Novel devices--a tangential flow filter and an "H-Filter™"--were developed and tested.
• These are applicable for extraction of analytes from many particle-laden fluids.

Principle of the H-Filter™*

In a thin channel at low Reynolds number, turbulence cannot exist. Flow streams can run next to each other without mixing; the reversible apposition of two fluids as at left is possible only in devices with small dimensions. Diffusion is the dominant transport mechanism between adjacent flow streams. Large particles diffuse slowly and stay on their side of the channel; small molecules rapidly diffuse across.

*trademark of Micronics, Inc.

SEM of the Si portion of an early prototype "mock" H-filter™. The two outlets have been combined for simplicity of testing. Use of anisotropic wet etching causes walls shapes to be nonideal, particularly at corners. These facets act as optical mirrors. RIE and LACE can solve these shape problems.	H-filter™ in Action: The "mock H-filter™" sealed by a Pyrex cover slip is shown in a fluorescence optical micrograph functioning to separate fluorescent dye from a mixture of beads and dye. Beads and dye enter from inlet 1, and buffer from inlet 2. After the diffusion channel all beads remain in the upper channel; fluorescent dye is distributed between the upper and lower channels.

For Larger Volumes: The "Flat Filter"

By rotating the flow direction of the H-filter™, much greater contact area between the adjacent streams is possible, and scaling up throughput is just a matter of increasing the device width. It can be fabricated in Si or moldable materials.

Absorption-Enhanced Field Flow Fractionation (AEFFF)

Concentrating Dilute Analytes

Giddings' FFF can be particularly well applied using our Flat Filter design.

Any field applied perpendicular to the flow direction can be used to concentrate dilute analytes many-fold. For example, consider concentration of DNA from cellular lysates by forcing DNA to one wall of the channel by electrophoresis, followed by removal in a reduced flow from that side of the channel. At small distances involved, only small voltages need be used.

Fields Applicable for Sample Extraction and Fraction (as in FFF)

• Concentration gradients, leading to diffusion, which is the primary embodiment of this invention investigated to date.
• Electrical fields, leading to electrophoresis to move molecules/particles to one side of the channel against a concentration gradient (consider that this is an excellent method for concentrating DNA).
• Temperature gradient, leading to diffusion of particles to the cooler side of the channel
• Cross flow, in which particles are subject to a flow of liquid perpendicular to the longitudinal flow direction, allowing separation by size, density, or diffusion coefficient
• Magnetic fields, which can be used to move ferromagnetic particles to one side of the channel
• Hydrodynamic forces such as the lift effect, that tend to concentrate particles in certain regions of a flow stream.

The T-Sensor™

Discovery of the H-Filter™ and its observation by optical microscopy led to the development of a novel sensor based on controlled diffusion of analytes in small spaces. The T-Sensor™ can be used to measure analytes in simple or particle-laden fluids. It is particularly well suited to optical detection, although other sensing modes are also possible.

Analyte from a sample stream diffuses into an indicator steam. Approach to steady state down the channel creates a pattern of indicator color or intensity change. The position of the boundary is sensitive to flow rate and changes in analyte concentration. The indicator can be free, complexed with a slowly-diffusing polymer, or immobilized on particles.

pH measurement can be made in whole or diluted blood, as shown above.	How to Pack a Lot of T-Sensor™ into a Small Space. Microfluidic devices don't have to be straight to have orderly laminar flow. Compact forms of devices can use less Si real estate.

Variants on the Simple T Sensor	The same microfluidic devices allow one to bring into adjacent interdiffusive flow more than two streams, allowing sequential processing of analytes without mixing of various components.	Sample AND Reference? The Control T-Sensor
	For example, one could bring together sample, a non-diffusing enzyme solution, and an indicator stream, so that only the small molecule product of an enzymatic reaction would reach the indicator region of the device. The intermediary fluid stream can act as a continuously renewable enzyme-containing membrane!		Alternately, a central indicator stream could be flanked by the sample and a reference stream This "Control T-Sensor" allows real-time calibration with a single image. This is a particularly important advantage for biomedical instrumentation.

3 Ports and Beyond--Referencing and Parallel Analysis

at the mixing point	just below the mixing point	several mm downstream of the mixing point

We have fabricated and tested devices with 3 or 5 input ports. Laminar flow is achieved and streams can remain distinct far downstream if the colored species are slow to diffuse. Shown is a pH sensor in which a bromcresol purple solution is flowing between buffer at pH 7 (left) and pH 9.5 (right).

Control-T-Sensor: Imaging of Interacting Streams and Data Reduction

We are currently using optical microscopy to monitor the concentrations of analytes and controls in T-sensors.

The channels are imaged using a color or B&W camera, the images digitized and processed to produce a data set consisting of intensity vs. position across the channel. Often a finite length along the channel is imaged to improve the S/N ratio.	To calculate the concentration of the unknown we collect values for: A: control blank B: control peak C: reagent blank D: sample peak E: sample blank

T-Sensor determination of albumin (HSA) in serum using the fluorescent indicator AB580

Modeling of optical signals observed in T-Sensors™

By combining knowledge of the absorption and/or fluorescence intensities of each species, their respective diffusion coefficients, and the chemical reaction stoichiometry and kinetics, it is possible to model the peaks observed in the T-sensor™ and Control-T-Sensor™. Shown are an example of real data (green) and a modeled fit (blue).

T-Sensor Evaluation

Several different types of biomedically relevant analytes have been measured using the Control-T-Sensor™. These include pH, Ca⁺⁺, as well as small molecules, enzymes and antibodies.

The T-Sensor™ also converts reaction kinetic data into spatial information. Time-independent kinetic concentration determinations can be made long T-sensors under continuous flow. Reaction rates are determined by measuring the dye absorption as a function of distance from the T-joint. This has been applied to measurement of alkaline phosphatase in serum.

Strengths	Challenges
Most standard fluorescent or absorption methods applicable which were designed for macro-scale serum analysis are applicable for µ-quantities directly in whole blood Applicability to wide range of particle-laden samples Multiple reactions can be carried out side-by-side. Experimental controls and calibration possible in real time.	An optical path length of <50 µm requires adjustment of some reagent concentrations from those used in larger samples Precise flow control is advantageous Control of surface adsorption is required for low-concentration analytes.

---

Flow Cytometer and Fluorescence Cell

Additional technologies have also been developed for the original Portable Stat Lab concept. Among these is the use of a v-groove in silicon as the basis for a low-volume flow cytometer, as shown below. Cells are confined in a groove etched in silicon using an anisotropic etchant. Subsequently the addition of a glass cover by anodic bonding creates a defined cavity large enough to pass all blood cells. These can be monitored by light scattering (both near-forward and 90°) or fluorescence, if the cells are labeled.

Demonstration of Cell Discrimination Capacity of the Flow Cytometer

Cells can be discriminated by their unique ratio and absolute intensities of angular scattering. Below is a 2-D histogram of cell counts in a red cell-depleted sample of blood taken with the instrumentation shown above.

This sample of blood was sedimented to decrease the proportion of RBCs before sending it through the V-groove. Note that platelets (Plts), red blood cells (RBCs) lymphocytes (L), monocytes (M) and neutrophils (N) are distinguishable.

The flow cytometer technology can also be used in conjunction with chemically sensitive beads to create a high-throughput chemical monitor.

For further information, consider linking to the home page of Micronics, Inc, to which these technologies have been licensed.