1701ISSN 1757-6180Bioanalysis (2010) 2(10), 1701–171010.4155/BIO.10.140 © 2010 Future Science Ltd

Over the past 15 years, an important trend has emerged in analytical chemistry, that is, the development of miniaturized measurement devices or micro-total ana lysis systems (µ-TAS). These systems aim to integrate several ana lysis steps (e.g., sample preparation, DNA/RNA amplification and detection) into a single, auto-mated device or ‘chip’ [1–3]. To accomplish this goal, these systems usually consist of a set of microfabricated chambers and channels, com-bined with a method of microfluidic pumping for fluidic movement [4]. As a result of this min-iaturization, µ-TAS feature several advantages over traditional benchtop ana lysis systems, including lower power consumption and space requirements, minimized sample and reagent size and shortened ana lysis times [5].

Microfluidics is the scientific/engineering discipline underlying µ-TAS and constitutes a promising field that enables automation of min-iaturized and multiplexed complex bio assays. Despite the many advantages µ-TAS presents, the emphasis on miniaturization has made it difficult to handle real-world sample sizes. Microfluidics is indeed ideal for handling very small samples and reagents (microliter range), but ‘real-world’ or clinical samples are often of the milliliter-scale. The reason that real-world samples of blood or saliva often have to be this large is that they may contain only very small concentrations of the target analyte (nucleic acid, protein, bacteria or viruses), perhaps only ten copies per milliliters of blood sample [6]. As a consequence, most detection schemes, for exam-ple those involving real-time polymerase chain

reactions (PCR) that have a limit-of-detection of ten copies, require testing of a sample larger than 1 ml. A smaller sample cannot guarantee positive detection even if the sample is positive for the target pathogen.

Real-world sample-driven analytical systems present new challenges for the field of micro-fluidics. Somewhat contradictory, these µ-TAS devices must be able to handle several milliliters of sample while keeping most other functions within a microstructured architecture. This real-ization forces one to put a heightened emphasis on the development of µ-TAS devices capable of interfacing with meso-scale fluidic samples in order to meet real-world demands.

The first section of this article briefly covers the background of microfluidic platforms while highlighting previous work on blood separation with such devices. Next, we discuss the motiva-tion for a large-volume blood separation device and the fluidic processes utilized in one spe-cific platform implemented to accomplish this: a centrifugal microfluidic platform. Then we detail our research methods and the experimental validation of a large-volume blood plasma separa-tion device based on such a centrifugal microflu-idic platform. We conclude with a discussion of the challenges and future work in sample-driven microfluidics based on real-world requirements.

Microfluidic systems for blood separationThe separation of blood cells from blood plasma is an important first step in many bioanalytical assays. For example, the cellular components of

Large-volume centrifugal microfluidic device for blood plasma separation

Background: Microfluidic-based systems are ideal for handling small microliter volumes of samples and reagents, but ‘real-world’ or clinical samples for bioana lysis are often on the milliliter scale. We aimed to develop and validate a large-volume centrifugal or compact disc-based device for blood plasma separation, capable of processing 2 ml undiluted blood samples. Results: This automated blood sample preparation device was shown to yield high purity plasma in less than half the time of commercial plasma preparation tubes, while enabling integration with downstream ana lysis and detection steps. Conclusion: This article draws upon a novel large-volume device to further illustrate the challenges in combining microfluidics structures with large-volume samples and the implications for sample-driven microfluidics systems.

Mary Amasia†1 & Marc Madou2,3

1Department of Chemical Engineering & Materials Science, University of California, Irvine, CA, USA 2Department of Chemical Engineering & Materials Science, Department of Biomedical Engineering, Department of Mechanical & Aerospace Engineering, University of California, Irvine, CA, USA 3World-Class University (WCU), Ulsan National Institute of Science & Technology, Ulsan, Republic of Korea †Author for correspondence:Tel.: +1 949 824 4143 E-mail: mamasia@uci.edu

Special FocuS: MicroFluidicS

reSearch article

For reprint orders, please contact reprints@future-science.com

reSearch article | Amasia & Madou

Bioanalysis (2010) 2(10)1702 future science group

blood can inhibit proper PCR amplification of targeted DNA and interfere with absorbance measurements, and must therefore be separated from the plasma before downstream ana lysis steps can take place [7]. Traditionally, blood plasma separation occurs in a hospital setting, in large-scale floor-based centrifuges. This pro-cess is time-consuming and requires several procedural steps to be performed by a skilled laboratory technician.

Many microfluidic systems to automate blood separation have been developed, using various separation mechanisms to achieve a cell-free plasma sample. Blood separation in micro-fluidic devices has been demonstrated based on dielectrophoresis [8,9], diffusion [10], filter/micro-filter [5,11], Zweifach–Fung Effect (at a bifurca-tion point, blood cells tend to travel into the channel with higher flow rates) [12–14], bends in microchannels [15], cross-flow filtration [16] and acoustic waves [17].

Centrifugal or compact disc (CD)-based microfluidics offer a simple approach to blood sample preparation as they allow for the multiplexing, automation and miniaturiza-tion of the classical blood-separation technique based on sedimentation. By exploiting density and size differences between the various blood components, one can sediment the denser cel-lular components of blood and is left with a cell-free plasma sample. A thorough descrip-tion of the separation mechanisms of blood cell sedimentation within centrifugal microfluidic devices have been detailed by Haeberle et al. [18]. Several groups have demonstrated plasma sepa-ration of small-volume raw and diluted blood samples using centrifugal microf luidic sys-tems [1,18–22]. Recently, Zhang et  al. demon-strated plasma separation from diluted blood samples using curved microchannel structures on a centrifugal device [23].

Absorbance-based blood chemistry analy-ses and immunoassays may require only small volumes of plasma for target detection, start-ing from diluted blood samples. However, no microfluidic systems, particularly no centrifugal microfluidics, have been developed to process raw blood sample sizes in the order of several milliliters to meet the needs of most real-world analytical methods. For example, to diagnose sepsis, a bacterial infection in the bloodstream (sometimes called blood poisoning), one needs to test several milliliters of whole blood to detect the bacterial cause of infection. In addi-tion, large sample volumes are necessary to

detect most viruses or blood-borne pathogens in a plasma sample via standard nucleic acid ana lysis techniques.

Compact disc-like centrifugal microfluidic platformsCompact disc or centrifugal microfluidics con-tinue to be of interest for integrated bioana lysis devices as they hold many advantages over other microfluidic platforms due to the unique pump-ing and valving mechanisms involved [20,21,24,25]. When a CD is spun, the centrifugal force exerted upon fluids in the reservoirs and channels pro-vides a means of propulsion from the CD center towards the rim of the disc. Centrifugal flow rates are dependent on angular velocity (rota-tions per min [rpm]), radial location from the disc center, channel geometries and fluidic prop-erties. Centrifugal pumping has been demon-strated to result in well-characterized flow rates from 5 nl/s to 0.1 ml/s [24]. For this reason, both large and small samples are equally simple to pump within a centrifugal device. Unlike AC and DC electrokinetic and electro-osmotic pumping, physicochemical properties of the fluid have little impact on centrifugal pump-ing efficiency. The centrifugal platform’s rela-tive insensitivity to sample properties is highly beneficial in sample preparation aspects, where a sample’s pH and ionic strength vary greatly. Biologically relevant samples such as blood, urine and milk, as well as various solvents and surfactants, have all been pumped successfully in this platform [20].

Valving within centrifugal devices is com-monly performed based on the competition between centripetal force and surface tension in a fluidic conduit such as in hydrophobic con-duit constrictions, hydrophilic conduit expan-sions (capillary valves) and siphons [20,21,24,26]. Hydrophobic valves rely on hydrophobic con-strictions or patches of a more hydrophobic material that hold back fluid until the rotational speed exceeds a critical value. Capillary valves, a typical design of which is a sudden expansion of a microchannel with hydrophilic surfaces, work in a similar way to hydrophobic valves: the fluid will not move past the conduit expansion until the centrifugally induced pressure overcomes the capillary barrier pressure, at the ‘burst fre-quency’ [27,28]. Siphon valving, employed in the device presented below, is ideal for situations that require downstream processes at lower centrifu-gal spin rates than upstream processes. When the device is spun at high rpm, the centrifugal

Key terms

Sample preparation: Steps needed to process a collected sample so that it can be analyzed, often the most time-consuming aspect of a biological assay.

Centrifugal microfluidic platform: Microfluidic platform that utilizes centrifugal force for fluid propulsion within compact disc-like devices.

Multiplexing: Simultaneously processing many samples in a single assay run.

Large-volume centrifugal microfluidic device for blood plasma separation | reSearch article

www.future-science.com 1703future science group

force resists the capillary action and the meniscus front does not go past the siphon crest; only when the spin speed is dropped to a low rpm does the capillary force overwhelm the centrifugal force, and surface tension pulls the meniscus over the crest, or ‘primes’ the siphon. The rotational speed can then be increased again for fluid transfer into an outer chamber [22]. Recently, serial siphon valves have been demonstrated, in which mul-tiple in-line siphon valves allow for a more con-trolled, sequential release of fluids, less depen-dent on dynamic changes in surface tension [29]. In addition to the passive valves discussed above, active valving techniques, which must be inde-pendently actuated and do not rely on surface characteristics (e.g., wax melting or magnetics), have been utilized in centrifugal platforms [30].

In addition to pumping and valving solu-tions, other f luidic functions such as mix-ing, metering, volume definition and coriolis flow switching have all been implemented on centri fugal platforms. Sample lysis and homo-genization, PCR amplification and cell-based assays and culturing have been demonstrated, and electrochemical and fluorescent or absorp-tion-based optical analytical measurements have been used for detection means on a centrifugal platform [20,21,24].

Although most of the recent developments in centrifugal-platform technology have occurred in academia, there are several centrifugal devices that have been commercialized. In 1995, Abaxis North America (CA, USA) introduced the Piccolo® system for blood ana lysis, and currently sells a wide range of blood-panel products for both medical and veterinary diagnostics [101]. Gyros AB (Uppsala, Sweden) has commercial-ized the GyroLab® workstation and Bioaffy CD, which automates fluorescence-based sand-wich immunoassays for antibody detection and pharmacokinetic studies [102]. Samsung (South Korea) has developed an automated bead-based ELISA CD for hepatitis detection from whole-blood samples [31], and launched a centrifugal blood analyzer product in June 2010.

Experimental section One of the major remaining challenges in the development of µ-TAS, or sample-to-answer devices in general, is the interfacing of large sample volumes with miniaturized downstream fluidic devices. The motivation for this current study is to address the proper integration of real-world large sample volumes with microfluidic devices. In the following sections, we will describe

our research on a centrifugal micro fluidic device for the sample preparation of large-volume blood samples for pathogen detection via nucleic acid ana lysis. Along with the biological validation of our large-volume centrifugal device, we discuss the problems that arise from interfacing large volumes and microscale features.

�� Device fabricationThe large-volume centrifugal blood–plasma sep-aration device we present in this study consists of layers of plastic held together by layers of double-sided pressure-sensitive adhesives (PSAs). The plastic layers are cut from stock polycarbonate sheets (McMaster, Carr, CA, USA) using a com-puter-numerical control machine (T-tech, GA, USA – QuickCircuit 5000). A cutter-plotter (Graphtec, Japan – Graphtec CE-2000) is used to cut the PSA layers from 0.004 inch-thick PSA (FLEXcon, MA, USA – DFM 200 Clear V-95 150 POLY H-9 V-95 4). After the five layers are machined, they are aligned using a jig with both a center spindle and outer-radius pin. Once aligned, the set of five layers are pressed together with a manual press (Desert Laminator, CA, USA) and the layers of polycarbonate are per-manently bonded together with the PSA layers to form a monolithic device. This plastics fabrica-tion technique does not require any photolitho-graphic methods, and can be extended to low-cost manufacturing processes such as injection molding or hot embossing.

The plastic surfaces must be subjected to oxy-gen plasma treatment to ensure that all surfaces in contact with liquids are rendered hydrophilic. This surface treatment process is carried out in a Technics II O2 Asher at 200 W and 200 mTorr for 2.5 min. All devices are tested within 3 days of surface treatment to minimize effects due to dynamic contact angle variations.

The device presented here, and shown in Figure 1, consists of five layers: the top poly-carbonate disc with inlet and vent holes (1/16-inch thick), top adhesive layer with chan-nels (0.004-inch thick), middle polycarbonate disc with chambers (1/8-inch thick), bottom adhesive layer with chambers and channels (0.004-inch thick) and bottom polycarbonate disc (1/16-inch thick).

�� Design testing & image acquisitionPrior to biological sample evaluation, the disc design was tested by strobe-imaging of DI water and contrast agents (McCormick, MD, USA – Neon food dyes)-filled disc to ensure

Key term

Sample-to-answer devices: Devices designed to integrate and automate all steps of a biological assay, including sample preparation, amplification and detection.

reSearch article | Amasia & Madou

Bioanalysis (2010) 2(10)1704 future science group

proper fluidic movement and to experimen-tally determine the spin procedure. The disc is securely affixed onto an aluminum chuck that is attached to a servo motor (Pacific Scientific Servo Motor) with corresponding motor con-troller (PAC SCI Programmable Servo Drive). This motor/controller system is used to control the spin speed, acceleration and deceleration rates, and absolute position of the disc. The software ToolPAC is used to program the motor to achieve individual spin profiles, tailored for a specific application.

The imaging system is composed of a camera (Basler A301bc, 640 × 480 pixels, 80 fps maxi-mum, 10× zoom lens mounted), a strobe light (PerkinElmer MVS-4200, 6-µs duration), and a retro-reflective fiber-optic sensor (Banner D10 Expert Fiber-Optic Sensor). A small piece of reflective tape is placed along the outside radius of the disc. Once per rotation, the reflective tape passes under the fiber-optic sensor, which actuates the strobe light and triggers the camera to capture one image frame. The still frames images of a particular area on the disc can then be compiled into a single video file for fluidic movement observations.

�� Biological testingGender unspecif ied porcine whole blood samples were ordered (Lampire Biological Laboratories, Inc., #7204907) and arrived pre-treated with anticoagulant K

2EDTA. Blood

samples were brought to room temperature by sitting on the laboratory bench for 30 min prior to their insertion into the centrifugal device. The hematocrit percentage was experimentally determined through high-speed sedimenta-tion for 20 min and visual inspection to be 49.5 ± 2% for four individual samples of por-cine blood from lot #91116PK

2EDTA. For each

run, 2.0 ml of whole-blood sample was added to the device prior to testing, with no dilution of the sample. All experiments were performed within a span of 3 h to minimize discrepancies between samples.

The plasma separation spin procedure was performed as follows: the 2.0 ml sample was input into the large sedimentation chamber, identified in Figure 2, and all vent and inlet holes were sealed off from the environment by an adhesive-backed thin film. The disc was then spun up at an acceleration rate of 500 rpm/s up to a sedimentation frequency of 3800 rpm. In order to study the effect of centrifugation time on plasma purity, the sedimentation por-tion of the run was programmed as 2.5-, 5-, 10- or 20-min durations. After sedimentation, the device was decelerated at 50 rpm/s down to 350 rpm. This relatively slow deceleration was necessary to minimize perturbations to the plasma–blood cell interface. If the decelera-tion were set to occur too quickly, sedimented cells would mix with the separated plasma, negatively impacting the purity of the plasma sample. The siphon-priming rotational speed, 350 rpm, was experimentally found to be the highest rpm that allows for the siphon valve

Top plastic disc with inlet/vent holes (1/16-inch thick)

Middle plastic disc with chambers (1/8-inch thick)

Bottom adhesive with chambers (0.004-inch thick)

Bottom plastic disc (1/16-inch thick)

Top adhesive with microchannels (0.004-inch thick)

1 inch

Figure 1. Exploded view of the multilayer fabricated centrifugal device.

Sample inlet hole

Channel for fluidicself-venting

Transfer siphon

Collection chamberPlasma collection hole

Sedimentation chamber

1 inch

Figure 2. Close-up of a single device on the disc.

Large-volume centrifugal microfluidic device for blood plasma separation | reSearch article

www.future-science.com 1705future science group

to prime. Choosing the highest siphon prim-ing speed is necessary because if the rpm drops too low, the interface between blood cells and plasma can be compromised and purity will drop. After 30 s at 350 rpm/s, the siphon will have fully primed, and the spin speed can be increased at 20 rpm/s to 600 rpm to pump the 750 µl of separated plasma sample into the lower collection chamber. This rotational speed profile versus time is depicted in Figure 3, which also illustrates the regimes for the plasma separation process.

After centrifugation, the plasma samples were collected from the device by removing all sealant films, and pipetting out the 750 µl plasma sample via the plasma collection hole. These samples were immediately pipetted onto a hemocytometer slide (Hausser Scientific, PA, USA) for cell counting purposes. A total of 10 µl was added to each chamber of the hemocytometer and the cell counting was performed in triplicate for all the collected plasma samples.

Results In this section we provide device design jus-tifications as they pertain to processing large-volume samples, present the results from the biological validation experiments, and con-clude with a discussion on sample-driven approaches and the challenges remaining for these microfluidic systems.

�� Device design & justificationIn small-sample volume CD-based devices, the liquid sample is of minimal mass and will not exert much force on the device walls. When dealing with devices capable of handling large volumes of sample, the pressures created at the outermost radius of the chamber are large enough to have a damaging effect on the multi-layer strength of the prototype discs – in effect, prying the layers apart. The liquid volume is 20-times larger than typical small-sample CD-based devices, and this volume is expected to exert a similarly greater pressure on the outer edge of the sedimentation chamber. In pre-liminary testing of this device, we found sample leakage to occur at high rotational velocities, as the sample would seep out from between two layers of the device. To counteract any negative effects from the heightened pressure, the top and bottom polycarbonate layers of the device were designed to be relatively thick so that the chamber is less susceptible to deformation and sub sequent delamination. As shown and detailed in Figure 4, three additional finger-like struc-tures, approximately 0.255 × 0.074 inches, were added to the large chamber design to increase the surface area between the disc layers. These finger-like structures help to lessen the pres-sure along the outer radius of the chamber by distributing the centrifugal force over a larger surface area. This design also increases surface area between the plastic layers of the disc device,

Time [sec]

Rota

tiona

l Vel

ocity

[RPM

]

Cell Sedimentation SiphonPriming

Plasma Transfer

0 50 100 150 200 250 300 3500

500

1000

1500

2000

2500

3000

3500

4000

Ro

tati

on

al v

elo

city

(R

PM

)

Time (s)

Cell sedimentationSiphon priming

Plasma transfer

Figure 3. Spin profile graph detailing rotational velocity over time during the on-disc plasma separation process. The phases of sedimentation, siphon priming and siphon transferring are depicted in this process.

reSearch article | Amasia & Madou

Bioanalysis (2010) 2(10)1706 future science group

allowing for more adhesive contact and, there-fore, a stronger bond. Along with the thicker top and bottom plastic layers, these design changes were enough to ensure leak-free processing of the large-volume samples.

Hematocrit levels in blood vary depending on the gender, heath and species of donor, with human blood ranging from 30 to 52% hemato-crit. To account for these expected variations in the amount of red blood cells, the siphon height for the device was fixed at a level above the vol-ume of blood cells according to a 52% hematocrit level, so that only the cell-free plasma portion is transferred into the collection chamber. The inte-gration of ‘tunable’ siphon channel heights would enable more complete plasma volume collection while maintaining high plasma-purity levels.

Additional channels were designed into the device to allow for ‘self-venting.’ As the fluid fills up a chamber, the now-displaced air is able

to move back towards the disc center via the self-venting channel. This design allows the bio-logical samples to be self-contained within the device during the entire testing procedure. In the development of point-of-care devices, the diag-nostic testing will not be performed by highly trained individuals in a sterile hospital environ-ment, but instead by nonprofessional clinicians at an airport, classroom or third-world clinic. In these environments, it is extremely important that the diagnostic systems are self-contained, thereby minimizing exposure to device users and possible contamination from the environment.

�� Biological validationExperiments were performed to validate the efficiency and the purity of the plasma separa-tion from raw blood samples on the CD fluidic platform. We compared plasma purity from our device with that of commercially avail-able plasma preparation tubes (BD Biosciences 362788, NJ, USA). Although the commercial tubes are designed for 5 ml of blood, we input 2 ml of whole blood in both the tube and disc device testing, so that a clear comparison could be made of their relative designs. In both the tube and CD device testing, the relative centrifugal force (RCF) was normalized to the procedure for commercial tubes, specified at 1100 RCF. A formula detailing the relationship between RCF and rpm is:

rpm 2.839 10RCF

r5=$# -

equation 1

where r is the radius of the rotor, in inches. In this work, r corresponds to the outer radius of the separation chamber for the disc device, and the outer radius of the tube within the table-top centrifuge. As prescribed by equation 1, the angular frequency of the CD device was set at 3800 rpm, corresponding to the suggested RCF for plasma separation in the tubes. The angular velocity of plasma separation must be high enough for efficient separation, yet not so high that hemolysis will occur [32]. In order to study the effect of centrifugation time on plasma purity, the sedimentation portion of the run was programmed for durations of 2.5, 5, 10 or 20 min. These sedimentation times were chosen to span the typical range for sedimen-tation, encompassing the specified centrifuga-tion time of 10 min for the commercial plasma preparation tubes.

2.69 inch

2.04 inch

1.44 inch

0.074 inch

0.255 inch0.255 inch

1 inch ω

Fcf

Figure 4. Device design detailing the position and radial distance of the various design elements, including the finger-like structures. The direction of applied centrifugal force is depicted as F

cf. The finger-like structures aid in

distributing centrifugal force and increasing surface area for improved adhesive strength of the multilayer device.

Key term

Relative centrifugal force: Force acting on a particle within a centrifugal field.

Large-volume centrifugal microfluidic device for blood plasma separation | reSearch article

www.future-science.com 1707future science group

Through visual inspection, the device was successful in processing the raw blood sample, as blood cells were observed to sediment below the siphon channel height after centrifuga-tion for 2.5 min. In Figure 5, a chronological still-frame timeline of the sedimentation pro-cess on-CD is shown. The cell sedimentation (Figure 5B & c), siphon priming (Figure 5d), and plasma transfer (Figure 5e & F) phases of the plasma separation scheme occur within 320 s, matching the programmed spin profile in Figure 3.

The results from tube and CD-based device comparison experiments are shown in Figure 6. All sedimentation runs resulted in very high purity plasma samples, much greater than 99% purity, defined as the percentage of blood cells removed from the starting raw blood sample. In comparing the tube and CD devices for the 5-min sedimentation run, the CD-based devices resulted in separated plasma of higher purity. Specifically, sedimentation within the CD device for 2.5 min resulted in plasma of purity equivalent to plasma resulting from a tube device run of 5 min.

Plasma samples of purity 99.9% and above are all likely to respond similarly to bio-analytical detection techniques. Although not shown in the graph, a tube sedimentation run

of 2.5 min resulted in plasma purity levels of 58%, compared with 99.9997% purity from the CD device. This result was expected since the protocol for the commercial tube requires centrifugation for at least 10 min. In situations where blood cell sedimentation needs to occur very rapidly, only the CD device would produce plasma with required high purity levels.

The total process time, including accelera-tion and deceleration times for the CD-based system, is 320 s, much faster than a floor centri-fuge, which requires a long deceleration phase so that the separated plasma is not disturbed from rapid deceleration of the rotor. This CD-based system automates the plasma col-lection into a separate chamber, removing the additional step of manual pipetting, thereby removing the need for this slow deceleration final step since the plasma is already separated into another chamber and is no longer at risk of cell resuspension.

Discussion�� Moving from tubes to flat substrates

The move from tubes to flat CD-based structures leads to some, perhaps unexpected, benefits. For example, by distributing a 15 µl tube-based PCR reaction volume over a large, flat surface, one is able to heat and cool that volume much faster,

t = 0 s

t = 280 st = 240 s

t = 150 s

t = 310 s

t = 20 sA B C

D E F

Figure 5. Still-frame images of blood-separation process within a compact disk-based plasma-separation device. The plasma processing steps include: (A) sample input, (B) partial sedimentation at 3800 rpm, (C) total plasma separation after 2.5 min, (D) siphon valve priming, (E) plasma transfer into collection chamber and (F) complete transfer of cell-free plasma sample.

reSearch article | Amasia & Madou

Bioanalysis (2010) 2(10)1708 future science group

resulting in shortened thermocycling times [33]. A similar argument can be made for velocity sed-imentation of blood samples within microfluidic structures. In the traditional tube format, the blood sample volume is processed in a tall con-figuration with a large radial span. By designing the chamber to be more wide and flat, as is typi-cal in CD-based devices, the separation distance of the blood cells can be decreased (Figure 4). For optimum performance of the device, due to minimized sedimentation distance, the cham-bers would be much wider than taller, but the disc was designed to also allow multiplexing of five samples on a single disc. The increased width when compared with that of a tube, along with the flat shape of the chamber, was sufficient in accelerating the sedimentation process.

As described by Stokes’ Law, the sedimentation rate of a particle, ν

r, due to centrifugal force is:

vr dtdr

18 rr d r

m

p m 2 2

= = -$

$ $ $

h

t t s

^

^^

h

hh

equation 2

where ρp is the particle density, ρ

m(r) is the media

density at a given r, d is the diameter of the par-ticle, r is the radial distance from the rotation center, ω is the angular velocity and η

m(r) is the

media viscosity at a given radius [34].

From equation 2, it is clear that the sedi-mentation rate is highly dependent on particle diameter, radial position and angular velocity. Importantly, for a given sedimentation rate, the shorter the distance a particle must travel, the less time it will take to travel that distance, thus making for a faster sedimentation process. This established relationship explains how the simple action of decreasing separation distance, by moving from a tube design to a flat cham-ber design, is crucial in reducing the separation time needed.

�� Challenges of sample-driven systemsWhile continuous flow systems constitute an attractive solution for processing large volumes with microfluidic devices, the existing separa-tion mechanisms are not well suited for rapidly handling samples of high hematocrit values. Kersaudy-Kerhoas et al. recently published a summary chart of microfluidic plasma-sepa-ration devices, comparing the achievable flow rates with several factors, including starting hematocrit percentage and plasma purity [12]. At a hematocrit value of 45%, the average value of human blood, the plasma separation efficiency drops to the point at which PCR is greatly inhibited by the cellular components

100

99.999

99.998

99.997

99.996

99.995

99.994

99.993Tube

20 minCD

10 minTube

10 minTube 5 min

CD5 min

CD2.5 min

Pla

sma

pu

rity

, as

per

cen

tag

e o

f b

loo

dce

lls r

emov

ed f

rom

inp

ut

sam

ple

Figure 6. Purity comparison graph between commercial plasma preparation tubes and the large-volume centrifugal microfluidic device. For the same duration of sedimentation time, the compact disk-based device yielded separated plasma of equal or greater purity than the tube format. Error bars represent one standard deviation from three samples (n = 3). CD: Compact disc.

Large-volume centrifugal microfluidic device for blood plasma separation | reSearch article

www.future-science.com 1709future science group

Executive summary

�� A multilayered centrifugal disc device has been developed for automated blood–plasma separation of large-volume raw blood samples.

�� In biological validation experiments, the device was shown to yield separated plasma of high purity (>99.99%) in less time than the commercial tube format.

�� Design challenges in the development of large-volume devices were discussed with an outlook on future blood-based sample-to-answer micro-total ana lysis systems devices.

of blood [35]. Only the centrifuge-based and CD-based methods – both batch processes – were shown to sufficiently separate raw blood samples. To use the various other microfluidic devices that require hematocrit levels ranging from 3 to 25%, the raw blood sample needs to be diluted with buffer. This extra dilution step not only requires longer device run times, but the target molecule concentration is also diluted. An additional step of target enrichment or solid-phase capture is then needed before the target can be detected via nucleic acid ana lysis.

One approach to batch centrifugation is to implement a modular-cartridge system, where cartridges can be designed for sample volumes in the milliliter range, and can be directly attached atop the disposable centrifugal device. In this sample-driven approach, the device por-tion containing the downstream amplification and detection mechanisms would remain con-stant, and the additional cartridge selection would depend on the sample volume (microliter to milliliter) and sample matrix (blood, saliva or urine).

Conclusion & future workThis article presented a large-volume blood plasma separation device capable of process-ing 2 ml of a raw blood sample. This prototype device was fabricated using traditional plastics machining techniques, with an extension to low-cost manufacturing techniques (i.e., injec-tion molding). Experimental results show that within 2.5 min, this novel CD-based device resulted in plasma samples of equal or greater purity when compared with traditional tube format. The relatively rapid blood separation is likely due to the increased surface area at the blood–plasma interface and minimized sedimentation distance, a consequence of the move from tubes to flat substrates. The CD-based device presented here illustrates not only the motivation and rationale for sample-driven systems, but some of the challenges in designing microfluidic systems to process large sample volumes.

In future work, we will perform in-depth analyses of the finger-like structural design. In particular, we will need to investigate the effect of various finger shapes and layouts on the structural integrity of these multilayer devices. In addition, we will integrate the large-volume sample preparation device with analyte/patho-gen enrichment, in an effort to develop an auto-mated CD-based device for nucleic acid ana lysis or other blood-based ana lysis.

Future perspectiveThe field of microfluidics holds much potential, and medical device and pharmaceutical compa-nies are beginning to utilize some of the tech-nologies developed with the use of miniaturiza-tion science. In moving towards point-of-care devices, there needs to be a heightened emphasis on the sample collection and preparation aspects so that true sample-to-answer µTAS devices can be realized.

Ethical conduct of research The authors state that they have obtained appropriate insti­tutional review board approval or have followed the princi­ples outlined in the Declaration of Helsinki for all human or animal  experimental  investigations.  In addition,  for investi gations involving human subjects, informed consent has been obtained from the participants involved.

AcknowledgementsThe authors would like to thank Dhivya Sridhar and Will Southard for their help in the fabrication and testing of the devices. Many thanks also to Jim Zoval, whose support was integral to the shaping of this project.

Financial & competing interests disclosureThe authors gratefully acknowledge financial  support  for this research from the DARPA MF3 center at University of California, Irvine, CA, USA. The authors have no other relevant  affiliations  or financial  involvement with  any organization or entity with a financial interest in or finan­cial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed. 

No writing assistance was utilized in the production of this manuscript. 

reSearch article | Amasia & Madou

Bioanalysis (2010) 2(10)1710 future science group

BibliographyPapers of special note have been highlighted as:� of interest�� of considerable interest

1 Duffy D, Gillis H, Lin J, Sheppard N Jr, Kellogg G. Microfabricated centrifugal microfluidic systems: characterization and multiple enzymatic assays. Anal Chem. 71(20), 4669–4678 (1999).

2 McCreedy T. Fabrication techniques and materials commonly used for the production of microreactors and micro total analytical systems. Trends Anal. Chem. 19(6), 396–401 (2000).

3 Whitesides GM. The origins and the future of microfluidics. Nature 442(7101), 368–373 (2006).

4 Madou M. Fundamentals of Microfabrication: The Science of Miniaturization (2nd Edition). CRC Press, Boca Raton, FL, USA (2002).

5 Beebe DJ, Mensing GA, Walker GM. Physics and applications of microfluidics in biology. Annu. Rev. Biomed. Eng 4(1), 261–286 (2002).

6 Siegrist J, Peytavi R, Madou M. Microfluidics for biological ana lysis: triumphs and hurdles of CD platforms. IVD Technology 111–127 (2009).

7 Toner M, Irimia D. Blood-on-a-chip. Annu. Rev. Biomed. Eng 7, 77–103 (2005).

8 Gascoyne P, Vykoukal J. Particle separation by dielectrophoresis. Electrophoresis 23(13), 1973 (2002).

9 Nakashima Y, Hata S, Yasuda T. Blood plasma separation and extraction from a minute amount of blood using dielectrophoretic and capillary forces. Sens Actuators B Chem. 145(1), 561–569 (2010).

10 Weigl B, Yager P. Microfluidics: microfluidic diffusion-based separation and detection. Science 283(5400), 346 (1999).

11 Crowley T, Pizziconi V. Isolation of plasma from whole blood using planar microfilters for lab-on-a-chip applications. Lab Chip 5(9), 922–929 (2005).

12 Kersaudy-Kerhoas M, Dhariwal R, Desmulliez M, Jouvet L. Hydrodynamic blood–plasma separation in microfluidic channels. Microfluid. Nanofluidics 8(1), 105–114 (2010).

�� Good overview and comparison of microfluidic devices for blood separation.

13 Yang S, Undar A, Zahn J. A microfluidic device for continuous, real time blood plasma separation. Lab Chip 6(7), 871–880 (2006).

14 Rodríguez-Villarreal AI, Arundell M, Carmona M, Samitier J. High flow rate microfluidic device for blood plasma separation using a range of temperatures. Lab Chip 10(2), 211–219 (2010).

15 Blattert C, Jurischka R, Schoth A, Kerth P, Menz W. Separation of blood cells and plasma in microchannel bend structures. Proc. SPIE 5591, 143 (2004).

16 Chen X, Cui D, Liu C, Li H. Microfluidic chip for blood cell separation and collection based on crossflow filtration. Sens  Actuators B Chem. 130(1), 216–221 (2008).

17 Nilsson A, Petersson F, Jönsson H, Laurell T. Acoustic control of suspended particles in micro fluidic chips. Lab Chip 4(2), 131–135 (2004).

18 Haeberle S, Brenner T, Zengerle R, Ducrée J. Centrifugal extraction of plasma from whole blood on a rotating disk. Lab Chip 6(6), 776–781 (2006).

19 Schembri C, Burd T, Kopf-Sill A, Shea L, Braynin B. Centrifugation and capillarity integrated into a multiple analyte whole blood analyser. J. Automat. Chem. 17, 99–99 (1995).

20 Madou M, Zoval J, Jia G et al. Lab on a CD. Annu. Rev. Biomed. Eng 8, 601–628 (2006).

21 Ducrée J, Haeberle S, Lutz S et al. The centrifugal microfluidic bio-disk platform. J. Micromech. Microeng 17(7), S103–S115 (2007).

22 Steigert J, Brenner T, Grumann M et al. Integrated siphon-based metering and sedimentation of whole blood on a hydrophilic lab-on-a-disk. Biomedical Microdevices. 9(5), 675–679 (2007).

23 Zhang J, Guo Q, Liu M, Yang J. A lab-on-CD prototype for high-speed blood separation. J. Micromech. Microeng 18(12), 125025 (2008).

24 Duffy DC, Gillis HL, Lin J, Sheppard NF, Kellogg GJ. Microfabricated centrifugal microfluidic systems: characterization and multiple enzymatic essays. Anal. Chem. 71(20), 4669–4678 (1999).

25 Gorkin R, Park J, Siegrist J, et al. Centrifugal microfluidics for biomedical applications. Lab Chip 10, 1758–1773 (2010).

��� Recently published critical review of centrifugal microfluidics, including academic and commercial research in this field.

26 Haeberle S, Zengerle R. Microfluidic platforms for lab-on-a-chip applications. Lab Chip 7(9), 1094–1110 (2007).

27 Chen JM, Huang P, Lin M. Analysis and experiment of capillary valves for microfluidics on a rotating disk. Microfluid. Nanofluidics 4(5), 427–437 (2007).

28 Madou M, Kellogg G. LabCD: a centrifuge-based microfluidic platform for diagnostics. Proc. SPIE 3259, 80–93(1998).

29 Siegrist J, Gorkin R, Clime L et al. Serial siphon valving for centrifugal microfluidic platforms. Microfluid. Nanofluidics 9, 55–63 (2009).

30 Park J, Cho Y, Lee B, Lee J, Ko C. Multifunctional microvalves control by optical illumination on nanoheaters and its application in centrifugal microfluidic devices. Lab Chip 7(5), 557–564 (2007).

31 Lee BS, Lee J, Park J et al. A fully automated immunoassay from whole blood on a disc. Lab Chip 9(11), 1548–1555 (2009).

��� Excellent example of an automated centrifugal device for blood separation and immunoassay detection.

32 Kroll M, Elin R. Interference with clinical-laboratory analyses. Clin. Chem. 40(11 Pt 1), 1996 (1994).

33 Jia G, Siegrist J, Deng C et al. A low-cost, disposable card for rapid polymerase chain reaction. Colloids Surf. B Biointerfaces 58(1), 52–60 (2007).

34 Pretlow TG, Pretlow TP, Cheret A. Cell Separation: Methods and Selected Applications. Academic Press, NY, USA (1987).

35 Kersaudy-Kerhoas M, Kavanagh DM, Dhariwal RS, Campbell CJ, Desmulliez MP. Validation of a blood–plasma separation system by biomarker detection. Lab Chip 10, 1587–1595 (2010).

�� Websites101 Abaxis Medical Diagnostics

www.abaxis.com/medical/piccolo.html

102 Gyros AB www.gyros.com/en/products/gyrolab_bioaffy_cds/index.html