Particle/Cell Separation on Microfluidic Platforms Based on Centrifugation

Microfuid Nanofuid (2017) 21:102 DOI 10.1007/s10404-017-1933-4

REVIEW

Particle/cell separation on microfuidic platforms based on centrifugation effect: a review

Wisam Al‑Faqheri1 · Tzer Hwai Gilbert Thio2 · Mohammad Ameen Qasaimeh3 · Andreas Dietzel4 · Marc Madou5 · Ala’aldeen Al‑Halhouli1

Received: 5 December 2016 / Accepted: 6 May 2017 © Springer-Verlag Berlin Heidelberg 2017

Abstract Particle/cell separation in heterogeneous mix- space on the platform is the main disadvantage, especially tures including biological samples is a standard sample when high sample volume is required. On the other hand, preparation step for various biomedical assays. A wide inertial microfuidics (spiral and multi-orifce) showed range of microfuidic-based methods have been pro- various advantages such as simple design and fabrication, posed for particle/cell sorting and isolation. Two promis- the ability to process large sample volume, high through- ing microfuidic platforms for this task are microfuidic put, high recovery rate, and the ability for multiplexing for chips and centrifugal microfuidic disks. In this review, we improved performance. However, the utilization of syringe focus on particle/cell isolation methods that are based on pump can reduce the portability options of the platform. In liquid centrifugation phenomena. Under this category, we conclusion, the requirement of each application should be reviewed particle/cell sorting methods which have been carefully considered prior to platform selection. performed on centrifugal microfuidic platforms, and inertial microfuidic platforms that contain spiral channels and Keywords Microfuidic platforms · Cells separation · multi-orifce channels. All of these platforms implement Particle separation · Lab-on-a-chip · Lab-on-a-disk · a form of centrifuge-based particle/cell separation: either Centrifugal effect physical platform centrifugation in the case of centrifugal microfuidic platforms or liquid centrifugation due to Dean drag force in the case of inertial microfuidics. Cen- 1 Introduction trifugal microfuidic platforms are suitable for cases where the preparation step of a raw sample is required to be inte- Particle and cell isolation and analysis have drawn more grated on the same platform. However, the limited available and more interest due to its perceived importance in many different felds and applications (Sajeesh and Sen 2014). In biomedical and biological research, the ability to iso- * Ala’aldeen Al‑Halhouli late specifc particles and cells from a heterogeneous back- [email protected] ground is considered a key tool for the study of individual cells or particles. This technology has resulted in great dis- 1 NanoLab, School of Applied Technical Sciences, German Jordanian University, Amman, Jordan coveries in cell biology and could be utilized to precisely predict a patient’s health status (Tomlinson et al. 2013; 2 Faculty of Engineering and Quantity Surveying, INTI International University, 71800 Nilai, Negeri Sembilan, Sajeesh and Sen 2014). A case in point is the development Malaysia of methods to isolate circulating tumor cells (CTCs). CTCs 3 Division of Engineering, New York University Abu Dhabi, are rare cells that originate from primary cancer tumors and Abu Dhabi, United Arab Emirates travel through the blood to other sites. This is how cancer 4 Institut fur Mikrotechnik, Technische Universität spreads to secondary sites and organs in the body. CTCs Braunschweig, Braunschweig, Germany were frst observed in 1869 by Ashworth (1869) and have 5 Department of Biomedical Engineering, University become a topic of keen interest again in the mid 1990’s. of California Irvine, Irvine, CA, USA The isolation of CTCs can be called “liquid biopsy” as it

1 3 102 Page 2 of 23 Microfuid Nanofuid (2017) 21:102 refects important and early information about tumor status is focused only on particle/cell isolation methods that and progress. Other than in the feld of cell biology, the use are based on centrifugation approaches. Centrifugation of particle/cell separation has been applied to other appli- approaches presented include methods that utilize the phys- cations including the separation of magnetic particles from ical centrifugation process on microfuidic platforms, and a background mixture (Kirby et al. 2012), separation and the approaches that utilize liquid centrifugation effects due observation of bacteria from host blood cells (Hou et al. to Dean effect for cell/particle separation. The platforms 2015), malaria enrichment (Warkiani et al. 2015b), and reviewed include the microfuidic CD, and platforms with separation of DNA (Zhao et al. 2015). spiral microfuidics and multi-orifce microfuidics. For a The demand for compact, inexpensive, disposable, and clear view of the categories, please refer to Fig. 1. high throughput devices that can be implemented for particle/cell separation has led many researchers to consider the various advantages of microfuidic platforms (Burger et al. 2 Centrifugal‑base microfuidics for particle/cell 2012a). The ability of microfuidic platforms to manipu- isolation late small amounts of fuid (on the scale of microliters and below) in micromachined channels has signifcantly The frst step in almost any particle/cell-based research is decreased the required volumes of samples and reagents. the isolation and purifcation of the particles/cells from raw Although microfuidic devices do feature many advantages, samples such as whole blood or serum. This frst step is an a challenge comes about when large sample volumes must important one that needs to produce results in high purity, be processed (few milliliters) to increase the chance of iso- high recovery rates, and high viability numbers (please lating rare cells such as CTCs (Amasia and Madou 2010). refer to the defnitions of these three terms in Sect. 1). Vari- Furthermore, in most cases some type of enrichment/con- ous techniques have been implemented to perform particle/ centration step is often required. cell isolation and purifcation on microfuidic platforms. In Various microfuidic methods and procedures have been this review, these methods and techniques will be presented developed for the isolation of targeted particles and/or cells and the advantages and disadvantages of each method will based on their unique characteristics such as geometry, be highlighted. According to an extensive review for cur- physical, chemical, or genetic properties. The proposed rently available isolation methods, both centrifugal micro- microfuidic platforms fall under one of two categories fuidic platforms and inertial microfuidic platforms have according to the operation mechanism: stationary micro- implemented some form of liquid centrifugation/spinning fuidics, also known as lab-on-chip (LOC) or centrifu- phenomenon to separate particles/cells from heterogeneous gal microfuidics, more commonly known as lab-on-disk backgrounds. For the centrifugal microfuidic platform, the (LOD). Under the LOD category, different passive and whole platform is spun with computer-controlled motors to active methods were proposed for particle/cell sorting and perform different separation methods. On the other hand, isolation. On the other hand, LOC methods mainly rely inertial microfuidic platforms utilize special structure of on the inertial effect for particle/cell manipulation. These microchannel such as curved channels or channels with methods have the advantages of high throughput, simple sudden expansion–contraction arrays to generate secondary design, easy operation, and passive activation without the centrifugation fow to separate particles/cells based on their need for external force for particle/cell manipulation. The physical properties. different isolation methods are usually evaluated by crite- The next section of this paper is divided into two main ria such as purity, recovery rate, and viability (Tomlinson sections: centrifugal microfuidic platforms and inertial et al. 2013). Purity represents the degree of contamination microfuidic platforms (see Fig. 1). In the frst section, of the recovered targeted cells with unwanted background the separation methods are classifed into passive meth- cells/particles. Recovery rate refects the ratio of number of ods (separation without the need for an additional external recovered targeted cells to the number of cells in the origi- force) and active methods (separation using an external nal sample, and viability is the number of recovered live force). In the following section, i.e., inertial microfuidics and heathy cells, which is very important for subsequent methods, the methods are all passive, and the classifcation cell analysis stages. Therefore, ideal cell isolation methods is based only on channel structure. The methods rely on should have short processing time, with high recovery and either spiral or multi-orifce channels to execute particles/ purity rates that result in viable cells. cells separation. Many recent detailed reviews are available that high- light available methods for particle/cell isolation with dif- 2.1 Centrifugal microfuidic platforms ferent types of microfuidic platforms (Warkiani et al. 2015c; Hyun and Jung 2014; Tomlinson et al. 2013; Burger Centrifugal microfuidic platforms are circular-shaped et al. 2012a; Sajeesh and Sen 2014). However, this review platforms containing a network of microchannels and

1 3 Microfuid Nanofuid (2017) 21:102 Page 3 of 23 102

Fig. 1 Cell separation methods on microfuidic platform based on License, and multi-orifce fgure is adopted from Lee et al. (2009a) centrifugation effect. Spiral fgure is adopted from Warkiani et al. with permission from AIP Publishing LLC (2015a) under a Creative Commons Attribution 4.0 International

Fig. 2 Centrifugal microfuidic platform preloaded with a volume of liquid in a straight microfuidic channel. The fgure highlights the main forces acting on the loaded liquid, i.e., centrifugal force (pushing liquid toward the outer edge of the platform), capillary force (acting against liquid fow), Coriolis force (perpendicular to the liquid fow and opposite to rotation direction), and Euler force (perpendicular to the CD rotation direction)

1 3 102 Page 4 of 23 Microfuid Nanofuid (2017) 21:102

chambers. Figure 2 shows a simple design of the platform the capillary pressure, Pcap. Most parameters that affect with a single straight microchannel leading away from the Pcen and Pcap are constants based on valve location on center of the platform toward a chamber. As the platform the CD, and also on fuid and CD material properties. resembles a compact disk (CD), it is commonly referred to The only parameter that can be freely adjusted to disrupt as the centrifugal microfuidic CD. the equilibrium between Pcen and Pcap is the CD spinning The main fuidic propulsion force on the CD is the speed. To design a network of microchannels and valves centrifugal force resulting from the spinning of the CD that allow for a series of fuid movement on the CD, the (Madou et al. 2006; Burger et al. 2012a; Strohmeier et al. minimum CD spinning speed where the centrifugal pres- 2015a; Smith et al. 2016; Tang et al. 2016). As the CD is sure exceeds the capillary pressure for each valve needs spun, the centrifugal force applies outward pressure on fu- to be determined. This CD spinning speed is commonly ids in the CD and propels fuids toward the CD edge. The referred to as the burst frequency and is usually presented centrifugal force per unit volume and the resulting pressure in revolutions per minute (rpm) as follows (Thio et al. can be determined as follows (Madou et al. 2006; Ducrée 2013): et al. 2007): Pcap 30 F = ρω2r (1) rpm = (5) cen ρ�rr π P = 2 rr cen ρω � (2) On the CD, apart from the centrifugal force, there is where ρ is the density of preloaded fuid, ω is the CD spin- also the Coriolis, the Euler, and a secondary centrifugal ning speed in radians per second (rad/s), Δr is the differ- forces acting on fuids in the platform. The Coriolis force ence between the top and bottom levels of a volume of fuid is perpendicular to the velocity of the moving fuid and is relative to the CD center, and r¯ is the average distance of frequently used for switching the lateral fow direction on the column of fuid from the CD center. the CD, and also for density-based particle separation and Under the centrifugal pressure, the average velocity of sorting (Burger and Ducrée 2012; Kim et al. 2008; Bren- fuid on the CD can be determined as follows (Smith et al. ner et al. 2005). The Coriolis force per unit volume can be 2016): determined as follows (Ducrée et al. 2007): D2P F = 2ρωU U = h cen corr (6) L (3) 32µ The direction of the Euler force is opposite of the CD where Dh is the hydraulic diameter of the channel or cham- spinning acceleration and perpendicular to the centrifugal ber the fuid is fowing in, μ is the viscosity of the fuid, force and can be used to create lateral motion of fuid dur- and L is the length of the column of fuid on the disk. ing disk acceleration for mixing applications (Smith et al. To prevent all fuids from running toward the CD edge 2016). The Euler force per unit volume can be determined without any order or sequence, specially designed passive as follows (Ducrée et al. 2007): valves are placed along channel paths (Thio et al. 2013). dω On hydrophilic CDs, the most commonly employed pas- F = ρr eur dt (7) sive valve is the capillary valve. A capillary valve is char- acterized by any sudden opening along the microchannel, The secondary centrifugal force is only present in curved such as the microchannel opening to a chamber shown in microchannels and supplements the main centrifugal force. Fig. 2. As fuid fows under the effect of centrifugal force This secondary force can be used for particle separation toward the boundary of a capillary valve, the capillary force and can be determined as follows (Zhang et al. 2008): applies opposing pressure on the fuid. This capillary pres- F = ρu2/R sure can be determined with the following mathematical sec (8) expression (Thio et al. 2013): where u is the velocity of the fuid fow in the channel and R is the radius of the curvature of the curved channel. 4 cos θ γ P = c la Many advantages especially for biological sample han- cap D (4) h dling and cell manipulation were demonstrated on the where θc is the fuid to surface contact angle and γla is the centrifugal platform (Burger et al. 2012a). Three of these fuid–air surface energy. advantages that apply to cell manipulation are listed here: When centrifugal and capillary pressures are at equilibrium, fuid is stopped at the boundary of the micro- 1. Simple implementation of sedimentation for sample channel and valve. For any fuid to fow past any capil- separation and cell enrichment (concentration) by uti- lary valve, the centrifugal pressure, Pcen must overcome lizing the centrifugation process (Smith et al. 2016).

1 3 Microfuid Nanofuid (2017) 21:102 Page 5 of 23 102

2. Centrifugation pumping is highly independent of sam- 2010). This confguration of the sedimentation chamber ple properties such as viscosity, electrical conductivity, forces the higher density portion of the blood (blood cells) and pH, which is ideal for handling biological sam- to settle at the bottom of the chamber during centrifugation. ples such as blood (Madou et al. 2006; Warkiani et al. The lighter portion of the blood (plasma) stays at the top of 2015c). the blood cells layer. Kim et al. (2013) and Kinahan et al. 3. Clear separation between the microfuidic networks (2014b) investigated the effect of the sedimentation cham- and the pumping/detection components prevents sam- ber geometry on the purity, and the time it takes to perform ple contamination and improves the disposability of the plasma separation (see Fig. 3). Kim et al. (2013) found that platform (Burger et al. 2012a; Warkiani et al. 2015c). by narrowing the sedimentation channel and enlarging the tilt angle of that channel with respect to the radial direction, In the following two sections, the main passive and the separation process could be sped up to 8 times faster. active methods for particle/cell separation on the centrifu- This can be explained by Boycott’s discovery in 1920 when gal microfuidic platforms are discussed. The main pro- he found that particles sediment faster in tilted chambers posed methods are summarized in Table 1 with their uti- due to the larger available surface area for particles to settle lized methods, processed samples, and reported results. on (the side wall and base of the chamber) (Boycott 1920). Another reason the sedimentation process will be faster is 2.1.1 Passive particle and cell separation methods due to the shorter distance that the particles/cells need to travel to reach chamber wall compared to the chamber base. Many passive techniques to perform particle/cell isola- This is similar in narrower chambers where the side walls tion on the centrifugal microfuidic platform have been will be closer to each other. Figure 3a–c shows the differ- reported. In several of these reports, centrifugal microfu- ence between straight and tilted sedimentation channels. idic platforms were proposed as an alternative to conven- Kinahan et al. (2014b) proposed a curved design of the sedi- tional benchtop centrifugation devices to separate blood mentation chambers using a logarithmic spiral or mirabilis components (Riegger et al. 2007; Steigert et al. 2006, design (see Fig. 3d–f). The authors claimed that this design 2007; Haeberle et al. 2006; Lee et al. 2011a; Zhang et al. can improve the separation process speed by 39% compared 2008; Kim et al. 2013; Kinahan et al. 2014b; Park et al. to straight sedimentation chambers. This is due to the fact 2014; Amasia and Madou 2010). The various techniques that spiral designs provide not only more surface area for can be categorized into three types (1) sedimentation- sedimentation (similar to a tilted chamber), and there is also or density-based blood fractionation (Park et al. 2008, constant centrifugal force applied along the entire channel 2014; Kinahan et al. 2014a, b; Nwankire et al. 2015b; length compared to varying centrifugal force along tilted Burger et al. 2012a; Haeberle et al. 2006; Li et al. 2010; chambers. Kim et al. 2013; Steigert et al. 2007; Strohmeier et al. In separate studies, Li et al. (2010) discussed that 2015b), (2) cells/particles separation based on physical after plasma separation is performed using the centrifuge properties, and (3) separation based on immunoaffnity method, blood cells may diffuse into the already separated processes. plasma when the spinning process is stopped (at the end of the centrifuge process). To overcome this problem, the 2.1.1.1 Density‑based blood fractionation (sedimenta‑ authors proposed out-of-plane microvalves and triangular tion) The importance of blood as the biological sample for obstacle structures (TOS). Zhao et al. (2015) proposed a most diagnostic assays, and the need for portable devices different passive method of using stored pneumatic energy with high throughput to handle blood samples is the main for particles/liquid sedimentation, resuspension, and trans- driver behind the high number of research articles in this port on a centrifugal microfuidic platform. The proposed area. In particular, one of the most investigated blood sepa- method is independent of particle size and does not require ration methods on centrifugal microfuidic platforms is the any external force or special coating of the microfuidic isolation of the plasma portion from the rest of the blood cells structure. The whole process is purely controlled by the portion (Strohmeier et al. 2015b). According to Strohmeier spinning speed of the platform. This method is proposed et al. (2015b), the plasma separation process can be divided to improve particle-based assays such as Deoxyribonucleic into two main steps: (1) cell sedimentation by centrifuga- acid (DNA) extraction and immunoassays. tion and (2) plasma extraction from the layer-separated sam- After the sedimentation process is completed, purifed ple. The plasma separation step is usually performed in a plasma decantation or extraction and subsequent steps are sedimentation chamber or in another microstructure that is carried out. The extraction step is usually accomplished by positioned radially on the centrifugal microfuidic platform utilizing a siphon channel (Steigert et al. 2007; Strohmeier (Park et al. 2008, 2014; Kinahan et al. 2014a; Nwankire et al. et al. 2015b; Nwankire et al. 2015b), or a straight channel, 2015b; Burger et al. 2012a; Haeberle et al. 2006; Li et al. the latter is controlled by a valve (Park et al. 2008, 2014;

1 3 102 Page 6 of 23 Microfuid Nanofuid (2017) 21:102 centrifugal platform for improved platform for improved centrifugal fltration effciency two unit operations: micromixer unit operations: micromixer two (for binding process), and inertial (for focusing process) fow noaffnity and DGM for CTCs capturing and separation fltration only. Staining cells can fltration only. require 1 h ute different cluster size of cancer ute different preprocessed to cells. Sample was other type of cells (white/ extract red blood cells) direct extraction of plasma without direct extraction the need for siphoning process Spiral camber improves process - Spiral camber improves ing time up to 39% compared straight tilted chambers – Carbon-electrode is integrated on Carbon-electrode is integrated The proposed platform integrates The proposed platform integrates – The proposed setup integrates immu - The proposed setup integrates – – 30 s is the processing time for cells This method only utilized to distrib - Low sample volume, but fast and fast but sample volume, Low – DGM is used in spiral chambers. Remarks 1020 rpm, BCE 90–96%, rpm, fow rate 0.01– 100–175 rpm, fow 0.03 mL/min, fltration effciency ~100% binding effciency, 98% separa - binding effciency, tion effciency incubation time, SS 1200 rpm, BCE = 99% rpm, Recovery and viabil - ~2300 rpm, Recovery ity >90% 3000–5000 rpm 1500–2000 rpm 1200–3600 rpm, 61% capture effciency 1550–1800 rpm, SP 96–99% 3800 rpm, SP 99% PSV 3 µL, PT 10 min, SS PSV 50 µL, PT 1–16 min, SS SS 225 rpm, 97.1% cells/beads PSV 2 µL, PT ~10 min + 90 PSV 5 mL, PT 78 min, SS PSV 24 µL, PT 30–40 s, SS PSV 4 µL, PT 40 min, SS 720 rpm PSV 5–20 µL, PT 300–720 s, SS PSV up to 3 mL, PT 20–30 s, SS PT 30–180 s, SS 3600 rpm PSV 0.5–1 µL, PT 1–6 s, SS Reported results PSV 2 mL, PT 2.5–10 min, SS particles and beads Jurkat cell, and NIH3T3 cell MCF-7 cancer cells/magnetic Yeast cells Yeast Cells (MCF7) HeLa, MCF7, RPMI-8226 cells, Cells (CTCs) , baker’s yeast, coli , baker’s Escherichia Cells (CTCs) Cells/particles Cells (CTCs) Blood (fractionation) Blood (fractionation) Sample Blood (fractionation) magnets (Siegrist et al. 2011 ; magnets (Siegrist Kirby et al. 2012 , 2015 ) electrodes (Martinez-Duarte et al. 2010 ) irre et al. 2015 ) 2012b ) medium (Park et al. 2014 ) medium (Park traps (Kubo et al. 2009 ; Furutani traps (Kubo et al. 2010 ) (CCE) (Morijiri et al. 2013 ) ber (Kim et al. 2013 ; Kinahan et al. 2014b ) ; Kuo and Chen and Li 2014 ; Kuo Kuo 2015 ) (Amasia and Madou 2010 ) integrated chamber + integrated Fork-like Straight channel + carbon-DEP + inertial sorter (Agu - Micromixer V-shaped traps (Burger et al. 2011 , traps (Burger V-shaped + density gradient Centrifugation Zig-Zag channel with u-shaped Serial gaps (Glynn et al. 2015 ) Serial gaps Counterfow centrifugal elutriation centrifugal Counterfow Filter (Lee et al. 2014 ) Tilted/spiral sedimentation cham - Tilted/spiral Curved channel (Zhang et al. 2008 ; Curved Structure Straight sedimentation chamber (DEP) Magnetic feld Dielectrophoresis Immunoaffnity Physical properties Physical Method Sedimentation Particle/cell separation methods on centrifugal microfuidic platforms separation methods on centrifugal Particle/cell Active methods Active 1 Table Category BCE bead capture effciency PT processing time, SS spinning speed, SP separation purity, PSV processed sample volume, Passive methods Passive

1 3 Microfuid Nanofuid (2017) 21:102 Page 7 of 23 102

Fig. 3 Isolation of plasma portion from the rest of blood based on bilis sedimentation chamber. a–c is adopted from Kim et al. (2013) centrifugation/sedimentation process; a–c effect of tilted sedimenta- with permission from Elsevier, and d–f Is adopted from Kinahan tion chamber on plasma separation, d–f effect of using a spiral mira- et al. (2014b) with permission from Elsevier

Kinahan et al. 2014a). The intersection of the extraction networks, the blood cells will be forced to fow closer to channel with the sedimentation chamber is positioned the outer wall of the microchannel which is nearer to the slightly above the interface between the layers of plasma CD edge, while plasma fows closer to the inner wall of and red blood cells (RBC). Compared to a straight channel the microchannel which is nearer to the CD center. This where an active valve is required, the siphon channel can is because blood cells have higher density, while plasma be passively actuated. When the spinning speed of the plat- has a lower density. At the fnal Y channel junction, the form is decreased, the hydrophilic property of the siphon two parallel fowing layers of blood cells and plasma will channel pulls the plasma into the channel and through the fow to separate collection chambers (see Fig. 4). The crest of the channel. Increasing the spinning speed then authors reported a separation effciency of 90% for blood push the plasma into the collection chamber. with 5% hematocrit level, while separation effciency In different mechanism for plasma/cells separa- drops to 65% when whole blood is processed (48% tion, Zhang et al. (2008) proposed a simple microfuidic hematocrit). (Kuo and Li 2014; Kuo and Chen 2015) uti- design to separate blood plasma from the other compo- lized the same basic design and concept by Zhang et al. nents of the blood. The proposed design consists of a (2008), but with an extended secondary microfuidic pro- short straight microchannel that leads to a curved micro- cess where the extracted plasma is mixed with specifc channel, followed by two collection reservoirs, one for reagent. Kuo and Li 2014; Kuo and Chen 2015 utilized plasma and the other for red blood cells (RBC) reservoir siphoning and mixing structures to perform prothrom- (see Fig. 4). In this system, the blood components will bin time (PT) tests. In their follow-up work, the authors also be under the effect of not just the centrifugal force implemented a slightly different design with decanting

(fω) and the Coriolis force (fC), but also the secondary chamber to perform creatinine test. The authors reported centrifugal force (fR) due to the curvature structure of the high separation effciency of 96% with a short process- microchannel. According to Zhang et al. (2008), the three ing time of 5–6 s, which is dramatically faster when com- forces contribute in different degree to the separation pared to conventional benchtop procedures. Compared to process. As blood sample fows through the microfuidic the earlier described sedimentation methods, the curved

1 3 102 Page 8 of 23 Microfuid Nanofuid (2017) 21:102

multi-step assays in automated fashion without the need for external interaction such as high trained clinical technician.

2.1.1.2 Cells/particles separation based on physical prop‑ erties In the previous section, we presented different microfuidic methods for isolating cells in general from blood plasma. Here, we present isolation methods for isolating specifc types of cells/particles from a relevant background population. These methods are usually implemented for the enrichment of specifc targeted cell type while dis- carding unwanted cells. Geometry-based methods for cells and particles isolation on centrifugal microfuidic platforms have been reported in various works. Among the proposed mechanisms are: (1) membrane flter (Lee et al. 2014), (2) flter- like microstructures such as obstacle arrays (Nwankire et al. 2015a) and scale-matched gaps (Glynn et al. 2015), (3) counterfow centrifugal elutriation (CCE) (Morijiri

Fig. 4 Blood fractionation using curved channel on centrifugal microfuidic platform is utilized to separate blood plasma from the other components of the blood, fω centrifugal force, fR centrifugal force due to channel curvature, fC Coriolis force

channel method does not require complex microfuidic design with siphon channels to extract plasma from the layered blood sample. In a later work by Shamloo et al. (2016), a curved channel with one initial container and three output containers was implemented for cells sorting from blood samples. In their work, a diluted blood sample containing a mixture of RBCs and neutrophils attached to activated magnetic beads is injected into the initial container. As the platform starts spinning, the effects of the centrifugal force, Coriolis force, Euler force, magnetic force and other forces correlated with the curvature structure will transfer the RBCs, bonded neutrophils, and free magnetic Fig. 5 Particle/cell separation based on its physical properties. a Cell beads into different destination containers. The authors separation using membrane flter, microfuidic CD contains three sets, reported 100% separation effciency in an optimum angu- each set with loading chamber, fltration chamber, and waste chamber lar velocity of 45 rad/s. (Lee et al. 2014). b A rail of gaps with increasing opening size is pro- In summary, passive sedimentation of plasma/cells is posed to measure the size distributions of CTCs clusters. a is adopted from Lee et al. (2014) with permission from American Chemical one of the most important features of the centrifugal micro- Society, b Glynn et al. (2015) under a Creative Commons Attribution fuidic platforms that allow for full integration of biological 4.0 International License

1 3 Microfuid Nanofuid (2017) 21:102 Page 9 of 23 102 et al. 2013) (4) microchannels with gaps-array (Kubo to fow near the chamber inlet while smaller particle tend et al. 2009; Furutani et al. 2010), (5) centrifugal deter- to fow toward the chamber outlet. The authors success- ministic lateral displacement (c-DLD) (Jiang et al. 2015), fully isolated polymer particles of 1, 3, and 5 µm in diam- density gradient free nanoparticles separation (Arosio eter, and also demonstrated the ability of separate blood et al. 2014), and separation based on density gradient cells (erythrocytes and leukocytes) from diluted blood sections (Moen et al. 2016). sample. To reach single-cell level analysis on centrifugal Lee et al. (2014) proposed a centrifugal platform for platforms, Kubo et al. (2009) proposed a zig-zag-shaped CTCs isolation with three processing sets, where each set microfuidic channel with U-shaped trapping cham- contains a sample loading chamber, a fltration chamber, bers lined along the sides of the channels. The proposed and a waste chamber (Fig. 5a). A commercially available platform was made of PDMS with a total of 24 micro- membrane flter with pore size of 8 µm was integrated channels, each with 530 trapping chambers. Various cell in the fltration chamber to trap the targeted CTCs. The types, i.e., Escherichia coli, baker’s yeast, Jurkat cell, platform is validated with MCF-7 breast cancer cell line and NIH3T3, were isolated on the proposed platform at spiked in phosphate-buffered saline (PBS) and blood 3000 rpm and 30 s processing time. In a separate study, samples collected from healthy donors. In addition, blood Jiang et al. (2015) utilized deterministic lateral displace- samples from patient with lung and gastric cancers were ment (DLD) on centrifugal platform for passive parti- tested for clinical validation. The process was reported cle isolation. On the platform, the authors integrated a to takes only 20 s to fltrate 3 mL of sample. However, square array of cylindrical posts that are in specifc tilting the postprocessing which includes the washing, block- angles with respect to the centrifugal force direction. By ing, incubation, staining, and cell analysis steps takes testing the effect of different tilting angles, the authors around 50 min. The authors reported high CTCs captur- found the best scenario when big particles are trapped by ing effciency of 84% at relatively low spinning speed arrays with small migration angles while small particles of 600 rpm. However, operating the platform at low are free to travel following the direction of the centrifugal spinning speed results in a high count of 3092 captured force. The integration of DLD on the centrifugal platform WBC. Meanwhile, at high spinning speed of 3600 rpm, improves the portability of the developed platform with the CTCs capturing effciency drops to 50% while the the elimination of external pumping methods and physi- WBC count drops signifcantly to 181 cells. Finally, the cal connections. This allows the integration of DLD as a authors validated their system experimentally by compar- preparation step for cell labeling and/or analysis on the ing it with ScreenCell®, the only commercially available platform. On a nanoparticle level, Arosio et al. (2014) microfuidic platform for CTCs isolation. In compari- proposed the installation of a simple curved channel with son, the capture effciencies for the proposed platform seven collection bins/chambers onto a centrifugal base. and ScreenCell system were 56 and 69%, respectively. The collection bins/chambers are placed in an orienta- The authors believe that the slightly higher effciency of tion perpendicular to the centrifugal force direction. As ScreenCell® system is due to (1) the smaller pore size of a result, the separation mechanism is based on the bal- the utilized membrane flter of 7.5 µm and (2) the imple- ance of the fuid drag force on one side, and the centrifu- mentation of dilution FC2 buffer which is known to stabi- gal and buoyancy forces on the other side. As the analyte lize the cells. This proposed microfuidic platform is the with dispersed nanoparticles fow through the separation only system which was validated by experimentally and channel, the centrifugal force will drive big and high den- compared to an commercially approved CTCs separa- sity particles to move laterally toward the outer wall of tion system. Figure 5b presents a novel design proposed the channel. Smaller and low density particles will expe- by Glynn et al. (2015) to measure the size distributions rience slower laterally fow speeds and will stay closer to of CTCs clusters in a blood sample. The authors imple- the channel center. Separation of 50, 100, and 200 nano- mented a rail of gaps with increasing opening size to sized particles, at 5000 rpm spinning speed and 5 min isolate different clusters based on size variation before processing time, was performed to demonstrate the capa- sending them to different destination chambers. Morijiri bility of the proposed system. et al. (2013) implemented the counterfow centrifugal In a more recent study, Moen et al. (2016) reproduced elutriation (CCE) method on a centrifugal microfuidic the traditional density media and relative centrifugal platform for particles sorting based on particle size. The forces (RCF) procedure for cell separation on the centrif- method relies on the balancing of the centrifugal force ugal microfuidic platform. Moen et al. (2016) proposed which pushes particles toward the platform edge, and liq- a single straight channel that divided into fve multiple uid drag force which pushes particles toward the platform sections. Each section has a different density media and center. As the net force is higher near the chamber inlet separated from the neighboring sections by passive capil- and lower near the chamber outlet, bigger particles tend lary valves. This multi-density setup forces different cells

1 3 102 Page 10 of 23 Microfuid Nanofuid (2017) 21:102 to sediment at different sections based on their respective et al. 2011, 2012b), and waved microchannel for inertial density. This is due to the different effect of the centrifu- sorting (Aguirre et al. 2015). gal force and liquid drag force on particles of different Park et al. (2014) from Samsung Biomedical Research densities. The authors reported 95.15% recovery rate for Institute developed a fully automated centrifugal micro- leukocytes while excluding 99.8% of red blood cells. fuidic platform for isolating CTCs from blood samples (see Fig. 6a). The platform designed consists of a blood 2.1.1.3 Separation based on or for immunoaffnity pro‑ chamber, a density gradient medium (DGM) chamber, cesses In general, immunoaffnity is defned as the uti- and a collection chamber. Inside the blood chamber, a tri- lization of surfaces or particles that are pre-activated with angular obstacle structure (TOS) is introduced to prevent specifc antigens/antibodies which only capture specifc blood cells from fowing backward during the plasma targeted cells/proteins. This trapping process facilitates evacuation process. The proposed platform has the abil- the isolation and sorting of the targeted cells/proteins from ity to handle relatively high volume of fresh blood, up heterogeneous background mixtures. In most proposed to 5 mL, to perform CTCs isolation without any pre- sorting platforms, immunoaffnity is performed together processing or external intervention. The process starts with other sorting mechanisms such as density gradient with the introduction of 5 mL of blood sample mixed medium (DGM) (Park et al. 2014), capture-array (Burger with 100 µL of 4.5-μm-diameter superparamagnetic

Fig. 6 Immunoaffnity on centrifugal platform for particle/cell sepa- V-cup array, and waste chamber, c micromixer and inertial sorter ration a microfuidic design is proposed for CTCs isolation from are integrated for cancer cells separation based on immunoaffnity. blood sample, the design consist of blood chamber, DGM chamber, a is adopted from Park et al. (2014) with permission from American collection chamber, and waste chamber, b microfuidic design with Chemical Society, b is adopted from Burger et al. (2012b) with per- V-shaped traps array is proposed for bead-based assays, the design mission from Royal Society of Chemistry, c is adopted from Aguirre consist of cell reservoir, washing buffer reservoir, IgG reservoir, et al. (2015) with permission from Springer

1 3 Microfuid Nanofuid (2017) 21:102 Page 11 of 23 102 activated microbeads used for CTCs trapping. Then, the a background mixture of yeast and latex cells with 100% platform is spun for few minute to separate plasma from separation effciency. In separate researches by Kirby et al. the rest of the blood. Plasma is then transferred to the (2012, 2015) and Siegrist et al. (2011), magnetic feld was waste chamber, and CTCs is left to bind with the acti- imposed on the centrifugal platform for cells manipulation vated microbeads under a simple shaking process. After- in stopped-fow sedimentation mode. The presented sepa- ward, the mixture is released to reach the DGM layer ration methods contain common features such as a source where the CTC-bead complex moves down to reach the chamber, a focusing channel, and a fork-shaped separa- collection chamber. The authors reported a high recover tion chamber that splits different cell types into different rate of >95% with >90% cells viability. In separate stud- destination sub-chambers (see Fig. 7b). Under the infu- ies, Burger et al. (2011, 2012b) proposed a centrifugal ence of Stokes’ drag, centrifugal, and magnetic forces, microfuidic platform for effcient trapping and analysis different cells are guided into the different destination of particles, which can facilitate bead-based assays (see sub-chambers. Using functionalized magnetic beads with Fig. 6b). The operation principle of the platform is based anti-EpCAM antibodies, Kirby et al. (2015) reported the on stopped-fow sedimentation, where microparticles are isolation of rare MCF7 cancer cells from 3 µL of whole injected into the designated chamber that is flled with blood at a recovery rate of 90–96%. stagnant fuid. When the platform is spun, the resultant centrifugal force will force the randomly dispersed parti- 2.2 Inertial microfuidics cles to fow though the array of V-shaped traps. To demonstrate the ability of the proposed system, the authors 2.2.1 Spiral microfuidic platforms performed a single step antibody assay. A more recent work by Aguirre et al. (2015) reported Segre (1961) observed that particles with 1-mm diam- the integration of two operation units on the centrifugal eter are randomly dispersed when fowed through an microfuidic platform. The frst unit is a micromixer used arc-shaped pipe of 1-cm in diameter (Segre 1961; Segre to create breast cancer cells and beads complex MCF7-PS. and Silberberg 1962a; Segre and Silberberg 1962b). This The second unit is an inertial sorter used to isolate cancer- discovery has led various researchers to investigate this beads complex from the background mixture. The plat- phenomenon. form design is shown in Fig. 6c. The biological sample of In a later study by Karnis et al. (1966), it was shown MCF-7 DMEM culture media, and the anti-Epcam func- that large particles can be focused near the center of a + tionalized beads are frst injected through inlet 1 and 2. The straight channel in higher fow rate, leaving smaller platform is then spun at an angular velocity of 3.75 Hz to particles fowing near channel wall (see Fig. 8a). Parti- fow the biological sample and microbeads through the cle migration is caused by the effect of shear gradient micromixer. The proposed micromixer works based on the lift force (FLS) that attracts particles toward the chan- principle of secondary fow generated by Dean drag force, nel walls, and wall induced lift force (FLW) that repulses while cell sorting mechanism is based on lateral migration them back toward the middle of the channel (see Fig. 8a). effect. The authors reported mixing and recovery rate of 97.1 However, the effects of these two forces are dependent on and 98.7%, respectively. particle mass, and particles of different sizes settle at different equilibrium positions within the channel. 2.1.2 Active particles/cells separation methods In 2008, Di Carlo et al. (2007) and Bhagat et al. (2008) have separately shown that adding curvatures to the fow In contrast to passive separation methods which mainly channel can lead to secondary vortices, i.e., Dean vorti- depend on the microfuidic structure and particle physical ces that are perpendicular to the main fow stream (see properties for separation process, active separation meth- Fig. 8b). Curved channels cause velocity mismatches ods usually utilize external forces or actuators to facili- between liquid elements that are near channel walls and tate the separation process (Madou et al. 2006; Strohm- those that are closer to the channel center (Di Carlo 2009; eier et al. 2015a). In a study by Martinez-Duarte et al. Zhang et al. 2016). Liquid elements near the channel (2010), dielectrophoresis (DEP) was used on the centrifu- center have higher inertia while the elements near the gal microfuidic platform for active fltration and trapping channel walls are relatively stagnant. This velocity mis- of targeted cells (see Fig. 7a). The authors reported the match results in two symmetrical secondary fows that are fabrication and integration of 3D carbon electrodes that perpendicular to the liquid main fow. Therefore, particles resulted in high fltration effcacy with low fabrication in inertial microfuidics with spiral-shaped channel will cost. The carbon electrodes were supplied with a voltage follow these vortices in addition to the main fow. With supply of 200 kHz and 20 Vpp. At a sample fow rate of this discovery and by following certain adjustments to 35 µL/min, the authors successfully separated yeast from the dimension and shape, relatively large particles can be

1 3 102 Page 12 of 23 Microfuid Nanofuid (2017) 21:102

Fig. 7 Active cells/particles sorting on centrifugal microfuidic plat- tination chamber A, B, and C, the fgure also show the main forces form a dielectrophoresis (DEP) is integrated on the centrifugal micro- effect particles movement, i.e., centrifugal force and magnetic force. fuidic platform for selective particle/cell separation, slip rings and a is adopted from Martinez-Duarte et al. (2010) with permission from metal pads are utilized to provide the carbon electrodes with power Royal Society of Chemistry, b is adopted from Kirby et al. (2012) b particle/cell sorting utilizing magnetic feld, the design consist of with permission from Springer loading chamber, focusing channel, separation chamber, and des- focused near the channel inner wall while leaving smaller density, U is average velocity, µ is dynamic viscosity, and particles to fow closer to the outer wall of the channel H is the channel characteristic dimension). On the other (see Fig. 8b) (Warkiani et al. 2014, 2016). hand, inertial microfuidics operate in approximate Re In most typical cases, microfuidic platforms are oper- range from 1 up to 100, in other words between Stokes ated in the Stokes regime with negligible fuid inertia and regime and turbulent regime (Re ~2000) (Zhang et al. Reynolds number (Re) (where Re ρ UH/µ; ρ is fuid 2016). For this kind of platforms, the infuence of both = f f 1 3 Microfuid Nanofuid (2017) 21:102 Page 13 of 23 102

Fig. 8 Particle/cell migration in straight and curved channel (a) in back toward the center line of the channel (b) in spiral channel, the straight channel, shear gradient lift force (FLS) push particles toward balance between net inertial lift force (FL) and Dean drag force (FD) the channel walls, and wall induced lift force (FLW) that repulses them causes particle migration to equilibrium position inertia and viscosity of the fuid exist, causing particle be calculated by the following equation (Warkiani et al. migration during fuid fow in the microfuidic channel. 2016): The two effects on suspended particles: inertial migration U = × −4De1.63 and secondary fow are observed in this regime and are Dean 1.8 10 (10) related to fnite inertial forces (see Fig. 8). When a particle travels in one direction between two When randomly dispersed particles with different opposite channel walls, it is said to have completed half a sizes are injected and start fowing through a spiral, Dean Dean cycle (LDC) (Warkiani et al. 2016). A particle trave- vortices develop and the resulting drag forces cause the ling from one wall to the opposite wall and then returning particles to follow the direction of these vortices in addi- to the initial wall is considered to have completed a full tion to the main stream fow. LDC. LDC can be calculated as follows: Dean vortices strength are dependent on the Dean L = 2w + h number (De), Reynolds number (Re), and channel aspect DC (11) ratio (AR) (Zhang et al. 2016). Dean number is a dimen- where w is channel width and h is channel height. For sionless parameter which is a function of fow Reynolds particles to achieve an equilibrium position within a number, channel hydraulic diameter (DH), and the cur- channel, the channel must have a minimum height calcu- vature of spiral channel (RC), and it can by calculated by lated as follows: the following equation (Dean 1928): UF lC = LDC (12) UDean DH De = Re (9) 2RC where UF is the average fuid velocity. Finally, the Dean drag force can be calculated according Due to Dean vortices, particles are pushed back and to Stokes’ law as: forth between the side walls of the spiral. The velocity of FD = 3πµUDeanac (13) this lateral fow is called Dean velocity (UDean) and it can

1 3 102 Page 14 of 23 Microfuid Nanofuid (2017) 21:102

where ac is particle diameter. In curved channels, particles will be under the effect of both secondary fow due to Dean drag force, and inertial force (FL). Inertial force is the balance of shear gradient lift force (FLS) which push the particles toward channel walls, and wall induced lift force (FLW) that repulses particles back toward channel center (Bhagat et al. 2008; Di Carlo et al. 2007). The net inertial force cells) (cells), 90% (viability) 96% (particles), 77% (Algae 96% (particles), 77% (Algae ≥ 85% >90% >65% 80–90% 100% 85% recovery, >98% viability 85% recovery, Focusing only Focusing 90% (particles) and 80% Complete separation is reported is separation Complete (FL,nett) can be calculated as following (Bhagat et al. rate Recovery/separation 2010):

2ρU2a4 F = F c L,nett D2 (14) H

In spiral channels, particles fowing within a curved subsection will experience both lift force and drag force (which is caused by Dean vortices), leading to particle centrifuga- (~500 mL/min for 84 spiral chips) 0.4–4 mL/min 2.1 mL/min 6 mL/min for a single spiral 1.7 mL/min 1.7 mL/min ~1.5 mL/min 0.05 mL/min 0.1 mL/min or Re ∼ 20 mL/min fow rate, Re 14.4 ~3 mL/min fow 0.02 mL/min, Re ~10 tion (Zhang et al. 2016). Furthermore, since the channel rate Flow curvature direction of the spiral is either outward toward the platform edge, or inward toward the platform center, the Dean vortices spinning direction is constant within each subsection. In a spiral, the Dean drag force changes as the particles move from one point in the spiral to another due to a change in the spiral curvature [see Eq. (9)]. However, as the spirals are relatively small, the change is negligible. Algae cells Algae As a result, the inertial force keeps particles in a specifc roblastoma and glioma cells 5, 10, 20 µm particles, and Blood (CTCs) CHO and Yeast cells Yeast CHO and Blood (bacteria isolation) Blood (CTCs isolation) Blood (CTCs isolation) Blood (CTCs isolation) 6 µm particles 10, 15, 20 µm particles; neu - 7.32 and 1.0 µm particles cross-sectional equilibrium position and this position is Sample under the effect of Dean drag force according to particle size. In other words, the ratio of inertial force to the Dean drag force, R a3R /D3 determines the relative posi- f = c c H tion where particles of various sizes will gather within the channel. On a range of R from 0 to , Dean drag force is f ∞ dominant when Rf approaches 0, and this condition is true for particles with size much smaller than channel hydrau-

lic diameter. In this case, Dean force drives small particles W for CHO cells, while 450, H inner/outer 30/70 for yeast cells 350/100 500/170 W 600, H inner/outer 80/130 500/80 W 600, H inner/outer 80/130 500/160 500/160 100/50 500/130 100/50 closer to the channel outer wall that is closer to the plat- (µm) Width/height form edge. On the other hand, as R approaches , inertial f ∞ force is dominant for large particles with diameter similar to channel hydraulic diameter. Therefore, large particles are fowing close to the inner wall of the curved channel which is closer to the platform center. Spiral microfuidic platforms have been widely investigated by the Papautsky group (Bhagat et al. 2008, 2010;

Kuntaegowdanahalli et al. 2009) and Lim and Han groups together 3 2 with 3 stacked layers 2 with 3 stacked 8 with four spiral connected 2 8 2 with 3 stacked layers 2 with 3 stacked 2 10 5 5 Turns (Hou et al. 2013, 2015; Guan et al. 2013; Warkiani et al. 2014, 2015a, 2016) for particle/cell separation. A summary of the fabrication design, sample, applied fow rate, and reported results of the various platforms are summarized in Table 2. In two different works, Bhagat et al. (2008, 2010) from Papautsky group proposed sheath-less spiral channel for the separation of different size particles. The authors separation methods on spiral platforms Particle/cell developed 5-loop and 10-loop spiral channels which have a ( 2009 ) Schaap et al. ( 2016 ) Warkiani et al. ( 2016 ) Warkiani Warkiani et al. ( 2015a ) Warkiani Hou et al. ( 2015 ) Warkiani et al. ( 2014 ) Warkiani Warkiani et al. ( 2014 ) Warkiani Hou et al. ( 2013 ) Bhagat et al. ( 2010 ) Bhagat Kuntaegowdanahalli et al. Kuntaegowdanahalli Bhagat et al. ( 2008 ) Bhagat rectangular cross section of width 100 µm by height 50 µm. 2 Table First author

1 3 Microfuid Nanofuid (2017) 21:102 Page 15 of 23 102

The 5-loop design was used to separate 7.32 and 1.9 µm the need for culturing or enzymatic amplifcation. The particles, while the 10-loop design was implemented to authors reported that in a test for bacteria identifcation focus 6 µm particles. Bhagat et al. (2010) reported high from whole blood sample, an improved processing time focusing throughput results of 2100 particles/s. Moreo- of around 8 h was achieved compared to methods involv- ver, the system performance was validated for cell counting culturing or amplifcation. Also, the sensitivity of the ing using SH-SY5Y neuroblastoma cells. In a related work platform was the same as that of methods involving cul- from the same group, Kuntaegowdanahalli et al. (2009) turing or amplifcation. A multiplexed setup of the same developed a 5-loop spiral microchannel with a fork-shaped design was clinically validated and reported in a parallel outlet for the separation of polystyrene particles of 10, research (Khoo et al. 2014), and the detailed procedure 15, and 20 µm in sizes. As with any inertial platform, the of the fabrication and implementation of multiplexed proposed device utilizes the balance between inertial lift spiral microfuidics for CTCs isolation from blood sam- force and Dean drag force to focus particles near the side ple was later reported in 2015 by Warkiani et al. (2016) walls of the spiral channel. The authors reported respec- (see Fig. 9a). The multiplexed platform achieved 85% tive recovery rates of 90 and 80% for polystyrene particles recovery rate of CTCs cells at relatively high fow rate of and neurogenic tumor cells. Moreover, the authors claimed 1.5 mL/min. that the platform achieved a throughput of around 1 mil- In an extensive theoretical and experimental study, lion cells/min, which is higher than any cell sorting tech- Guan et al. (2013) has shown that spiral channels with nique method currently commercially available at the time rectangular cross sections has the limitation of low separa- of publication. tion resolution, especially in close range of particle sizes Hou et al. (2013) from the Lim and Han group utilized (see Fig. 9a, b). Therefore, the authors proposed the use spiral microfuidics, named as Dean fow fractionation of channels with trapezoidal cross sections which gener- (DFF), for continuous CTCs isolation from blood sam- ate stronger Dean drag force in the outer half of the chan- ples collected from patient with lung cancer. For the frst nel, i.e., half opposite to the spiral channel center. This time, sheath buffer was used in a spiral platform to facili- has led to higher separation distances between the dif- tate the processing of blood samples with hematocrit ferent sized particles, i.e., better separation effciency. In levels of 20–25% compared to other works that reported a continuation study, Warkiani et al. (2014) implemented as low hematocrit as 5%. The implementation of sheath trapezoidal spiral microfuidics for ultra-fast and label- buffer had resulting in a high throughput of 3 mL/h with free CTCs isolation (see Fig. 9b). The proposed platform the elimination of clogging issues, and CTCs recovery was able to achieve an 80% recovery rate for cancer cells rate and viability of 85 and 98%, respectively. In a fol- MCF-7, T24, and MDA-MB-231 from 7.5 mL of spiked low-up study, Hou et al. (2015) utilized a modifed ver- blood in just eight minutes. Warkiani et al. (2015a) further sion of the platform for label-free bacteria isolation from demonstrated the implementation of multiplexing multi- host blood cells. Ribosomal RNA detection was then uti- trapezoid spiral with membrane-less microfltration to lized to capture samples with low abundance pathogens eliminate the clogging limitation of using membrane fl- (~100 per mL) from the processed blood sample without ters. In their work, 48 spiral chips were integrated in one

Fig. 9 Spiral microfuidic channel for passive particle/cell sorting (ii) cross-sectional drawing (just before channel outlets) demonstrates a multi-layers spiral microfuidics with rectangular cross section, focusing of different size particles (iii) experimental results of parti- design used for CTCs isolation from blood sample b spiral micro- cles separation. a and b is adopted from Zhang et al. (2016) with per- fuidic with trapezoidal cross sections, design used for CTCs isola- mission from Royal Society of Chemistry tion from WBC (i) full spiral design with one inlet and two outlets

1 3 102 Page 16 of 23 Microfuid Nanofuid (2017) 21:102 setup to increase the fow rate from 6 mL/min for single 2.2.2 Multi‑orifce microfuidic channel (microvortices) chip up to 500 mL/min. The integrated setup was successfully utilized for the separation of CHO (10–20 µm) and Beside the utilization of curved microfuidic channel (spi- yeast (3–5 µm) cells with 90% separation effciency. In a ral) to generate secondary fow, microvortices with inertial recent work by Schaap et al. (2016), a 3-loop spiral micro- migration was also reported on microfuidic channel with fuidic with rectangular cross section of 350 µm width and single or multiple cavity/chamber on one or both sides. For 100 µm was proposed for particles and Algae cells sort- summarized view, Table 3 lists the reported microvortices ing. The utilized polystyrene particles are spherical in platforms according to their developing research group, shape with 5, 10, 20 µm diameter. The Algae cells sorted platform structure, application, fow rate, and reported are from Chlorella (spherical shape), Cyanothece (Prolate results. This idea was frst invented and investigated for spheroid), and Monoraphidium (Cylindrical). The authors different application by the Chiu group (Shelby et al. reported 96 and 77% separation rate for polystyrene parti- 2003; Shelby and Chiu 2004; Chiu 2007). Shelby et al. cle and Algae cell, respectively. (2003) proposed 30 µm height by 30 µm width straight

Table 3 Particle/cell separation methods on multi-orifce platforms Group leader Structure Application Flow rate Remarks

Shelby et al. (2003) Single microchamber Microvortex generation 1.08 mL/min With low fow rate, the proposed design can achieve 12 m/s rotational velocity Shelby and Chiu (2004), Single microchamber Particle/cell manipulation 0.001–2 mL/min The proposed design was Chiu (2007) employed for cell separation and to study the centrifugation effect on cells Lee et al. (2009a) Multi-chamber on single Mixing 0.08 mL/min, Re 7.2 Mixing range of 90% can side = be achieved in wide range of Reynold number (4.3–28.6) Lee et al. (2009b) Multi-chamber on single Focusing (red blood cells) 0.001–0.01 mL/min The proposed platform side requires sheath fuid to achieve 3D focusing Lee et al. (2011c) Multi-chamber on single Particles separation (4 and 0.016 mL/min 100% separation rate was side 10 µm) achieved with a throughput of 111 particles/s. The proposed platform requires sheath liquid Lee et al. (2011b) Multi-chamber on single Blood plasma separation 0.02 mL/min 62% separation rate was side achieved with a throughput of ~1.0 108 cells/ min × Lee et al. (2013) Multi-chamber on single Cancer cells separation 0.05–0.2 mL/min 99.1% recovery rate, 88.9% side blood cells rejection rate, 1.1 108 cells/min throughput× Park et al. (2009) Multi-chamber on two Particle focusing (7 µm) 0.08 mL/min Particles are focused near sides both sidewalls (at 0.6 from the centerline) Park and Jung (2009) Multi-chamber on two Particle separation (7, 0.1 mL/min ~1–5 104 separation sides 15 µm) throughput,× high recover rate of 90% but low purity 15% Sim et al. (2011) Multi-stage/multi- chamber Particle separation (7, 0.061 and 0.102 mL/min Compared to single stage, on two sides 15 µm) recover rate of 15 µm is improve to 88.7% with 89.1% purity Moon et al. (2011) Multi-chamber on two Cancer cells separation 0.126 mL/min 75.18% recovery rate of sides DEP MCF-7 cells. 162-fold of + enrichment was reported

1 3 Microfuid Nanofuid (2017) 21:102 Page 17 of 23 102

Fig. 10 Multi-orifce microfuidic channel (microvortices) a single permission Royal Society of Chemistry, b is adopted from Lee et al. microchamber (single vortex) by Chiu group, b multi-chamber on a (2009a) with permission from AIP Publishing LLC, c is adopted from single side of the channel by Park group, c multi-chamber on both Park et al. (2009) with permission from Royal Society of Chemistry side of the channel. a Is adopted from Shelby and Chiu (2004) with microchannel with a single diamond-shaped chamber inte- organelles, and shear stress that can cause physical changes grated on one side of the channel to generate high radial to cell surface. acceleration microvortices (Fig. 10a). The introduction of A series of microchambers, or as it also known as con- the microchamber creates detachment of fuid fow at the traction–expansion array (CEA), integrated on a single side opening of the microchamber that will create recirculation/ of the microchannel was frst reported and investigated by vortex fow in the diamond chamber. With optimum micro- the Park group (Fig. 10b). In their frst work, Lee et al. chamber dimensions and opening angle, fow velocity can (2009a) proposed CEA on a single side of the microchan- be accelerated from 3 m/s in the main channel to 12 m/s nel to perform laminar mixing between two liquids. Lee’s in the microchamber. This microcirculation/microvortex theoretical and experimental analysis showed that the pro- was frst reported to improve passive liquid mixing in a posed CEA structure creates acceleration and decelera- straight channel. To illustrate one of the potential application of streamlines that result in Dean-like vortices at the tions of microvortex, Shelby employed red polystyrene entrance of contraction region. As what happen in a curved/ beads and green slice beads with two different densities spiral channel, two vortices are generated, upper counter- of 1.5 and 1.8–2.0 g/cm3, respectively. When the fow rate clockwise vortex and lower clockwise vortex. As shown in of the main stream was increased from 1.5 to 20 m/s, the Fig. 10b, the two counter-rotating vortices shift the de-ion- green beads was centrifuged toward the outer edge of the ized water (DIW) toward the center of the channel where microchamber, while the red beads were concentrated at it will be surrounded by the fuorescein isothiocyanate the center of the chamber. In later works, the group utilized (FITC/FYE). The author reported a mixing rate of 90% by the same idea to study the effect of centrifugation/vortices utilizing 30 contraction–expansion array within a relatively on different types of cells and molecules (Shelby and Chiu wide range of Reynolds number of 4.3–28.6. In their fol- 2004; Chiu 2007). Compared to traditional centrifugation lowing works, the group implemented the same microfu- methods, microvotex method was able to reveal the dif- idic structure for three-dimensional hydrodynamic focusing ferent effect of centrifugation on a single-cell level, such of red blood cells (Lee et al. 2009b), inertial separation of as tensile stress that causes the relocation of intracellular different size polystyrene beads (Lee et al. 2011c), blood

1 3 102 Page 18 of 23 Microfuid Nanofuid (2017) 21:102 plasma separation (Lee et al. 2011b), and label-free can- MOFF platform. On the other hand, Moon et al. (2011) cer cell separation from whole blood (Lee et al. 2013). For integrated MOFF with dielectrophoretic (DEP) to improve inertial focusing, the group utilized the balance between the recovery rate of MOFF. This integrated platform was Dean force and inertial force to focus cells/particles based implemented to separate breast cancer cells (MCF-7) from on their size. spiked blood sample. The authors reported 162-fold of Multi-orifce fow fractionation (MOFF) with axisym- MCF-7 enrichment at a fow rate of 126 µL/min. metric CEA on both sides of the microchannel was frst For further improvement in term of throughput of CEA proposed by the Jung group in 2008 (Fig. 10c). Similar to platform, Mach et al. (2011) proposed an array of CEA the mechanism of single-sided CEA, double-sided CEA where multiple CEA channels are connected in parallel utilizes the balance between inertial force and microvorti- with a single input and a single output. However, Mach ces on both sides to focus cells/particles in a specifc path. et al. implemented the microvortices to trap larger particle This structure was frst reported by Park et al. (2009) where and not to focus it. In other words, as soon as larger par- they proposed a microchannel with 80 repeated contrac- ticles/cells reach a microvortex, it is centrifuged near the tion–expansion cycles for 7 µm polystyrene divinylbenzene vortex center and stays there, while small particles fow (PS-DVB). According to the authors, this design offers with the main stream. This idea was successfully imple- continues separation of cells/particles with high through- mented to trap targeted particles, then fuorescent labe- put. Moreover, the focusing position was shown to move ling it by a medium exchange process without the need from the channel sides to the channel center when parti- for manual pipetting and washing steps. Moreover, this cle Reynolds number Rep was increased from the range of method achieved high throughput in the range of mL/min. 0.8–2.3 up to the range of 3.0–3.5. In their following work, To release trapped particles/cells, the fow rate is decreased Park implemented this MOFF platform to separate parti- to weaken microvortices and let particles/cells escape to the cles of different sizes, i.e., 2, 7 and 15 µm that replicates main fow. This method was utilized to isolate CTCs and platelet, red blood cell, and white blood cell, respectively mesothelial cells from background mixture (Mach et al. (Park and Jung 2009). For this application, the authors 2011; Sollier et al. 2014; Che et al. 2013). report that MOFF has the advantages of being a continues process, does not need sheath fuid, and having an interme- diate fow rate of 1–5 104 particles/s compared to other 3 Summary and future outlook × separation methods (Park and Jung 2009). On the other hand, this platform showed low purity of 36.4% maximum This paper has reviewed the microfuidic platforms and for 15 µm particles. Furthermore, it was demonstrated that centrifugation approaches for particle/cell separation. Cen- while enlarging the collection region increases the recov- trifugation approaches are methods that utilize the physical ery rate, the level of separation purity dropped further to a centrifugation process on microfuidic platforms such as low of 15.5%. To improve system performance, the authors in microfuidic CD, or approaches that utilize liquid cen- suggested to have specifc fow rate to separate each type of trifugation effect which results from the Dean effect for particles from the background mixture. Recently, Warkiani cell/particle separation such as in spiral and multi-orifce et al. (2015b) reported a CEA platform for Malaria para- microfuidics. site enrichment as a preprocessing step for qPCR assay. The various implementation of particle/cell separation The CEA is utilized to focus and deplete the larger WBC on centrifugal microfuidic platforms are either passive cells, whereas unfocused smaller Malaria parasites are or active in nature. Table 1 in Sect. 2.1 shows the various recovered from the center of the focusing channel. Warki- methods reviewed. Passive methods include density-based ani et al. reported an impressive depletion rate of WBCs of sedimentation, separation by physical size, and separation 99.99%, with a throughput of 1 mL of lysed blood sample through immunoaffnity, whereas active methods include per 15 min. DEP or magnetic based separation. Most reported tech- Sim et al. (2011) and Moon et al. (2011) from the niques showed that the centrifugal microfuidic platform is Park group proposed two different design improvements most suitable when a preparation step of a raw sample is to the MOFF platform proposed by Park et al. 2009. Sim required to be integrated on the same platform. For exam- et al. (2011) developed a multi-stage-multi-orifce fow ple, the fractionation of blood sample to its sub-component fractionation (MS-MOFF) platform where samples fow can be easily performed with simple one-chamber designs. through two stages of CEA for better recovery rate. The Nevertheless, centrifugal microfuidic platforms have also design includes one CAE in the frst stage, and two CAR demonstrated the ability to perform single-particle/cell sep- in the second stage. With this MS-MOFF design, the fnal aration and particle/cell observations using simple passive achieved recovery rate of 15 µm particle size was 88.5% structure such as V-shaped cups, inertial focusing, or by the with 89.1% purity, which were unachievable with a normal implementation of a DGM.

1 3 Microfuid Nanofuid (2017) 21:102 Page 19 of 23 102 sample concentration HCC827, respectively fow rate and sample fow type) 99.1 80 85 100 85 80–88 according to 90–95 for MCF-7 and 98.7 44–84 (according to CTCs recovery rate (%) CTCs recovery cer and 15 lung cancer 3 10 15 breast can - 56 20 5 3 3 3 Test repetition Test ( n ) sample time only without CTCs-beads incu - bation time) 50.5 min for block - washing, ing, staining and cells analysis 1 h for 6 mL of 8 min 12.5 min 5 min 1 h 10 min (focusing 78 min Not reported 20 s fltration time. Processing time ume (system can process 6 mL in 1 h) - No specifc vol 7.5 mL 7.5 mL 7.5 mL 3 mL 3 µL each set 5 mL 1 mL 3 mL Processed volume hematocrit was spiked with MCF-7, spiked hematocrit was SK-BR-3, and HCC70 breast cancer cells spiked with MCF-7, T24 and MDA- with MCF-7, spiked MB-231 CTC line cells (breast and lung cancer cells) 10, 15 µm and lysed blood sample CTCs lines with different spiked and lung cancers at 2 × concentra - tion were fractionated, lysed, and resuspended in 3.75 mL PBS whole blood from patient with lung diluted to 20% Blood was cancer. hematocrit (only 2 dilution of whole blood) MCF-7 breast cancer. The sample MCF-7 breast cancer. 4.5 µm with activated is mixed magnetic beads MCF-7 breast cancer and HCC827 The blood was lung cancer cell. with microbeads. Microbeads mixed with anti-EPCAM to are covered capture CTCs spiked in diluted blood. Dilution spiked rate is 2 and 5% v/v in PBS buffer. 200:1 beads to cancer cells ration implemented was in PBS and whole blood sample (Laboratory Whole experiment). blood sample from patient with lung cancer (Clinical cancer and gastric experiment) Whole blood sample with 45% Whole blood sample was lysed and Whole blood sample was System evaluated with: microbeads 6, System evaluated Blood from advance metastatic breast Blood from advance Whole blood spiked with CTCs, and Whole blood spiked Whole blood sample spiked with Whole blood sample spiked Whole blood sample spiked with Whole blood sample spiked 250–1050 MCF-7 breast cancer cells MCF-7 breast cancer cells spiked MCF-7 breast cancer cells spiked Sample type/preparation fuidic with focus - (sheath ing fow fow) sheath fow sheath fow sheath fow phoresis, continu - ous sedimentation density gradient medium (DGM) inertial cell sorting Multi-orifce micro - Trapezoidal spiral Trapezoidal Spiral channel, Spiral channel, Spiral channel, Centrifugo-magneto - Immunoaffnity, and Immunoaffnity, Immunoaffnity, and Immunoaffnity, Membrane flter Separation method fuidic ics (trapezoidal spiral) ics (spiral) ics (spiral) ics (spiral) fuidic platform fuidic platform fuidic platform fuidic platform Multi-orifce micro - Inertial microfuid - Inertial microfuid - Inertial microfuid - Inertial microfuid - Centrifugal micro - Centrifugal Centrifugal micro - Centrifugal Centrifugal micro - Centrifugal Centrifugal micro - Centrifugal Platform type Microfuidic platforms for CTCs isolation Lee et al. ( 2013 ) Warkiani et al. ( 2014 ) Warkiani Warkiani et al. ( 2016 ) Warkiani Khoo et al. ( 2014 ) Hou et al. ( 2013 ) Kirby et al. ( 2015 ) Park et al. ( 2014 ) Park Aguirre et al. ( 2015 ) Lee et al. ( 2014 ) 4 Table References

1 3 102 Page 20 of 23 Microfuid Nanofuid (2017) 21:102

The main advantages of using the centrifugal microfuidic platforms for particle/cell separation include simple design and fabrication process, allowance for the integration of multi-processing stages of mixing, valving, centrifugation…,etc., improved portability of the proposed methods, requires less human interaction, and includes a wide ~10–20 CTCs recovery rate (%) CTCs recovery 75.18 ~20 range of implementable unit/operations that can be utilized in different applications. On the other hand, the main disadvantage of using centrifugal microfuidic platforms for particle/cell separation is the limited space available on the

cancer and 8 lung cancer platform for sample, reagents, and other fuids/components. 6 Test repetition Test ( n ) 4 breast Even though some authors reported the ability to process sample volume of around 5 mL, processing such sample volumes limits the number of processes that can be multiplexed on the same platform. Inertial microfuidic platforms are categorized into

sample spiral microfuidics and multi-orifce microfuidics, or Less than 30 min Processing time 1 h for 0.7 mL of 20 min microvotices. Table 2 in Sect. 2.2.1 shows the various implementation of spiral microfuidics, while Table 3 in Sect. 2.2.2 shows the works performed using multi-orifce. In general, inertial microfuidics is more suitable than centrifugal microfuidics when it comes to process- processed ume (system can process 0.7 mL in 1 h) 10 mL can be Processed volume - No specifc vol 7.5 ing samples of high volumes, e.g., for CTCs isolation. This is due to the fact that samples are loaded using syringe pumps which can cater to a wider range of sample volumes required for any kind of application.

) for RBC, Many works have demonstrated that inertial micro- − 1 fuidics is suitable for particles/cell separation. Some µL 3 examples are the separation of CTCs, WBC, Algae, and bacteria with high recovery and viability rate. Some studies have also highlighted that spiral microfuidic designs and spiked with breast or and spiked

X can be easily adapted for different applications by adjust-

–40 ing the spiral cross-sectional, dimension, length, and X donors was diluted in a range of donors was 5 have the following cells concentra - the following have tion: 10, 1, 1 ( × 10 diluted to 5% v/v (i.e., 0.5 mL with of whole blood), and spike MCF-7 cancer cell WBC, and MCF-7, respectively lung cancer. Blood from patients lung cancer. with breast and lung cancer was used for system evaluation Whole blood sample from healthy Whole blood sample from healthy Whole blood sample was processed to Whole blood sample was mL of whole blood sample was 10 mL of whole blood sample was Sample type/preparation fow rate. Meanwhile, multi-orifce has been reported to be also designed with different structures to suit specifc application. Variations of the designs include single chamber (single vortex), multi-chamber on a single side of the channel, and multi-chamber on both sides of the channel. fuidic, multi-line in parallel fuidic with DEP fuidic, multi-line in parallel The many advantages of inertial, or spiral and multi- Multi-orifce micro - Multi-orifce micro - Multi-orifce micro - Separation method orifce microfuidics include simple design and fabrication, the ability to process large sample volumes, high throughput, high recovery rate, and the ability for multiplexing for improved performance. However, the use of syringe pump has reduced the fexibility of the plat- fuidic fuidic fuidic form in integrating preprocessing steps for more involved Multi-orifce micro - Multi-orifce micro - Multi-orifce micro - Platform type processes. As both types of microfuidic platforms, namely centrifugal microfuidic platform and inertial microfuidic platform, have their advantages and disadvantages, the requirement of the application should be carefully considered prior to platform selection. For example, Sollier et al. ( 2014 ) Moon et al. ( 2011 ) Mach et al. ( 2011 ) References 4 Table continued to design a point-of-care application for low resource

1 3 Microfuid Nanofuid (2017) 21:102 Page 21 of 23 102 environment, and preprocessing is required, then centrif- Burger R, Kirby D, Glynn M, Nwankire C, O’Sullivan M, Siegrist ugal microfuidic platforms should be a better choice. On J, Kinahan D, Aguirre G, Kijanka G, Gorkin RA (2012a) Cen- trifugal microfuidics for cell analysis. Curr Opin Chem Biol the other hand, if the desired platform is the miniaturi- 16(3):409–414 zation of commercially available machines with reduced Burger R, Reith P, Kijanka G, Akujobi V, Abgrall P, Ducrée J processing time, with relatively high volume of sample, (2012b) Array-based capture, distribution, counting and mul- then inertial microfuidic platform is the better choice. tiplexed assaying of beads on a centrifugal microfuidic platform. Lab Chip 12(7):1289–1295 Due to the importance of CTCs isolation and retrieval for Che J, Mach AJ, Go DE, Talati I, Ying Y, Rao J, Kulkarni RP, Di early cancer diagnosis, the proposed microfuidic meth- Carlo D (2013) Microfuidic purifcation and concentration of ods for CTCs isolation are summarized in Table 4. malignant pleural effusions for improved molecular and cyto- In summary, microfuidic platforms are capable of (1) morphological diagnostics. PLoS ONE 8(10):e78194 Chiu DT (2007) Cellular manipulations in microvortices. Anal Bio- having high portability, (2) wide range of applicable oper- anal Chem 387(1):17–20 ations, (3) integrating preprocessing stages, (4) process- Dean W (1928) Fluid motion in a curved channel. In: Proceedings ing relatively high sample volume, (5) high throughput, of the Royal Society of London A: mathematical, physical and and (6) high recovery and viability rate. With proper con- engineering sciences, vol 787. The Royal Society, pp 402–420 Di Carlo D (2009) Inertial microfuidics. Lab Chip 9(21):3038–3046 sideration to the requirement of the application, various Di Carlo D, Irimia D, Tompkins RG, Toner M (2007) Continu- particle/cell separation platforms can be accomplished. ous inertial focusing, ordering, and separation of particles in microchannels. Proc Natl Acad Sci 104(48):18892–18897 Acknowledgements This research was fnancially supported by Ducrée J, Haeberle S, Lutz S, Pausch S, Von Stetten F, Zengerle the German Academic Exchange Service (DAAD), project enti- R (2007) The centrifugal microfuidic Bio-Disk platform. J tled “Inertial focusing for continuous nanoparticles separation Micromech Microeng 17(7):S103–S115 in femtosecond laser 3D micromachined curved channels,” and Furutani S, Nagai H, Takamura Y, Kubo I (2010) Compact disk Seed-fund number SATS 27/2016 provided by German Jordanian (CD)-shaped device for single cell isolation and PCR of University—Amman. a specifc gene in the isolated cell. Anal Bioanal Chem 398(7–8):2997–3004 Glynn M, Nwankire C, Lemass K, Kinahan DJ, Ducrée J (2015) Cluster size distribution of cancer cells in blood using stopped-fow centrifugation along scale-matched gaps of a References radially inclined rail. Microsyst Nanoeng 1. doi:10.1038/ micronano.2015.18 Aguirre GR, Efremov V, Kitsara M, Ducrée J (2015) Integrated Guan G, Wu L, Bhagat AA, Li Z, Chen PC, Chao S, Ong CJ, Han micromixer for incubation and separation of cancer cells on a J (2013) Spiral microchannel with rectangular and trapezoi- centrifugal platform using inertial and dean forces. Microfuid dal cross-sections for size based particle separation. Scientifc Nanofuid 18(3):513–526 reports 3 Amasia M, Madou M (2010) Large-volume centrifugal micro- Haeberle S, Brenner T, Zengerle R, Ducrée J (2006) Centrifugal fuidic device for blood plasma separation. Bioanalysis extraction of plasma from whole blood on a rotating disk. Lab 2(10):1701–1710 Chip 6(6):776–781 Arosio P, Müller T, Mahadevan L, Knowles TP (2014) Density-gra- Hou HW, Warkiani ME, Khoo BL, Li ZR, Soo RA, Tan DS-W, dient-free microfuidic centrifugation for analytical and prepar- Lim WT, Han J, Bhagat AAS, Lim CT (2013) Isolation and ative separation of nanoparticles. Nano Lett 14(5):2365–2371 retrieval of circulating tumor cells using centrifugal forces. Ashworth T (1869) A case of cancer in which cells similar to those Scientifc reports 3 in the tumours were seen in the blood after death. Aust Med J Hou HW, Bhattacharyya RP, Hung DT, Han J (2015) Direct detec- 14(3):146–149 tion and drug-resistance profling of bacteremias using inertial Bhagat AAS, Kuntaegowdanahalli SS, Papautsky I (2008) Con- microfuidics. Lab Chip 15(10):2297–2307 tinuous particle separation in spiral microchannels using dean Hyun K-A, Jung H-I (2014) Advances and critical concerns with fows and differential migration. Lab Chip 8(11):1906–1914 the microfuidic enrichments of circulating tumor cells. Lab Bhagat AAS, Kuntaegowdanahalli SS, Kaval N, Seliskar CJ, Chip 14(1):45–56 Papautsky I (2010) Inertial microfuidics for sheath-less Jiang M, Mazzeo AD, Drazer G (2015) Centrifugal determinis- high-throughput fow cytometry. Biomed Microdevice tic lateral displacement separation system. arXiv preprint 12(2):187–195 arXiv:150706027 Boycott A (1920) Sedimentation of blood corpuscles. Nature 104:532 Karnis A, Goldsmith H, Mason S (1966) The fow of suspen- Brenner T, Glatzel T, Zengerle R, Ducree J (2005) Frequency- sions through tubes: v. Inertial effects. Can J Chem Eng dependent transversal fow control in centrifugal microfuidics. 44(4):181–193 Lab Chip 5(2):146–150. doi:10.1039/b406699e Khoo BL, Warkiani ME, Tan DS, Bhagat AA, Irwin D, Lau DP, Burger R, Ducrée J (2012) Handling and analysis of cells and Lim AS, Lim KH, Krisna SS, Lim WT, Yap YS, Lee SC, Soo bioparticles on centrifugal microfuidic platforms. Expert Rev RA, Han J, Lim CT (2014) Clinical validation of an ultra high- Mol Diagn 12(4):407–421. doi:10.1586/erm.12.28 throughput spiral microfuidics for the detection and enrich- Burger R, Kijanka G, Sheils O, O’Leary J, Ducrée J (2011) Arrayed ment of viable circulating tumor cells. PLoS One 9(7):e99409. capture, assaying and binary counting of cells in a stopped- doi:10.1371/journal.pone.0099409 fow sedimentation mode. In: 15th International conference on Kim J, Kido H, Rangel RH, Madou MJ (2008) Passive fow switch- miniaturized systems for chemistry and life sciences (uTAS). ing valves on a centrifugal microfuidic platform. Sens Actua- Seattle, pp 2–6 tors B Chem 128(2):613–621. doi:10.1016/j.snb.2007.07.079

1 3 102 Page 22 of 23 Microfuid Nanofuid (2017) 21:102

Kim T-H, Hwang H, Gorkin R, Madou M, Cho Y-K (2013) Geom- Moen ST, Hatcher CL, Singh AK (2016) A centrifugal microfu- etry effects on blood separation rate on a rotating disc. Sens idic platform that separates whole blood samples into multiple Actuators B Chem 178:648–655 removable fractions due to several discrete but continuous den- Kinahan D, Kilcawley N, Glynn M, Kirby D, Ducree J (2014a) Iso- sity gradient sections. PLoS ONE 11(4):e0153137 lation of white blood cells using paper-triggered dissolvable- Moon H-S, Kwon K, Kim S-I, Han H, Sohn J, Lee S, Jung H-I (2011) flm valves on a centrifugal platform. In: 18th International Continuous separation of breast cancer cells from blood samples conference on miniaturized systems for chemistry and life sci- using multi-orifce fow fractionation (MOFF) and dielectropho- ences (uTAS). Texas, pp 1425–1427 resis (DEP). Lab Chip 11(6):1118–1125 Kinahan DJ, Kearney SM, Glynn MT, Ducrée J (2014b) Spira mira- Morijiri T, Yamada M, Hikida T, Seki M (2013) Microfuidic counter- bilis enhanced whole blood processing in a lab-on-a-disk. Sens fow centrifugal elutriation system for sedimentation-based cell Actuators, A 215:71–76 separation. Microfuid Nanofuid 14(6):1049–1057 Kirby D, Siegrist J, Kijanka G, Zavattoni L, Sheils O, O’Leary J, Nwankire CE, Maguire I, Kernan D, Glynn M, Kirby D, Ducree Burger R, Ducrée J (2012) Centrifugo-magnetophoretic parti- J (2015a) SIZE-and deformability-based particle sorting by cle separation. Microfuid Nanofuid 13(6):899–908 strategic design of obstacle arrays in continuous centrifugal Kirby D, Glynn M, Kijanka G, Ducrée J (2015) Rapid and cost-eff- sedimentation mode. In: Transducers-2015 18th international cient enumeration of rare cancer cells from whole blood by low- conference on solid-state sensors, actuators and microsystems loss centrifugo-magnetophoretic purifcation under stopped-fow (TRANSDUCERS). IEEE, pp 1854–1856 conditions. Cytom Part A 87(1):74–80 Nwankire CE, Venkatanarayanan A, Glennon T, Keyes TE, Forster Kubo I, Furutani S, Nagai H (2009) The activity determination of sin- RJ, Ducrée J (2015b) Label-free impedance detection of can- gle cell by isolation and cultivation on a centrifugal fow disk. cer cells from whole blood on an integrated centrifugal micro- ECS Trans 16(17):1–8 fuidic platform. Biosens Bioelectron 68:382–389 Kuntaegowdanahalli SS, Bhagat AAS, Kumar G, Papautsky I (2009) Park J-S, Jung H-I (2009) Multiorifce fow fractionation: con- Inertial microfuidics for continuous particle separation in spiral tinuous size-based separation of microspheres using a microchannels. Lab Chip 9(20):2973–2980 series of contraction/expansion microchannels. Anal Chem Kuo J-N, Chen X-F (2015) Plasma separation and preparation on cen- 81(20):8280–8288 trifugal microfuidic disk for blood assays. Microsyst Technol Park J-M, Kim B-C, Lee J-G, Ko C (2008) One-step white blood 21(11):2485–2494 cell separation from whole blood on centrifugal microfuidic. Kuo J-N, Li B-S (2014) Lab-on-CD microfuidic platform for rapid Paper presented at the NSTI-nanotech Boston, 1–5 June 2008 separation and mixing of plasma from whole blood. Biomed Park J-S, Song S-H, Jung H-I (2009) Continuous focusing of micro- Microdevice 16(4):549–558 particles using inertial lift force and vorticity via multi-orifce Lee MG, Choi S, Park J-K (2009a) Rapid laminating mixer using microfuidic channels. Lab Chip 9(7):939–948 a contraction-expansion array microchannel. Appl Phys Lett Park J-M, Kim MS, Moon H-S, Yoo CE, Park D, Kim YJ, Han K-Y, 95(5):051902 Lee J-Y, Oh JH, Kim SS (2014) Fully automated circulating Lee MG, Choi S, Park J-K (2009b) Three-dimensional hydrodynamic tumor cell isolation platform with large-volume capacity based focusing with a single sheath fow in a single-layer microfuidic on lab-on-a-disc. Anal Chem 86(8):3735–3742 device. Lab Chip 9(21):3155–3160 Riegger L, Grumann M, Steigert J, Lutz S, Steinert C, Mueller C, Lee BS, Lee YU, Kim H-S, Kim T-H, Park J, Lee J-G, Kim J, Kim Viertel J, Prucker O, Rühe J, Zengerle R (2007) Single-step H, Lee WG, Cho Y-K (2011a) Fully integrated lab-on-a-disc for centrifugal hematocrit determination on a 10-$ processing simultaneous analysis of biochemistry and immunoassay from device. Biomed Microdevice 9(6):795–799 whole blood. Lab Chip 11(1):70–78 Sajeesh P, Sen AK (2014) Particle separation and sorting in micro- Lee MG, Choi S, Kim H-J, Lim HK, Kim J-H, Huh N, Park J-K fuidic devices: a review. Microfuid Nanofuid 17(1):1–52 (2011b) Inertial blood plasma separation in a contraction– Schaap A, Dumon J, Den Toonder J (2016) Sorting algal cells by expansion array microchannel. Appl Phys Lett 98(25):253702 morphology in spiral microchannels using inertial microfuid- Lee MG, Choi S, Park J-K (2011c) Inertial separation in a con- ics. Microfuid Nanofuid 20(9):125 traction–expansion array microchannel. J Chromatogr A Segre G (1961) Radial particle displacements in Poiseuille fow of 1218(27):4138–4143 suspensions. Nature 189:209–210 Lee MG, Shin JH, Bae CY, Choi S, Park J-K (2013) Label-free cancer Segre G, Silberberg A (1962a) Behaviour of macroscopic rigid cell separation from human whole blood using inertial microfu- spheres in Poiseuille fow part 1. Determination of local con- idics at low shear stress. Anal Chem 85(13):6213–6218 centration by statistical analysis of particle passages through Lee A, Park J, Lim M, Sunkara V, Kim SY, Kim GH, Kim M-H, Cho crossed light beams. J Fluid Mech 14(01):115–135 Y-K (2014) All-in-one centrifugal microfuidic device for size- Segre G, Silberberg A (1962b) Behaviour of macroscopic rigid selective circulating tumor cell isolation with high purity. Anal spheres in Poiseuille fow part 2. Experimental results and Chem 86(22):11349–11356 interpretation. J Fluid Mech 14(01):136–157 Li T, Zhang L, Leung KM, Yang J (2010) Out-of-plane microvalves Shamloo A, Selahi A, Madadelahi M (2016) Designing and mod- for whole blood separation on lab-on-a-CD. J Micromech Micro- eling a centrifugal microfuidic device to separate target blood eng 20(10):105024 cells. J Micromech Microeng 26(3):035017 Mach AJ, Kim JH, Arshi A, Hur SC, Di Carlo D (2011) Automated Shelby JP, Chiu DT (2004) Controlled rotation of biological micro- cellular sample preparation using a centrifuge-on-a-chip. Lab and nano-particles in microvortices. Lab Chip 4(3):168–170 Chip 11(17):2827–2834 Shelby JP, Lim DS, Kuo JS, Chiu DT (2003) Microfuidic systems: Madou M, Zoval J, Jia G, Kido H, Kim J, Kim N (2006) Lab on a CD. high radial acceleration in microvortices. Nature 425(6953):38 Annu Rev Biomed Eng 8:601–628 Siegrist J, Burger R, Kirby D, Zavattoni L, Kijanka G, Ducrée J Martinez-Duarte R, Gorkin RA III, Abi-Samra K, Madou MJ (2011) Stress-free centrifugomagnetic 2D-separation of cancer (2010) The integration of 3D carbon-electrode dielectropho- cells in a stopped-fow mode. In: 15th International confer- resis on a CD-like centrifugal microfuidic platform. Lab Chip ence on miniaturized systems for chemistry and life sciences 10(8):1030–1043 (uTAS). Seattle, pp 2–6

1 3 Microfuid Nanofuid (2017) 21:102 Page 23 of 23 102

Sim TS, Kwon K, Park JC, Lee J-G, Jung H-I (2011) Multistage- of burst frequency equations for passive valves on centrifugal multiorifce fow fractionation (MS-MOFF): continuous size- microfuidic platforms. Med Biol Eng Comput 51(5):525–535 based separation of microspheres using multiple series of con- Tomlinson MJ, Tomlinson S, Yang XB, Kirkham J (2013) Cell sep- traction/expansion microchannels. Lab Chip 11(1):93–99 aration: terminology and practical considerations. J Tissue Eng Smith S, Mager D, Perebikovsky A, Shamloo E, Kinahan D, Mishra 4:2041731412472690 R, Torres Delgado SM, Kido H, Saha S, Ducrée J (2016) CD- Warkiani ME, Guan G, Luan KB, Lee WC, Bhagat AAS, Chaudhuri based microfuidics for primary care in extreme point-of-care PK, Tan DS-W, Lim WT, Lee SC, Chen PC (2014) Slanted spi- settings. Micromachines 7(2):22 ral microfuidics for the ultra-fast, label-free isolation of circu- Sollier E, Go DE, Che J, Gossett DR, O’Byrne S, Weaver WM, lating tumor cells. Lab Chip 14(1):128–137 Kummer N, Rettig M, Goldman J, Nickols N (2014) Size- Warkiani ME, Tay AKP, Guan G, Han J (2015a) Membrane-less selective collection of circulating tumor cells using Vortex microfltration using inertial microfuidics. Scientifc reports 5 technology. Lab Chip 14(1):63–77 Warkiani ME, Tay AKP, Khoo BL, Xiaofeng X, Han J, Lim CT Steigert J, Grumann M, Dube M, Streule W, Riegger L, Bren- (2015b) Malaria detection using inertial microfuidics. Lab Chip ner T, Koltay P, Mittmann K, Zengerle R, Ducrée J (2006) 15(4):1101–1109 Direct hemoglobin measurement on a centrifugal microfu- Warkiani ME, Wu L, Tay AKP, Han J (2015c) Large volume microfuidic platform for point-of-care diagnostics. Sens Actuators, A idic cell sorting. Annu Rev Biomed Eng 17(1):1–34 130:228–233 Warkiani ME, Khoo BL, Wu L, Tay AKP, Bhagat AAS, Han J, Steigert J, Brenner T, Grumann M, Riegger L, Lutz S, Zengerle Lim CT (2016) Ultra-fast, label-free isolation of circulating R, Ducrée J (2007) Integrated siphon-based metering and tumor cells from blood using spiral microfuidics. Nat Protoc sedimentation of whole blood on a hydrophilic lab-on-a-disk. 11(1):134–148 Biomed Microdevice 9(5):675–679 Zhang J, Guo Q, Liu M, Yang J (2008) A lab-on-CD prototype Strohmeier O, Keller M, Schwemmer F, Zehnle S, Mark D, von for high-speed blood separation. J Micromech Microeng Stetten F, Zengerle R, Paust N (2015a) Centrifugal microfu- 18(12):125025 idic platforms: advanced unit operations and applications. Zhang J, Yan S, Yuan D, Alici G, Nguyen N-T, Warkiani ME, Li W Chem Soc Rev 44:6187–6229 (2016) Fundamentals and applications of inertial microfuidics: a Strohmeier O, Keller M, Schwemmer F, Zehnle S, Mark D, von review. Lab Chip 16(1):10–34 Stetten F, Zengerle R, Paust N (2015b) Centrifugal microfu- Zhao Y, Schwemmer F, Zehnle S, von Stetten F, Zengerle R, Paust N idic platforms: advanced unit operations and applications. (2015) Centrifugo-pneumatic sedimentation, re-suspension and Chem Soc Rev 44(17):6187–6229 transport of microparticles. Lab Chip 15(21):4133–4137 Tang M, Wang G, Kong S-K, Ho H-P (2016) A review of biomedical centrifugal microfuidic platforms. Micromachines 7(2):26 Thio THG, Soroori S, Ibrahim F, Al-Faqheri W, Soin N, Kulinsky L, Madou M (2013) Theoretical development and critical analysis

1 3