Analytical and Bioanalytical Chemistry

, Volume 387, Issue 1, pp 17–20

Cellular manipulations in microvortices


    • Department of ChemistryUniversity of Washington

DOI: 10.1007/s00216-006-0611-2

Cite this article as:
Chiu, D.T. Anal Bioanal Chem (2007) 387: 17. doi:10.1007/s00216-006-0611-2


The ability to tailor the size of microfluidic systems to match the length scale of single biological cells has led to a plethora of new microfluidic techniques geared towards the manipulation, culturing, and analysis of single cells [111]. Examples include the use of laminar flows for the rapid exchange of the solution environment around single cells [2] and the targeting of small molecules to localized regions of single cells [3]. By miniaturizing the channels to cellular or subcellular dimensions [4, 5], single cells can be trapped in defined regions of the fluidic system for subsequent detailed studies. Miniaturized devices for the high-throughput manipulation and detection of streams of single cells have also been demonstrated, and include such widely used methods as flow cytometry and cell sorting [6]. By combining electrical or optical methods with microfluidics such as laser trapping [7, 11], single cells and particles can be precisely positioned. In addition to single-cell manipulation, a range of integrated microfluidic platforms has been designed for the analysis of the contents of single cells [8, 9] or even that of single subcellular components [10, 11].

Some of these microfluidic methods originated from the miniaturization of their widely used macrofluidic counterparts. Many other new microfluidic techniques, however, are based on the unique fluidic behavior in microchannels; these latter techniques are especially interesting because they exploit the physics of low-Reynold’s number flow and offer new capabilities that are otherwise unattainable. Here we describe one unique microfluidic system that we have explored in the past years for exerting rotational control over single cells [1216]. This system is based on the use of microvortices, which are tiny re-circulating flows that we controllably generate in microchannels. This system illustrates the intriguing possibilities when unique microfluidic behaviors are applied towards biological and cellular manipulations.

Controlled generation of microvortices

Re-circulating flows or vortices are formed when flow detachment occurs, and can be generated under a wide range of conditions. Figure 1a–d shows some simple channel geometries in which microvortices form. For example, Fig. 1c is a diamond-shaped chamber designed adjacent to a straight channel. As the fluid passes by the chamber, flow detaches into the chamber where it re-circulates. Re-circulating flows are common in 90° corners (Fig. 1a), and in dead volumes located adjacent to straight-channel flows (Fig. 1c,d).
Fig. 1

a–d Microfluidic designs that cause flow detachment and the formation of microvortices; Femlab simulations are shown to the top of each channel design. The arrows indicate the direction of fluid flow. All scale bars are 30 μm. e–g Generation of ultra-high radial acceleration in microvortices. e Schematic that shows the design and dimensions of the microchannels. f A three-dimensional fluid simulation shows re-circulation of flow in the microvortex at up to ∼15 m/s. g Experimental measurements showing the speed of fluid flow and radial acceleration in a microvortex generated in the trapezoidal-shaped chamber depicted in panel E. The rotational frequency (left ordinate) was measured up to 250 kHz with a corresponding radial acceleration (right ordinate) of ∼107 m/s2. Rotational velocity (upper abscissa) and volumetric flow rate (lower abscissa) are also plotted. Experimental measurements were represented by (◁) and simulation values by (○). (Reprinted with permission from Shelby et al., Anal. Chem. (Chapel Hill, NC) 76, 2492, Copyright 2004 American Chemical Society and with permission from Shelby et al., Nature (London, England) 425, 38, Copyright 2003 Nature Publishing Group

As radial acceleration scales quadratically with velocity and inversely with radius, the combination of high flow velocities in microchannels and small radii of microvortices enables the generation of high radial accelerations [12]. Although many designs produce re-circulating flows in microchannels, only a few designs generate stable vortices whose center of rotation stays relatively spatially confined. Furthermore, we found the average rotational velocity as well as the formation and the shape of the microvortex to depend critically on the geometries of the microchamber, especially the angle of the chamber opening and the aspect ratio of the channel width to the width of the opening. Microvortices formed in secondary chambers (Fig. 1d), that is, chambers where the re-circulation is driven by another microvortex, often provide the stable conditions required for single-cell rotation.

Fluidic behavior of microvortices

To study the fluidic properties of microvortices and to verify its ability to generate high radial accelerations, we must be able to measure fast fluid flow in such small dimensions. Using single-molecule detection and spatially defined uncaging of caged fluorescein, we were able to map flows at speeds up to tens of meters per second within confines of micrometers [15]. With this sensitive flow mapping method, we were able to measure experimentally rotational velocities as high as ∼12 m/s at 10 μm from the vortex core, which corresponds to a radial acceleration of ∼1×107 m/s2 or ∼1×106 g (Fig. 1g) [12].

This ultra-high rotational velocity was attained in a specially designed trapezoidal microchamber connected to an asymmetrically constricted microchannel (Fig. 1e). There were two effects that made this design attractive for generating high rotational velocities. The first effect was that, by reducing the cross-sectional area, the channel constriction increased the overall flow velocity at the opening of the trapezoid chamber, provided that sufficient pressure was applied to counter the pressure-drop associated with the constriction. The second effect arose from the asymmetric nature of the constricted channel, which skewed the velocity maxima towards the chamber opening and drove the microvortex more effectively.

Separation and fractionation

Because of the centrifugal force generated by the microvortex, we could demonstrate the concentration and separation of red polystyrene beads (ρ=1.05 g/cm3) and green silica beads (ρ=1.8–2.0 g/cm3) in a solution of 50% w/w of CsCl (ρ=1.56 g/cm3), where the low-density red beads were driven inward and concentrated in the center of the microvortex while the high-density green beads occupied the outer region of the vortex (Fig. 2a–c). The ability to use fluid flow with no moving parts to generate high centrifugal forces may prove useful for integrated microanalytical separations such as in sample preparation and analyte concentrations. Microvortices may also have potential towards the fractionation of the subcelluar contents of single cells.
Fig. 2

a–c Density-based bead separation in a microvortex. d–f Fast rotation of a single B-lymphocyte at ∼200 Hz, shown before (d), immediately after (e), and one minute after (f) rotation of the cell; arrows point to the nuclear membrane. g–i Effect of shear stress on a rotating mouse mast cell. Cell-surface molecules were clearly visible prior to rotation (g), but rotation at ∼60 Hz for 10 mins (h) caused shear-induced detachment of these surface molecules, which can be seen in the image taken immediately after rotation (i). j A single λ DNA molecule in the globular state was optically trapped then rotated in the microvortex. The circular streak in the image was caused by a free λ DNA tracing the flow. k–m A sequence of images showing the rotation of a small aggregate of three DNA molecules. Scale bars=10 μm. (Reprinted with permission from Shelby et al., Nature (London, England) 425, 38, Copyright 2003 Nature Publishing Group and with permission from Shelby et al., Lab on a Chip (Cambridge UK) 4, 168, Copyright 2004 Royal Society of Chemistry

Rotation of single cells and molecules

In addition to using the microvortex as a method of separating or fractionating particles, it offers a new venue for exerting rotational control over single cells [14]. To demonstrate controlled rotation and orientation, we optically trapped single cells and positioned them at the center of the microvortex. Because of the no-slip boundary condition at the fluid-particle interface, the shear stress from the re-circulation flow in the microvortex caused the rotation of the cell placed at the center of the vortex.

Figure 2d–f shows the fast rotation of a mouse B-lymphocyte, where the appearance of the cell is shown before (Fig. 2d), immediately after (Fig. 2e), and 1 min after (Fig. 2f) rotation at ∼200 Hz. The area defined by the nuclear membrane (arrow) measured ∼125 μm2 (∼70 % of the cell area) prior to spinning. Immediately after spinning (Fig. 2e), the area within the nuclear membrane was observed to expand to ∼150 μm2 (∼85% of the cell area). The expanded nucleus slowly relaxed back to its original size, and measured ∼125 μm2 after 1 min (Fig. 2f). Although we have achieved ultra-high rotational frequencies of 2×105 Hz with a corresponding radial acceleration of greater than 106 (Fig. 2g) in a microvortex, the rate at which we can spin a single cell is currently limited by our ability to use optical trapping to maintain the cell at the center of rotation. Because of the instability of the vortex center at high rotational rates, we can only retain a spinning cell with optical trapping up to an estimated 200 Hz (∼12,000 rpm). Nevertheless, this rate is still much faster than current electrical or optical methods of single-cell rotation.

In addition to the presence of centrifugal force, the steep velocity gradients present in microvortices can cause significant shear stress on a fast rotating cell. Figure 2g–i shows a mouse mast cell before (Fig. 2g), during (Fig. 2h), and after rotation at ∼60 Hz (Fig. 2i). The cell was rotated periodically over a period of 10 mins. The shear stress at the surface of the cell, which was produced by the steep velocity gradient present in the microchamber, caused a shear-induced detachment of the glycocalyx at the membrane-fluid interface. We estimated the local shear stress to be as high as ∼100 dyn/cm2 [14]. Endothelial cells can respond to shear stress as low as 0.5 dyn/cm2; however typical in-vivo arterial shear stress level is ∼12 dyn/cm2.

Besides single cells, microvortices also can be applied towards the manipulation of subcellular structures, such as single DNA molecules. Figure 2j illustrates the rotation of a single λ-DNA molecule that has been compacted into a globular state [7]. A free DNA molecule tracing the flow in the microvortex caused the circular streak in the image. Rotation of a single DNA is difficult to observe under fluorescence; thus, we trapped simultaneously three DNA molecules, which caused these DNAs to form an aggregate within the optical trap. Figure 2k–m shows the rotation of this DNA aggregate. One advantage of using the microvortex to rotate objects is the great dynamic range (less than one Hz to hundreds of Hz) available and the precision with which one can rotate single cells and microparticles.


Microvortices offer intriguing potential in exerting controlled and precise rotation and shear stress on single cells, both for mechanical control and for fundamental studies of cellular behavior. The high radial acceleration attainable in microvortices may be harnessed for the fractionation of single-cell contents. To achieve these goals, however, is not without challenges. Besides high radial acceleration, also present in microvortices are hydrodynamic forces and shear stress derived from the steep velocity gradient inherent in the vortices [16]. To fully exploit the utility of this interesting microfluidic phenomenon will require both an in-depth understanding of its properties and behavior as well as clever microfluidic designs that can tailor this system with the desired cellular applications.


Support from the NSF and NIH is gratefully acknowledged.

Copyright information

© Springer-Verlag 2006