We demonstrate a new technique for trapping, sorting, and manipulating cells and micrometer-sized particles within microfluidic systems, using a diode laser bar. This approach overcomes the scaling limitations of conventional scanned laser traps, while avoiding the computational and optical complexity inherent to holographic optical trapping schemes. The diode laser bar enables us to control a large trapping zone, 1 μm by 100 μm, without the necessity of scanning or altering the phase of the beam.
©2004 Optical Society of America
Current methods for manipulating micrometer-sized colloids and cells with optical trapping require rapid-scanning mirrors [1,2,3,4] or the use of holographic array generators . A common method for creating multiple particle traps is to control a diffraction-limited beam with a piezoelectric mirror, which scans rapidly around an area of interest. The beam jumps from one particle to another in a cycle, but must complete a cycle before Brownian motion significantly displaces the particle . Trapping a single particle and slowly translating the beam transfers momentum to particles, inducing motion. This has been exploited in the creation of optically based microfluidic pumps  and cell sorters . Despite its utility, scanning laser optical trapping (SLOT) is restricted by the piezoelectric elements that translate the mirror. Thus, this technique is limited in scalability in applications such as microfluidics.
A comparable method for manipulating multiple particle systems uses holographic optical tweezers (HOTs). HOTs utilize a single beam and pass it through a diffractive optical element (DOE), usually an LCD . A computer writes patterns to the DOE, which in turn induces phase (and amplitude) modulations in the beam [5,6]. An objective focuses the altered beam where the modulations create a substantial number of traps within the sample. This technique has proven its ability to manipulate many particle arrays in an area on the order of 100 μm by 100 μm, in 2D as well as 3D .
Analogous to the scanning mirror technique, HOTs may be used in optically actuated pumps for microfluidics . The holographic method uses an array of optical vortices, spinning a large number of particles to create fluid flow. Optical vortices are single beam optical gradient force traps created by focusing helical modes of light, and are limited by their great sensitivity to aberrations [9,10].
To overcome the scaling limitations of conventional scanning systems and avoid altering the basic laser mode we developed a new method, utilizing a diode laser bar for the rapid manipulation of multiple particles in microfluidic and biological applications . The standard of controlling micrometer sized particles since the 1970’s concentrates on trapping at single points [12,13]; however, we control objects within a 1 μm by 100 μm line, which are the dimensions of the diode laser bar.
This paper describes the manipulation of particles in static and flowing environments using a simple mask at the intermediate image plane. This simple one to one control scheme, as well as angling the beam with respect to the channel enables us to demonstrate cell sorting. Through the use of diode laser bars we will maneuver vast arrays of independently controlled particles and cells.
2. System schematic
2.1 Basic laser trapping setup
Figure 1 shows our diode laser bar trapping system. The diode laser bar emitter (JDS Uniphase, SDL-6300), capable of producing 3W of average power, is 100 μm by 1 μm centered at a wavelength of 980 nm. The output of the emitter is imaged by a 10x 0.25 numerical aperture (NA) objective. This image is relayed into the sample by an identical 10x, 0.25 NA objective (delivery objective), consequently preserving the original dimensions of the diode laser bar. A 45-degree high reflector (HR at 1064 nm) mirror is located between the two objectives and serves to couple white light into the system from a fiber lamp. The white light source is used to simultaneously view the sample and align the optical trap.
An infinity corrected 40x, 0.65 NA objective (imaging objective) is located above the sample. A 160 mm focal length tube lens located 100 mm above this objective images the sample and trap onto a CCD camera.
2.2 Sample preparation
In order to test the basic trapping capability of the diode bar, water cells containing colloids of varying size were created. A single Parafilm layer is sandwiched between standard thickness glass coverslips to create the cells. The center of the Parafilm layer is removed, creating a well. This well is filled with water and either 1.8 μm colloids, 10 μm diameter colloids or bovine red blood cells. A sample created in this manner is then of the appropriate thickness for both the delivery objective and the imaging objective, minimizing spherical aberration.
3. Trapping and manipulating microscopic objects in a static flow environment
3.1 Stationary trapping of multiple objects
Careful attention to the imaging of the diode laser bar enables the trapping of many colloids and cells simultaneously along the entire length (100 μm) of the trap. In Fig. 2(a), 30 1.8 μm polystyrene colloids (Interfacial Dynamics Corporation, Batch Number 1-2000.1213,1) are shown trapped along the length of the focus. Likewise, Fig. 2(b) shows 6 colloids with a diameter of 10 μm (Molecular Probes Lot # 6981-1) and Fig. 2(c) shows 15 bovine red blood cells trapped along the laser line. This clearly illustrates trapping and alignment of multiple microscopic objects of varying size and refractive index without the use of scanning optics or alteration of the beam. For all three figures the laser was operated at an average power of 0.1 W (measured at the sample).
A transmission mask, inserted at the intermediate image plane between the two 0.25 NA objectives, is used to move particles within the trap. This mask consists of two razorblades mounted on a 1-D translation stage with an adjustable gap between the edges of the blades. The stage moves perpendicular to the beam path, as shown in Fig. 3. The razorblades obstruct the beam, while the small 5 μm wide window, allows only a fraction of the beam to pass through to the sample. We can trap at any point along the length of the trap by simply sliding the mask through this intermediate image plane.
This mask has the capability of manipulating each of the aforementioned samples. Figure 4 shows a 1.8 μm a polystyrene colloid manually translated the length of the laser trap. Figure 5 shows a 10 μm colloid, and Fig. 6 shows a bovine red blood cell, manually moved along the trapping zone. In these image series, the laser was operated at an average power of 0.28 W, which resulted in a transmission of 1.4 mW through the 5 μm window (measured at the sample).
4. Manipulation of microscopic objects in microfluidic channels
4.1 Microfluidic trapping and imaging system
For trapping and manipulating objects in microfluidic channels, a setup incorporating a Nikon Optiphot microscope was employed. The side port of the Nikon was used in conjunction with a 1x to 7x optical zoom CCD camera to align and image the system. The diode laser bar was once again imaged one-to-one into the sample.
4.2 Microfluidic Sample
Microfluidic networks were fabricated in polydimethylsiloxane (PDMS) using soft-lithography , a technique pioneered by the Whitesides group at Harvard University. This method enables microfluidic networks to be quickly replicated from a permanent, reusable master with fidelity on the order of single nanometers and a feature size less than 100 nm. The networks are created by first transferring the pattern of a shadow mask to a negative photoresist film (SU-8 50, MicroChem Corp., Newton, MA) spun upon a silicon wafer to a depth of approximately 20 μm. A two-part mixture of PDMS (Sylgard 184, Dow Chemical, Midland, MI) is then poured and cured upon the silicon master to produce an optically transparent replica. A glass coverslip was bonded to the PDMS channel network by a brief exposure to oxygen plasma, producing an irreversible seal .
The microfluidic system was filled with an aqueous suspension of 9 μm polystyrene colloids as well as dilute bovine red blood cells. As seen in Fig. 7, the colloids flow freely through the channel until they become trapped in the laser line. Since the trap line is oriented at an angle with respect to the direction of flow, the particles move like an “optical conveyor belt”; they enter at one part of the trap, flow down the line, and are released at the end of the trap. A gradient in the fluid velocity profile pushes the colloids or cells along the entire length of the trap, until finally driving them out at the end.
We also actively manipulate objects within a flowing environment by incorporating the razorblade mask located at the intermediate image plane, as in the static trapping experiments. We are able to trap a single colloid and move it along the length of the trapping zone as other particles flow by in the channel (See Fig. 8). This demonstrates that one-to-one amplitude masks make it possible to manipulate trapped objects over large distances in a straightforward manner.
4.3 Cell sorting
In addition to manipulating polystyrene colloids, we control bovine red blood cells in our microfluidic network. Figure 9 demonstrates the optical conveyor belt phenomena with the blood cells. As shown in the movie, the trap collects all of the cells entering it, leaving a clear path down stream. Notably, this is the first optical sorting technique used for cells that does not require scanning or altering the phase or amplitude of the trapping beam in any way.
We can also release aliquots of cells by blocking the beam (See Fig. 10). When the beam is on in the channel, the cells trap, flow, and release into the desired region.
One way to send individual cells to a specific destination is to allow the beam to translate a cell to a specific point in the channel, and then block the beam. When the cell releases from the trap, laminar flow and slow diffusion keep the cell in the same streamline, thus determining which outlet it enters (See Fig. 11). In Fig. 12 bovine blood cells enter the channel at the edge. When a cell encounters the optical trap it is pulled away from the wall until reaching the desired position. We then block the beam, releasing the cell into the desired streamline. Keeping the beam blocked, the second cell in the image series stays at the bottom edge of the channel.
We also use a single razorblade to block one edge of the beam as another way to control the placement of cells. This restrains the length of the trap in the channel, thus releasing the cells into desired streamlines. An example of this technique can be seen in Fig. 13. When the first five cells encounter the trap it is completely open and the cells flow to the center of the channel. When the next cell reaches the trap the razorblade covers one half of the beam, directing this cell to a lower part of the channel. The final two cells arrive at the trap when the razorblade covers nearly the entire beam. This results in only a slight deflection from the initial streamline. One can clearly see that the different cells are sent to different, chosen, streamlines.
In conclusion, we demonstrate the use of a diode laser bar to trap, manipulate, and sort a variety of microscopic objects that vary in size and refractive index. We control movement and trapping of these objects in stationary and flowing environments. Notably, it was shown that objects could be positioned into targeted streamlines by angling the trap with respect to the channel and blocking/unblocking different sections of the beam. Transmission of the trap is in all instances controlled by a simple amplitude mask.
This new laser trapping technique is significant in that it addresses design issues that presently prohibit the development of truly practical, economical, optically-actuated microfluidic cell sorters and analyzers. Only a fraction of the available laser power was used, indicating that a single diode laser bar could be multiplexed and used to simultaneously drive flow and manipulate objects over a vast array of channels. Manipulative tasks within the microfluidic system are clearly a straightforward process using this laser diode geometry. For many of these tasks scanning can be entirely eliminated, and no complex holographic processes to reshape and control the beam location are required.
This work was supported by the National Institutes of Health under grant R21 EB001722-01.
1. C. Mio, T. Gong, A. Terray, and D. W. M. Marr, “Design of a scanning laser optical trap for multiparticle manipulation,” Rev. Sci. Instrum. 71, 2196–2200 (2000). [CrossRef]
3. C. Mio and D. W. M. Marr, “Optical trapping for the manipulation of colloidal particles,” Adv. Mat. 12, 917 (2000). [CrossRef]
4. J. Oakey, J. Allely, and D. W. M. Marr, “Laminar-flow-based separations at the microscale,” Biotechno. Prog. 18, 1439–1442 (2002). [CrossRef]
5. E. R. Dufresne and D. Grier, “Optical tweezer arrays and optical substrates created with diffractive optics,” Rev. Sci. Inst. 69, 1974–1977 (1998). [CrossRef]
6. E. R. Dufresne, G. C. Spalding, M. T. Dearing, S. A. Sheets, and D. G. Grier, “Computer-generated holographic optical tweezer arrays,” Rev. Sci. Instrum. 72, 1810–1816 (2001). [CrossRef]
10. J. Curtis and D. G. Grier, “Structure of optical vortices,” Phys. Rev. Let. 90, 133901 (2003). [CrossRef]
11. K. Svoboda and S. M. Block, “Biological applications of optical forces,” An. Rev. Biophys. and Biomol. Struc. 23, 247–285 (1994). [CrossRef]
12. A. Ashkin, “Optical levitation by radiation pressure,” Appl. Phys. Let. 19, 283–285 (1971). [CrossRef]
14. Y. Xia and G.M. Whitesides, “Soft lithography,” An. Rev. Mat. Sci. 28, 153–184 (1998). [CrossRef]
15. D. C. Duffy, O J. A. Schueller, S. T. Brittain, and G. M. Whitesides, “Rapid prototyping of microfluidic switches in poly(dimethyl siloxane) and their actuation by electro-osmotic flow,” J. Micromech. Microeng. 9, 211–217 (1999). [CrossRef]