HK1145335B - Cell sorting system and methods - Google Patents

Description

Cell sorting system and method

Technical Field

The present invention relates to methods and devices for using forces within a microfluidic channel network to provide a switch capable of selectively sorting target cells through the network to separate the target cells from non-target cells and collect them. Of particular interest are optical switching forces or fluid switching forces.

RELATED APPLICATIONS

This application claims priority from U.S. application No.11/781,848 filed on 23/7/2007, and from U.S. provisional application No.60/925.563 filed on 20/4/2007.

Background

Traditional Fluorescence Activated Cell Sorter (FACS) is widely used in research and clinical applications 1. These instruments are capable of very rapid, multi-parameter analysis and sorting, but generally require large sample volumes, require trained operators for operation and maintenance, and are difficult to sterilize. FACS instruments are capable of analyzing as few as 10,000 and up to tens of millions of cells. However, below 100,000 cells, sorting performance is weakened¹. Other separation methods such as magnetic beads do not require as many cells as FACS, but they suffer from non-specificitySexual binding, cell and bead aggregation, and the possibility that the beads themselves may interfere with subsequent processing steps. Thus, to sort valuable small samples or cells from primary tissues, a cell sorter that is capable of handling small sample volumes containing low cell numbers and allows for efficient recovery of sorted cell populations solves a unique scientific problem.

Microfabricated cytometers are capable of sorting as few as 1,000 cells while consuming less reagents in a closed system that is easy to use. The latter is important because, unlike conventional FACS instruments, no aerosol is generated, the risk of contamination of the sorted cells is reduced, and the risk of handling with biohazardous materials is reduced. Several microfabricated cell sorters have been described, but most are "proof of concept". Fu, et al²A 30-fold enrichment of escherichia coli (e.coli) at a flux of 17 cells/sec was reported. Only 20% of the bacteria survived after sorting and the sorting purity in the target reservoir was 30%. In the subsequent study³In (b), the flux increased to 44 cells/sec, but the target purity decreased to less than 10%, with a recovery reported at 39%. Wolff, et al⁴Beads (beads) can be separated from chicken erythrocytes at a throughput of 12,000 events/sec, achieving 100-fold enrichment. However, the purity within the target well was about 1%. In these studies, enrichment was defined as the increase in concentration of the target population in the collection wells compared to the starting concentration. Purity indicates the accuracy of sorting and is the percentage of sorted target cells sorted into the collection well relative to all cells. Recovery is defined as the ratio of the number of cells counted by the fluorescence detector to the number of cells recovered from the collection well. The latter two studies use pressure switches in the microfluidic device that switch the entire flow path and thus any particles contained within the fluidic plug. Mechanical compliance in these switches results in the fluid switching speed being the rate limiting step in throughput³. Electrokinetic flow control, e.g. electroosmosis, has also been reported^2，5，6Or dielectrophoresis^7，8，9But high electric field gradient and physicalization of buffered ionic strengthChemical constraints are non-ideal conditions for cells.

Buica et al⁹It was first proposed to use optical forces for deflecting particles through a fluid channel. The force exerted on the particles by the light beam is determined by the optical power and the relative optical properties of the particles and their surrounding fluid medium. For biological cells having a diameter of about 10 μm, a force of about 1pN/mW can be obtained. Although the optical forces are small, the force necessary to deflect cells into an adjacent flow stream is also small, e.g., 900pN moves 10 μm diameter cells, 20-40 μm cells traverse the stream in a few milliseconds. The force is the force necessary to overcome the viscous drag experienced by the cell at the velocities involved in this lateral movement.

See, e.g., U.S. patent 6,744,038, which is incorporated herein by reference in its entirety, for principles of optical force and general background.

A variety of air pressure modulation systems for particle sorting in microfluidic devices are known in the art. By alternating the gas pressure contacting the microfluidic device, the cell-containing particles flowing in the microfluidic channel can be directed to the desired branch or branches, thereby allowing for a low-cost sorting function.

Disclosure of Invention

These forces are used to effect switching within the microfluidic channel network, operable as a cell sorting system, as described below. The switching is triggered by detecting a fluorescent signal from a target cell flowing within the microfluidic channel network upstream of the switch location, although other detection means such as light scattering may be used equally to activate the switch. Switching is used to direct cells or particles into one of the multiple outlet flow streams without altering the basic flow, thereby collecting the desired cells for other applications. It is desirable that the flow within the microfluidic channel is typically laminar at a very low reynolds number. Thus, any cell flowing in a particular laminar layer or flow stream will stay in that flow stream in the absence of any force transverse to that laminar layer. This is achieved by the switch exerting a force on the cell which is displaced laterally from the laminar flow layer to move the cell from a flow stream exiting the bifurcation through one outlet to a flow stream exiting the bifurcation through a second outlet.

In one embodiment, the cell sorter comprises: a cell inlet adapted to receive one or more cells in a fluid medium; first and second buffer inlets fluidly connected to the cell inlet to provide a buffer solution to the sorter; a fluid channel fluidly connected to the cell inlet and the first and second buffer inlets; a first cross flow channel fluidly connected to the fluid channel; first and second outlets fluidically connected to the flow channel, the outlets being downstream of the flow channel, a detector adapted to detect cells of a given state and to generate a signal in response to the cells of the given state, the detector being arranged to detect cells at a location upstream of the first flow channel, a transverse force switch connected to the detector and actuatable in response to the signal to cause fluid to move within the flow channel; such that when a cell of a given state is detected, the lateral force switch is activated to provide a lateral force to the cell, thereby moving the cell so that it selectively exits and enters the first or second outlet.

The present invention details the methodology for generating the switch and the method for optimizing the switch, the design of the microfluidic channel network and the properties of the cell or particle flow in the microfluidic network in the following paragraphs in order to achieve enhanced sorting performance. In the case of optical switching, optical switching generally works by projecting an optical illumination field into the microfluidic channel network, near the cell tracks of established flow within the microfluidic channel. The interaction of the cell with the optical field generates a force on the cell that causes the cell to be laterally transferred to the established flow rate, such that the cell moves from one flow stream to another at the established flow rate without cell entrapment or without significantly altering the movement of the cell at the original flow rate.

In the following, the terms cell and particle are both understood to mean any biological cell, biological particle, natural organic or inorganic particle and man-made organic or inorganic particle. The size range of the cells sorted in the microfluidic channel network is typically that of biological cells, with diameters of about 1 micron to about 50 microns. More generally, cells having a diameter of about 100 nanometers to about 100 micrometers are candidates for sorting by means of valves of a microfluidic channel network.

In one embodiment, an optical switch is used. Typically, lasers have been used to generate light beams for use in optical switches. The laser currently used for optical switches is a near infrared continuous wave laser, which is known to not harm the viability of biological cells at the energy density and exposure time used to demonstrate optical switching. Alternative laser sources may be considered for different applications if damage to the particles is not an issue, including visible or near ultraviolet wavelength lasers, or pulsed lasers where a large optical flow can be used to move the particles very rapidly. However, the beam source is not limited to lasers, even though other discussions of the present invention use lasers to produce optical switching.

In yet another embodiment, a fluid switch may be employed. Preferably, a pneumatic based fluid switch may be utilized. In another embodiment, a microfluidic wafer designed for a cell sorter uses air pressure modulation. Single-sided channels for flow switching or cell sorting may be used within a microfluidic device. The channel geometry provides for efficient pneumatic flow switching within the microfluidic device. In yet another aspect, a pair of double switching valves on a single-sided channel minimizes pneumatic switching response time. In yet another aspect, a cartridge holder is provided for pneumatic control of a microfluidic wafer.

With respect to specific applications of the fluid switching system, the fluid sorter switching system comprises: an inlet adapted to receive a fluid medium; a fluid channel fluidly connected to the inlet; first and second cross flow channels fluidly connected to the fluid channel; first and second pneumatic valves connected to the first and second cross flow passages; first and second outlets fluidly connected to the fluid channel; these outlets are located downstream of the cross flow channel; and a control system connected to the first and second valves, the control system providing timed control signals to actuate the first and second pneumatic valves, characterized in that the first valve opens before the second valve opens. In yet another aspect, the switching is further characterized in that the first valve is closed after the second valve is opened and before the second valve is closed. In yet another aspect, the second cross-flow channel and the second pneumatic valve may be eliminated to enable sorting using a single side channel with 1 or 2 pneumatic valves.

In yet another embodiment, the microfluidic cell sorter uses optical switching and pneumatic pressure modulation. Optical and opto-mechanical designs are also provided. Preferably, a dual laser illumination module and a dark illumination module are used. A high efficiency high NA (numerical aperture) objective lens is preferably used to collect the fluorescence.

In other aspects of the invention, fluorescence signal detection and processing systems and methods are employed. An ADC is preferably used to convert the analog signal to a digital signal. In one embodiment, the digital signal processing algorithm is implemented in an FPGA.

Optionally, self-alignment of the cartridge is provided. The present invention solves the problem of self-alignment of the cassette loaded in the system.

In other aspects, the invention relates to application software designed for microfluidic sorter instruments. Raw data is shunted from an external device between multiple executable applications and processes in a recordable, lossless, and high-speed (10 megabits/second or higher) format.

In other embodiments, a microfluidic cartridge loading station is provided. A filling station for filling a microfluidic device is described.

In other aspects, dilution cloning methods for growth monitoring and selection of cells, such as fetal cells, are provided. The present invention describes a dilution cloning method for growth monitoring and selection of said cells.

Drawings

Figure 1 is a plan view of a "Y" shaped sorting junction in a microfluidic channel network.

Fig. 2 is a plan view of a microfluidic channel network incorporating sheath flow-pinching junctions and "Y" -shaped sorting junctions connected by a main channel and cells in the flow with 50/50 shunts, collectively referred to as a 50/50 optical network.

Fig. 3a and 3b are plan views of microfluidic channel networks incorporating sheath flow pinching junctions and "Y" type sorting junctions connected by primary channels and cells in the flow being differentially shunted by differential sheath flow rates, which together with optical switches are referred to as sheath flow deflecting optical networks.

Fig. 4a and 4b are plan views of microfluidic channel networks incorporating sheath flow pinching junctions and "Y" type sorting junctions connected by primary channels and cells in the streams being skewed by differential exit widths, which together with optical switches are referred to as sheath flow skewed optical switch networks.

Fig. 5 is an 50/50 optical switch network with a bi-directional laser line optical switch.

Fig. 6 is a 50/50 optical switch network with bi-directional laser spot optical switches.

Fig. 7a, 7b and 7c are plan views of laser line optical switches in a larger microfluidic channel network with more than two outlets.

Fig. 8a, 8b and 8c show optical designs designed for modulating and/or interrupting an optical switch.

Fig. 9a and 9b are plan views of sheath flow deflecting optical switch networks of laser spot optical switches displaced parallel to or at an angle to the cell flow.

10a, 10b, 10c and 10d show detector arrangements and timing/trigger diagrams using a single laser source and trigger determination method for cell detection.

11a, 11b, 11c and 11d show a detector arrangement and timing/trigger diagram using two laser sources and a trigger determination method for cell detection.

Fig. 12 is an illustration of a representative design of a photolithographic mask for microfluidic channel networks in both bottom and top glass substrates that provide 2-dimensional sheath flow pinching of cell flow within the main channel when the substrates are combined to form a single network.

FIG. 13 shows a 3-dimensional example of the design described in FIG. 12.

Fig. 14 is an illustration of a side view of a microfluidic channel network that provides sequential sheath flow pinching of cell flow in a vertical direction and then in a horizontal direction, resulting in complete 2-dimensional sheath flow pinching of cell flow within the main channel.

Fig. 15 is a 3-dimensional example of the microfluidic channel network described in fig. 14.

Fig. 16 is an illustration of a representative photolithographic mask design for both the bottom and top glass substrates when bonded together to form the microfluidic channel network shown in fig. 14 and 15.

FIG. 17 is a representative embodiment of a photolithographic mask for a complete microfluidic channel network having T-pinching junctions and T-diverging junctions connected to outlets for implementing an optical switch-based cell sorting method.

FIG. 18 is a representative embodiment of a photolithographic mask for a complete microfluidic channel network having triangular-shaped pinched junctions and Y-forked junctions connected to outlets for implementing an optical switch-based cell sorting method.

Figure 19 shows a preferred embodiment of a microfluidic channel network in a complete microfluidic cell sorting wafer.

Figure 20 shows a preferred embodiment of a self-contained disposable cartridge for an optical switch-based microfluidic channel network cell sorter.

Figure 21 shows a preferred embodiment of an optical system for an optical switch-based microfluidic channel network cell sorter.

Figure 22 shows representative performance of an optical switch-based microfluidic channel network cell sorter for various implementations of optical switches.

Figure 23 is a plan view of a prior art "Y" shaped sorting junction in a microfluidic channel network showing pneumatically based switched fluid particle trajectories.

Fig. 24a and 24b are plan views of "Y" -shaped sorting junctions in microfluidic channels with two-channel fluidic switches showing possible particle trajectories at different pneumatic partial pressures.

Fig. 25a and 25b are plan views of "Y" -shaped sorting junctions in microfluidic channels with single-channel fluidic switches showing possible particle trajectories at different pneumatic partial pressures.

Figure 26 shows a preferred embodiment of a "Y" shaped sorting junction in a microfluidic channel with a single channel fluidic switch.

Fig. 27 shows a pair of double switching valves for minimizing pneumatic switching response time and controlling pneumatic mode in a single channel fluid switch.

Fig. 28a and 28b show a pair of double-switching valves and a graph of flow across the channels as a function of time to minimize pneumatic switching response time and control pneumatic modes in a two-channel fluid switch.

Fig. 29 is a cassette holder for pneumatic control of microfluidic wafers.

FIG. 30 shows fluorescence calibration bead measurements performed on the instrument described herein.

Fig. 31a shows a fluorescence microscope image of Jurkat cells stained with DAP1-, and fig. 31b shows a fluorescence microscope image of target Jurkat cells with CellTracker Green.

Figure 32 shows a station for adding microfluidic cartridges. A filling station for filling a microfluidic device is described.

FIG. 33 shows a dilution cloning method for growth monitoring and fetal cell selection. The present invention describes a dilution cloning method for growth monitoring and selection of foetal cells.

Fig. 34 is a shallow microfluidic channel for 1-dimensional flow concentration. The present invention describes a method to overcome the inherent velocity dispersion in 1-dimensional flow concentration for microfluidic flow cytometry or sorting.

Fig. 35 shows cell/particle detection within a multi-phase microfluidic droplet. Several methods for detecting cell/particle counts within a multi-phase fluid droplet are described.

Fig. 36 is a microfluidic outlet port configuration for efficient sample collection. The present invention describes methods of several sample collections after cell or particle sorting using microfluidic devices.

Fig. 37 is a microfluidic device for multiple sample analysis or sorting. Several microfluidic devices for multiple sample analysis or sorting are described.

Fig. 38 is a microfluidic analysis wafer for sample analysis and recovery. Several microfluidic devices for sample analysis and recovery at lower dilutions are described.

Detailed Description

Figure 1 shows one embodiment of an optical switch 10 for sorting cells in a 1 x 2 microfluidic channel network (i.e., a network having one primary input channel 11 and two outlets 12 and 13 extending from a bifurcated junction). The "Y" geometry of the bifurcation junction is shown in fig. 1, but other shaped bifurcations such as the "T" geometry may also be used. Typically, these microfluidic channels are formed in an optically transparent substrate so that optical switches and other cell detection optics can be projected into the channels. The substrate is typically, but not limited to, glass, quartz, plastics such as Polymethylmethacrylate (PMMA), and the like, and other castable or workable polymers (e.g., polydimethylsiloxane, PDMS, or SU 8). The depth of the microfluidic channel is typically, but not limited to, 10 μm to 100 μm. The width of the microfluidic channel is typically, but not limited to, 1 to 5 times the depth. In the case of making microfluidic channels by photolithographic masking of a glass substrate followed by isotropic etching of the channels, the cross-section is typically rectangular, or rectangular with quarter-corners rounded.

The flow conditions are set such that when the light beam (in this case from a laser) is switched off or blocked so that the light beam is not incident on the junction area, all cells will preferably flow into one of the outlets, for example the outlet 13 on the right. When the beam is turned on or switched on, the beam is incident on the junction area and the cell is directed into the left outlet 12 by the optical forces created by the cell's interaction with the beam. In this embodiment, the optical pattern selected to direct the cells is an elongated line of laser illumination at an angle relative to the direction of fluid flow. Optical gradient forces (Optical gradient forces) cause the cells to leave the main stream line of cells laterally, so that the switched cells then leave the main channel into one outlet, e.g. 12, while the unswitched cells leave the main stream of cells and enter another outlet, e.g. 13. Setting and controlling of flow conditions in a microfluidic channel network can be accomplished by direct drive pumps, pneumatic pumping, electrodynamics, capillary action, gravity, or other means by which fluid flow can be generated.

The performance of the sorting mechanism in terms of throughput (the time rate at which cells enter the sorting region at the top of the bifurcation junction), yield (the fraction of target cells within the target outlet 12) and purity (the ratio of the number of target cells/total number of cells within the target outlet 12) is affected by various factors, each of which affects the execution of the optical switch. The optical switch may be characterized by several parameters such as: the shape of the optical pattern projected into the sorting junction area of the microfluidic channel network, the position of the pattern relative to the diverging junctions, any movement of the optical pattern relative to its original position and shape, the duration of the optical switch activation, the wavelength and power of the laser source used to generate the optical switch pattern, and the like. The choice of the specific values of these parameters of the optical switch is determined, inter alia, by the following factors: the topology and geometry of the microfluidic channel system, the flow rate (cell velocity) within the microchannel system, the ability to control the position of the cell flowing within the primary channel (whether the cell flows in the center or off to one side of the primary channel), the amount of cell displacement necessary to achieve reliable switching, the depth of the channel, the shape of the channel, and the force generated by the cell interacting with the optical switch.

Generally, when cells are introduced into a flow located within a primary channel, they may move down the channel at any lateral position within the flow. Thus, depending on the lateral position of the cell, the cell may move at different velocities due to the well-known parabolic (for cylindrical microfluidic channels) or quasi-parabolic (for more general cross-sections) velocity profile of the pressure-driven flow within the microfluidic channel. This would make it difficult to bias all of the cell flow to one outlet, 13, as shown in figure 1. Any optical switching using this flow geometry will necessarily result in low fluence and inefficient use of the laser power available to produce the optical switching. Any fluid switching using this flow geometry will result in low throughput, purity and yield due to variable cell flight time between the detection and switching regions. Using appropriate flow conditions can help alleviate these limitations on the performance of optical or fluidic switching.

Appropriate flow conditions can be established in many ways. In one embodiment, pinching of the cell input channel stream 20 by adding buffer streams from both the left 21 and right 22 sides of the cell input channel stream 20, using the sheath flow scheme as shown in FIG. 2, achieves a single row (horizontally in plan view) of cells in the central 1-dimension of the main channel. By having equal flow from each side it is achieved that the cells are kept in the center of the main channel. This flow effectively creates a fluid diversion surface 23, as shown in FIG. 2, and which will ultimately create a diversion of fluid and cells at the bifurcation junction 50/50. Optical switching to sort target cells from a mixed population of cells using this microfluidic channel design and flow conditions requires an optical switch that actively switches target cells to one outlet, 12 as shown in figure 1, and non-target cells to another outlet, 13 as shown in figure 1. Fluid switching to sort target cells from a mixed population of cells using this microfluidic channel design and flow conditions requires fluid switching that actively switches target cells to one outlet, 12 shown in fig. 1, and non-target cells to another outlet, 13 shown in fig. 1. One example of such an implementation is a two-channel fluid switch as in fig. 24.

Alternatively, the concentrated cell line location can be offset from the center of the primary channel by applying unequal flows to the sides of the sheath flow channel, as shown in FIGS. 3 a-b. This effectively causes a cell diagonal flow deviating from the input channel 30 to the side of the flow dividing surface 33 in the main channel. The side of the main channel where the cell flow is inclined is opposite to the side where the sheath flow has a higher flow velocity. That is, when the right sheath buffer 32 flows faster than the left sheath buffer 31, the cell line is deflected in the main channel towards the left side of the flow, as shown in fig. 3 a-b. However, the left sheath flow may also have a higher flow rate, thereby pushing the cell lines to deflect towards the right side of the main channel. Fig. 3a-b also show fluorescence detector 34 and optical switch 35. The fluorescence detector is used as a mechanism to determine which cells are to be sorted and will be discussed in detail below. As is apparent from fig. 3b, efficient sorting involves moving cells across the plane of the split, from the flow stream exiting the bifurcation junction to the fluorescence negative non-target cell microfluidic channel 36 into the flow stream exiting the bifurcation junction to the fluorescence positive target cell microfluidic channel 37. Manipulation of the sheath buffer flow rate can be achieved by: the flow rates in the respective side channels are individually controlled using direct drive pumps, pneumatic pumping, electrodynamics, capillary action, gravity or other means of generating fluid flow, or by carefully balancing the pressure drop in each microfluidic channel by specifically designing the microfluidic sheath network to ensure that a central flow (50/50 split) or bias flow occurs.

An alternative means of achieving that all cells flow preferentially from the input stream 40 in the primary channel into one output microfluidic channel, the fluorescence negative channel 46, prior to fluorescence detection 44 is: central pinching is achieved using equal sheath buffer flow rates 41 and 42, and then cells are preferentially biased into the fluorescence negative channel by having a larger volume of fluid flow from the bifurcation junction relative to the fluorescence positive outlet 47 into the fluorescence negative outlet 46. This approach is illustrated in fig. 4a-b, where the left outlet 46 is wider than the right outlet 47. This configuration effectively positions the diverging surface 43 to the right of the centrally located cell stream. Thus, when the cell is in the desired position, the optical or fluidic switch 45 is then used to transfer the target cell across the shunt face into the right-hand fluorescence positive exit of the target cell. This segment is equally effective by the right outlet being wider than the left outlet, whereby the target cell is diverted by an optical or fluidic switch across the shunt face, which is now to the left of the centrally located cell stream and is thus sorted into the left outlet. Thus, by specifically designing the microfluidic channel outlet network, or by actively controlling the outlet back-pressure within the respective outlet, the flow of cells into the desired outlet can be controlled.

Whether a central flow or an offset flow is used, the respective distances of the concentrated cell flow from the fluid splitting plane ultimately dictate the magnitude of displacement of the cells necessary to achieve reliable switching. This further dictates the length of the laser line and laser power required to achieve reliable optical switching or the amplitude and duration of the pneumatic pulse required to achieve reliable fluid switching. The closer the cell flow is to the shunting surface, the shorter the required displacement and the more efficient the sorting process becomes. For purity enhancement of the sorted population and for high throughput, a single switch arranged unidirectionally, either as an optical switch or a fluidic switch, requires the sample flow to be offset from the plane of the split stream. In this way, the incidence of erroneous sorting is minimized. For samples of non-uniform size, such as single cell suspensions of primary tissues where the diameter of cells and debris can vary from 1 micron to 50 microns, it is advantageous to facilitate greater deviation at the expense of throughput. For more homogeneous samples, such as cell lines or polystyrene beads, smaller deviations can be selected to allow for greater throughput.

An alternative to the design is to use a bi-directional optical switch that employs two laser lines. With this approach, one laser line sorts the desired cells into one outlet and the other laser line sorts all other cells into the other outlet. This segment may be used with the 50/50 split configuration shown in fig. 2 or the offset split configuration shown in fig. 3 and 4. In the latter case, when the cell is not in the switching region, it is possible to select either one of the two positions of the laser to turn on the laser, or also to interrupt the laser during this period. The optical switch may also be made as a bi-directional optical switch as follows: there are two mirrored laser lines incident on the switching region just above the bifurcation junction, which are independently turned on to direct the cell to either of the two outlets emanating from the bifurcation junction.

A schematic of a bi-directional optical switch using laser lines in a 1 x 2 microfluidic network is shown in fig. 5. A similar bi-directional optical switch can also be implemented using laser spots directed to either side of the channel, as shown in fig. 6. As with the unidirectional optical switch, a single laser source may be used in the bidirectional optical switch, or the bidirectional optical switch may use two separate laser sources. Bi-directional designs may provide certain performance benefits over unidirectional designs. The first benefit is that purity can be maximized because each cell is guided by the laser. A second benefit is that fluid flow is simplified because equal flow can be directed out of each of the two output ports instead of using a certain predetermined flow ratio.

Although only a 1 × 2 microfluidic channel design flowing through one input main channel into a bifurcation to two outlets has been considered so far in this specification, microfluidic networks with 1 × N or M × N outputs may also be used ". In these larger networks, optical or fluidic switching can be achieved by having an arbitrarily large number of independently modulated laser lines or independent cross flow channels. Some embodiments are shown in figures 7 a-c. In addition, cells may also be fed back multiple times through the same sorter to increase the purity of the sorted cells, or channels may also be arranged for a cascade of multiple stages of sorting.

When operating an optical switch arranged unidirectionally or bidirectionally, two different activation modes can be considered: passive or active. The passive approach is such that the state of the optical switch is either on or off, regardless of what cells can flow through the channel. In this case, it is not necessary to know when or how many cells enter the switching region, and therefore, all cells in the switching region are switched depending on the state of the laser light. Alternatively, in the active mode, the cells are first detected as they enter the detection/selection zone and then switched according to a certain decision process. Fig. 3a-b and fig. 4a-b show examples of this approach using a fluorescent detector arranged just before the switching region. In this case, all fluorescent cells are directed to one outlet and all non-fluorescent cells are directed to the other outlet. Other non-fluorescent detection/selection techniques for the determination process include Time-Of-Flight (Time-Of-Flight), scattering, imaging, capacitance, or any detection means that can identify the desired cell. Regardless of the detection/selection method, switching using an active approach may be employed to sort one cell population from another according to a certain decision process.

In order to utilize an active approach, the beam must be modulated to be on or off in response to the decision process. Regardless of the number of lasers used, or whether the optical switch is unidirectional or bidirectional, the lasers may be modulated in a variety of ways including using electro-optic modulators, modulating laser power, chopping lasers, using liquid crystal modulators, using galvanometers, and using acousto-optic modulators. For a bi-directional optical switch using two lasers, the individual lasers can be independently turned on or off; however, when a single laser source is used, two different orientations of the optical switch lines can be achieved by using a polarization rotator (such as a liquid crystal modulator) and each of the two different line patterns is each of two separate polarizations. Similarly, an acousto-optic modulator or galvanometer mirror may be used to adjust the position of a single spot used as an optical switch, or a dual axis acousto-optic modulator or dual axis galvanometer mirror may be used to map two different line shapes used as a bi-directional optical switch.

Fig. 8 shows three different possible optical designs for adjustment and/or interruption of the optical switch. In fig. 8a, a bidirectional optical switch is fabricated from a single beam (laser) directed toward and through a Liquid Crystal Modulator (LCM). The LCM is a polarization rotator, so if the beam is polarized in one direction, it will pass straight through a Polarizing Beam Splitter (PBS), through a cylindrical lens to create a line shape, through another PBS, and then through some focusing optics that focus the line onto the microfluidic switching region. This effectively creates a single line, bi-directional optical switch that is used to switch the cell into one of the bifurcated channel outputs. In order to switch the cell into another outlet, a mirror line must be created. This is achieved by rotating the LCM which changes the polarization of the beam. Thus, when the beam is incident on the first PBS, it is directed into another path: it passes through a different cylindrical lens (creating a line shape), through another PBS, which directs the beam back through focusing optics that focus the mirrored line onto the microfluidic switching region. It is noted that cylindrical lenses are used to create line shapes for bi-directional optical switches; alternatively, the cylindrical lens can be eliminated, producing a light spot for optical switching. In fig. 8b, instead of using a combination of LCM and PBS, an acousto-optic modulator (AOM) can be used, with or without cylindrical lenses, to generate a line or spot for use in the bi-directional optical switch. This is achieved by constructing the AOM to achieve the desired line shape required. In addition, the AOM can be used to chop the beam on/off, directing the beam to a stop for the closed condition of the optical switch. Fig. 8c shows a combination of the systems described in fig. 8a and 8 b. Instead of an AOM, any configuration may be used that employs an AOM to change the direction of the beam, employing a galvanometer mirror, either uniaxial or biaxial, depending on the desired beam motion.

Many variations of the optical pattern may be considered when achieving the best switching efficiency of a unidirectional or bidirectional optical switch. As described above, laser lines have been used as optical switching patterns. The laser line can be generated by the following technique: cylindrical lenses, scanning galvanometer mirrors or acousto-optic modulators, diffractive optics, custom refractive optics, or any other technique. Heretofore, cylindrical lenses have been used to generate laser lines by scanning galvanometers or by using acousto-optic modulators. The length of the laser line may be any length or may be as short as a single point. The laser line may have a higher intensity at the top of the line and a gradually decreasing intensity towards the end of the line. In addition, the laser line may be curved in an arc that optimizes the output direction of the cell. In addition, the angle or shape of the laser line can be changed in real time (i.e., rotated to achieve optimization of the output). For implementations using multiple exits, it is possible to generate any arbitrary pattern of 2D spatial laser lines to achieve directional optimization of each output cell. Alternatively, the laser lines may be generated by an array of discrete spots.

To further improve the performance of the sorting mechanism in terms of throughput, yield efficiency and purity, optical switches have been constructed so that the laser spot scans side-by-side with the selected cell as it flows down the main channel towards the bifurcation junction, thereby increasing the total interaction time between the cell and the laser. The laser spot used by the optical switch is translated in a straight line down the length of the main channel toward the bifurcation junction. The lines scanned by the spots can be parallel to the walls of the main channel (fig. 9a) or can be at an angle with respect to the cell flow stream (fig. 9 b). Thus, the angle may be 0-90 degrees. The ability to scan the spot is achieved using an AOM or a scanning galvanometer mirror. The determination based on the detection of the desired cell using fluorescence or other detection means that can identify the desired cell, such as time-of-flight, scattering, imaging, or capacitance techniques, triggers the optical switch to scan. The cell position may be an offset position or a central position in the main channel, which dictates the length of the line scanned by the spot and the laser power used to achieve efficient switching/sorting. Thus, when the desired cell is detected, the optical switch is turned on and the light spot appears alongside the desired cell. The spot then follows a trajectory alongside the selected cell and uses optical forces to direct the selected cell into the desired exit.

Two means of assisting in the efficient triggering of an optical switch or a fluidic switch are described below. Both methods typically use a time signal (temporal signal) to analyze the moving cells and use this information to generate a decision to switch or not switch. The time signal is basically a measure of the signal as a function of time, which makes it possible to obtain a characteristic time fingerprint in terms of both peak intensity and peak width. The signal may be fluorescence, scattering (e.g., forward scattering), capacitance, imaging or any means by which the desired cell can be identified. One approach is to employ a single laser source in combination with two or more detectors to achieve both cell detection and cell identification. Fig. 10a-d show this approach using a laser source in combination with a fluorescence detector and a forward scatter detector. The time signals obtained by these detectors are used as information for handover decision. The presence of cells is detected by a forward scatter signal and when the signal is combined with a fluorescence signal intensity within a predetermined range; this "gating" information is used to trigger the optical switch. It is noted that only a single fluorescence detector is shown, however, multiple fluorescence detectors may be used for further refinement of cell identification. In the case described, the cell flow is centered with equal flow rate sheath buffer, and the use of outlets with different widths creates a splitting plane on the right side of the cell flow. However, as noted above, any configuration may be used to manipulate the position of the cell flow and the plane of flow separation. In addition, both configurations present an error detector that checks whether the cells are switched or not. In this case, detection may be based on fluorescence, scattering (e.g., forward scattering), capacitance, imaging, or any means by which the desired cells may be identified.

Fig. 10a-b show detector arrangements and timing/trigger diagrams when the sort parameter is negative and the optical switch or fluidic switch is not triggered. These cells enter the main fluid channel and are collected into a single row by sheath buffer from both sides. As the cell travels through the laser in the detection/selection region, a fluorescent signal and a forward scatter signal are detected at or near the same time. Although the presence of cells was successfully detected by the forward scatter signal (at time t)₁) However, the fluorescence signal is below the gating level and the optical switch is not triggered (at time t)₂). Thus, no error-checking signal is obtained (at time t)₃) Since no cells are switched. Alternatively, fig. 10c-d show a detector arrangement and timing/trigger diagram when the sort parameter is positive and the optical switch or fluidic switch is triggered. Here, as the cell travels past the laser within the detection/selection region, both the fluorescence signal and the forward scatter are again detected at or near the same time (at time t)₁) But the fluorescence signal is within the gating level and the optical or fluidic switch is triggered (at time t)₂). Obtaining an error-checking signal (at time t)₃) As the cells are switched. In this method, the time is triggered (at time t)₂) Is from the initial detection time (t)₁) A preset value (Δ t) is measured and the value of Δ t is determined by the speed of the cell and the position of the optical switch relative to the detection/selection area. The method satisfactorily achieves effective sorting; however, as a means for further improving the trigger accuracy, the second method is used.

Fig. 11a-d show this second approach, where two laser sources are used instead of one. In addition, as with the single laser method described above, the time signals from these detectors are used as information for the switching decision. Using a laser in the examination areaCell detection was achieved separately before the regions were fixed/selected. In this case, detection can be based on fluorescence, scattering (e.g., forward scattering), capacitance, imaging, or any detection means that can identify the desired cells. The second laser is combined with two or more detectors and used to effect cell detection and cell identification. In addition, in this case, the identification can be based on fluorescence, scattering (e.g., forward scattering), capacitance, imaging or any detection means that can identify the desired cells. The purpose for the two sequential cell detection steps is to allow a cell flow rate to be derived from the first detection (at time t)₁) And a second detection (at time t)₂) The time difference (Δ t) therebetween. Knowing the spacing (d) between the detector windows will yield the flow rate (v ═ d/Δ t), which value is then used in combination with the known distance (x) of the optical or fluidic switch from the evaluation window to calculate the time to trigger (t) of the optical switch₃X/v). In addition, switching occurs again when a particular gating level is obtained for the cell identification step. Although only a single fluorescence detector is shown for identification, multiple fluorescence detectors may be used. In this case, the cell flow is centered with an equal flow rate of sheath buffer, and the splitting plane is created on the right side of the cell flow using outlets with different widths. However, as noted above, any configuration may be used to manipulate the position of the cell flow and the plane of flow separation. In addition, both configurations present an error detector that checks whether the cells are switched or not. In this case, detection can be based on fluorescence, scattering (e.g., forward scattering), capacitance, imaging, or any detection means that can identify the desired cells.

11a-b show a detector arrangement and timing/trigger diagram when the sort parameter is negative and the optical switch or fluidic switch is not triggered. These cells enter the main fluid channel and are collected into a single row by sheath buffer from both sides. The presence of cells is signaled by forward scattering (at time t) as they pass through the detection window region₁) And (6) checking. When the cell passes the identification/selection window, a second forward scatter signal is obtained (at time t)₂) However, this signal is strong with a fluorescent signal that is not within the gated levelDegree (at time t)₂) Combined and the optical or fluid switch is not triggered (at time t)₃). No error-checking signal is obtained (at time t)₄) Since no cells are switched. Even without sorting the cells, use (t)₁)、(t₂) And detecting and identifying the known distance (d) between the windows to obtain the flow velocity (v) of the cell stream. This is obtained using the following relation: Δ t ═ t (t)₂)-(t₁) And v ═ d/Δ t.

11c-d show detector arrangements and timing/trigger diagrams when the sort parameter is positive and the optical or fluidic switch is triggered. Here, the presence of the cell is again signaled by forward scattering (at time t) as it passes through the detection window region₁) And (6) checking. When the cell passes the identification/selection window, a second forward scatter signal is obtained (at time t)₂) However, the signal is correlated with the intensity of the fluorescence signal within the gating level (at time t)₂) Combined and the optical or fluid switch is triggered (at time t)₃). Now an error signal is obtained (at time t)₄) As the cells are switched. In the method, a time (t) is triggered₃) Instead of a preset value, the trigger time is calculated using the cell flow rate (v) and the known distance (d) between the optical switch and the evaluation window. This is obtained using the following relation: Δ t ═ t (t)₂)-(t₁)；v＝d/Δt；(t₃) X/v. This method allows sorting to be performed more efficiently because it can account for fluctuations in cell flow rate and thus trigger an optical switch or fluidic switch more accurately. An additional benefit of this approach to optical switching is that for each individual cell, it is possible to adjust the rate at which the laser spot moves down the channel so that it matches the velocity of the cell as determined above, thereby maximizing the interaction time between the cell and the laser spot of the optical switch. The displacement speed of the laser spot will be changed by changing the drive of the AOM.

Another approach to improve sorting efficiency while introducing the triggering method described above is to concentrate cells in the main channel using a channel design that produces a true sample core such that it is completely surrounded by sheath buffer. Variability in the position of the cells along the height of the channel can cause variability in cell detection and fluorescence intensity. Ensuring that the cells are in the core stream in the center of the main channel improves sorting efficiency, as this minimizes any variability due to cell radial distribution and controls the distance the cells need to move to achieve efficient sorting. Such core flow may be achieved using a sheath buffer for 2-dimensional pinching of the input flow stream.

The method requires a bottom substrate and a top substrate; each having a microfluidic channel network formed therein. Figures 12a-b and 13 show a method of achieving this in which the channel design on one substrate is a mirror image of the design on the other substrate. Thus, when two substrates are assembled together, with the channel design facing each other, the network of channels covers and forms the complete fluid conduit. One type of design using this method is shown in fig. 12a-b, with the sample channel shown as a dashed line. A key feature of this method is to ensure that the sample channel is shallower than the sheath channel so that the sample conduit appears to enter the junction as a hole when the substrates are assembled together. This is shown in FIG. 13, where it can be seen that the cells enter the junction and then are pinched from four sides, creating a core of sample that flows into the center of the main channel. It is noted that the channels may be formed by wet chemical etching or laser etching of glass or quartz, or by molding or embossing in a plastic or polymer.

Another approach involves having a series of cross channels arranged such that at a first junction/intersection, the cell is pushed vertically towards one wall of the main channel, the next junction/intersection forces the cell flow vertically into the center of the main channel, and then at the third junction/intersection the final pinching flow from both sides creates a complete sheath buffer cover around the center of the sample flowing within the main channel. This is shown in fig. 14 and 15, and one possible path diagram is shown in fig. 16. In this embodiment, at junction (a), the sample flows from the top substrate into the junction and down into a channel located within the bottom substrate, with the lateral sheath buffer flowing laterally into the junction. The sample is somewhat concentrated and pushed to the upper wall of the bottom channel as it continues to flow towards the next junction (B). At junction (B), the sample flows from junction a to junction (B) along the top of the bottom channel. At this point the second sheath buffer flows from the top substrate into junction (B) and the sample is pushed down to the middle of the channel located in the bottom substrate. The sample continues to flow along the middle of the bottom channel towards the next junction (C). At this point the third sheath buffer flows into junction (C) from both sides and the sample is pinched into a single file. The sample is now surrounded as a core of the sample by the sheath buffer as it continues to flow, centered both horizontally and vertically within the primary input channel.

Another approach employs a shallow channel with its axis perpendicular to the 1-dimensional sheath buffer to minimize the effects of parabolic velocity dispersion. During 1-dimensional flow concentration within the microfluidic channel, particles or cells in the inlet-center stream are pinched in only one direction. In the direction perpendicular to the pinching and flow direction, the parabolic velocity mode still remains. The particles near the center of the channel flow fast, while the particles near the channel walls flow slowly. As a result, the particles or cells move down the flow channel with a velocity profile that complicates the synchronization of the detection event and the switching event. Fig. 34 shows a method to overcome the problems inherent during 1-dimensional microfluidic flow concentration, using channel depths that are vertically comparable to the particle size of the particles or cells. For example, if the diameter of the particle or cell is 10 microns, the channel depth may be 15 microns. In this case, the particles or cells occupy a large portion of the depth direction of the channel, so that they experience velocities that average the velocity distribution and narrow the velocity dispersion within the channel.

All of the microfluidic channel network designs described in fig. 1-16 have been created in glass substrates using conventional photolithographic masking and isotropic etching of masked glass substrates. Isotropic etching typically produces a microfluidic channel having a depth d in the center of the channel_eAnd on the top of the channelWith a width w ═ w_p+2×d_eWherein w is_pIs the width of the lithographic pattern defining the channel. The bottom profile of the via has a radius d at each edge due to isotropic etching_eAnd the top of the etched channel is open. A glass substrate, typically a glass cover slip, is thermally bonded to the substrate with the etched microfluidic channels to seal the top of the channels and complete the microfluidic channel network. Holes are typically turned in the top substrate prior to thermal bonding to provide through holes for fluid flow to and from the microfluidic channel network, but may alternatively be drilled in the etched bottom substrate rather than in the top substrate. Depth d of the channel_eThe microfluidic channels typically, but not limited to, are 10 to 100 microns deep, the width of the microfluidic channels is typically, but not limited to, 2 to 5 times the depth, this is achieved by using lines on the photolithographic mask, typically, but not limited to, 5 to 400 microns, as previously described, other substrates, such as plastic or formable or castable polymers, may be used in these cases, the microfluidic channels typically have rectangular cross-sections, yet otherwise similar to the channels within the glass substrate.

The microfluidic channel networks shown in fig. 1-16 generally depict only the local geometry of the inlet microfluidic channel, the sheath buffer-pinch junction channel, the cell identification and optical switch main channel, and the bifurcation of the main channel to the outlet. This description needs to be extended to provide a region within each channel to create a connection to a reservoir within a macro-scale fluidic device or cartridge that provides an interface to the through-holes described above, thereby providing ingress and egress of fluid flow from the network. The cross-section and length of each of these microfluidic channels needs to be adjusted to ensure that flow within the entire microfluidic channel network is controlled according to the technology selected to achieve flow within the channel. Both the cross-section and the length of these channels are determined by the pattern used to create the lithographic mask.

Figure 17 shows one embodiment of a mask for a complete microfluidic channel network with an inlet channel, two sheath channels to a T-pinch junction and two outlets branching off from a T-bifurcation junction. The mask was designed to provide a volume pinching ratio of 7: 1 (sheath flow rate seven times the cell inlet flow rate). The length of the channel is designed to provide a sufficient pressure drop to enable the use of a standard low flow syringe pump or low pressure pneumatic controller to establish flow. The design also reflects the need to equalize the pressure to be able to use only two pumps, one for the cell inlet channel and one for both sheath channels, with the outlet maintained at atmospheric pressure. The sheath channel inlet is located at the end of the design top, the cell inlet channel is initially lower than this, is central to both sheath channels, and is long enough to provide the appropriate pressure drop set to a 7: 1 pinching ratio, and the two outlets are located at the ends of the bottom left and right sides.

FIG. 18 shows another embodiment incorporating a triangular junction for the pinch junction and the Y-branch junction, designed to provide a 10: 1 volume pinch ratio. The design is otherwise geometrically similar to the design shown in fig. 17. Many other designs are obviously possible, but they share a common feature of the need to provide fluid ingress and egress, as well as appropriate pressure drop and pressure balance for the method selected to establish fluid flow. Similar design conditions were used to generate photolithographic masks for fabricating microfluidic channel networks of 2-dimensional constricting flow networks as previously described.

Figure 19 shows a preferred embodiment of a microfluidic channel network in a complete microfluidic sorting wafer. Two inlet ports for the cell sample stream and for the sheath buffer stream were identified, two outlet ports for fluorescence positive target cells and for fluorescence negative non-target cells, i.e., waste, were identified. The wafer is 24 mm x 40 mm. The thickness of the etched substrate was 1.1 mm. The thickness of the bonded cover plate was 550 μm. The microfluidic channels were 50 microns deep. The cell inlet microfluidic channel was 110 microns wide. The sheath flow and outlet microfluidic channels were 150 microns wide, as were the main microfluidic channels. The sheath flow pinching junction is an inverted equilateral triangle, 300 microns on each side, with the cell inlet channel connected by the base of the triangle at the top of the junction and the two sheath flow pinching channels from each side of the main channel at the bottom of the junction by the vertices of the triangle. The microfluidic channel network design was optimized to use pneumatic control of fluid flow at all four ports to establish network flow.

Microfluidic connections to the wafer can be made in a variety of ways. One way is to use flexible microfluidic tubing that is directly connected to the port, by gluing or using various tubing adapters that can be attached to the wafer surface at the port. The tubing may be directly connected to a syringe pump or similar system that provides volumes for handling the cell sample and sheath buffer, and provides pressure for flowing these volumes through the wafer. Using a syringe pump for handling sample volumes requires that the pump be washed and reloaded with each sample and creates the possibility of carrying one sample or contamination of one sample with the next.

An improved method for microfluidic attachment to a wafer employs a cartridge that is attached directly to the wafer using a uv curable adhesive, PSA (pressure sensitive adhesive) adhesive sheet, or other conventional bonding methods such as thermal bonding. The cartridge has four embedded reservoirs that provide an interface to each of the cell inlet channel, two sheath channels (from one reservoir) and two outlets, respectively. Such cartridges offer the possibility of aseptically manipulating cell samples and sorted target cells and waste products, as they can be completely confined within the volume of the cartridge before and after cell sorting. Fluid flow for such cartridge and wafer systems can be provided by using two pneumatic controllers that individually pressurize the cell inlet and sheath buffer reservoir to induce flow through the microfluidic channel network of the wafer to the outlet reservoir at atmospheric pressure.

Improved flow control methods are provided by using four pneumatic controllers that individually pressurize each of the cell inlet, sheath buffer, target cell collection reservoir, and waste collection reservoir. Such a flow control system provides the ability to individually adjust the volume pinching ratio at the sheath pinching junction, adjust the cell flow rate within the primary microfluidic channel for fluorescence analysis and optical switching, and adjust the split ratio at the switching bifurcation to enable bias flow, as previously described.

Figure 20 shows a preferred embodiment of a self-contained disposable cartridge that provides fluid reservoirs for a cell sample volume, a sheath buffer volume, and two outlet collection volumes for target cells and waste, respectively. The cassette is made of acrylic plastic or may be machined or cast. If appropriate, other plastics or suitable materials may be used in place of the acrylic plastic. The cell sample volume is generally conical in shape, tapering towards the port of the inlet microfluidic channel. In a preferred embodiment, the inlet reservoir comprises polypropylene patches to minimize cell adhesion and thus maximize cell yield. The wafers were bonded to the optical window area with a UV adhesive and the exit ports from the wafers interfaced with their respective reservoir volumes. The reservoir volume was sealed with snap-on caps having drilled holes for connection between the pneumatic controller and the individual reservoirs. The lid includes a silicone gasket to help seal against the cartridge body. A 0.1 micron polypropylene filter was also incorporated to create a gas permeable, liquid impermeable interface between the cartridge volume and the external environment. This maintains sterile conditions on the cartridge and minimizes any biohazardous contamination to the user or the instrument.

Preparing cartridges for cell sorting: the microfluidic channel network was first loaded through the sheath port with a sheath buffer solution using a conventional syringe with luer fitting. In this way, the channel was loaded and the sheath reservoirs were filled with 800 microliters and each outlet reservoir with 200 microliters. The cell sample reservoir is aspirated of excess buffer liquid, and then a 5-25 microliter sample of cells is placed into the sample input reservoir using a pipette. The cassette lid is then applied and pressed into place, providing a self-contained system in which cell sorting operations are performed.

The cassette is designed to rest on a support that arranges the main channels of the wafer so that the optical imaging system projecting the optical switch beam into the channels is properly aligned and focused into the channels. The cassette holder also includes a pressure manifold plate with 4-6 ports that are connected to four pneumatic controllers through external tubing. Each manifold port is sealed to its respective tray lid port using an O-ring, and these seals are made leak-tight by pressing the manifold against the tray lid using a cam-lock mechanism. Fig. 29 shows a preferred embodiment of a cassette holder for fluid switching that integrates two pneumatic valves on the manifold to shorten the volume and path length of the pneumatic pulses. The manifold enables the coupling of air pressure from a source to an air pressure port without the need for intervening tubing. This allows faster switching speeds for higher performance and supplements the performance benefits derived from the switching configurations shown in fig. 27-28.

A preferred embodiment of an optical system for optical switching is shown in fig. 21. The body with the pneumatic manifold connected to the snap-on lid is arranged so that the optical switching area is located at the focal point of both the lens system seen from above the cassette and the lens system seen from below. An output beam from the 488nm laser is projected through the imaging system into the main channel as shown in figures 3-7 and 9-11 just upstream of the sorting region to provide excitation for detection of fluorescence from fluorescence positive target cells. The fluorescence emission is collected by the same lens and imaged onto a photomultiplier tube through a dichroic mirror and appropriate fluorescence emission filters. The signal from the photomultiplier tube is processed by electronics to measure the level of fluorescence from the cells and determine the presence of fluorescence positive target cells in the flow stream within the primary channel. The fluorescence excitation wavelength is not limited to the 488 nanometer wavelength and can also be any wavelength of fluorophore suitable for identifying the target cell. If different excitation illuminations are used, the wavelength of the fluorescence emission filter must be changed accordingly. When fluorescence positive cells are identified, the electronics trigger the AOM to direct a beam from an IR-laser, typically from 1070nm laser operation at an output power of between 5W and 20W, into the main channel at an optically switched position. In a preferred embodiment, the AOM is controlled to produce an optical switching pattern as depicted in fig. 9b, although any of the optical switching methods previously described may be implemented. The 488 nanometer excitation light is imaged on the photodiode by the lens below the box body. The signal detected by the photodiode is used to help distinguish fluorescently labeled cells from smaller fragments that may carry the fluorescent label, and to identify clumps of cells that may form. These events were eliminated as candidates for sorting to the target outlet.

Yet another preferred embodiment would incorporate appropriate imaging and optical filtering to provide forward scatter signals upon illumination of the cells by the 488nm laser used to excite fluorescence. The optical arrangement will provide a range of angular sensitivity, such as but not limited to the range of 0.8 ° to 10 ° for forward scatter signal detection. This signal can help characterize cells in addition to fluorescent signals, and help distinguish cells from debris. The forward scatter illumination is not limited to fluorescence excitation laser light and may also be any other wavelength provided by an additional light source suitably imaged within the primary channel.

Yet another preferred embodiment would incorporate additional fluorescence detection channels that are sensitive to fluorescence emissions at different wavelengths, typically fluorescence emissions using a single excitation wavelength such as, but not limited to 488 nm. Each detection channel incorporates a PMT with an appropriate dichroic mirror and an emission filter for the fluorescence emission wavelength of the additional fluorophore. Two to four fluorescence detection channels can be easily accommodated in this way. The use of more than one fluorophore in this manner provides the ability to multiplex detection standards to identify target cells for sorting using an optical switch.

Yet another preferred embodiment would incorporate error proofing capabilities that provide optical illumination, typically as a narrow line across one of the channels within the network, and typically at a longer wavelength, but not limited to 785nm wavelength from solid state lasers, that is outside the wavelength range for fluorescence detection and forward scatter detection, but shorter than the optical switching wavelength typically at 1070 nm. The light source may be suitably imaged in a microfluidic channel network to provide a line that can be used to detect the travel of particles through any vertical plane within the network. Which provides additional capability to check the performance of the optical switch performance and provides additional capability for timing of the trigger for the optical switch as described in fig. 11.

Yet another preferred embodiment of the optical system would incorporate additional optical illumination at, but not limited to, 750nm, such as that generated by bandpass filtering of the light from the LED, and use that light to illuminate the microfluidic channel region. This region will be imaged through a 750nm band pass filter onto a CCD camera to provide a visual observation of the performance of cells flowing within the microfluidic channel network at the bifurcation junctions and/or at the pinching junctions. The filter before the camera will be sufficient to block any shorter wavelength radiation associated with the excitation or detection of fluorescence and associated with the forward scatter/side scatter optics and the error detection optics. The filter also blocks light from the optical switch having a longer wavelength of 1070 nm.

The preferred embodiment of the cartridge shown in fig. 20 is designed to hold the microfluidic channel network in a horizontal configuration so that all channels and inlet/outlet ports are in the same vertical water levelAnd (7) flattening. This minimizes the effect of gravity on the pressure drop through the microfluidic channels, resulting in more stable and controlled flow within the network. However, gravity will still have an effect on the cells within the fluid, particularly as the cells pass from the cell sample reservoir into the cell inlet microfluidic channel. To help control the effect of gravity on settling of cells in the reservoir and the effect of gravity on settling within the relatively slower fluid within the inlet microfluidic channel before the cell flow rate at the pinch junction is increased, a preferred embodiment of the sorter is to increase the buoyancy of the cells, thereby minimizing settling of the cells. Increasing buoyancy may be accomplished by using additives in the sample buffer. Examples of such rheology control additives, especially those having pseudoplasticity or shear thinning or both, are xanthan gum, carrageenan, sodium carboxymethylcellulose, methylcellulose, hydroxypropylmethylcellulose, hydroxyethylcellulose, hydroxypropylcellulose, hydroxypropylguar gum, gum arabic, gum tragacanth, alginates, polyacrylates, carbomers. Other additives include Histopaque^TMWhich is a mixture of polysucrose and sodium diatrizoate, and Optiprep^TMIt is a 60% w/v solution of iodixanol in water. The concentration of these additives used varies depending on the density of the cells to be sorted. For example, in Optiprep^TMIn the case of (3), the concentration may be 5% to 40%. Finally, the salinity of the sample buffer and the addition of sucrose can also be used to regulate the buoyancy of the cells.

The buffer for the cell sample volume and for the sheath flow may be any buffer that is biocompatible with the sorted cells and compatible with the optical illumination for the fluorescence detection mode and for the optical switch, that is, has sufficiently low absorbance at the fluorescence excitation/detection and optical switch wavelengths. A preferred embodiment of sheath buffer uses PBS/BSA, a Phosphate Buffered Saline (PBS) with 1% Bovine Serum Albumin (BSA) fraction 5, pH 7.2. A preferred embodiment of the cell buffer uses PBS/BSA with 14.5% Optiprep for live cell samples and 27% Optiprep for various formalin-fixed cell samples.

The performance of the cell sorting optical switching method in a microfluidic channel network was evaluated by the throughput, purity and recovery rate of sorting as described previously. The cartridge depicted in figure 20 was optimized to allow measurement of the performance as it was made of acrylic plastic, the bottoms of the target collection reservoir and waste collection reservoir were transparent, and the cells sorted into these reservoirs could be quantified both in terms of number and fluorescent label using an inverted fluorescence microscope. Several switch configurations as described in figures 3-11 were evaluated. These evaluations were performed using a 50: 50 mixture of live HeLa: HeLaGFP cells sorted using either a 1-sided or 2-sided fixed laser spot or a 0 ° or 8 ° 1-sided laser scan. The laser was scanned at 240 Hz. The laser on time is 4 milliseconds and the laser power is 20W for all switching modes. For the scanning spot approach, the focused IR laser spot is moved about 70 microns along the main channel.

As shown in fig. 22, the bi-directional optical switch using laser spots as shown in fig. 6 provides good purity and recovery results for a 50: 50 mixture of target cell-non-target cell at fluxes as high as 50 cells/second. However, at lower subpopulation concentrations (data not shown), laser power was not effective for switching non-target cells, and overlay error increased at higher cell throughput rates. In addition, small particles that are not switched will contaminate the target reservoir.

Fig. 22 also shows the performance of the 1-side switching method shown in fig. 9 with a fixed laser spot or a spot that is parallel to the flow in the direction of flow or shifted at a slight angle to the flow. The sample core flow stream is deflected to the waste outlet such that all cells are defaulted to flow to the waste outlet in the absence of the optical switch. Both of these methods provide improved performance, as shown. The fact that the performance of these two approaches cross implies that the triggering of the optical switch is not optimal and implies that the active triggering of the optical switch as shown in fig. 10 and 11 will improve performance.

Figure 23 shows an embodiment of a pneumatic fluidic switch for sorting cells or particles in a 1 x 2 (i.e. a network with one main input channel and two outlets extending from a bifurcated junction). The "Y" geometry of the bifurcation junction is shown in fig. 23, but other bifurcation geometries such as "T" geometry may also be used. Typically, these microfluidic channels are formed in an optically transparent substrate so that the cell detection optics can be projected into the channels. The substrate is typically, but not limited to, glass, quartz, plastics such as Polymethylmethacrylate (PMMA), and the like, and other castable or workable polymers (e.g., polydimethylsiloxane, PDMS, or SU 8). The depth of the microfluidic channel is typically, but not limited to, 10 microns to 100 microns. The width of the microfluidic channel is typically, but not limited to, 1 to 5 times the depth. In the case of microfluidic channels fabricated by photolithographic masking of a glass substrate followed by isotropic etching of the channel, the cross-section is typically rectangular, or rectangular with quarter-corners rounded.

The flow conditions are set such that the gas pressure P0 > P1 > P2 and the cells in the fluid flow preferentially from the highest pressure to the lowest pressure. When the pressure P1 increases such that P0 > P2 > P1, then the fluid flow is disturbed and equilibrium is reestablished as the fluid preferentially flows down the opposite branch of the "Y" junction. The system can restore the initial state by repairing the relation of P0 & gtP 1 & gtP 2.

The performance of the sorting mechanism in terms of throughput (time rate of cells entering the sorting region at the top of the bifurcation junction) is limited by the backward-propagation of the flow as shown in figure 23. When the flow switch is activated, the direction of particle motion is observed to reverse and flow upstream before a new flow path is established. The time delay induced by this motion limits the switching speed and the frequency of events that achieve acceptable yield efficiency (fraction of target cells within the target outlet) and purity (ratio of number of target cells/total number of cells within the target outlet).

Figure 24 shows one embodiment of a two-channel fluidic switch for sorting cells in a 1 x 2 microfluidic channel network (i.e., a network with one main input channel and two outlets extending from a bifurcation junction). The "Y" geometry of the bifurcation junction is shown in fig. 24, but other bifurcation such as "T" geometries may also be used. A double crossflow channel is disposed between the analysis zone and the bifurcation junction. The cross flow channels may be symmetrical in size, or one side larger than the other to support faster switching in one direction. Typically, these microfluidic channels are formed in an optically transparent substrate so that the cell detection optics can be projected into the channels. The substrate is typically, but not limited to, glass, quartz, plastics such as Polymethylmethacrylate (PMMA), and the like, and other castable or workable polymers (e.g., polydimethylsiloxane, PDMS, or SU 8). The depth of the microfluidic channel is typically, but not limited to, 10 microns to 100 microns. The width of the microfluidic channel is typically, but not limited to, 1 to 5 times the depth. In the case of making microfluidic channels by photolithographic masking of a glass substrate followed by isotropic etching of the channels, the cross-section is typically rectangular, or rectangular with quarter-corners rounded.

The flow conditions are set such that when the fluid switch (derived from air pressure in this case) is switched off, so that the pressure P3 in the region of the junction point is P4, all cells will preferentially flow into one of the outlets, for example the outlet on the right. When the fluid switch is opened under varying pressure such that P3 > P4, the fluid plug displaces the flow stream such that cells near the fluid plug are directed into the left outlet. Setting and controlling of flow conditions in a microfluidic channel network can be accomplished by direct drive pumps, pneumatic pumping, electrodynamics, capillary action, gravity, or other means by which fluid flow can be generated.

The performance of the sorting mechanism in terms of throughput (the time rate at which cells enter the sorting region at the top of the bifurcation junction), yield efficiency (the fraction of target cells within the target outlet 12) and purity (the ratio of the number of target cells/total number of cells within the target outlet 12) is affected by various factors, each of which affects the performance of the optical switch. Fluid switching can be characterized by several parameters such as: pressure differences projected into the sorting junction regions of the microfluidic channel network (P4 versus P3), the position of the switching channel relative to the diverging junctions, the duration of fluid switching activation, the maximum pressure for generating fluid displacement, and the like. The selection of specific values of these parameters of the fluid switch is determined in particular by: the topology and geometry of the microfluidic channel system, the flow rate (cell velocity) within the microchannel system, the ability to control the position of the flowing cells within the primary channel (whether the cells flow in the center or off to one side of the primary channel), the amount of cell displacement necessary to achieve reliable switching, the depth of the channel, the shape of the channel, and the force generated by the cell interaction with the optical switch.

Figure 25 shows an embodiment of a single channel fluidic switch used to sort cells in a 1 x 2 microfluidic channel network (i.e., a network with one main input channel and two outlets extending from a bifurcated junction). The "Y" geometry of the bifurcation junction is shown in fig. 25, but other bifurcation such as "T" geometries may also be used. Single cross-flow channels may be located on either side of the main channel between the analysis region and the bifurcation point. Typically, these microfluidic channels are formed in an optically transparent substrate so that the cell detection optics can be projected into the channels. The substrate is typically, but not limited to, glass, quartz, plastics such as Polymethylmethacrylate (PMMA), and the like, and other castable or workable polymers (e.g., polydimethylsiloxane, PDMS, or SU 8). The depth of the microfluidic channel is typically, but not limited to, 10 microns to 100 microns. The width of the microfluidic channel is typically, but not limited to, 1 to 5 times the depth. In the case of making microfluidic channels by photolithographic masking of a glass substrate followed by isotropic etching of the channels, the cross-section is typically rectangular, or rectangular with quarter-corners rounded.

The flow conditions are set such that when the fluid switch (in this case derived from the air pressure) is switched off so that the pressure in the region of the junction is neutral, all cells will preferentially flow into the outlet on the same side as the side channel, in this case the outlet on the right side, for example. When the fluid switch is opened at varying pressures such that P3 > P2, the fluid plug displaces the flow stream such that cells near the fluid plug are directed into the left outlet. Setting and controlling of flow conditions in a microfluidic channel network can be accomplished by direct drive pumps, pneumatic pumping, electrodynamics, capillary action, gravity, or other means by which fluid flow can be generated.

A preferred embodiment of a single channel fluid switch is shown in fig. 26. The bias flow is achieved by the right side channel being wider than the left side channel. The switching channel is also wider than the main channel along most of its path length and tapers at the point of fluid insertion. This reduces the peak amplitude of the applied air pressure necessary to achieve sorting by concentrating the force over a smaller cross-sectional area. This allows for faster operation and improved throughput.

The performance of the sorting mechanism in terms of throughput (the time rate at which cells enter the sorting region at the top of the bifurcation junction), yield efficiency (the fraction of target cells in the target outlet) and purity (the ratio of the number of target cells/total number of cells in the target outlet) is affected by various factors, each of which affects the execution of the optical switch. Fluid switches can be characterized by several parameters such as: maximum pressure projected into the sorting junction area of the microfluidic channel network, the position of the switching channel relative to the diverging junction, duration of fluid switch activation, etc. The selection of specific values of these parameters of the fluid switch is determined in particular by: the topology and geometry of the microfluidic channel system, the flow rate (cell velocity) within the microchannel system, the ability to control the position of the cell flowing within the primary channel (whether the cell flows in the center or off to one side of the primary channel), the amount of cell displacement necessary to achieve reliable switching, the depth of the channel, and the shape of the channel. For mammalian cells, a channel with a cross-section of about 50 microns by 150 microns and a pressure drop of 0.5 to 1.0psi over a 20-50 mm path length will achieve a flow rate of 1-5 microliters per minute and a switching pressure of 1-3psi is sufficient to achieve a switching rate of hundreds of events per second. To be generally applicable to particles of 100 nm to 100 μm, typical pressure drops for sample and sheath flow are 0.1 to 10psi over a path length of 10 to 100 mm. Sheath selection in current wafer design: the ratio of target pressures was 2: 1 to ensure proper pinching of the sample stream. The channel cross-section can be made 5 to 150 microns deep by 10 to 1,000 microns wide, depending on the type of cells or particles flowing across the wafer.

Fig. 27 shows a two-valve switch configuration that allows for control of the maximum pressure, rise time, fall time and total time of the pneumatic pressure pulse to actuate the fluid switch. The three ports may be connected to an air pressure source, ambient air, or vacuum. For fast switching, P1 is kept at a high pressure greater than ambient air, and P2 and P3 are kept at ambient air. In the neutral position, both valves are closed or off. When a switching event is required, the first valve is opened, which applies air pressure to the switching channel. After a short delay (e.g., less than 5-10 milliseconds) the second valve is opened, which quickly vents excess pressure from P1 out of port P3. The first valve is closed to vent the remaining pressure to ambient level in the switching channel and then the second valve is closed to return the system to its neutral position. This allows the pneumatic pulse to be shorter than the off-on-off cycle time of a single valve (e.g., about 5 to 10 milliseconds). Selecting a pressure P1 that is several times higher than the maximum pressure required (e.g., 8psi for a pulse with a required 2psi peak) and in combination with ambient air or vacuum at ports P2 and P3, the air pressure pulse may have a shape for maximum performance of the fluid switch.

Fig. 28 shows a dual valve configuration for a two-channel fluid switch as shown in fig. 24. In the first passage, a single valve regulates the high pressure source, and the second passage is in ambient air. In this configuration, the shortest possible pulse is determined by the off-on-off cycle period of the valve (e.g., 5-10 milliseconds). In the second configuration, a valve is located on each side passage of the fluid switch. In this case, both valves may be used to regulate the external pressure source. The second valve may be opened after the first valve after a variable time delay shorter than the off-on-off cycle time of the valves. In this manner, the pneumatic pulse can be shortened to increase the speed and performance of the fluid switch.

Cell sorting instruments based on a complete fluidic switch have been constructed. Examples of measurements obtained using the instrument are shown in fig. 30-31. In fig. 31, fluorescence calibration beads (e.g., sphereotech SPHERO beads or BDQuantibrite beads) were measured and showed a clear separation of the two intensities. The mean intensity, standard deviation and CV statistics were consistent with the manufacturer's recommendations. Fig. 31 shows an example of sorting a cell population. In this example, CellTracker was used^TMGreen (Invitrogen Corp.) stained target population of 1% Jurkat cells was incorporated into the unstained Jurkat cell population. Cells were sorted using a pressure of 0.3psi for the sample, 0,6psi for the sheath buffer, and 1.5psi for the switch valve. Approximately 35000 cells were analyzed and 336 target cells were identified and sorted into target wells. Examination of the target reservoir found 331 target cells out of a total of 369 cells. This resulted in a recovery calculated to be 99% and a purity of 90%.

To operate a microfluidic wafer, initial fluidic loading of all microfluidic channels is required in a manner that does not trap bubbles or introduce dust particles that block the flow. This requires the application of a reproducible volume of air at a steady pressure from a clean, high pressurized air source. Typically, a dust free system such as a clean bench or laminar flow clean room is necessary to add microfluidic devices. The apparatus shown in fig. 32, i.e. the filling station, meets these requirements in a simple and economical manner. A clamp on the filling station physically holds the cassette and provides a seal for high pressure access. A small submicron pore filter is attached to the clamp to supply high pressure air to the fluid-filled reservoir in a dust-free manner. A removable syringe may be used to provide a controlled volume of air or an automated valve connected to a high pressure source may be used. This evenly distributes a predetermined volume of bubble-free fluid throughout the microfluidic channel.

FIG. 33 shows a cell dilution cloning method for monitoring growth and isolating clonal populations. Micropores having a volume of a few nanoliters can be made using a variety of materials including, but not limited to, glass, quartz, plastics, such as Polymethylmethacrylate (PMMA), and the like, and other castable or workable polymers (such as polydimethylsiloxane, PDMS, or SU 8). The depth of the microfluidic channel is typically, but not limited to, 10 microns to 100 microns. The width of the microfluidic channel is typically, but not limited to, 1 to 5 times the depth. In the case of making microfluidic channels by photolithographic masking of a glass substrate followed by isotropic etching of the channels, the cross-section is typically rectangular, or rectangular with quarter-corners rounded. An example of a nanopore array is shown in fig. 33. In this case, the holes are 100 micron by 100 micron squares with a depth of 70 microns and are fabricated using PDMS flex lithography. The wells were sterilized in plasma and inserted into 24-well tissue culture plates. Cells suspended in media are loaded into each well and settled by gravity into the nanopores where they can be monitored periodically to observe growth kinetics or to harvest preferred clonal colonies at a single time point. The holes may also be incorporated into the output reservoir of the sorting apparatus as fixed elements or as movable inserts.

Fig. 35 shows an optical detection method for performing cell/particle counting within a multi-phase microfluidic droplet (e.g., cells/particles within a droplet encapsulated within a microfluidic channel). Applying appropriate signal processing, the number of cells/particles within the droplet can be detected based on the signal peaks from the cells/particles. In one approach, the entire channel is illuminated and a single detector measures a signal, such as a forward scatter signal, that indicates the presence or absence of a cell within a droplet. More sophisticated detection employs an array of forward scatter detectors arranged across the channel and perpendicular to the flow direction to provide better spatial resolution for counting and localizing cells/particles within the droplet. In yet another approach, additional dark field illumination and imaging may be performed on a single detector scatterometry. When the forward scatter signal detects the leading edge of the droplet, the CCD camera can be synchronized to obtain a dark field image of the droplet. The cells/particles will be bright spots in the image, which can be processed and identified for counting.

In many sorting cases, the target population of cells or particles may be smaller, but the surface area of the collection reservoir is larger. Particles may stick to the reservoir walls, which reduces collection efficiency. Figure 36 shows means to minimize cell or particle interaction with a surface. In one method, sorted target cells are collected in a containment area that displaces a smaller volume than the fluid, minimizing contact between the cells and the collection reservoir wall. Due to surface tension, the collected liquid forms wetted droplets on the enclosed area. In another approach, the surface of the collection reservoir can be made hydrophobic by coating with hydrocarbon or fluorocarbon based silanization. Because of the hydrophobic coating, the collected cell or particle containing liquid will be enclosed within the droplet near the exit orifice. In another method, a separately fitted but removable container may be used to collect the sorted cells. This separate container can be used for further processing of the cells, which will minimize cell loss while transferring the sorted cells from the collection reservoir to another reservoir (e.g., a multi-well culture plate). The collection container has an aperture at the bottom leading to an outlet aperture in the microfluidic device. The diameter of the hole may be less than one millimeter so that surface tension prevents leakage through the hole. A gasket made of a flexible material (e.g., silicone-based rubber, PDMS) is inserted between the collection container and the microfluidic device for sealing and bonding.

It is often desirable to recover the sample after cytometric analysis. Figure 38 shows a microfluidic device that pinches a flow of cells for analysis but recovers the cells at a lower dilution. The sample was subjected to focused flow using a sheath buffer as shown in fig. 2 and 38. After the sample passes through the analysis zone, the microfluidic channel is split into three channels. The central channel is the sample recovery channel, and the left and right channels collect excess sheath buffer to reduce sample dilution. By appropriate selection of the channel width and applied pressure, the degree of sample dilution can be adjusted to 1: 1 dilution to that of the sheath flow (typically 1: 20). Under certain conditions, i.e., conditions of extremely dilute samples, the collected sample may instead be concentrated.

Figure 37 shows a microfluidic device designed for multiple sample analysis or sorting. This design does not require changing the cartridge or rearranging after loading different samples that may cause a change in the fluorescent background. The present invention includes three sample reservoirs, two for the samples and one for the buffer, to keep the two samples away from cross-contamination. By adjusting the pressure, each sample can be injected and pinched for flow cytometry analysis. For example, with two samples a and B, for analysis of a, the highest pressure is applied to the reservoir of sample a, the buffer reservoir is at an intermediate pressure, and the reservoir of sample B is held at the lowest pressure. By alternating the channel with the highest pressure, the desired channel will either flow to the analysis zone (highest pressure) or its flow is prevented (lowest pressure). Pure buffer may be introduced between each sample to allow for clear separation of the signals in the analysis zone. The number of samples that can be loaded and manipulated in sequence can be increased to 2 for each individual partitioned buffer channel attached to the microfluidic wafer.

Fig. 38 also shows an embodiment of a microfluidic device that can perform multiplexing of two samples with adjustable dilutions. The device employs two sample channels a and B and a separation buffer. To analyze sample a, the following pressure settings were used: p _ sw _ A < P _ sw _ B < P _ sb < P _ A < P _ sth. To analyze sample B, the following pressure settings were used: p _ sw _ B < P _ sw _ A < P _ sb < P _ B < P _ sth. To prevent cross-contamination between samples, separation buffer may flow down the channel between sample a and B measurements. With additional pressure ports, the design can be scaled up for three or more samples. Fig. 38 shows an embodiment of the device that switches between two samples.

Although the foregoing invention has been described in some detail by way of illustration of embodiments for purposes of clarity of understanding, it will be apparent to those of ordinary skill in the art in light of the teachings of this invention that certain changes and modifications may be made thereto without departing from the spirit or scope of the invention.

Claims (6)

1. A microfluidic cell sorter, comprising:

a cell inlet adapted to receive one or more cells in a fluid medium;

first and second buffer inlets fluidly connected to the cell inlet to provide a buffer solution to the sorter;

a fluid channel fluidly connected to the cell inlet and the first and second buffer inlets,

a first cross-flow channel fluidly connected to the fluid channel, the first cross-flow channel being downstream of the first and second fluid buffer inlets;

a second cross flow channel fluidically connected to the fluidic channel, the second cross flow channel being located downstream of the first and second fluid buffer inlets, and the second cross flow channel being located on an opposite side of the fluidic channel from the first cross flow channel;

first and second outlets fluidly connected to the fluid channel, the outlets being downstream of the first and second cross-flow channels,

a detector adapted to detect cells in a given state and to generate a signal in response thereto, the detector being arranged to detect cells at a position upstream of the first lateral flow channel,

a pneumatic transverse force switch connected to the detector and the first and second cross flow channels and actuatable in response to a signal to generate a pneumatic pressure pulse to cause fluid movement within the cross flow channels, the pneumatic transverse force switch comprising:

a first pneumatic valve connected to the first cross flow passage;

a second pneumatic valve connected to the second cross flow passage;

a control system connected to the detector and the pneumatic lateral force switch for providing a timed control signal to actuate the first and second pneumatic valves in a sequential manner, wherein the first pneumatic valve is opened before the second pneumatic valve is opened;

whereby when a cell of a given state is detected, the pneumatic lateral force switch is activated to provide a lateral force to the cell, thereby moving the cell so that it selectively exits into the first or second outlet.

2. The microfluidic cell sorter of claim 1 wherein the fluid flow through the fluid channel is laminar flow.

3. The microfluidic cell sorter of claim 1 wherein the first and second buffer inlets are of different sizes.

4. The microfluidic cell sorter of claim 1 wherein the first and second outlets are of different sizes.

5. The microfluidic cell sorter of claim 4 wherein the first outlet has a greater volume of fluid flow than the second outlet.

6. The microfluidic cell sorter of claim 1 wherein the first pneumatic valve is closed after the second pneumatic valve is opened and before the second pneumatic valve is closed.