WO2001019516A1

WO2001019516A1 - Suspension handling system

Info

Publication number: WO2001019516A1
Application number: PCT/EP2000/008908
Authority: WO
Inventors: Rudolf Buser; Daniel BÄCHI
Original assignee: Hoffmann La Roche; Rudolf Buser; Baechi Daniel
Priority date: 1999-09-13
Filing date: 2000-09-12
Publication date: 2001-03-22

Abstract

A device for handling particles like cells and macromolecules contained in suspensions, said handling including in particular transferring, counting or sorting of said particles, and said device comprising means for actuating over selected intervals selected microvalves (13) of a plurality of microvalves available in an array of microvalves located in microchannels (12) of a network of microchannels for allowing passage of fluid of said suspension and of particles contained therein along selected paths through said network of microchannels. The time interval between successive selected time intervals is e.g. 100 milliseconds.

Description

Suspension handling system

The invention concerns a device of the type defined by the preamble of claim 1, that is a device for handling particles like cells and macromolecules contained in suspensions, said handling including in particular transferring, counting or sorting of said particles.

The invention also concerns a process for fabricating a device of the above mentioned kind.

The invention further concerns a system for handling particles like cells and macromolecules contained in suspensions, said handling including in particular transferring, counting or sorting of said particles.

In the environment of pharmaceutical and biological tests there is an increasing interest in fluidic manipulation of cells and macromolecules. Mixing and separation at a bifurcation by electrophoresis has been reported [1].

The main aim of the invention is to provide a device of the above mentioned kind which provides desirable performance characteristics with high reliability and which can be manufactured at a relatively low cost, said desirable performance characteristics including e.g. the capability of being suitable for handling a large variety of fluids and particles. The kinds of particles that can be handled by a device according to the invention should include biological particles, like e.g. cells and macromolecules.

According to the invention this aim is attained with a device having the features defined by claim 1.

Preferred embodiments of a device according to the invention are defined by claims 1 to 8.

A preferred embodiment of a device according to the invention comprises (a) a network of microchannels,

(b) an array of microvalves integrated in said microchannels, said network of microchannels and said array of microvalves being formed in a silicon chip which is encapsulated between two glass chips, and

(c) means for selectively actuating some of the microvalves of said array for allowing passage of fluid of said suspension and of particles contained therein along selected paths through said network of microchannels.

According to another aspect of the invention a process for fabricating a device according to the invention comprises the fabrication step defined by claim 9.

According to a further aspect of the invention the latter proposes a system which comprises a device according to the invention and which in addition comprises the features defined by claim 10 or claim 11.

The main advantages of a device and a system according to the invention are the high reliability of its operation for performing the desirable kinds of suspension handling and the relatively low cost at which it can be manufactured.

In a preferred embodiment of a device according to the invention different selected sets of microvalves are actuated at different points of time in order to provide corresponding different networks of microchannels and corresponding different selected paths through which a suspension and the particles contained therein are allowed to flow through the device. With this embodiment it is therefore possible to control fluid flow in the network of microchannels in order to achieve the desired kind of suspension handling, i.e. a handling enabling e.g. transferring, counting, sorting or mixing particles contained in a suspension.

A preferred embodiment of a device according to the invention comprises extremely thin silicone rubber membranes manufactured as part of and within the fabrication process for making the device. This has the advantage that thermopneumatic actuation of the microvalves is effected with relatively moderate temperatures. This makes the device suitable for handling in particular temperature sensitive biological particles like e.g. cells or macromolecules. A further advantage of the low temperature required for the thermopneumatic actuation of the microvalves is that it also allows to prevent crosstalk between individual microvalves of the array by means of water-cooling.

A particular advantage of a system according to the invention over e.g. electrophoretic systems is that it can be used with a wide variety of working fluids and particles.

The array of microvalves is preferably a matrix array.

The particle sizes that can be handled range from the submicron scale to a few microns.

In a preferred embodiment each microvalve comprises a free standing silicone rubber membrane with a thickness of two microns. When the microvalve is actuated, this membrane is used to reduce the cross-sections of channels which have a size of e.g. 30 x 7 microns in a preferred embodiment. The flow of e.g. micron sized particles in the channels can be controlled by actuating the valves. The density of valves is e.g. 600 valves / square inch. The flow of suspension through the network of channels can be monitored e.g. by means of sensors which are directly integrated into the channels .

Examples of preferred embodiments of a device according to the invention are described hereinafter with reference to the accompanying drawings wherein

Fig. 1 shows a representation of a first embodiment of a device according to the invention for sorting and mixing of particles, said device comprising a network of microchannels formed by an array of microvalves,

Fig. 2 shows a mask layout of heater cavities and cooling channels for reducing cross-talk between microvalves,

Fig. 3 shows a diagram showing 3 curves of deflection vs. pressure calculated according to three different theories for membrane deflection of a 20 x 130 micrometer silicon membrane with a 2 micrometer thickness,

Figures 4a to 4c show particle paths for 3 valve positions (a) open, (b) 40% closed and (c) 80% closed,

Fig. 4d shows an Y-branch as used for computational fluid dynamics (CFD) symulation,

Fig. 5a shows a schematic representation of a fabrication process of a device according to the invention,

Fig. 5b shows details of the heater circuit schematically shown in Fig. 5a, Fig. 6 shows a representation of a typical valve switching cycle for a silicone rubber membrane,

Fig. 7a shows a IR thermography of a system without water cooling and with two valves actuated,

Fig. 7b shows a representation of a finite element simulation of the system to which Fig. 7a refers to,

Fig. 7c shows a IR thermography of a system with water cooling and with two valves actuated,

Fig. 7d shows a representation of a finite element simulation of the system to which Fig. 7c refers to,

Fig. 8a shows a partial representation of a system which allows to assess sorting efficiency in a network of channels and integrated microvalves of a device according to the invention,

Fig 8b shows a photographic time-motion study of motion of a particle moving within an enlarged view of a zone 96 shown in Fig. 8a.

Fig. 9 shows a schematic representation of a system for controlling the operation of a device according to the invention.

In a preferred embodiment described hereinafter in particular with reference to Fig. 1 a network of microchannels 12 is formed with a matrix of mechanical gates 13 represented by dots in Fig. 1, e.g. microvalves, integrated therein and the system 11 formed by the network of microchannels 12 and the matrix of mechanical gates 13 is used for mixing and sorting of small particles with sizes ranging from the sub micron scale to a few microns. The system 11 just mentioned is arranged in a housing 14 which has an inlet 15 and several outlets which correspond to respective channels of the network of microchannels.

The above described system 11 is fed from a pressurized tank 16 containing a particle suspension. Air pressure is fed to tank 16 via an inlet 17 thereof. In operation suspension carrying particles is thus forced by air pressure to leave tank 16 through an outlet 18 which is connected to inlet 15 of housing 14 by a tube 19, e.g. a teflon tube, and to pass though the above described system 11.

Tank 16 contains different kinds of particles, e.g. particles 21 of a first kind/size, particles 22 of a second kind/size, particles 23 of a third kind/size, each kind having a specific particle size. By selectively actuating the microvalves 13 of the system 11 and by selectively controlling to which extent these valves are open or closed, system 11 defines different paths, e.g. paths 24, 26, 27 in Fig. 1, and each of these paths allows the passage of particles of a given size.

This definition of paths is achieved by dynamically switching the valves for a short time whenever a particle reaches an intersection of channels. A system for controlling the valves as represented in Fig. 9 allows to switch valves at points of time separated by e.g. 100 millisecond intervals. The different paths, e.g. paths 24, 26 and 27 represent particle paths achieved by dynamically switching the valves. With the help of image processing or integrated optical sensors it is possible to distinguish between particles of a different kind. The system represented in Fig. 9 allows to assign a specific path to every single particle contained in system 11. Another way to define specific paths through system 11 is to control to which extent the valve membranes are deflected into the channel. This allows to mechanically select a certain upper limit of particle size for a specific path. The dynamic definition of particle paths is more specific, because it does not only define upper limits for particle size for a specific path, but it also allows to define lower limits and it allows to select particles because of other properties, e.g. optical properties.

As can be appreciated from Fig. 1, system 11 can define for instance a first fluid path 24 which leads to a first outlet 25 of system 11 and which allows passage of particles of a first size. These particles are thus separated and thereby isolated from the other particles fed to inlet 15 of system 11. As shown by Fig. 1, system 11 can also simultaneously define other paths for particles carried by a suspension fed through inlet 15, e.g. a second path 23 and a third path 24, which allow passage of particles of a second and a third size respectively, and which are brought together into a fourth path 28 which leads to a second outlet 29 of system 11. System 11 thus allows e.g. to separate particles fed into the system via inlet 15 according to their size, or to mix particles of different kinds and sizes by bringing together fluid paths.

As can be appreciated from Fig. 1, the above described system 11 allows parallel processing of particles thus increasing throughput of the system (see Fig. 1) .

In order that the valves 13 can be integrated into the channels 12 of the network of microchannels represented in Fig. 1 the valves have to be very small. In the examples described hereinafter the valves 13 are thermopneumatically actuated microvalves. It should be noted however that any other suitable type of microvalve can also be used within the context of the instant invention.

A thermopneumatic valve consists of air confined in a recipient covered by a flexible membrane. By heating the bottom of the cavity the gas is expanded and the pressure in the chamber rises. This bends the membrane made e.g. of silicone rubber. The silicones major advantages are that it has a low Young modulus and a high maximum strain.

A first preferred embodiment of a device according to the invention comprises e.g. a matrix array of microvalves, such array having e.g. a valve density of 600 valves / square inch.

In order to reduce cross-talk between the valves, a device according to the invention preferably includes heater cavities and channels for the evacuation of cooling water as represented in Fig. 2. The mask layout shown by Fig. 2 comprises a channel network 31 for cooling water. Channel network 31 receives cooling water via a channel 32 connected to an inlet for cooling water and delivers water to a cooling water outlet via a channel 33. As shown by Fig. 2, the mask also comprises V-trenches 34 and heater cavities 35. In Fig. 2 the position of each valve shown therein is represented by a small square 36.

The convex corners, which are needed for cooling channels, cause an etching of silicon crystal planes in KOH with higher etch rates than the {111} planes. For this reason, if cooling channels are part of the device, an increased spacing of the valves is required, because before the wafer is etched through, these faster etching planes will attack the convex corners considerably. Therefore, due to the anisotropic etching properties of silicon, the valve spacing has to be increased considerably in order to integrate the cooling channels into the valve structure. If this is done, the result is a spatial valve density which is about six times lower than with a valve structure without cooling channels. However by using the technology of deep reactive ion etching (DRIE) instead of etching in KOH, the valve density will be even higher than for the valve structure without cooling channels and the valve density can be up to 2000 valves per square inch.

For a system without cooling channels, the anisotropic etching properties of silicon are not a problem, because only concave corners exist, and the valve spacing can be reduced.

In order that the operation of the thermopneumatic valve structure used in a device according to the invention be properly understood some theoretical aspects are described hereinafter.

THEORY OF THERMOPNEUMATIC VALVE OPERATION Nomenclature

T temperature / K To ambient temperature / K t thickness / m a length / m b width / m p pressure / N m -2 po ambient pressure / N m -2

E Young' s modulus / N m -2 n Poisson' s ratio y deflection / m

1 thermal conductivity / / WW mm ^_ -1¹ K I Pixel brightness, value range: 0...255 Membrane bending

A uniform pressure generated by heating of confined air deflects the membrane. Membrane bending for large deflections can be calculated by Equation (1) proposed by Roark [4] .

16 p (1 - v²) / (π⁶ E t⁴) = (1/12) (a^-2 + b^-2)² y/t +

(1/16) (4 v a^"2 b^"2 + (3 + v²) (a^-4 + b^"4)) (y/t)³ (1)

Membrane bending values computed by means of Equation (1) and expressed as a membrane deflection in micrometer versus pressure in kPascal are represented in the diagram of Fig. 3 as points on a full line 41. In the same diagram values obtained by an equation proposed by Timoshenko [5] are represented as points on a broken line 42, and values obtained by a simple finite element simulation are represented by dots 43. In all these three approaches it has been assumed that the Young modulus of the silicone rubber membrane is 200 kPascal according to values published by Bousse [1] . As can be appreciated from Fig. 3, all three theoretical results show good agreement with each other.

Thermopneumatic pressure

The volume of the heater cavities can be assumed to be constant since the increase in volume by the deflection of the membranes is negligible compared to the volume of a cavity at ambient pressure. The pressure generated by temperature change of the air confined in the cavities can be calculated with Equation (2) . Po T / o ( 2 :

A pressure of 3.5 kPascal can be generated if the air is heated to 10 °C above room temperature. Theoretically, this results in a membrane deflection of 6 micrometer for a valve membrane with a length of 130 micrometer, a width of 20 micrometer and a thickness of 2 micrometer. Therefore, it can be expected to get a sufficient reduction of the cross section of the channels necessary for mixing and separation of microparticles .

Thermal simulation

The thermal characteristics of the system are apt to be determined by using a finite element simulation. For this purpose a model is first built and meshed with I-DEAS, a computer aided design tool. The tetrahedron mesh generated with the automatic mesher of I-DEAS is then transferred to MARC, a finite element analysis program, where a thermal analysis is performed. For the embodiment described hereinafter a model consisting of 63505 elements and 13546 nodes has been built and meshed. The thermal conductivities of the materials used in the simulation carried out with this model are listed in Tab. 1.

Table 1 Thermal conductivity of materials (supplier data sheets)

The above mentioned model has 3 boundary conditions. The top of the channel chip see Figures 4a to 4c is cooled by free convection of the surrounding air. - 12 -

The simulation is based on the following values:

- A calculated heat transfer coefficient of 8.8 W rrf² K^"1 between the top of the channel chip and the surrounding air.

- The temperature of the surrounding air is 293 K.

- The bottom of the heater chip is glued to an epoxy print with a thickness of 1.5 mm which is part of the model. - The bottom of the print has a fixed temperature of 293 K. The cooling channels are filled with water with a temperature of 293 K.

- A calculated heat transfer coefficient between silicon and cooling water of 300 W m^"2 K^"1 .

Based on the simulation results it can be said that it is possible to operate two neighboring valves continuously without crosstalk to the other valves of an array if active cooling is used. A comparison of the results of the simulation and of measurements is described hereinafter with reference to Figures 7a and 7b, respectively 7c and 7d.

Fluid flow simulation

The influence of membrane bending on fluid flow through the channels is apt to be determined by a Computational Flow Dynamics (CFD) analysis performed for a Y Junction in a channel network. For the embodiment described hereinafter such an analysis has been performed with the program STAR CD. Because the channel height is small compared to the channel width, a 2D model consisting of 1180 hexahedral elements and 2718 nodes has been used. The inlet was simulated with a pressure boundary with a static pressure of 3 kPascal the pressure at the two outlets was set to 0 kPascal. The walls were simulated with a no slip boundary. Water was used as the working fluid.

Table 2 : Material properties of working fluid at 20 C

The program STAR CD allows to generate particle paths along the velocity vectors of the simulated flow field. These particles are not part of the simulation and do not interact with the flow field. They are only used for visualization.

Fig. 4d shows an Y-branch 44 as used for computational fluid dynamics (CFD) simulation. Y-branch 44 has a flow speed 45 at its inlet 51 and two branches 46 (or branch A) and 47 (or branch B) . At point 48 of branch 46 an activated valve is modeled by a reduced cross section B-B for branch 46 at that point. Branch 46 has at its outlet a flow speed 49. Branch 47 has at its outlet a flow speed 50. Close to inlet 51 the section is A-A as represented. For comparison sections C-C are shown at other points of branches 46 and 47.

Figures 4a to 4c shows particle paths obtained by such a simulation for 3 valve positions. Fig. 4a shows particle paths when a valve is open. Fig. 4b shows particle paths when a valve is 40% closed. Fig. 4c shows particle paths when a valve is 80% closed.

Figures 4a to 4c show flow speed vectors 53 at half the channel depth, a path 54 of a first particle, a path 55 of a second particle, and a path 56 of a third particle. In view of the simulation results represented in Figures 4a to 4c it can be expected that a 40 % valve closing should be sufficient to affect the path of a particle moving in the middle of a channel, whereas a 80 % valve closing should allow to affect also particles which are moving closer to the channel walls.

In the following, the basic components and steps of a process for manufacturing a device according to the invention are described.

FABRICATION OF A DEVICE ACCORDING TO THE INVENTION

As schematically shown by Fig. 5a, a device according to the invention comprises three chips. A single lithography step on each wafer produces the necessary structures. The advantage of this process is that changes on one layer can be made very easily and that the fabrication process has a very high yield. In the embodiment described hereinafter as example, S1813SP15 photoresist is used as the masking layer for all lithographs.

Char. wafer

A channel wafer 61 made of glass AF 45 is first coated with a CVD polysilicon layer having a thickness of 500 nanometer which is structured by wet etching in a mixture of 1 part HF 48 %, 2 parts CH3COOH 98.5 %, and 3 parts HN03 70 %. Channels, which are the microchannels 12 of the network of microchannels, are then etched to the desired depth in 12 % HF.

The channels in the network have e.g. a cross-section of 30x7 micrometer. The masking polysilicon layer is stripped in KOH. The holes which are necessary to connect inlet and outlet tubes 64 to/of the network of microchannels are formed by electro discharge machining. The network of channels has channel outlets 70, one of such outlets is shown in Fig. 5a.

The following steps of the processing of the channel wafer are represented on top and in the central part of Fig. 5a:

- process step 71: deposition of CVD polysilicon layer,

- process step 72: channel formation by lithography.

Heater wafer

A heater meander 69 is a composite layer composed of a Cr layer having a thickness of 300 nanometer and an Al layer having a thickness of 200 nanometer is evaporated on the heater wafer 63 which is made of glass AF 45. The heating meanders consist of 10 micrometer wide conductor paths having a length of 12 mm per heater element.

Heater cavities 66 and cooling channels 67 are formed in heater wafer 63.

The following steps of the processing of the heater wafer are represented on top and on the left side of Fig. 5a: - process step 73: deposition of Cr/Al layer,

- process step 74: circuitry formation by lithography.

Fig. 5b shows details of the heater means schematically shown in Fig. 5a and in particular, contact pads 60 for wire bonding, circuitry wide wire 59 and a heater meander narrow wire 69. Valve wafer

A wet oxide having a thickness of 1500 nanometer is grown on a valve wafer 62 which is subsequently used as a mask for KOH etching of the heater cavities. The valve wafer is a silicon layer having a thickness of 380 micrometer.

After reaching an etch depth of 360 micrometer in KOH the valve wafer is spin coated with silicone rubber.

In order to be able to deflect the membranes sufficiently with low heating power it is essential to fabricate extremely thin membranes 65. One parameter which can be easily adjusted to achieve this is the spin speed. Another possibility is to reduce the viscosity of the spin on compound. This is achieved e.g. by diluting the silicone rubber with toluene. DC 1-2577 has been used for some embodiments, because of its good solubility with toluene. However, for membrane thicknesses below 4 micrometer, the homogeneity of the layers was not satisfactory. With Sylgard 184 an alternative has been found, which makes possible to manufacture 2.9 micrometer thick membranes without any defects on a wafer surface having a diameter of 3 inches. Table 3 displays the influence of toluene concentration on membrane thickness for Sylgard 184.

DC 1-2577 and Sylgard 184 are silicone rubbers manufactured by Cow Corning.

The following steps of the processing of the valve wafer are represented on top and on the right side of Fig. 5a:

- process step 75: cavity formation by lithography and backside etch KOH, 360 micrometer,

- process step 76: spin coating with silicone rubber, - process step 77: SF6 etch silicon. Table 3: Membrane thickness for diluted Sylgard 184, spin time 30 s, spin speed 5000 rp

The remaining silicon membrane having a thickness of 20 micrometer is then etched with a plasma etch process at 270 K using sulfur hexafluoride (SF6) .

Assembly

To obtain the individual chips, the wafers are diced with a wafer saw. After dicing, the chips are ultrasonically cleaned. The heater, valve and channel chips are aligned in a single step with an especially developed aligner. Then an UV curing adhesive e.g. Vitralit 6127 is deposited on the edges of the stack. This adhesive is crosslinked with an UV dose of 300 mJ/cm² .

A microchannel network 68 with integrated microvalves is formed in this way between the channel wafer 61 and the membrane wafer 62.

In the following some measurements are described which allow to assess the performance of a microvalve of the type used in a device according to the invention. MEASUREMENTS

Membrane bending

In order to determine the mechanical properties of the silicone rubber membranes, the deflection under static pressure is measured. The deflection is measured with a non- contact type surface profiler. When a pressure of 30 kPascal is applied on a square shaped silicone rubber membrane having a side length of 130 micrometer a deflection of up to 24 micrometer is measured.

Any differences between theoretical predictions and actually measured membrane deflection values may be explained by intrinsic stresses in the silicone rubber layer which are built during evaporation of the toluene when the silicone rubber is already crosslinked to some extent. An oxide layer below the silicone rubber, which was not completely etched with SF6, may cause differences between membrane samples.

Valve switching cycle time

Thermopneumatic actuation of the silicone rubber membranes provides a very reliable deflection and operation thereof. Switching the valves up to 1000 times has not shown any deterioration of the valve's characteristics. The following valve switching times were measured (see Fig. 6) : rise time 150 ms; fall time 300 ms .

After a first sharp rise caused by heating of the air, the silicon walls heat up which causes a further increase in the temperature of the enclosed air. Because of thermal conductivity, this rise in temperature and thus pressure is much slower. IR thermography

For IR thermography a thermal imaging system was used. The camera of this system compares the radiation from a reference body heated to a known temperature with the radiation received from the test sample with a resolution of 500 x 600 micrometer.

Figures 7a and 7c show an IR thermography of systems with two valves each of which is actuated with a heating power of 200 mW. The temperatures indicated in these diagrams are temperatures in degrees Celsius.

Fig. 7a shows a thermography of a system with two valves 81 and 82 without water cooling.

Fig. 7b shows a representation of a finite element simulation of the system to which Fig. 7a refers to.

Fig. 7c shows a thermography of a system with two valves 83 and 84 with water cooling.

Fig. 7d shows a representation of a finite element simulation of the system to which Fig. 7c refers to.

Since silicon is transparent to the near IR the camera also records IR radiation from the glass below the silicon. The temperature measurements from very small thermocouples mounted directly on the chip showed good agreement with the IR images.

As can be seen when comparing the thermal simulation results in Fig. 7b (without cooling) and Fig. 7d (with cooling), the cooling limits the heating of the chip to the close surroundings of the valves in Fig 7d, whereas Fig. 7b shows a situation where neighboring valves are also affected by heating of the two valves. These findings were confirmed with IR thermography as can be seen in Fig. 7a and 7c. It can thus be seen that water-cooling reduces cross talk between valves significantly.

Particle tracing

Particles as small as 1 micrometer moving through the etched channels in the glass chips can be detected with standard video microscopes. The numerical aperture of the objective (Mitutoyo Plan APO 10) used is 0.28 giving a resolution of 1 micrometer with visible light. The working distance is 48 millimeter.

Fig. 8a shows a partial representation of a network of channels 91 and integrated microvalves 92, 93 of a device according to the invention. Also represented in Fig. 8a are an inlet tube 94, an outlet tube 95 and a zone 96 examined by a FOV video microscope.

Fig 8b shows a photographic time-motion study of motion of a single particle 97 moving within an enlarged view of a zone 96 shown in Fig. 8a and in particular of particle 97 as it moves through a T-bifurcation. The positions of single particle 97 shown by Fig. 9 were obtained by photographic recording at 50 ms intervals, the average flow speed and thereby the average particle speed being of about 50 micrometers per second when the valves 92 and 93 shown in Fig. 8a are open. During actuation of valve 93 shown in Fig. 8a, the flow speed and thus particle speed in the left channel branch is reduced to 2.5 micrometer per second in the zone indicated by arrow 98. In Fig. 8b arrow 99 indicates the flow direction. As can be appreciated from Fig. 8b, particle flow in the channels can be controlled by actuating a valve. For straight channels with a length of 2 millimeter and a cross section of 7 x 80 micrometer a flow speed of e.g. 530 micrometer/s is measured for a driving pressure of 800

Pascal. The particles moving through a network of channels of a device according to the invention has e.g. an average speed of 50 micrometer/s with no valve actuated and a driving pressure of 500 Pascal. This speed is reduced to e.g. 2.5 micrometer/s by actuation of a valve.

In the following a system for controlling the operation of a device according to the invention is described.

SYSTEM CONTROL

Fig. 9 shows a schematic representation of a system for controlling the operation of a device 101 according to the invention.

The valve array of device 101 is controlled with a standard personal computer 102, which is e.g. a Compaq Deskpro EP, with a Pentium II 450 Mhz processor, 128 MB RAM, and Windows NT4.0. The software for this purpose is programmed e.g. with Lab View 5.0 and is schematically represented by software block 103 in Fig. 9.

An output signal of a CCD camera 104 looking at the inlet channel of the system is used as the input of a frame grabber, e.g. a National Instruments NI 1408 PCI frame grabber 105. CCD camera is a ^ inch camera e.g. a Bischke CCD 4012, equipped with a microscope tube 106 (e.g. a NAVITAR Ultra Zoom) and a Mitutoyo PLAN APO lOx objective 107. A region of interest in the captured picture is defined via a software module 111 which provides control signals to frame grabber 105. After filtering the picture for the gray values of a particle by means of a software module 112, the pixel brightness I is averaged over the selected region according to Equation (3) by means of a software module 113

I average = ( ∑ lj,k ) / j * k (3)

Particles moving through the region of interest cause a change in the average brightness. This change is detected by comparing the current value to the value of the previous control cycle by means of a software module 114. The binary output of this comparison , which indicates whether particles are or not moving through the region of interest, is then used as a trigger for a sorting logic module 115 which via a valve select module 116 causes valve switching. This latter module generates suitable control signals that are transmitted via a standard digital I/O card 117 (e.g. a card NI-DIO 24) driving external relay cards 118 (e.g. cards ER-16, ER-8) .

The algorithm used for particle detection is very simple and thus allows a control cycle to be completed in less than 100 milliseconds. By using two or more wavelengths for illumination it is possible to distinguish between particles with different optical properties.

A software module 121 controls the heating power necessary for the thermopneumatic actuation of the valves. Software module 121 provides the necessary control signals via a GPIB (General Purpose Instrumentation Bus) 122 to a power supply 123, which provides the necessary energy to device 101 via relay cards 118. GPIB is a bus standard describing communication between computer controlled instruments and includes a hardware specification.

The program represented by software block 103 also controls the feed pressure on the suspension fed from suspension tank 16 to the inlet 15 of device 15 in Fig. 1.

In a preferred embodiment of a device according to the invention optical sensors are directly integrated into the channels. These sensors make possible tracking of particles through the entire system.

In a preferred embodiment of a device according to the invention optical sensors, e.g. photodiodes are directly integrated into the channels as part of the wafers and during the fabrication process. These photodiodes work as detectors in a light barrier. When a light source is placed above the channel wafer, a particle moving over a photodiode in the valve wafer will cast its shadow onto the light sensitive area of the photodiode and therefore reduce its photocurrent . The measured reduction of the photocurrent can be used to detect individual particles as they pass the sensors. The measured reduction of the photocurrent can also be used to distinguish between different kinds of particles. E.g. for a first kind of particles which is bigger than a second kind the reduction of the photocurrent is also bigger than for the second kind. It is also possible to distinguish between particle kinds with different optical properties, e.g. between particle kinds with different absorption coefficients. These sensors make possible tracking and identification of particles through the entire system.

In preferred embodiments of the invention the valve density of the system can be increased to more than 1000 valves per square inch, e.g. up to 2000 valves per square inch, by use of deep reactive ion etching in the process of fabrication of the system.

REFERENCES

[1] Bousse. L., Kopf Sill A., Parce J.W., An electrophoretic serial to parallel converter, Transducers 97, Chicago, June 16-19, 1997, pp. 499-502

[2] X. Yang, C. Grosjean, Y-Ch. Tai Ch.-M. Ho, A MEMS thermopneumatic silicone membrane valve, MEMS 97, Nagoya, 26-30 Jan, 1997, pp. 114-118.

[3] C. Vieider, 0. Ohman, H. Elderstig, A pneumatically actuated micro valve with a silicone rubber membrane for integration with fluid handling systems, Transducers 95, Stockholm, June 25-29, 1995, pp. 284-286.

[4] J. Roark, 1943, Formulas for Stress and Strain_, 2 nd edition McGraw Hill, pp. 213-218

[5] S. Timoshenko, 1940, Theory of Plates and Shells_, McGraw-Hill Book Company, Inc., New York, pp. 342-350

List of reference numbers

11 system

12 microchannels 13 mechanical gates (microvalves)

14 housing

15 inlet

16 tank

17 inlet for air pressure 18 outlet for suspension under pressure

19 tube 20

21 particle of a first kind/size

22 particle of a second kind/size 23 particle of a third kind/size

24 first path

25 first outlet

26 second path

27 third path 28 fourth path

29 second outlet 30

31 channel network for cooling water

32 channel for incoming cooling water 33 channel for outgoing cooling water

34 V-trench

35 heater cavity covered with valve membrane

36 valve position 37 38 39 40

41 line (membrane deflection vs. pressure, Roark)

42 line (membrane deflection vs. pressure, Timoshenko) 43 dots (membrane deflection vs. pressure, FE model) Y-branch as used for computational fluid dynamics

(CFD) simulation Flow speed inlet Branch A Branch B Activated valve modeled by reduced cross section for branch A 46 Flow speed branch A (46) Flow speed branch B (47) Inlet

Flow speed vectors at half the channel depth Path of particle one Path of particle two Path of particle three

circuitry, wide wire contact pad for wire bonding channel chip valve chip heater chip inlet tube membrane of a valve heater cavity cooling channel microchannel network heater meander, narrow wire channel outlet process step: deposition of CVD polysilicon layer and channel formation by lithography process step: electro discharge machining of adapter holes process step: deposition of metal (Cr/Al) layer 74 process step: circuitry and heater formation by lithography

75 process step: cavity formation by lithography and backside etch KOH, 360 micrometer, 76 process step: spin coating with silicone

77 process step: SF6 etch silicon 78 79 80 81 actuated valve without water cooling

82 actuated valve without water cooling

83 actuated valve with water cooling

84 actuated valve with water cooling 85 86

87

88

89

90 91 channel

92 valve

93 valve

94 inlet tube

95 outlet tube 96 zone examined by a video microscope

97 particle

98 arrow indicating the position of an actuated valve

99 arrow indicating the flow direction in a channel 100 101 device according to the invention

102 personal computer

103 software of control system

104 CCD camera

105 frame grabber 106 microscope tube 107 objective 108 109 110 111 software module: select region of interest 112 software module: filtering picture for gray values 113 software module: forming average over selected region 114 software module: comparing brightness values 115 software module: sorting logic 116 software module: valve select 117 digital I/O card 118 relay cards 119 120 121 software module: control of heating power 122 GPIB interface card 123 power supply 124 125

Although preferred embodiments of the invention have been described above using specific terms, such description is for illustrative purposes only, and it is to be understood that changes and variations may be made without departing from the spirit or scope of the following claims.

Claims

1. A device for handling particles like cells and macromolecules contained in suspensions, said handling including in particular transferring, counting or sorting of said particles, said device comprising means for actuating selected microvalves of a plurality of microvalves (13) available in an array of microvalves located in microchannels (12) of a network of microchannels for allowing passage of said suspension and of particles contained therein along selected paths (24, 26, 27, 28) through said network of microchannels.

2. A device according to claim 1 wherein all valves of the array are normally open, and wherein said means for selectively actuating a part of a plurality of microvalves (13) are adapted to be able to close selected valves over selected intervals of time, and to let the valves open between said selected intervals of time, thereby providing an array of valves having a configuration which dynamically varies with time.

3. A device according to claim 1 further comprising (a) a network of microchannels (12), (b) an array of microvalves (13) integrated in said microchannels, said network of microchannels and said array of microvalves being formed in a silicon chip which is encapsulated between two glass chips.

4. A device according to claim 1 wherein said network of microchannels and said microvalves are formed in a silicon layer which includes silicone rubber membranes which are each part of a microvalve.

5. A device according to claim 1 and further comprising means for thermopneumatic actuation of said microvalves.

6. A device according to claim 5 and further comprising means for individually cooling said microvalves and thereby preventing thermal crosstalk between the valves.

7. A device according to claim 1 and further comprising optical sensors which are directly integrated into the channels during the fabrication process of the device and enable tracking of particles.

8. A device according to claim 7, wherein said optical sensors are adapted for enabling particle identification.

9. A process for fabricating a device according to claim 1, which process comprises deep reactive ion etching.

10. A system for handling particles like cells and macromolecules contained in suspensions, said handling including in particular transferring, counting or sorting of said particles, said system comprising

(a) a device according to claim 1, and

(b) means for optically tracing particles carried by a suspension contained in said device.

11. A system for handling particles like cells and macromolecules contained in suspensions, said handling including in particular transferring, counting or sorting of said particles, said system comprising (a) a network of microchannels, (b) an array of microvalves integrated in that network of channels,

(c) valve actuation means for actuating selected microvalves of said array of microvalves in order to define various paths of flow of suspension carrying particles through the device,

(d) particle sensing means (e.g. optical particle sensing means) apt to provide control signals, (e) control means which in response to said control signals provided by said particle sensing means actuate selected valves of said array of valves in order to define said various paths of flow.

12. A system according to claim 11 wherein all microvalves valves of the array of microvalves are normally open, and wherein said valve actuation means include means which are adapted to close selected valves over selected intervals of time, and to let the valves open between said selected intervals of time, thereby providing an array of valves having a configuration which dynamically varies with time.

13. A system according to claim 11, wherein said control means are adapted to cause closing of close selected valves over selected intervals of time, said selected intervals of time being defined by said control signals provided by said particle sensing means.