Present day chemical engineering has a tendency to perform chemical reactions in a fluid (ie gas and/or liquid) phase in sub-millimetre sized microreactors.

Typically, a large number of such microreactors are arranged to operate in parallel and are formed in a common metal plate or, for low temperature applications, a box or plate manufactured from other materials such as plastics. An advantage of the small size of the microreactors is the capability of controlling very precisely the chemical reaction. Due to the small volume of reactants in relation to the relatively large heat or mass transfer surface, a fast control or system response is possible. The parameters that are controlled in such microreactors include temperature and/or pressure in the reactor and, particularly, the fluid flow rate of the fluid or fluids passing through the reactor.

In large reactors the fluid is controlled by mechanically operated valves. The valves are typically arranged to restrict the fluid flow rate at the reactor exit to maintain a pre-determined or required pressure level inside the reactor.

It is in principle also possible to use such mechanical valves for flow rate control in microreactors.

However, the use of such mechanical valves is problematical. This is particularly the case if the' chemical reaction in the reactor is exothermic so that

the generated products are at an elevated temperature.

This leads to extreme difficulty in the design and operation in the micro-size valves based on the use of moving components. If the micro-size valves are supported by elastic members, such as springs formed (eg by etching), from the common plate in which the reactor cavities are made, the moving components have a severely limited useful life. The limited useful life follows as a consequence of the elastic members being highly stressed to achieve relatively long motion.

Alternatively, the moving components of such a micro- sized valve may be constructed from freely moveable components that are unsupported. This leads to assembly difficulties and there is a danger of the valves becoming stuck.

The above problems are further compounded by the fact that many electro-mechanical phenomena, such as piezoelectric effect or magnetostrictive effect, which are used for generating the required mechanical motion in such a micro-sized valve, usually cease above a certain temperature.

Conventionally, the above problems may be, at least in part, addressed by substituting the mechanical valves by fluidic flow rate control valves. Fluidic devices do not have mechanically moveable components. The absence of such moveable components in fluidic devices, which are based upon aerodynamic or hydrodynamic phenomena in fixed geometry cavities, makes these devices relatively simple to manufacture. Furthermore, such fluidic devices have a significant service life. The risk of a moving component becoming seized or breaking is removed.

A typical fluidic device that is used for restricting fluid flow in a large-scale reactor is the vortex amplifier. Unfortunately, the use of such a vortex amplifier is limited to situations which result in or require relatively high values of Reynolds numbers. At low Reynolds numbers, the vortex amplifiers, as well as other known fluidic devices, do not work effectively or do not work at all. Typically, vortex amplifiers do not operate well at Reynolds numbers of below Re = 1000.

For good performance, a vortex amplifier should be operated with a Reynolds number of at least Re = 10, 000.

It is well known that in Wormley's experiments, the vortex amplifiers stopped working completely at Re = 750.

It is an object to at least mitigate some of the problems of the prior art fluidic control valves or fluidic devices.

Accordingly, a first aspect of the current invention provides a fluidic control device for controlling the flow rate. of a controlled fluid using a control fluid, the device comprising an outlet chamber having mutually facing inlet nozzles that are arranged to form mutually opposing streams of control fluid and controlled fluid to control the rate of flow of the controlled fluid.

In microreactors using fluidic devices according to the present invention, the operating Reynolds numbers tend to be very small due to the small overall dimensions, small fluid flows (dictated by the requirements of the reactant residence time in the reactor) and the often high fluid viscosities (especially

if the fluid is a gas at a high temperature).

A second aspect of the present invention provides a reactor or microreactor comprising at least one inlet arranged to feed a first reactant to or into an entrance end of a reaction chamber; the reaction chamber having an exit end comprising a fluidic control device as claimed in any proceeding claim for controlling the flow rate of fluid through the reaction chamber.

A third aspect of the present invention provides a method for controlling the rate of flow of a controlled fluid using a control fluid, the method comprising the steps of forming the controlled fluid and the control fluid into mutually facing streams of fluid injected into a fixed geometry chamber.

Embodiments of the present invention will now be described, by way of example only, with reference to the accompanying drawings in which: figure 1 shows a longitudinal section through a microreactor comprising a fluidic device according to an embodiment; figure 2 shows schematically a microreactor in which flow control is realised by electrically generated thermal effects; figure 3 illustrates a plurality of microreactors operating in parallel; figure 4 shows a further plurality of microreactors operating in parallel; figure 5 illustrates a microfluidic value according to an embodiment; figure 6 illustrates a flow control at the exit of a chemical microreactor by a turning-down collision valve

according to an embodiment of the present invention; figure 7 illustrates a microphotograph of the flow in a fluidic valve in an open state at a very low Re=11.2; figure 8 shows a microphotograph of visualised flow in a microfluidic valve at a higher velocity; figure 9 depicts the definition of the relative turning-down determined by the position of the stagnation point in a microfluidic valve according to an embodiment; figure 10 illustrates favourable turning-down characteristics of a microfluidic valve according to an embodiment; figure 11 shows bifurcations characterised by the variations of the inclination angle of the stagnation streams at progressively increasing Reynolds numbers; figure 12 depicts flow at a higher Reynolds number, Re = 75; figure 13 shows large inclinations of the stagnation streamline of oscillating collision jets at Reynolds number Re = 510; figure 14 shows the time dependence of colliding jets in oscillation; figure 15 illustrates the stagnation point where the location of the point of collision does not remain stationary; figures 16 depicts an almost zero magnitude of increasing angle; figure 17 illustrates an almost zero angle but with decreasing magnitude; and figure 18 illustrates two extreme positions of oscillating colliding jets at large Reynolds numbers near to Re = 100; figure 19 shows a microreator using a fluidic valve; figure 20 illustrates an experimentally investigated turning down effect;

figure 21 illustrates opposing jet flows; figure 22 shows a visualised interaction of opposing jets at a slightly higher control flow rate. The reactor exit is still not practically affected; figure 23 shows a photograph was taken later after starting the control flow-nearly reaching already the steady state situation; figure 24 shows a photograph of the visualised interaction of opposing jet flows with the stagnation point near the halfway between the two nozzles; figure 25 shows a photograph'of the visualised interaction shows that the stagnation point S has already moved beyond the mid-distance position; figure 26 shows a situation in which the reactor is seriously influenced; figure 27 shows a situation in which, at this velocity ratio, which is still not very high, the reactor exit becomes nearly closed; the marked asymmetry of the flow field in due to unstable character of the central balanced position; figure 28 depicts conditions in which the stagnation point was actually pushed upstream into the very reactor exit; figure 29 illustrates total closure of the reactor at a still higher control flow velocity; figure 30 shows a position in which the stagnation point was moved further'upstream ; figure 31 illustrates the relationship between the relative position of the stagnation point at the velocity ratio of the flows.

Referring to figure 1, there is shown a microreactor 20 in longitudinal section. The chemical reaction takes place in a cavity that is made by etching a thin basic plate 1 of stainless steel. The basic plate 1 is covered

by a similar flat cover plate ('not shown) to form a closed reactor cavity. The rate at which fluid flows through the microreactor (that is, the reactor flow rate) is controlled by a fluidic device. The flow through the reactor cavity is from left to right.

The microreactor 20 shown in figure 1 is an example of a catalytic reactor. Inside the reactor cavity there are, for example, packed beads 21 having a catalyst coated on their surface. Grids are formed at the entrance to and exit from the microreactor cavity to prevent the loss of the packed beads 21. At the entrance to the reactor cavity there is an inlet grid 22a. The inlet grid 22a comprises a row of projections. The projections can be made during an etching process by arranging for several islands of stainless steel to remain after etching. Between the islands of stainless steel are channels which are too narrow for the packed beads 21 to pass through. Similarly, an outlet grid 22b is situated at the exit end of the reactor cavity. The outlet grid 22b prevents any of the packed beads from escaping downstream from the reactor cavity.

Two inlets, that is, a first reactant inlet 6a and a second reactant inlet 6b, are provided to allow the supply of a reactant fluid or reactant fluids into the reactor cavity. The reactant fluids enter the cavity where they mix and, together with contact with the catalyst, undergo a required reaction.

At the exit end of the reactor cavity is a fluidic device comprising outlet chamber 9 that is connected to a product outlet 7 through which the products of the reaction are output.

The fluidic device comprises mutually facing reactor 2 and control 3 nozzles. The reactor nozzle 2 is formed by appropriate shaping of the exit of the microreactor 20. In a preferred embodiment, the appropriate shape of the reactor nozzle 2 is that of a planar sub-sonic nozzle comprising an'upstream contraction part of converging or progressively reducing cross-section and a constant cross-section exit channel. The constant cross-section exit channel opens into the outlet chamber 9. The reactor nozzle 2 is preferably coaxial with the control nozzle 3. The control nozzle 3 has the same or a substantially similar planar sub-sonic shape with an upstream contraction part of converging or progressively reducing cross-section and a constant cross section exit channel. The exit channel of the control nozzle 3 opens into the outlet chamber 9. The control nozzle is arranged to receive a control fluid from a control fluid source 4 and an actuator 43 is used to control the flow rate of the control fluid.

It will be appreciated that the nozzles do not need to be coaxial. It is sufficient is the nozzles are arranged such that a the flow of the control fluid interferes with or obstructs the flow of the controlled fluid.

The fluidic device is used, in the embodiment shown, to maintain pre-determined operating conditions within the reactor. Preferably, the operating conditions can be sensed by appropriately positioned and selected sensors 11. The signals from the sensors are fed, via a feedback line 10, to a controller 42 which controls the actuation of the actuator 43.

The operation of the fluidic device is based upon

jet collision. The control fluid leaves the control nozzle 3 as a jet that is directed to oppose the jet of the reaction products, that is, the controlled fluid, leaving the reactor nozzle 2. The jets from the reactor nozzle 2 and the control nozzle 3 have opposing velocities. The respective jets meet in the gap 5. The velocities of the respective jets from the reactor nozzles 2 and control nozzle 3 decrease towards the zero value at a stagnation point. The flow rates of the controlled and control fluids are arranged to result in relatively low respective Reynolds numbers. Preferably, the Reynolds numbers are less than or equal to 40 and preferably less than or equal to ten. In an embodiment the Reynolds number is arranged to be substantially 9.

The Reynolds number, Re, for a fluid is calculated by Re = nozzle width x nozzle exit velocity/fluid viscosity, where nozzle width is the width of the nozzle measured in the plane of the basic plate, the nozzle exit velocity is the velocity of the fluid leaving the nozzle and the fluid viscosity is the viscosity of the fluid leaving the nozzle.

At such low Reynolds numbers, the relatively large viscous damping of fluid motion ensures that the position of the stagnation point remains substantially fixed. The location of the stagnation point depends upon the relative magnitudes of the flow rates of the control and control fluids. Generally, if the velocity of the control fluid is increased relative to that of the controlled fluid, the stagnation point moves towards the reactor nozzle 2. If the velocity of the control fluid relative to that of the controlled fluid decreases, the stagnation point moves towards the control nozzle 3. The stagnation point may be positioned, under particular circumstances or under particular operating conditions, so that the

flow rate of reactants through the microreactor 20 is decreased and the residence time in the reactor cavity is increased. In effect, the cross-section through which the reactor products can leave the reactor is varied according to the flow rate of the control fluid. The above described mechanism of varying the available cross- section through which the reaction products in or of the controlled fluid can flow, has been found to be effective even at extremely low Reynolds numbers, that is at Reynolds numbers which are equal to or less than 10, where standard methods of fluidic control are not applicable or effective.

Referring to figure 2, there is shown a further microreactor having a fluidic device according to an embodiment of the present invention. It can be appreciated that the geometry of the basic part of the reactor 20 with its two inlets for reactants 6a and 6b is the same as for figure 1. Again, the microreactor of figure 2 also comprises a similar inlet grid 22a and a similar outlet grid 22b. A reactor heater, in the form of conductive foils, is arranged to maintain, upon connection to a current source 15, the temperature of the reactor cavity at a predetermined level. The predetermined level is selected according to the reaction to be performed. It can be appreciated that electric heating of a fluid is an effective method of controlling fluid flow rate in a low Reynolds number regime inside a microchannel. Such a flow is characterised-by very large hydraulic losses due to the strong influence of viscosity. Since viscosity depends upon temperature, that is, it increases with temperature in gasses and decreases with temperature in liquids, similar Ohmic heating by electric current in a conductive material exposed to the fluid flow leads to a change in hydraulic

losses and therefore a change in flow rate. However, it will be appreciated that the same heating element cannot be used to achieve concurrently the two different purposes. On the one hand, heating is required to maintain a particular temperature level inside the reactor cavity for proper kinetics of the chemical reaction. In contrast, the changes in heating to achieve variation of fluid flow rate due to thermal changes in viscosity cannot be, in the prior art, implemented simultaneously in the same microreactor channel.

Accordingly, an embodiment provides, as shown in figure 2, a second reactant tap in the second reactant inlet 6b. Although the tap 12 is taken from the second reactant inlet 6b, it can be appreciated that the tap could equally well be taken from the first reactant inlet 6a. The second reactant tap 12 leads to the actuator 43, which takes the form of a reactor heater that is substantially similar to reactor heater 13. The reactor heater 43 comprises conductive foils on the walls of a control chamber 39 connected to a current controller 42 via first 431 and second 432 terminals. Changes in viscosity of the control fluid are realised by heating the control fluid contained within the control chamber 39. This change in viscosity leads, in turn, to a change in the flow rate of the control fluid. As described above, a change in the flow rate of the control fluid leads to a change in the flow rate of the reactants through the microreactor, that is, to a change in the flow rate of the controlled fluid.

Referring to figure 3, there is shown a plurality of microreactors that operate in parallel. The flow rate of each microreactor 20 can be controlled independently by the controller 42. The control'fluid may be supplied

by, for example, a pressurised cylinder of control fluid serving as a control fluid source 4.

It can be seen that the microreactors 20, in this embodiment, have substantially the same configuration as the microreactor shown in figure 1. The microreactors are of a planar design, manufactured by etching and packed with catalyst beads. It can be appreciated that the three microreactors shown in figure 3 may form part of a larger number of parallel microreactors. The same structural features in the microreactors as shown in figure 3 will not be described in detail in relation to corresponding feature of figure 1. However, it will be appreciated that the same structures having the same reference numeral perform substantially the same function. It can be appreciated from figure 3 that a single reactant inlet 6 is provided rather than the first 6a and second 6b reactant inlets shown in figure 1. The single reactant inlet 6 can be used to carry pre-mixed reactants to the reactor cavity. At the exit of each reactor cavity there is a fluidic control device according to an embodiment of the present invention. The fluidic device comprises a reactor nozzle 2 formed by shaping the ends of the microreactors 20 into a constriction or contraction followed by a constant cross- section exit channels which open into the corresponding outlet chambers 9. Disposed opposite to and coaxial with the reactor nozzles 2 are control nozzles 3. The reactor nozzles 2 and control nozzles 3 are separated by a predetermined distance. Preferably, the predetermined distance is a multiple of the nozzle width. Still more preferably, the predetermined distance is substantially 2.7 times a corresponding nozzle width. The rate of flow of control fluid into each control nozzle 3 is adjusted by a respective one of the actuators 43 in response to a

respective control signal from the controller 42.

Although not shown in figure 3, it can be appreciated that the actuators 43 may be substantially similar to the actuator 43 shown and described in relation to figure 2.

The microreactor cavities and outlet chambers 9 are covered by a cover plate (not shown). The cover plate comprises openings, preferably oval openings, which are positioned to provide an exit for the fluids contained within outlet chambers 9. The oval openings are arranged to feed an outlet channel 19. The outlet channel directs the fluid contained therein into a separator 8 which separates the control fluid from the controlled fluid, that is, the reactor products are separated by separator 8 from the control fluid. It can be appreciated that the oval openings allow fluid flow from the outlet chamber 9 in a direction that is substantially normal to the plane containing the reactor cavity and/or the outlet chamber.

Referring to figure 4, there is shown the three microreactors 20. The microreactors 20 operate parallel and may form part of a larger number of microreactors.

The microreactors 20 in the embodiment shown in figure 4 have or can have substantially the same configuration as those shown in figures 1,2 and 3. Again, the cavities have a planar design and are realised by etching. The cavities of the microreactors are packed with catalyst beads.

It can be appreciated that the reactors are heated by the heat generated by the reaction. A separate means of electrically heating the microreactors 20 is not provided. However, it will be appreciated that separate means of heating the microreactors can be provided substantially as shown in figure 2. Similarly, means for

heating the control fluid, and thereby controlling velocity, can also be provided.

In the embodiment shown in figure 4, the flow rate in each of the microreactors is again controlled independently by actuators 43 in response to control signals output from controller 42.

It can be seen that the exit ends of the microreactors 20 are provided with respective fluidic devices according to an embodiment of the present invention. There is provided an inlet grid 22a and an outlet grid 22b in each reactor cavity for keeping the catalyst beads within the reactor cavities.

The flow rate in each of the microreactors 20 is again controlled independently in response to signals from the controller 42. However, in this embodiment, it can be appreciated that the outlet channel 19 comprises a product tap 14 via which the reactor product fluid can also be used as a control fluid. In effect, an amount of the reaction product is removed from the main flow of reaction product within the outlet channel 19 via product tap 14 and fed. back to the control nozzles 3 as a control fluid. It can be appreciated that this embodiment removes the need for a fluid separator 8. Preferably, there is also provided a pump 15 that is used for controlling or establishing the pressure of the control fluid. The control fluid available at product tap 14 is at a lower pressure than the pressure required at. the control nozzle 3. Therefore, the pump can be arranged to increase the pressure of the control fluid.

Furthermore, there is also, preferably, provided a heat exchanger 16 for controlling the temperature of the

control fluid. The heat exchanger 16 is used, in this embodiment, to lower the temperature of the control fluid by removing the heat generated during the exothermic reactions in the microreactors 20.

Preferably, the heat exchanger 16 is disposed between the product tap 14 and the pump 15.

In the embodiment shown in figure 4, it can be seen that the two reactants fed into the microreactors are not pre-mixed but are kept separate until a short time before they enter the microreactors 20 where they do not need intensive mixing. It can be seen that the separate inlets, the first reactant inlets 6a and the second reactant inlet 6b, are, in this embodiment, arranged in substantially the same manner as the separate product outlets 7, that is the oval openings, as described in relation to figure 3.

It can be appreciated that the oval openings at the entrance and exit ends of the microcavities allow fluid flow in a direction that is substantially normal to the plane containing the microreactor cavities. At the entrance end of the microreactors, the oval openings form the first reactant inlet 6a. At the exit end of the microreactors, the oval openings 7 allow reactant product 37 to escape from the outlet chamber 9 into the outlet channel.

It can be seen that the control nozzles 3 also open into the outlet chambers 9 such that the streams of the controlled and control fluids are mutually opposing. As described earlier, the flow rate of the control fluid is controlled by actuators 43 in response to control signals from controller 42. The actuators 43 in this and in the

other embodiments may be mechanically actuable valves such as, for example, solenoid valves. It can be appreciated that these mechanically actuable valves can be disposed sufficiently remotely from the control nozzles 13 so as not to be affected adversely by the exothermic reactions occurring in the microreactors 20.

The present invention can be used to stop completely the flow of controlled fluid.

The depth of the etching of the basic plate for the above embodiments can be set at any predetermined depth which can still allow a relatively low Reynolds number to be realised. In the above embodiments, the depth of all cavities, including the nozzles and outlet chambers, was 0.44 nozzle widths.

The outlet chambers 9 in the above embodiments preferably have an increasing width with transverse distance from the axis of the two inlet nozzles.

Preferably, the width of the outlet chamber is ten nozzle widths. The width of the oval openings also corresponds to ten nozzle widths.

Although the above embodiments have been described in relation to catalytic reactors, the present invention is not limited thereto. Other types of reactors could equally well be realised.

Although the above embodiment utilise control and reactor nozzles that have comparable dimensions, the present invention is not limited thereto. Embodiments can be realised in which the reactor and control nozzles have different dimensions. An embodiment has been realised in which the control nozzle was smaller than the

reactor nozzle. In particular, the a control nozzle having a width of, 70.6% of the width of the reactor nozzle still operates effectively.

It will be appreciated that a"subsonic nozzle"is characterised by a decreasing cross section in the direction of fluid flow. A planar nozzle is characterised by a rectangular cross section which includes a square cross-section.

Microreactors and fluid devices are described in "No-Moving Part Turn-Down Control of Flow in Microreactors", by Prof. Vaclav Tesar, Department of Chemical and Process Engineering University of Sheffield, a copy of which is included and incorporated herein.

A significant feature of technology in the 90ties is an interest in microscopically small devices, manufactured by techniques originally developed for microelectronics. This trend is recognisable also in chemical engineering, where it has led to an interest in chemical microreactors. It has been said for some time that they are"a solution in search of a problem to be solved", because there were only a few real applications (even though certainly not uninteresting ones: e. g. in chemical analysis of extremely small amounts of fluids, the"on-chip analysis"makes possible for any medical doctor availability of a DNA analysis laboratory nowadays possessed only by the largest of hospitals).. Recent development, however, has shown that microreactors are likely to play an important role in automobiles powered by fuel cells. This is because they may lead to a substantial (by up to two decimal orders of magnitude) decrease of the systems for preparation of gaseous fuels (required by fuel cells) from petrol (or another common

present liquid fuel), see Tonkovich A. L. Y., Zilka J.

L., MaMont M. J., Fitzgerald S. :"Microchannel Reactors for Automotive Fuel Processing", Proc. Of Conf. IMRET 3, Frankfurt, Apr. 1999, [1]. So far, such fuel reforming systems are unacceptably large and heavy. The application of microreactors seems to be the key step which should make possible in a horizon sometimes given as short as only five years, see"IC Engine's Days Are Numbered", Automobile engineer, February 1999 [2], to replace piston engines in cars by the new power units friendly to the environment. Intensification of the chemical processes is achieved because at dimensions of the order of micrometers there are substantially improved conditions for heat transfer and diffusive transport to the catalyst layer on walls. The way towards achieving the particular required large fuel flow rate is not scaling up but increasing the number of parallel channels ("numbering up"). An essential question for success is the individual control of conditions inside each microreactor. This is exactly what micromechanics and microelectronics can achieve: there is no problem in having sensors and perhaps even dedicated control circuits at each of the parallel channels, perhaps made on the same chip. One of the basic problems is, however, how to control the fluid flow rate through the reactor. Microvalves with moving components have been demonstrated, but much more promising seem to be application of principles of the modern fluidics-controlling fluid flow without moving parts.

The present author has been recently active in this field. The basic problem encountered is the fact that the flows to be controlled are at extremely low Reynolds numbers, as low as Re = 1.0. This makes impossible using the usual fluid dynamic mechanisms as used in Power

Fluidics, see Tesar V.,"Power Fluidics", Proceedings of the All-state conference of Technical Universities and Industry TRANSFER 98, Prague, June 1998, Part 2, page 137 [3] the content of which is incorporated by reference herein, where the flow closure into an"unwelcome" channel, instead of blockage by a mechanical component, is usually achieved by utilisation of inertia of fluid accelerated into the"welcome", direction. Since Reynolds number determines the ratio of inertial forces acting on fluid particles to viscous forces, it is apparent that if its value is small, viscous damping slows the accelerated fluid down and using the inertial effect is not successful. One of the promising solutions how to control fluid flow at low Re, tested by this author [4], is the rather brutal way of providing a collisional control jet flow directed opposite to the flow from the reactor, see figures 5 and 6. This requires relatively large control flows so that this principle would be certainly uninteresting in Power Fluidics [3] (where the aim is controlling by a control signal that is as small as possible). In microreactors, however, the absolute values of flows are so small that their magnitude is not important. It is possible to get (see figure 10) high flow gain (= ratio of the change of the controlled flow to the magnitude of the control action) and it is possible to obtain a complete closure (= i. e. stagnation point in position x > 1.0), of course only with a considerable control power input and a"soft"supply source having a large output dissipance.

Interesting hydrodynamic questions The head-on collision flows of submerged fluid jets at extremely low Reynolds numbers exhibits several interesting aspects, particularly well observable on the

shapes of the stagnation streamline (passing through the stagnation point S at the location of the mutual impact of the two jets, Fig. 9). Investigation was performed on a model in miniature size. It was a planar geometry, with constant depth of all cavities, which were laser cut in a foil of thickness h = 0. 15 mm, placed between transparent plane plates (to make possible observation of visualised flowfields). The width of the supply nozzle was b = 0. 34 mm, the width of the control nozzle was 0.24 mm. The separation distance between the two opposing nozzle exits was s = 0. 884 mm. Visualisation was achieved by short duration addition of light scattering particles into the control flow so that the video recordings capture the position of the propagating interface-and since the information about the time is stored, this makes possible evaluation of the local flow velocity.

Bistable inclination of the stagnation streamline One of the interesting facts is the observation of the flow field being symmetric (Fig. 7, Fig. 8) only at the lowest Reynolds numbers, Re < 40, where the flow has, in fact, a character of creeping flow. This, of course, is exactly the region of interest in the present application in microfluidics. The scaling laws of time and dimensions (equal Strouhal as well as Reynolds number) at the small dimensions of the order of micrometers cause that the transitional"creeping"flows in the microfluidic valves actually take place very fast, in time intervals of the order of milliseconds (the length of the intervals decreases with square of the linear scale). Theoretically interesting is the fact that above the values Re = 40 is observed (increasing with increasing Re) asymmetry-as shown in Fig. 12. The phenomenon is a bistable one:

inclinations with positive and negative angles a are found with practically the same probability. The fact is the more important that according to V. Tesar aj. : "Bistabilni proudeni se stojatym virovym prstencem", Fluid Mechanics and Thermodynamics, Proceedings of XVIIIth Intern. Conf. Of Department of Fluid Mechanics and Thermodynamics, str. 141, Praha 1999, ISBN 80-902714- 1-3 [5] the bifurcation transition to bistability as shown in Fig. 11 should be the starting point-via further bifurcation stages-of a scenario of transition to the chaotic turbulent behaviour.

Oscillations One of such bifurcation steps should be a, transition to regular periodic phenomena-monochromatic first, but gradually adding further harmonic components and during next bifurcations losing it regularity.

Indeed, the experiments at increased flow rates indicated that after crossing a boundary at about Re around 100- 200 the colliding jets are prone to oscillation. The oscillation is well observable as periodic changes of the inclination angle a of the stagnation streamline, Fig. 13, Fig. 14.At the relatively low Re = 410 prevailing in the experiment shown in Fig. 14, the time history of the angle variation could be well represented by a simple sinusoid. It was, nevertheless, not a perfectly monochromatic process. There are interesting position changes of the stagnation point S and its locations measured from the video recordings in Fig. 15 exhibit a marked regular deviations from the sinusoid (e. g. the points"a"lie regularly below the sinusoid), which testifies to the presence of higher harmonic components.

Remarkable is also the surprisingly low Strouhal number Sh-e. g. the value Sh = 0. 041 in the case shown in

Fig. 14 and Fig. 15 is just ten times smaller than the one corresponding to the basic columnar instability of a single jet, see Tesar V.:"Character of the Tesar-Ho Structure in an Excited Axisymmetric Jet Inferred from Anemometric Traverses", Sb. XV. Symp. O anemometrii a mezinarodni uncastic, Uvlay, kveten 1998, ISBN 80-86020- 23-1 [6]. This permits to conclude that there is a different feedback mechanism acting in the mutual interaction of the jets. Also the fact that increasing the velocity ratio u leads to a significant decrease of Strouhal number value indicates a different mechanism: e. g. the experiment with ratio u = 1. 81 exhibited Sh = 0. 0217. Further increase of Reynolds number, however, may lead to a complete disappearance of the stagnation points-Fig. 18.

Characteristic present-day development in modern chemical engineering is the tendency to perform chemical reactions in gas and liquid phase in submillimetre sized microreactors. One of the important advantages of the small size is the capability of precise control of the reaction-the small volume of reactants and relatively large surface area make possible a fast response to control actions. Among the controlled parameters, such as temperature or pressure, a very important one is the flow rate through the reactor. This is controlled in large reactors by mechanically operated valves, usually turning down the flow at the reactor exit (the reason for this location is the requirement to maintain a certain pressure level inside the reactor). In microreactors- especially in the case of exothermic reactors, where the generated products are at elevated temperatures-the layout based on moving components is not suitable. If held by some elastic members, the moving components have a severely limited useful life. If they are to move

freely, they may be difficult to assembly and there is a danger of their becoming stuck in their cavity. Many electromechanical phenomena-such as piezzoeffect or magnetostrictive effect-which are used for generating the mechanical motion usually ceases to operate above a certain temperature levels.

A modern possibility how to control fluid flow is substituting the mechanically operated valves by fluidic devices. These have no moving components, being based upon aerodynamic or hydrodynamic interaction with an auxiliary control flow in fixed geometry cavities. As a result they are relatively easy to manufacture and exhibit unlimited service life. Somewhat complicating feature is the unavoidable mixing of the reactor products with control fluid. In some cases, this fluid may be selected so that it is easily separated-ideally it should form an immiscible pair with the reaction products, the flow of which is controlled. Another solution is to use directly the reaction products as the control fluid-a pair of their flow is compressed and returned to the fluidic valve.

Typical fluidic device used for turning down fluid flows is the vortex amplifier. Unfortunately, its operation is limited to rather high values of Reynolds numbers. In microreactors, their small overall dimensions, small flows (dictated by the requirements of the reactant residence time in the reactor) and often high fluid viscosity (especially if the fluid is gas at high temperature) lead just to very low Reynolds numbers, at which the vortex amplifiers do not work properly.

The new solution, shown schematically in Fig. 19, utilises the idea of colliding jets. The control flow is

directed to oppose the flow of products from the microreactor. Both flows are fed into nozzles oriented so that their exits are facing one another. The position of the stagnation point S (Fig. 20) which may be adjusted by variations of the control flow rate, then leaves a limited extent of space through which the products may leave the reactor. This mechanism of varying the available cross section was found experimentally to be sufficiently effective even at extremely low Reynolds numbers, where standard methods of fluidic control are not applicable.

Feasibility studies of the novel fluidic valves were performed with neutral liquid (water) not undergoing chemical changes, in a geometry simulating the parallel reactor layout as shown in Fig. 3. Extensive use of flow visualisation on water model rig has provided useful insight. By contrast with historical developments in fluid-and aero-dynamics, the laboratory models are scaled up-they are larger than the final products. An interesting aspect is that time and distance scaling laws lead to much larger time intervals in the model, so that very short transition processes during valve closure and opening could be on the rig conveniently recorded by slowly running video camera.

The valves were of planar design and were made from transparent material, which permitted. visualisation of liquid motion in transmitted light. Geometry of the model is characterised by all cavities being of the same depth, 1.5 mm. The width of the reactor exit nozzle was 3.4 mm, so that the aspect ratio was quite low, 0.44.

The separation between the opposing nozzle exits was equal to 2.7 nozzle widths. In the configuration shown in the accompanying illustrations Fig. 21 to Fig. 30 the

control nozzle width was 0.706 of the reactor exit width.

In other experiments, equally good performance was found with reactor and control nozzles of the same width.

Fig. 21 shows photograph of the visualised interaction of opposing jet flows: At this lowest used control flow rate, the control fluid is just leaving the control nozzle. The reactor exit its practically not affected. The relative position to the stagnation point is = 0. 0.95. The ratio of the control jet to reactor jet maximum velocity velocities (on the nozzle axis u = 0.056) Fig. 22 shows a visualised interaction of opposing jets at a slightly higher control flow rate. The reactor exit is still not practically affected. The relative position of the stagnation point is = 0. 181. The ratio of the control jet to reactor jet maximum velocity velocities (on the nozzle axis): u = 0.075.

Fig. 23 shows a photograph was taken later after starting the control flow-nearly reaching already the steady state situation. Note how at this control flow rate, only slightly further increased, the relative position of the stagnation point increased significantly over the previous value, = 0. 285. The ratio of the control jet to reactor jet maximum velocity velocities (on the nozzle axis) is u = 0.080.

Fig. 24 shows a photograph of the visualised interaction of opposing jet flows with the stagnation point near the halfway between the two nozzles. In this position there is the highest sensitivity of the stagnation point S position on velocity ratio. The relative position of the stagnation point is 4 = 0. 384.

The ratio of the control jet to reactor jet maximum velocity velocities (on the nozzle axis): u = 0.081.

Fig. 25 shows a photograph of the visualised interaction shows that the stagnation point S has already moved beyond the mid-distance position. The relative position of the stagnation point is 4 = 0.529. The ratio of the control jet to reactor jet maximum velocity velocities (on the nozzle axis): u = 0.089.

Fig. 26 shows a situation in which the reactor is seriously influenced. The stagnation point S has already moved to 75% of the distance towards the reactor exit.

The relative position of the stagnation point is 4 = 0.748. The ratio of the control jet to reactor jet maximum velocity velocities (on the nozzle axis) : u = 0.138.

Fig. 27 shows a situation in which, at this velocity ratio, which is still not very high, the reactor exit becomes nearly closed; the marked asymmetry of the flow field is due to unstable character of the central balance position. The relative position of the stagnation point is 4 = 0. 891. The ratio of the control jet to reactor jet maximum velocity velocities (on the no. u = 0.221).

Fig. 28 depicts conditions in which the stagnation point was actually pushed upstream into the very reactor exit. Although substantially restricted, the flow rate through the reactor is not zero and some fluid gets out (note the asymmetry). The relative position of the stagnation point is 4 = 1.043. The ratio of the control jet to reactor jet maximum velocity velocities (on the nozzle axis): u = 0.3541.

Fig. 29 illustrates total closure of the reactor at a still higher control flow velocity. The relative position of the stagnation point is = 1. 102, The ratio of the control jet to reactor jet maximum velocity velocities (on the nozzle axis): u = 0.530.

Fig. 30 with the used geometry of the control nozzle, it was possible to move the stagnation point even further upstream-of course, the reactor flow was closed already at smaller control flow and this state is outside any normal operating requirements.

The relative position of the stagnation point 4 = 1.368. The ratio of the control jet to reactor jet maximum velocity velocities (on the nozzle axis): u = 0.751.

To make the extent occupied by the control fluid clearly recognisable, the water admitted into the control nozzle was coloured (using a mixture of malachite green and Victoria blue dyes). Transparent, non-coloured water was supplied into the simulated reactors. Full control of the reactor flow was demonstrated within the range of Reynolds number valves of the reactor'exit flow- evaluated from the nozzle width and maximum exit velocity -between Re = 9 and Re = 40. It was possible to turn down the flow completely (which, of course, is rarely required in microreactor operation). In the experiments shown in the accompanying illustrations Fig. 21 to 30, the reactor exit Reynolds number was Re = 75.

At each setting of the reactor and control flows, the control flow was first adjusted to zero, so that the field of view was filled by clear fluid (from the reactor

exit). The control flow was then started and a still photograph of the flow field was taken after, some time required for the conditions already having settled down.

Thus although most of the photographs (with the exception of Fig. 23) show the initial stage of the valve operation, the conditions basically are those of a steady state. Some bubbles of air released from the water during the experiment are visible in the photograph.

These are attached to the cavity walls and are believed to be of no consequence for the investigated process.

The dependence between the observed relative position of the stagnation point and the ratio of the nozzle exit velocities for this particular geometry is shown in Fig. 31. The closing effectiveness of the new valve may be estimated from the results of the performed flow visualisation shown here as a dependence of the observed position of the stagnation point (Fig. 20) on measured velocities. Evidently, this is a quite effective geometry: control jet velocity equal to only 30% of the reactor exit velocity suffices to move the stagnation point towards the"fully closed"position. Of particular interest is the steepness of the dependence: quite large motion of the stagnation point results from relatively small changes in the control flow. Of course, in possible applications, very much will depend upon the relative"stiffness"of both sources from which the fluid is to be supplied.

Operation of microfluidic valves for control of chemical microreactors was demonstrated on scaled-up models operated at very low Reynolds numbers, with flow visualisation providing very useful insight. The valves are of planar shape, suitable for manufacturing methods which are usual in microdevices. The capability to turn

down the flow without use of moving components at the extremely low values of Reynolds number is a unique property, not realisable by any other known means.

References : [1] Alépée Ch., Paratte L., Renaud P. :"High Temperature Chemical Microreactors", Institute of microsystems, DMT- IMS, EPFL, Lausanne, November 97 [2] Datta M. :"Applications of Electrochemical Microfabrica'tion : An Introduction", IBM Journal of Research and Development, p. 563, Vol. 42, Nr. 5. Sept.

1998 [3] Ehrfeld W., Hessel V., Moebius H., Richter T., Rusow K. :"Potentials and Realization of Microreactors", Dechema Monographs, Vol. 132, VCH Verlagsgesellschaft pp. 1,1996 [4] Joeckel K-P.. :"Microtechnology : Application Opportunities in the Chemical Industry", BASF AG Ludwigshafen, D, Dechema Monographs, Vol. 132-VCH Verlagsgesellschaft, p. 29,1996 [5] Tesar V. :"Valvole fluidiche senza parti mobili" (No Moving Part Fluidic Valves-in Italian), Invited survey in an anniversary issue Oleodinamica-pneumatica revista delle applicazioni fluidodinamiche e controllo del sistemi, Numero 3, p. 216,1998 [6] Tesar V. :"Fluidic Load-Switched Valve for Hot Gas Flow Control", Proc FLUCOME'94-the 4th Triennal International Symposium on Fluid Control, Fluid Measurement, and Visualization, p. 741, Toulouse, France, Sept. 1994