Microfluidics offers the potential of implementing many existing and novel liquid handling techniques within integrated substrates, commonly described as'chips'.

Such integration can offer considerable advantages in terms of performance, cost, environmental impact and indeed provision of new functionalities. By way of example, broad application areas such as chemical synthesis, chemical analysis and chemical species isolation, to identify but a few, have the prospect of being integrated into such'chips'. In order to facilitate this prospect, a number of generic liquid processing techniques, common to some or all of these applications, need to be implemented in microfluidic form. Amongst these techniques, pumping and valving elements have particular importance but have proved challenging to integrate.

Various approaches to designing micro-fluidic pumps and valves have been attempted. One method of producing microelectromechanical (MEMS) structures such as pumps and valves are silicon-based bulk micro-machining (which is a subtractive fabrication method whereby single crystal silicon is lithographically patterned and then etched to form three-dimensional structures), and surface micro- machining (which is an additive method where layers of semiconductor type materials such as polysilicon, silicon nitride, silicon dioxide, and various metals are sequentially added and patterned to make three-dimensional structures).

Recently, there has been considerable interest in employing Silicone Elastomer structures as an alternative to existing materials, and their use has been described in the Article by Duffy, DC McDonald, C. J. , Schueller, J. A. , Whitesides, G. M. Rapid Prototyping of Microfluidic Systems in Poly (dimethylsiloxane), Anal. Chem. Vol 70, pp 4974-4984, 1998.

The use of Silicone Elastomers in conjunction with conventional external pneumatic valves to actuate integrated valves and pumps has been described in the Article by Quake, S. R. , Scherer A. From Micro-to Nanofabrication with Soft Materials, Science, Vol 290, pp 1536-1539,2000.

Elastomeric-glass fluid control elements are disclosed in Patent Applications WO 97/00125, WO 02/43615 and US Patent 6408878.

In WO 02/43615, a system for fabricating and operating microfabricated structures such as on/off valves, switching valves, and pumps e. g. made out of various layers of elastomer bonded together is disclosed and used for controlling and channelling fluid movement. A preferred method of actuating the valves in such devices is by applying pressure on a membrane to close the flow channel and then removing the pressure to open the channel again. A range of means of applying the pressure is given and these cover pressurising a second channel adjacent to the flow channel by gas or liquid pressure; electrostatic actuation; by applying electrical potential to materials whose physical properties change in an electromagnetic field e. g. mercury; using a polymer which changes shape with the application of an electric current e. g. polypyrrole; magnetic actuation; electrokinetic systems; electrolytic systems and thermal expansion of fluid in an adjacent channel.

In these systems the means for closing the valves and shutting off the fluid are actuated positively, but the method for opening the valves relies on releasing the closing force and the channels resuming their open configuration by means of their natural elasticity. This can be disadvantageous when more rapid opening is required.

We have now devised an improved device and system for MEMS.

According to the invention there is provided a microfluidic structure which comprises sequentially (i) a substrate through which there are a plurality of conduits (ii) a first

layer of elastomeric material having a flat face and a patterned face having recesses formed therein, the flat face being bonded to the substrate and there being holes through the elastomeric layer connected to the recesses and aligned with the conduits in the substrate (iii) a second layer of a flexible material overlaying the patterned face of the elastomeric so that the holes and recesses in the elastomeric layer and the conduits in the substrate form channels through which fluid can flow (iv) an actuating means for driving a sealing means attached to or forming part of the second layer into at least one recess to modulate or seal the recess, which actuating means is operated by a piezo electric component.

Preferably the elastomeric material is a polydimethylsiloxane (PDMS) which is commercially available.

There can be recesses in the second layer overlaying the recesses in the substrate to form conduits. If a membrane is used this membrane can be attached to or form part of the second layer.

The piezo electric component can be connected to a second or top substrate to which the second layer is bonded.

The piezo electric component can include a piezoelectric element which is connected directly or indirectly to the second elastomeric layer.

Such an element includes a single sheet of material poled so as to contract or expand radially when a voltage is applied across the electrodes on its opposing faces or a pair of such devices bonded together and arranged so that with the applied voltage one contracts whilst the other expands or any other arrangement.

With the single piezoelectric element case the piezoelectric element and the flexible material acting together form a unimorph which deforms normal to the surface of the

second substrate in the manner of a drum membrane. With the pair of bonded elements, the piezo electric discs react against each other as a bimorph, again resulting in a drum-like deformation, which is transmitted to the second layer resulting in a similar deformation to the first case, albeit with greater magnitude if optimally implemented.

The invention also provides a fabrication sequence for microfluidic devices and systems consisting of a sandwich structure comprising a rigid base substrate, a first elastomeric intermediate layer and a flexible top layer.

By methods which include casting, punching and other related layering procedures it is possible to form microfluid components such as channels and chambers on either surface of the elastomer and within it. Such components may be interconnected. By employing piezoelectric elements it is possible to selectively deform a thin top substrate and thereby, with appropriate design, control fluid flow. By this means a microfluidic system incorporating control, pumping, and other components may be implemented.

By employing elastomers it is possible to exploit their elasticity in certain design aspects. Specifically: valve structures may employ flexible seals which permit flow channel closure without risk of damage to their constituent materials as would tend to be the case with a rigid-rigid seal such as silicon-glass. Furthermore, the seal is tolerant of suspended phase elements such as particulate contamination in its resistance to damage and in the sense of, by virtue of seal flexibility, its ability to still provide a seal under such circumstances. In such a case, a rigid-rigid seal would not fully close. Additionally, the flexible nature of the sealing surface reduces the probability of cell lysis within microfluidic cytometric or cytomanipulative systems.

The elasticity of the material may be exploited within the design process to give specific functionality. For instance, flexible columns may be employed to support and

locate large spans of the top substrate above a shallow chamber formed in the adjoining elastomer whilst not restricting the ability for the substrate to flex under the influence of an external force such as that resulting from piezoelectric action.

Alternatively, the elasticity may be employed, as described below to form throttles- flow elements whose cross sections, and hence flow resistance, are modulated by force emanating from the piezo electric element.

Mould Making The process or forming microstructures in the elastomer is usually by casting and this involves solidifying the elastomer in a mould. Many microengineering mould making techniques may be employed such as, but not exhaustively, classical silicon etching, deep plasma etching, silicon-silicon bonding, silicon-glass bonding, ultrasonic drilling, laser ablation, layer deposition and others discussed in the literature. In this work we have employed SU8 structural photo resist (Microchem Corp, MA, USA) on a silicon wafer substrate.

Elastomer Composites Elastomer mechanical properties may be modified by a number of established techniques. For example, they may be reinforced with fine silica powder. Elastomers may also be applied in thin layers onto substrates by means of spin-coating. Used together, these techniques may be employed to create an initial thin layer of elastomer with certain properties on the mould surface followed by further casting, of a second batch of elastomer with different properties thereby creating a composite.

For example, a thin layer of un-reinforced polymer may be spun onto a substrate forming a mould to confer desirable surface properties. The layer is then cured, the mould may then be assembled as discussed, and a second batch of elastomer reinforced with silica powder may be poured on top of the spun layer to provide a mechanically advantageous rigidity. The composite is then cured and employed in the normal manner discussed here.

Components Components which can be fabricated include valves, throttles, pumps, throttle-pumps, seal-pumps valve-pump hybrid components. These are described in more detail below.

Valves and Throttles We distinguish between valves which, in a perfect implementation, can entirely stop flow in one state and minimally impede it in the other and throttles which can modulate the flow rate without necessarily being able to stop it. More complex valves allow redirection of flow between an input and a plurality of outputs or a plurality of inputs and an output or a plurality of inputs to a plurality of outputs. Some valves may also be able to be used within their extremities of flow rate control range as throttles. It is advantageous that the depth of the throttle channel should be as large a part of the PDMS thickness as possible so that less force is"wasted"in deforming the bulk of the PDMS below the level of the throttle. The PDMS layer can be below 2mm e. g. 1. 6mm or as thin as 0. 5mm ; in both cases the channel is 100, umdeep.

Throttles are of particular value as blockage-tolerant, component elements in pumps.

Throttles-Channel Throttles The channel throttle exploits the elastic deformation of an element of elastomer in the direction normal to the applied tensile or compressive force. By providing a flow channel containing a'pinch point'at one or more locations and arranging for the elastomer forming the pinch point to preferentially deform into the flow channel when compressed normal to the flow plane, it is possible to reduce the width of the flow channel at the pinch point by controlled compression via the top substrate.

Alternatively, the pinch point width may be controlled by an applied extension, via the top substrate.

Advantageously, a channel throttle may be formed with one or a plurality of islands within the gap. By this means isovolumetric deformation of the island's elastomer towards both sides of the channel wall is achieved enhancing the constriction. Such a concept can be further developed by employing elongated islands whose length is greater than their width. By this means the majority of isovolumetrically deformed elastomer will be forced into the flow channel rather than deforming equally in a direction congruent to the flow, as such a large proportion of the deformed material will be employed to constrict the flow.

Throttles formed in elastomers may undergo secondary deformation by virtue of the operating fluid pressure differentials developed across them in operation. Such deformation may reduce their flow rate control range. Such effects may be mitigated by employing a number of throttle elements in series. By virtue of their being identical and still allowing flow they will each develop an equal fraction of the composite pressure and thus operate more optimally.

The flow resistance of single and series elements, as distinct from their control range of flow rate, may be reduced by forming arrays of such elements operating in a parallel manner fluidically. Such arrays may be advantageously implemented in a radial form thereby exploiting the pressure field of a discoidal force generating element such as a piezoelectric disc.

Channel throttles are less demanding in fabrication terms because they do not require selective preparation of the surface with hydroxyl groups via corona or other appropriate techniques ; the entire channel throttle surface-top substrate plane is bonded.

Planar Throttles The planar throttle exploits deformation of the upper substrate by controlled

compression, or if appropriately designed, extension, or a combination of both, via the top substrate, generated by force generating element such as a piezoelectric disc.

The deformation is arranged to constrict flow through a channel, or array of channels, by virtue of reducing the height of the channel. Beneficially, the elastic properties of the elastomer may be employed to provide supporting columns to define the resting position and deflection rate of the top substrate. The use of a piezoelectric actuator with positive opening makes it easier to control the position of the throttle valve.

Seal Valves Such a valve can employ an annular seal around a central feed shaft. Advantageously the seal is aligned with the centre of a discoidal force generating element such as a piezoelectric disc thereby exploiting the maximum and symmetric displacement of the top substrate. The seal may be arranged, by appropriate two level mould making, to be planar with the surface plane of the elastomer-top substrate interface resulting in a normally-closed element or it is below the plane, thereby resulting in a normally- open element. Activation of the force generating element provides for a change of state.

An important part of the fabrication process involves selectively avoiding formation of hydroxyl groups on the seal or selectively removing them subsequent to treatment in order to prevent adhesion of the elastomer to the substrate. A straightforward technique is to physically mask these elements during corona or related treatment.

Advantageously the fluid volume of the valve may be reduced by a number of means including: forming the outer area of the flow chamber from a shallow mould depth; delineating the fluid flow area with a sealed wall outside which a sponge-like array of columns locates the top substrate whilst allowing it to travel freely. Those skilled in the art will appreciate that a variety of elastic-mechanical structures could fulfill a similar role to columns.

Correct choice of seal compliance in the axis of compression is important to ensure good sealing whilst controlling secondary adhesion between the seal surface and the top substrate. This may be controlled by elastomer formulation or composite formation, as discussed previously, or by surface treatments to the top substrate and/or the elastomer, including silanisation or by optimal mechanical design. In the latter case two level mould making can be exploited to form a seal whereby the lower 'pedestal'confers resistance to deformation whilst the upper, narrower seal, mounted on the'pedestal'minimises contact area and hence parasitic attachment forces. By virtue of the reduced height of the upper seal element the top substrate deformation will be proportionately large, resulting in increased proportional elastic deformation and hence restorative force to counteract the parasitic attachment. Such an increase in the restorative force in respect to the contact-related parasitic forces will enhance seal separation behaviour. Another means to reduce parasitic adhesion between the valve seal and glass is to deliberately microstructure the seal surface. This can be via photolithography or by deliberately employing a reversed ('upside down') silicon wafer and using the finely ground, but unpolished side as the basis of the mould. By this means the seal can be cast with a microscopically rough, textured surface Planar Valves Planar valves are advantageous in circumventing the requirement for a punched or otherwise formed channel that characterises seal valves. In general they consist of a pair of separate fluid networks with a high interface perimeter length across a narrow elastomer wall. Advantageously the interface area is aligned with the centre of a discoidal force generating element such as a piezoelectric disc thereby exploiting the maximum and symmetric displacement of the top substrate. The two networks are sealed by the top substrate, but an essential part of the fabrication process involves selectively avoiding formation of hydroxyl groups on the network interface area or selectively removing them subsequent to treatment. A straightforward technique is to physically mask these elements during corona or related treatment. Activation of the force generating element serves to lift the top substrate away from the elastomer

surface thereby allowing fluid to flow between the two networks.

Advantageously, the non-hydroxilised network area may lay in the centre of the area subjected to force. Around this a hydroxyl bonded perimeter may be formed, itself surrounded by an area in which a sponge-like array of columns locates the top substrate whilst allowing it to travel freely. Those skilled in the art will appreciate that a variety of elastic-mechanical structures could fulfil a similar role to columns.

By this means the elastomer will exert a lower restorative force on the top substrate and hence allow a greater lift for a given actuating force. Variations on this concept will be obvious to those skilled in the art.

A throttle pump consists of a pump chamber with a throttle valve controlling the inlet flow and another controlling the outlet flow. By appropriate synchronisation of the respective throttling signals and the pump signal it is possible to exploit the differential flow resistance between the two throttles to provide a net transfer of fluid in a preferred direction.

A pump would consist of throttle valves, as described previously, and a pump chamber. Such a chamber would consist of a simple recessed chamber aligned with the centre of a discoidal force generating element such as a piezoelectric disc acting via the top substrate thereby exploiting the maximum and symmetric displacement of the top substrate. Fluid coupling may be via one connection leading to a bifurcation or, preferably two distinct connections that may be planar or in the form of punched through-channels.

Advantageously, especially in the case of liquid, the pump chamber will be formed by a mould level deeper than the throttle elements. This exploits the incompressibility of the liquid and hence a chamber volume-independent inflow and outflow in conjunction with minimising any pumping losses associated with the top plate being forced close to the chamber floor, hence creating a restricted flow path, at the lower

extreme of the pump stroke.

Advantageously, the pump chamber may be of greater diameter than the throttle valves thereby optimising chip area and minimising the cost of the force generating element e. g. the piezoelectric material.

Those experienced in the field will recognise that optimum performance may require a subtle timing relationship between the valves and pump chamber; such timing may involve an temporal overlap of operating waveforms.

Seal-Pumps A seal pump consists of a pump chamber with one seal valve controlling the inlet flow and another controlling the outlet flow. By appropriate synchronisation of the respective seal valve signals and the pump signal we obtain a transfer of fluid in a preferred direction.

A pump would consist of seal valves, as described previously, and a pump chamber.

In a preferred implementation, such a chamber would consist of a simple recessed chamber aligned with the centre of a discoidal force generating element such as a piezoelectric element acting via the top substrate thereby exploiting the maximum and symmetric displacement of the top substrate. Fluid coupling may be via one connection leading to a bifurcation or, preferably two distinct connections that may be planar or in the form of channels.

Advantageously, especially in the case of liquid, the pump chamber will be formed by a mould level deeper than the throttle elements. This exploits the incompressibility of the liquid and hence a chamber volume-independent inflow and outflow in conjunction with minimising any pumping loses associated with the top plate being forced close to the chamber floor, any hence creating a restricted flow path, at the lower extreme of the pump stroke. In many circumstances the pump may

advantageously function as a positive displacement device with a flow rate that can be accurately predicted.

It is also possible to configure a pump with a plurality of inlet or outlet valves, thereby combining a pumping role with the role of delivering fluid through selected ports.

Valve-pump Hybrid Components By using a force generating element with the ability to deflect the top substrate upwards or downwards, depending upon the polarity of the control signal, it is possible to form combined pump-valve components. A particularly suitable force generating element would be a 2 layer piezoelectric stack such as those manufactured by Piezo Systems Inc. Cambridge, Massachusetts, USA. whose constituent elements are arranged back-to-back in terms of polling, that is to say: positive to negative, An example of such a component has a similar form to a normally-closed seal valve, such as that shown in fig 5 below, but is operated with a bipolar electrical drive to the piezoelectric element such that the top substrate may be deflected upwards from its unactivated position, in which case fluid is drawn in through both ports. As the substrate is subsequently brought downwards, until the point is reached where the seal is closed, the fluid is forced out via both ports again. However, once the seal is made and the polarity reverses thus continuing the downstroke, the majority of the fluid is forced out of the port connecting to the larger outer chamber. Upon the commencement of the upstroke fluid is drawn back via the ports into the respective chambers. Whilst no net pumping results from this cycle, it is the case that when combined with a second similar component it is possible, with appropriate actuator sequencing, to cause a net transfer of fluid through the combination.

An alternative implementation of this component may be based upon a modified planar valve. In a preferred implementation, the planar valve structure is modified so

that a volume asymmetry within the two portions of the chamber is introduced, so that, as with the seal valve based component previously described, upon further compression, more fluid is displaced through one port than the other.

The argument for a volume asymmetry is also valid in the modified seal valve component, but is generally the case anyhow with this component.

Such a combination of components may be used in a number of pump configurations intended to minimise device size and number of actuators and, depending upon configuration, accept a reduction of one or other metric of efficiency such as pressure, flow rate etc.

Those skilled in the art will recognise that other timing sequences and configurations of chambers can be employed to implement pumps based on these principles.

Example-Test device moulds.

An embodiment of the invention was fabricated in the following general manner to create a two layer mould. The device mould design was defined by two photomasks via a CAD package and then implemented in 4040dpi laser-plotted acetate photomasks. The first mask layer defined all topological features both the'shallow' layer of 40um thickness and the'deep'layer of 100, um thickness. The second layer represents a topological sub-set of the first. The overall result of the two masks is to offer three effective design depths: (0,40, 140) um. The first pattern was transferred into SU8 epoxy structural photoresist that had been spun 40 um thick onto a 3 inch silicon wafer substrate. The SU8 was then post exposure baked (but not developed) and coated with a further layer of 100 um SU8. After pre exposure bake, the second photomask was aligned to the (visually apparent) first pattern and the second layer was exposed. The SU8 was then developed according to well established procedures.

Other materials could be employed for the mould construction.

Silicone Elastomer Casting and Device Assembly The resulting structure was then employed as a negative mould for casting a Polydimethylsiloxane (PDMS) (Dow Coming Corp. ) structure. Well-established procedures known to those skilled in microfabrication were employed. In addition to these established procedures, we employed a'picture frame'of 1. 6mm thick PMMA around the perimeter of the substrate and sealed the top, after introduction of the uncured PDMS by means of another PMMA sheet which is clamped to the picture frame. After curing according to the manufacturers recommendations, the PDMS was removed from the mould by peeling off the top sheet of PMMA and removal of the PDMS from the'picture frame'. The individual PDMS chips were cut from the casting and their lower surfaces corona treated to create hydroxyl groups on the surface and enhance adhesion. They were then aligned and bonded to the lower substrate of diamond drilled glass. Each composite assembly was then corona treated on its upper face and the thin upper substrate of circa 150 micrometres thick glass was positioned. Each resulting assembly was cured at 80°C for 45 minutes and subsequently had nylon tubes epoxy-glued into the drill holes to provide fluid connectivity.

The invention is illustrated in the drawings in which Fig. 1 shows a single throttle element Fig. 2 shows a channel throttle element Fig. 3 shows an array of elements Fig. 4 shows a planar throttle Fig. 5 shows a seal valve Figs. 6 and 7 show a means for reduction in the fluid volume of a valve Fig. 8 shows a seal valve whose seal has a narrow'lip'seated on a broader shoulder Figs. 9 shows a network perimeter scheme

Figs. 10 shows a network perimeter scheme with a compliant outer support area Fig. 11 shows a modified planar valve with asymmetric internal volumes when closed Fig. 12 details the operating sequence of a two-actuator pump employing asymmetrically-closing valves.

Referring to fig. 1, fig. la shows a plan view of the element and fig. lb shows a side view, the piezo electric actuator (1) acts on upper layer (4) which causes pinch (3) in flow channel to close. The channel (2), in which there is bi-directional flow is formed in PDMS layer (6), which is bonded to lower substrate (5). In use a voltage is applied to (1) which causes pinch (3) to close channel (2), when the current is turned off the actuator opens and flow resumes.

Referring to fig. 2 this shows PDMS islands (9) by this means isovolumetric deformation of the island's elastomer towards both sides of the channel wall is achieved enhancing the constriction the majority of isovolumetrically deformed elastomer will be forced into the flow channel rather than deforming equally in a direction congruent to the flow, as such a large proportion of the deformed material will be employed to constrict the flow.

Referring to fig. 3 this shows a radial array of elements (10) thereby exploiting the pressure field of a discoidal force generating element such as a piezoelectric disc (11).

Referring to fig. 4 this shows a plan view of a planar throttle in which (1) shows the perimeter of the actuator, (2) is the bi-directional flow channel, (12) is the PDMS support columns and (13) shows the planar throttling region. In use the actuator acts on the planar throttling region (13) to construct the flow channel (2).

Referring to fig. 5, fig. 5a shows a side view of a seal valve and fig. 5b shows a plan view. There is an annular seal (14) acted on by actuator (1) in top glass layer (15)

there is a PDMS cast chamber (16) through which are channels (17) and (20) and there are drill holes (18) through lower substrate (19). There is bi-directional flow as shown by the arrows. The annular seal fits around a central feed shaft (fig. 5b). The seal (14) is aligned with the centre of piezoelectric disc (1) thereby exploiting the maximum and symmetric displacement of the top substrate. The seal is planar with the surface plane of the top substrate interface resulting in a normally-closed element or it is be below the plane, thereby resulting in a normally-open element. Activation of (1) provides for a change of state.

Referring to fig. 6 fig. 6a shows a side view of a seal valve and fig. 6b shows a plan view of the valve of fig. 5 in which the mould depth is in two steps, an outer shallow area (22) and an inner deeper area (23).

Referring to fig. 7 in this embodiment the device has the same configuration as in fig. 6 but with delineation of the fluid flow area with a sealed wall (24) outside which a sponge-like array of columns (25) locates the top substrate whilst allowing it to travel freely.

Referring to fig. 8 there is a two level mould to form a seal and 8a is a plan view and 8b is a view along the line Z-Z'of 8a. In 8b the actuator (1) acts in an upward direction, and the planar seal (26) is open. The lower'pedestal'"shoulder seal" (27) confers resistance to deformation whilst the upper, narrower seal, "lip seal"mounted on the'pedestal' (28) minimises contact area and hence parasitic attachment forces.

By virtue of the reduced height of the upper seal element the top substrate (4) deformation will be proportionately large, resulting in increased proportional elastic deformation and hence restorative force to counteract the parasitic attachment. Such an increase in the restorative force in respect to the contact-related parasitic forces will enhance seal separation behaviour.

Referring to fig. 9, fig. 9a is plan view and fig. 9b is a view along line Z-Z'of fig. 9a.

There are a pair of separate fluid networks with a high interface perimeter length across a narrow elastomer wall. The interface area is aligned with the piezoelectric disc (1) thereby exploiting the maximum and symmetric displacement of the top substrate. The two networks are sealed by the top substrate (4). Activation of the piezo-electric actuator serves to lift the top substrate away from the elastomer surface thereby allowing fluid to flow between the two networks.

Referring to fig. 10 the concept of fig. 9 is further developed by filling the outer, non fluid carrying region outside with a sponge-like array of columns thereby locating the top substrate whilst allowing it to travel freely.

Referring to figs. 11, 11 a, lib and 11 c show the displacement of the top layer and fig. lid shows a side view. The figs. show a planar valve whose sealing point has been offset so as to result in an unequal displacement of fluid as the piezoelectric element continues to depress the top layer beyond the point where a seal has resulted. The piezo electric element (31) acts on glass layer (30). The chamber (32) in the PDMS layer on substrate (34) is divided into two unequal chambers (36) and (37) by column (33). Fig. 1 la shows the valve is open, lib is closed and l lc depressed beyond the sealing point.

By way of example, fig 12 shows a two actuator arrangement comprising a pair of planar pump-valves (as shown in fig 11) in diagrammatic-skeleton form to illustrate the actuator timing sequence to achieve pumping. It is to be noted that the larger volume semi-chambers resulting from closed sealing must be connected to each other. It is to be noted that the equal volume and temporally opposing semi-chambers are necessary to provide for the recovery stroke of the first chamber by virtue of the second chamber temporarily displacing fluid back to facilitate this.

It is to be noted that the configuration of fig 12 can also be operated at lower efficiency with a single actuator on the first chamber and the second chamber operating as a passive'slave'valve.