WO2014031532A1 - Microfluidic device for filtering fluids and dialysis - Google Patents

Abstract

Disclosed is a microfluidic device that includes a sealed body having an inlet port and an outlet port; and a plurality of channels formed in the sealed body, each of the channels being in fluid communication between the inlet port and the outlet port, the channels being defined by a support material and a nanoporous or microporous membrane connected to the support material; and a chamber inside the sealed body separated from the fluid communication of the inlet port, the plurality of channels, and the outlet port via the nanoporous or microporous membranes of the channels. Use of these devices for filtration of fluids, including for dialysis, is also disclosed.

Description

MICROFLUIDIC DEVICE FOR FILTERING FLUIDS AND DIALYSIS

[0001] This application claims the priority benefit of U.S. Provisional Patent Application Serial No. 61/684,798, filed August 19, 2012, which is hereby incorporated by reference in its entirety.

FIELD OF THE INVENTION

[0002] The present invention relates to a microfluidic device and its use for filtering fluids, particularly for dialysis.

BACKGROUND OF THE INVENTION

[0003] Clinical End-Stage Renal Disease (ESRD) affects 2 million people worldwide (Blagg, C. R., J. Nephrol. Italy (2011)) and this population has been growing at a rate greater than 8% (Schieppati & Remuzzi, Kidney Int. Suppl. United States (2005)). In the United States in 2010, nearly 600,000 patients received Renal

Replacement Therapy (RRT), almost 400,000 of which underwent dialysis (USRDS, 2012, "USRDS 2012 Annual Data Report: Atlas of Chronic Kidney Disease and End- Stage Renal Disease in the United States," National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD (2012)). The standard of care for ESRD, when renal transplantation is not available or feasible, is lifelong hemodialysis (HD) treatments at a frequency of 3 to 4 times per week. Even with frequent HD, life expectancies for ESRD patients aged 30 to 85 years are typically less than 10.5 years (Turin et al, Nephrol. Dial. Transplant, England (2012)). In an effort to improve both the access to dialysis and the quality of life for those on RRT, research groups are working on technologies for wearable HD (Fissell et al, Kidney Int. (2013); Lee & Roberts, "Automated Wearable Artificial Kidney (Awak): A Peritoneal Dialysis Approach," Proc. World Congress on Medical Physics and Biomedical Engineering: Diagnostic and Therapeutic

Instrumentation, Clinical Engineering, September 7, 2009 - September 12, 2009, Munich, Germany, 25 : 104- 107 (2009); Gura et al, "Technical Breakthroughs in the

Wearable Artificial Kidney (WAK)," Clin. J. Am. Soc. Nephrol. 4(9): 1441-1448 (2009)). These advances would not only provide lifestyle benefits of mobility and convenience, they could improve outcomes by reducing extracorporeal blood volumes and enabling more frequent or continuous dialysis. [0004] A prerequisite for wearable HD technology is the development of highly efficient membranes that can achieve standard toxic clearance rates with far less membrane. Clinical HD currently uses membranes that are ~10 microns thick with tortuous flow paths. These characteristics slow diffusion and convection through membranes and consequently dialyzers have long flow channels (15" to 18") and large

2 2

membrane surface areas (1.4 m to 2.4 m ) to achieve target clearance values. The extended extracorporeal circulation increases the risk of hemolysis (Vieira et al., "Hemolysis in Extracorporeal Circulation: Relationship Between Time and Procedures," Rev. Bras. Cir. Cardiovasc. 27:535-41 (2012)), the break down of red blood cells, and thrombus formation (Cosemans et al., "The Effects of Arterial Flow on Platelet

Activation, Thrombus Growth, and Stabilization," Cardiovasc. Res. 99(2):342-352 (2013)). Thus shorter dialysis flow paths not only enable portability, they have the potential to ameliorate other complications. Current work on wearable HD devices has focused on miniaturization of the fluidics, controls, and improved membrane selectivity (Fissell et al., "High-Performance Silicon Nanopore Hemofiltration Membranes," J. Membrane Science 326(l):58-63 (2009)) but has not addressed the need for increased membrane efficiency.

[0005] Thus, there remains a need for the development of an improved filtration device that can be used for HD treatments and overcome the above -identified deficiencies. The present invention is directed to overcoming these and other deficiencies in the art.

SUMMARY OF THE INVENTION

[0006] A first aspect of the invention relates to a microfluidic device that includes a sealed body having an inlet port and an outlet port; and a plurality of channels formed in said sealed body, each of said channels being in fluid communication between the inlet port and the outlet port, said channels being defined by a support material and a nanoporous or microporous membrane connected to the support material; and a chamber inside the sealed body separated from the fluid communication of the inlet port, the plurality of channels, and the outlet port via the nanoporous or microporous membranes of said channels.

[0007] In accordance with one embodiment, the device also includes a retention agent inside the chamber adjacent to the nanoporous or microporous membrane. The retention agent is suitable for retaining urea or creatinine removed from blood or plasma, and the microfluidic device is capable of use for microdialysis.

[0008] In accordance with another embodiment, the device also includes a second inlet port and a second outlet port formed in the sealed body and in fluid communication with the chamber. In this embodiment, dialysate flows through the chamber whereas blood or plasma flows through the microchannels on the opposite sides of the nanoporous or microporous membranes.

[0009] In preferred embodiments, the device is characterized as having microchannels with a Peclet number of about 1 or less than 1.

[0010] A second aspect of the invention relates to a method of filtering a fluid sample that includes: providing a microfluidic device according to the first aspect of the invention; and passing through the microfluidic device, from the inlet port to the outlet port, a fluid comprising one or more agents smaller than the dimension of nanopores or micropores formed in the nanoporous or microporous membrane, whereby the one or more agents are removed from the fluid sample and pass through the nanopores or micropores, and filtered fluid is recovered from the outlet port.

[0011] A third aspect of the invention relates to a method of performing dialysis that includes providing a microfluidic device according to a first aspect of the invention, such as the first embodiment or the second embodiment described above; and coupling the inlet port and the outlet port to the blood supply of a patient, and forcing blood to flow through the microfluidic device and over the nanoporous or microporous membranes, whereby biological contaminants are removed from the blood and pass through the nanopores or micropores, and filtered blood is delivered to the outlet port.

[0012] The present invention demonstrates the application of porous

nanocrystalline silicon (pnc-Si) membranes for inclusion in a small format HD device. These membranes, which were first described six years ago (Striemer et al, "Charge- and Size-Based Separation of Macromolecules Using Ultrathin Silicon Membranes," Nature 445 (7129): 749-53 (2007), which is hereby incorporated by reference in its entirety), are 100 to 1000 times thinner than conventional membranes and are therefore orders-of-magnitude more efficient for dialysis (Snyder et al., "An Experimental and Theoretical Analysis of Molecular Separations by Diffusion Through Ultrathin

Nanoporous Membranes," J. Membrane Sci. 369(1-2): 119-129 (2011), which is hereby incorporated by reference in its entirety). In fact, given the molecular thickness of these membranes (~15 nm) and their appreciable porosity (-15%), pnc-Si membranes operate near the maximum permeability that is achievable for a nanoporous membrane.

Additionally, the membrane pore sizes can be tuned to match specific molecular separation goals (Fang et al., "Methods for Controlling the Pore Properties of Ultra-Thin Nanocrystalline Silicon Membranes," J. Phys. Condens. Matter 22(45) :454134 (2010), which is hereby incorporated by reference in its entirety) and the silicon platform allows for scalable manufacturing and straightforward integration with fluidics. Material advancements that have allowed the manufacture of pnc-Si membranes with 26 fold more active membrane area than previously possible are now available. The design and operation of a benchtop microfluidic system that achieves target urea dialysis goals predicted from finite element models of the system is described herein. The use of surface functionalization to reduce both protein and cellular attachment to pnc-Si and render the membranes hemocompatible (Muthusubramaniam et al., "Hemocompatibility of Silicon-Based Substrates for Biomedical Implant Applications," Ann. Biomed. Eng. 39(4): 1296-305 (2011), which is hereby incorporated by reference in its entirety) is also demonstrated.

[0013] For embodiments designed for dialysis, the compact dialyzer will enable fabrication of a wearable dialysis device and embodiments characterized by a Peclet number of one or less will facilitate evaluation of various chip configurations. These devices afford advantages over traditional dialysis machines due to their compact form, thereby allowing them to be easily wearable while still performing the same level of dialysis. This will allow patients greater mobility and ease of use.

BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 illustrates a microfluidic device according to the present invention, which includes the device installed within a sealed body having inlet and outlet ports.

[0015] Figure 2A is a side elevational view of the microfluidic device according to one embodiment of the invention, with the entire front portion of the sealed body broken away to illustrate the microfluidic device contained therein. This embodiment utilizes a retentive agent. Figure 2B is a cross-sectional view along line 2B-2B in Figure 2A. Although only six microfluidic channels are illustrated, it should be appreciated that the device can include any number of microfluidic channels. [0016] Figure 3A is a side elevational view of the microfluidic device according to a second embodiment of the invention, with the entire front portion of the sealed body broken away to illustrate the microfluidic device contained therein. This is a parallel flow device. Figure 3B is a cross-sectional view along line 3B-3B in Figure 3A.

Although only six microfluidic channels are illustrated, it should be appreciated that the device can include any number of microfluidic channels.

[0017] Figure 4 is a perspective view of a parallel flow microfluidic filtration device that includes side ports, and transverse channels that feed or collect from all of the microchannels.

[0018] Figure 5 is a graph illustrating the results of an ELISA assay which confirmed the reduction in complement activation level of PEG-treated pnc-Si as compared to DEAE cellulose membrane (positive control) and Teflon (negative control).

[0019] Figures 6A-6D show the fabrication process for forming a silicon nitride scaffolding on one side of the pnc-Si membrane. A silicon substrate with a 20 nm thick silicon nanomembrane is coated with a 400 nm thick LCPVD nitride. Figure 6A shows the wafer is spin coated with photoresist ("PR"), which is patterned with the scaffolding structure. Figure 6B shows a reactive ion etch is used to transfer the photo-pattern of the PR to the nitride. Figure 6C shows the PR is removed, via plasma etching, leaving the patterned silicon nitride scaffolding on top of the membrane material. Figure 6D shows a micrograph of silicon nitride scaffolding supporting the silicon nanomembrane. The scaffolding is a repeating pattern of hexagons with 5 μιη wide supports. The width of the hexagons is 82 μιη. The porosity of the scaffolding is -90%.

[0020] Figures 7A-7D illustrate single-channel and multi-channel dialysis chips.

Figure 7A shows a single-slot dialysis chip used for conducting urea dialysis tests. The channel is 500 μιη wide on the membrane side and 10 mm long. Figure 7B shows a multi-channel dialysis chip (22 mm x 24 mm) with 13 parallel dialysis channels. The channels are 700 μιη wide and 13 mm long on the membrane side. Figure 7C shows a cross sectional drawing of a membrane chip in use and shows how the channel etched through the silicon becomes a microfluidic channel with the nanomembrane as a boundary. Figure 7D is a 3-D drawing of single-slot dialysis test device. Flexible tubing (1 mm ID) connects to glass capillaries inserted into PDMS cap with channels connecting to a fluidic channel in the bulk silicon over silicon nanomembrane. [0021] Figures 8A-8C show a design of multi-channel dialysis fluidics. Inlet and outlet connect to a single fluidic bus delivering fluid to the individual dialysis

microchannels. Figure 8A shows a COMSOL model with 100 μιη deep fluidic bus bar. Velocity field indicates flow will move more quickly down center channel and have nearly zero flow along the outermost channels. Figure 8B shows a COMSOL model with 300 μιη deep fluidic bus bar having even flow across all microchannels. The large volume of the bus bar allows for even fluid flow to all of the channels. Figure 8C shows an image of 300 μιη bus bar fabricated with silicone gasket and PDMS cap, and having nearly even flow in all microchannels as predicted by the model.

[0022] Figure 9 shows a drawing of experimental set-up. A syringe pump was used to flow sample though the single slot dialysis chip suspended in a stirred beaker of PBS then collected in a fraction collector that changes collection vials at 30 minute intervals.

[0023] Figures 1 OA- IOC show membrane selection. Figure 10A is a TEM image of ultrathin porous silicon nanomembrane. Light areas are pores. Figure 10B is a histogram of pore diameter (nm) of the membrane. The mean pore diameter is 13.6 nm and the porosity is 5.7%. The sharp cut-off of the histogram allows for high resolution separations. Figure IOC is an image of SDS-polyacrylamide gel showing that the cytochrome c passes from the retentate through the membrane to the filtrate, while the membrane retains the albumin and even the cytochrome c dimer.

[0024] Figure 11 shows long term urea dialysis testing results. A solution of

30% serum and 70%> PBS with 0.5 mM urea was passed through the single slot dialysis device at 5.6 μί/ηιίη. The device was evaluated over time to examine if steady dialysis rate at target concentrations could be achieved. The concentration entering the device is shown in red (upper dashed line) and the predicted exit (goal) concentration is shown green (lower dashed line). The black line shows urea dialysis rate when the system is pretreated, whereas the blue line shows fluctuating dialysis rate when the system is not pretreated. With priming and degassing pretreatment, the exit concentration initially falls short of the predicted value (Co = 0.34 mM) but settles to value over a day of continuous use.

[0025] Figures 12A-12B show protein and cell adhesion on functionalized membranes. Figure 12A shows fluorescence of absorbed protein normalized to bare pnc- Si. All treatments reduce protein binding to more than 5% of control. Figure 12B shows platelets are double-labeled with platelet marker (CD41 in green) and activated platelet marker (CD62P in red). ADP is used as a positive control to induce platelet activation and aggregation. PEG treatment significantly reduces platelet binding and activation.

[0026] Figures 13A-D illustrate the 13 microchannel dialysis device of Figures 7B and 8A-C. The ability of the dialysis device to facilitate dye transfer to the dialysate is illustrated by passage of dye in the dye-labeled sample, into the microchannels, and then the dialysate over the time course illustrated in Figures 13A-D.

DETAILED DESCRIPTION OF THE INVENTION

[0027] The present invention relates to microfluidic devices that can be used for filtering fluid samples, including the performing of dialysis on a patient's blood.

[0028] Referring to Figures 1-3, the microfluidic device 10 of the present invention includes a sealed body 12 having an inlet port 14 and an outlet port 16, a plurality of channels 20 formed in said sealed body, each of said channels being in fluid communication between the inlet port and the outlet port, said channels being defined by a support material 22 and a nanoporous or microporous membrane 24 connected to the support material; and a chamber 26 inside the sealed body separated from the fluid communication of the inlet port, the plurality of channels, and the outlet port via the nanoporous or microporous membranes of said channels.

[0029] As shown in Figure 1, the sealed body includes two halves 13a, 13b. The body halves can be formed of any suitable thermoplastic material, and can be secured together via suitable adhesives, sonic welding of the joints, or a coupling system and gasket. The sealed body should be fluid tight.

[0030] The support 22 that defines the channels 20 includes a substrate 30 and an elastomeric polymer layer 32 bonded to the substrate.

[0031] Although silicon is by far the most common substrate used in forming microfluidic devices of this type, persons of skill in the art should appreciate the other materials can be used including, without limitation, undoped germanium, p-doped silicon or germanium, n-doped silicon or germanium, a silicon-germanium alloy, and Group III element nitrides. Dopants are well known in the art and may include, without limitation, (CH₃)₂Zn, (C₂H₅)₂Zn, (C₂H₅)₂Be, (CH₃)₂Cd, (C₂H₅)₂Mg, B, Al, Ga, In, H₂Se, H₂S, CH Sn, (C₂H₅) S, SiH₄, Si₂H₆, P, As, and Sb. The dopants can be present in any suitable amount. [0032] The membrane formed on the substrate can be the same material or an additional material applied thereon, such as silicon nitride.

[0033] Nanoporous nanocrystalline (pnc) silicon membranes between about 2 nm to about 500 nm are described in U.S. Patent Application Publ. No. 20060278580 to Striemer et al., which is hereby incorporated by reference in its entirety. Pnc-silicon membranes of less than about 100 nm (thickness) are particularly desirable, although membranes less than 50 nm, less than 45 nm, less than 40 nm, less than 35 nm, less than 30 nm, less than 25 nm, less than 20 nm, less than 15 nm, or less than 10 nm are both feasible and useful. Depending on the fabrication conditions, these membranes can be formed with a porosity of at least about 1 percent and up to about 40 percent, preferably about 10 to 20 percent; a pore size cutoff of less than 30 nm (or less than 25 nm, less than 24 nm, less than 23 nm, less than 22 nm, less than 21 nm, less than 20 nm, less than 19 nm, less than 18 nm, less than 17 nm, less than 16 nm, less than 15 nm, less than 14 nm, less than 13 nm, less than 12 nm, less than 11 nm, or less than 10 nm); or a combination of any of these two features. Figures 10A-B illustrate one such

combination (porosity 5.7%; mean pore diameter 13.6 nm; and a cutoff ~22 nm).

[0034] Membranes that are less than about 100 nm thick can be supported on the side opposite of the microfluidic channel by deposition of a supporting structure having, e.g., a grid-like or honeycomb configuration. Silicon nitride is a preferred material for use in these supporting structures.

[0035] Microporous membranes are preferably, though not necessarily, formed of silicon nitride. The pores in silicon nitride membranes can be formed by electron beam lithography following by reactive ion etching or focused ion beam etching (Tong et al, "Silicon Nitride Nanosieve Membrane," Nano Letters 4(2):283-287 (2004), which is hereby incorporated by reference in its entirety. Pore diameters larger than about 50 nm can be generated by these pore-forming techniques, including pores in the microporous range of 100 nm to about 500 nm.

[0036] When the membrane is formed of any of the materials described above, the membrane can be surface treated to render the membrane resistant to biofouling. One example of this, illustrated in the accompanying examples, is the use of

polyethylene glycol that is tethered to the surface of the membrane via silanization. PEG-ylation can be performed on only the side of the membrane this is exposed to the interior of the microfluidic channel, or alternatively PEG-ylation can be performed on both exposed surfaces of the membrane. Alternatively, ethanolamine can be used as described in the examples.

[0037] The elastomeric polymer material is preferably a silicone elastomeric material such as polydimethylsiloxane ("PDMS", e.g., Dow Corning Sylgard^® 184) (McDonald et al, "Fabrication of Micro fluidic Systems in poly(dimethylsiloxane)," Electrophoresis 21 :27-40 (2000), which is hereby incorporated by reference in its entirety). PDMS is a particularly well studied material for the construction of microfluidic systems. It is optically transparent, and has a refractive index that is much lower than that of silicon. PDMS has a hydrophobic surface after polymerization, but the surface of PDMS can be treated with a surfactant, oxygen and plasma, or

atmospheric RF to become hydrophilic (Hong et al., "Hydrophilic Surface Modification of PDMS Using Atmospheric RF Plasma," Journal of Physics: Conference Series 34:656-661 (2006), which is hereby incorporated by reference in its entirety). This hydrophilicity assists not only in bonding the polymer layer to the substrate, but also to decrease surface tension and bio fouling within the microchannels to allow fluids to move easily along those channels. Chemical treatment methods are also available for improving the performance of PDMS (Lee and Voros, "An Aqueous-based Surface Modification of poly(dimethylsiloxane) with poly(ethylene glycol) to Prevent

Biofouling," Langmuir 21 : 11957-11962 (2004), which is hereby incorporated by reference in its entirety).

[0038] The use of these materials to form a microfluidic device having reduce channel heights and, in particular, low channel volume to membrane surface area ratios is described in provisional U.S. Patent Application Serial No. 61/684,796, entitled "Method For Preparing Microfluidic Device With Reduced Channel Height," to Johnson and McGrath, filed August 19, 2012, as well as corresponding PCT Application

PCT/US 13/55541, filed August 19, 2013, each of which is hereby incorporated by reference in its entirety.

[0039] As noted above, the active area of the membrane is preferably maximized, and this also preferably achieves a low microchannel volume to membrane surface area ratio. For example, a microfluidic device with 300 μιη deep, 1mm wide channel (at membrane), has a V/A ratio of 0.345, whereas a channel reduced to 100 μιη but having the same width has a V/A ratio of 0.105 and a channel reduced to 10 μιη but having the same width has a V/A ratio of 0.01005. [0040] The dimensions of the device are preferably designed to ensure the Peclet number is about 1 , or even less than 1. A device with a Peclet number of about 1 means that the time for the urea to diffuse across the channel (i.e., pass over the membrane), Td, is roughly equal to the time it takes the fluid to traverse the dialyzer (i.e., diffuse from the sample channel, through the membrane, and arrive in the outlet channel), T_r. Thus, where T_d = T_r, the ratio is 1. Due to the nature of diffusion, e^"1 (37%) of the urea remains in the retentate while 1-e^"1 (63%) has diffused into the permeate. In one embodiment, using a membrane width of 700 μιη, channel heights of 100 μιη, and a length of 15 mm, a Peclet number of 1 is achieved with a device that contains about 290 channels to achieve a clinically relevant flow rate of 10 mL/min. In another embodiment, using a membrane width of 700 μιη, channel heights of 50 μιη, and a length of 15 mm, a Peclet number of 1 is achieved with a device that contains about 170 channels to achieve a clinically relevant flow rate. In these embodiments, increasing the number of channels, while maintaining the flow rate, will reduce the Peclet number to less than 1. Other embodiments, with different channel dimensions and membrane dimensions are, of course, also possible.

[0041] The support 22 is installed within the body 12 such that it is sealed about its perimeter via gasket or sealant 34. This separates the device into the upper chamber 36 and the lower chamber 26. In certain embodiments, the device may include a common bus or manifold 38 that delivers liquid from the inlet port into the micro fluid channels. As noted in the Examples, the bus or manifold can be dimensioned to create uniform or nearly uniform flow through each of the microchannels. In other

embodiments, the upper chamber 36 can be divided into an inlet side and outlet side by a gasket or the like, and the microfluidic channels share a common opening (see Figure 4) that is exposed to the inlet side of the upper chamber as well as a common opening (also see Figure 4) that is exposed to the outlet side of the upper chamber. In the latter embodiment, the entire inlet side of the upper chamber will fill with fluid entering the device from the inlet port before it passes into the microchannels for filtration/cleansing, and the entire outlet side of the upper chamber will fill with filtered/cleansed fluid before exiting the device through the outlet port.

[0042] Referring specifically to Figures 2A-B, one embodiment of the microfluidic device includes a retention agent 40 positioned inside the chamber 26 adjacent to the nanoporous or microporous membrane 24. The retentive agent can be suitable for the capture and retention of any type of component that is desired for removal from the fluid entering the microfluidic device. In general, the retentive agent is an adsorbent or similar agent that can bind specifically or nonspecifically to a precipitate dissolved or suspended in the fluid.

[0043] In one embodiment, the retentive agent is specific for urea, creatinine, or both, and the device is suitable for use as a microdialysis device to remove these impurities from blood/plasma. Exemplary retentive agents for this purpose include, without limitation, activated carbon adsorbent, carboxen 563 adsorbent, oxidized activated carbon adsorbent, chitosan/Cu(II) adsorbent (Liu et al., "Preparation and Characterization of Chitosan/Cu(II) Affinity Membrane for Urea Adsorption," JAppl Polym Sci 90: 1108-1112 (2003), which is hereby incorporated by reference in its entirety) an oxidized, cross-linked β-cyclodextrin polymer (Shi et al, "Novel Composite Adsorbent for Adsorption of Urea," Polymers Adv. Technol. 10(l-2):69-73 (1999), which is hereby incorporated by reference in its entirety); starch 3,5-dinitrobenzoate (Yu et al., "Synthesis of a Novel Starch Derivative and its Absorption Property for

Creatinine," Yao Xue Xue Bao 38(3): 191-195 (2008), which is hereby incorporated by reference in its entirety), charcoal, and combinations thereof. Other materials that serve this same purpose can also be used alone or in combination with the materials identified below.

[0044] In this embodiment, the lower chamber 26 can also be loaded with sodium and/or potassium-dense buffer solution that inhibit the retention of these blood/plasma ions within the lower chamber.

[0045] In another embodiment, the retentive agent is specific for retention of industrially valuable precipitates or environmentally dangerous contaminants in an industrial water supply, and the device is suitable for use in recovering such agents.

Exemplary retentive agents for this purpose include, without limitation, organo-ceramic materials disclosed in U.S. Patent Nos. 7,358,318 to Tavlarides et al, which is hereby incorporated by reference in its entirety. Precipitates that can be removed include, without limitation, cerium, neodymium, praseodymium, ruthenium, rhodium, palladium, osmium, iridium, platinum, silver, gold, antimony, arsenic, beryllium, cadmium, chromium, copper, lead, mercury, nickel, selenium, thallium, tin, zinc, molybdenum, cobalt, technetium, rhenium, cesium, and strontium. [0046] Referring specifically to Figures 3A-B, one embodiment of the

microfluidic device includes lower chamber 26 that can be provided with a polymer (e.g., PDMS) block that defines a common passage 48 filled with a dialysate solution. The common passage 48 is exposed to all of the membranes 24. In this embodiment, the common passage 48 is in fluid communication with an inlet port 44 and an outlet port 46. The inlet port 44 can be coupled to a dialysate supply reservoir and the outlet port can be coupled to a waste reservoir. The dialysate can flow through the common passage 48 under influence of gravity or via pump. The direction of flow through common passage 48 is preferably opposite the direction of flow through the channels 20.

[0047] Figure 4 shows an alternative construction for this type of device.

According to this embodiment, the microfluidic device 80 includes a device substrate 90 formed of silicon and having a plurality of parallel, reduced-height microchannels 83 with a porous nanocrystalline membrane (preferably less than 50 nm thick, more preferably less than 25 nm thick) formed at the base of each microchannel. Above substrate 90 is a polymer layer 82 that has a conforming lower surface that partially defines the plurality of microchannels. Formed in the polymer layer 82 is a pair of transverse microchannels 84, 86, which allow for coupling of the plurality of parallel, reduced-height microchannels to a common inlet 88 that communicates with the microchannel 84 and a common outlet 90 that communicates with the microchannel 86. As explained in the Examples, the transverse microchannels 84, 86 can be dimensioned to achieve uniform flow throughout the plurality of reduced-height microchannels. The polymer layer 82 is capped by a layer 92, which is formed by PDMS, glass, or a thermoplastic material. Below device substrate 90 is a polymer layer 94 that defines a common chamber communicating with each of the plurality of parallel, reduced-height microchannels via the porous nanocrystalline membranes. The common chamber is also capped by a layer 96, which is formed by PDMS, glass, or a thermoplastic material. The common chamber includes an inlet 98 and an outlet 100. This embodiment can be used as a parallel flow filtration system for, e.g., dialysis. A fluid to be filtered is delivered via inlet 88 through each of the microchannels, where the fluid flows above the porous membrane, and exits via outlet 90 as a filtered fluid. A counter- flow fluid is delivered via inlet 98 to the common chamber, where it collects the filtered materials before exiting via outlet 100. In this embodiment, it should be appreciated that multiple devices can be connected in parallel at their inlet ports to a source of fluid. [0048] In each of these embodiments, a pump can be placed in fluid

communication with inlet 14 or outlet 16, inlet 44 or outlet 46, inlets 88/98 or outlets 90/100 or one of each, to promote the delivery of fluid through the microchannels 20 (Figures 2A-B, Figures 3A-B), passage 48 (Figure 3B), and/or microchannels 83 (Figure 4).

[0049] In a further aspect of the invention, filtration of a fluid can be achieved by passing through a microfluidic device of the present invention, from the inlet port to the outlet port, a fluid containing one or more (filtrate) agents smaller than the dimension of nanopores or micropores formed in the nanoporous or microporous membrane, whereby the one or more agents are removed from the fluid sample and pass through the nanopores or micropores, and filtered fluid is recovered from the outlet port. The filtrate remains entrapped within the retentive agent or is passaged through the outlet port 46 or 100 with the dialysate.

[0050] In yet another aspect of the invention, filtration of a patient's blood supply in a microdialysis procedure can be carried out by coupling the inlet port and the outlet port to the blood supply of a patient, and forcing blood to flow through the microfluidic device and over the nanoporous or microporous membranes, whereby biological contaminants (e.g., urea and creatinine) are removed from the blood and pass through the nanopores or micropores, and filtered blood is delivered to the outlet port. In this embodiment, the patient's blood pressure may be sufficient to cause the blood to flow through the device. Alternatively, a pump can be supplied to facilitate redelivery of the cleansed blood to the patient's vascular system.

[0051] In this aspect of the invention, the microdialysis can be performed continuously except for periodic replacement of dialysate or retention agent.

Alternatively, the microdialysis can be performed continuously for a period of time, e.g., overnight or for 24 hours, after which the patient can halt dialysis for a limited duration of time before repeating the process.

EXAMPLES

[0052] The following examples are intended to illustrate practice of the invention, and are not intended to limit the scope of the claimed invention. Example 1 - Fabrication of High Area Porous Nanocystalline Silicon Membrane

[0053] The fabrication of pnc-Si membranes has been previously described in detail (Snyder et al., "An Experimental and Theoretical Analysis of Molecular

Separations by Diffusion Through Ultrathin Nanoporous Membranes," J. Membrane Sci. 369(1-2): 119-129 (2011); Fang et al, "Methods for Controlling the Pore Properties of Ultra-Thin Nano crystalline Silicon Membranes," J. Phys. Condens. Matter

22(45):454134 (2010), which are hereby incorporated by reference in their entirety). Briefly, the membrane material is formed by annealing an ultrathin layer of amorphous silicon deposited on a silicon wafer by sputter deposition. Annealing creates both nanocrystals and adjacent nanopores that span the thickness of the annealed layer {see Figure 10A). Annealing temperature, layer thickness, substrate bias, and other fabrication parameters are used to control pore sizes. The membrane is made

freestanding by chemical etching through the backside of the supporting wafer. This etching process naturally creates channels that conform to the area of the freestanding membrane.

[0054] To increase the freestanding area of membranes while maintaining mechanical integrity, a silicon nitride (SiN) based support scaffold (Figures 6A-D) was developed. The technique involved the use of low-pressure chemical vapor deposition (LPCVD) to place a 400 nm thick low stress (250 MPa) silicon nitride layer above the amorphous silicon layer. Standard photolithography to pattern tessellated hexagons with 42 μιη openings and 5 μιη wide frames was used. Reactive ion etching (RIE) was used to transfer the photo-pattern into the nitride, and the Si0₂ layer above the pnc-Si (Striemer et al., "Charge- and Size-Based Separation of Macromolecules Using Ultrathin Silicon Membranes," Nature 445(7129):749-53 (2007); Snyder et al, "An Experimental and Theoretical Analysis of Molecular Separations by Diffusion Through Ultrathin

Nanoporous Membranes," J. Membrane Sci. 369(1-2): 119-129 (2011), which are hereby incorporated by reference in their entirety) served as an etch- stop. The resulting support structure reduces the effective porosity of the hybrid membrane by only -12%. Example 2 - Construction of Single Channel Dialysis Device

[0055] Single channel devices were created by incorporating fluidic components formed in polydimethylsiloxane (PDMS) with 11 mm x 20 mm membrane chips (Figure 7A) in the test fixture. The test fixture consisted of two acrylic plates that were used to clamp the flow channels to the hybrid membrane (Figure 7D). The fluidic channel on the backside of silicon-etched channel was bounded by pnc-Si/SiN hybrid membrane on one side and the cap of PDMS on the other side (Figure 7C). The flow channels, inlet/outlet, were fabricated with the Sylgard 184 PDMS (Dow Corning, Midland, MI) patterned and cured on a custom-ordered SU-8 mold with a feature height of 300 μιη (Stanford

Micro fluidics Foundry, Stanford, CA). Holes for the inlets and outlets were punched into the cured PDMS using a blunt 20-gauge needle (Small Parts Inc., Logansport, IN). Glass microcapillaries with inner/outer diameter (ID/OD) of 500 μηι/900 μιη (Friedrich & Dimmock, Inc., Millville, NJ) were cut into shorter segments and inserted into the punched holes in the flow compartment as the adaptors for tubing attachment. Tygon tubing with ID/OD of 1/32" / 3/32" (Saint-Gobain Performance Plastics Corporation) were used to connect the glass microcapillaries to the syringe and the fraction collector (see Figure 9). Example 3 - Construction of Multichannel Dialysis Device

[0056] Multichannel devices were constructed in a similar fashion as single channel devices, except the membrane chips were 22 mm x 24 mm with 13 parallel microfluidic channels (Figure 7B). An important consideration for the multichannel device was the design of a fluid bus or manifold that distributes flow evenly across all 13 channels. The bus system was designed with the use of a finite elements model of the multichannel device (COMSOL Multiphysics, Stockholm SWEDEN). Figure 8A illustrates that a shallow straight bus (height = 100 μιη) results in flow rate that is much higher in the central channels near the input than at the periphery. Using a flow rate of 5.6 μΕ/ηιίη in each channel, it was found that uniform flow can be achieved using a thin bus that expands gradually from the source or by increasing the height of a straight bus bar. Using a bus bar of height 300 μιη for example, simulations showed a flow of 5.6 μΕ/ηιίη in all channels with less than a 1% deviation (Figure 8B). A fluidic chip with a 300-μιη deep bus bar was fabricated, and used to visualize water flow with dye.

Experimental tests verified a relatively uniform dye front across all 13 channels (Figure 8C). As shown in Figures 13A-D, the dye is rapidly transferred from the sample to the dialysate over a short period of time. Example 4 - Urea Dialysis Studies

[0057] Solutions containing urea and fetal bovine serum (FBS) were pumped through single channel devices to measure the rate of urea dialysis in the presence of a complex protein background. The inlet port was connected to a syringe pump with a 10 mL syringe filled with a solution of 0.5 mM urea in 30% FBS and 70% PBS. Keeping serum at levels below 50%> allowed the impact of viscosity on fluid flow and diffusion to be neglected while still presenting a complex protein environment to the membrane (Yadav et al., "Viscosity Analysis of High Concentration Bovine Serum Albumin Aqueous Solutions," Pharm. Res. 28(8): 1973-83 (2011), which is hereby incorporated by reference in its entirety). The syringe pump moved the solution though the single slot dialysis chip which was suspended in a stirred beaker of 100%) PBS exposing the backside of the membrane to a vigorously stirred beaker containing 100% PBS. Dialyzed samples were collected with a fraction collector that switched collection vials every 30 minutes (Figure 9). The experiment was conducted for over 30 hours pumping at a rate of 5.6 μί/ηιίη. Experiments were performed at 4 °C to minimize evaporation of the collected samples. Urea concentrations in collected fractions were measured by absorbance using a urea assay kit as described by the manufacturer (Abeam, Cambridge, MA). The membranes used in these experiments were not functionalized to prevent fouling. This was done to collect base line data on the untreated membranes. Once the static functionalization studies have been completed, additional experiments will be run with modified membrane surfaces to show the effects, if any, of the functionalization on long term clearance (see Example 5 below).

[0058] Dialysis experiments were done with and without priming of the device and degassing of solutions. Priming was performed by first wetting the nanomembrane with isopropanol and drawing additional isopropanol through the channel filling the tubing connected to the inlet port. A syringe was filled with degassed PBS and pumped through the device for >1 day before the introduction of serum and urea. In this manner the isopropanol was chased with more than 4,000 volumes of PBS and should have had no residual effect when protein was introduced. The FBS/PBS solution for this experiment was also degassed prior to pumping through the device.

[0059] Urea clearance rates are calculated with the following expression

(Pisitkun et al., "A Practical Tool for Determining the Adequacy of Renal Replacement Therapy In Acute Renal Failure Patients," Contrib. Nephrol. 144:329-49 (2004), which is hereby incorporated by reference in its entirety):

where Ci and Co are the concentration of solute in the blood at the input and output of the dialyzer, and Qi and Qo are the flow rates at the input and output. Because the dialysis systems are open and the pressure in the beaker is lower than the pressure in the channel, the exit flow rate is lower than the input flow rate as urea is removed by convection through the membrane (ultrafiltration) in addition to diffusion based dialysis. Another metric for evaluating the dialysis membrane is the instantaneous urea dialysis rate.

The instantaneous urea dialysis rate is defined as:

where Ci and Co are defined as above for Eq. 1. Because ultrafiltration through the membrane removes urea and fluid at the same rate, it does not lower the sample concentration. Thus the instantaneous urea dialysis rate only measures the effectiveness of the diffusive component of urea removal.

[0060] A finite element convection/diffusion model was developed to guide the device design and predict urea dialysis rates. The model was developed with COMSOL with the addition of the microfluidics module. For urea dialysis studies with single channel devices as shown in Figure 7A, the COMSOL simulation considers a structure with a single rectangular channel with inlet and outlet ports. The channels are 300 μιη tall, 500 μιη wide, and 10 mm long. Counter flow through the underlying dialysate chamber is configured to mimic the conditions in a stirred beaker. The starting point for the COMSOL model was the 2D equations for flow and diffusion in a rectangular channel. It was assumed the stirred beaker, with a volume of 250 mL, would act as a perfect sink for the urea diffusion for the 5 mL of 0.5 mM urea. The COMSOL model was tuned by adjusting the dialysate flow until it matched the results of the 2D analytical solution. Urea diffusion was considered to take place at the rate of free diffusion (Snyder et al., "An Experimental and Theoretical Analysis of Molecular Separations by Diffusion Through Ultrathin Nanoporous Membranes," J. Membrane Sci. 369(1 -2): 119-129 (2011), which is hereby incorporated by reference in its entirety).

[0061] The ability of pnc-Si membranes to dialyze urea at rates predicted by a

COMSOL-based convection-diffusion model that assumes uninhibited diffusion of urea through the membrane (Snyder et al., "An Experimental and Theoretical Analysis of Molecular Separations by Diffusion Through Ultrathin Nanoporous Membranes," J. Membrane Sci. 369(1-2): 119-129 (2011), which is hereby incorporated by reference in its entirety) was assessed. Negligible resistance to diffusion of toxins is the key to enabling high clearance rates in portable devices. Simulations were configured to match the physical dimensions of single channel dialysis chips (Figure 7A) with flow rates of

[0062] Figure 11 shows urea dialysis as a function of time in a case where the channel was first primed with isopropanol and PBS flushes of the system to ensure properly wetting and a case in which it was not. Without priming it was found that urea dialysis values included episodes of near zero urea dialysis, which was assumed to be due to incomplete wetting of the membrane (Kim et al., "Stabilizing Nanometer Scale Tip-to-Substrate Gaps in Scanning Electrochemical Microscopy Using an Isothermal Chamber for Thermal Drift Suppression," Anal. Chem. 84(8):3489-3492 (2012), which is hereby incorporated by reference in its entirety) and/or trapping of gas bubbles beneath the membrane. With priming and degassing, it was found that the exit concentration initially falls short of the predicted value (Co = 0.34 mM) then settles to value over a day of continuous use. Thus, instead of finding that membrane fouling slowed urea dialysis with time, dialysis rate actually improved. It is believed that ultrafiltration (passage of fluid by convection through the membrane) slows as serum proteins adsorb to the membrane over time and, thus, transport occurs more by pure diffusion, as assumed in the COMSOL model. Based on recovered volumes after the experiments, about 17% of the volume entering the membrane passes through by ultrafiltration.

[0063] Consistent with the above approach, the 13 microchannel dialysis device with a surface protected in accordance with the following Examples will be used to assess dialysis rates.

Example 5— Membrane Selectivity

[0064] The separation characteristics of membranes with average pore sizes between 10 nm and 30 nm were examined for the ability to separate bovine serum albumin (BSA) (MW = 66 kD) from β2 microglobulin (MW = 12 kD), the largest protein toxin in hemodialysis. Cytochrome c (MW = 12 kD) was used as a surrogate for β2 microglobulin because the molecule's strong visible absorbance at 415 nm facilitates detection in the presence of albumin. Cytochrome c concentrations were 1 mg/mL and BSA concentrations were 1.33 mg/mL in phosphate buffered saline (PBS). Cytochrome c and BSA were purchased from Sigma- Aldrich (St. Louis, MO).

[0065] pnc-Si membranes that had been functionalized with EA with average pore diameters between 10 nm and 20 nm were tested for the ability to fractionate solutions of cytochrome c and BSA using both diffusion (Snyder et al, "An

Experimental and Theoretical Analysis of Molecular Separations by Diffusion Through Ultrathin Nanoporous Membranes," J. Membrane Sci. 369(1-2): 119-129 (2011), which is hereby incorporated by reference in its entirety) and pressurized flow (Gaborski et al., "High-Performance Separation of Nanoparticles with Ultrathin Porous Nanocrystalline Silicon Membranes," ACS Nano. 4(11):6973-6981 (2010), which is hereby incorporated by reference in its entirety) as transport modes. It was found that membranes with a cutoff at the low end of this range (Figure 10A-B) exhibited the desired separation characteristics in both modes. While the membrane pore sizes are far larger than the molecular sizes of cytochrome c (~2.4 nm) and BSA (~6.8 nm) it has been consistently found that electrostatic interactions and protein adsorption reduce the effective pore sizes of pnc-Si membranes compared to pore sizes measured by electron microscopy (Snyder et al., "An Experimental and Theoretical Analysis of Molecular Separations by Diffusion Through Ultrathin Nanoporous Membranes," J. Membrane Sci. 369(1 -2): 119-129 (2011); Gaborski et al., "High-Performance Separation of Nanoparticles with Ultrathin Porous Nanocrystalline Silicon Membranes," ACS Nano. 4(11):6973-6981 (2010), which are hereby incorporated by reference in their entirety).

[0066] Figure IOC shows a SDS-PAGE gel displaying the results of a diffusion experiment in which a small pore membrane transmits monomeric cytochrome c and retains both albumin and dimeric cytochrome c. SDS resistant dimeric cytochrome c is a known contaminant in commercial cytochrome c, because of its tendency to aggregate during detergent based purification (Hirota et al., "Cytochrome c Polymerization by Successive Domain Swapping at the C-Terminal," Proc. Nat'l. Acad. Sci. U.S.A.

107(29): 12854-9 (2010), which is hereby incorporated by reference in its entirety). Here, the dimeric contaminant provides a useful measure of the resolution of pnc-Si, indicating the ability of the membranes to discriminate between monomeric and dimeric

cytochrome c. Example 6 - Membrane Surface Functionalization and Its Affect

[0067] Protein and cellular binding has the potential to trigger immune or coagulation cascades. Therefore, as Muthusubramaniam et al. ("Hemocompatibility of Silicon-Based Substrates for Biomedical Implant Applications," Ann. Biomed. Eng. 39(4): 1296-305 (2011), which is hereby incorporated by reference in its entirety) investigated, the potential of short hydrophilic molecules to block protein and cellular adhesion to silicon membranes was investigated. Successful attempts to make inert surfaces with PEG polymer brushes (Hucknall et al, "Versatile Synthesis and

Micropatterning of Nonfouling Polymer Brushes on the Wafer Scale," Bio interphases 4:FA50-FA57 (2009), which is hereby incorporated by reference in its entirety) are not appropriate, because the 50 nm to 100 nm thick coatings will occlude pores of the pnc-Si membranes. Thus, the objective was to establish coatings that are much smaller than pore sizes, yet still prevent biological adhesion.

[0068] Methods for grafting short PEG (6 mer to 10 mer) to silicon at high density (Shestopalov et al., "Soft-Lithographic Approach to Functionalization and Nanopatterning Oxide-Free Silicon," Langmuir 27(10):6478-6485 (2011), which is hereby incorporated by reference in its entirety) were compared with the proprietary approach of a local diagnostics device manufacturer (Adarza Biosystems, West

Henrietta, NY), and both were tested protein and cellular binding.

[0069] Two chemical modifications of the silicon surface to reduce nonspecific protein binding were compared. In the first approach, Adarza Biosystems Inc. was contracted to vapor deposit a proprietary monolayer silane (sLink) to create an amine- reactive surface that was subsequently reacted with (9 mer to 12 mer) amino-terminated PEG or ethyl-amine (EA). The amino-PEG molecules were reacted with activated membranes in either toluene or dichloromethane (DCM), while EA reactions were performed entirely in vapor phase.

[0070] Second, procedures were used for creating highly stable PEG linkages by replacing unstable silicon-oxygen bonds with silicon-carbon bonds. First, oxide-free silicon was created and subsequently reacted to yield an inert methyl-terminated primary layer. A secondary layer was attached to the primary monolayer via carbene insertion to yield stable C-C bonds. This overlayer is comprised of densely spaced N- hydroxysuccinimide (NHS) esters that can react with amine-terminated molecules (Shestopalov et al., "Soft-Lithographic Approach to Functionalization and Nanopatterning Oxide-Free Silicon," Langmuir 27(10):6478-6485 (2011), which is hereby incorporated by reference in its entirety). Specifically, a liquid based reaction of either 2 mmol or 4 mmol dodecaoxaheptatriacontane (m-dPEG®i2-amine; Quanta Biodesign Limited) was used to yield the desired PEG overlayer.

[0071] Surface functionalized membranes were then studied for their ability to prevent protein adhesion, platelet activation and adhesion, as well as Complement C3a activation. To test the ability of surface functionalization to prevent protein and cell adhesion, 'windowless' 1 cm x 1 cm membrane chips were fabricated. These chips were formed without performing a backside etch through the supporting silicon and therefore are identical surfaces to dialysis chips. This format was designed to simplify handling as drying steps can fracture membranes on standard chips.

[0072] After a pre-wetting rinse in PBS, test chips were incubated with 5 mg/mL fluorescein-conjugated bovine serum albumin (f-BSA) in PBS in a humidified chamber for 12 hours at 4 °C. The samples were rinsed in PBS then DI water and dried with filtered compressed air. Measurements were taken with a Zeiss Axiovert fluorescent microscope (Romulus, MI) and analyzed with ImageJ software for image intensity. Background measurements were determined from untreated chips and subtracted from sample data. The background signal subtracted is an order of magnitude lower than the signal from the positive control ensuring that protein binding is not significantly underestimated.

[0073] Figure 12A shows the results of protein binding studies. In these experiments, fluorescently tagged BSA was incubated with membrane chips in a humidified chamber for 12 hours at 4 °C and binding was assessed by fluorescence microcopy. All surface chemistries reduced protein binding to less than 5% of untreated controls. Higher concentrations of amine-reactive PEG yielded slight improvements for the in-house reactions, while DCM gave slightly better results than toluene when used as a solvent. The most impressive results were obtained for ethanolamine (EA), which leaves the surface with only a terminal hydroxyl group connected to the surface through a diamine linkage. Because of the small size of EA (~1 nm), the entire process is done in vapor phase and likely has the highest density of any of the test surfaces. The temporal stability of this chemistry, however, will be further examined as compared to surfaces introduced via liquid-based processes. [0074] The ability of PEG-coatings to reduce platelet adhesion and activation was also examined. Platelet rich plasma (PRP) was collected from healthy donors following established protocols (Landesberg et al., "Quantification of Growth Factor Levels Using a Simplified Method of Platelet-Rich Plasma Gel Preparation," J. Oral Maxillofac. Surg. 58(3):297-300; discussion 300-1 (2000), which is hereby incorporated by reference in its entirety). Platelet activation and adhesion protocols largely followed those of Muthusubramaniam et al. ("Hemocompatibility of Silicon-Based Substrates for Biomedical Implant Applications," Ann. Biomed. Eng. 39(4): 1296-305 (2011), which is hereby incorporated by reference in its entirety), but are detailed here for accuracy. PRP (200 μΐ) was dispensed on membrane chips with or without prior oxygen-plasma treatment or onto control surfaces for 2 hours at 37 °C. Adenosine diphosphate (ADP; Sigma Aldrich), a calcium stimulator, was used as a positive control for platelet activation. ADP was reconstituted in PBS and added to a final concentration of 40 μΜ. Teflon was chosen as a negative control because of its ability to repel protein and cell binding. Glass cover slips were used as a positive control. The PRP was incubated on test surfaces for 2 hours at 37 °C. The surfaces were removed from PRP, rinsed gently by dipping in PBS, and then fixed with 4% paraformaldehyde for 15 minutes followed with 1% BSA blocking for 30 minutes. The surfaces were again washed with phosphate buffered solution (PBS) to remove unabsorbed BSA molecules. Test surfaces were incubated with antihuman CD62P (P-selectin) mouse mono-clonal antibody for 1 hour followed by Alexa Fluor 546 donkey anti-mouse secondary antibody for another hour. Lastly, the surfaces were incubated with antihuman CD41 FITC-conjugated mouse monoclonal antibody diluted 300x for 1 hour. The surfaces were washed after every labeling step with PBS for avoiding any non-specific binding. Finally, the surfaces were imaged under an inverted fluorescent microscope to assess any platelet adhesion (green channel) and/or platelet activation (red channel). All antibodies were purchased from Life Technologies (Grand Island, NY). As seen in Figure 12B, PEG treatment significantly reduced both platelet adhesion and activation compared to uncoated surfaces and glass.

[0075] Biochemical studies have also proven that PEG treated pnc-Si membrane has excellent hemocompatibility over the standard materials currently used in

commercial membranes. Platelet adhesion and activity was overwhelmingly reduced along with a significant decrease in complement and thrombin activity as well (see Figure 5).

Discussion of Examples 1-6

[0076] The development of silicon-based membranes for the bioartificial kidney

(BAK) has been a focus of the pioneering efforts of Roy, Fissell, and colleagues (Fissell et al., "High-Performance Silicon Nanopore Hemo filtration Membranes," J. Membrane Science 326(l):58-63 (2009); Muthusubramaniam et al., "Hemocompatibility of Silicon- Based Substrates for Biomedical Implant Applications," Ann. Biomed. Eng. 39(4): 1296- 305 (2011); Fissell et al., "Bioartificial Kidney Alters Cytokine Response and

Hemodynamics in Endotoxin-Challenged Uremic Animals," Blood Purif. 20:55-60 (2002); Kanani et al., "Permeability - Selectivity Analysis for Ultrafiltration: Effect of Pore Geometry," J. Memb. Sci. 349:405 (2010), which are hereby incorporated by reference in their entirety). Although many system components must come together to make the BAK a reality (Humes et al., "The Bioartificial Kidney: Current Status and Future Promise," Pediatr. Nephrol. Epub (2013), which is hereby incorporated by reference in its entirety), a highly efficient membrane platform is fundamental. Assuming membrane pore sizes are fixed to give molecular discrimination that mimics the healthy kidney, only the porosity and the thickness of a membrane can be adjusted to improve membrane permeability. Thus, as molecularly thin nanoporous membranes with porosities that can approach -20% (Kavalenka et al., "Ballistic and Non-Ballistic Gas Flow Through Ultrathin Nanopores," Nanotechnology 23: 145706 (2012), which is hereby incorporated by reference in its entirety), the silicon nanomembranes being developed operate close to the maximum permeability that can be achieved for a passive hemodialysis membrane. This report demonstrates the ability to: 1) manufacture nanomembranes with appropriate separation characteristics; 2) the capacity to modify these membranes to improve hemocompatibility; and 3) the ability to integrate into dialysis devices that achieve predicted urea dialysis rates after a day of use. Importantly, the predicted dialysis rates assume the membrane offers no resistance to the diffusion of urea, confirming the expectation of ultrahigh efficiency.

[0077] Typical ESRD patients experience the equivalent renal urea clearance of

~30 mL/min for a 4 hour period three times in a week at dialysis centers. The goal is a device that achieves this level of total clearance in a small format and at reasonable cost. Simulations indicate that the optimal clearance through the current 13 channel chip format is ~0.1 mL/min (at total chip flow rate of -0.45 mL/min). Assuming daily dialysis for 10 hours, a device containing 50 chips would be needed to achieve the target weekly clearance. These can be packaged into a device the size of a modern smart phone. In addition, the preceding Examples demonstrate the availability of useful surface chemistries that minimize activating plasma or immune systems, and inhibit biofouling of membrane surfaces.

[0078] Having thus described the basic concept of the invention, it will be rather apparent to those skilled in the art that the foregoing detailed disclosure is intended to be presented by way of example only, and is not limiting. Various alterations,

improvements, and modifications will occur and are intended to those skilled in the art, though not expressly stated herein. All of the features described herein (including any accompanying claims, abstract and drawings), and/or all of the steps of any method or process so disclosed, may be combined with any of the above aspects in any

combination, except combinations where at least some of such features and/or steps are mutually exclusive. Additionally, the recited order of processing elements or sequences, or the use of numbers, letters, or other designations therefore, is not intended to limit the claimed processes to any order except as may be specified in the claims. These alterations, improvements, and modifications are intended to be suggested hereby, and are within the spirit and scope of the invention. Accordingly, the invention is limited only by the following claims and equivalents thereto.

Claims

WHAT IS CLAIMED:

1. A micro fluidic device comprising:

a sealed body comprising an inlet port and an outlet port;

a plurality of channels formed in said sealed body, each of said channels being in fluid communication between the inlet port and the outlet port, said channels being defined by a support material and a nanoporous or microporous membrane connected to the support material; and

a chamber inside the sealed body separated from the fluid communication of the inlet port, the plurality of channels, and the outlet port via the nanoporous or

microporous membranes of said channels.

2. The micro fluidic device according to claim 1, wherein the device comprises a Peclet number of about 1.

3. The micro fluidic device according to claim 1 or 2, wherein the

nanoporous membrane is less than about 100 nm thick.

4. The micro fluidic device according to any one of claims 1 to 3, wherein the nanoporous membrane is present and comprises a porosity of at least about 1 percent, a pore size cutoff of less than 100 nm, or a combination thereof.

5. The micro fluidic device according to any one of claims 4, wherein the nanoporous membrane comprises pnc-Si.

6. The microfluidic device according to claim 1 or 2, wherein the

nanoporous or microporous membrane further comprises polyethylene glycol groups bound to the surface of the membrane.

7. The microfluidic device according to any one of claims 1 to 3, wherein the microporous membrane comprises pores in range of 100 nm to 500 nm.

8. The microfluidic device according to claim 7, wherein the microporous membrane comprises SiN.

9. The micro fluidic device according to any one of claims 1 to 8, wherein the nanoporous or microporous membrane further comprises a patterned reinforcement applied to one side of the membrane.

10. The microfluidic device according to claim 9, wherein the patterned reinforcement comprises silicon nitride.

11. The microfluidic device according to claim 1 further comprising a retention agent inside the chamber adjacent to the nanoporous or microporous membrane.

12. The microfluidic device according to claim 11, wherein the retention agent comprises activated carbon adsorbent, carboxen 563 adsorbent, oxidized activated carbon adsorbent, chitosan/Cu(II) adsorbent, or charcoal.

13. The microfluidic device according to one of claims 1 to 12 further comprising a second inlet port and a second outlet port formed in said sealed body, the second inlet and outlet ports being in fluid communication with the chamber.

14. The microfluidic device according to claim 13, wherein the direction of flow from the second inlet port to the second outlet portion is opposite of the direction of flow from the first inlet port to the first outlet port.

15. The microfluidic device according to one of claims 1 to 14 further comprising a pumping device in fluid communication with one of the inlet or outlet ports.

16. The microfluidic device according to claim 15 further comprising a first pumping device in fluid communication with the inlet port or the outlet port; and

a second pumping device in fluid communication with the second inlet port or the second outlet port.

17. A method of filtering a fluid sample comprising:

providing a micro fluidic device according to one of claims 1 to 14; and passing through the microfluidic device, from the inlet port to the outlet port, a fluid comprising one or more agents smaller than the dimension of nanopores or micropores formed in the nanoporous or microporous membrane,

whereby the one or more agents are removed from the fluid sample and pass through the nanopores or micropores, and filtered fluid is recovered from the outlet port.

18. The method according to claim 17, wherein the fluid sample is whole blood or plasma.

19. The method according to claim 17, wherein the fluid sample is a buffer solution comprising the one or more reagents.

20. The method according to claim 17, wherein the fluid sample is an industrial water source and the one or more agents are industrial water contaminants.

21. A method of performing dialysis comprising:

providing a microfluidic device according to one of claims 1 to 16; and coupling the inlet port and the outlet port to the blood supply of a patient, and forcing blood to flow through the microfluidic device and over the nanoporous or microporous membranes,

whereby biological contaminants are removed from the blood and pass through the nanopores or micropores, and filtered blood is delivered to the outlet port.

22. The method according to claim 21, wherein the dialysis is performed continuously for a duration of more than 24 hours.

23. The method according to claim 21 , wherein the biological contaminants comprise urea and creatinine.

24. The method according to claim 21, wherein the microfluidic device is a microfluidic device according to claim 11 or 12, the method further comprising periodically replacing the retention agent inside the chamber.