Using hydration water to enhance the selectivity of membrane seperations
Professor Alberto Striolo, University College London
We report atomistic molecular simulation results for the transport of various gases, including methane, H2S, and ethane, through narrow pores saturated by liquid water. The slit-shaped pores are of width of approximately 1 nm. This pore width is selected because the pore surface strongly affects the structure of confined water throughout the entire pore volume. Different solid materials are chosen for the pores, because they are known to affect differently the structure of interfacial water. While some materials are realistic, e.g., the cristobalite mineral of silicon dioxide, others are model ones used as proof of concept in our calculations. Equilibrium molecular dynamics simulations are used to assess the adsorption of various gases within the confined water as a function of bulk pressure. The results show that when confined water is characterized by relatively high fluctuations in density, the solubility of hydrophobic gases can be significantly enhanced compared to the solubility of the same gases in bulk water at comparable thermodynamic conditions. We analysed the free energy barriers experienced by the various gases to enter the pores saturated with water by conducting umbrella-sampling simulations, and we find that the density of water near the pore entrance determines such barriers. We then analysed the transport of the various gases within the pores using a number of statistical techniques, and we compared the selective transport of various gases through the pores filled with water. The results suggest that water could be used to enhance selectivity in membrane separations, although the permeability is reduced significantly compared to values achievable for the pores without water present.
Assessing the permeability, selectivity and applications of carbon nanotube membranes using molecular simulation
Dr Ben Corry, The Australian National University, Australia
Materials containing carbon nanotubes (CNTs) have been suggested to have enormous potential in membrane filtration technology due to the observation of rapid water, liquid and gas transport. However, reports of the transport and rejection capabilities of these membranes differ widely amongst exiting experimental and simulation studies. In order to assess the potential uses of these materials, it is essential to have an accurate estimation of their permeability and selectivity as well as of the physical origins of these properties. In this talk I aim to outline what we have learnt about the potential permeability and rejection capabilities of these membranes using molecular simulations and to highlight questions that need further study.
We have been improving the accuracy of molecular dynamics simulations to gain the best estimation of the influence of pore size, chemical functionality pore density, pressure and salt concentration on water flux and salt rejections. Water flows through these pores at much greater rates than in polymeric membranes, due to the near frictionless walls of the CNTs which by their physical smoothness and lack of localised electrostatic interactions. Salt rejection arises in narrow CNTs as a consequence of the energetic costs of ions having to dehydrate in order to fit inside the pores. An upshot of this is that salt rejection is strongly related to the CNT diameter as well as being highly dependent on pressure. Appropriate chemical modification of the CNTs can improve salt rejection by simple charge repulsion or attractive binding interactions. However any chemical modification is likely to reduce the water transport rate by increasing the friction of the walls.
From this data we are able to assess the design principles and potential energy savings in using CNT membranes for water filtration applications. Given the enormous water permeability of CNTs, there are limited gains to be achieved from increasing the pore density in the membrane, or in reducing the friction of functionalised CNTs. The use of highly permeable membranes has little impact on the energy required for reverse osmosis desalination, however, much less membrane area is required and so the door is open for much smaller plants or portable desalination devices. The uniformity of the pores in these membranes may also be exploited to aid in the removal of toxic contaminants that are difficult to eliminate with polymeric membranes.
Voltage activated carbon nanotube membranes as biomimetric platforms
Professor Bruce Hinds, University of Washington, USA
A grand challenge for the membrane community is to mimic the dynamic activity of natural protein channels giving significant advantages over passive man-made systems based on pore size and coarse chemical selectivity. Instead of using chemical reactions, such as ATP cycle, a more facile method to activate pore entrances is to use applied bias to actuate charged gatekeepers and induce ionic pumping. Carbon nanotubes have three key attributes that make them of great interest for novel membrane applications 1) atomically flat graphite surface allows for ideal fluid slip boundary conditions and extremely fast flow rates 2) the cutting process to open CNTs inherently places functional chemistry at CNT core entrance for chemical selectivity and 3) CNT are electrically conductive allowing for electrochemical reactions and application of electric fields gradients at CNT tips. The CNT membrane, with tips functionalized with charged molecules, is a nearly ideal platform to induce electro-osmotic flow with high charge density at pore entrance and a nearly frictionless surface for the propagation of plug flow. Through diazonium electrochemical modification we have successfully bound anionic surface charge to CNT tips and along CNT cores. High electro-osmotic flows of 0.16 cm/s-V at are seen by the pumping of neutral caffeine molecules. Improvements in electroosmotic power efficiency of 25-112 fold are seen in CNTs compared to conventional nanoporous materials with atomically rough interfaces . Use of the electro-osmotic phenomenon for responsive/programmed transdermal drug delivery devices is discussed with the voltage gated delivery of clonidine and nicotine across CNT membrane at therapeutically useful fluxes . In small diameter SWCNTs ion mobilities are seen to be ~6 fold enhanced due to induced electroosmotic flow. Electroosmotic flow enhancements of 10,000 fold are seen  and are consistent with pressure driven flow enhancements. An efficient 2-cycle voltage voltage pulse system to selectively pump his-tagged proteins across membrane/electrode stack are also demonstrated 
1 "Nanoscale hydrodynamics: Enhanced flow in carbon nanotubes" Majumder, M.; Chopra, N.; Andrews, R; Hinds, B.J * Nature 2005, 438, 44.
2 ‘Mass Transport through Carbon Nanotube Membranes in three different regimes: ionic diffusion, gas, and liquid flow’ Mainak Majumder, Nitin Chopra, B.J. Hinds* ACS Nano 2011 5(5) 3867-3877
3 ‘Highly Efficient Electro-osmotic Flow through Functionalized Carbon Nanotubes Membrane’ Ji Wu, Karen Gerstandt, Mainak Majunder, B.J. Hinds*, RCS Nanoscale 2011 3(8) 3321-28
4 " Programmable transdermal drug delivery of nicotine using carbon nanotube membranes" J. Wu, K.S. Paudel, C.L. Strasinger, D. Hamell, Audra L. Stinchcomb*, B. J. Hinds* Proc. Nat. Acad. Sci. 2010 107(26) 11698-11702.
5 "Electrophoretically Induced Aqueous Flow through sub-Nanometer Single Walled Carbon Nanotube Membranes" Ji Wu, Karen Gerstandt, Hongbo Zhang, Jie Liu, and Bruce. J. Hinds Nature Nano 2012 DOI: 10.1038/NNANO.2011.240
6 "Dynamic Electrochemical Membranes for Continuous affinity protein separation", Z. Chen, T. Chen, X. Sun and B.J. Hinds. Advanced Functional Materials, in press DOI: 10.1002/adfm.201303707
Multiscale modelling of nano-confined fluid flows
Dr Duncan Lockerby, University of Warwick, UK
Nano-confined fluid flows, such as those in nano-structured membranes, are characterised by non-continuum effects that place them beyond the modelling scope of conventional Computational Fluid Dynamics (CFD). Typically a molecular or particle treatment of the liquid or gas, and any bounding solid surface, is required to accurately resolve such flows. However, the cost of these particle-based simulations is prohibitive for all but the simplest geometries. In this talk, the opportunity in combining the efficiency of conventional fluid modelling with the accuracy of a particle simulation, in a 'hybrid’ code, is explored. Key to forming an effective hybrid is the identification of scale separation where it exists (both in time and space) for the particular flow configuration considered. Examples of this from a range of internal-flow configurations will be considered, including the investigation of water transport through aligned carbon nanotubes.
Transport in carbon nanotube nanochannels under different driving forces
Dr Francesco Fornasiero, Lawrence Livermore National Laboratories, USA
The recently-reported exceptionally-fast fluid transport rates in carbon nanotubes (CNT) spurred great interest for their application as nanofluidic channels in several areas ranging from membranes for water purification and carbon capture, to drug delivery and protein separation. However, in spite of considerable recent advances, the origin and precise magnitude of the large transport rates through CNTs under different driving forces remain unclear.
Here, we highlight experimental work performed in our laboratory directed toward: a) a precise CNT-membrane flux quantification by a through characterization of the CNT pore structure with synchrotron x-ray techniques; b) a fundamental understanding of the selectivity of these pores for electrolyte solutions under a pressure driving force; c) elucidating electric-field driven ion transport in a single CNT nanopore; d) the exploitation of CNT fast flow for the realization of ultrabreathable and protective CNT fabrics. For our studies, we used ceramic or polymeric membranes with well-aligned, a-few-nm wide CNTs as only through-pores. With a combination of x-ray scattering and attenuation measurements, we demonstrate the quantification of CNT pore diameter, density, and tortuosity needed for a precise CNT flux determination. We provide evidences of a pH-tunable ion selectivity in narrow CNT pores that is dominated by electrostatic interactions between carboxylic groups at the CNT tips and the ions in solution. With single-pore ionic conductance measurements, we reveal giant ionic currents in CNT pores that follow an unusual power-law concentration dependence. Finally, we show that CNT membranes provide diffusion-driven water-vapor transport rates that are comparable to or exceeding state-of-art breathable fabrics at all relative humidities.
This work performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344.
Graphene oxide membranes for gas seperation
Professor Ho Bum Park, Hanyang University, South Korea
Graphene and graphene oxide (GO), as two-dimensional nanosheets, have a great potential in many applications, including energy storage, sensors, barrier, fuel cell, and flexible electronics. More recently, these materials have been regarded molecular filters (membranes) to separate ions as well as gases. Graphene is intrinsically impermeable to gas molecules, but if subnanometric pores can be technologically created with high porosity, the porous, atom-thick-graphene will be ever an ideal membrane in terms of selectivity and permeability. However, such graphene membrane with subnanometric pore size (based on size-exclusion) and high porosity (for high flux) will be a great challenge in the future. Alternatively, GO sheets have been also considered as promising membrane materials for gas and liquid separation. GO can be easily obtained from chemical exfoliation of graphite, and mass production is also available. Since GO has hydrophilic nature, GO is highly dispersible in aqueous media so that it can be few-layer-coated by using existing coating methods, including spin-casting, spray, vacuum filtration and direct evaporation. A key idea to use two-dimensional GO nanosheets as membranes is to generate slit-like pores or channels by tailoring interplanar spacing between GO sheets, as it is not the case for typical porous structures in the conventional membranes. The interplanar spacing of stacked GO membranes ranges from 0.6 to 1.0 nm, depending on stacking manner and the amount of intercalated water molecules. That is, small gases or water molecules can travel through the spacing between GO sheets. However, since GO sheets have high aspect ratio and layered structures, the GO size and the stacked GO thickness should be carefully considered to reduce high tortuosity, so leading to high permeability. More recently our group has demonstrated the potential of thin GO membranes for selective gas transport by changing stacking manner, and pore evolution by thermal treatment (Science, 342, 95-98 (2013)). As a result, the GO membranes have great ability to separate gas molecules very effectively. The gas transport behavior through GO membranes can be also tuned; H2-selective (H2/CO2) or CO2-selective (CO2/N2 or CO2/CH4) membranes can be prepared by different stacking methods. Since many industrial gas streams, such as post-combustion, natural gas purification, and syngas adjustment, contain water vapor, the effect of water vapor on membrane separation performance is also crucial. Usually, water vapor tends to deteriorate the membrane performance; the reduction of both permeability and selectivity by water condensation on the membrane surfaces or pores, so an energy-costly water vapor removal before membrane unit is necessary. In this regard, we also evaluated the effect of water vapor on the gas transport and separation performance of GO membranes in this study for practical applications.
Convergence of molecular dynamics and experiments in the engineering design of aligned nanotube membranes
Professor Jason Reese, FREng FRSE, University of Edinburgh, UK
Dr Davide Mattia, University of Bath, UK
Rapid progress in computational power and in code optimization is bringing a convergence of molecular dynamics simulations and experiments in the study of fluids dynamics under nanoscale confinement. In this presentation we discuss several examples where the two investigation methods have been applied at the same time, to the same materials, with the results from molecular dynamics guiding experiments, and experiments validating simulations.
We describe the effects that nanotube surface chemistry and structure, including defects and geometrical characteristics (diameter and length), have on flow enhancement and permeability of nanotubes of different materials, including carbon, silicon carbide and boron nitride. We show that entrance and exit effects are significant and lead to large losses in these regions - although the application of a localised electric field can create a liquid-crystal-like ordering of water molecules entering carbon nanotubes and thereby control the water flow rate by up to a factor of three.
We extend the results from individual nanotubes to nanotube membranes, and present a comparison with other results in the literature as well as with commercial polymeric and ceramic membranes. Finally, we discuss a model that makes explicit how certain nanotube properties affect flow enhancement and permeability, with validation from both experiments and simulations.
Nanofluidic transport and noise across individual carbon and boron-nitride nanotubes: experiments and theory
Professor Lyderic Bocquet, Ecole Normale Superieure, France
Fluid transport at the nanoscales is one of the remaining virgin territory in fluid dynamics, in spite of hydrodynamics being a very old and established domain. Over the last years, a number of striking phenomena have been unveiled, such as superfast flows in carbon nanotubes, hydrodynamic slippage, fluidic diodes, nanobubble superstability, … and many of them are still awaiting an explanation . A major challenge to adress the fundamental properties at the nanoscales lies in building distinct and well-controlled nanosystems, amenable to the systematic exploration of their properties. To this end, we have developed new methods based on the manipulation of nano-objects, displacing, cutting, and glueing these elementary building blocks. This allows us to fabricate original fluidic and mechanical systems involving single nanotubes, made of carbon as well as boron-nitride materials.
In this talk, I will discuss various experiments on the fluidic transport inside single nanotubes.
First, putting osmotic transport and its fundamental origins into perspective, I will show how to harvest this powerful mechanism beyond the classical van’t Hoff law. Experiments of osmotic transport across boron-nitride show unprecedented energy conversion from salt concentration gradients. This points to new avenues in the field of osmotic energy harvesting from salinity gradient . These results will be then compared to those obtained with carbon nanotubes. They point to some key differences between these two materials which exhibit the same crystallography, but very different electronic properties.
I will finally extend this systematic comparison to the properties of the electrical noise across BN and C nanotubes. Experimental data exhibit low frequency 1/f noise, with an amplitude which depends on the applied voltage. However, while the spectra recorded in CNT follows the empirical Hooge’s relation, this is not the case of the insulating BNNT. I will discuss potential theoretical mechanisms at the origin for such behaviors.
Graphene-based fluidic systems: From compact micro/nano-fluidic devices to large area filtration membranes
Dr Mainak Majumder, Monash University, Australia
Graphene-based multilayer thin films are exciting new materials for fluidic systems because these films form ensemble nano-capillaries between each individual graphene sheet of ~ 1nm regardless of the size of the graphene microplates and the size of the continuous films. Permeation is determined to a large extent by the ~ 1 nm spacing, although transport may occur through defects in graphene sheets. In addition these materials can be fabricated from fluid-phase dispersed graphene oxide which makes these systems suitable for industrial-level manufacturing. Our research efforts are directed towards two front: firstly, we develop methodologies to integrate these nano-capillaries into micro-/nano-fluidic devices and secondly, construct large area membranes. The micro-/nano-fluidic devices have provided us a platform to study fluid transport phenomena at the boundary of micro-/nano-scales. For e.g. we have reported the finding of a novel form of ion rectification in graphene oxide (GO) and reduced graphene oxide (RGO) films. Rectification is imparted by introducing geometric asymmetry in fluidic inlets to the counter-ion selective nanochannels of GO/RGO which creates asymmetry in the enrichment/depletion effects at the macro-/nano-interface. The devices are made simply by cutting a GO or RGO film into a trapezoid and sealing the film within a Polydimethylsiloxane (PDMS) block so that fluid may only enter through one of two inlets. These devices exhibit rectification ratios larger than 20 (in 1 mM NaCl) while operating at modest voltages [-1V,+1V]. We have also developed a mass-producible method for fabricating supported Graphene-based membranes. Rheological-tailored Graphene Oxide (GO) dispersions is cast on microfiltration Nylon (pore size ~ 0.2 µm) support membrane by knife coating – a very simple and rapid method to produce ~10 cm×15 cm membranes in less than a minute. Scanning electron and polarized light microscopy reveal well-packed and ordered structure. The microstructure can be tuned by shear rate and rheological characteristics of the coating fluid. Filtration studies have shown large water flux (25 litres/m2-hour-bar) and 100% rejection of Direct Yellow 50 (957 g/mol), > 99% for Methylene blue (320 g/mol), > 92% for Rhodamine B (479), ~ 86% for Methyl Viologen (257 g/ mol) with a molecular weight cut-off of ~ 320 indicating the suitability as a novel nanofiltration membrane. Rejection mechanisms are reliant on physical sieving and electrostatic interactions with minimal adsorption. This new process of ideal asymmetric membranes on macroporous supports is scalable and has the potential to be translated to large areas easily.
Understanding molecular transport through single, isolated nanocarbon pores: single walled carbon nanotubes and graphene
Professor Michael Strano, Massachusetts Institute of Technology, USA
Our lab at MIT has been interested in how nanopores constructed from 1D and 2D nanomaterials can solve longstanding challenges in molecular separations. Nanostructured carbon represents a relatively new type of material for potential high-selectivity and high-permeance membrane applications. In our group, we have carried out fundamental studies on transport through both graphene and carbon nanotube nanopores. At a thickness of one atomic layer, single layer graphene membranes (SLG) represent the ultimate limit of membrane materials. Pristine SLG has been experimentally shown to be impermeable to even the smallest gas molecules, providing a platform to create porated SLG that can achieve both high permeance and selectivity. While membrane materials are typically defined by their thickness and bulk chemistry, the characteristics of SLG membranes are defined by the interactions of the permeate molecule at a single point, the pore mouth. Therefore, to evaluate SLG membranes, new analyses are needed to quantify the interaction energy and new models are needed to analytically describe the transport based on that energy. Our work has focused on developing such analytical models and applying them to the emerging experimental results1. Alternatively, unlike porated SLG, carbon nanotubes are high-aspect ratio, straight-line pores of a well-defined diameter. Along with the selectivity derived from the size-based exclusion and pore-mouth chemistry, simulations have also predicted enhanced water and proton transport. At the scale of 1-2 nm, the diameter of the nanopore has a significant and non-monotonic effect on transport properties. In our work2, we have experimentally measured voltage-driven proton transport rates through individual, isolated carbon nanotube nanopores of a well-defined diameter (0.94 - 2.01 nm). The results reveal a local maximum in proton pore-blocking current around 1.6 nm which can be explained by the predicted transition point to subcontinuum water flow.
1. Drahushuk, L. W. & Strano, M. S. Mechanisms of Gas Permeation through Single Layer Graphene Membranes. Langmuir 28, 16671-16678, doi:10.1021/la303468r (2012).
2. Choi, W. et al. Diameter-dependent ion transport through the interior of isolated single-walled carbon nanotubes. Nat Commun 4, doi:10.1038/ncomms3397 (2013).
The influence of carbon nanotubes and graphene on the transport properties of the high free volume polymer PIM-1
Professor Peter Budd, University of Manchester, UK
Carbon nanomaterials, such as carbon nanotubes, graphene, and graphene derivatives, may be utilised in membrane technology in various ways. One approach is the formation of mixed matrix membranes, in which the carbon nanomaterial serves to enhance the performance of a polymeric phase. We shall discuss the effects of carbon nanotubes and graphene on the transport properties of a polymer of intrinsic microporosity (PIM).
PIMs are glassy polymers which possess high free volume and high internal surface area as a consequence of their relatively inflexible, contorted macromolecular backbones. Typically, they comprise fused ring sequences interrupted by spiro-centres. The archetypal solution-processable PIM, referred to as PIM-1, forms membranes that are of interest for a variety of separation processes, including gas separations (e.g., CO2 recovery) and organophilic liquid separations (e.g., bioalcohol recovery). In gas separation, PIM-1 was shown to surpass the 1991 upper bound of performance for several gas pairs, and contributed to the revision of the upper bound by Robeson in 2008.
Water flows in carbon nanotube membranes: Simulations and uncertainties
Professor Petros Koumoutsakos, ETH Switzerland
In the last decade experiments and simulations have reported flows in Carbon Nanotube (CNT) membranes that span over 5 orders of magnitude. While earlier reported ultra-fast flow rates are now revised downwards in various experimental studies and discrepancies remain among Molecular Dynamics (MD) and multiscale flow simulations, the fundamental question of what drives water flow in CNTs and how transport rates depend on CNT and membrane parameters remain rather open issues. In this talk I will discuss possible reasons for discrepancies between experiments and simulations and present a Bayesian framework for quantifying the uncertainties in MD simulations of flows in CNTs.
Effects of dielectric and viscosity profiles on electro-osmotic flow and conductivity in nanotubes
Professor Roland Netz, Freie Universitaet, Germany
The electro-osmotic mobility and conductivity in a nanotube follows from a modified Poisson-Boltzmann equation that includes spatial variations of the dielectric function and the viscosity that are extracted from molecular dynamics simulations of aqueous interfaces. The low-dielectric region directly at the interface leads to a substantially reduced surface capacitance. At the same time, ions accumulate into a highly condensed interfacial layer, leading to saturation of the electro-osmotic mobility at large surface charge density regardless of the hydrodynamic boundary conditions.
Nanotubular and nanoporous membranes: Processing strategies and applications
Professor Sankar Nair, Georgia Tech, USA
This talk will describe our recent progress in the synthesis and properties of nanotubular and nanoporous materials for molecular separation applications, and their subsequent use in the fabrication of ‘molecular sieving’ membranes. New separations technology is key to increased energy efficiency in the chemical sector, efficient production of clean and renewable fuels, and water treatment. It is hypothesized that the above kinds of materials may find successful applications in this area. Towards this goal, we will overfiew three areas of investigation. We will first discuss the synthesis of single-walled metal oxide nanotubes and their incorporation into polymeric matrices to obtain membranes for use in organic/water separation processes. We will then turn our attention to nanoporous metal-organic framework (MOF) materials, particularly the fine control of their properties via “mixed-linker” synthesis and their processing into gas and hydrocarbon separation membranes supported on polymeric hollow fibers. Finally, we will address the fabrication and use of carbon molecular sieve (CMS) membranes for applications involving extreme conditions such as the treatment of black liquor in the Kraft process.
4" CNT membranes: Fabrication and Characterization
Dr Valentin Lulevich, Porifera Inc, USA
Carbon nanotube (CNT) membranes have unique nanometer size smooth hydrophobic CNT pores that are also electrically conductive. Well-established carbon chemistry allows for relatively simple CNT pore functionalization, enabling engineering of selective membranes. We have expanded a significant effort in designing process for making carbon nanotube membranes, which can be used for chemical separations, selective electroosmotic pumping, and drug delivery. In of our work we demonstrate reproducible fabrication of CNT membranes on 4" diameter wafer substrates. Gaps between CNTs in aligned array grown on silicon wafer are blocked with polymer (epoxy resin). CNT ends are exposed by using CMP (Chemical Mechanical Polishing) on one side, and by simple wafer release in HF on other side. We show that even the subtle differences in membrane manufacture result with considerable differences in membrane transport properties. We will discuss effects of following parameters on membrane transport properties: CNT quality (single v.s. triple wall CNTs, different G/D ratio), the way tube ends were treated during CMP process, degree of tube damage during polymer infiltration etc. We will also discuss the challenges of proper characterization of these membranes and the ways a researcher may be fooled to misinterpret the transport properties of this material.