Nanoscale Bubble Valves on MWCNT Membranes for Chemical Energy Storage

Abstract

A new class of valves for membranes is based on the formation of nanobubbles at the pore entrances of carbon nanotube (CNT) membranes. Nanobubble stabilization is achieved by electrochemically etching CNTs into a polymer matrix to form a well that can be reversibly filled. Such valves have applications in flow battery systems where high-energy chemicals can be stored indefinitely. Shifting to renewable energy resources such as solar and wind energy will require innovative approaches in energy storage.1 Chemical energy storage is appealing due to a very large energy density and scalable capacity by simple increase in storage volume.2 However this requires stable electrodes with low overpotential, fast mass transport, and a valve system to stop diffusion while in the storage mode. Carbon nanotube (CNT) membranes3 are an intriguing platform for energy storage since the mass transport through CNT cores is 1000-fold faster than pores of conventional materials, graphite is highly conductive yet corrosion resistant, and CNT surfaces can be functionalized with catalyst metals or complexes for a low electrochemical overpotential.4 However, an effective valve to turn off the membrane during storage time needs to be developed. Molecular gate keepers have been developed for CNT membrane but cannot completely turn off CNT membrane flux due to incomplete coverage.5 Proposed here is that electrochemically generated nanobubbles can be used to produce a barrier with complete coverage of pore entrance and act as a switchable valve. Microscale bubbles have been used as valves for micropumps in microchannel devices6 and attributed to ionic current noise in nanopores,7 but have not been applied as valves on nanometer-scale pores of membranes. The Laplace equation predicts that nanoscale bubbles are unstable due to the high pressure inside them; with 140 atm for bubbles 20 nm in diameter.8 However, the nanoscale bubbles have been observed with unexpected stability on hydrophobic surface due to small contact angle.9 It was found that the average pressure in the flat nanoscale bubbles (large radius of curvature) is only about 1 atm,[9] which allows the long-term stability of nanoscale bubbles. Though nanobubbles can be transient in nature, they can be pinned by contact line,10 thus can be stabilized by a supporting geometry such as a “well.” These bubbles can form inside of the nanochannels,11 on CNTs,12 be formed by electrochemically generated hydrogen,13 and control nanofluidics.14 Demonstrated here is that bubbles generated at the entrances of conductive3, 5 Multiwalled CNT (MWCNT) pores by electrochemical will act as an effective valve for flow batter energy storage. Ninety-two percent blocking efficiency is achieved when the nanoscale bubbles are stabilized in the 30–60 nm diameter “wells” at the tips of MWCNTs. These wells are formed by the electrochemical oxidation of the conductive MWCNTs partially into the polymer matrix of the MWCNT membrane.3 The nanoscale bubbles can be removed by applying 0.004 atm pressure to recover the flow through the MWCNT membrane. Our initial efforts to block as-made MWCNT membranes with nanobubbles at CNT tips were not successful, due to the predicted instability of small diameter bubbles. In order to stabilize nanobubbles, nanometer-scale wells were formed by electrochemical oxidation of CNTs following the schematic of Figure 1a. This was accomplished by 5 h of electrochemical oxidation at a potential of 2.5 V versus Ag/AgCl, and in a solution of 0.1 m KCl. After the electrochemical oxidization, the MWCNT membrane became visibly more transparent due to the loss of highly light absorbing MWCNTs. The measured UV–vis absorbance of light through the MWCNT membrane corresponds to a 27% reduction in the CNT volume/length in the 5 μm thick membrane. Scanning electron microscopy (SEM) images of the top view of MWCNT membrane before and after electrochemical oxidization (Figure 1b,c) confirmed that the tips of MWCNTs were burned away leaving larger polymer wells. Under a constant potential of −1.2 V versus Ag/AgCl at the MWCNT membrane working electrode, only large macrobubbles are generated and a readily visible to the eye (Figure S4, Supporting Information). The transport of [Ru(bipy)3]2+ across the membrane was not blocked at pore entrances under this constant bias until a macrobubble covered the entire membrane surface (Figure S5, Supporting Information). Presumably H2 bubbles are generated at a rate too fast to remain nanometer-scale bubbles and coalesce to form larger bubbles. To explore the best potential for generating the nanoscale bubbles on the MWCNT membrane with polymer wells, cyclic voltammograms (CVs) measurements were performed with the MWCNT membrane as the working electrode in the same setup as the previously mentioned blocking experiment. The potentials range of CVs is from −0.8 to +0.8 V versus Ag/AgCl (Figure S6, Supporting Information). There is hydronium reduction (H2 production) at low rates in the voltage range from −0.8 to −0.3 V, which is consistent with the Ag/AgCl potential of 0.28 V versus standard hydrogen electrodrode (SHE). At larger negative potentials above the overpotential high current levels are reached resulting in coalescence of macroscopic bubbles. In order to prevent the generation of macrobubbles seen under applied constant potential, the sine potential pulse was applied to minimize time at high H2 generation rate and reduce charge build-up. Table 1 shows that nanoscale bubbles generated by 200 Hz sine potentials on MWCNT membrane without polymer wells have modest (21%) blocking of the ionic transport through the membrane. Blocking efficiency is defined as (J0 −Jb)/J0, where J0 is initial Ru(bipyr)32+ (5 mmol feed) diffusive flux across membrane and Jb is flux after bubble generation. However, with polymer wells generated from electrochemical oxidation of CNTs, 92% of ionic current is blocked by nanobubbles generated using the sine potential. The polymer wells stabilize the nanoscale bubbles by reducing the bubbles radius of curvature. Since the sine function has both positive and negative bias on the CNT membrane working electrode, it is important to study the polarity of the bubble formation (O2 or H2). For a 0.4 V sine amplitude with a baseline of +0.4 V (sine potential range 0 to +0.8 V), the blocking efficiency is seen to be 0%. However, when the baseline of sine potentials is at 0 or −0.4 V, the blocking efficiencies under both conditions are similar and above 80%. Thus only negative part of the sine potential can generate bubbles indicating that the reductive H2 generation is responsible for membrane gating. Gas bubbles are inherently hydrophobic and would be stabilized by a hydrophobic polymer well at the pore entrance. The oxidized CNT membrane polymer well was converted into a highly hydrophobic surface by the reaction of carboxylate groups with butyltrichlorosilane. The hydrophobic well gave 100% blockage with gas absorbed from air. With a second electrochemical oxidation, to render the well hydrophilic, the flux rate through the MWCNT membrane was recovered. This is consistent with the CNT wells stabilizing nanometer-scale bubbles and an ability to tune bubble stability with the degree of hydrophobicity. However with too hydrophobic a well, the membrane was difficult to switch from blocked state. In analogous surface modifications of quartz surfaces with methylated siloxanes, a change in surface contact angle of 40° is seen with ≈40% coverage.15 However it should be stressed that we are unable to directly measure surface functionalization coverage or contact angle in the nanometer-scale wells, but are able to show the operational limits of no-bubble support/blockage to indefinite blockage. To make a useful valve, it is critical to demonstrate that small pressures can open the blocked CNT membranes. A pressure difference of 0.004 atm was applied on the feed side (4 cm water column height) of the U-tube cell (Figure S1b, Supporting Information and Table 2 showed 84% recovery of initial flux. It should be noted that there are two major variables that are the object of current study: (1) the degree of hydrophobicity of polymer well and (2) the rate of H2 generation. A highly hydrophobic well stabilizes nanobubbles and cannot reopen channels while hydrophilic channels require very little pressure to regenerate channels. The degree of stability and required switching pressure can be tuned between these extremes. Slow H2 generation helped stop bubble coalescence but can be further optimized to shorten the time required to block membranes. In conclusion, nanoscale bubbles can be formed on the tips of the MWCNTs to act as membrane valves enabling a new class of active membranes and nanofluidic devices. Alternating (sine) potentials are needed to generate nanoscale bubbles that do not coalesce into macroscale bubbles. Polymer wells formed by electrochemical oxidization in the MWCNT into polymer matrix of the membrane stabilize the nanoscale bubble generated at MWCNT tips with blocking efficiencies up to 92% being seen. These bubble valves can be removed using a 0.004 atm pressure for hydrophilic wells, while hydrophobic wells remain permanently blocked. These nanoscale bubbles valves can be used for controlling the transport through membrane systems with applications in energy storage and drug delivery. Fabrication of Aligned MWCNT Membrane: The fabrication of aligned MWCNT membrane followed prior protocol.[5] Briefly, aligned multiwalled CNTs with an average core diameter of ≈7 nm and a length of 150 μm were prepared via a chemical vapor deposition using ferrocene/xylene as the feed gas.16 Next, Epon 862 epoxy resin (Miller Stephenson Chem. Co.), hardener methylhexahydrophthalic anhydride (MHHPA, Broadview Tech. Inc.), and catalyst 1-Cyanoethyl-2-ethyl-4-methylimidazole (2E4MZ-CN, Shikoku Chemical) were thoroughly mixed using a ThinkyTM Mixer, which was employed to fabricate epoxy/MWCNT composite and cast into a 2 mL polyethylene centrifuge tube. As-prepared epoxy/MWCNT composite was then appropriately cured and removed from cast before being cut into MWCNT membranes using a microtome equipped with a glass blade. The typical thickness of as-cut MWCNT membranes (≈5 × 5 mm) is about 5 μm. Finally, the residual epoxy on the tips of MWCNTs was removed by H2O plasma oxidation. Electrochemical Oxidization on MWCNT Membrane for Polymer Wells: The polymer wells at the tips of MWCNTs were created by electrochemical oxidization on the MWCNT membrane, performed in a three-electrode cell using a potentiostat (PAR Model 263A) with the reference electrode of Ag/AgCl (Bioanalytical) and Pt wire counter electrode. The electrolyte solution was 0.1 m KCl. The bottom of MWCNT membrane was coated with ≈25 nm gold film by DC sputtering (Cressington coating system 308R) to achieve sufficient conductivity between the working electrode and MWCNTs dispersed across the membrane. The plane resistance across membrane after sputtering gold on it is 5 Ω sq−1. For MWCNT electrochemical oxidization, the top surface of the membrane was oxidized for 5 h, at a potential of 2.5 V versus Ag/AgCl, and in a solution of 0.1 m KCl. To render polymer well hydrophobic to stabilized nanobubbles, a partially oxidized MWCNT membrane was immersed in the solution of 0.2 g butyltrichlorosilane (Sigma-Aldrich) in 10.9 g hexane (EM Science) with Ar gas protection for 24 h. After the inner surfaces of polymer wells were rendered hydrophobic, the membrane was screened in the setup of Figure S1b (Supporting Information) to test the flux rate of [Ru(bipy)3]2+ through the membrane. Then the MWCNTs in the polymer wells were electrochemically oxidized again under the same conditions as before to convert the inner surfaces of polymer wells from hydrophobic to hydrophilic to recover the original flux rate. Characterizations of the MWCNT Membrane: The surface morphology of MWCNT membrane with/without electrochemical oxidization was characterized by scanning electron microscopy (SEM) (Hitachi S-4300) under the operating voltage of 20 kV mounted on double-sided conductive tape. The absorbance of light through MWCNT membrane before/after electrochemical oxidization was measured by UV–vis spectrophotometer (USB400-ISS-UV-vis, Ocean Optics Inc.). Electrochemical Bubble Generation on MWCNT Membranes: Constant potential for generating macroscopic bubbles on the MWCNT membrane was carried out in the three-electrode cell using a potentiostat (PAR Model 263A), the MWCNT membrane as the working electrode, Ag/AgCl from Bioanalytical as the reference electrode, and Pt wire as the counter electrode to demonstrate the hydrogen reduction. The setup is shown in Figure S1b (Supporting Information). The constant potential of −1.2 V versus Ag/AgCl was applied on the MWCNT membrane. The electrolyte solution was 5 × 10−3 m [Ru(bipy)3]2+/0.05 m H2SO4. CV measurements used the same apparatus with an electrolyte solution of 5 × 10−3 m [Ru(bipy)3]2+/0.05 m H2SO4. For ac-potential nanobubble generation, an E-corder 410 potentiostat (EDAQ) was used to apply sine potentials on the MWCNT membrane with the same setup and solution as with CV measurements. The authors would like to thank Dr. Mainak Majumder and Dr. Xinhua Sun for their helpful discussion and Prof. Rodney Andrews and Dr. Dalli Qian from the Center for Applied Energy Research, University of Kentucky, for supplying MWCNTs. Facility support was provided by the Center for Nanoscale Science and Engineering and Electron Microscopy Center at the University of Kentucky. X.S. and J.W. acknowledge financial support from DOE EPSCoR (DE-FG02-07ER46375) and NIH NIDA (R01DA018822), respectively. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.