rexresearch.comDavid MITLIN, et al.
NanoHemp SuperCapacitor
NanoHemp SuperCapacitor
http://www.leafscience.com/2013/10/01/top-5-innovative-uses-hemp/
1 Oct, 2013
Top 5 Most Innovative Uses For Hemp
[ Excerpt ]In 1937, Popular Science published an article called “Hemp: The New Billion-Dollar Crop” that listed over 25,000 potential uses for the plant.
While this ancient crop has recently started to gain popularity around the world, it still hasn’t received the attention it deserves.
Which might be due to the fact that hemp is just a type of marijuana that can’t get you high.
But as more countries start to see the benefits of this incredibly eco-friendly crop, a lot of cool R&D has been happening behind the scenes. Here’s 5 of our favorite innovations being developed from hemp:
1. Bacteria-Fighting Fabric...
2. Housing Insulation...
3. Concrete...
4. Cars...
5. Graphene-Like Nanomaterial
Canadian engineers believe hemp could revolutionize nanotechnology
Graphene is often touted as the future of nanotechnology and the thinnest, strongest, and lightest material ever made. But how does hemp compare? Apparently, it’s even better.
Earlier this year, chemical engineers from the University of Alberta turned hemp fiber into a nanomaterial with similar properties as graphene, but a much lower price tag.
What’s more, when it comes to making energy storage devices like batteries and supercapacitors, the hemp nanomaterial showed “superior electrochemical storage properties” compared to graphene.
Research is still in its early stages, but if the results hold, hemp could eventually be used for a wide range of nanotechnology applications, from flashlights to solar cells.
http://cen.acs.org/articles/91/web/2013/05/Energy-Storing-Nanomaterial-Made-Hemp.html
Chemical & Engineering News
May 15, 2013
Energy-Storing Nanomaterial Made From
Hemp
Electronics: Researchers turn agricultural waste into a carbon nanomaterial for high-power supercapacitors
by
Katherine Bourzac
Electronics: Researchers turn agricultural waste into a carbon nanomaterial for high-power supercapacitors
by
Katherine Bourzac
Graphene might one day be used in batteries, solar cells, transparent electrodes, and a host of other electronic gadgets. But graphene is still quite expensive to make. Now researchers at the University of Alberta have demonstrated a low-cost process for turning agricultural waste into graphenelike nanomaterials for use in energy storage electronics (ACS Nano 2013, DOI: 10.1021/nn400731g).
With high surface area and conductivity, graphene is ideal for use as electrodes in batteries and supercapacitors, which are energy storage devices that excel at providing quick bursts of power. Supercapacitors charge and discharge faster than batteries can because they store energy in the form of fast-moving charges on the surfaces of their electrodes. Currently, supercapacitors are used in braking systems for buses and fast-charging flashlights.
Commercial supercapacitors use activated carbon electrodes, but experimental devices made with graphene can store more energy. Unfortunately, graphene’s production costs can’t come close to competing with the price for activated carbon, about $40 per kilogram, says University of Alberta chemical engineer David Mitlin.
Part of Mitlin’s research is finding ways to use plant waste as feedstocks for commercial materials. He thought he could transform waste from the cannabis plant (Cannabis sativa) into a carbon nanomaterial that had similar properties to graphene and with a much smaller price tag. The cannabis plant’s notorious use is for producing marijuana, but people also grow the plant to use its fibrous parts for products such as rope, clothing, oil, and plastics. The plants used for these industrial applications are referred to as hemp, and have lower levels of psychoactive compounds. Hemp is relatively inexpensive, since the plant grows rapidly in a variety of climates without the need for fertilizer and pesticides.
Mitlin and his colleagues focused on a barklike layer of the plant called the bast, which is usually incinerated or sent to landfills during industrial hemp production. “Hemp bast is a nanocomposite made up of layers of lignin, hemicellulose, and crystalline cellulose,” Mitlin says. “If you process it the right way, it separates into nanosheets similar to graphene.”
The Alberta researchers start the process by heating the bast at 180 °C for 24 hours. During this step, the lignin and hemicellulose break down, and the crystalline cellulose begins to carbonize. The researchers then treat the carbonized material with potassium hydroxide and crank up the temperature to 700 to 800 °C, causing it to exfoliate into nanosheets riddled with pores 2 to 5 nm in diameter. These thin, porous materials provide a quick path for charges to move in and out, which is important when a supercapacitor charges and discharges.
The team built a supercapacitor using the nanosheets as electrodes and an ionic liquid as an electrolyte. The best property of the device, Mitlin says, is its maximum power density, a measure of how much power a given mass of the material can produce. At 60 °C, the material puts out 49 kW/kg; activated carbon used in commercial electrodes supplies 17 kW/kg at that temperature.
Liming Dai, a chemical engineer at Case Western Reserve University, says the hempbased material shows promise as a low-cost substitute for graphene. Yury Gogotsi, a materials scientist at Drexel University, sees room for improvement. He points out that the 24-hour high-temperature process will have some associated costs when scaled up. But he’s impressed by this first step. Finding scalable production methods like this one will be key, Gogotsi says, if researchers want to move nanostructured materials out of the lab and into the marketplace.
http://pubs.acs.org/doi/abs/10.1021/nn400731g
Interconnected Carbon Nanosheets Derived
from Hemp for Ultrafast Supercapacitors with High Energy
Huanlei Wang †‡, Zhanwei Xu †‡, Alireza Kohandehghan †‡, Zhi Li †‡*, Kai Cui ‡, Xuehai Tan †‡, Tyler James Stephenson †‡, Cecil K. King’ondu †‡, Chris M. B. Holt †‡, Brian C. Olsen †‡, Jin Kwon Tak §, Don Harfield §, Anthony O. Anyia §, and David Mitlin †‡*
† Chemical and Materials Engineering, University of Alberta, Edmonton, Alberta T6G 2 V4, Canada
‡ National Institute for Nanotechnology (NINT), National Research Council of Canada, Edmonton, Alberta T6G 2M9, Canada
§ Bioresource Technologies, Alberta Innovates-Technology Futures, Vegreville, Alberta, T9C 1T4, Canada
ACS Nano, 2013, 7 (6), pp 5131–5141; DOI: 10.1021/nn400731g
Abstract
We created unique interconnected partially graphitic carbon nanosheets (10–30 nm in thickness) with high specific surface area (up to 2287 m2 g–1), significant volume fraction of mesoporosity (up to 58%), and good electrical conductivity (211–226 S m–1) from hemp bast fiber. The nanosheets are ideally suited for low (down to 0 °C) through high (100 °C) temperature ionic-liquid-based supercapacitor applications: At 0 °C and a current density of 10 A g–1, the electrode maintains a remarkable capacitance of 106 F g–1. At 20, 60, and 100 °C and an extreme current density of 100 A g–1, there is excellent capacitance retention (72–92%) with the specific capacitances being 113, 144, and 142 F g–1, respectively. These characteristics favorably place the materials on a Ragone chart providing among the best power–energy characteristics (on an active mass normalized basis) ever reported for an electrochemical capacitor: At a very high power density of 20 kW kg–1 and 20, 60, and 100 °C, the energy densities are 19, 34, and 40 Wh kg–1, respectively. Moreover the assembled supercapacitor device yields a maximum energy density of 12 Wh kg–1, which is higher than that of commercially available supercapacitors. By taking advantage of the complex multilayered structure of a hemp bast fiber precursor, such exquisite carbons were able to be achieved by simple hydrothermal carbonization combined with activation. This novel precursor-synthesis route presents a great potential for facile large-scale production of high-performance carbons for a variety of diverse applications including energy storage.
CNT COMPOSITES
WO2012129690
WO2012129690
Inventor(s): ZHANG LI [CN]; MITLIN DAVID [CA]; HOLT CHRIS [CA] +
Abstract
Synthesis of three-dimensional (3D) electrochemically supercapacitive arrays is disclosed. The arrays comprise multi-walled carbon nanotubes conformally covered by nanocrystalline functional materials such as vanadium nitride and firmly anchored to glassy carbon or Inconel electrodes. These nanostructures demonstrate a respectable specific capacitance of 289 F/g, which is achieved in 1 M KOH electrolyte at a scan rate of 20 m V/s. The well-connected highly electrically conductive structures exhibit a superb rate capability; at a very high scan rate of 1000 m V/s there is less than a 20% drop in the capacitance relative to 20 m V/s. Such rate capability has never been reported for VN and is highly unusual for any other oxide or nitride. These 3D arrays also display nearly ideal triangular voltage profiles during constant current charge-discharge cycling.
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application claims the benefit under 35 USC 119(e) of United States provisional application serial no. 61/468,020 filed March 27, 2011.
TECHNICAL FIELD
[0002] Nanomaterials.
BACKGROUND
[0003] Electrochemical capacitors (ECs), have attracted wide interest as energy storage/conversion devices owing to their outstanding properties of high power delivery, rapid charge/discharge rates, high cycle efficiency, and extremely long cycle life. Such predominant advantages make them good candidates in a large variety of applications including portable electronics, hybrid and fully electric vehicles, and load leveling for stationary power. Depending on the charge-storage mode, electrochemical capacitors are generally classified into two major types: electrochemical double layer capacitor (EDLC) and pseudocapacitor. The capacitance in the former is electrostatic in origin, arising from charge separation at the interface of the high specific- area electrode and electrolyte. Activated, mesoporous, and carbide-derived carbons, graphene, carbon fabrics, fibers, nanotubes, onions, and nanohorns as well as various nanostructured polymers with high specific surface area and moderate cost have been widely investigated for EDLC applications.
[0004] Pseudo-capacitors that utilize transition metal nitride electrode materials are attracting increasing scientific attention. These nitrides were recently named as one of the three greatest recent achievements in the electrochemical capacitor field. Compared to transition metal oxides, transition metal nitrides exhibit superior properties in some respects, such as markedly better electrical conductivity and in some cases excellent chemical stability. Various materials such as molybdenum nitrides (MoN and Mo2N) and vanadium nitride (VN) have been investigated for electrochemical capacitor use. Among the nitrides, VN is considered to be one of the more promising materials. In previous studies of VN, however, the capacitance would significantly degrade at high voltage scanning rates/ current densities. For example, the superb capacitance reported in reference 13 degraded by more than 50% when an intermediate scan rate of 100 mV/s was utilized.
[0005] Another promising supercapacitive technology is based on nanocomposites, with an inner electrically conducting tubular core and an outer nanoscale functional coating. Due to their excellent electrical conductivity, inherent nanoscale dimensions, and high surface area, carbon nanotube (CNT) arrays have attracted great interest as three-dimensional (3D) current collector materials for such applications. Unfortunately the extremely high packing density and intrinsic hydrophobicity of ordered CNT arrays prevents the conformal chemical or electrochemical deposition and utilization of the capacitive materials. The capacitive materials wind up only being deposited near the CNT tips or at their contact junctions. Alternatively, loose CNTs may be mixed with a binder and a functional material, and subsequently applied onto a current collector. This approach is more involved and often yields unsatisfactory results. For example, vanadium nitride/carbon nanotube composites were prepared by sonicating nanocrystalline VN with multiwalled carbon nanotubes (MWCNTs) in methanol. The capacitance values reported were modest: the absolute best performance reported was a capacitance of roughly 160 F/g at scan rate of 100 mV/s.
[0006] Ideally one would create supercapacitive nanocomposites directly anchored onto electrically conducting current collectors. This not only would reduce the complexity of the synthesis process but also would ensure improved power performance due to elimination of interparticle resistance. In many cases, however, CNTs are grown on non-conducting substrates (especially Si02 and A1203), requiring their removal and transfer. An A1203 barrier layer is also extensively utilized to grow CNTs on a variety of other substrates such as Ni and Fe foils, glassy carbon (GC), and Ta. Ajayan and co-workers have developed a novel strategy for direct growth of CNT arrays on an Inconel 600 metallic alloy. The authors proposed that a tenacious nanoscale passivating layer of Cr203 on the Inconel surface plays a key role for CNT growth.
SUMMARY
[0007] A method of creating an energy storage or energy generation device is disclosed comprising growing anchored carbon nanotubes (CNTs) on an electrically conducting substrate with a spacing suitable for functionalization, and functionalizing the CNTs by conformally coating the CNTs with an energy storage or generation material.
[0008] The resulting device is also claimed comprising anchored carbon nanotubes (CNTs) extending from an electrically conducting substrate with a spacing suitable for functionalization, and an energy storage or generation material conformally coating the CNTs. [0009] In one embodiment, the electrically conducting substrate comprises a CNT growth support film with growth catalyst particles deposited onto, built into or forming part of the growth support film.
[0010] Exemplary bulk substrates include Inconel, copper, stainless steel, glassy and vitreous carbon, graphite, aluminum, magnesium, conventional steel, niobium, titanium, molybdenum, tungsten, zirconium, silicon carbide, and tungsten carbide.
[001 1 ] Exemplary growth support films include oxides of metals, of non-metals and of alloys including oxides of Si, Mg, Fe, Cr, Cr-Ni, Ni, Ir, Zr, Sc, Ti, Mo, and Ta, such as Si02, MgO, Ti02, NbO, Nb02, Nb205, Ti-doped Nb02, Cr203, etc. Support films also include nitrides such as TiN and TixV1-x N (TiVN).
[0012] Exemplary growth catalysts include metals as Ni, Fe, Cr, Co, Cu, Au, Ag, Pt, Pd, Mn, Mo, Sn, Mg, and alloys that combine two or more of these elements; oxides of these metals and of their alloys; and nitrides of these metals and of their alloys.
[0013] In various embodiments, there may be included any one or more of the following features: The electrically conducting substrate comprises growth catalyst particles. The electrically conducting substrate comprises a CNT growth support film with the growth catalyst particles deposited onto, built into or forming part of the CNT growth support film. The growth catalyst particles have a wetting angle of 30-90 degrees with the CNT growth support film. The as-deposited thickness of the growth catalyst layer has a thickness of 1-10 nm before it breaks up into growth catalyst particles. The electrically conducting substrate comprises one or more of Inconel, copper, stainless steel, glassy and vitreous carbon, graphite, aluminum, magnesium, conventional steel, niobium, titanium, molybdenum, tungsten, zirconium, silicon carbide, and tungsten carbide. The CNT growth support film comprises one or more of oxides or nitrides of metals, non-metals and alloys. The CNT growth support film comprises one or more of TiN and TixVi-x N and oxides of Si, Mg, Fe, Cr, Cr-Ni, Ni, Ir, Zr, Sc, Ti, Mo, and Ta. The growth catalyst particles comprise one or more of Ni, Fe, Cr, Co, Cu, Au, Ag, Pt, Pd, Mn, Mo, Sn, Mg, alloys that combine two or more of these metals, oxides of these metals, oxides of alloys that combine two or more of these metals, nitrides of these metals, and nitrides of alloys that combine two or more of these metals. The energy storage or generation material comprises a host for Li ions. The energy storage or generation material comprises a provider of reversible faradaic electrochemical oxidation/reduction reactions. The energy storage or generation material comprises one or more of Si, Al, Sn, Co02, Fe203, Mn02, Fe304, FeOOH, MgH2, sulfur, FeS, TixVi-xN, NiCo204, Co304, cobalt hydroxide, iron hydroxide, and VN. The spacing of the CNTs falls within the range of 200 nm to 300 nm. The apparatus is formed as a lithium-ion battery or supercapacitor.
[0014] These and other aspects of the device and method are set out in the claims, which are incorporated here by reference.
BRIEF DESCRIPTION OF THE FIGURES
[0015] Embodiments will now be described with reference to the figures, in which like reference characters denote like elements, by way of example, and in which:
[0016] Fig. 1A shows an SEM micrograph of an as-grown array of CNT/Inconel/GC.
[0017] Fig. IB shows TGA (axis on left) and DSC (axis on right) analysis allowing estimation of amorphous carbon fraction.
[0018] Fig. 2A is an SEM micrograph of VN/GC.
[0019] Fig. 2B is an X-ray diffraction pattern of VN/Si.
[0020] Figs. 2C and 2D are increasing magnification SEM micrographs of
VN/CNT/Inconel/GC.
[0021] Fig. 2E is a schematic summarizing the synthesis process and ultimate morphology of the VN/CNT/3D array showing CNTs before coating, Cr-Fe-Ni bases to the CNTs, VN coated CNTs, oxidized Inconel upper substrate, Inconel 600 intermediate substrate and bulk GC or Inconel substrate. In all the figures 2A-2E, unless indicated otherwise, the VN mass loading is 0.037 mg per geometrical cm<2>.
[0022] Fig. 3Aa is a bright-field TEM micrograph on the VN covered CNT tips; Figs. 3B and 3C are bright field and HAADF images of the VN/CNT structure, respectively; and Fig. 3D is an indexed SAED pattern of the VN/CNT structure imaged in Figs. 3B and 3C.
[0023] Fig. 4A is an SEM micrographs of CNT/A1203; and Fig. 4B is an SEM micrograph of VN covering only the top surface of CNT/A1203/GC.
[0024] Fig. 5A is a graph showing a CV curve of CNT/Inconel with data shown at cycle 1 and at cycle 10000; Fig. 5B is a graph showing CV curves of GC and of VN/GC; Fig. 5C is a graph showing CV data for CNT/Inconel, and for the remnant catalyst on Inconel, after the CNTs were removed; and Fig. 5D is a graph showing CV data for CNT/Inconel and for VN/CNT/Inconel.
Cycling for Fig. 5A was at 50 mV/s, while cyclings for Figs. 5B-5D were at 100 mV/s.
[0025] Fig 6A shows CV curves of VN/CNT/Inconel/GC in the potential range of -1.0 to
0.06 V, at different scan rates from 20 to 1000 mV/s; Fig. 6B shows the relationship between the specific capacitance versus scan rate for various mass loadings of VN; Fig. 6C shows the charge- discharge cycling curves at different current densities; and Fig. 6D shows the specific capacitance versus current density.
[0026] Fig. 7A shows Nyquist plots of CNT/Inconel/GC and Fig. 7B shows Nyquist plots for VN/CNT/Inconel/GC electrodes. The insets are the enlarged Nyquist plots in the high frequency region. The electrolyte is 1 M KOH.
[0027] Fig. 8A shows the relationship between the specific capacitance versus cycle number with different VN mass loading, using a scan rate of at 50 mV/s; Figs. 8B and 8C are SEM micrographs of the pre-cycled and post-cycled (250 cycles) nanostructures.
[0028] Figs. 9A and 9B are XPS spectra of the VN thin films before electrochemical cycling. Figs. 9C and 9D are XPS spectra after 250 electrochemical cycles.
[0029] Fig. 10 shows Raman spectra of CNTs grown on Inconel.
[0030] Fig. 11A shows an HRTEM image of CNTs grown on Inconel and Fig. 1 IB is an
HRTEM image of a CNT grown on A1203.
[0031] Fig. 12A is a graph showing CV curves of GC and of VN/GC and
Fig. 12B shows CV data for CNT/Inconel and for VN/CNT/Inconel. Cycling for both Figs. 12A and 12B was at 20mV/s.
[0032] Fig. 13A shows cyclic voltammogram (CV) curves of CNT/A1203. Data is shown at cycle 1 and at cycle 10000. Cycling was performed at a sweep rate of 50 mV/s. Fig. 13B shows CV data for CNT/A1203and for VN/CNT/A1203 at 20 mV/s.
[0033] Fig. 14 shows the electrically equivalent circuit used for fitting impedance spectra.
DETAILED DESCRIPTION
[0034] A method of creating an energy storage or energy generation device is disclosed comprising growing anchored carbon nanotubes (CNTs) on an electrically conducting substrate with a spacing suitable for functionalization, and functionalizing the CNTs by conformally coating the CNTs with an energy storage or generation material. The resulting device is also disclosed comprising anchored carbon nanotubes (CNTs) extending from an electrically conducting substrate with a spacing suitable for functionalization, and an energy storage or generation material conformally coating the CNTs. In one embodiment, the electrically conducting substrate comprises growth catalyst particles. For example, the electrically conducting substrate may comprise a CNT growth support film with the growth catalyst deposited onto, built into or forming part of the CNT growth support film.
[0035] The principle of selection of the growth support film and growth catalyst is as follows: The support film should 1) prevent excessive catalyst wetting or dewetting, i.e., have the appropriate film-support interaction to generate a desired catalyst particle size, distribution and spacing, which in turn dictates the CNT size, distribution, spacing and height; 2) prevent or reduce to an acceptable level the interdiffusion of the catalyst into the support film and/or the underlying substrate; 3) be sufficiently electrically conductive and electrochemically stable for the desired application; 4) maintain a strong bond with the bulk substrate, not delaminating or spalling; and 5) maintain a strong bond with the grown CNTs as to prevent CNT delamination or separation from the substrate. The spacing and dimensions of the dewetted catalyst particles and hence of the CNTs may be tailored by the use of the force-balance wetting equation, knowing that the catalyst mass is conserved, and knowing that the as-deposited catalyst thickness is in the 1 - 10 nm range. The wetting angle for the film on the support should be greater than 30[deg.] but less than 90[deg.]. Balancing the forces on the wetted catalyst results in the following range of energies, i.e., [gamma]: y{substrate-vapor} = y{catalyst-vapor}*cos(30[deg.]) + y{substrate-catalyst} to y{substrate-vapor} = y{catalyst- vapor}*cos(90[deg.]) + y{substrate-catalyst} .
[0036] For those materials and methods for which experimental data is not provided herein, in general, the predicted utility is based on the disclosed results and application of known principles of vapor and liquid-phase deposition technologies and of supercapacitor and battery materials. The materials for which predictions are provided have been demonstrated to work in bulk or in other nanoscale embodiments. Functionalizing the surfaces of the CNTs with the predicted materials will improve their performance over bulk or other nanoscale embodiments. This is because the CNT supports will provide an excellent electrical conductivity path to the underlying substrate while enabling an ultra-high surface area-to-volume ratio for the battery or supercapacitor material coating. Predictions are also based upon the thermodynamics and phase diagrams of the respective materials systems.
[0037] Exemplary bulk substrates include Inconel, copper, stainless steel, glassy and vitreous carbon, graphite, aluminum, magnesium, conventional steel, niobium, titanium, molybdenum, tungsten, zirconium, silicon carbide, and tungsten carbide. Each of these materials has excellent electrical conductivity, which is in or near the metallic range. This makes them suitable as electrodes. Exemplary growth support films include oxides of metals, of non-metals and of alloys including oxides of Si, Mg, Fe, Cr, Cr-Ni, Ni, Ir, Zr, Sc, Ti, Mo, and Ta, such as Si02, MgO, Ti02, NbO, Nb02, Nb205, Ti-doped Nb02, Cr203, etc. Support films also include nitrides such as TiN and TixVi_x N (TiVN). These support growth films would work because 1) they are stable at the CNT growth temperatures (i.e. Tmeltmg of film " growth temperature), 2) they may be used to block diffusion of the CNT growth catalyst into the substrate, and 3) they have some thermodynamic chemical interaction with the growth catalyst, therefore causing partial wetting and preventing catalyst agglomeration.
[0038] Exemplary growth catalysts include metals as Ni, Fe, Cr, Co, Cu, Au, Ag, Pt, Pd, Mn, Mo, Sn, Mg, and alloys that combine two or more of these elements, oxides of these metals and of their alloys, and nitrides of these metals and of their alloys. These growth catalysts would work because they chemically interact with the precursor hydrocarbon gases (such as ethylene). In addition they all possess some solubility (in solid or liquid phases) for atomic carbon and/or they form carbides, the presence of at least one of the two being a necessary requirement for CNT growth.
[0039] Exemplary energy storage or generation devices include lithium-ion batteries and supercapacitors. The achieved CNTs may be functionalized with a variety of energy storage materials that either act as hosts for Li ions (batteries) or provide reversible faradaic electrochemical oxidation/reduction reactions and therefore act as supercapacitors. In regards to batteries, these materials may be used in the positive (cathode) and/or the negative (anode) electrodes. In regards to supercapacitors, these materials may be used for symmetrical or asymmetrical electrodes. In regards to battery materials such embodiments include, but are not limited to, elemental materials such as Si, Al, Sn, and their alloys, oxides and hydroxides such as Co02, Fe203, Mn02, Fe304 and FeOOH, hydrides such as MgH2, sulfur and sulfides such as FeS, and elemental and alloy nitrides such as TiVN. In the case of supercapacitor applications such embodiments include, but are not limited to, oxides such as NiCo204, Co304, Fe203 and Mn02, hydroxides such as cobalt and/or iron hydroxide, and nitrides such as VN and TiVN. These materials would work because it is possible to deposit them onto the carbon nanotubes using a variety of techniques such as physical vapor deposition. Furthermore these materials would work because they have been demonstrated to function as an energy storage phase in the CNT-unsupported state. However in the current embodiment we expect these materials to possess improved performance since they are intimately coupled to an electrically conductive skeleton, which would not only accelerate the electronic conductivity but also the ionic conductivity through field effects. Furthermore the CNT arrays provide a natural way for the coating materials to achieve nanoscale dimensions. This would improve the kinetics via a reduction of the diffusion distances of the diffusing species. It would also improve the gravimetric energy density due to a higher surface area-to-volume ratio of the energy storage phase created by the underlying support skeleton.
[0040] Deposition techniques for depositing a growth support film and growth catalyst include one or a combination of physical vapor deposition (PVD) methods such as sputtering, evaporation, and pulsed laser deposition; chemical vapor deposition (CVD); atomic layer deposition (ALD); electroplating and electroless deposition; and wet chemical techniques. These deposition techniques would work since they are capable of coating the CNT arrays with the desired materials with high mass-loading accuracy (some techniques such as ALD and sputtering down to microgram) and uniformity.
[0041 ] Growing of the CNTS may be carried out by all suitable methods including and related to chemical vapor deposition (CVD) such as conventional CVD, thermal CVD, and plasma- enhanced CVD. These methods would work since they are scientifically established and widely technologically demonstrated methods for CNT growth.
[0042] Functionalization techniques include one or a combination of physical vapor deposition (PVD) methods such as sputtering (including reactive sputtering), evaporation, and pulsed laser deposition; chemical vapor deposition (CVD); atomic layer deposition (ALD); electroplating and electroless deposition; and wet chemical methods such as precipitation, hydrothermal processing, and ionothermal processing. These methods would work since they have been scientifically established and technologically demonstrated for deposition of these functional materials on other types of carbon-based supports (not anchored CNT arrays), such as loose CNTs, graphene flakes, and activated carbon powder. It is expected that it will be possible to utilize all of the above techniques given the exquisite control of the CNT array spacing and height demonstrated here.
[0043] The device may be used, for examples, as a capacitor, battery or a hybrid battery- capacitor. Each of these applications requires a good electrical path from the functional material to the electrode. In each case a significant benefit is achieved when the functional materials and current collectors (dual role of the CNT array as a skeleton support and a current collector) are directly anchored to the underlying substrate (electrode) as they are shown herein.
[0044] Spacing of the CNTs is preferably greater than 100 nm, more preferably in the range 200-300 nm. Dry deposition techniques such as PVD and CVD are preferred for depositing the support film and catalyst, growing the CNTs and conformally coating the CNTs.
[0045] A specific example of the method and device is disclosed in this patent document. However, we can soundly predict that the additional examples given above will work in like manner to the specific examples disclosed due to the similar nature of the processes and materials.
[0046] We approach electrochemical capacitor synthesis in a new way. In some embodiments our methodology involves no wet chemical routes and creates in one embodiment 3D arrays of nanostructures that are directly anchored on bulk GC or bulk Inconel electrodes. This is achieved using three steps: First, we use physical vapor deposition (PVD) or other suitable methods disclosed in this document to deposit a CNT growth support film such as an Inconel-based growth support film and a CNT growth catalyst onto a bulk substrate. By this example, we create an electrically conducting substrate with a spacing suitable for functionalization. For a baseline, we also examine the feasibility of using a "conventional" A1203 CNT growth support. Second, we grow anchored 3D arrays of CNTs via conventional chemical vapor deposition (CVD) or other suitable methods disclosed in this document. We demonstrate that the achieved CNT distribution is ideally suited for subsequent functionalization. This functionalization is achieved through a conformal coating of the CNTs by a nanocrystalline VN film or other suitable materials disclosed in this document. The nitride is deposited via PVD or other suitable method and constitutes the third and last step in this example of the synthesis process. These structures are then electrochemically tested. The CNT growth support film may have the growth catalyst particles deposited onto, built into or forming part of a CNT growth support film.
[0047] One significant advantage of our synthesis methodology is its potential commercial scalability. Both CVD and PVD are well-established advanced manufacturing technologies, being widely employed for batch-process manufacturing of microelectronics and solar cells. In addition to, or instead of, PVD, atomic layer deposition (ALD) may be utilized to grow the conformal functional oxide or nitride coatings on the CNT arrays. In regards to aligned CNT arrays (not loose tubes), a CVD-based method has the most potential for industrial -scale production because it is the technique that is capable of growing nanotubes directly on desired substrates. CVD-based techniques for growth of CNTs are also rapidly evolving in their efficiency and process control accuracy, with several notable examples. Recent advances include water-assisted supergrowth of dense and aligned CNT forests; additional dramatic process improvements which allow CVD to be utilized to grow centimeter tall, aligned CNT arrays, as well as arrays that fully cover 210 x 297 mm (A4 paper) substrates; and work by researchers at our institute, who produced CNT films on a large scale using a commercial CVD reactor. We utilize the same manufacturing-scale CVD reactor for the current study. The nanocomposite 3D array methodology detailed here may also be utilized for other applications where electrical conductivity to the substrate is essential, such as proton exchange membrane (PEM) fuel cells and Li -ion battery electrodes.
[0048] The following discloses experiments that provide a factual basis for the predictions in this patent document.
[0049] Two types of bulk substrates were utilized in this study: polished Inconel and GC, both being round disks 11.32 mm in diameter and 2 mm in thickness. These materials had the dual role of also being utilized as the electrodes in the electrochemical experiments. The disks were polished to a mirror finish with 6.0, 1.0, 0.25, and 0.05 um polycrystalline diamond paste (Allied High Tech Products, Inc.) in sequence. After each polishing step, the electrodes were ultrasonically cleaned in Milli-Q (Millipore) water. Finally, the electrodes were ultrasonically cleaned in 8 M HN03, acetone, isopropanol, and Milli-Q water in sequence, and dried under argon flow.
[0050] Inconel support films of 150 nm in thickness were synthesized via PVD onto the electrodes. This was achieved via radio frequency (RF) magnetron sputtering (AJA International, Inc.), from an Inconel-600 target (VBM Precision Machine), in ultra-pure argon (99.999%) plasma. Prior to all depositions the sputtering chamber reached a base pressure of less than or equal to 5 [chi] 10 <s> Torr. A substrate temperature of 800 [deg.]C was used to obtain crystalline Inconel films. The films possessed good adhesion to either electrode. For this PVD step, the sputtering rate was accurately measured in-situ using a crystal monitor at the substrate plane. A deposition rate of 2.5 nm/min was utilized. Inconel-coated electrodes were then placed in the center of a tube furnace (MTI, GSL1100X). Oxidation was accomplished by heating the samples to 700 [deg.]C at a rate of 15[deg.]C/min under a flow of dried air (100 seem total). The sample was held at that temperature for 60 s. Then the furnace was slowly cooled to room temperature under an argon flow (200 seem total). This oxidation process produced roughly a 100 nm thick passivation oxide (blue in color), which acted as the diffusion barrier layer for the CNT growth. A 1 nm Cr/1 nm Fe-30 atom % Ni bilayer CNT growth catalyst was subsequently deposited via direct current (DC) magnetron sputtering and co-sputtering, using elemental targets.
[0051 ] In order to investigate the effect of CNT density on the resulting VN/CNT composite structure, CNT growth catalyst was deposited on an A1203 support as well. Twenty five nanometers of A1203 was reactively sputtered by dc magnetron sputtering from an Al target (99.995% purity) in mixed argon-oxygen plasma. The substrate was held at 350 [deg.]C during deposition. The total pressure in the sputtering chamber was maintained at 5 <[chi]> 10<"3> Torr for all the experiments, and the volume flow rate ratio of argon to oxygen was 9.5: 1.5 (total 11 seem) . After sputtering, the oxide films were annealed at 400 [deg.]C for 60 min. Deposition rate calibration was performed ex situ through the use of a profilometer (KLA-Tencor Alpha-Step IQ) to measure step heights. The 1 nm Cr/1 nm Fe-30 atom % Ni bilayer CNT growth catalyst was subsequently deposited on top.
[0052] The CNT arrays were synthesized via a commercial CVD reactor (Tystar, Inc.). The mass-production designed system handles up to fifty 150 mm wafers simultaneously, with industry standard process control. The basic process flow is load, heat and reduce, grow, cool, and unload. The electrodes were loaded into the reactor at 550 [deg.]C; the reactor was then purged with argon and heated to the growth temperature of 750 [deg.]C using a rate of 10 C7min under a flow of 10% hydrogen in argon (3000 seem total).
[0053] Table 1. Legend for the Specimen Configurations Utilized in the Supporting Experiments
[0054] The electrodes were then held at 750 [deg.]C for a total time of 40 min. During the growth stage, 15% ethylene and 10% hydrogen in argon (3300 seem total) flowed into the reactor for a time of 4-6 min. The reactor was then cooled to 350 [deg.]C under a flow of argon (2700 seem total), and the samples were removed.
[0055] Vanadium nitride thin films were grown using a similar approach to that by Zasadzinski et al. The methodology consisted of reactive RF magnetron sputtering from an elemental vanadium target (99.995% purity) in mixed argon-nitrogen plasma. The optimum substrate temperature for well-crystallized films was found to be 580 [deg.]C. With the heater at temperature, the base pressure in the chamber was less than 5 [chi] 10 <s> Torr. High -purity argon and nitrogen were injected directly into the chamber with a volume flow rate ratio of 16.8:3.2 (total 20 seem). The deposition pressure in the sputtering chamber was maintained at 4 [chi] 10<"3> Torr. The VN films were subsequently annealed in the chamber at 610 [deg.]C for 60 min in a nitrogen flow (10 seem, 4 [chi] 10<"3> Torr). Deposition rate calibration was performed ex situ through the use of a profilometer (KLA- Tencor Alpha-Step IQ) to measure the step heights. The deposition rates of VN were on the order of 1 nm/min. The VN mass loading was calculated from the film thickness and the known density of crystalline VN. The loading is presented in units of milligram per geometric (not real) surface area.
[0056] The phases present were identified using glancing angle X-ray diffraction (XRD) using a Bruker AXS D8 Discover diffractometer with a GADDS area detector. Phase identification was performed on -350 nm thick VN films grown on 4 in. Si substrates. The wafer was oriented to avoid the primary Si reflections in the detector. Sheet electrical resistivity measurements were taken using a four-point probe configuration (Lucas Laboratories). Scanning electron microscopy (SEM) was performed using a Hitachi S-4800 operated at 20 kV accelerating voltage. Transmission electron microscopy (TEM) was performed using a JEOL 2010 operated at 200 kV accelerating voltage. For TEM analysis the structures were mechanically removed from the electrodes and dry-dispersed onto grids. Electron diffraction simulation was performed using the commercial software Desktop Microscopist(TM), with the input of the well-known space group information for the constituent cubic VN phase.
[0057] X-ray photoelectron spectroscopy (XPS) measurements were performed using an Axis Ultra spectrometer (Kratos Analytical) with a base pressure of 5 [chi] 10<"10> Torr. X-rays were generated by an Al mono (Ka) source operated at 210 W. The spectra were collected at a 90[deg.] takeoff angle. For the survey spectra, the analyzer pass energy was 160 eV, and for the high -resolution spectra the pass energy was 20 eV. The bonding energy scale in XPS measurements was corrected for the charging effect by assigning a value of 284.6 eV to the C Is peak of adventitious carbon. Peak-fitting identification was performed using the system software and the NIST XPS database as a cross-check.
[0058] Simultaneous thermogravimetric analysis (TGA) and differential scanning calorimetery (DSC) (SDT Q600, TA Instruments) were applied to precisely determine the weight fractions of carbon nanotubes, amorphous carbon, and catalysts. As-prepared carbon nanotubes were scraped from the substrates, and the analysis was performed in an aluminum oxide crucible under a constant dried air flow (2 mL/min). The samples were heated from room temperature to 1000 [deg.]C at a heating rate of 5 C7min.
[0059] All electrochemical measurements including cyclic voltammetry and constant current charge-discharge cycling were performed using a standard rotation disk electrode (RDE) system (Princeton Applied Research model 616 RDE), a potentiostat (Versa STAT 3), and an in-house fabricated three-compartment Teflon cell. A helical Pt wire counter electrode and an Hg/HgO reference electrode with taper joint (CHI 152, CH Instruments, Inc.) were used. The GC and Inconel working electrodes were directly assembled into Teflon holders. All electrochemical measurements were carried out in 1 M KOH (analytic grade, Fisher Scientific) at room temperature without any rotation. All CV curves depicted in the results represent the steady state after -40-50 cycles unless specifically indicated otherwise.
[0060] Electrochemical impedance spectroscopy (EIS) measurements were carried out to further verify the merit of a 3D CNT/Inconel support on the electrode performance. All the tests were conducted (three-electrode system, with Hg/HgO as reference electrode) on a Solartron 1470 electrochemical test station, applying an alternating current in the frequency range from 1 Hz to 20 kHz, with 5 mV amplitude, around 0 V vs open circuit potential (OCP, around 0.16 V vs reference electrode).
[0061 ] We use a series of shorthand notations for the types of samples examined in this study. The abbreviation along with the sample description and purpose are presented in Table 1.
[0062] Fig. 1 A provides an SEM micrograph of the as-synthesized CNT array on Inconel/GC. The diameters of the CNTs on Inconel/GC are in the 60 nm range. The nanotubes are relatively coarsely distributed on the substrate, with spacing on the order of 200-300 nm. The height of these CNTs is 45 [mu][iota][eta]. The number density of the CNTs is (1.01.5) <[chi]> 10<9>/cm<2>. These dimensions and spacings yield a ratio of -15 for the true substrate surface area versus the geometric surface area.
[0063] Simultaneous differential scanning calorimentry (DSC) and thermogravimetric analysis (TGA) were performed in a dry flowing air environment. Such analysis allows for an estimate of the mass proportion of carbon nanotubes, amorphous carbon, and residual catalysts. Fig. IB shows the results. The decomposition onset temperature for MWCNTs is around 515 [deg.]C. All the amorphous carbon was burnt before this onset temperature, with its mass fraction being on the order of 5%. By 710 [deg.]C, 98% of the total mass is lost, which indicates that 93% of the material by mass is MWCNT. At higher temperatures only the growth catalyst remains.
[0064] Fig. 2A presents an SEM micrograph of a 60 nm thick VN film grown on a GC substrate. The film had a mass loading of 0.037 mg/cm<2> by geometric area of the GC substrate. In the subsequent results, unless indicated otherwise, the VN mass loading is that value. The film was continuous and had good electrical conductivity. Its average measured resistivity was 43.5 [Gamma][Pi][Omega] cm. It is well-known that good electrical conductivity is key for fast and reversible redox reactions at high charging/discharging rates. Fig. 2B shows the XRD pattern of the synthesized vanadium nitride thin film. Three strong diffraction peaks are observed at 2[Theta] value of 38. , 44.3[deg.], and 64.4[deg.]. These diffraction peaks can be ascribed to the crystal planes of (111), (200), and (220). The as-prepared VN film can be considered to be a cubic crystal system (Fm3 m [225]), with a unit-cell parameter a = 4.085 A, which is almost identical to the 4.09 A of the stoichiometric VN (space group [225] Fm3 m , Joint Committee on Powder Diffraction Standards (JCPDS) Power diffraction File Card No. 25- 1252). We can conclude that the VN synthesized by reactive sputtering in ultrapure argonnitrogen mixture is nearly stoichiometric.
[0065] It also can be concluded that the diffraction peaks of VN are broadened, indicating a small size of individual crystallites. While the grain size is expected to be different between the several tens of nanometer thick (for example, 60 nm VN on GC) and the hundreds of nanometer thick samples, the structure of the nitride phase should be nearly identical.
[0066] Fig. 2C and Fig. 2D highlight the VN/CNT/Inconel/GC system. The case of VN/CNT/Inconel/Inconel is not shown, since the electrochemical capacitor structure (i.e., CNT diameter, height and spacing, VN coverage) is nearly identical to that of VN/CNT/Inconel/GC. For both the blanket VN films and for the CNT-supported VN, a faceted VN nanocrystallite crystallographic morphology is observed. Such morphology is expected for VN films deposited at high temperatures, where crystals nucleate in their equilibrium Wulff shape, due to fast long-range mass transport on the film surface. Fig. 2E provides a summary schematic of the 3D array of the VNCNT nanocomposites, with all the support and catalyst layers included. Conceptually the figure is similar to the "ideal" electrochemical capacitor core-shell structure proposed by Simon and Gogotsy, though in our case the supercapacitive film is fully conformal to the CNT array whereas in their case, it is depicted as partially wetted. Though we did not investigate this in detail, it is reasonable to assume that the CrFeNi catalyst film will dewet on the oxidized Inconel, forming catalyst islands that are roughly on the same scale as the diameter of the resulting CNTs. What is not captured in the schematic is that limited tip growth may also occur. In both the base-growth (depicted) and the tip- growth cases, the catalyst is likely to be ultimately encapsulated by a graphitic layer. Further investigation is needed to fully understand the structure of this particular catalyst and its interaction with the oxidized Inconel support.
[0067] Figs. 3A-3D show the TEM characterization results of the VN/ CNT/Inconel/GC 3D arrays. Fig. 3A shows a bright field micrograph of the top of the CNTs, highlighting their coverage by the nanocrystalline VN. Figs. 3B and 3C show bright field and high angle annular dark field (HAADF) (Z -contrast) micrographs, respectively. The nanostructures were sheared from the electrode right at the base. The portions imaged in Figs. 3B and 3C were close to the fracture surface, i.e., near the bottom. These images confirm that the VN nanocrystallites conformally cover the CNTs. The coverage is not perfectly uniform, however. The CNTs were not fully vertically aligned and hence received somewhat varying levels of coverage depending on their orientation to the atomic flux. The corresponding selected area electron diffraction (SAED) pattern and accompanying simulation is shown in Fig. 3D. The pattern displays the expected VN diffraction ring pattern. Additionally there are the 0002 and 0004 "graphitic" MWCNT reflections.
[0068] CNT arrays grown on A1203/GC are both much denser and much longer compared to those on Inconel (see Figs. 4A and 4B). A typical height and diameter of a CNT is 35 um and 10 nm, respectively. The spacing of CNTs is quite fine, being on the order of 10-15 nm. The coverage of the VN on CNTs grown using the A1203 support films is quite different from that of the Inconel case. Carbon nanotubes grown using Inconel are covered by the VN. In the case of the CNTs grown using the A1203 support, the VN covers only the very top fraction of the nanotubes.
[0069] The difference in the morphology between CNT arrays grown using the Inconel versus the A1203 films is likely due to varying catalyst substrate interaction. The CNT height, diameter, and spacing difference between the two arrays can be ascribed to catalyst stability. It is well known that for thin-film -based catalysts, CNT growth is a two-step process. First is the restructuring of a thin film into separated nanoclusters of catalyst particles and subsequent particle coarsening. The kinetics of thin film de wetting and breakup have been recently detailed by Luber et al. The second stage occurs upon the introduction of the carbon feedstock gas and involves graphitic network formation and actual CNT growth.
[0070] An A1203 support layer is one of the more effective substrates for CNT growth. Its relatively strong chemical interaction with the Fe-based catalyst particles results in a narrow and fine particle size distribution that is relatively stable against coarsening. A narrow and stable catalyst size distribution, in turn, is known to lead to well-ordered and uniform arrays of vertically aligned CNTs, as is observed in this study. Conversely, Inconel-based films have a weak interaction with Fe-based catalysts. Catalyst substrates with a weak chemical interaction will result in catalyst islands that are widely spaced and dimensionally coarse at elevated temperatures. During CNT growth these particles will also undergo accelerated Ostwald ripening relative to the case of the reactive substrates, e.g., see ref 42 and citations therein. Coarsened catalyst particles will grow CNTs with correspondingly large diameters and spacing. The CNTs' coarse spacing will lead to them folding over during growth due to low steric hindrance as is evident from Fig. 1A. Because the CNTs grown on A1203 are too dense to become fully coated by VN during sputtering, the remainder of the paper will focus on CNTs grown on Inconel as the electrochemical properties are also clearly superior.
[0071 ] We utilized cyclic voltammetry (CV) and constant current charge-discharge (CD) cycling to evaluate the supercapacitive behavior of VN on the different substrates. The specific capacitance (SC), in farads per gram, based on CV and CD cycling can be calculated by the following equations:
"; - - 2 ) where Q is the integrated cathodic (or anodic) charge, m is the mass of the active material, [Delta][Epsilon] is the potential window, i is the discharge (or charge) current, and At is the discharge time. All SC values were calculated by averaging the capacitance of symmetric anodic and cathodic halves. We first investigated the electrochemical stability of CNTs on Inconel/GC, without the presence of VN. Since the actual CNTs are stable in KOH, these results provide insight regarding the electrochemical stability of the oxidized Inconel and of the catalyst particles that anchor the CNTs at their base. Degradation in the capacitance could then be largely attributed to corrosion at sites where the CNTs are anchored, and subsequent CNT detachment from the electrodes. The samples were tested with 10000 CV cycles at a 50 mV/s sweep rate. Fig. 5 A shows the CV curves of CNTs on Inconel at cycle 1 and at cycle 10000 where only minimal degradation is observed. Fig. 5B shows the CV curves for GC and for a VN/GC film tested at a 100 mV/s sweep rate. According to the CV data the specific capacitance of the VN on GC was 84.3 F/g. In the case of exposed GC there is a clear redox peak around -0.25 V. This is indicative of two-electron reduction of oxygen to H02<" 43> Naturally, this redox peak is detected in all CV curves, albeit at slightly different potentials due to the different nature of the electrode surface, GC, CNT, and VN. The CV curve of VN/GC shows a broad relatively symmetric redox peak centered at -0.61 V during charging and at -0.63 V during discharging. An additional smaller peak is observed at -0.94 V during charging, and at -0.98 V during discharging. These features are unique to the samples with VN, and can be related to redox processes on the VN surface.
[0072] The CV curve of CNTs on Inconel/GC, scanned at 20 mV/s, is shown in Fig. 5C. After the CV measurement, the carbon nanotubes were removed, and a CV test of the remnant CNT growth catalyst particles on Inconel/GC was performed under identical conditions. This curve is also shown in Fig. 5C. It can be seen that the current density of catalyst/Inconel/GC is much lower than that of CNT/Inconel/GC, and the peak positions in these two curves are different. Thus we can safely conclude that the remnant catalyst particles have a negligible contribution to the overall electrochemical supercapacitance. As was already demonstrated, the carbon nanotubes are thoroughly covered by the VN. Since the CNTs are not in contact with the electrolyte, their capacitive contribution is also zero (apart from acting as a support for the VN). Similarly, any catalyst particles that wound up giving tip growth would be fully encapsulated by the VN. Fig. 5D shows the CV data for the CNT/Inconel/GC vs VN/CNT/Inconel/GC, at 100 mV/s. The geometrical area normalized electrical double layer capacitance for CNT/Inconel is 0.001216 F/cm<2>.The CNTs grown on Inconel were too light to accurately ascertain their mass per geometrical squared centimeter (and hence to obtain specific capacitance per mass). The geometrical area normalized capacitance of the VN/CNT/Inconel is 0.01022 F/cm<2>. The specific capacitance of VN/CNT/Inconel is 276 F/g. Since the VN is conformal to the CNTs, the electrochemically accessible surface areas are similar for both types of structures. Therefore the electric double layer contribution for the VN/CNT/Inconel can be estimated as 12% of the total capacitance.
[0073] Fig. 6A shows the cyclic voltammetry diagrams of VN/ CNT/Inconel/GC nanostructures. The potential range of -1.0 to 0.06 V is sampled at different scan rates. There is a slight shift of peak positions due to the voltage drop caused by electric resistance at high sweep rates. Otherwise, in the sweep rate range of 20 mV to 1000 mV/s, the shape of all the curves is quite similar. Fig. 6B gives the relationship between the specific capacitance and the scan rate, in the case of VN/CNT/Inconel/ GC, using three different nitride loadings. The specific capacitance, in F/g, degrades with increasing material loading. This is an expected result: with increasing loading the films get thicker, while their total surface area still follows that of the underlying CNT array. Since pseudocapacitance arises from charge transfer reactions at the materials surface, thicker VN films essentially waste material. For each mass loading, when the sweep rate was increased from 20 to 1000 mV/s, the specific capacitances decreased by less than 20%. Our material shows superior rate capability over the early studies on and comparable performance to Glushenkov et al. and Zhou et al.
[0074] It should be mentioned here that at low scan rate (e.g., 20 mV/s as shown in Figs. 12A and 12B) the cathodic current is higher than anodic current, although the differences are much smaller than for the bare GC and CNT electrodes shown in Figs. 5A-5E. These results show that the oxygen reduction reaction has much slower kinetics and is of much smaller influence for the VN functionalized electrodes compared to bare carbons and will have only very minor influence on the calculated value for the specific capacitance of our VN coated CNT arrays. Our results are in good agreement with the results of a recent publication by Zhou et al, who indicated that the CV of VN nanoparticles is also far from "symmetric" with respect to zero current at low scan rates. As shown in Figs. 5B and 5D, VN/GC and VN/CNT/Inconel/GC exhibit excellent reversibilities at 100 mV/s, the CV curve is symmetric with respect to zero current, indicating a dominant fast and reversible redox reaction. At 20 and 100 mV/s, the capacitances of VN/CNT/Inconel/CNTs are 289 and 276 F/g, respectively.
[0075] The charge-discharge cycling test is considered to be an alternative and arguably better quantitative method to evaluate the supercapacitive nature of an electrode material as opposed to the CV measurement. The supercapacitive behavior of VN/CNT/Inconel/GC is further analyzed by constant current charge-discharge cycling, with current densities from 0.3 mA/cm (i.e., 8 A/g) to 2 mA/cm (i.e., 54 A/g). These results are shown in Fig. 6A. As shown in the figure, all the charge- discharge curves possess close to an ideal triangular shape. The variation of the specific capacitance value deduced by charge-discharge curves at different current densities is shown in Fig. 6D. The specific capacitance decreases gradually from 295 to 214 F/g as the current density increases from 0.3 to 2 mA/cm . This result shows the same trends as the CV measurements. In addition, it can be seen from Fig. 6C that the charge time and discharge time are different during galvanostatic charge and discharge at low current density. At 0.3 mA/cm (i.e., 8 A/g), the charge time and discharge time are 24.8 and 39.5 s, respectively. This difference arises from some slow and irreversible reactions, probably involving oxidation of the active material although more work will be needed to conclusively establish the root cause of such reactions. Glushenkov et al. encountered a similar problem, who indicated that the VN electrodes fail to charge and discharge in 1 M KOH at low current densities. However at 0.7 mA/cm , the charge time and discharge time are 11.1 and 11.5 s, respectively, which are quite close. At higher current density, the charge time is almost the same as the discharge time, suggesting a good reversibility at high charging-discharging current density.
[0076] Although there are some slow and irreversible reactions at low scan rates and low charging-discharging current density, for supercapacitor applications, high charge-discharge rate is very important. Therefore we believe the VN/CNT/Inconel composite is a promising material for fast charge-discharge applications.
[0077] Figs. 7A and 7B show the Nyquist plots of the CNT/Inconel/GC and VN/CNT/Inconel/GC electrodes in 1 M KOH, as well as the simulated curves. The spectra were fitted to the equivalent electronic circuit shown in Fig. 14. The calculated equivalent series resistance (ESR) and charge-transfer resistance (Rct) of CNT/Inconel/GC are 2.76 and 0.11 [Omega], respectively. The 0.11 [Omega] for the CNT array probably corresponds to the charge transfer resistance of the oxygen reduction reaction, which, as was already mentioned, is much faster on the carbons than on VN. After coating with VN, the ESR and R^ are 2.94 [Omega] and 0.28 [Omega], respectively. Our value for R;t is almost a factor of 10 lower than that reported by Zhou et al., which shows how the CNT array improves the rate capability compared to powder electrodes. Low charge-transfer resistance helps to achieve good rate capability because cycling stability is an important property of a supercapacitor, the VN/CNT/Inconel/ Inconel electrode was subjected to a long-term cycling test. The results are shown in Fig. 8A. After 600 CV cycles at a scan rate of 50 mV/s, the specific capacitance values remain at 64%, 63%, and 64% of the original values for the 0.037, 0.092, and 0.135 glc VN films, respectively. To further understand the electrochemical stability of VN in alkaline solution, SEM analysis was carried out on the pre- and post-electrochemically cycled VN/CNT/Inconel/Inconel samples. Figs. 8B and 8C show the SEM micrographs of the pre-cycled and the post-cycled (250 cycles) materials. The SEM micrographs reveal that no substantial changes in the VN nanocrystallites' morphology or size are incurred by the electrochemical cycling.
[0078] X-ray photoelectron spectroscopy (XPS) analysis of the post-cycled materials was carried out on 160 nm VN/Si. These planar samples received the same 250 electrochemical cycles. XPS spectra of the VN thin films before and after 250 electrochemical cycles are shown in Figs. 9A- 9D. Besides the experimental peaks (black), the figures also show the model predictions of individual and summation peaks and background fits (lower lines). The peaks were identified using the aforementioned NIST XPS database. Fig. 9A shows the V peaks and O peaks obtained from the as- synthesized film, which was stored at ambient conditions for approximately 24 h prior to analysis. Both the V 2p and V 2p appear to be a sum of the spectral lines from several valence states of vanadium. The V 2p is fitted with three peaks. The peak at 513.75 eV can be attributed to vanadium 3+ in the vanadium nitride structure. The peaks at 514.7 and 516.8 eV most likely correspond to V 5+(V203) and V (V205) oxidation states of vanadium, respectively. We can therefore conclude that vanadium oxide is present in multiple stoichiometrics. The O Is peak at 530.2 eV is also shown. In the V205 structure the O Is spectral line is in the 529.8-530.6 eV range. In the V204, V203, and V02 structures, the O Is spectral peaks are at 530.4, 530.3, and 530.0 eV, respectively. Since these peaks are quite close in energy, it is difficult to ascribe a particular oxidation state from the oxygen peak data. Fig. 9B shows the portion of the spectrum containing the nitrogen peak, before cycling. The nitrogen peak, which is at 397.3 eV, is in the position expected for a metal nitride (N spectral line Is: 396.8-398.9 eV). Since both the oxide and the nitride peaks are present in the as-synthesized sample, we can conclude that even prior to electrochemical cycling the VN crystals are at least partially oxidized. Most likely the coverage is uniform, with the nanometer-scale thickness of the mixed oxides still enabling detection of a signal from the underlying VN.
[0079] Figs. 9C and 9D show XPS results for the post-electrochemically cycled VN films. The V curves are different from those of the pre-cycled films. The V 2p line shifts to 515.1 eV (V ) and 517.1 eV (V ), with an increase in intensity that indicates a change in the relative amounts of each oxide during electrochemical cycling. The underlying nitride, at 397.4 eV, is still clearly present after the cycling. The nitrogen peak N Is, which is at 397.4 eV, is still clearly present after the cycling with a minor shift. It should be mentioned that the nitrogen peak is also in the position expected for oxidized nitrogen species like N03<"> (N Is around 397.4 eV). As discussed above, the nitrogen peak N Is at 397.3 eV can be definitely attributed to the underlying VN prior to the cycling, but because of the strong overlap with the nitrate peak position, we cannot entirely exclude formation of nitrate species in the cycled material.
[0080] The possible charge-storage mechanism of VN has been studied by several groups. Choi et al. discussed in detail the surface reactions of the VN nanocrystals in aqueous electrolyte of high pH (-14). The authors proposed that in the presence of OH<"> ions an equilibrium reaction could occur on the nitride or oxynitride surfaces. The reaction is a combination of electrochemical double layer formation and a Faradaic redox reaction VNxOy + OH <-> VNxOy¦¦OH + VNxOy- OH + e (3) where the VNxOy¦¦OH<"> represents the electrical double layer formed by the hydroxide ions physisorbed on nonspecific sites, whereas VNxOy- OH represents chemisorption in the faradaic redox reaction. Impedance measurements in the study by Zhou et al. revealed that a charge transfer reaction is indeed happening. More precise determination of the reaction mechanisms would be possible using spectroscopic techniques such as FTIR and XPS in situ, but that would require a specialized electrochemical setup. Theoretically, a lot of information about the redox processes could be gained from detailed studies of CV peak positions and currents as a function of scan rate, electrolyte concentration, etc., but this is outside the scope of the present paper.
[0081 ] The maximum capacitance obtained in this study is different from the values reported by Zhou et al. (161 F/g), Glushenkov et al. (186 F/g), and Choi et al. (1340 F/g). The extraordinary capacitance 1340 F/g reported by Choi et al. was achieved at a very slow scan rate of 2 mV/s. Subsequent studies by the same group reported more modest capacitance values for nanocrystalline VN synthesized nominally in the same manner. The best performance that the authors could obtain was 160 F/g. These study-to-study variations may be related to the differences in both the surface area-to-volume ratios and in the surface state of each material. Glushenkov and coworkers argued that for the VN system the composition structure of the surface oxide layer has a tremendous influence on the capacitive performance. This is a reasonable conclusion since the vanadium oxide phases are generally poor electrical conductors. Their presence in sufficient thicknesses would degrade the capacitance through resistive losses. Furthermore, the various valence vanadium oxides are expected to have differing electrical conductivities. For a given oxide, its electrical conductivity is generally expected to improve with increasing oxygen deficiency.
[0082] Choi et al. proposed that the cycling performance was determined by several parameters like crystallite size, oxide layer, material loading, electrolyte concentration, and potential 13
window. Our cycling results support the above argument regarding the importance of surface structure. The observed change in the oxide may be the factor leading to the electrochemical cycling- induced capacitance loss. There is discernible difference between the as-synthesized (289 F/g) and the post-cycled (185 F/g) samples in the composition and/or relative amount of the various mixed oxide phases on the material's surface. If during cycling the oxide layers grew thicker and/or became less electrically conducting due to a change in stoichiometry or the relative volume fractions of each phase, the capacitance would degrade. It is unlikely that the contribution from electrical double layer capacitance varies due to cycling. As we showed in Fig. 5 A, the electrical double layer capacitance of CNTs on Inconel does not evolve. Since the VN conformally covers the CNTs, it is not expected that the total electrochemically accessible surface area will be drastically different in the case of VN/CNT/Inconel versus CNT/Inconel. During cycling, there is little evidence of the VN/CNT/Inconel nanostructures physically degrading in a way as to reduce that area.
[0083] It has been previously argued that poor rate capability is a key disadvantage of utilizing VN-based electrodes for electrochemical capacitor applications. For example Choi et al. found that the capacitance of highly porous VN powders decreased dramatically (more than 50%) when the sweep rate was increased from 2 to 100 mV/s. Such rate -dependent loss is attributed to the morphology of the as-prepared VN nanocrystallites, which normally consist of weakly interconnected nanoparticles. Ohmic losses due to these interparticle contacts would increase linearly with current density, resulting in a sharp drop in capacitance at higher currents and/or scan rates. In this study, conversely, the capacitance decreases only very slowly with increasing current density at higher currents and scan rates. For an electrochemical reaction involving charge transfer, the overpotential increases only logarithmically with current according to the Butler-Volmer equation. Hence, in our case, the behavior of the material seems to be determined entirely by the charge- transfer kinetics of the relevant redox processes. The fact that the VN nanocrystallites are in the form of a continuous shell around a conducting core that directly anchored to a conductive substrate eliminates ohmic losses, resulting in excellent rate capability.
[0084] Among previous studies, only Glushenkov et al. and Zhou et al. obtained rate capability comparable to ours albeit at lower specific capacitance. Variation in the other studies involved a slightly different ratio between VN, Super P conductive additive, and poly(vinylidene fluoride). This suggests that the final performance of the supercapacitor is very sensitive to small variations in the electrode preparation process. As mentioned in the Introduction, our approach has the big advantage of being a "drop-in" for an industrial thin film fabrication environment and involves no handling or mixing of the active materials. Additionally, CNTs grown on conductive electrodes are just one of many highly electrically conductive, high surface area substrates, such as carbon paper, carbon cloth, CNT paper, or graphene that VN can be vapor-deposited onto. As we demonstrated in this study when these materials consist of a three-dimensional open network, uniform coverage by VN would be achievable given a sufficient spacing of the features.
[0085] This study demonstrates a direct synthesis methodology to achieve three-dimensional arrays of carbon nanotube-vanadium nitride nanostructures. Since the resultant arrays come firmly anchored to the electrically conductive electrodes, the developed methodology is highly advantageous for electrochemical capacitor applications. The skeleton of multiwalled carbon nanotubes was grown via chemical vapor deposition using an Inconel-based support film. The outer layer of vanadium nitride was subsequently synthesized using physical vapor deposition (reactive sputtering). Transmission electron microscopy analysis indicates that the vanadium nitride layer consists of a shell of interconnected sub-50-nm scale nanocrystallites that conformally cover the nanotubes.
[0086] For the three-dimensional arrays grown using the Inconel-based support film, a specific capacitance of 289 F/gwas achieved in 1 M KOH at a scan rate of 20 mV/s. Cyclic voltammetry measurements indicated that the nitride-nanotube composites exhibit superb high rate capability (up to 1000 mV/s) and a long cycle life. At a sweep rate of 1000 mV/sthe specific capacitances decreased by less than 20%. After 600 cycles the specific capacitance remains at roughly 64% of the original value. Electrochemical analysis performed by constant current charge-discharge cycling, in the range of 0.3-2 mA/cm , demonstrates close to an ideal triangular shape of voltage versus time behavior.
[0087] X-ray photoelectron spectroscopy analysis was performed on the as-synthesized and the post-electrochemically cycled blanket films of vanadium nitride. Vanadium oxide peaks of several valences and vanadium nitride peaks were present in the as-synthesized and in the post-cycled samples. Hence we conclude that even prior to electrochemical cycling the nitride crystals are at least partially oxidized. Scanning electron microscopy of the post-electrochemically cycled samples revealed minimal morphological changes of the nanocrystalline nitride.
[0088] We also examined the feasibility of creating similar nanostructures using a much denser array of carbon nanotubes grown on "conventional" A1203 support films. The high CNT density did not allow for conformal coating by the nitride.
[0089] Raman spectroscopy is widely used for examining the structure of carbon nanotubes. In this study, first-order Raman spectroscopy (532 nm) was employed on the 3D CNT array grown on Inconel/GC. Raman measurements were carried out with a confocal microprobe Raman system (Thermo Nicolet Almega XR Raman Microscope) using an air-cooled charge-coupled device (CCD) and a green diode laser operating at 532 nm. A 2000 groove/mm holographic reflection grating was used. The spatial resolution, confocal resolution and spectral resolution are 1 um, 2 [mu][iota][eta] and 2 cm<"1>, respectively. The Raman -scattered light was collected normal to the sample surface where at least five positions were randomly chosen on each sample.
[0090] As shown in Fig. 10, the Raman spectrum normalized to D peak demonstrates two Raman bands, the G band and the D band. The G band, at -1571 cm<"1>, is related to the graphite tangential E2g Raman active mode, which is due to the stretching vibration of 5/J> -hybridized carbon. The D band, at -1343 cm<"1>, is a breathing mode of Aig symmetry, which only becomes active in the presence of disorder and defects. Accordingly, the intensity ratio of IG I ID can provide semiquantitative information about the CNT quality. The average value of calculated IG I ID is 1.60, indicating good crystallinity of the as-prepared CNT array, with low levels of disordered carbonaceous impurities.
[0091 ] Fig. 11 A shows HRTEM image of a CNT grown on Inconel. Fig. 1 IB is an HRTEM image of a CNT grown on A1203. Clearly, the diameter is much smaller and the resulting packing density much higher (see Figure 4) for the nanotubes grown on A1203. Fig. 12A shows the CV curves of GC and of VN/GC at a slow scan rate of 20 mV/s. Fig. 12B shows the CV data for CNT/Inconel and for VN/CNT/Inconel at a slow scan rate of 20 mV/s. It is clear that the cathodic current is higher than the anodic current. These results suggest that at low scan rates there are some slow and irreversible reactions that only occur in the cathodic region. Fig. 13A shows the CV results (scan rate of 50 mV/s) for the case of the CNT/A1203. After 10000 cycles the specific capacitance (Csp) is approximately 1/3 of the original value. When the GC disk was removed from the solution after testing, we observed that a significant amount of the CNTs had peeled off the electrode. This electrochemical instability of the CNT/A1203 system can be attributed to the well-known amphoteric property of A1203, which is unstable in both highly acid and highly alkaline environments. Fig. 13B shows the CV results for CNT/A1203 and for VN/CNT/A1203 at a sweep rate of 20 mV/s. There is only a minimal difference between the two curves: with the presence of VN film the current density increases only marginally, while the shape of the CV remains similar to that of the uncoated CNT array. A small symmetrical redox peak is observed at -0.65 V. The total capacitance of the VN-covered array is the sum of the capacitance due to double layer charging and the electrochemical pseudocapacitance of the VN overlayer. Furthermore since the VN covers a relatively small fraction of the CNTs, one can assume that the mass available for double layer charging is about the same in each case. The pseudocapacitive contribution due to the VN surface film can then be obtained from the difference of the two CV curves. This value is calculated to be 131 F/g. This is roughly 50% higher than that of VN/GC, indicating only limited improved utilization of the nitride.
[0092] The above results demonstrate that although such dense CNT arrays have large surface areas and are useful as electrochemical double layer capacitors, they are difficult to functionalize via PVD. Interestingly, a similar limitation was reported by Zhang et al.<17> for the case of electrodeposition of manganese oxide on a dense array of parallel CNTs. The authors attributed the inability to fully coat the CNTs to limited ion access to the CNT surface and electrical shielding effects.
[0093] The measured impedance spectra were analyzed using the complex nonlinear least- squares (CNLS) fitting method on the basis of the equivalent circuit,<2> which is given in Figure 14. At very high frequencies, the intercept at real part ([Zeta]') represents a combined resistance of ionic resistance of electrolyte, intrinsic resistance of substrate, and contact resistance at the active material/current collector interface ( Re or ESR ). Rct represents the charge-transfer resistance caused by the faradaic reactions and Cdi is the double-layer capacitance on the electrode surface. Zw corresponds to the Warburg resistance, which is a result of the frequency dependence of ion diffusion/ transport in the electrolyte to the electrode surface. CL is the limit capacitance.
[0094] (1) Simon, P.; Gogotsi, Y. Nat. Mater. 2008, 7, 845.
[0095] (2) Conway, B. E. In Electrochemical Supercapacitors, Scientific Fundamentals and Technological Applications; Kluwer Academic/Plenum Press: New York, 1999.
[0096] (3) Algharaibeh, Z.; Pickup, P. G. Electrochem. Commun. 2011, 13, 147.
[0097] (4) Cormie, A.; Cross, A.; Hollenkamp, A. F.; Donne, S. W. Electrochim. Acta 2010, 25, 7470.
[0098] (5) Algharaibeh, Z.; Liu, X. R.; Pickup, P. G. J. Power Sources 2009, 187, 640.
[0099] (6) Jacob, G. M.; Yang, Q. M.; Zhitomirsky, I. J. Appl. Electrochem. 2009, 39, 2579.
[00100] (7) Li, J.; Yang, Q. M.; Zhitomirsky, I. Nanoscale Res. Lett. 2010, 5, 12.
[00101] (8) Zhang, J.; Ma, J.; Zhang, L. L.; Guo, P.; Jiang, J.; Zhao, X. S.J. Phys. Chem. C 2010, 114, 13608. [00102] (9) Levi, M. D.; Salitra, G.; Levi, N.; Aurbach, D.; Maier, J. Nat.Mater. 2009, 8, 872.
[00103] (10) Ardilla, D.; Estrany, F.; Al!eman, F. J. Phys. Chem. C 2011,115, 8430.
[00104] (11) Liu, T. C; Pell, W. G.; Conway, B. E. J. Electrochem. Soc. 1998,145, 1882.
[00105] (12) Choi, D. W.; Kumta, P. N. Electrochem. Solid-State Lett. 2005,8, A418.
[00106] (13) Choi, D. W.; Blomgren, G. E.; Kumta, P. N. Adv. Mater. 2006, 18, 1178.
[00107] (14) Glushenkov, A. M.; Hulicova-Jurcakova, D.; Llewellyn, D.; Lu,G. Q.; Chen, Y. Chem. Mater. 2010, 22, 914.
[00108] (15) Zhou, X. P.; Chen, H. Y.; Shu, D.; He, C; Nan, J. M. J. Phys.Chem. solids 2009, 70, 495.
[00109] (16) Hou, Y.; Cheng, Y.;Hobson, T.; Liu, J. Nano Lett. 2010, 10, 2727.
[001 10] (17) Zhang, H.; Cao, G. P.; Wang, Z. Y.; Yang, Y. S.; Shi, Z. J.; Gu,Z. N. Nano Lett. 2008, 8, 2664.
[001 1 1] (18) Liu, J. W.; Essner, J.; Li, J. Chem. Mater. 2010, 22, 5022.
[001 12] (19) Jiang, Y. Q.; Wang, P. B.; Zhang, J.; Li, W.; Lin, L. W. Proceedingsof 23th IEEE Micro Electro Mechanical Systems Conference, Hong Kong,2010; p 1171.
[001 13] (20) Hanumantha P. J.; Kumta, P. N. ECS Meeting of The Electrochemical Society, 218th 2010; Abstract #124
[001 14] (21) Jung, Y. J.; Wei, B. Q.; Vajtai, R; Ajayan, P. M. Nano Lett. 2003,3, 561-564.
[001 15] (22) Takagi, D.; Homma, Y.; Hibino, H.; Suzuki, S.; Kobayashi, Y.Nano Lett. 2006, 6, 2642-2645.
[001 16] (23) Mattevi, C; Wirth, C. T.; Hofmann, S.; Blume, R; Cantoro, M.;Ducati, C; Cepek, C; Knop-Gericke, A.; Milne, S.; Castellarin-Cudia,C; Dolafi, S.; Goldoni, A.; Schloegl, R.; Robertson, J. J. Phys. Chem. C 2008, 112, 12207.
[001 17] (24) Hiraoka, T.; Yamada, T.; Hata, K.; Futaba, D. N.; Kurachi, H.;Uemura, S.; Yumura, M.; Iijima, S. J. Am. Chem. Soc. 2006, 128, 13338.
[001 18] (25) Zhang, H.; Cao, G. P.; Wang, Z. Y.; Yang, Y. S.; Gu, Z. N. Carbon 2008, 46, 818.
[001 19] (26) Futaba, D. N.; Hata, K.; Yamada, T.; Hiraoka, T.; Hayamizu, Y.; Kakudate, Y.; Tanaike, O.; Hatori, H.; Yumura, M.; Iijima, S. Nat. Mater. 2006, 5, 987.
[00120] (27) Talapatra, S.; Kar, S.; Pal, S. K.; Vajtai, R; Ci, L.; Victor, P.; Shaijumon, M. M.; Kaur, S.; Nalamasu, O.; Ajayan, P. M. Nat. Nanotechnol. 2006, 1, 112. [00121] (28) Pal, S. K.; Talapatra, S.; Kar, S.; Ci, L.; Vajtai, R.; Borca-Tasciuc, T.; Schadler, L. S.; Ajayan, P. M. Nanotechnology 2008, 19, 045610.
[00122] (29) Bult, J. B.; Sawyer, W. G.; Ajayan, P. M.; Schadler, L. S. nanotechnology 2009, 20, 085302.
[00123] (30) Pal, S. K.; Kar, S.; Lastella, S.; Kumar, A.; Vajtai, R.; Talapatra, S.; Borca-Tasciuc, T.; Ajayan, P. M. Carbon 2010, 48, 844.
[00124] (31) Donders, M. E.; Knoops, H. C. M.; van de Sanden, M. C. M.; Kessels, W. M. M.; Notten, P. H. L. J. Electrochem. Soc. 2011, 158, G92.
[00125] (32) Profijt, H. B.; Kudlacek, P.; van de Sanden, M. C. M.; Kessels, W. M. M. J. Electrochem. Soc. 2011, 158, G88.
[00126] (33) Hata, K.; Futaba, D. N.; Mizuno, K.; Namai, T.; Yumura, M.; Iijima, S. Science 2004, 306, 1362.
[00127] (34) Yasuda, S.; Futaba, D. N.; Yamada, T.; Satou, J.; Shibuya, A.; Takai, H.; Arakawa, K.; Yumura, M.; Hata, K. ACS Nano 2009, 3, 4164.
[00128] (35) Dai, L.; Wang, P.; Bosnick, K. J. Vac. Sci. Technol., A 2009, 27 (4), 1071.
[00129] (36) Peron, J.; Edwards, D.; Haldanea, M.; Luo, X.; Zhang, Y.; Holdcroft, S.; Shi, Z. J. Power Sources 2011, 196, 179.
[00130] (37) Liu, J.; Corny, T. E.; Song, X.; Doeff, M. M.; Richardson, T. J. Energy Environ. Sci. 2011, 4, 885.
[00131] (38) Zasadzinski, J.; Vaglio, R.; Rubino, G.; Gray, K. E.; Russo, M. Phys. Rev. B 1985, 32, 2929.
[00132] (39) Sekerka, R. F. Cryst. Res. Technol. 2005, 40, 291.
[00133] (40) Luber, E. J.; Olsen, B. C; Ophus, C; Mitlin, D. Phys. Rev. B 2010, 82, 085407.
[00134] (41) Comas, M.; Caussat, B.; Ayral, A.; Durand, J.; Kihn, Y.; Kalck, P.; Serp, P. Chem. Eng. Sci. 2003, 58, 4475.
[00135] (42) Kim, S. M.; Pint, C. L.; Amama, P. B.; Zakharov, D. N.; Hauge, R. H.; Maruyama, B.; Stach, E. A. J. Phys. Chem. Lett. 2010, 1, 918.
[00136] (43) Qu, D. Carbon 2007, 45, 1296.
[00137] (44) Denuault, G.; Sosna, M.; Williams, K. J. In Handbook of Electrochemistry; Elsevier: Amsterdam and Boston, 2007; Chapter 11.