Journal of Power Sources 277 (2015) 59e63
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Short communication
Ultra-fast aqueous Li-ion redox energy storage from vanadium oxide-carbon nanotube yarn electrodes Jesse Smithyman*, Quyet H. Do, Changchun Zeng, Zhiyong Liang High-Performance Materials Institute, Florida State University, 2005 Levy Ave, Tallahassee, FL 32310, USA
h i g h l i g h t s Fabrication of yarn electrodes for high-performance electrochemical energy storage. Carbon nanotube yarn used as deposition substrate for vanadium oxide. Supercritical fluid deposition maintains open pore structure and large surface area. Reversible Liþ intercalation at scan rates as high as 5 V s1. Charge storage decreased 10% after 100 cycles in large potential window of 1.2 V.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 August 2014 Received in revised form 21 October 2014 Accepted 21 November 2014 Available online 22 November 2014
Half-cell electrochemical characterizations were conducted on carbon nanotube-vanadium oxide (CNTVOx) yarn electrodes in an 8 M LiCl aqueous electrolyte. A supercritical fluid deposition and in-situ oxidation process was utilized to deposit nanoscale coatings of vanadium oxide on carbon nanotube (CNT) surfaces throughout the porous structure of CNT yarns. The high surface area, interconnected pore structure and high electrical conductivity of the CNT yarn enabled extraordinary rate capabilities from the high capacity Li/VOx system. High-rate cyclic voltammetry scans, requiring current densities of hundreds of amperes per gram of electrode mass, produced rectangular voltammograms with distinguishable redox peaks from Li-ion intercalation/deintercalation. Capacitances of over 150 F g1 were achieved at a scan rate of 5 V s1 over a 1.2 V potential window resulting in an energy density of >32 Wh kg1 (>30 Wh L1) for the yarn electrode. The charge storage also showed good reversibility when cycled over this large potential window, maintaining 90% of the capacitance after 100 cycles at a scan rate of 2 V s1. Electrochemical impedance spectroscopy shows the frequency dependent behavior is distinctly lacking of the characteristic responses from the rate-limiting processes associated with faradaic charge storage in VOx. © 2014 Elsevier B.V. All rights reserved.
Keywords: Carbon nanotube Vanadium oxide Yarn electrode Aqueous Li-ion Energy storage Supercritical fluid deposition
1. Introduction Vanadium oxides are promising material for electrochemical energy storage and are investigated as the active material in electrochemical capacitor and metal-ion (e.g. Lithium, Sodium) batteries in both aqueous and non-aqueous electrolytes. It is a particularly interesting candidate for aqueous energy storage such as Li-ion supercapacitors and aqueous rechargeable Li-ion batteries (ARLB), which are gaining attention recently due to the safety and toxicity concerns with non-aqueous Li-ion batteries. Vanadium
* Corresponding author. E-mail address:
[email protected] (J. Smithyman). http://dx.doi.org/10.1016/j.jpowsour.2014.11.095 0378-7753/© 2014 Elsevier B.V. All rights reserved.
oxides are attractive for ARLBs due to the ability to intercalate large amounts of cations into the structure and the low redox potentials, enabling large energy densities [1,2]. Progress has been made on improving the cycle life of vanadium oxides in aqueous media through the investigation of different phases/morphologies and electrolyte selection [3e7]. However, VOx based electrodes still generally suffer from low rate capabilities, minimizing the power density achievable. This is a significant drawback as high power is one of the main advantages of aqueous batteries and electrochemical capacitors. In a recent review of aqueous rechargeable Li-ion batteries with “superfast charging”, the highest current densities reported are on the order of several amperes per gram, with the majority of experiments conducted in the mA g1 range [8]. In this report, we
60
J. Smithyman et al. / Journal of Power Sources 277 (2015) 59e63
show that the Li/VOx redox reactions are not only accessible, but capable of providing significant energy storage at current densities of several hundred amperes per gram. Composite electrodes were fabricated using carbon nanotube (CNT) yarns as a threedimensional substrate for the supercritical fluid deposition (SFD) of vanadium oxides. The zero surface tension and extremely low viscosity of supercritical fluids allows for rapid penetration and conformal deposition of dissolved species throughout the porous nanotube network while maintaining the open pore structure and large surface area provided by the yarn. Atomic layer deposition (ALD) has been used on CNT sheets to create VOx/CNT electrodes with similar microstructures [9], however ALD requires hundreds of deposition cycles due to the low gas-phase solubility of the precursors. On the other hand, this report demonstrates that a single SFD cycle deposits enough VOx to provide an order of magnitude increase in the energy storage capabilities of CNT yarn. In addition to the impressive electrochemical performance, the unique yarn architecture makes these results potentially attractive for niche energy storage applications such as wearable and microscale devices. The porous CNT yarn functions as both the structural support and the primary electron conductor of the composite electrode while providing a high surface area deposition surface for the active redox material. This multifunctional role eliminates the need for metallic substrates commonly used in the fabrication process of traditional particle based electrodes. Such part integration strategies are an attractive approach to reducing the volume/ mass of energy storage devices for applications with strict size and weight restrictions. 2. Experimental 2.1. Carbon nanotube yarn Carbon nanotube yarn (CTex™) were purchased from Nanocomp Technologies, Inc. and underwent a light purification process to remove non-fullerene organics (e.g. amorphous carbons, polymers, or other residuals from the manufacturing process) and exposed metal catalyst particles. The iron catalyst can be removed through a hydrochloric acid bath (12e16 h at room temperature) if exposed and accessible to the liquid solution. However, a significant portion of the catalyst particles are typically protected by a carbon coating, requiring physical or chemical oxidation to “crack” the protective coatings and allow accesses to the metal surface. However full removal of the catalyst is difficult and may require several iterations of the oxidation/acid bath process or more severe processing conditions which can degrade the structural integrity of the CNT networks. Therefore, no oxidation process was conducted so that any unexposed iron that was not removed during the HCl bath continued to remain inaccessible and unable to contribute to the electrochemical response of the electrode. The yarn has an average diameter of 50 microns, BET specific surface area of 200 m2 g1 and linear density of ~1.43 mg m1. 2.2. Supercritical fluid-deposition of vanadium oxide Vanadium oxide was deposited using a two-stage supercritical fluid-deposition process with in-situ oxidation of the vanadium(III) acetylacetonate, V(acac)3, precursor [10]. V(acac)3 was dissolved to saturation in a mixture of high-purity CO2 and O2 under temperature and pressure conditions such that CO2 was in the supercritical fluid state but the precursor was not yet oxidized (~70 C and 200 bar). The solution was then transferred from the saturation chamber into a reaction chamber heated to ~150 C containing the CNT yarn substrate where simultaneous adsorption and oxidation of the precursor occurs on CNT surfaces. After deposition, the
sample underwent a 5 h calcination process in air at 250 C. The final composite yarn electrodes contained ~27 wt.% VOx. Scanning electron microscope images of the CNT yarn are shown in Fig. 1. 2.3. Electrochemical characterization Single yarn filaments of ~1.5 cm length were attached to 3 mm wide Nickel tabs with silver paste. Approximately 1 cm of the yarn was attached to the tab, leaving ~0.5 cm exposed length extending from the tab as the “active region”. A vinyl-based sealant (Performix Liquid Tape) was used to cover the silver paste to further secure the connection, provide insulation and prevent exposure to the electrolyte. An image of an example electrode is provided in the supplementary information. The active region was fully submerged in 8 M LiCl electrolyte (pH 6e7). A saturated calomel electrode (SCE) was used as the reference electrode and a glass frit isolated platinum wire as the counter electrode. Nitrogen gas was bubbled in the electrolyte for several minutes prior to testing each sample and a N2 blanket flow on top of the electrolyte continued during testing. All results are based on the total electrode mass of the active region. 3. Results and discussion Nitrogen gas adsorption measurements show that the general textural characteristics of the porous CNT networks (i.e. pore structure, surface area) remain, albeit reduced, after oxide deposition. Fig. 2 shows the BJH pore size distribution (PSD) for the CNT yarn before and after deposition of (27 wt.%) VOx. The PSD of both samples show the presence of a broad range of pore sizes, peaking at each end of the spectrum. The broad yet bimodal distributions contain one peak around 2e3 nm and a secondary peak around 50e70 nm. The VOx-CNT yarn electrodes possessed a BET specific surface area (SSA) of 78 m2 g1. This electrode surface area is comparable to that of high-surface area vanadium oxide powders (i.e. before electrode fabrication) obtained from aerogel synthesis process [11]. The cyclic voltammetry (CV) results at scan rates of 100, 250 and 500 mV s1 are shown in Fig. 3. While particularly large differential capacitances are observed around the redox peaks, large integral capacitances are also achieved over the entire 1.2 V potential window. Plotting the CV results in terms of differential capacitance indicates two different behaviors from the enhanced charge storage provided by the VOx. Most obvious is the large redox peaks observed around 0.4 V to 0.6 V, believed to be due to the intercalation of Liþ into the VOx lattice [5]. A gradual broadening and shifting of the peaks is observed as the scan rate is increased and is consistent with the kinetic effects expected for ion insertion/ de-insertion [12,13]. In addition to this diffusion-controlled process of charge storage, there is also a significant contribution of charge storage from nondiffusion-controlled surface redox reactions with the vanadium oxide. This psuedocapacitive charge storage is evident from significant enhancement of differential capacitance (over the CNT yarn prior to deposition) that occurs outside of the intercalation peaks, but shows essentially no variation with scan rate. Fig. 4 shows the integral capacitance at various scan rates for two different VOx-CNT yarn electrode samples. The capacitance reaches 200 F g1 at low scan rates (100 mV s1) and remains above 150 F g1 even at the extremely fast scan rates of 5 V s1. The energy density (E ¼ ½CV2) was calculated based on the 1.2 V potential window of the CV scans and is shown normalized by the electrode mass on the right Y-axis of Fig. 4. The electrode density is close to 1 (0.95 g cm3) and thus the energy expressed volumetrically is only slightly lower than the gravimetric form, reaching a maximum of
J. Smithyman et al. / Journal of Power Sources 277 (2015) 59e63
61
Fig. 1. SEM images showing the yarn morphology at low magnification and the densely packed CNT networks on the surface of the yarn.
here show a strong similarity to those reported by Li et al. from single crystal H2V3O8 (or V3O7$H2O) nanowires in 5 M LiNO3 and 0.001 M LiOH [5]. Both sets of CV results show large redox peaks centered around 0.4 V to 0.6 V vs. SCE with various degrees of smaller secondary peaks occurring at higher potentials. The V:O ratio of 0.375 (V per O) for the V3O7$H2O nanowires is within the range obtained for the VOx-CNT yarn electrodes from energydispersive X-ray spectroscopy (EDS) and electron energy loss spectroscopy (EELS) analysis (Supplementary information). The
Fig. 2. Pore size distribution calculated using the BJH method on the adsorption branch of the N2 isotherm for the CNT yarn before and after deposition of vanadium oxide.
~38 Wh L1 (as demonstrated in the Supplementary information file). This rapid, high energy charge storage also shows good cycle-life as can be seen by the ~10% reduction in capacitance over 100 cycles at a scan rate of 2 V s1, shown in Fig. 5. The CV cures of the 5th, 50th and 100th cycle are shown in the inset of Fig. 5. Even after 100 cycles over a 1.2 V potential window, the redox peaks are still clearly visible. While a wide variety of electrochemical responses from VOx materials can be found in the literature, the CV results presented
Fig. 3. Cyclic voltammetry results of VOx-CNT yarn electrodes at three different scan rates shown in terms of differential capacitance. Results from a CNT yarn electrode before deposition are provided as a baseline (10 mV s1 scan rate).
62
J. Smithyman et al. / Journal of Power Sources 277 (2015) 59e63
Fig. 4. Capacitance and energy density vs. scan rate for two VOx-CNT yarn electrodes.
Fig. 5. Cycle life of CNT-VOx yarn electrode at 2 V s1. The figure inset shows the CV results for the 5th, 50th and 100th cycles.
average V:O ratio from EDS was 0.44 with a large standard deviation of ±0.1535. While the large range supports the notion of a mixed oxide composition, the similarity of the Liþ insertion characteristics observed in the CV results of V3O7$H2O nanowires may indicate the presence of localized crystalline phases with similar structures. Analysis of the chemical and physical composition of the VOx phase is non-trivial due to the unique properties of the composite electrodes. Difficulty in determination of the chemical/physical properties of the deposited VOx material arise due to several factors, including: 1) the nanoscale dimensions of the VOx species, 2) the likelihood of mixed oxides and non-crystalline formations of the vanadium species, and 3) the low-mass density of the CNT yarn substrate. The small amount of VOx present both locally on CNT surfaces and cumulatively over a given length of the sample results in weak spectroscopic signals that often containing appreciable noise or background interference. Furthermore, any lack of distinguishable domains or crystallinity would severely limit the utility of characterization methods such as X-ray diffraction or transmission electron microscopy to obtain information on the physical structure of the material. While characterizations of the chemical, physical and morphological properties of the VOx phases are ongoing, we have decided to report the electrochemical properties of the VOx-CNT yarn
electrodes. The ability to enable redox reactions between VOx/Liþ at current densities of several hundred amperes per gram of total electrode material is quite impressive and to our knowledge unique among the literature. This performance is achievable due to the combination of several features of the composite electrode system that stem from both the CNT yarn substrate and results of the novel oxide deposition process. The mesoporous pore structure of the CNT yarn mitigates the slow liquid phase ion diffusion that occurs in highly tortuous micropore structures. At the same time the large surface area and nanoscale dimensions of the VOx coatings optimize the faradaic surface charge storage processes by maximizing available surface sites and minimizing the transport length through the highly resistive oxide that electrons must travel through in order to participate in the surface reactions. The versatility of the VOx/Liþ redox system is not limited to surface processes and the ability to also store charge via Liþ insertion/intercalation increases the extent of charge storage capable from the electrode. The extremely large surface-to-volume ratio of the VOx active material results in the observed electrochemical performance of capacitance-like characteristics. The porous 3D current collector functionality of the CNT networks attempts to maximize the cumulative amount of these large surface-to-volume ratio VOx phases in a given volume. It is in this way that the multifunctional CNT yarn substrate works synergistically with the nanoscale VOx coatings to enable contributions to the charge storage from either surface redox reactions or Liþ insertion processes at extremely high rates. Based on the relatively low mass percent of VOx in the sample, along with the comparatively large density of vanadium and the abundant surface sites available during deposition, it cannot be ruled out that a significant portion of the VOx species may exist as chemical functional groups attached to CNT surfaces sites as opposed to film or particle like domains. Reversible redox activity from oxygen containing functional groups on CNT surfaces has been shown to have the ability to provide high power, high energy electrodes [14]. Addition of vanadium to the functional groups in such a structure could allow for an increase in the energy density due to the wide range of oxidation states achievable from vanadium, with little to no detrimental effects on the power capabilities. The unique characteristics of the electrode microstructure are also reflected in the results from potentiostatic electrochemical impedance spectroscopy (EIS). Recently, Chen et al. have shown the importance of the mesoporosity in CNT-V2O5 electrodes and the effects of porosity reduction on the frequency-dependent
Fig. 6. Nyquist plot from potentiostatic electrochemical impedance spectroscopy results with equivalent model and associated fitting results. Rs is the solution resistance and the yarn electrode is represented by the constant phase element (CPE).
J. Smithyman et al. / Journal of Power Sources 277 (2015) 59e63
impedance spectrum [15]. The microstructure of the VOx-CNT yarn electrode minimizes the effects of the rate-limiting processes observed in traditional VOx electrodes to such an extent that the characteristic EIS responses of these processes cannot be found in the EIS results of the VOx-CNT yarn electrodes. The Nyquist plots in Fig. 6 for EIS results at 0.2 V and 0.6 V vs. SCE show essentially featureless vertical lines with no semi-circle nor 45 regions characteristic of charge transfer resistances, diffusion or other highly rate-dependent (and thus frequency dependent) processes. To further demonstrate this, an equivalent circuit consisting only of a modified RC circuit is fit to the experimental results with high accuracy. The resistor of the RC circuit describes the electrolyte solution resistance, and the capacitor of the traditional RC circuit is replaced by a constant phase element (CPE) to represent the VOxCNT yarn electrode. A CPE is often used to represent any non-ideal capacitance behavior in, for example, porous electrodes and is equal to a capacitor when the “P” parameter is equal to 1. Fig. 6 provides a drawing of the equivalent circuit associated fitting result from the Rs-CPE equivalent circuit model for the EIS response at each potential. An in depth investigation into the potential-dependent EIS properties of CNT yarn electrodes is underway, but the initial results appear to indicate that at least some of the potentialdependent variation is inherent to the CNT yarn and independent of the presence of VOx. The fit parameters obtained from the equivalent circuit model results are shown in Table 1 for the VOxCNT yarn along with results from CNT-only electrodes (i.e. prior to VOx deposition). The potential-dependent behavior inherent to the CNT yarn is likely due to the potential-dependent injection of charge carries, or “electrochemical doping”, that occurs in the semimetal electronic structure of carbon nanotubes [16]. Further investigations of these phenomena, with more in-depth EIS modeling are ongoing but beyond the scope of the current report. The comparison of the modified simple RC circuit is presented only to further demonstrate that while the addition of VOx provides a significant increase in the extent of charge storage capable from the CNT yarn electrodes, the accompanying ratelimiting effects that typically accompany this increased charge storage are surprisingly absent. In addition to the aforementioned beneficial properties from the electrode microstructure, the unique cylindrical geometry of the yarn electrode may also contribute to the excellent rate capabilities. Studies using supercapacitor electrodes to perform capacitive energy extraction from salinity gradients have shown that the use of an electrode with cylindrical geometry can increase the mass transport by diffusion resulting in faster charging and higher overall power when compared with a flat plate geometry [17,18].
4. Conclusion Supercritical fluid deposition of vanadium oxide on carbon nanotube yarns is shown to produce electrodes capable of accessing the Li/VOx redox couple at extremely high rates with good cycle life in an aqueous electrolyte. The zero surface tension of the Table 1 Parameters obtained from fitting the potentiostatic EIS results to an equivalent circuit consisting of a resistor representing the electrolyte solution resistance, Rs (ohm), in series with a constant phase element (CPE). The error percentage for all parameters shown in the table was 40 Wh kg1 (>38 Wh L1). The energy density remained greater than 30 Wh kg1 (capacitance of 150 F g1) after increasing the scan rate by more than two orders of magnitude to 5 V s1. The charge storage also showed good reversibility, particularly considering the large surface area and large voltage window, retaining 90% capacitance after 100 cycles at 2 V s1. Electrochemical impedance spectroscopy was used to further demonstrate the attractive properties of the VOx-CNT yarn electrodes. Although the deposition of VOx resulted in nearly an order of magnitude increase in capacitance, analysis of the impedance results indicates a distinct lacking of the characteristic features associated with rate limiting processes (e.g. resistance/ diffusion) that typically accompany the faradaic storage of charge in vanadium oxides. Acknowledgment Support provided from the High-Performance Materials Institute, the Florida State Research Foundation GAP Grant, and the NSF SNM award #1344672. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2014.11.095. References [1] J.-Y. Luo, W.-J. Cui, P. He, Y.-Y. Xia, Nat. Chem. 2 (9) (2010) 760e765, http:// dx.doi.org/10.1038/nchem.763. [2] W. Li, J.R. Dahn, D.S. Wainwright, Sci. (New York, NY) 264 (5162) (1994) 1115e1118, http://dx.doi.org/10.1126/science.264.5162.1115. [3] X. Chen, H. Zhu, Y.-C. Chen, Y. Shang, A. Cao, L. Hu, G.W. Rubloff, ACS Nano 6 (9) (2012) 7948e7955, http://dx.doi.org/10.1021/nn302417x. [4] J.-M. Li, K.-H. Chang, T.-H. Wu, C.-C. Hu, J. Power Sources 224 (2013) 59e65, http://dx.doi.org/10.1016/j.jpowsour.2012.09.007. [5] H. Li, T. Zhai, P. He, Y. Wang, E. Hosono, H. Zhou, J. Mater. Chem. 21 (6) (2011) 1780, http://dx.doi.org/10.1039/C0JM02788J. [6] G. Wang, X. Lu, Y. Ling, T. Zhai, H. Wang, Y. Tong, Y. Li, ACS Nano 6 (11) (2012) 10296e10302, http://dx.doi.org/10.1021/nn304178b. [7] J.-M. Li, K.-H. Chang, C.-C. Hu, Electrochem. Commun. 12 (12) (2010) 1800e1803, http://dx.doi.org/10.1016/j.elecom.2010.10.029. [8] W. Tang, Y. Zhu, Y. Hou, L. Liu, Y. Wu, K.P. Loh, Energy & Environ. Sci. 6 (7) (2013) 2093, http://dx.doi.org/10.1039/C3EE24249H. [9] S. Boukhalfa, K. Evanoff, G. Yushin, Energy & Environ. Sci. 5 (5) (2012) 6872, http://dx.doi.org/10.1039/c2ee21110f. [10] Q.H. Do, Florida State University, (Ph.D. Dissertation), 2013. http://diginole.lib. fsu.edu/etd/8678/. [11] H. Li, P. He, Y. Wang, E. Hosono, H. Zhou, J. Mater. Chem. 21 (29) (2011) 10999, http://dx.doi.org/10.1039/C1JM11523E. [12] B.E. Conway, Kluwer Academic/Plenum Publishers, New York, 1999, ISBN: 978-1-4757-3058-6. [13] M. Sathiya, A.S. Prakash, K. Ramesha, J.-M. Tarascon, A.K. Shukla, J. Am. Chem. Soc. 133 (40) (2011) 16291, http://dx.doi.org/10.1021/ja207285b. [14] S.L. Woo, N. Yabuuchi, B.M. Gallant, C. Shuo, B.S. Kim, P.T. Hammond, S.H. Yang, Nat. Nanotechnol. 5 (7) (2010) 531e537. [15] X. Chen, et al., J. Mater. Chem. A 1 (28) (2013) 8201. [16] A.A. Zakhidov, D.-S. Suh, A.A. Kuznetsov, J.N. Barisci, E. Munoz, A.B. Dalton, et al., Adv. Funct. Mater. 19 (14) (2009) 2266e2272. [17] O.S. Burheim, F. Liu, B.B. Sales, O. Schaetzle, C.J.N. Buisman, H.V.M. Hamelers, J. Phys. Chem. C 116 (36) (2012) 19203e19210, http://dx.doi.org/10.1021/ jp306522g. [18] B.B. Sales, O.S. Burheim, F. Liu, O. Schaetzle, C.J.N. Buisman, H.V.M. Hamelers, Environ. Sci. Technol. 46 (21) (2012) 12203e12208, http://dx.doi.org/10.1021/ es302169c.
