Nanorod Niobium Oxide as Powerful Catalysts for an All Vanadium

Nov 26, 2013 - Recent emerging issues, such as fuel diversification, resource scarcity, energy security, and regulatory market opportunities, and so f...
4 downloads 14 Views 2MB Size
Download PDF
Letter pubs.acs.org/NanoLett

Nanorod Niobium Oxide as Powerful Catalysts for an All Vanadium Redox Flow Battery Bin Li, Meng Gu, Zimin Nie, Xiaoliang Wei, Chongmin Wang, Vincent Sprenkle, and Wei Wang* Pacific Northwest National Laboratory, P.O. Box 999, Richland, Washington 99352, United States S Supporting Information *

ABSTRACT: A powerful low-cost electrocatalyst, nanorod Nb2O5, is synthesized using the hydrothermal method with monoclinic phases and simultaneously deposited on the surface of a graphite felt (GF) electrode in an all vanadium flow battery (VRB). Cyclic voltammetry (CV) study confirmed that Nb2O5 has catalytic effects toward redox couples of V(II)/V(III) at the negative side and V(IV)/V(V) at the positive side to facilitate the electrochemical kinetics of the vanadium redox reactions. Because of poor conductivity of Nb2O5, the performance of the Nb2O5 loaded electrodes is strongly dependent on the nanosize and uniform distribution of catalysts on GF surfaces. Accordingly, an optimal amount of W-doped Nb2O5 nanorods with minimum agglomeration and improved distribution on GF surfaces are established by adding water-soluble compounds containing tungsten (W) into the precursor solutions. The corresponding energy efficiency is enhanced by ∼10.7% at high current density (150 mA·cm−2) as compared with one without catalysts. Flow battery cyclic performance also demonstrates the excellent stability of the as prepared Nb2O5 catalyst enhanced electrode. These results suggest that Nb2O5-based nanorods, replacing expensive noble metals, uniformly decorating GFs holds great promise as highperformance electrodes for VRB applications. KEYWORDS: Energy storage, redox flow battery, catalysts, electrode, vanadium, nanorod

R

promising redox systems because of its reduced crossover of active redox species and excellent electrochemical reversibility.4 In particular, compared with the traditional sulfuric acid VRB, a mixed-acid VRB recently invented by Li and co-workers at Pacific Northwest National Laboratory has achieved a significantly improved energy density and a wider operational temperature window.5 Despite its compelling merits, VRBs have achieved only limited market presence after the continuous development over the last 20 years. High cost is still one of the major factors that has prevented VRB technology from a broader market penetration.6 In this regard, decreasing stack size is the most effective way to reduce the system cost because a recent cost evaluation suggested that the stack-related cost accounts for ∼60% of the whole system cost for a 1 MW/0.25 MWh VRB system.7 Reducing the stack size requires the flow cell to be operated under a higher charge/discharge current density for a nominal system power output, which would inevitably increase the cell polarization and result in declining cell performance manifested as low efficiencies and a low vanadium-utilization ratio. Therefore, it is critically important to reduce the

ecent emerging issues, such as fuel diversification, resource scarcity, energy security, and regulatory market opportunities, and so forth, have dramatically changed today’s global energy landscape, which in turn has fueled interest from both the industrial and research communities in the development of enhanced stationary energy storage. The underlying theme, indeed, is increasing recognition that reliable access to clean and affordable energy is crucial to current economic growth and to a sustainable and prosperous future.1,2 Underscoring these considerations is the renewed interest in redox flow energy storage systems, which offer great promise to provide transformational improvements in system energy density, scalability, cost, and so forth, because of their fundamentally different architectures from currently available solid-state rechargeable batteries.3 Capitalizing on the separation of the energy-carrying electrolyte and the power-generating stack, the decoupling of energy capacity and power output is the most fundamental characteristic that differentiates flow-based batteries from solid-state batteries. This difference allows the energy-to-power ratio to be independently tuned to meet specific applications. Because of the unique architecture, other appealing advantages are the intrinsic high safety, easy thermal management, long service life, and compatibility with modular manufacturing approaches. Among the various flow batteries, the all-vanadium redox flow battery system (VRB) invented by Skyllas-Kazacos and co-workers is probably one of the most © 2013 American Chemical Society

Received: October 2, 2013 Revised: November 16, 2013 Published: November 26, 2013 158

dx.doi.org/10.1021/nl403674a | Nano Lett. 2014, 14, 158−165

Nano Letters

Letter

overpotential of flow cells especially at high charge/discharge current density to obtain better rate capacity and efficiency. The overpotentials of a flow cell include ohmic polarization (membrane, electrodes, and electrolytes), charge transfer polarization (electrodes), concentration polarization resistances, and so forth.8 In VRBs, because the redox reactions between couples of V(II)/V(III) and V(IV)/V(V) take place on the electrode surfaces, the electrode materials and surface chemistry determine the charge transfer polarization. Because of their wide operating potential range, stability as both negative and positive electrodes and availability with high surface areas at low cost, carbon-based materials such as graphite felt (GF) typically are used as electrode materials for VRBs. However, pristine carbon materials normally exhibit poor kinetic reversibility and electrochemical activity, which limits the power capability of the VRBs (∼50 mA/cm2).9 So far, several approaches have been reported to effectively improve the electrochemical properties of carbon electrodes in VRBs. First, carbon-based electrodes were treated to form special functional groups on the surface, such as carboxylic, hydroxyl, and pyridinic nitrogen groups that were confirmed to afford faster charge transfer.9−14 Graphite oxide (GO) has recently attracted more attention because of the presence of surface oxygen functional groups, large specific surface areas, and active defects on the surfaces or around edges.15 The conductivity of GO usually needs to be improved by adjusting the carbon-tooxygen ratio,16,17 mixing with a conductive matrix (such as graphite powders and carbon nanotubes),18,19 or converting to nitrogen functional groups (such as pyridinic nitrogen).20−22 Second, noble metals, such as Pt, Au, Ir, Pd, and Ru, were deposited onto the surface of carbon electrodes to enhance the electrochemical catalytic activity of electrodes; however, this approach is not practical because of the high cost, limited mineral resources, and susceptibility to hydrogen evolution.10,23−26 Recently, we reported on the use of low-cost Bi nanoparticles as catalysts, which led to significantly improved cell performance.27 In addition, low-cost metal oxides such as Mn3O4 and WO3 were reported as catalysts for VRBs.28,29 However, Mn3O4 is soluble in HCl, while the reported WO3 seems to have a limited catalytic effect with low energy efficiency.29 Nb2O5 has attracted much interest because of its remarkable properties that are suitable for a wide range of applications such as gas sensing, catalysis, electrochromics, photoelectrodes, fieldemission displays, and microelectronics.30−34 Particularly, Nb2O5 shows a great promise in providing strong surface acidity and stability in aqueous medium for various acidcatalyzed reactions. In this paper, we describe our success in synthesizing low-cost, Nb2O5-based nanorods as novel catalysts that were loaded onto the surface of GFs at both the positive and negative electrodes of a VRB. Addition of these Nb2O5 nanorods enhanced the electrochemical activity of electrodes, thus enabling high current-density operation. We investigated the effects of synthesis conditions, size, and distribution of the catalyst on the surface of GFs and the electrochemical performance of VRBs. Nb2O5 powders were synthesized using a hydrothermal method (details in Experimental Section). Because Nb2O5 is known to have an n-type semiconducting property with a band gap of about 3.4 eV (exhibiting poor conductivity),35 it was mixed with graphite powders and then deposited on glassy carbon electrodes for cyclic voltammetry (CV) testing. Figure 1 shows the CV curves on glassy carbon working electrodes in 2

Figure 1. Cyclic voltammograms with or without Nb2O5 nanorods onto glassy carbon as working electrodes in solutions of 2 M VOSO4 + 5 M HCl.

M VOSO4 and 5 M HCl electrolytes with and without Nb2O5 powders at the same scanning rate (50 mV·s−1). It can be clearly seen from Figure 1 that four main peaks are exhibited in both curves, representing redox reactions between two redox couples of V(II)/V(III) and V(IV)/V(V). The potential separation between the oxidation and the reduction peaks for each redox couple is an indication of the redox reaction kinetics and reversibility. As demonstrated in Figure 1, the oxidation and reduction peak potential separations corresponding to redox couples V(II)/V(III) and V(IV)/V(V) are around 0.498 and 0.423 V, respectively, for the working electrodes without Nb2O5. However, upon adding Nb2O5, the peak potential separations corresponding to redox couples V(II)/V(III) and V(IV)/V(V) are decreased by 61 and 100 mV, respectively, suggesting that Nb2O5 has an electrocatalytic effect on redox reactions at both the negative and positive electrodes, which is completely different from the effects previously reported for bismuth catalysts.27 Bismuth as a catalyst was confirmed to be present only in the form of bismuth metal at the negative side and to facilitate the redox reaction between V(II) and V(III). However, Nb2O5 can afford catalytic effects on both V(II)/ V(III) and V(IV)/V(V) redox couples. Nb2O5 nanorods distributed uniformly on the GF electrode are synthesized using a hydrothermal method in which the GFs were placed in a Teflon-sealed autoclave containing ammonium niobium oxalate solutions. Nb2O5-based powders were synthesized on the surface of GFs after hydrothermal treatment at 443 K, followed by calcination at 773 K in Ar. The amount of catalysts on the surface of GFs was calculated based on the weight difference of GFs before and after loading the Nb2O5. Because of their low conductivity, the amount and distribution of Nb2O5 powders on the surface of conductive GFs significantly influence the catalytic effect toward active materials in VRBs. In Figure 2a, the amount of obtained solid precipitate on the surface of GFs is plotted as a function of the concentration of Nb in the precursor solutions. With increasing concentrations of Nb, more precipitates were found on the GFs with the same synthesis time and temperature. Figure 2b,c shows field emission scanning electron microscope (FE-SEM) images of Nb2O5-modified GFs with different Nb concentrations in the precursor solutions. Consistent with the results 159

dx.doi.org/10.1021/nl403674a | Nano Lett. 2014, 14, 158−165

Nano Letters

Letter

Figure 2. The amount of nanoparticles on the surface of GFs as a function of the concentration of Nb in the precursor solutions with or without W source (a); the FE-SEM images of Nb2O5-modified graphite felts with different Nb concentrations in the precursor solutions, 0.05 M-Nb (b), 0.10 M-Nb (c), and 0.05 M-Nb(W) (d). The inset of panel d shows the FE-SEM image of nanoparticles on GFs at larger magnification.

of Figure 2a, it can be seen from Figure 2b,c that as the concentration of Nb increases the amount of particles on GFs increases, which however also is accompanied by more severe agglomeration. As shown in Figure 2b,c, the particle sizes grew from ∼700 nm to ∼7 μm as the concentration of Nb increased from 0.1 to 0.15 M. To further increase the amount of Nb2O5 while preventing the agglomeration of nanorods, a small amount of ammonium paratungstate, (NH4)10W12O41·H2O, was added into the precursor solutions. The atomic ratio of W to Nb in the precursor solutions was kept at 1:10. The introduction of a W source promoted the precipitation of active materials.36 In Figure 2a, a steeper increase in the catalysts loading on GFs was observed with increasing Nb concentration in the precursor solutions with the addition of the W source. At the same concentration of 0.05 M Nb, the catalyst weight ratio on GFs was increased from 1.67 to 3.80 wt % after introducing the W source. Consistent with the results shown in Figure 2a, nanoparticles with a more uniform size of ∼250 nm are synthesized and uniformly distributed on GFs as demonstrated in Figure 2d. The inset in Figure 2d shows that the nanoparticles are composed of many nanorods. The composition of the obtained nanopowders was measured by the inductively coupled plasma (ICP) method after melting the powders with KOH at 500 °C for 6 h and then dissolving the obtained residue in deionized water. The W-to-Nb atomic ratio was measured to be around 1:3, which is larger than that in the

precursor solutions (1:10), possibly because the presence of the W source expedites the formation of solid precipitates, which was consistent with the results observed by other investigators.36 The structure integrity of the as prepared Nb2O5 nanorods loaded GF electrode was also investigated after repetitive electrochemical cycling. As shown in Supporting Information Figure S1, the presence of Nb2O5 nanorods catalysts demonstrated no significant change compared to Figure 2d at both electrodes of flow cell, suggesting excellent electrochemical and mechanical stability of the as synthesized Nb2O5 catalysts under the constant cycling and continuous flow (20 mL·min−1) condition in the mixed-acid vanadium flow battery. The morphology and structure of the obtained nanopowders containing W were analyzed further by scanning transmission electron microscope (STEM). Figure 3a shows the STEM Zcontrast image of the nanoparticles. It can be seen from this figure that the nanoparticles are composed of nanorods with diameters in the range of 3−4 nm. The STEM image of one typical nanorod is shown in Figure 3b, which exhibits a single crystal structure. The composition of the nanorods was studied using electron energy loss spectroscopy (EELS). Integrated Nb M-edge and W M-edge EELS spectra along the green line in Figure 3b are shown in Figure 3c, indicating a uniform distribution of Nb and W along the axis of the nanorod, which suggests that W was uniformly incorporated into the lattice of 160

dx.doi.org/10.1021/nl403674a | Nano Lett. 2014, 14, 158−165

Nano Letters

Letter

Figure 3. STEM images of W-doped Nb2O5 nanorods at lower (a) and higher (b) magnifications; the distributions of Nb and W elements in one EELS line scan (the line is shown in panel b) (c); corresponding FFT of one atomic resolution image (panel b) (d).

concentration polarizations remain the same. In a flow cell, redox reactions happen in close proximity of the electrode surface. The surface properties of the carbon electrodes therefore have critical influence in not only the redox reaction kinetics but also the charge transfer. Higher charge transfer resistance leads to higher overpotential. We carried out AC impedance analysis to identify the charge transfer resistance. Figure 5 shows Nyquist plots for different electrodes obtained at the state of charge (SOC) of 50%. All the Nyquist plots include a semicircle part at the high frequency arising from the charge transfer reaction at the electrolyte/electrode interface, and a linear part at low frequency associated with the diffusion of vanadium ion through solutions.29 The radius of the semiarc reflects the charge transfer resistance, which gradually decreases in the order of the pristine, Nb2O5-modified, and welldistributed W-doped Nb2O5-modified GF electrodes in good agreement with the results of Figure 4a. On the basis of the impedance measurement, equivalent circuit modeling was carried out to better understand the differences in electrochemical processes on the pristine and Nb2O5 catalyst loaded GF electrodes. The numerical fitting results indicate that the charge transfer resistances (R2) are 0.059, 0.031, and 0.022 Ω, respectively, for pristine, 0.1 M Nb and 0.05 M Nb(W) electrodes. The detailed analysis is provided in the Supporting Information (impedance analysis, Figure S2, and Figure S3). These results further illustrate the catalytic effect of Nb2O5 and the importance of uniform distribution of poor-conductivity Nb2O5 on GFs in VRBs.

Nb2O5. Fast Fourier transform (FFT) analysis of one atomic resolution image is shown in Figure 3d, which confirmed the C12/C1 monoclinic structure of the nanorods. Other nanorods also were identified with the same result attained, suggesting that the nanoparticles are composed of W-doped Nb2O5 single crystal nanorods rather than mixtures of WO3 and Nb2O5 nanorods. Charge/discharge cycling using a VRB single flow cell with Nafion 115 as the membranes was performed to further understand the catalytic effect of Nb2O5 deposited on GFs on the electrochemical performance of the VRB flow cell. Figure 4a shows the charge−discharge voltage profiles at the same charge−discharge current density of 150 mA·cm−2 for the pristine, Nb2O5-modified and W-doped Nb2O5-modified GF electrodes. The introduction of Nb2O5 on the surface of GFs leads to reduced overpotential in the corresponding charge and discharge processes by 45 and 126 mV, respectively, compared with those of pristine GFs, suggesting the electrocatalytic effect of Nb2O5 materials. Thus, the charge voltage is decreased, and the discharge voltage is enhanced. The overpotentials further decrease by 22 and 39 mV in the corresponding charge and discharge process for the W-doped Nb2O5 GF electrodes mainly owing to the more uniform distribution of the nanocatalysts on the surface of GFs (see Figure 2d), resulting in lower charge voltage and higher discharge voltage and thus higher charge/discharge capacity given a fixed voltage window between 0.8 and 1.6 V. In our case, identical electrolytes, membranes, and flow rates were used in the flow cell test. Accordingly, the overpotentials arising from ohmic and 161

dx.doi.org/10.1021/nl403674a | Nano Lett. 2014, 14, 158−165

Nano Letters

Letter

Figure 4. Electrochemical performances of VRBs employing different electrodes as a function of charge/discharge current density: charge−discharge curves at 150 mA·cm−2 (a); CE and VE (b); EE (c); cycling performances of VRB cells with pristine and W-doped Nb2O5 nanorods-decorating GFs as both electrodes (d).

Figure 5. Nyquist plots of VRBs cell with different catalysts-modified GFs as both electrodes at SOC of 50%.

To improve cell performance, the amounts and distribution of catalysts on the surface of GF electrodes were systematically optimized by adjusting the concentration of Nb from 0 to 0.15 at 0.05 M intervals in the precursor solutions with and without the W source. The catalyst modified GFs electrodes are thus labeled as x M-Nb or x M-Nb(W), which means the

corresponding electrode is prepared from the precursor solution containing x M Nb without or with a W source. The Coulomb efficiency (CE), voltage efficiency (VE), and energy efficiency (EE) were measured under current densities varying from 50 to 150 mA·cm−2. CE is the ratio between the flow cell charge and the discharge capacity, while VE is 162

dx.doi.org/10.1021/nl403674a | Nano Lett. 2014, 14, 158−165

Nano Letters

Letter

Figure 6. Specific discharge capacity (a) and discharge energy density (b) of VRBs as a function of cycle number at different charge/discharge current densities employing different surface-modified electrodes.

the GFs, resulting in a higher tendency of catalysts to be displaced from surfaces of GFs in the flowing electrolyte. Therefore, to avoid agglomeration of nanocatalysts the soluble compound containing W, ((NH4)10W12O41·5H2O), was introduced into the original precursor solutions. Nucleation centers seem to be more easily formed and evenly distributed, thus reducing the agglomeration of nanorods during the same hydrothermal reaction time (48 h). However, as the concentration of Nb increases, the agglomeration of catalysts still occurs. As a result, as shown in Figure 2d and Figure 4b,c, an optimal concentration of Nb (0.05 M) with the W source gives the best flow battery performance. This is because there are uniformly distributed nanoparticles with smaller sizes on the surfaces of GFs due to the addition of the W source. In addition, the cycling performances of VRBs with pristine GFs and well-distributed W-doped Nb2O5 GFs (0.05 M Nb(W)) as electrodes under a current density of 150 mA·cm−2 are shown in Figure 4d. It can be seen from Figure 4d that EE values are enhanced by the introduction of uniformly distributed Wdoped Nb2O5 nanocatalysts. Furthermore, almost no EE fading was observed, and a large amount of nanoparticles could still be found on the surfaces of both the negative and positive GFs electrodes (see Supporting Information Figure S1) after more than 50 cycles, which, as discussed earlier in this paper, suggests good stability of both the Nb2O5 and its electrocatalytic effect on both electrodes in the mixed acid vanadium electrolyte over repetitive cycling. It can be seen from Supporting Information Figure S4 that within a fixed voltage window the cells operated at a higher current density will demonstrate reduced charge and discharge capacities because of the elevated overpotential. Therefore, as shown in Figure 6a, specific discharge capacity decreases with increasing discharge rate for different electrodes. However, the overpotential can be significantly reduced by introducing appropriate, well-dispersed Nb2O5-based nanocatalysts onto GFs, which affords faster charge transfer. Consistent with the results shown in Figure 4c, the optimal electrodes in our case were the GFs decorated with well-distributed W-doped Nb2O5 nanocatalysts, treated in the precursor solutions containing W with 0.05 M Nb. These electrodes deliver the highest specific discharge capacity of 14.4 Ah·L−1 at 150 mA·cm−2, which corresponds to a ∼1.8 times increase compared to the cell using

determined by the voltage difference between the charge and discharge processes. The EE value is a derivative of the CE and VE (EE = VE × CE). Figure 4b shows the CE and VE values as function of current density for flow cells with different electrodes. For a given GF electrode used in a flow cell, CE, with values of around 97% increases slightly by ∼1.5% with the increasing current density attributed to the reduced time of vanadium ion crossover through membranes. Nevertheless, a fast charge/discharge rate can cause the substantial increase in charge and discharge overpotential, leading to significant VE drop. At the same charge/discharge rate, the surface modification of GF electrodes has little effect on the CE improvement. However, it significantly influences the VE values of the flow cell. On one hand, as the concentration of Nb in precursor solutions containing only Nb increases, corresponding to the increasing catalyst amount on GF electrode, the VE values first increases and then decreases, with the highest VE value of 74.3% at 150 mA·cm−2 at 0.1 M Nb in the precursor solutions, which is around 7.8% higher than the one without catalysts. On the other hand, upon adding W source into the precursor solutions, the VE is further improved and reaches the highest value of 77.6% at 150 mA·cm−2 at 0.05 M Nb in the precursor solutions. As shown in Figure 4c, the trend of EE values with Nb concentration and current density is very similar to that of VE, which is attributed to minor variations of CE. At a current density of 150 mA·cm−2, the optimized Nb2O5 (0.1 M Nb) and W-doped Nb2O5 (0.05 M Nb(W)) GF electrodes at both sides improve the EE values of VRBs by 7.6 and 10.7%, respectively, compared with cells with pristine GF electrodes (electrodes with no catalyst addition). A discussion of why the optimal concentration of Nb in the precursor is needed to achieve the highest EE values follows. In our case, as previously discussed, Nb2O5 has an electrocatalytic effect toward both V(II)/V(III) and V(IV)/V(V) redox couples, which facilitate the corresponding redox reactions, especially at high charge/ discharge rate, presumably by lowering the kinetic activation energy. A small amount of Nb2O5 catalyst introduced on the surface of GFs can promote faster charge transfer, leading to improved performance of electrodes. However, as shown in Figure 2a−c, it is evident that the increasing concentration of Nb in the precursor solutions leads to more severe Nb2O5 nanorod agglomeration into larger particles on the surface of 163

dx.doi.org/10.1021/nl403674a | Nano Lett. 2014, 14, 158−165

Nano Letters

Letter

h prior to use it for catalytic reaction. The obtained catalystmodified GFs were used as electrodes in VRBs. Characterizations. A CV test was conducted in a threeelectrode cell using a CHI660C workstation (CH Instruments, USA) in a nitrogen-filled glovebox. A Pt wire and Ag/AgCl electrode were used as the counter and reference electrodes, respectively. The working electrodes were prepared by applying highly dispersed graphite or Nb2O5/graphite ink onto the prepolished glassy carbon disk electrode. A Nafion thin film was formed on the surface by dropping 10 μL of 0.05 wt % Nafion on it after the ink was dried. The ink was prepared by dispersing carbon powder (graphite, MPC, N-MPC powder) or the mixtures of Nb2O5 nanorods and carbon in ethanol under strong ultrasonication. The active materials loading on the disk electrode is 20 μg. Testing was performed from −0.85 to 1.5 V versus Ag/AgCl reference electrode in solutions of 2 M VOSO4 and 5 M HCl with or without Nb2O5 catalysts on glassy carbon electrode at a scan rate of 50 mV·s−1. In addition, the morphology and microstructure of modified GFs and the resultant nanorods were characterized by a FE-SEM (JEOL JSM-7600F) and an aberration-corrected STEM (TEM/STEM, FEI Titan) at 300 kV, respectively. The EELS of the resultant nanorods was collected using a Quantum Gatan Image filter with 2K × 2K charge-coupled device. The ac impedance of the flow cells was measured using a Solartron 1250 frequency response analyzer in a frequency range from 0.1 Hz to 1 MHz with an ac amplitude of 20 mV. The W-to-Nb mole ratio of the W-doped Nb2O5 nanorods was measured by ICP/atomic emission spectrometry (ICP/AES, Optima 7300DV, PerkinElmer) techniques. Flow-Cell Test. The original electrolytes were prepared by dissolving 2 M VOSO4 (Aldrich, 99%) in 5 M HCl solutions (Aldrich, 37%). The corresponding anolytes and catholytes can be prepared through an effective “balance” process. The setup of the flow cells is described in detail in a previous publication.38 The areas of active electrodes at both sides are kept to be 5 cm2. The single cell was connected to two glass reservoirs containing balanced 50 mL catholytes and anolytes, respectively. Both reservoirs were purged with nitrogen gas and then sealed prior to the electrochemical tests to minimize oxidation of V2+. Nafion 115 was used for the membranes. Electrolytes were pumped at a flow rate of 20 mL·min−1 (0.33 cm·s−1 face velocity) through a peristaltic pump. The electrochemical performance of the flow cell was carried out using a potentiostat/galvanostat (Arbin Instrument, U.S.A.) within a fixed voltage window between 0.8 and 1.6 V under a constant current mode operated under current densities ranging from 50 to 150 mA·cm−2.

pristine GFs as electrodes. The corresponding fuel (V) utilization ratio was improved by around 25%, which will contribute significantly to cost reduction for VRBs stacks because V electrolyte cost is nearly 40% of the VRBs system cost.37 Moreover, discharge energy density was determined by both discharge capacity and discharge voltage. As shown in Figure 6b, compared with the pristine GFs, the presence of appropriate catalysts on GFs resulted in a 112% improvement of the cell discharge energy density at 150 mA·cm−2 because of the improvement in both discharge capacity and voltage (see Figure 4a). On the basis of the experimental results, Nb2O5 based nanopowders were proved to be powerful catalysts toward redox couples of both V(II)/V(III) and V(IV)/V(V). The GFs decorated with well-distributed Nb2O5 based nanocatalysts substantially improved the overall VRB performance, in terms of critical operation parameters such as VE, EE, utilization of fuels, specific discharge capacities, discharge energy densities, etc. In summary, monoclinic Nb2O5 nanorods, which were synthesized by the hydrothermal method and deposited onto GF surfaces, can act as a powerful electrocatalyst toward both V(II)/V(III) and V(IV)/V(V) redox couples to enhance the electrochemical activity of GF electrode in VRB systems. We investigated and optimized the amount, size, and distribution of the Nb2O5 nanorod catalyst on GF surfaces. Adding soluble salts containing W in the precursor solutions during synthesis was found to produce more uniformly distributed and smallersized nanoparticles. We determined that the W element was doped into the lattice of monoclinic Nb2O5 nanorods. The Nb2O5-based nanocatalysts exhibit excellent stability at both sides in the vanadium mixed-acid electrolyte and on electrocatalytic effect over repetitive cycling. Therefore, VRBs, assembled with well-distributed, W-doped, Nb2O5-nanoroddecorated GFs as positive and negative electrodes show significantly enhanced performance in terms of VE, EE, utilization of fuels, discharge capacities, and discharge energy densities due to lower charge transfer resistance, particularly under high current density. These results suggest that the nanosized Nb2O5-based catalyst offers great promise as highperformance electrodes for VRB applications.

■

EXPERIMENTAL SECTION Synthesis of Nb2O5 and W-Doped Nb2O5 Nanorods on GFs. Nb2O5 and W-doped Nb2O5 nanorods were synthesized by a hydrothermal method. Typically, the precursor solutions were prepared by dissolving appropriate amount of ammonium niobium oxalate, NH4[NbO(C2O4)2(H2O)]·xH2O, or ammonium paratungstate, (NH4)10W12O41·5H2O into deionized water. The concentration of Nb in the precursor solutions were varied from 0 to 0.15 M at 0.05 M intervals. The W-to-Nb mole ratio of in the precursor solutions was kept at 1:10. The obtained mixed transparent solution and GFs (GFD5EA, SGL Carbon Group, Germany) were simultaneously placed in a Teflon-sealed autoclave and heated at 443 K for 48 h. After the synthesis, white solid precipitates were observed in the autoclave and on the surfaces of the GFs. Then, the precursor powders were collected by vacuum filtration and washed three times with deionized water and three times with alcohol. Similarly, the GFs were taken out from the autoclave and washed with deionized water three times. Both precursor powders and GFs with precursors on the surfaces were dried and then calcined at 773 K in an Ar flow (150 mL·min−1) for 2

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information is available. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: 1-509-372-4097. Notes

The authors declare no competing financial interest. 164

dx.doi.org/10.1021/nl403674a | Nano Lett. 2014, 14, 158−165

Nano Letters

■

Letter

(26) Huang, R. H.; Sun, C. H.; Tseng, T.; Chao, W.; Hsueh, K. L.; Shieu, F. S. J. Electrochem. Soc. 2012, 159 (10), A1579−A1586. (27) Li, B.; Gu, M.; Nie, Z.; Shao, Y.; Luo, Q.; Wei, X.; Li, X.; Xiao, J.; Wang, C.; Sprenkle, V. Nano Lett. 2013, 13 (3), 1330−1335. (28) JaeáKim, K.; JináLee, N. Chem. Commun. 2012, 48 (44), 5455− 5457. (29) Yao, C.; Zhang, H.; Liu, T.; Li, X.; Liu, Z. J. Power Sources 2012, 218, 455−461. (30) Wang, Y.-D.; Yang, L.-F.; Zhou, Z.-L.; Li, Y.-F.; Wu, X.-H. Mater. Lett. 2001, 49 (5), 277−281. (31) Carniti, P.; Gervasini, A.; Marzo, M. J. Phys. Chem. C 2008, 112 (36), 14064−14074. (32) Mujawar, S.; Inamdar, A.; Patil, S.; Patil, P. Solid State Ionics 2006, 177 (37), 3333−3338. (33) Jose, R.; Thavasi, V.; Ramakrishna, S. J. Am. Ceram. Soc. 2009, 92 (2), 289−301. (34) Varghese, B.; Haur, S. C.; Lim, C.-T. J. Phys. Chem. C 2008, 112 (27), 10008−10012. (35) Tsang, E.; Zhou, X.; Ye, L.; Edman Tsang, S. C. Nano Rev. 2012, 3. (36) Okumura, K.; Tomiyama, T.; Shirakawa, S.; Ishida, S.; Sanada, T.; Arao, M.; Niwa, M. J. Mater. Chem. 2011, 21 (1), 229−235. (37) Kamath, H.; Rajagopalan, S.; Zwillenberg, M. Vanadium Redox Flow Batteries: An In-Depth Analysis; Electric Power Research Institue: Palo Alto, CA, 2007. (38) Li, B.; Li, L.; Wang, W.; Nie, Z.; Chen, B.; Wei, X.; Luo, Q.; Yang, Z.; Sprenkle, V. J. Power Sources 2013, 229, 1−5.

ACKNOWLEDGMENTS The authors would like to acknowledge financial support from the U.S. Department of Energy’s (DOE) Office of Electricity Delivery and Energy Reliability (OE) (under Contract No. 57558). The S/TEM work was conducted in the William R. Wiley Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility sponsored by DOE’s Office of Biological and Environmental Research and located at PNNL. Pacific Northwest National Laboratory is a multiprogram national laboratory operated by Battelle for DOE under Contract DE-AC05-76RL01830.

■

REFERENCES

(1) Kim, S.; Vijayakumar, M.; Wang, W.; Zhang, J.; Chen, B.; Nie, Z.; Chen, F.; Hu, J.; Li, L.; Yang, Z. Phys. Chem. Chem. Phys. 2011, 13 (40), 18099−18106. (2) Dunn, B.; Kamath, H.; Tarascon, J.-M. Science 2011, 334 (6058), 928−935. (3) Wang, W.; Luo, Q.; Li, B.; Wei, X.; Li, L.; Yang, Z. Adv. Funct. Mater. 2013, 23 (8), 970−986. (4) Skyllas-Kazacos, M.; Rychick, M.; Robins, R. All-vanadium redox battery. U.S. Patent, 4786567, 1988. (5) Li, L.; Kim, S.; Wang, W.; Vijayakumar, M.; Nie, Z.; Chen, B.; Zhang, J.; Xia, G.; Hu, J.; Graff, G.; Liu, J.; Yang, Z. Adv. Energy Mater. 2011, 1 (3), 394−400. (6) Kear, G.; Shah, A. A.; Walsh, F. C. Int. J. Energy Res. 2012, 36 (11), 1105−1120. (7) Viswanathan, V.; Crawford, A.; Stephenson, D.; Kim, S.; Wang, W.; Li, B.; Coffey, G.; Thomsen, E.; Graff, G.; Balducci, P.; KintnerMeyer, M.; Sprenkle, V. J. Power Sources 2014, 247, 1040−1051. (8) Liu, Q.; Turhan, A.; Zawodzinski, T.; Mench, M. M. Chem. Commun. 2013, 49, 6292−6294. (9) Sun, B.; Skyllas-Kazacos, M. Electrochim. Acta 1992, 37 (7), 1253−1260. (10) Sun, B. T.; Skyllas-Kazacos, M. Electrochim. Acta 1991, 36 (3− 4), 513−517. (11) Sun, B.; Skyllas-Kazacos, M. Electrochim. Acta 1992, 37 (13), 2459−2465. (12) Kim, K. J.; Kim, Y.-J.; Kim, J.-H.; Park, M.-S. Mater. Chem. Phys. 2011, 131 (1−2), 547−553. (13) Zhu, H.; Zhang, Y.; Yue, L.; Li, W.; Li, G.; Shu, D.; Chen, H. J. Power Sources 2008, 184 (2), 637−640. (14) Wu, T.; Huang, K.; Liu, S.; Zhuang, S.; Fang, D.; Li, S.; Lu, D.; Su, A. J. Solid State Electrochem. 2012, 16 (2), 579−585. (15) Han, P.; Wang, H.; Liu, Z.; Chen, X.; Ma, W.; Yao, J.; Zhu, Y.; Cui, G. Carbon 2011, 49 (2), 693−700. (16) González, Z.; Botas, C.; Alvarez, P.; Roldan, S.; Blanco, C.; Santamaría, R.; Granda, M.; Menéndez, R. Carbon 2012, 50 (3), 828− 834. (17) Li, W.; Liu, J.; Yan, C. Carbon 2013, 55, 313−320. (18) Li, W.; Liu, J.; Yan, C. Electrochim. Acta 2011, 56 (14), 5290− 5294. (19) Han, P.; Yue, Y.; Liu, Z.; Xu, W.; Zhang, L.; Xu, H.; Dong, S.; Cui, G. Energy Environ. Sci. 2011, 4 (11), 4710−4717. (20) Shao, Y.; Wang, X.; Engelhard, M.; Wang, C.; Dai, S.; Liu, J.; Yang, Z.; Lin, Y. J. Power Sources 2010, 195 (13), 4375−4379. (21) Wang, S.; Zhao, X.; Cochell, T.; Manthiram, A. J. Phys. Chem. Lett. 2012, 3 (16), 2164−2167. (22) Jin, J.; Fu, X.; Liu, Q.; Liu, Y.; Wei, Z.; Niu, K.; Zhang, J. ACS Nano 2013, 7 (6), 4764−4773. (23) Flox, C.; Rubio-Garcia, J.; Nafria, R.; Zamani, R.; Skoumal, M.; Andreu, T.; Arbiol, J.; Cabot, A.; Ramon Morante, J. Carbon 2012, 50 (6), 2372−2374. (24) Tsai, H.-M.; Yang, S.-J.; Ma, C.-C. M.; Xie, X. Electrochim. Acta 2012, 77, 232−236. (25) Wang, W. H.; Wang, X. D. Electrochim. Acta 2007, 52 (24), 6755−6762. 165

dx.doi.org/10.1021/nl403674a | Nano Lett. 2014, 14, 158−165