The Net Discharge Mechanism of the VB₂/Air Battery

Nanoscopic VB₂ was synthesized mechanochemically using vanadium powder (Alfa Aesar) and boron powder (Alfa Aesar) based on methods previously published by Licht et al. [8]. In a controlled argon atmosphere glove box system (Vacuum Atmospheres, Nexus-II), 0.500 g of vanadium and 0.212 g of boron power were measured and transferred into a tungsten carbide milling chamber along with ten tungsten carbide balls, ten millimeters in diameter. The milling vessel was sealed, removed from the glove box, and placed into a Retsch PM 100 planetary ball mill set to 600 rpm and allowed to run for four hours. After the vanadium diboride cooled to room temperature, the material was then collected in the glove box and the mass was measured.

The electrochemical performance of the VB₂ anode, as well as the composite anode, was investigated. Coulombic efficiency is the percentage of the measured capacity compared to the theoretical (intrinsic) anode discharge capacity of the VB₂ anode.

Anodes were prepared using nanoscopic VB₂, synthesized in-laboratory as described above, or commercially available, macroscopic VB₂ (American Elements Organo-Metallics), carbon black (TIMCAL C-NERGY SUPER C65), and KOH pellets (Alfa Aesar). Panasonic PR675H batteries were used as a test bed for electrode fabrication to perform electrochemical tests. The cell fabricated is as previously delineated.²¹ In brief, the air electrode from the Panasonic battery is kept intact and reused. The Zn anode of the Panasonic cells was removed by opening the existing cell, removing the anode active material, and using the separator and cathode as received for the VB₂/air battery. VB₂ electrodes were prepared first by mixing 70 wt% active material (VB₂) and 30 wt% carbon black, then measuring the appropriate amount of dry material, and lastly spreading the mixture with isopropyl alcohol and allowing the working electrode to dry. An 8M KOH electrolyte solution was used. The cell was then closed with the cap in reverse so that the anodic material was in contact with the electrolyte.

Once the fabrication process had been completed, cells were discharged at constant loads as indicated in the results and discussion. The change in voltage with time during discharge was measured using NI LabVIEW 2010 and NI USB-6210 multifunction data acquisition.

X-ray diffraction (XRD) measurements were performed with a Rigaku MiniFlex. Samples were prepared by loading the crushed powders on a round zero background holder containing a depression in the middle and spread by either isopropyl alcohol or petroleum jelly. Spectra were obtained using the Rigaku powder diffraction analysis package, PDXL. Further analysis was completed with comparison of the obtained spectra to the Inorganic Crystal Structure Database (ICSD) online.

X-ray photoelectron spectroscopy (XPS) measurements were obtained at the University of South Carolina on a Kratos AXIS Ultra DLD XPS system equipped with a hemispherical energy analyzer and a monochromatic Al K source, which was operated at 15 keV and 150 W. XPS fitting was carried out by using XPSPEAK Version 4.1 software. Anodic material of the macroscopic and nanoscopic VB₂ was tested prior to and after discharging at a load of 3,000 ohm. The anodic composite consisted of a mixture of 70% VB₂, equivalent to a capacity of 5 mAh, and 30% C65 carbon. Cells used for XPS analysis were discharged using an Arbin Instruments battery tester controlled by MITS software.

5 mL of 8 molar KOH were transferred to two 10 mL beakers and placed on a stir plate under constant stirring. Potassium metavanadate, KVO₃, or boron oxide, B₂O₃, were weighed and periodically transferred to either beaker, and dissolved until the two solutions became saturated. Both beakers were covered and left to stir for 48 hours. Vacuum filtration was used to collect KVO₃ (Alfa Aesar 99.9%) and B₂O₃ (Alfa Aesar 99%) that remained undissolved. Filter papers with the solid excess were left overnight to dry in a hood then weighed. The solubility was determined by subtracting the undissolved KVO₃ and B₂O₃ from the total mass added to each solution then dividing by volume of KOH. FTIR were measured for the salts and for K₂B₄O₇^.4H₂O (Alfa Aesar 99%), KVO₃ (Alfa Aesar 99.9%) and K₃VO₄ (Alfa Aesar 99.9%).

Characterization of spent battery material and battery discharge products was completed via Fourier-Transform Infrared Spectroscopy (FT-IR). Weighed samples to be analyzed were ground thoroughly with KBr by mortar and pestle. By means of a hydraulic press, 150 mg of the sample mixture was pressed under 7 tons of pressure for 5 minutes using a Perkin-Elmer 13 mm die set to form a KBr pellet containing the sample for analysis. Composition of the KBr pellets varied between 0.1 weight percent and 1 weight percent of the sample depending on the opaqueness of the sample. Filtrates for B₂O₃ in KOH and H₂O were measured using attenuated total reflectance infrared spectroscopy (ATR-FTIR), as an alternative method to transmission spectroscopy. FTIR spectra were measured by Perkin-Elmer Spectrum 100 with a resolution of 4 cm⁻¹ and between 16 and 32 scans per spectrum. A baseline correction was performed with the automatic baseline correction tool offered with the Perkin-Elmer Spectrum software package.

The unusually high gravimetric anodic capacity coupled with the characteristic high density of VB₂, 5.10 g cm⁻³, results in a large volumetric capacity of 20,700 mAh cm⁻³. Hence, only small quantities of VB₂ are required for extended battery discharge. A five mAh (1 cm diameter) button cell contains only ∼1 mg of VB₂, which yields little discharge product. This small amount of discharge product is a direct result of the attractive high capacity of the anode. However, the small quantities posed challenges to effective characterization of that discharge product. Initial attempts to scale-up were ineffective resulting in rapid decreases in coulombic efficiency as the anode thickness was increased several fold.

Recent improvements to the cell fabrication process including upgrades to the gas regeneration in the glove box to better exclude oxygen and better control of the ball mill process have significantly improved scaled-up cell performance. Here, we show that a 6-fold increase in the packed capacity of the 1 cm diameter cell results in a reproducible cell with a relatively small decrease in coulombic efficiency, and further enhancement in the initial cell discharge potential for that of the nanoscopic VB₂ anode (greater than 1.0 V) as shown in Figure 2. The coulombic efficiency of the nanoscopic material has fallen from ∼70 to ∼52% with scale-up; however, the 600% increase in packed anode post-discharge provides sufficient discharge product for analysis. Using the 30 mAh cells that contained thicker anodes, the nanoscopic material still showed significantly higher voltages and capacities compared with the macroscopic material.

Zoom In Zoom Out Reset image size

To further understand the significant difference in the electrochemical discharge of the nanoscopic and macroscopic material, XPS measurements were carried out to characterize the concentrations and oxidation states of the elements at the surface region (top ∼10 nm of the surface) of the unreacted macroscopic and nanoscopic VB₂ materials (prior to discharge).

The XPS showed clear differences in the atomic concentration of V, B, and O at the surface region of the macroscopic and nanoscopic VB_2, as shown in Table I. From the data, comparable amounts of V were observed for both macroscopic and nanoscopic VB₂. However, the concentration of B at the surface of macroscopic VB₂ is much lower than that on nanoscopic VB₂. In contrast, the oxygen concentration on macroscopic VB₂ is much higher compared to nanoscopic VB₂.

The V2p, B1s, and O1s regions of the XPS spectra were evaluated to determine the specific species present at the surface region of the macroscopic and nanoscopic VB₂ materials, as presented in Figure 3. The peak positions from fitting, their relative concentrations, and assignments for the V, B, and O species are shown in Table II.

Zoom In Zoom Out Reset image size

The V2p, B1s, and O1s regions of the XPS spectra show significant differences in the type and relative concentration of species present at the surface of macroscopic and nanoscopic VB₂. As shown in Figure 3a, in the V2p region, the spectra are composed of a pair of V2p^1/2 and V2p^3/2 peaks due to spin-orbit peak splitting. As presented in Figure 3c, the V2p^3/2 peak of the macroscopic VB₂ can be fitted to two peaks at 516.3 eV and 517.6 eV, which are attributed to VO₂/V₂O₄ (V⁴⁺) and V₂O₅ (V⁵⁺) species respectively.²⁴ In contrast, the V2p^3/2 peak of nano-VB₂ contains three peaks from fitting analysis as shown in Figure 3b, and the peak positions of 512.5 eV, 513.6 eV, and 515.4 eV are attributed to elemental vanadium, VB₂, and a V₂O₃ (V³⁺) species respectively.²⁴ The peak position of 513.6 eV attributed to VB₂ is similar to the reported value of 513.7 eV for VB₂ prepared from a high temperature synthesis process.²³ The existence of elemental V at the surface of nano-VB₂ is due to the un-reacted vanadium precursor. Interestingly, while the V-containing species at the surface of macroscopic VB₂ are vanadium oxides, nanoscopic VB₂ contains VB₂ at the surface in addition to elemental vanadium and a vanadium oxide species.

As shown in Figures 3d and 3f, the XPS spectra of macroscopic VB₂ in the B1s region can be fitted to peaks at 193.9 eV and 192.8 eV, assigned to H₃BO₃ and B₂O₃, respectively.²⁴ For nano-VB₂, shown in Figure 3e, the B1s spectra can be fitted to three peaks at 187.3 eV, 187.8 eV, and 192.3 eV, which are attributed to B(0), VB₂, and B₂O₃, respectively. The B1s peak position of 187.8 eV for boron in VB₂ is in the same range but slightly lower than the peak position reported for VB₂ (188.2 eV),²³ which indicates a slightly different chemical environment for boride at the surface of nano-VB₂. The existence of boron in nano-VB₂ is due to the un-reacted boron precursor.

The XPS spectra in the O1s region was also analyzed and is shown in Figures 3g–3i. Oxygen detected by XPS on the surface of VB₂ may be introduced by surface oxidation of B and V, as well as the adsorption of H₂O. As shown in Figure 3i, fitting analysis of macroscopic VB₂ shows three peaks at 530.3 eV, 531.0 eV, and 532.7 eV attributed to oxygen in VO₂(V⁴⁺) oxide, V₂O₅ (V⁵⁺) oxide, and B₂O₃ respectively.²⁴ Nano-VB₂ exhibits peaks at 530.5 eV and 532.1 eV attributed to oxygen within VO₂(V⁴⁺) oxide or V₂O₃(V³⁺) oxide, and B₂O₃ respectively.²⁴ It should be noted that, the O1s peak of oxygen from H₂O has a binding energy of 534.4 ± 1.8 eV, which is not observed in our case. Therefore, the oxygen present in both macroscopic and nanoscopic VB₂ is not from adsorbed water.

As shown in Figure 3 and Tables I and II, it is clear from the XPS analysis that nanoscopic VB₂ has a very different surface structure compared with macroscopic VB₂. Interestingly, the macroscopic VB₂ does not show the presence of VB₂ on the surface, but rather it shows oxides. In contrast, the nanoscopic VB₂ shows the presence of VB₂ at the surface region.

The surface structure of both macroscopic and nanoscopic VB₂ is related to the synthetic conditions, additional surface treatments, and atmospheric storage conditions (oxygen and/or water). The low observed surface oxygen content of nanoscopic VB₂ is related to the synthesis and storage conditions used: elemental vanadium and boron were combined in a glove box with ≤10 ppm H₂O and O₂, sealed and transferred to the ball mill, and then transferred back into the glove box. Although handled in our laboratory in the glove box similar to nanoscopic VB₂, commercial macroscopic VB₂ may have higher surface oxygen content due to synthetic or prior storage conditions.

The data and analysis shows that the as-prepared materials have very different surface structures, which in turn affects the electrochemical properties. Since the macroscopic VB₂ has a significantly higher concentration of oxides present at the surface region compared with nanoscopic VB₂, this may result in an additional resistive layer that affects the ability of hydroxides to interact with the underlying VB₂ structure. The additional resistance resulting from the oxide layer is supported by our prior work, which showed that the interfacial charge transfer resistance of the macroscopic material was higher for the macroscopic VB₂ compared to nanoscopic VB₂.²⁰

The ∼30 mAh cells yield sufficient discharge product for effective FTIR and XRD characterization. For these analyses, a pure 8 M KOH electrolyte was employed, rather than the previous 8 M electrolyte containing 4 M KOH and 4 M NaOH.²¹ This change of electrolyte leads to no discernable change in the discharge voltage or to the coulombic efficiency, but provides a simpler (single cation) basis for the discharge product analysis. Effective XRD of the product requires a slow 2Θ scan (6 hours from 2Θ = 5 to 80) due to the high carbon (30% Timcal) content. Additionally, the data is sequentially averaged every 10 data points, which does not result in any peak loss and improves the signal to noise ratio.

The XRD of the deeply discharged 30 mAh VB₂/air cell product does not exhibit remaining VB₂ and provides a strong match to the Inorganic Crystal Structure Database (ICSD) online spectra of KVO₃ shown as the blue curve in Figure 4. It is clear that as predicted in the initial discharge mechanism, pentavalent vanadium is present, V(V), and in this case takes the form of the salt KVO₃. Interestingly, the boron product is not evident, presumably existing as an amorphous rather than a crystalline product. In the subsequent section the boron product will be characterized by IR spectroscopy.

Zoom In Zoom Out Reset image size

The XRD analysis provides useful information of the vanadium product of the VB₂/air cell discharge. In order to determine the predominate boron species, further testing was performed via solubility analysis. Interestingly, one viable product, KVO₃, was found to be highly soluble (5.3 molar), whereas a boron product was found to exhibit a solubility behavior similar to zincate discharged in zinc/air batteries. In that latter case the solubility is dynamic, initially dissolving and then precipitating. Similarly, the equation 3 generalized predicted boron product, B₂O_3, was initially soluble in an 8M KOH, but after 48 hours had precipitated back out from solution. In order to determine the predominate boron species, further testing was performed prior to FTIR analysis. Vacuum filtration was used to collect the precipitate that was identified, via FTIR absorption analysis, to be potassium tetraborate tetrahydrate, K₂B₄O₇, shown in Figure 5. The formation of the precipitate from the battery electrolyte suggests that K₂B₄O₇, and not B₂O₃, is the principal B(+3) product when the VB₂/air cell is discharged in the KOH electrolyte.

Zoom In Zoom Out Reset image size

Figure 6 compares FTIR absorption of the 30 mAh VB₂/air battery discharge product. As seen in Figure 6, the discharge product spectrum contains a mix of the separate pure component spectra of pure KOH, K₂B₄O₇ and KVO₃ (figure top), and the battery discharge product is a poor fit to the alternate alkali cation free and B₂O₃ and V₂O₃ products (figure bottom). The K₂B₄O₇ and KVO₃ products strongly overlap in the 650 to 1650 cm⁻¹ domain. In the next section ATR FTIR is used to distinguish between these two product species.

Zoom In Zoom Out Reset image size

Although absent from the XRD pattern of the discharge product in Figure 4, ATR FTIR provides further evidence (in addition to the IR absorption evidence of Figures 5 and 6) of the boron product from the original VB₂ reactant. We hypothesize that the boron product is highly amorphous in character and it is therefore difficult to characterize with XRD. This hypothesis is supported by results comparing the ATR FTIR spectrum of the VB₂ discharged anode with that measured for KOH, B₂O₃, K₂B₄O₇, V₂O₅ and KVO₃. The ATR FTIR spectra in Figure 7 show the dominant feature between 1900 cm⁻¹ and 2300 cm⁻¹ is from the carbon added to the anode to improve electrical conductivity. The spectrum of this carbon, Timcal C65, is observed before the discharge and with or without mixing 70% by weight nanoscopic VB₂. Similarly as with the XRD analyses the high degree of absorption by this carbon black additive adds noise.

Zoom In Zoom Out Reset image size

ATR FTIR over a broader frequency range was used for further comparison of the spent anodic material. The data shown in Figure 8 presents evidence for KVO₃ and K₂B₄O₇ as the primary discharge products extracted from an approximate 30 mAh discharged nanoscopic VB₂/air battery. Despite large variations to the baseline, due to Timcal there are notable absorption peaks, which provide evidence of the isolated discharge products. Spectral features located at approximately 825 cm⁻¹, 910 cm⁻¹, and 995 cm⁻¹ show strong correlations between the spent battery material and K₂B₄O₇. Specifically, peaks at 680 cm⁻¹ and 970 cm⁻¹ may be attributed to KVO₃, rather than K₂B₄O₇. Additionally, features attributed to KVO₃ are located at 895 cm⁻¹ and 910 cm⁻¹. This provides corroboration to the XRD evidence that KVO₃ is the principal vanadium discharge product of the VB₂/air battery, and is separate from the principal boron product, K₂B₄O₇.

Zoom In Zoom Out Reset image size

There is little evidence for the IR absorption of B₂O₃ in the discharge product as shown in Figure 8, while the V₂O₅ spectrum in the figure exhibits general features evident in the discharge product. Absorption peaks in the 825 cm⁻¹ and 995 cm⁻¹ region overlap with features attributed to KVO₃ and K₂B₄O₇, respectively. Specific absorption peaks observed in the discharge product at 895 cm⁻¹ and 910 cm⁻¹ are not evident in the V₂O₅ spectrum. Despite uncertainties in the absorption baseline of the spent anodic material due to the large global IR absorption of the Timcal, these additional features are accounted for when KVO₃ is considered as the vanadium discharge product. From FTIR and ATR FTIR analysis of the spent anodic material and expected discharge products, it is evident that the VB₂/air discharge reaction is:

Future work will show analysis of XPS measurements of charged and spent macroscopic VB₂. The discharged VB₂ samples are obtained from anodes equivalent to a capacity of 5 mAh of VB₂ and discharged at a load of 3000 ohm.