SEARCH

SEARCH BY CITATION

Keywords:

  • lithium-sulfur battery;
  • lithium sulfide;
  • carbon nanotube paper;
  • sandwiched electrodes;
  • electrochemical performance

As the petroleum resources deplete and the concern on environmental pollution increases, utilization of renewable energies (e.g., solar and wind) and the adoption of electric vehicles are becoming more desirable. However, the development of an electrical energy storage system that can meet the rigorous requirements on the weight and volume for electric vehicles or on the cycle life and cost for stationary storage is a challenge. Li-ion batteries, representing the highest energy density battery chemistry, are believed to be one of the most promising technologies. However, the capacities of the current insertion oxide cathodes (e.g., layered and spinel oxides) have reached their limits of <300 mAh g−1,[1] which forces the materials community to develop alternative high-capacity cathode materials that can support multiple electrons per molecule, such as sulfur and oxygen.[2]

Lithium-sulfur (Li-S) batteries, first developed in 1960s,[3] have attracted much attention in recent years as sulfur offers a high theoretical capacity of 1672 mAh g−1.[4] However, the low electronic conductivity of sulfur and its reaction products, along with the dissolution of polysulfide intermediates into liquid electrolyte and the consequent shuttling effect between the anode and cathode, makes it difficult to achieve high capacity and practical cycle life. Significant improvements in the utilization of sulfur and cyclability have been achieved by smartly designed sulfur-carbon nanomaterials,[5] core-shell structured composites,[6] and efficient trapping configurations for polysulfides.[7] However, although the use of sulfur as the cathode material has many advantages (e.g., low cost, abundance, and high energy), lithium metal or lithiated anodes are required as no lithium is present in the initial stage of the sulfur cathode. Unfortunately, lithium metal anode poses a significant safety hazard and it can react with the polysulfide ions diffused from the cathode, which limits capacity retention and cycle life. Lithium sulfide (Li2S), the end discharge product of sulfur with a theoretical capacity of 1169 mAh g−1, is more desirable to be the cathode material than sulfur. Being in the fully lithiated state, Li2S cathode will allow the use of lithium-free anodes such as Si, Sn, Sb, or metal oxides. Although a few approaches have been pursued with Li2S recently, showing promising results,[8, 9] a more facile and scalable strategy is needed to tap the full potential of the Li2S cathode in practical cells.

Here, we report a novel sandwiched cathode configuration composed of pristine Li2S powder in between two layers of self-weaving, binder-free carbon nanotube (CNT) electrodes, as shown in Figure 1a. This approach is inspired from our previous work on the interlayer electrode configuration with slurry-cast sulfur electrodes,[10] but with an ability to use pristine Li2S powder instead of sulfur and with a simpler electrode structure. The difference in the chemical state of sulfur in the initial electrodes makes the first electrochemical reactions to be either the oxidation of Li2S or reduction of sulfur. The functions of the two CNT electrodes are three-fold: i) efficient electron conduction within the sandwiched electrode and between the current collector; ii) fast ion transport through the nanospace within the carbon nanotube electrode; and iii) trapping of charged and discharged products within the sandwiched electrode upon cycling. Since soluble lithium polysulfides are produced and could diffuse out of the sandwiched electrode during cycling, LiNO3 additive is still used in the electrolyte to reduce the shuttle effect and ensure good Coulombic efficiency and long cycle life.[9]

image

Figure 1. a) Schematic showing the structure of a CNT/Li2S/CNT sandwiched electrode (CNT: carbon nanotube) with a scanning electron microscopy (SEM) image of the CNT electrode. The bold arrows depict the flow of lithium ions during charge and discharge in cells. b) Cyclic voltammograms of a half cell with the sandwiched electrode at a potential sweep rate of 0.025 mV s−1 between 1.8 and 3.0 V, including an initial potential sweep from the open circuit voltage to 4 V.

Download figure to PowerPoint

Li2S has a high electronic resistivity and low lithium ion diffusivity, which leads to a high overpotential at the beginning of charging to overcome the energy barrier, as reported by Yang et al.[9] We adapted this approach to activate Li2S within the electrodes due to the large particle size of the pristine Li2S powder. The CV plot shown in Figure 1b reveals the electrochemical behavior of the sandwiched electrode in a half cell. The potential was swept from the open-circuit voltage (≈2.28 V) to 4.0 V initially followed by a sweeping between 1.8 and 3.0 V, which is the working voltage range of the cells cycled at low rates. Two obvious cathodic peaks can be seen with onset potentials at approximately 2.45 and 2.15 V, corresponding, respectively, to the transitions of sulfur/high-order polysulfides to low-order polysulfides and then to the end discharge product Li2S.[11] The following anodic sweep accompanies a broad peak starting at 2.3 V with a distinguishable small shoulder peak, which is believed to be a feature of the active material within confined structures.[12] The CV results show that the cells behave like Li-S batteries after the initial activation, which agrees well with Yang's report.[9] In addition, the 10th cycle matches well with the 1st, 2nd, and 5th cycles without any peak shift and noticeable peak decrease, indicating excellent electrochemical stability of the sandwiched electrode. The superior stability is attributed to the favorable CNT electrode structure mentioned above.

Figure 2a shows the voltage profile of the cell as a function of time. The sharp peak at the beginning of 1st charge represents the energy barrier of phase nucleation of lithium polysulfides to overcome with the micron-sized Li2S particles to render Li2S to be active.[9] The short voltage dip followed by the two prolonged high-voltage plateaus suggests a two-phase reaction, which is distinct from that in the conventional Li2S/carbon/PVdF electrodes in Yang's report.[9] Once the polysulfide phase nucleates after the initial barrier, the two-phase reaction occurs within the network of intertwined CNT electrodes. The diffusion of lithium ions would be slow due to the highly localized and viscous lithium polysulfides formed around un-reacted Li2S for further de-lithiation within the sandwiched electrode. This kinetic barrier attributes to the high overpotential after the short voltage dip in the 1st charge. The 1st discharge as well as the subsequent discharges show two voltage plateaus which resemble the CV plots and are similar to those of conventional sulfur electrodes. The following charges show a complete two-plateau charging process unlike the 1st charge indicating the discharged product is much more active than the pristine Li2S within the sandwiched electrode. The shortening of the second charge voltage plateau with increasing cycle number indicates an evolving optimized process of the electrode.

image

Figure 2. a) Voltage vs. time profile of the first three cycles of the cell. The cell was initially charged to 4 V at C/20 rate and then cycled at 1.8–3.0 V at C/10 rate. b) X-ray diffraction (XRD) patterns of the freshly made, 1st charged, and 1st discharged electrodes.

Download figure to PowerPoint

To have further understanding, two additional experiments with low 1st charge voltage and un-sandwiched electrode configuration were conducted. Figure S1 (Supporting Information) shows the voltage profile of the cell that was charged to 3.2 V. The subsequent discharges show two voltage plateaus, but only ≈50% of the discharge capacities were achieved in comparison to those obtained with the cells charged to 4 V, indicating a high 1st charge voltage is needed to activate most of the Li2S within the sandwiched electrode. Figure S2 (Supporting Information) shows the voltage profile of the cell with only one layer of CNT electrode in between the separator and Li2S powder. After the 1st charge, only the high voltate plateau was obtained in the 1st discharge and the following charge and discharge show high overpotential with partial capacities. This experiment demonstrates that the sandwiched electrode configuration is essentioal for achieving good performance with Li2S.

To assess the phase change of Li2S within the sandwiched electrodes, ex situ XRD was performed and the pattern is shown in Figure 2b. The peaks below 25° in the three samples are due to the Kapton films. In the initial stage, sharp peaks corresponding to the (111), (220), and (311) planes of Li2S are clearly shown, indicating that the pristine Li2S is highly crystalline. It has to be noted that the peak itensities are much less pronounced than those with the pristine Li2S powder sample (Figure S3, Supporting Information), which is due to the coverage of the Kapton films on the sample, shadow effect of the CNT electrode, and residual liquid electrolyte. After the 1st charge, these peaks almost disappear indicating the conversion of most crystalline Li2S particles to nearly amorphous polysulfides or elemental sulfur. Due to the weakened intensities, un-reacted Li2S crystals could be present within the oxidized products in the sandwiched electrode. The following discharge (1st discharge) results in a reappearance of the peaks of crystalline Li2S. The much lowered peak intensities indicate that the re-formed Li2S is less crystallized than the pristine Li2S in the freshly made electrode.

The morphology change within the sandwiched electrodes at different stages during the 1st cycle was monitored by SEM, as shown in Figure 3a, b, and c. A big chunk of Li2S powder can be seen in Figure 3a on the CNT layer in the freshly made electrode. After the 1st charge, no Li2S particles are seen, but melt-like materials within the carbon nanotube matrix can be seen (Figure 3b), which could be a mixture of polysulfides, sulfur, and lithium salt from the electrolyte. The EDS elemental (sulfur and carbon) mapping shown in Figure S4 (Supporting Information) indicates a uniform distribution of these materials. Particles appear again within the electrode after the 1st discharge as shown in Figure 3c, which are Li2S as indicated by the XRD (Figure 2b) and EDS sulfur mapping (Figure S5, Supporting Information). This conversion process can be depicted as in the schematic diagram in Figure 3d. Solid Li2S powder is mostly converted to soluble polysulfides or sulfur, which occupy the spaces between nanotubes within the sandwiched electrode after the 1st charge. Li2S particles re-nucleate after the following discharge.

image

Figure 3. a–c) SEM images of the inner area within the freshly made electrode, 1st charged electrode, and 1st discharged electrode. d) Schematic showing the structure and morphology changes of the sandwiched electrode during charge and discharge. e) Electrochemical impedance analysis of the cycled cell at different stages.

Download figure to PowerPoint

The EIS results of the cell at different stages, shown in Figure 3e, also support this transition process. The freshly made cell shows a small semicircle in the high-medium frequency range, which is attributed to the interfacial and charge transfer resistance,[11] indicating the excellent electrochemical environment provided by the sandwiched electrode structure. After the following charge, the semicircle significantly increases, which is due to the increase in interfacial impedance associated with the drop in surface area within the electrode occupied by the soluble lithium polysulfides and the increased charge transfer resistance of polysulfides/sulfur diffused within the sandwiched electrodes occupying most of the spaces between the CNTs. The slope tail in the low frequency range, which is associated with the diffusion impedance, indicates the presence of un-reacted Li2S within the charged electrode. After the 1st discharge, the semicircle decreases, which is indicative of increase in the surface area due to the re-crystallization of Li2S. The re-crystallization of Li2S could occur at the un-reacted Li2S sites within the sandwiched electrode. The larger impedance compared to that before cycling is also because of the increase in charge transfer resistance associated with the diffused active materials within the sandwiched electrode; in other words, more active materials are exposed to be in contact with the CNTs after the 1st cycle.

The cells with the sandwiched electrodes were cycled at different rates. Representative voltage profiles of the stable 5th cycles are shown in Figure 4a. At the low rate of C/10, a high capacity of 838 mAh g−1 is achieved without noticeable overpotential during charge and discharge, which corresponds to ≈72% of the theoretical capacity of Li2S. As the rate increases, the capacity decreases to 787 mAh g−1, 652 mAh g−1, and 578 mAh g−1, respectively, at C/5, C/2, and 1C rates. All curves show clear two-voltage plateaus indicating a favorable electrochemical environment in which the cathode reactions occur. The increased overpotential with increasing rate is mainly due to the long Li-ion diffusion path in the thick electrode. The extended cycle life at different rates is shown in Figure 4b. These cells exhibit steady cycle life after the 1st cycles and maintain capacities of 680 mAh g−1, 613 mAh g−1, 565 mAh g−1, 502 mAh g−1, respectively, at C/10, C/5, C/2, and 1C rates over 100 cycles. The Coulombic efficiency is low in the first cycles due to the prolonged activation charge. However, all cells retain a Coulombic efficiency of over 80% in most cycles. As the rate increases, the Coulombic efficiency increases, which is >95% at 1C rate over 100 cycles. At low rates, the soluble polysulfides could diffuse out, resulting in low Coulombic efficiency. At high rates, only Li2S on the surface of the bulky Li2S could be active, which can be completely trapped in the middle of the sandwiched electrode, resulting in high reversibility.

image

Figure 4. a) Voltage vs. specific discharge capacity profiles of the 5th cycles of the cells at C/10, C/5, C/2, and 1C rates. b) Cyclability and Coulombic efficiency of the cells at C/10, C/5, C/2, and 1C rates. The capacity values are in terms of the Li2S active mass within the sandwiched electrodes.

Download figure to PowerPoint

The sandwiched electrodes with different mass loadings of Li2S were also evaluated. Figure 5 presents the voltage profiles and cycle life at C/2 rate with loadings of approximately 1, 2, and 3 mg cm−2, corresponding to the mass percentages of Li2S within the electrodes of, respectively, 23, 37, and 47 wt%. The capacities obtained are 752 mAh g−1, 654 mAh g−1, and 442 mAh g−1 respectively, as shown in Figure 5a. The low capacities obtained with high loadings mean a significant amount of Li2S remain inactive at C/2 rate, which, however, does not affect the cyclability as shown in Figure 5b. The utilization of Li2S can be further improved with a uniform distribution of Li2S powder within the sandwiched electrode, increasing the contact area of Li2S with carbon nanotubes. In addition, the loading can be raised more by applying layer-by-layer configuration during the electrode fabrication. The superior performance obtained with different loadings of Li2S indicates this simple electrode fabrication method is versatile and scalable.

image

Figure 5. a) Voltage vs. specific discharge capacity profiles of the 5th cycles of cells with different Li2S loadings (1 mg cm−2, 2 mg cm−2, and 3 mg cm−2) at C/2 rate. b) Cyclability of the cells with different Li2S loadings at C/2 rate. The capacity values are in terms of the Li2S active mass within the sandwiched electrodes.

Download figure to PowerPoint

In summary, we have demonstrated a sandwiched electrode with bulky Li2S powder within two layers of self-weaving carbon nanotube electrodes. The high capacities and excellent cyclability obtained at C/10, C/5, C/2, and 1C rates are attributed to the unique sandwiched electrode architecture, facilitating ion and electron transport and trapping of cycled products within the electrode. This electrode allows ready use of pristine Li2S powder and lithium-free anodes, which is a critical problem facing the use of rechargeable Li-S batteries with Li metal anode. The results shown here also provide a guideline for designing electrode architectures and cell configurations to achieve high performance Li-S batteries. With further improvement in the distribution of Li2S powder within the sandwiched electrode, this method could be readily adapted in commercial Li-S batteries without metallic lithium anode for practical applications.

Experimental Section

  1. Top of page
  2. Experimental Section
  3. Acknowledgements

Chemicals and Materials: Lithium trifluoromethanesulfonate (LiCF3SO3, 98%, Acros Organics), lithium nitrate (LiNO3, 99+%, Acros Organics), dimethoxy ethane (DME, 99+%, Acros Organics), 1,3-dioxolane (DOL, 99.5%, Acros Organics), and lithium sulfide (Li2S) powder (99.5%, Acros Organics) were purchased and used as received. CNTs were obtained from a commercial source, which were synthesized by a chemical vapor deposition process, possessing a “self-weaving” property.[10]

Materials Preparation and Cell Fabrication: The CNT electrode was prepared as follows: CNTs (75 mg) were dispersed in de-ionized water (750 mL) by high-power ultrasonication for 15 min with the addition of isopropyl alcohol (20 mL) to wet the CNTs. The CNTs were collected by vacuum filtration and washed with de-ionized water, ethanol, and acetone several times. The CNT paper was formed and dried for 24 h at 100 °C in an air-oven and then was peeled off the filter paper. Circular CNT discs (1.2 cm in diameter, 40–50 μm in thickness, and 1.9–2.3 mg in mass) were then cut. The sandwiched electrodes were fabricated in CR2032 coin cells in an Argon-filled glove box. A known amount of Li2S powder, e.g., 1.2 mg (1 mg cm−2), 2.3 mg (2 mg cm−2), and 3.4 mg (3 mg cm−2), was weighed on a CNT disc, and the Li2S powder was spread out by a spatula on the disc as uniformly as possible. Another CNT disc was the added onto the top of the Li2S powder. For the fabrication of the un-sandwiched electrode with only one layer of CNT electrode, the Li2S powder was spread directly onto the coin cell case followed by a layer of CNT disc. A certain amount of electrolyte (1 M LiCF3SO3 and 0.1 M LiNO3 in DME/DOL (1:1 v/v)) was added into the sandwiched electrode, followed by a Celgard 2400 separator, additional electrolyte, and lithium metal anode. Finally the cells were crimped for electrochemical evaluation.

Characterization: The XRD data were collected on a Philips X-ray diffractometer equipped with CuKα radiation in a step of 0.02°. Electrode samples were collected from cells that were freshly made, after 1st charge, and after 1st discharge in the glove box. The XRD pattern of pristine Li2S powder was also collected for a comparison. These XRD samples were covered by Kapton films. Morphological characterizations were carried out with a FEI Quanta 650 SEM. The SEM samples were collected by peeling off a layer of the CNT electrode from the XRD samples. The morphology was examined in the area in between the two CNT electrodes. The elemental mapping results were examined with an energy-dispersive spectrometer (EDS) attached to the FEI Quanta 650 SEM.

Electrochemical Characterization: The cells for electrochemical evaluation contained a Li2S loading of approximately 2 mg cm−2 except those indicated. Cyclic voltammetry data were collected with a VoltaLab PGZ402 with an assembled coin cell initially from open circuit voltage to 4.0 V and then between 1.8–3.0 V at a scan rate of 0.025 mV s−1. In the battery cycling measurements, cells were galvanostatically charged to 4.0 V or 3.2 V at C/20 rate (1C = 1169 mA g−1) initially with an Arbin battery test station, then cycled between 1.8–3.0 V at C/10 and C/5 rates, 1.7–3.0 V at C/2 rate, and 1.6–3.0 V at 1C rate. The decreasing cutoff voltage with increasing rate is to ensure full discharge voltage profiles due to the large overpotential at high rates as discussed later. The capacity values shown in this paper are calculated based on the mass of Li2S within the sandwiched electrodes. EIS data were collected with a computer interfaced HP 4192A LF Impedance Analyzer in the frequency range of 1M Hz to 0.1 Hz with an applied voltage of 5 mV and Li foil as both counter and reference electrodes.

Acknowledgements

  1. Top of page
  2. Experimental Section
  3. Acknowledgements

This work was supported by Seven One Limited.