High-Capacity Mg–Organic Batteries Based on Nanostructured

6 days ago - State Key Laboratory of High Performance Ceramics and Superfine Microstructure, Shanghai Institute of Ceramics, Chinese Academy of Scienc...
7 downloads 7 Views 3MB Size
Subscriber access provided by - Access paid by the | UCSB Libraries

High Capacity Mg-Organic Batteries Based on Nanostructured Rhodizonate Salts Activated by Mg-Li Dual-Salt Electrolyte Jing Tian, Dunping Cao, Xuejun Zhou, Jiulin Hu, Minsong Huang, and Chilin Li ACS Nano, Just Accepted Manuscript • DOI: 10.1021/acsnano.7b09177 • Publication Date (Web): 04 Apr 2018 Downloaded from http://pubs.acs.org on April 4, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

High Capacity Mg-Organic Batteries Based on Nanostructured Rhodizonate Salts Activated by Mg-Li Dual-Salt Electrolyte

Jing Tian,a,b Dunping Cao, a Xuejun Zhou, a Jiulin Hu, a,b Minsong Huang, a,b and Chilin Li a,* a

State Key Laboratory of High Performance Ceramics and Superfine Microstructure, Shanghai Institute of Ceramics, Chinese Academy of Sciences, 1295 Ding Xi Road, Shanghai, 200050, China. E-mail: [email protected]

b

University of Chinese Academy of Sciences, Beijing 100039, China.

Abstract: Magnesium battery is a promising candidate for large-scale transportation and stationary energy storage due to the security, low cost, abundance and high volumetric energy density of Mg anode. But there are still some obstacles retarding the wide application of Mg batteries, including poor kinetics of Mg-ion transport in lattices and low theoretical capacity in inorganic framework. Mg-Li dual-salt electrolyte enables the kinetic activation by dominant intercalation of Li-ions instead of Mg-ion in cathode lattices without the compromise of stable Mg anode process. Here we propose a Mg-organic battery based on renewable rhodizonate salt (e.g. Na2C6O6) activated by Mg-Li dual-salt electrolyte. The nanostructured organic system can achieve a high reversible capacity of 350-400 mAh/g due to the existence of high-density carbonyl groups (C=O) as redox sites. Nanocrystalline Na2C6O6 wired by reduced graphene oxide enables a high-rate performance with 200 and 175 mAh/g at 2.5 (5 C) and 5 A/g (10 C) respectively, which also benefits from high intrinsic diffusion coefficient (10-12-10-11 cm2/s) and pesudocapacitance contribution (> 60%) of Na2C6O6 for Li-Mg co-intercalation. The suppressed exfoliation of C6O6 layers by a firmer non-Li pinning via Na-O-C or Mg-O-C and dendrite resistive Mg anode lead to a long-term cycling for at least 600 cycles. Such an extraordinary capacity/rate performance endows Mg-Na2C6O6 system high energy and power densities up to 525 Wh/kg and 4490 W/kg (based on active cathode material) respectively, exceeding the

ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

level of high-voltage insertion cathodes with typical inorganic structures.

Keywords: nanostructured rhodizonate salts, Mg-organic batteries, Mg-Li dual-salt electrolyte, multi-electron transfer reaction, pseudocapacitance effect

Although rechargeable lithium-ion batteries (LIBs) have achieved a huge success in mobile equipment, the application of LIBs in large-scale grid and transportation energy storages is blocked due to the safety concern (Li dendrite propagation) and high cost (Li resource shortage). Compared with lithium, magnesium resource in earth is more abundant and metal Mg has a higher volumetric capacity (3833 mAh cm-3 vs. 2205 mAh cm-3 for Li).1-3 In addition, the plating and stripping of Mg metal as anode are dendrite-free with less risk of short circuit. Since the 0.6 V higher reduction potential of Mg than Li would not seriously compromise the energy density of Mg-based batteries,1-3 they have received much more attention recently as potential alternative systems/technologies beyond LIBs. Nevertheless, a big challenge is that highly polar Mg2+ limits the pick-out of well-matched crystal structures including open framework mineral prototypes.4 Even though Mg2+ can be fortunately inserted into the lattices during the first discharge, its extraction process is discouraging usually accompanying with poor reversibility and structural collapse. For instance, tunnel-framework-type α-MnO2 prefers to undergo conversion reaction with the formation of Mn and Mg oxides especially at grain surface, rather than complete intercalation reaction to MgxMnO2.5,6 Apart from structure destruction, the poor reversibility may be also caused by large voltage polarization (or overpotential) and accumulative trapping of Mg-ions by interaction with channel ligands as a consequence of sluggish diffusion and high charge density of Mg2+.4 Till now, only few materials (e.g. Chevrel Mo6S8 and thiospinel Ti2S4) allow reversible insertion and extraction of Mg2+, which however still require critical running conditions, e.g. low current density (< 1C) and enhanced temperature (60oC).7,8 Most recently, the design of cation-deficient oxide appears to be an effective route to enable reversible and fast

ACS Paragon Plus Environment

Page 2 of 34

Page 3 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

intercalation of Mg.9 The Mg-insertable sites are limited in these unusual phases, leading to a reversible capacity less than 160 mAh/g with moderate discharge voltages between 1.0 and 1.3 V. The low energy storage retards the practical application of Mg batteries dominantly driven by Mg-ions. In order to circumvent the sluggish Mg2+ transport in host lattices and activate the utilization of traditional high-voltage structure prototypes, a dual-salt hybrid system characterized by Li-intercalation or Mg-Li co-intercalation at cathode and Mg plating/stripping at anode is proposed recently.10,11 The ratio of co-intercalated Mgand Li-ion numbers depends on the choice of phase structures.12-16 The electrolyte solution simultaneously contains lithium and magnesium salts and serves as Li-ion reservoir for kinetically modified cathode reaction. When charging, Li+ is extracted from the cathode side and Mg2+ is preferentially electrodeposited over Li+ on the anode, and the discharge process is just the opposite. Therefore this system is expected to be safer than LIBs in terms of intrinsically dendrite-free anode reaction. The alleviated anode roughening is also beneficial to the achievement of higher rate charging compared with the use of Li metal anode.17 The high-voltage (> 2 V) Mg-Li hybrid battery (MLB) is enabled by using Mo current collector or pouch cell architecture as long as its reaction voltage does not exceed the electrochemical window of dual-salt electrolyte.18-21 LiFePO4, LiMn2O4 and prussian blue analogues have been successfully activated in MLB system most recently, however their capacities (1.5 V), where the peaks become weak and more (roughly from two to three ones) with some of them appearing at higher voltage closing to 2 V. From the 2nd to 5th cycle, the reduction peak at 2 V become intensive at the cost of the weakening of peak at 1.5 V. Meantime the peaks around 0.7 V undergo less change apart from the appearance of shoulder peak towards higher voltage. The corresponding anodic peaks are highly reversible and locate around 1.2 and 2.2 V, leading to two sets of evident redox couples based on 1 and 2.1 V. The latter (high-voltage redox couple) has the smaller overpotential. For LR, there are also three reduction peaks but with different positions from n-SR (i.e. around 1.8, 0.9 and 0.7 V) during the first cathodic process. Although these peaks are sharper than for n-SR, they broaden seriously from the second cycle and their intensity becomes weak continuously with the progress of cycling. For the anodic process, the low-voltage peak is still sharp during the early cycles and however quickly degrades into broad peak when reaching to the 5th cycle. Similarly, there are also roughly two sets of redox couples for LR, but with smaller overpotential for the low-voltage one. When narrowing the upper cut-off voltage to 2 V, the remaining redox peaks do not undergo striking change including their intensity and position (Figure S10). The electrochemical profiles of rhodizonate salts in MLBs appear to be similar as in Li-metal batteries (LMBs) apart from the discrepancy of reaction voltage vs. Mg2+/Mg or Li+/Li (Figure S11). It indicates a dominant Li-ion insertion coupled with Mg-ion stripping from Mg anode.11 Unfortunately these LMBs degrade seriously from ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

the early cycles owing to a substantial dissolution of organic active species and its reaction with Li anode. Less corroded MLBs with dendrite-free anode behavior enable a significant improvement on cyclability. XRD, X-ray photoemission spectra (XPS) and Raman spectra were used to analyze the cycled electrodes in order to disclose the reaction mechanism of n-SR, the best cathode for MLBs discussed here. The XRD pattern undergoes significant broadening and position change of peaks with the appearance of some new peaks after initial discharge (Figure S12). The different XRD patterns between pristine and cycled electrodes are caused by irreversible evolution of phase structure. The pristine Na2C6O6 crystal has a space group of Fddd. The original C6O6 stacking is no longer energetically favourable with the injection of additional cation and electron, and the C6O6 layers lie over each other with changed relative location and their layer spacing is expanded. This leads to different stacking manner and space group.34,35 However the accurate stacking evolution is complex especially when smaller sized Li-ions (or Mg-ions) intercalate into the channels pre-supported by Na-ions. The co-existence of heterogeneous cations in lattices causes an additional difficulty on space group assignment at different reaction stages before the loss of crystallinity, compared with exclusive intercalation of Na-ions in Na2C6O6 in previous reports.34,35 The discharged pattern corresponding to new phase is well maintained after the electrode is recharged, resulting in a reaction path different from the initial cycle. The structure irreversibility is also reflected in the curve profile evolution between the first and following discharge, agreeing with previous reports by Yu and Bao et al. even based on sodiation rather than lithiation/magnesiation.32,35 No substantial phase transformation occurs during recharge process apart from slight positive shift of XRD peaks. XPS is a suitable tool to disclose the bonding evolution during electrochemical process (Figure 3a and b). The C 1s spectra consist of the peaks denoting C=C/C-C, C-O and C=O at 285.0, 286.1 and 287.2 eV respectively.26 After discharge, the fraction of C-O component increases substantially at the cost of shrinkage of C=O moiety. The intensity ratio of C-O and C=O decreases after recharge, and but is still higher than for the pristine electrode. It indicates that a fraction of Li or Mg does not debond from already ACS Paragon Plus Environment

Page 10 of 34

Page 11 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

formed Li-O-C or Mg-O-C during recharging, agreeing with the relatively low CE for the first cycle. The O 1s of pristine electrode consists of two peaks denoting C-O-Na and C=O at 532.1 and 533.9 eV respectively.33 After discharge, two new peaks denoting C-O-Li (at 533.1 eV) and C-O-Mg (at 530.4 eV) appear with the shrinkage of C=O peak, indicating a co-intercalation of Li and Mg.26 The Li-intercalation is dominant from the higher intensity of C-O-Li peak than C-O-Mg. Both the new peaks are still residual after recharging due to irreversible trapping of Li and Mg, agreeing with the result of C 1s. The peak at higher binding energy (BE) around 537-538 eV should be assigned to Na KLL signal.35 The potential adsorption of H2O or O2 on electrode surface during sample transfer may also be responsible for the higher BE peak.40,41 Raman spectra (Figure 3c) disclose the existence of vibrations of ring skeleton (from 225 to 750 cm-1) and C=O bonding (at 1500 cm-1).38 These vibration signals do not disappear after discharge, indicating a structure stability of ring skeleton without full occupation of C=O reaction sites. The slight evolution of ring skeleton peaks is likely caused by the hetero-cation-insertion driven ring distortion, which appears to be irrecoverable after charging. The D/G band peaks stem from the carbon paper substrate. Note that the Raman signal of C=O partially overlaps with that of G band in pristine electrode, resulting in that its visualization is not evident. Actually its relative intensity should be comparable to that in the charged state. Its slight displacement after cycling and relative intensity increase after discharging should be associated with the irreversible structure change and the accompanied symmetry evolution of C6O6 layers.42 The galvanostatic intermittent titration technique (GITT) was performed to estimate the cationic diffusion behavior in organic lattices (Figure 4a and b).43,44 The MLB was firstly charged and discharged for 5 cycles at a constant current density of 50 mA/g before GITT measurement, then the battery was operated at a low current density of 10 mA/g for an intermittent time of 2 h (7200 s) and followed by an open circuit relaxation for 10 h. The electrochemical diffusion process of Li+/Mg2+ follows the second Fick's law of diffusion, and their diffusion coefficient D can be calculated by the following equation:43-45 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 34

4   Δ =    ( ≪  ⁄ )   ( ⁄ √) where mB, MB and Vm are mass, molecular weight (214.04 g/mol), molar volume (4.91×10-5 m3/mol) of cathode active material respectively, L and S are the thickness and area of electrode respectively, τ is the intermittent time (7200 s), ∆ES is the difference of open circuit voltages after two adjacent relaxation of 10 h. Since the transient potential (Eτ) shows a linear relationship with the square root of τ as shown in Figure S13, the above equation can be simplified as follows: =

4   Δ   ( ≪  ⁄ )   Δ

where ∆Eτ is the voltage difference between the beginning and termination of one single step under 10 mA/g. The capacity under GITT is close to 400 mAh/g, corresponding to about 3 electron transfer. The voltage hysteresis is likely associated with different bonding sequence between discharge and charge. The estimated diffusion coefficient (D) depends on the reaction voltage and mainly ranges from 10-13 to 10-11 cm2/s. During the discharge process, the D value fluctuates between 10-12-10-11 cm2/s depending on the evolution of quasi-plateaus before the voltage drops below 0.75 V. In the charge process, the D value is stable (around 4.8×10-12 cm2/s) in a wide voltage range of 0.75-2 V before its dropping when closing to the upper cut-off voltage. These high D values for n-SR are responsible for the high-rate performance of this Mg-organic hybrid system. This high-rate performance appears to be related to the pseudocapacitive effect as indicated by the capacitive-like sloped curves (Figure S4b). CV measurement at various scan rates from 0.4 to 1 mV/s was performed to quantitatively estimate the pseudocapacitance contribution (Figure 4c).46,47 The type of electrochemical process can be deduced from log i(V) = b log v + log a (derivative from i(V) = avb), where current (i) as a function of potential (V) obeys a power law relationship with scan rate (v) with a and b as adjustable parameters. In view of overpotential change with scan rate, the independent variable potential is slightly adjusted based on equidistribution principle. The b-value is determined from the slope of the plots of log i vs. log v (e.g. ACS Paragon Plus Environment

Page 13 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

respective CV peak positions shown in Figure 4d). This value is close to 1.0 when the process is capacitance-dominated, whereas close to 0.5 if it is diffusion-controlled (i.e. faradaic intercalation).48,49 Most of the b-values at peak positions (apart from those from the low-voltage couple) are evidently higher than 0.5 (some even approaching 1) from the near-linear fitting of log i-log v plots in the whole rate range, indicating a substantial contribution of capacitance effect. The equation i(V) = k1v + k2v1/2 is used to calculate the respective current contributions from capacitance effect (k1v) and intercalation process (k2v1/2). Both the parameters k1 and k2 are determined from the linear relationship of iv-1/2 and v1/2 based on i(V)/v1/2 = k1v1/2 + k2. Figure 4e displays a typical CV curve at 0.6 mV/s, where the outlined capacitive current (k1v) can be integrated into a large area and is distinguished from the total current (k1v + k2v1/2). The ratio of stored charge contributed by capacitive current at different scan rates is about 60-65 % (Figure 4f). This fraction of capacitive contribution is high and responsible for the good rate performance of organic MLBs. The pseudocapacitive effect for b-SR and LR is also analyzed as a comparison (Figure S14). Note that b-SR electrode displays the highest capacitive fraction (>80%). However this positive effect is counteracted by its relatively low absolute capacity. LR electrode has the lowest capacitive fraction (40-50%), agreeing with its interior rate performance. The discrepancy of pseudocapacitive contribution is in accordance with the capacity retention ratio at higher rate (Figure S14g). The b-SR shows slightly better ratio values than for n-SR, and these values are evidently worst for LR. Note that our MLB system enables a lithiation/magnesiation of Na2C6O6 up to three electron transfer corresponding to 380 mAh/g, which is higher than that based on maximum two electron reaction (~250 mAh/g) when sodiation of Na2C6O6. This improvement is attributed to less steric hindrance for small-sized Li and Mg.32 In MLBs, the residue of Mg-O-C (as proved by XPS) as a consequence of co-intercalation of Mg appears to be helpful to alleviate the exfoliation of C6O6 layers (or electrode pulverization) and achieve a much better cycling performance of rhodizonate compounds than in LMBs. The stable Mg anode process is also responsible for the better cyclability of MLBs than LMBs with undesired Li dendrite ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

growth and anode side reaction, which would consume more active species and accelerate organics dissolution. The electronic conductivity is another bottleneck for performance activation, although Na-substitution in rhodizonate salt has improved the conductivity of Na2C6O6 with a narrower bandgap of 1.1 eV than that of Li2C6O6.38 Therefore we attempted to displace the conventional carbon additive (Super P) by equal amount of rGO in order to strengthen the conductive (wiring) network and electric contact with n-SR. Compared with Super P, rGO is expected to have better capability of electron penetration, and thereby the capacity and rate performance of n-SR-rGO are significantly improved. Note that the reversible discharge capacity is increased to 450 mAh/g at 50 mA/g, and the rate performance is further upgraded with a reversible capacity of 300, 250, 200 and 175 mAh/g at 0.5 (1 C), 1 (2 C), 2.5 (5 C) and 5 A/g (10 C) respectively (Figure 5a). By rGO wiring, the capacitance-like charge/discharge profiles present smaller kinetic polarization under high rates (Figure 5b). The kinetic modification is also reflected from the smaller charge transfer resistance of Mg/n-SR-rGO cell compared with Mg/n-SR-Super-P cell (Figure S15) Binder network is important to improve the cycling stability of organic electrodes in Li or Na batteries, however it takes marginal effect on our case (Figure S16).33,50-53 Renewable rhodizonate-based cathodes show strong advantages in terms of both the energy and power densities over typical high-voltage inorganic structures (e.g. LiFePO4, VO2, LiMn2O4 and Prussian blue analogues) also based on intercalation reaction (Figure 5c, S17 and Table S1).18-21,54 When the power density is 125 W/kg, the energy density of n-SR can even reach to 525 Wh/kg, which is much higher than those of best inorganic insertion materials (< 350 Wh/kg for LiFePO4 and VO2) based on the similar power density. Under the high rates of 2.5 (5 C) and 5 A/g (10 C), its energy densities are still maintained at 218 and 161 Wh/kg corresponding to high power densities of 2608 and 4490 W/kg respectively. We also performed the SEM measurement and energy dispersive spectrometer (EDS) mapping on cycled n-SR cathode and Mg anode after the Mg/n-SR cell underwent 100 cycles at 1A/g, in order to link the good cycling stability with the morphology evolution. One can find that the grain cracking is not serious even under ACS Paragon Plus Environment

Page 14 of 34

Page 15 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

this high current density, although the crossed rods are electrochemically grinded into discrete nods which still retain the original size and smooth surface (Figure 6). These nod-like grains are homogenously embedded in conductive carbon network, guaranteeing a stable cycling performance with sufficiently high electroactivity. In contract, the grain cracking in LR was reported to be drastic by Kang et al.,37 and even leaded to an exfoliation of C6O6 layers which could pulverize and delaminate the grains and accelerate the dissolution of active species. From the EDS mapping of cycled n-SR, Mg element is concentrated on the organic grains as the cases of Na and O for Na2C6O6, rather than homogeneous through the whole selected region as other electrolyte residual (e.g. Cl and Al). It also indicates a co-intercalation of Mg, agreeing with the XPS result of cycled cathode with the emergence of C-O-Mg peak. The pinning via Mg-O-C is likely firmer and takes effect on suppressing exfoliation of C6O6 layers and alleviating grain cracking. From the similarity of electrochemical profiles of cathode in Li-Mg hybrid batteries and Li-metal batteries, the content of intercalated Mg is not high. Otherwise the electrochemical profile would be evidently altered. However the accurate percentage is difficult to be estimated, since the surface residual of Mg-salt and Li-salt for cycled samples cannot be ruled out. The surface species and low concentration of intercalated Mg would retard the quantification especially for the surface-sensitive characterizations of XPS and EDX element mapping, although which enable the indication of the existence of Mg co-intercalation. The percentage of intercalated Mg depends more or less on the rate due to the discrepancy between Mg and Li intercalation kinetics. Based on the above analysis, we propose a redox mechanism of Na2C6O6 dominantly driven by Li (Figure 5d). The first discharge process enables a four-electron reaction from Na2C6O6 to Na2Li4C6O6, which is however oxidized to Na2LiC6O6 (rather than Na2C6O6) based on three-electron transfer. The following reversible reaction roughly occurs between Na2Li4C6O6 and Na2LiC6O6. The irreversibility in the initial cycle is also indicated by the XPS result with the residual of Li-O-C peak after recharging. The CV curves (Figure 9) enable a better discernment for the reaction stages with various electron-transfer number. For n-SR, ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

two-stage reaction is indicated from the appearance of two sets of roughly overlapped redox couples during the reversible cycles. Since the CV peak area of low-voltage (around 1 V) redox couple is roughly double larger than that of high-voltage (around 2.1 V) couple, the low-voltage stage is believed to achieve a two-electron transfer (between Na2Li4C6O6 and Na2Li2C6O6) and the high-voltage stage to enable a one-electron transfer (between Na2Li2C6O6 and Na2LiC6O6). The high-voltage peak intensity for the first cathodic process is roughly double higher than that for the following cathodic process, indicating an irreversible loss of one-electron transfer occurring at high-voltage region. The reaction between Na2Li4C6O6 and Na2Li2C6O6 is quite reversible regardless of the formation of Na2LiC6O6 as indicated from the highly overlapped CV profiles in 0.1-2 V and in 0.1-2.75 V (Figure S10). The better reversibility of CV curves for n-SR than for LR is in accordance with the better capacity retention for the former. SEM images also disclose that the Mg surface is smooth and dense and no dendrite can be observed even after long-term cycling (100 cycles) at high current density (1 A/g, Figure 7). The roughening of Mg anode appears to be not serious and the shallow pits are the interspaces left after Mg stripping. Few residues of glass fibers are also observed. The good preservation of Mg anode morphology is responsible for the much better cyclability of MLBs than LMBs. The Li metal batteries based on n-SR cathode degrade quickly from the early cycles as the dissolved organic species are continuously consumed by reaction with Li anode before Li dendrite growth dominates. We also performed the element analysis of electrolyte extracted from the Mg/n-SR cell after 100 cycles under 1 A/g by inductive coupled plasma emission spectrometer (ICP). The Na-ion concentration in electrolyte is detected to be 0.12 M, indicating a potential extraction of Na from cycled Na2C6O6 and its release into electrolyte. This can be further confirmed by the detectable Na signal on Mg anode disassembled from the corresponding Mg/n-SR cell via EDS mapping (Figure 7). The slight dissolution of (cycled) organic electrode is responsible for capacity fading of Na2C6O6 during the early cycling. However the use of Mg as dendrite-free anode can mitigate the anode roughening caused by the reaction with ACS Paragon Plus Environment

Page 16 of 34

Page 17 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

dissolved organic species. The residual of Na and electrolyte components (e.g. Cl, C) likely further protect Mg surface from side reaction. These coatings appear not to block Mg-ion transport and passivate the surface. These factors retard more dissolution and consumption of organics from cathode, benefiting to the cycling stability of MLBs in the later cycles. The Mg-Li hybrid battery based on organic cathode appears to be a promising candidate for storage batteries, due to the security and dendrite inhibition of Mg anode, low cost and abundance for both anode and cathode, as well as high volumetric energy density of Mg anode and high mass energy density of organics both based on multi-electron reaction. The stabilization of anode would significantly promote the cycling life of batteries, which is one of the key indexes for storage batteries. Since no Mg dendrite growth is found during low-voltage anode process even under harsh cycling conditions, the potential below 1 V can be trustingly utilized. If we raise the cut-off voltage from 0.1 to 0.5 V, the energy density compromise is actually not serious (Figure S18). The presence of Li electrolyte is necessary, since it would dominantly drive the cathode process in view of the better kinetics of Li insertion than for Mg. In this hybrid system, the Li electrolyte would not be seriously reduced and consumed with the appearance of electroplated Li. Therefore the usage of Li electrolyte would not compromise the safety and stability of storage batteries. The capacity would be lowered if much less Li-salt is used. For hybrid Mg-based batteries, a moderate concentration of 1.0-1.5 M Li-salt is enough to achieve the theoretical capacity even for the conversion cathode with more Li+-e couple consumption.17 This concentration is much lower than that (5-10 M) in Li-metal batteries with highly concentrated Li-salt to suppress anode dendrite growth.55,56 Therefore this hybrid system would not compromise serious mass addition compared with the high concentration system.

Conclusion In summary, we reported a kind of Mg-organic batteries with high capacity activated by Mg-Li dual-salt electrolyte. The rhodizonate salt cathodes of high-density carbonyl ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

groups (C=O) display improved capacity performance after substitution by Na-ions (i.e. Na2C6O6) and its nano-sizing. Nanocrystalline Na2C6O6 enables an initial discharge capacity closing to 450 mAh/g and reversible capacity as high as ~350 mAh/g at 50 mA/g. Under much higher 1A/g (2 C), its capacity still has ~200 mAh/g and is preserved at 125 mAh/g after 600 cycles. Wiring by reduced graphene oxide can lead to a further rate upgrade with a reversible capacity of 200 and 175 mAh/g at 2.5 (5 C) and 5 A/g (10 C) respectively. High intrinsic diffusion coefficient (10-12-10-11 cm2/s) and substantial pesudocapacitance contribution (> 60%) are responsible for the high-rate performance. Although the cathode reaction is mainly driven by Li+, the residue of Mg-O-C as a consequence of co-intercalation of Mg appears to be beneficial to alleviate the exfoliation of C6O6 layers and achieve a much better cycling performance than in LMBs. The stable Mg anode without serious dendrite growth is also responsible for the better cyclability of MLBs. The use of Mo current collector and rGO conductive network enables high energy and power densities up to 525 Wh/kg and 4490 W/kg (based on active cathode material) respectively.

Experimental Section Preparation of nanosized Na2C6O6 (n-SR): The nanosized sodium rhodizonate was synthesized by antisolvent crystallization method.32 12 mg Na2C6O6 (AR, Aladdin Reagent Co.) was dissolved in 20 mL deionized water, then the aqueous solution was poured into 100 mL ethanol with bath sonication. After sonication for 10 min, the precipitation was collected through filtration and then dried. Preparation of Li2C6O6 (LR): The lithium rhodizonate was synthesized by a metathesis method.36 0.247 g H2C6O6·2H2O (98%, Alfa Aesar) and 0.088 g Li2CO3 (AR, Shanghai Sinopharm Chemical Reagent Co.) were mixed together in a 5 mL centrifuge tube. Then 2 mL deionized water was added dropwise to the mixture during 5 min. The mixture was next stirred for 10 h at room temperature. The precipitation was collected by centrifugation and heated in an oven at 200 °C for 17 h. Preparation of reduced graphene oxide (rGO): Graphene oxide (GO) was ACS Paragon Plus Environment

Page 18 of 34

Page 19 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

synthesized by a modified Hummer’s method and then rGO was synthesized by a thermal reduction of GO in hydrogen contained atmosphere. Appropriate amount of GO was put into a quartz boat, and then heated at 900 °C under Ar/H2 (95:5) for 3 h with a heating rate of 3-5 °C min-1. Preparation of Mg electrolyte solutions: Two pure Mg electrolyte solutions were selected,

consisting

0.4

M

all-phenyl

complex

(APC)

or

0.1

M

Mg(HMDS)2-2AlCl3-3MgCl2 complex (MHCC). 0.4 M APC electrolyte was prepared by dissolving 0.1333 g AlCl3 (≥98%, TCI) in 1.5 mL tetrahydrofuran (THF), and then 1 mL PhMgCl (≥99%, 2.0 M in THF solution, Aladdin Reagent Co.) was added to the AlCl3/THF solution. The final solution was stirred overnight prior to use. 0.1 M MHCC

electrolyte

was

prepared

by

dissolving

0.0345

g

magnesium

bis(hexamethyldisilazide) (Mg(HMDS)2, 97%, Sigma-Aldrich), 0.0267 g AlCl3 (≥98%, TCI) and 0.0286 g MgCl2 (99.9%, Aladdin Reagent Co.) into 1 mL THF. The solution was stirred overnight prior to use. Preparation of hybrid electrolyte solution: One hybrid electrolyte solution consists of 1.0 M LiCl and 0.25 M APC in THF. This hybrid electrolyte (APC-LiCl) was prepared by adding 0.1696 g anhydrous LiCl (99.99%, Aladdin Reagent Co.) and 0.1333 g AlCl3 (≥98%, TCI) in 3 mL THF with stiring overnight, and then 1 mL PhMgCl (≥99%, 2.0 M in THF solution, Aladdin Reagent Co.) was added to the above solution. The final solution was stirred overnight prior to use. Another hybrid electrolyte solution consists of 1.0 M lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) and 0.2 M [Mg2Cl2][AlCl4]2 (MACC) in 1,2-dimethoxyethane (DME). The MACC-LiTFSI hybrid electrolyte was prepared by adding 0.2871 g LiTFSI (99.95%, Sigma-Aldrich), 0.0381 g MgCl2 (99.9%, Aladdin Reagent Co.) and 0.0533 g AlCl3 (≥98%, TCI) into 1 mL DME (99.5%, Sigma-Aldrich). The final solution was stirred overnight prior to use. Structural Characterizations: The morphology and grain size of rhodizonate salt samples and Mg anode were observed by scanning electron microscope (SEM, S-4800, Hitachi) with element mapping by inductive coupled plasma emission spectrometer (EDS). The structure and crystallinity of pristine samples and cycled ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 34

electrodes were analyzed by X-ray diffraction (XRD, D8 Discover, Bruker) in a 2-theta range of 10°-60° at a scan rate of 5° min-1 using Cu Kα radiation. The evolution of microstructure and bonding situations of pristine and cycled electrodes was disclosed by Raman spectroscopy (Thermo Nicolet) and X-ray photoelectron spectroscopy (XPS, ESCAlab-250, Thermo Fisher Scientific) with an Al anode source. In XPS fitting, the binding energy was aligned based on the reference C 1s peak at 285.0 eV. A Shirley-type background was used and the experimental data was fitted using a nonlinear least square fitting based on a mixed Gaussian/Lorentzian peak shape. To prepare the cycled electrodes for ex-situ XRD, SEM, EDS, Raman and XPS characterization, they were taken out from the cells in the Ar-filled glove box, and then carefully washed by THF to remove the residual electrolyte. The washed electrodes were then dried in Ar filled glove box before measurement. The cycled electrolyte was diluted in water by a dilution ratio of 100 for element detection by inductively coupled plasma optical emission spectrometry (ICP-OES, Agilent 725). Electrochemical Characterizations: Electrochemical experiments were tested in 2032-type coin cell which were assembled in an argon-filled glove box (