- Email: [email protected]
Failure mechanism of NaAlH4 negative electrodes in lithium cells
Accepted Manuscript Title: Failure mechanism of NaAlH4 negative electrodes in lithium cells Authors: L. Silvestri, M.A. Navarra, S. Brutti, P. Reale PII: DOI: Reference:
S0013-4686(17)31945-X http://dx.doi.org/10.1016/j.electacta.2017.09.074 EA 30273
To appear in:
Electrochimica Acta
Received date: Revised date: Accepted date:
5-6-2017 6-9-2017 13-9-2017
Please cite this article as: L.Silvestri, M.A.Navarra, S.Brutti, P.Reale, Failure mechanism of NaAlH4 negative electrodes in lithium cells, Electrochimica Actahttp://dx.doi.org/10.1016/j.electacta.2017.09.074 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Failure mechanism of NaAlH4 negative electrodes in lithium cells L. Silvestria,b, M. A. Navarraa, S. Bruttic,d,* and P. Realeb,*
a. Dipartimento di Chimica, Sapienza Università di Roma, P. le Aldo Moro 5, Roma (Italy); b. ENEA Centro Ricerche Casaccia, via Anguillarese 301, Roma (Italy); c. Dipartimento di Scienze, Università della Basilicata, V.le dell’Ateneo Lucano 10, Potenza (Italy); d. ISC-CNR, UOS Sapienza, Via dei Taurini, Roma (Italy) * Corresponding authors; e-mail [email protected], [email protected] We would like to dedicate this paper to our mentor Prof. Bruno Scrosati on the occasion of his 80th birthday.
Abstract Insights about the failure mechanism of negative electrodes based on NaAlH4 for lithium batteries are here discussed based on electrochemical-pressure measurements, galvanostatic cycling, infrared spectroscopy and impedance spectroscopy. The accumulation of irreversible capacity in the first cycle and the capacity fading have multiple origins. Besides the mechanical instability due to volume expansions/contractions upon lithium incorporation/de-incorporation, two additional reasons for electrode failure have been disclosed, both related to active material consumption: (1) the irreversible oxidation upon charge of the alanates that leads to H2 release, and (2) the spontaneous chemical reaction of the electrode active material with the electrolyte due to an unstable solid electrolyte interface. Both problems can be tackled by limiting the anodic voltage cutoffs in lithium cells and by the use of innovative electrolytes based on mixture of an ionic liquid and the standard carbonate-based solvent blend. The use of both strategies allowed us to assemble and demonstrate for few cycles the operation of a full Li-ion cell with a negative electrode based on NaAlH4 and a LiFePO4 positive electrode.
Keywords: Sodium alanates; NaAlH4; negative electrodes; anodes; li-ion batteries; lithium; failure mechanism; gas release.
1. Introduction Alanates are a new class of materials able to react electrochemically with lithium through a conversion reaction [1–4]. Recently, we reported the electrochemical behavior of either Li- and Naalanates and explained the electrochemical reaction mechanism. For both, a complex multistep conversion mechanism have been observed, achieving almost all the theoretical capacity upon first discharge, i.e. 1985 mAh/g and 2119 mAh/g for NaAlH4 and LiAlH4, respectively [2,3]. Despite the promising performance of these materials in lithium cell, various technological drawbacks have to be solved for a successful use in a lithium devices. In analogy with other materials undergoing to a conversion reaction[5], alanates suffer from large overpotentials leading to voltage hysteresis between discharge/charge, poor coulombic efficiency and low cycle life. Moreover alanate-based electrodes suffer big volumetric expansion/contraction resulting from discharge/charge conversion steps. In fact, the conversion processes entail massive structural reorganization and volumetric changes (for instance, 72% for NaAlH4). These changes can lead to active particles isolation and cracking thus justifying the poor charge efficiency and the massive capacity fading. In a previous publication, we developed a composite alanate/carbon able to mitigate the volumetric variations [3], thus improving, for instance, first charge efficiency from 30% to more than 70%. However, even in this case, the cell performance quickly drops in few cycle. In few words, alanate failure upon cycling must have other additional concurring reasons. Here we tackled two possible challenges. Considering the intrinsic strong reducing activity of alanates, we investigated its electrochemical stability upon oxidation/charge and we studied its chemical reactivity toward the electrolyte. As a negative electrode, alanate has been so far considered only for its reductive processes. However, starting from the fully reduced mixture of Na/Al and LiH, also the direct hydrogen evolution could take place upon oxidation. The electrochemical hydrogen evolution from alanates has not been widely studied in literature, because of the experimental complexity and the poor application interest: hydrogen evolution is generally achieved by thermal desorption or by water hydrolysis. Nevertheless there are very few thermodynamic studies from which the estimated equilibrium potential of hydrogen electrode reaction of NaAlH4 has been obtained [6]. Apparently alanate electrochemical oxidation can occur in potential ranges that matches those exploited for conversion reactions in lithium cells. In order to shed light on the possible hydrogen evolution upon alanate charge process, we performed electrochemical pressure tests (EPT) to monitor the internal pressure upon cycling of alanate-based lithium cells. Moreover, we also qualitatively checked by
gas chromatography using a thermal conductivity detector (GC-TCD) the formation of molecular hydrogen in the gas accumulated in NaAlH4 lithium cell after cycling. The strong chemical reactivity of alanates towards organic oxidized molecules is another really important issue to be tackled. It is well-known that alanates are strong reducing agents, typically used in organic chemistry for carbonyl reduction. Standard LP30 electrolyte used for electrochemical tests is based on alkyl carbonates, and thus, it may be easily reduced as soon as it comes into contact with the alanate electrode. Our study of electrode/electrolyte interface of LiAlH4 [2] showed the formation of reduction products accumulated on the electrode surface as a result of its high chemical reactivity. Our is a direct clue of the formation of a spontaneous surface film on the electrode after the cell assembly and before any electrochemical polarization. Furthermore, this observation fits with the low OCV value of activated alanates based electrodes, and with the absence of the signature of electrolyte decomposition typically observed around 0.8 V upon their discharge. Following the experimental approach previously reported for LiAlH4 [2], here we extend the study of surface reactivity to sodium alanates. In particular we highlight the unavoidable reactivity between NaAlH4 and the standard LP30 electrolyte (1M LiPF6 solution in a 1:1 mixture of ethylene carbonate and dimethylene carbonate) and we evaluate the effect of the use of an ionic liquids as co-solvent in the electrolyte, namely the N-n-butyl-N-methylpyrrolidinium hexafluorophosphate ([Py14]PF6). The obtained results allowed us to suggest a possible strategy to improve alanate cyclability. On this basis, in the last section of this paper we demonstrate for few cycles the concept of a full lithium ion battery based with a sodium alanate-base electrode at the negative side and LiFePO4 electrodes at the positive one.
2. Experimental section NaAlH4 based electrodes have been prepared following the procedure already described in our previous paper [3]. Basically, the powder of NaAlH4 (Hydrogen grade / Sigma Aldrich) has been activated by milling in a Spex 8000 shaker mill for 15 hours; then a carbon compositing has been realized by adding conductive Super P carbon (Timcal) and milling 5 hours further. Finally, the NaAlH4/Super P composite was mixed with the polymeric binder polyvinylidene difluoride (PVdF, Kynar) in order to obtain an electrodic mixture containing the 50% of alanate, 30% of Super P and 20% of binder. The latter was pressed with a Specac hydraulic press on 10 mm Cu disk to obtain the electrode used for the characterizations.
Three electrodes cells were assembled using NaAlH4 electrodes as working and metallic lithium foils either as counter and reference electrodes. Whatman borosylicate fiber disks were adopted to separate electrodes and support the electrolyte. Preliminary tests were carried out using a 1M LiPF6 solution in 1:1 v/v ethylene carbonate– dimethyl carbonate (LP30, Merck), therefore different electrolytes were prepared mixing to LP30 increasing
amount
of
N-n-butyl-N-methylpyrrolidinium
hexafluorophosphate,
[Py14]PF6
(Solvionic). To highlight the gas evolution upon alanate oxidation, electrochemical pressure tests (EPT) were performed. EPT experiments consist in the measure of the time evolution of the internal pressure within a two electrodes lithium cell, by using a designed electrochemical cell with a calibrated internal empty volume (3.024cm3) and a pressure gauge (ECC-Press EL-Cell), with a resolution of 0.05 mbar. In order to get rid of the possible spurious pressure signals, due to thermal fluctuations, the EPT measurements have been carried out in a thermostatic oven at 30.0±0.1°C including a 6 hours thermal preconditioning at OCV.[7] The study of the alanate/electrolyte chemical reactivity involved the investigation of the interphase composition and stability. Three techniques have been adopted: (a) evolution of the open circuit potential (OCP), (b) Electrochemical Impedance Spectroscopy (EIS) and (c) ex-situ Fourier Transformed Infrared Spectroscopy conducted in Attenuated Total Reflection (FTIR-ATR). A VSP Biologic Potentiostat equipped with Frequency Range Analyzer was used for OCP collection and EIS experiments. EIS spectra were acquired by applying a potential signal of 10mV amplitude in the frequency range 100kHz-50mHz on electrochemical cells stored at the open circuit for a week. OCP and EIS measurements were contemporary and independently carried on different test cells. FTIR-ATR was carried out using a Brucker Alpha instrument. Spectra were collected exposing the surface of binder free electrodes, previously soaked in the electrolytic solutions for 10 minutes, 1 hour, 5 hours and 3 days and then washed with dimethyl carbonate and dried under vacuum. For electrochemical tests, galvanostatic cyclations were carried out on a Maccor battery cycler, using a current density of 100 mA/g, nearly corresponding to C/20. Specific capacities in both lithium cells and Li-ion cells have been always calculated in respect to the mass of NaAlH4 in the negative electrode. The qualitative composition of the permenent gas accumulated in the first full discharge/charge galvanostatic cycle (OCV-10mV-2.0V at C/20) within a two electrodes lithium pouch cell with a NaAlH4-based electrode has been obtained by GC-TCD with a Hewlett Packard
6890 gas chromatograph, equipped with a thermal conductivity detector and a packed MS 13X stainless steel column to separate the permanent gases. Argon was used as carrier gas at 25 ml min1
. The gaseous sample has been extracted from the pouch cell after cycling in the Ar-filled glove
box by using a gas-tight lock syringe (Aldrich) and then injected into the gas chromatograph. A benchmark cell cycled in the voltage range (OCV-10mV-2.0V at C/20) has been also checked. The calibration of gas chromatograph was performed by using standard gas in the Ar carrier gas. LiFePO4 was purchased by CustomCells in the form of 1.0mAhcm-2 specific capacity electrode sheets.
3. Results and discussion 3.1 Gas release upon charge EPT measurements on Li/LP30/NaAlH4 cells have been carried out in order to highlight the possible gas evolution upon charge in parallel with the establish back-conversion reaction. Alanate electrodes have been cycled through a full galvanostatic discharge and charge at C/20 while pressure have been monitored. As benchmark, a similar test has been performed using a stainless steel electrode as working, in order to consider gas evolution related to electrolyte evaporation. The evolution of the internal pressure of both the alanate and the benchmark cells are shown in the figure 1a. In lithium cells, the hydride conversion reactions of sodium alanates have been already clarified in the literature [1,3], and imply: (a) upon discharge a complex multistep reduction to hexahydrides and to metals, both Na and Al and (b) upon charge the reverse formation of NaAlH 4. In the initial 6 hours rest at OCV, the pressure of the alanate cell matches the SS benchmark. A small pressure increase soon starts as discharge begins, and continues negligible upon charge up to 1.2V. Such a small pressure increase is expected in consideration of the possible electrolyte side reduction, which can cause gaseous products evolution.[8] In the last stage of the charge, above 1.2V, the internal pressure of the alanate cell noticeably rises. In order to clarify the identity of the constituents of the permanent gas produced above 1.2 V, a GC-TCD experiment has been carried out on the gaseous products accumulated within pouch cells after one galvanostatic cycle at C/20 with voltage cutoffs of 2.0 and 1.2 V, respectively. The comparison of the two gas chromatographic results are shown in the figure 1b. Carbon monoxide (CO) has been detected in the produced gas phase of both cells charged at 1.2 and 2.0 V whereas hydrogen is the most abundant constituent only within the cell charged at 2.0 V and is almost absent in the cell charged at 1.2 V.
The formation of CO can be related to the electrolyte degradation and the parallel growth of the SEI layer [5]. On the other hand hydrogen release from alanates is generally conceived upon thermal desorption or by hydrolysis [9], and very few studies can be found about electrochemical hydrogen evolution. Among the few published, Senoh et al.[6] in 2008 used alanates solutions in THF and Ni electrodes to experimentally evaluate the hydrogen electrode reaction of lithium and sodium aluminum hydrides, metallic lithium and sodium where used as references. The authors evaluated the onset of the hydrogen evolution and highlighted the necessity of an “activation” process. Thermodynamic estimations suggests that hydrogen could evolve following to two processes: MAlH4 = M+ + Al + 2H2 + e- (R1) MAlH4 = M+ + AlH3 + ½H2 + e-
(R2)
According to Senoh, despite reaction (R1) is the thermodynamic favorite process (0.5V vs Na), mechanism (R2) (0.98V vs Na) is the kinetically dominant. In fact 0.5 mol of hydrogen electrochemically generated occurred above 1.2V vs Na. Both processes are in the potential range explored during the electrochemical tests up to now reported for the alanate as conversion electrode in lithium cells, in fact anodic cut-off potential is usually set in between 2-3V vs Li.[1,10] Therefore we updated the thermodynamic E° by using more recent literature ΔrG°, [11,12] obtaining for reaction (1) and (2) 0.56 and 1.1V vs Na respectively, corresponding to 0.86 and 1.4V vs Li. Our EPT and GC-TCD data suggest that hydrogen evolution occurs in the alanate cell at potentials that nicely matches those estimated for reactions (R1) and (R2). Once formed, gaseous hydrogen leaves the electrode surface and cannot reduced back in the subsequent cycles. Therefore the NaAlH4 oxidation also involve a fully irreversible process (reactions R1 and R2) that causes the loss active material useful for conversion. An easy solution to limit the internal pressure increase and thus possible hydrogen evolution is the adoption of a smaller anodic cutoff voltage in charge, e.g. 1.2V. Figure 2 shows the galvanostatic cycling of alanate electrodes up the standard 2.50V anodic cutoff limit (benchmark) in comparison to the cycling limited to 1.20V. The effect of limited voltage range is clear from the second cycle, where the capacity in discharge is 30% larger compared to the benchmark. However in the subsequent cycled the trend of the capacity fade is similar for both anodic voltage cutoffs. This evidence suggests that the possible hydrogen evolution is not the unique failure mechanism. Nevertheless it must be emphasized that after 10 cycle, the cell charged at the lower cutoff is capable to exchange nearly 400mAh/g, a capacity 56% larger then what is delivered disregarding the oxidation process beyond 1.2V.
3.2 Instability of the electrode/electrolyte interface Among other possible failure mechanisms, the chemical reactivity of alanate towards the liquid electrolyte must be taken into account. Hitherto, all studies on electrochemical conversion of alanates in lithium cells have been carried out in conventional LP30 electrolyte, i.e. in a 1M LiPF6 solution in 1:1 v/v ethylene carbonate–dimethyl carbonate [1,3,9]. This electrolyte is not the best choice, considering the well-known chemical ability of alanates to reduce the carbonyl functional group [2]. In a typical lithium cell, the spontaneous formation of electrical double layers [13] at the electrodes leads to an open circuit potential between 2 and 3 V regardless of the electrode chemistry. On the other hand lithium cells containing LP30 electrolyte and NaAlH4 electrodes exhibit open circuit potentials of about 1.1V vs Li [1,3]. This small OCP value is a clear indication of the existence of a specific reactivity between the LP30 and alanate electrodes. Figure 3a shows the Nyquist plots obtained from the electrochemical impedance spectroscopy measurement of NaAlH4 electrodes stored in contact to LP30 at OCP for a week. Figure 3b shows the evolution of the OCP for a week. The impedance curve is characterized by a semicircle in the high frequencies region associated to the “solid electrolyte interface” (SEI) presence, followed by a large semicircle at mid frequencies relative to the electrodic process and finally a convoluted line at low frequencies. The observed not blocking interface can be described by a R(RQ)(RQ)Q circuit. At t=0 the high frequency resistance is approximately 30 ; after 1 day this value constantly increases until 78 and then decreases again to 30 after 6 days. These oscillations in the estimated resistance of the SEI suggest the formation of an unstable film on the electrode surface, that continuously grows, breaks and grows again. Moreover the smooth OCP increase confirms that the interphase tends to the equilibrium only very slowly. The chemical nature of the surface film has been investigated by FTIR-ATR spectroscopy. ATR mode, allows to collect the IR signal coming from the electrode surface, and therefore to highlight the absorbance due to the film growing on it. Spectra as function of aging time in the LP30 electrolyte solutions are given in figure 4. The pristine electrode exhibits the typical vibrational frequencies of NaAlH4: [AlH4]stretching mode at 1648 cm-1 and the bending at 900 and 712 cm-1. Beside them, Na3AlH6 peaks are also observable. Specifically, stretching at 1400 cm-1 and bending around 1189 and 1081 cm1
.[14,15] As evident from figure 4, LP30 decomposition starts as soon as it comes in contact with
NaAlH4. After only 10 minutes, vibrational frequencies associated with alanates disappear and wide vibrational modes are observable in the spectrum. As already observed for lithium alanate, the
electrode surface is covered by a complex film, which include insoluble organic lithium carbonates, polymeric anhydrides, fluoro-alkyl phosphates and aluminum oxides [16]. Specifically, at 1803 and 1750 cm-1 are visible the stretching modes of the C=O groups and at 1272 cm-1 the C-O stretching relative to carbonates. It's also observable the frequencies associated to the fluoro-alkyl phosphates: P=O at 1195 and 1159 cm-1; P-O-C at 1066 and 968 cm-1; and P-O at 796 and 774 cm-1. Also, at 640 cm-1 there is the appearance of aluminum oxides stretching mode. After 1 hour, peaks appear reduced in intensities. Also, prolonging the aging time, the signals intensity constantly changes. These evidences suggest that the electrode/electrolyte interface is highly reactive. Sodium alanate, thanks to its high reducing power, largely reacts with the electrolyte solution and several reaction products precipitate on the electrode surface. The large surface area of the alanate nanomaterial as the catalytic effect of the carbon coating enhances the surface reactivity and therefore the process continuously evolves upon time. Being the SEI layer unstable, the active material is progressively consumed upon time leading to an unavoidable capacity fading. In summary, besides the mechanical instability, two additional reasons for electrode failure have been disclosed, both related to active material consumption: its irreversible oxidation upon charge and its chemical reaction with the electrolyte, i.e. the hydrogen evolution and the production of an unstable solid electrolyte interface. Both problems could find technological solutions.
3.3 Ionic liquids as co-solvents for the stabilization of the interface In order to avoid the hydrogen evolution, a severe limitation of the anodic cutoff to 1.2V is the easiest solution. On the other hand in order to mitigate the alanate attack to the carbonyl, we tested the use of a carbonyl-free solvent as solvent or co-solvent for the liquid electrolyte. Therefore, the use of a different type of electrolyte systems has been considered. Ethers, like tetraethylen glycol dimethyl ether (TEGDME) or the mixture 1,3-dioxolane-1,2-dimethoxyethane (DOL-DME), failed: in fact ethers are able to dissolve alanates, thus boosting interface instability in terms of potential, impedance and redox activity (data not shown). On the contrary ionic liquids are promising solvents. Ionic Liquids have attracted a lot of attention as electrolyte systems thanks to their high ionic conductivity, low toxicity as well as their high thermal, chemical and electrochemical stability [17]. We tested the N-n-butyl-N-methylpyrrolidinium hexafluorophosphate ([Py14]PF6), an Ionic Liquid consisting of an alkylmethylpyrrolidinium cation and the same anion PF6- contained in LP30 as cosolvent. Previous studies [18] demonstrated that its addition to LP30 improved the electrochemical and thermal properties, reduced the flammability of the volatile carbonates, enhancing the safety of the overall system. For our purpose, [Py14]PF6 has been added at the standard electrolyte and three
different mixtures have been prepared increasing the amount of IL from 30 to 70 wt.%. Table 1 reports composition, labelling and molality of the resulting electrolytes.
Larger concentrations and the pure IL as solvent have not been investigated due to the high viscosity. In order to evaluate the effect on electrochemical performances, lithium cells using the enriched IL solutions as electrolyte and alanate as working electrode have been assembled and tested in galvanostatic regimes in the anodic limited voltage range 0.01-1.20V. Figure 5a shows the voltage profiles, while figure 5b reports capacity values and cell efficiency for the three IL added solution in comparison with the pure LP30 electrolyte. The total capacity exchanged at the end of the first discharge decreases with the amount of ionic liquid present in the electrolyte, moving from 2010 mAhg-1 with the pure LP30 to 840 mAhg-1 with the addition of 70 wt.% of IL. However, despite differences in the total exchanged capacity values, first discharge profiles appear similar for the three solutions apart a down shifting of 100200 mV to be related with the increased amount of IL added to electrolyte. Such overvoltages can be originated from the lower Li+ transference number of solutions after the addition of IL. Concerning charge profile, it's evident that the addition of ionic liquid has beneficial effects on the efficiency of the conversion process, that increases from 45% in bare LP30, to 71.4% in IL70 solution, see figure 4b. Charge evolves through two plateaus around 0.4 and 0.8 V, similarly to the profile in LP30, but the addition of ionic liquid leads to an extension of the plateau around 0.8V. Interestingly, conversely to discharge process, charge is not affected by larger overvoltages. The addition of 30% of [Py14]PF6 to LP30 appears the best compromise between capacity exchanged and recharge efficiency, and was therefore considered for further characterizations. EIS and FTIR-ATR were used to evaluate the effective improvement of interface properties. Figure 6a shows the Nyquist plots collected for the NaAlH4 electrode in contact with the IL30 electrolyte in a three electrode lithium cell. The impedance shape is similar to that already discussed of figure 4 and it has been also in this case fitted by using a R(RQ)(RQ)Q equivalent circuit. Upon time, the resistance associated with SEI constantly increases from 2 to 43 . In the same way, charge transfer resistance increases from hundreds to a thousand ohm, while the final linear trend, probably indicative of a diffusion kinetics limitation, do not basically change upon time. These trends suggest a noticeable different surface chemistry compared to the LP30 electrolyte. In fact while NaAlH4/LP30 interaction gave rise at the interphase to an unstable film, which continuously forms, breaks and creates again, in the IL containing solution the surface film continuously and slowly grows.
The figure 6b shows the trend upon time of the SEI resistance (RSEI) of the NaAlH4/IL30 interface, in comparison to that observed at the NaAlH4/LP30 interface. The replacement of carbonates based electrolyte with an ionic liquid produces beneficial effects on the film formation process. In the case of plain alkylcarbonate based electrolyte, a 30 resistive film is quickly formed and is therefore subject to continuous fluctuations. Conversely, the IL30 solution showed small initial RSEI values that continuously increase upon time. After 1 day, such growth becomes lower and the RSEI settle on 30 . Anyway, the trend observed in RSEI reveals that alanate still reacts largely with electrolyte but the resistivity and the instability of the SEI appears to be mitigated. In fact, as occurring for a LP30 soaked electrode, upon aging in the IL30 solution the electrode surface is enriched by reduction products resulting from electrolyte decomposition. Figure 7 shows FTIR-ATR spectra collected for the NaAlH4/IL30 system. In contrast with the experiments performed with pure LP30, in IL30 the film increases gradually upon time. Specifically, after 10 minutes of aging there is the appearance of the vibrational modes related to fluoro-alkyl phosphates at 850 cm-1 and the Al-O bending at 650 cm-1. These signals constantly increase in intensity and after prolonged time of aging it's observable also the stretching modes of P=O group at 1160 cm-1 and P=O-R group at 976 and 780 cm-1 while the intensity of peak at 850 cm-1 largely increase. Furthermore, stretching frequencies related to deposition of organic lithium carbonates are clearly visible at 1806, 1773 and 1306 cm -1, while the C-O stretching is at 1080 cm-1. In summary in the IL-enriched electrolyte IL30, the interaction between alanate and carbonates is still large, as evident from the several products deposited on electrode surface. However the addition of the IL allows the formation of a more stable film. Beneath the addition of IL is not ultimate to optimize the alanate electrochemical performance, it mitigates the alanate reactivity towards the electrolyte. One may speculate that the strong alanate reducing power really needs to be controlled by a careful coating of alanate particles or by their confinement into an inert matrix. This strategy demonstrates to be effective for several other electrode materials, especially for high potential cathodes capable to catalyze the electrolyte oxidation.[19]
3.4. Formulation of a full Li-ion cell with improved electrolyte and controlled voltage limits Even if optimization is still a distant goal, the here-reported improvements led us to assemble a lithium ion cell in order to demonstrate, even for few cycles, the use of NaAlH4 as negative electrode. The figure 8 shows the voltage profile in the first 3 cycles of a full Li-ion cell with an NaAlH4-based negative electrode. The IL30 solution has been used as electrolyte. As positive electrode a commercial LiFePO4 has been chosen because of its well-known behavior, in terms of
constant working potential and stability upon cycling. Electrodes masses have been balanced in order to match the capacity in the first discharge. Upon cycling a time cutoff has been used to limit the capacity to the 67% of the theoretical one. This capacity limitation has been adopted to avoid lithium plating at the negative side and overcharge at the positive collector. Considering the redox potential of NaAlH4 and LiFePO4, the applied cycling voltage range was 2.25-3.70V. A metallic lithium reference electrode has been also used to constantly check the alanate electrode potential. Figure 8 shows the (a) lithium ion cell voltage and (b) alanate negative electrode potential vs Li. as a function of delivered specific capacity. The first cycle cell voltage profile, highlighted in red, shows the signature of the multistep alanate conversion mechanism[3]; charge mostly develops between 2.9 and 3.25V (alanate conversion at 0.55 and 0.2V vs Li) while discharge occurs around 2.6V (recombination to alanate at 0.8V), the average cell voltage is around 2.8V. As expected, first charge finishes when the capacity limit is reached, i.e. 1330mAh/g, and more than 60%, i.e. 849mAh/g is returned upon the following discharge. In the second cycle efficiency markedly improves: 1282mAh/g are delivered during charge and 1000mAh/g in discharge. Upon further cycling, capacity progressively fades because still efficiency never exceeds 90%. Lithium plating does not occur during the firsts three cycles, but not even in the following, and the anode potential never drops below 0.04V. The performance of this Li-ion cell is far from being optimal: this demonstrator shows poor cycling performance, thus the alanate negative electrodes are far from being ready for application. On the other hand our Li-ion cell is the first ever reported demonstration of the use of the sodium alanate as negative electrode material in a full Li-ion cell and illustrates the current potentialities and open problems of this battery system.
4. Conclusion In this communication we discussed for the first time new insights about the failure mechanisms of NaAlH4 negative electrodes in lithium cells. Besides the well-known mechanical instability of the electrode due to the volume variation upon conversion also hydrogen gas evolution upon oxidation and the instability of the SEI layer have been proved and discussed. Possible mitigation strategies have been highlighted including the limitation of the anodic cutoff potential of the NaAlH4 electrode as well as the use of composite electrolytes based on mixtures of organic carbonates and an ionic liquid. Both these strategy allowed the demonstration of a full Li-ion cell with a NaAlH4 negative electrode, a LiFePO4 positive electrode and a composite electrolyte based on a EC:DME:LiPF6:Pyr1,4PF6 mixture. Although only three cycles have been reported, our prototype is, as far as we know, first ever reported full Li-ion formulation that exploits the alanate
conversion chemistry. Much further knowledge is needed to fully develop a reliable and longlasting configuration and an intense research activity is in progress in our laboratories to tackle this challenge.
Acknowledgements This work is carried out in the framework of the Italian project (FIRB-Futuro in ricerca 2010) “Hydrides as high capacity anodes for lithium ion batteries” (RBFR10ZWMO), supported by Italian Minister for University and Research. Thanks are due to F.Langerame for the kind.
References [1]
J.A. Teprovich, J. Zhang, H. Colón-Mercado, F. Cuevas, B. Peters, S. Greenway, R. Zidan, M. Latroche, Li-Driven Electrochemical Conversion Reaction of AlH3, LiAlH4, and NaAlH4, J. Phys. Chem. C. 119 (2015) 4666–4674. doi:10.1021/jp5129595.
[2]
L. Silvestri, S. Forgia, L. Farina, D. Meggiolaro, S. Panero, A. LaBarbera, S. Brutti, P. Reale, Lithium Alanates as Negative Electrodes in Lithium-Ion Batteries, ChemElectroChem. 2 (2015) 877–886. doi:10.1002/celc.201402440.
[3]
L. Silvestri, L. Farina, D. Meggiolaro, S. Panero, F. Padella, S. Brutti, P. Reale, Reactivity of Sodium Alanates in Lithium Batteries, J. Phys. Chem. C. 119 (2015) 28766–28775. doi:10.1021/acs.jpcc.5b10297.
[4]
L. Cirrincione, L. Silvestri, C. Mallia, P.E. Stallworth, S. Greenbaum, S. Brutti, S. Panero, P. Reale, Investigation of the Effects of Mechanochemical Treatment on NaAlH4 Based Anode Materials for Li-Ion Batteries, J. Electrochem. Soc. 163 (2016) A2628–A2635. doi:10.1149/2.0731613jes.
[5]
J. Cabana, L. Monconduit, D. Larcher, M.R. Palacín, Beyond intercalation-based Li-ion batteries: The state of the art and challenges of electrode materials reacting through conversion reactions, Adv. Mater. 22 (2010) 170–192. doi:10.1002/adma.201000717.
[6]
H. Senoh, T. Kiyobayashi, N. Kuriyama, Hydrogen electrode reaction of lithium and sodium aluminum
hydrides,
Int.
J.
Hydrogen
Energy.
33
(2008)
3178–3181.
doi:10.1016/j.ijhydene.2008.01.019. [7]
J. Manzi, S. Brutti, Surface chemistry on LiCoPO4 electrodes in lithium cells: SEI formation and
self-discharge,
Electrochim.
Acta.
222
(2016)
1839–1846.
doi:10.1016/j.electacta.2016.11.175. [8]
M. Holzapfel, A. Würsig, W. Scheifele, J. Vetter, P. Novák, Oxygen, hydrogen, ethylene and CO2 development in lithium-ion batteries, J. Power Sources. 174 (2007) 1156–1160.
doi:10.1016/j.jpowsour.2007.06.182. [9]
B. Sakintuna, F. Lamari-Darkrim, M. Hirscher, Metal hydride materials for solid hydrogen storage:
A
review,
Int.
J.
Hydrogen
Energy.
32
(2007)
1121–1140.
doi:10.1016/j.ijhydene.2006.11.022. [10] L. Silvestri, A. Paolone, L. Cirrincione, P.E. Stallworth, S. Greenbaum, S. Panero, S. Brutti, P. Reale, NaAlH4 nanoconfinement in mesoporous carbon for application in lithium ion batteries, J. Electrochem. Soc. 164 (2017) A1120–A1125. [11] C. Qiu, S.M. Opalka, G.B. Olson, D.L. Anton, Thermodynamic modeling of the sodium alanates and the Na–Al–H system, Int. J. Mater. Res. 97 (2006) 1484–1494. doi:10.3139/146.101410. [12] J. Graetz, J. Reilly, G. Sandrock, J. JOHNSON, W.M. Zhou, J. Wegrzyn, Aluminum hydride AlH3, as a hydrogen storage compound., Brookhaven National Laboratory, Upton, NY, 2006. doi:10.2172/899889. [13]
J.O. (John O.. Bockris, A.K.N. Reddy, M. Gamboa-Aldeco, Modern electrochemistry. Volume 1, Ionics, Kluwer Academic, 2002.
[14]
B.C. Smith, Infrared spectral interpretation : a systematic approach, CRC Press, 1999.
[15]
J.-C. Bureau, Z. Amri, P. Claudy, J.-M. Letoffe, Etude comparative des hexahydrido-et des hexadeuteridoaluminates de lithium et de sodium. I - Spectres raman et infrarouge de Li3-et Na3AlH6, et Li3AlD6, Mater. Res. Bull. 24 (n.d.) 23–31.
[16]
T. Kawamura, S. Okada, J. Yamaki, Decomposition reaction of LiPF6-based electrolytes for lithium
ion
cells,
J.
Power
Sources.
156
(2006)
547–554.
doi:10.1016/j.jpowsour.2005.05.084. [17]
D.R. MacFarlane, N. Tachikawa, M. Forsyth, J.M. Pringle, P.C. Howlett, G.D. Elliott, J.H. Davis, M. Watanabe, P. Simon, C.A. Angell, Energy applications of ionic liquids, Energy Environ. Sci. 7 (2013) 232–250. doi:10.1039/C3EE42099J.
[18]
A. Tsurumaki, M.A. Navarra, S. Panero, B. Scrosati, H. Ohno, N-n-Butyl-Nmethylpyrrolidinium hexafluorophosphate-added electrolyte solutions and membranes for lithium-secondary
batteries,
J.
Power
Sources.
233
(2013)
104–109.
doi:10.1016/j.jpowsour.2013.01.131. [19]
M. Agostini, A. Matic, S. Panero, F. Croce, R. Gunnella, P. Reale, S. Brutti, A mixed mechanochemical-ceramic solid-state synthesis as simple and cost effective route to high performing LNMO spinels, Electrochim. Acta. 235 (2017) 262–269.
Figure captions
Figure 1. (a) Galvanostatic EPTs of a Li/LP30/NaAlH4 (black) and a reference Li/LP30/SS (blue) cell in the 0.01-2.5V voltage range. The Li/LP30/NaAlH4 measure has been repeated in the 0.011.2V voltage range (red). C/20 current rate; (b) results of gas chromatographic analysis of the internal gases released within pouch cells discharged at 10 mV and then charged at 1.2 and 2.0 V, respectively.
Figure 2. Galvanostatic cyclation of Li/LP30/NaAlH4 cells in the 0.01-2.50V (black) and in the 0.01-1.20V (red) voltage range
Figure 3. Nyquist Plot (a) and OCP (b) of NaAlH4 in a lithium cell containing LP30 as electrolyte
Figure 4. FTIR-ATR spectra of NaAlH4 electrode upon aging in LP30 electrolyte
Figure 5. a) First discharge/charge voltage profiles obtained in galvanostatic mode as function of amount of ionic liquid into the electrolyte. Results obtained with LP30 are used as reference and b) Efficiency and Specific Capacity as function of electrolyte solution plot.
Figure 6. (a) Nyquist plot of NaAlH4 sample using IL30 as electrolyte; (b) RSEI as function of time at the NaAlH4/IL30 interface in comparison to that developed at the NaAlH4/LP30 interface.
Figure 7. FTIR-ATR spectra of NaAlH4 electrode upon aging in IL30 electrolyte
Figure 8. (a) Cell voltage profiles and (b) anode potential in a NaAlH4/LiFePO4 full cell.
Table 1. Composition of the electrolytes. Label
LP30 wt.%
IL wt.% LiPF6 Molality Ionic Conductivity i/S*cm-1
LP30
100
/
0.87
1.2 10-2
IL30
70
30
0.58
2.8 10-2
IL50
50
50
0.41
3.0 10-2
IL70
30
70
0.24
1.9 10-2
S0013-4686(17)31945-X http://dx.doi.org/10.1016/j.electacta.2017.09.074 EA 30273
To appear in:
Electrochimica Acta
Received date: Revised date: Accepted date:
5-6-2017 6-9-2017 13-9-2017
Please cite this article as: L.Silvestri, M.A.Navarra, S.Brutti, P.Reale, Failure mechanism of NaAlH4 negative electrodes in lithium cells, Electrochimica Actahttp://dx.doi.org/10.1016/j.electacta.2017.09.074 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Failure mechanism of NaAlH4 negative electrodes in lithium cells L. Silvestria,b, M. A. Navarraa, S. Bruttic,d,* and P. Realeb,*
a. Dipartimento di Chimica, Sapienza Università di Roma, P. le Aldo Moro 5, Roma (Italy); b. ENEA Centro Ricerche Casaccia, via Anguillarese 301, Roma (Italy); c. Dipartimento di Scienze, Università della Basilicata, V.le dell’Ateneo Lucano 10, Potenza (Italy); d. ISC-CNR, UOS Sapienza, Via dei Taurini, Roma (Italy) * Corresponding authors; e-mail [email protected], [email protected] We would like to dedicate this paper to our mentor Prof. Bruno Scrosati on the occasion of his 80th birthday.
Abstract Insights about the failure mechanism of negative electrodes based on NaAlH4 for lithium batteries are here discussed based on electrochemical-pressure measurements, galvanostatic cycling, infrared spectroscopy and impedance spectroscopy. The accumulation of irreversible capacity in the first cycle and the capacity fading have multiple origins. Besides the mechanical instability due to volume expansions/contractions upon lithium incorporation/de-incorporation, two additional reasons for electrode failure have been disclosed, both related to active material consumption: (1) the irreversible oxidation upon charge of the alanates that leads to H2 release, and (2) the spontaneous chemical reaction of the electrode active material with the electrolyte due to an unstable solid electrolyte interface. Both problems can be tackled by limiting the anodic voltage cutoffs in lithium cells and by the use of innovative electrolytes based on mixture of an ionic liquid and the standard carbonate-based solvent blend. The use of both strategies allowed us to assemble and demonstrate for few cycles the operation of a full Li-ion cell with a negative electrode based on NaAlH4 and a LiFePO4 positive electrode.
Keywords: Sodium alanates; NaAlH4; negative electrodes; anodes; li-ion batteries; lithium; failure mechanism; gas release.
1. Introduction Alanates are a new class of materials able to react electrochemically with lithium through a conversion reaction [1–4]. Recently, we reported the electrochemical behavior of either Li- and Naalanates and explained the electrochemical reaction mechanism. For both, a complex multistep conversion mechanism have been observed, achieving almost all the theoretical capacity upon first discharge, i.e. 1985 mAh/g and 2119 mAh/g for NaAlH4 and LiAlH4, respectively [2,3]. Despite the promising performance of these materials in lithium cell, various technological drawbacks have to be solved for a successful use in a lithium devices. In analogy with other materials undergoing to a conversion reaction[5], alanates suffer from large overpotentials leading to voltage hysteresis between discharge/charge, poor coulombic efficiency and low cycle life. Moreover alanate-based electrodes suffer big volumetric expansion/contraction resulting from discharge/charge conversion steps. In fact, the conversion processes entail massive structural reorganization and volumetric changes (for instance, 72% for NaAlH4). These changes can lead to active particles isolation and cracking thus justifying the poor charge efficiency and the massive capacity fading. In a previous publication, we developed a composite alanate/carbon able to mitigate the volumetric variations [3], thus improving, for instance, first charge efficiency from 30% to more than 70%. However, even in this case, the cell performance quickly drops in few cycle. In few words, alanate failure upon cycling must have other additional concurring reasons. Here we tackled two possible challenges. Considering the intrinsic strong reducing activity of alanates, we investigated its electrochemical stability upon oxidation/charge and we studied its chemical reactivity toward the electrolyte. As a negative electrode, alanate has been so far considered only for its reductive processes. However, starting from the fully reduced mixture of Na/Al and LiH, also the direct hydrogen evolution could take place upon oxidation. The electrochemical hydrogen evolution from alanates has not been widely studied in literature, because of the experimental complexity and the poor application interest: hydrogen evolution is generally achieved by thermal desorption or by water hydrolysis. Nevertheless there are very few thermodynamic studies from which the estimated equilibrium potential of hydrogen electrode reaction of NaAlH4 has been obtained [6]. Apparently alanate electrochemical oxidation can occur in potential ranges that matches those exploited for conversion reactions in lithium cells. In order to shed light on the possible hydrogen evolution upon alanate charge process, we performed electrochemical pressure tests (EPT) to monitor the internal pressure upon cycling of alanate-based lithium cells. Moreover, we also qualitatively checked by
gas chromatography using a thermal conductivity detector (GC-TCD) the formation of molecular hydrogen in the gas accumulated in NaAlH4 lithium cell after cycling. The strong chemical reactivity of alanates towards organic oxidized molecules is another really important issue to be tackled. It is well-known that alanates are strong reducing agents, typically used in organic chemistry for carbonyl reduction. Standard LP30 electrolyte used for electrochemical tests is based on alkyl carbonates, and thus, it may be easily reduced as soon as it comes into contact with the alanate electrode. Our study of electrode/electrolyte interface of LiAlH4 [2] showed the formation of reduction products accumulated on the electrode surface as a result of its high chemical reactivity. Our is a direct clue of the formation of a spontaneous surface film on the electrode after the cell assembly and before any electrochemical polarization. Furthermore, this observation fits with the low OCV value of activated alanates based electrodes, and with the absence of the signature of electrolyte decomposition typically observed around 0.8 V upon their discharge. Following the experimental approach previously reported for LiAlH4 [2], here we extend the study of surface reactivity to sodium alanates. In particular we highlight the unavoidable reactivity between NaAlH4 and the standard LP30 electrolyte (1M LiPF6 solution in a 1:1 mixture of ethylene carbonate and dimethylene carbonate) and we evaluate the effect of the use of an ionic liquids as co-solvent in the electrolyte, namely the N-n-butyl-N-methylpyrrolidinium hexafluorophosphate ([Py14]PF6). The obtained results allowed us to suggest a possible strategy to improve alanate cyclability. On this basis, in the last section of this paper we demonstrate for few cycles the concept of a full lithium ion battery based with a sodium alanate-base electrode at the negative side and LiFePO4 electrodes at the positive one.
2. Experimental section NaAlH4 based electrodes have been prepared following the procedure already described in our previous paper [3]. Basically, the powder of NaAlH4 (Hydrogen grade / Sigma Aldrich) has been activated by milling in a Spex 8000 shaker mill for 15 hours; then a carbon compositing has been realized by adding conductive Super P carbon (Timcal) and milling 5 hours further. Finally, the NaAlH4/Super P composite was mixed with the polymeric binder polyvinylidene difluoride (PVdF, Kynar) in order to obtain an electrodic mixture containing the 50% of alanate, 30% of Super P and 20% of binder. The latter was pressed with a Specac hydraulic press on 10 mm Cu disk to obtain the electrode used for the characterizations.
Three electrodes cells were assembled using NaAlH4 electrodes as working and metallic lithium foils either as counter and reference electrodes. Whatman borosylicate fiber disks were adopted to separate electrodes and support the electrolyte. Preliminary tests were carried out using a 1M LiPF6 solution in 1:1 v/v ethylene carbonate– dimethyl carbonate (LP30, Merck), therefore different electrolytes were prepared mixing to LP30 increasing
amount
of
N-n-butyl-N-methylpyrrolidinium
hexafluorophosphate,
[Py14]PF6
(Solvionic). To highlight the gas evolution upon alanate oxidation, electrochemical pressure tests (EPT) were performed. EPT experiments consist in the measure of the time evolution of the internal pressure within a two electrodes lithium cell, by using a designed electrochemical cell with a calibrated internal empty volume (3.024cm3) and a pressure gauge (ECC-Press EL-Cell), with a resolution of 0.05 mbar. In order to get rid of the possible spurious pressure signals, due to thermal fluctuations, the EPT measurements have been carried out in a thermostatic oven at 30.0±0.1°C including a 6 hours thermal preconditioning at OCV.[7] The study of the alanate/electrolyte chemical reactivity involved the investigation of the interphase composition and stability. Three techniques have been adopted: (a) evolution of the open circuit potential (OCP), (b) Electrochemical Impedance Spectroscopy (EIS) and (c) ex-situ Fourier Transformed Infrared Spectroscopy conducted in Attenuated Total Reflection (FTIR-ATR). A VSP Biologic Potentiostat equipped with Frequency Range Analyzer was used for OCP collection and EIS experiments. EIS spectra were acquired by applying a potential signal of 10mV amplitude in the frequency range 100kHz-50mHz on electrochemical cells stored at the open circuit for a week. OCP and EIS measurements were contemporary and independently carried on different test cells. FTIR-ATR was carried out using a Brucker Alpha instrument. Spectra were collected exposing the surface of binder free electrodes, previously soaked in the electrolytic solutions for 10 minutes, 1 hour, 5 hours and 3 days and then washed with dimethyl carbonate and dried under vacuum. For electrochemical tests, galvanostatic cyclations were carried out on a Maccor battery cycler, using a current density of 100 mA/g, nearly corresponding to C/20. Specific capacities in both lithium cells and Li-ion cells have been always calculated in respect to the mass of NaAlH4 in the negative electrode. The qualitative composition of the permenent gas accumulated in the first full discharge/charge galvanostatic cycle (OCV-10mV-2.0V at C/20) within a two electrodes lithium pouch cell with a NaAlH4-based electrode has been obtained by GC-TCD with a Hewlett Packard
6890 gas chromatograph, equipped with a thermal conductivity detector and a packed MS 13X stainless steel column to separate the permanent gases. Argon was used as carrier gas at 25 ml min1
. The gaseous sample has been extracted from the pouch cell after cycling in the Ar-filled glove
box by using a gas-tight lock syringe (Aldrich) and then injected into the gas chromatograph. A benchmark cell cycled in the voltage range (OCV-10mV-2.0V at C/20) has been also checked. The calibration of gas chromatograph was performed by using standard gas in the Ar carrier gas. LiFePO4 was purchased by CustomCells in the form of 1.0mAhcm-2 specific capacity electrode sheets.
3. Results and discussion 3.1 Gas release upon charge EPT measurements on Li/LP30/NaAlH4 cells have been carried out in order to highlight the possible gas evolution upon charge in parallel with the establish back-conversion reaction. Alanate electrodes have been cycled through a full galvanostatic discharge and charge at C/20 while pressure have been monitored. As benchmark, a similar test has been performed using a stainless steel electrode as working, in order to consider gas evolution related to electrolyte evaporation. The evolution of the internal pressure of both the alanate and the benchmark cells are shown in the figure 1a. In lithium cells, the hydride conversion reactions of sodium alanates have been already clarified in the literature [1,3], and imply: (a) upon discharge a complex multistep reduction to hexahydrides and to metals, both Na and Al and (b) upon charge the reverse formation of NaAlH 4. In the initial 6 hours rest at OCV, the pressure of the alanate cell matches the SS benchmark. A small pressure increase soon starts as discharge begins, and continues negligible upon charge up to 1.2V. Such a small pressure increase is expected in consideration of the possible electrolyte side reduction, which can cause gaseous products evolution.[8] In the last stage of the charge, above 1.2V, the internal pressure of the alanate cell noticeably rises. In order to clarify the identity of the constituents of the permanent gas produced above 1.2 V, a GC-TCD experiment has been carried out on the gaseous products accumulated within pouch cells after one galvanostatic cycle at C/20 with voltage cutoffs of 2.0 and 1.2 V, respectively. The comparison of the two gas chromatographic results are shown in the figure 1b. Carbon monoxide (CO) has been detected in the produced gas phase of both cells charged at 1.2 and 2.0 V whereas hydrogen is the most abundant constituent only within the cell charged at 2.0 V and is almost absent in the cell charged at 1.2 V.
The formation of CO can be related to the electrolyte degradation and the parallel growth of the SEI layer [5]. On the other hand hydrogen release from alanates is generally conceived upon thermal desorption or by hydrolysis [9], and very few studies can be found about electrochemical hydrogen evolution. Among the few published, Senoh et al.[6] in 2008 used alanates solutions in THF and Ni electrodes to experimentally evaluate the hydrogen electrode reaction of lithium and sodium aluminum hydrides, metallic lithium and sodium where used as references. The authors evaluated the onset of the hydrogen evolution and highlighted the necessity of an “activation” process. Thermodynamic estimations suggests that hydrogen could evolve following to two processes: MAlH4 = M+ + Al + 2H2 + e- (R1) MAlH4 = M+ + AlH3 + ½H2 + e-
(R2)
According to Senoh, despite reaction (R1) is the thermodynamic favorite process (0.5V vs Na), mechanism (R2) (0.98V vs Na) is the kinetically dominant. In fact 0.5 mol of hydrogen electrochemically generated occurred above 1.2V vs Na. Both processes are in the potential range explored during the electrochemical tests up to now reported for the alanate as conversion electrode in lithium cells, in fact anodic cut-off potential is usually set in between 2-3V vs Li.[1,10] Therefore we updated the thermodynamic E° by using more recent literature ΔrG°, [11,12] obtaining for reaction (1) and (2) 0.56 and 1.1V vs Na respectively, corresponding to 0.86 and 1.4V vs Li. Our EPT and GC-TCD data suggest that hydrogen evolution occurs in the alanate cell at potentials that nicely matches those estimated for reactions (R1) and (R2). Once formed, gaseous hydrogen leaves the electrode surface and cannot reduced back in the subsequent cycles. Therefore the NaAlH4 oxidation also involve a fully irreversible process (reactions R1 and R2) that causes the loss active material useful for conversion. An easy solution to limit the internal pressure increase and thus possible hydrogen evolution is the adoption of a smaller anodic cutoff voltage in charge, e.g. 1.2V. Figure 2 shows the galvanostatic cycling of alanate electrodes up the standard 2.50V anodic cutoff limit (benchmark) in comparison to the cycling limited to 1.20V. The effect of limited voltage range is clear from the second cycle, where the capacity in discharge is 30% larger compared to the benchmark. However in the subsequent cycled the trend of the capacity fade is similar for both anodic voltage cutoffs. This evidence suggests that the possible hydrogen evolution is not the unique failure mechanism. Nevertheless it must be emphasized that after 10 cycle, the cell charged at the lower cutoff is capable to exchange nearly 400mAh/g, a capacity 56% larger then what is delivered disregarding the oxidation process beyond 1.2V.
3.2 Instability of the electrode/electrolyte interface Among other possible failure mechanisms, the chemical reactivity of alanate towards the liquid electrolyte must be taken into account. Hitherto, all studies on electrochemical conversion of alanates in lithium cells have been carried out in conventional LP30 electrolyte, i.e. in a 1M LiPF6 solution in 1:1 v/v ethylene carbonate–dimethyl carbonate [1,3,9]. This electrolyte is not the best choice, considering the well-known chemical ability of alanates to reduce the carbonyl functional group [2]. In a typical lithium cell, the spontaneous formation of electrical double layers [13] at the electrodes leads to an open circuit potential between 2 and 3 V regardless of the electrode chemistry. On the other hand lithium cells containing LP30 electrolyte and NaAlH4 electrodes exhibit open circuit potentials of about 1.1V vs Li [1,3]. This small OCP value is a clear indication of the existence of a specific reactivity between the LP30 and alanate electrodes. Figure 3a shows the Nyquist plots obtained from the electrochemical impedance spectroscopy measurement of NaAlH4 electrodes stored in contact to LP30 at OCP for a week. Figure 3b shows the evolution of the OCP for a week. The impedance curve is characterized by a semicircle in the high frequencies region associated to the “solid electrolyte interface” (SEI) presence, followed by a large semicircle at mid frequencies relative to the electrodic process and finally a convoluted line at low frequencies. The observed not blocking interface can be described by a R(RQ)(RQ)Q circuit. At t=0 the high frequency resistance is approximately 30 ; after 1 day this value constantly increases until 78 and then decreases again to 30 after 6 days. These oscillations in the estimated resistance of the SEI suggest the formation of an unstable film on the electrode surface, that continuously grows, breaks and grows again. Moreover the smooth OCP increase confirms that the interphase tends to the equilibrium only very slowly. The chemical nature of the surface film has been investigated by FTIR-ATR spectroscopy. ATR mode, allows to collect the IR signal coming from the electrode surface, and therefore to highlight the absorbance due to the film growing on it. Spectra as function of aging time in the LP30 electrolyte solutions are given in figure 4. The pristine electrode exhibits the typical vibrational frequencies of NaAlH4: [AlH4]stretching mode at 1648 cm-1 and the bending at 900 and 712 cm-1. Beside them, Na3AlH6 peaks are also observable. Specifically, stretching at 1400 cm-1 and bending around 1189 and 1081 cm1
.[14,15] As evident from figure 4, LP30 decomposition starts as soon as it comes in contact with
NaAlH4. After only 10 minutes, vibrational frequencies associated with alanates disappear and wide vibrational modes are observable in the spectrum. As already observed for lithium alanate, the
electrode surface is covered by a complex film, which include insoluble organic lithium carbonates, polymeric anhydrides, fluoro-alkyl phosphates and aluminum oxides [16]. Specifically, at 1803 and 1750 cm-1 are visible the stretching modes of the C=O groups and at 1272 cm-1 the C-O stretching relative to carbonates. It's also observable the frequencies associated to the fluoro-alkyl phosphates: P=O at 1195 and 1159 cm-1; P-O-C at 1066 and 968 cm-1; and P-O at 796 and 774 cm-1. Also, at 640 cm-1 there is the appearance of aluminum oxides stretching mode. After 1 hour, peaks appear reduced in intensities. Also, prolonging the aging time, the signals intensity constantly changes. These evidences suggest that the electrode/electrolyte interface is highly reactive. Sodium alanate, thanks to its high reducing power, largely reacts with the electrolyte solution and several reaction products precipitate on the electrode surface. The large surface area of the alanate nanomaterial as the catalytic effect of the carbon coating enhances the surface reactivity and therefore the process continuously evolves upon time. Being the SEI layer unstable, the active material is progressively consumed upon time leading to an unavoidable capacity fading. In summary, besides the mechanical instability, two additional reasons for electrode failure have been disclosed, both related to active material consumption: its irreversible oxidation upon charge and its chemical reaction with the electrolyte, i.e. the hydrogen evolution and the production of an unstable solid electrolyte interface. Both problems could find technological solutions.
3.3 Ionic liquids as co-solvents for the stabilization of the interface In order to avoid the hydrogen evolution, a severe limitation of the anodic cutoff to 1.2V is the easiest solution. On the other hand in order to mitigate the alanate attack to the carbonyl, we tested the use of a carbonyl-free solvent as solvent or co-solvent for the liquid electrolyte. Therefore, the use of a different type of electrolyte systems has been considered. Ethers, like tetraethylen glycol dimethyl ether (TEGDME) or the mixture 1,3-dioxolane-1,2-dimethoxyethane (DOL-DME), failed: in fact ethers are able to dissolve alanates, thus boosting interface instability in terms of potential, impedance and redox activity (data not shown). On the contrary ionic liquids are promising solvents. Ionic Liquids have attracted a lot of attention as electrolyte systems thanks to their high ionic conductivity, low toxicity as well as their high thermal, chemical and electrochemical stability [17]. We tested the N-n-butyl-N-methylpyrrolidinium hexafluorophosphate ([Py14]PF6), an Ionic Liquid consisting of an alkylmethylpyrrolidinium cation and the same anion PF6- contained in LP30 as cosolvent. Previous studies [18] demonstrated that its addition to LP30 improved the electrochemical and thermal properties, reduced the flammability of the volatile carbonates, enhancing the safety of the overall system. For our purpose, [Py14]PF6 has been added at the standard electrolyte and three
different mixtures have been prepared increasing the amount of IL from 30 to 70 wt.%. Table 1 reports composition, labelling and molality of the resulting electrolytes.
Larger concentrations and the pure IL as solvent have not been investigated due to the high viscosity. In order to evaluate the effect on electrochemical performances, lithium cells using the enriched IL solutions as electrolyte and alanate as working electrode have been assembled and tested in galvanostatic regimes in the anodic limited voltage range 0.01-1.20V. Figure 5a shows the voltage profiles, while figure 5b reports capacity values and cell efficiency for the three IL added solution in comparison with the pure LP30 electrolyte. The total capacity exchanged at the end of the first discharge decreases with the amount of ionic liquid present in the electrolyte, moving from 2010 mAhg-1 with the pure LP30 to 840 mAhg-1 with the addition of 70 wt.% of IL. However, despite differences in the total exchanged capacity values, first discharge profiles appear similar for the three solutions apart a down shifting of 100200 mV to be related with the increased amount of IL added to electrolyte. Such overvoltages can be originated from the lower Li+ transference number of solutions after the addition of IL. Concerning charge profile, it's evident that the addition of ionic liquid has beneficial effects on the efficiency of the conversion process, that increases from 45% in bare LP30, to 71.4% in IL70 solution, see figure 4b. Charge evolves through two plateaus around 0.4 and 0.8 V, similarly to the profile in LP30, but the addition of ionic liquid leads to an extension of the plateau around 0.8V. Interestingly, conversely to discharge process, charge is not affected by larger overvoltages. The addition of 30% of [Py14]PF6 to LP30 appears the best compromise between capacity exchanged and recharge efficiency, and was therefore considered for further characterizations. EIS and FTIR-ATR were used to evaluate the effective improvement of interface properties. Figure 6a shows the Nyquist plots collected for the NaAlH4 electrode in contact with the IL30 electrolyte in a three electrode lithium cell. The impedance shape is similar to that already discussed of figure 4 and it has been also in this case fitted by using a R(RQ)(RQ)Q equivalent circuit. Upon time, the resistance associated with SEI constantly increases from 2 to 43 . In the same way, charge transfer resistance increases from hundreds to a thousand ohm, while the final linear trend, probably indicative of a diffusion kinetics limitation, do not basically change upon time. These trends suggest a noticeable different surface chemistry compared to the LP30 electrolyte. In fact while NaAlH4/LP30 interaction gave rise at the interphase to an unstable film, which continuously forms, breaks and creates again, in the IL containing solution the surface film continuously and slowly grows.
The figure 6b shows the trend upon time of the SEI resistance (RSEI) of the NaAlH4/IL30 interface, in comparison to that observed at the NaAlH4/LP30 interface. The replacement of carbonates based electrolyte with an ionic liquid produces beneficial effects on the film formation process. In the case of plain alkylcarbonate based electrolyte, a 30 resistive film is quickly formed and is therefore subject to continuous fluctuations. Conversely, the IL30 solution showed small initial RSEI values that continuously increase upon time. After 1 day, such growth becomes lower and the RSEI settle on 30 . Anyway, the trend observed in RSEI reveals that alanate still reacts largely with electrolyte but the resistivity and the instability of the SEI appears to be mitigated. In fact, as occurring for a LP30 soaked electrode, upon aging in the IL30 solution the electrode surface is enriched by reduction products resulting from electrolyte decomposition. Figure 7 shows FTIR-ATR spectra collected for the NaAlH4/IL30 system. In contrast with the experiments performed with pure LP30, in IL30 the film increases gradually upon time. Specifically, after 10 minutes of aging there is the appearance of the vibrational modes related to fluoro-alkyl phosphates at 850 cm-1 and the Al-O bending at 650 cm-1. These signals constantly increase in intensity and after prolonged time of aging it's observable also the stretching modes of P=O group at 1160 cm-1 and P=O-R group at 976 and 780 cm-1 while the intensity of peak at 850 cm-1 largely increase. Furthermore, stretching frequencies related to deposition of organic lithium carbonates are clearly visible at 1806, 1773 and 1306 cm -1, while the C-O stretching is at 1080 cm-1. In summary in the IL-enriched electrolyte IL30, the interaction between alanate and carbonates is still large, as evident from the several products deposited on electrode surface. However the addition of the IL allows the formation of a more stable film. Beneath the addition of IL is not ultimate to optimize the alanate electrochemical performance, it mitigates the alanate reactivity towards the electrolyte. One may speculate that the strong alanate reducing power really needs to be controlled by a careful coating of alanate particles or by their confinement into an inert matrix. This strategy demonstrates to be effective for several other electrode materials, especially for high potential cathodes capable to catalyze the electrolyte oxidation.[19]
3.4. Formulation of a full Li-ion cell with improved electrolyte and controlled voltage limits Even if optimization is still a distant goal, the here-reported improvements led us to assemble a lithium ion cell in order to demonstrate, even for few cycles, the use of NaAlH4 as negative electrode. The figure 8 shows the voltage profile in the first 3 cycles of a full Li-ion cell with an NaAlH4-based negative electrode. The IL30 solution has been used as electrolyte. As positive electrode a commercial LiFePO4 has been chosen because of its well-known behavior, in terms of
constant working potential and stability upon cycling. Electrodes masses have been balanced in order to match the capacity in the first discharge. Upon cycling a time cutoff has been used to limit the capacity to the 67% of the theoretical one. This capacity limitation has been adopted to avoid lithium plating at the negative side and overcharge at the positive collector. Considering the redox potential of NaAlH4 and LiFePO4, the applied cycling voltage range was 2.25-3.70V. A metallic lithium reference electrode has been also used to constantly check the alanate electrode potential. Figure 8 shows the (a) lithium ion cell voltage and (b) alanate negative electrode potential vs Li. as a function of delivered specific capacity. The first cycle cell voltage profile, highlighted in red, shows the signature of the multistep alanate conversion mechanism[3]; charge mostly develops between 2.9 and 3.25V (alanate conversion at 0.55 and 0.2V vs Li) while discharge occurs around 2.6V (recombination to alanate at 0.8V), the average cell voltage is around 2.8V. As expected, first charge finishes when the capacity limit is reached, i.e. 1330mAh/g, and more than 60%, i.e. 849mAh/g is returned upon the following discharge. In the second cycle efficiency markedly improves: 1282mAh/g are delivered during charge and 1000mAh/g in discharge. Upon further cycling, capacity progressively fades because still efficiency never exceeds 90%. Lithium plating does not occur during the firsts three cycles, but not even in the following, and the anode potential never drops below 0.04V. The performance of this Li-ion cell is far from being optimal: this demonstrator shows poor cycling performance, thus the alanate negative electrodes are far from being ready for application. On the other hand our Li-ion cell is the first ever reported demonstration of the use of the sodium alanate as negative electrode material in a full Li-ion cell and illustrates the current potentialities and open problems of this battery system.
4. Conclusion In this communication we discussed for the first time new insights about the failure mechanisms of NaAlH4 negative electrodes in lithium cells. Besides the well-known mechanical instability of the electrode due to the volume variation upon conversion also hydrogen gas evolution upon oxidation and the instability of the SEI layer have been proved and discussed. Possible mitigation strategies have been highlighted including the limitation of the anodic cutoff potential of the NaAlH4 electrode as well as the use of composite electrolytes based on mixtures of organic carbonates and an ionic liquid. Both these strategy allowed the demonstration of a full Li-ion cell with a NaAlH4 negative electrode, a LiFePO4 positive electrode and a composite electrolyte based on a EC:DME:LiPF6:Pyr1,4PF6 mixture. Although only three cycles have been reported, our prototype is, as far as we know, first ever reported full Li-ion formulation that exploits the alanate
conversion chemistry. Much further knowledge is needed to fully develop a reliable and longlasting configuration and an intense research activity is in progress in our laboratories to tackle this challenge.
Acknowledgements This work is carried out in the framework of the Italian project (FIRB-Futuro in ricerca 2010) “Hydrides as high capacity anodes for lithium ion batteries” (RBFR10ZWMO), supported by Italian Minister for University and Research. Thanks are due to F.Langerame for the kind.
References [1]
J.A. Teprovich, J. Zhang, H. Colón-Mercado, F. Cuevas, B. Peters, S. Greenway, R. Zidan, M. Latroche, Li-Driven Electrochemical Conversion Reaction of AlH3, LiAlH4, and NaAlH4, J. Phys. Chem. C. 119 (2015) 4666–4674. doi:10.1021/jp5129595.
[2]
L. Silvestri, S. Forgia, L. Farina, D. Meggiolaro, S. Panero, A. LaBarbera, S. Brutti, P. Reale, Lithium Alanates as Negative Electrodes in Lithium-Ion Batteries, ChemElectroChem. 2 (2015) 877–886. doi:10.1002/celc.201402440.
[3]
L. Silvestri, L. Farina, D. Meggiolaro, S. Panero, F. Padella, S. Brutti, P. Reale, Reactivity of Sodium Alanates in Lithium Batteries, J. Phys. Chem. C. 119 (2015) 28766–28775. doi:10.1021/acs.jpcc.5b10297.
[4]
L. Cirrincione, L. Silvestri, C. Mallia, P.E. Stallworth, S. Greenbaum, S. Brutti, S. Panero, P. Reale, Investigation of the Effects of Mechanochemical Treatment on NaAlH4 Based Anode Materials for Li-Ion Batteries, J. Electrochem. Soc. 163 (2016) A2628–A2635. doi:10.1149/2.0731613jes.
[5]
J. Cabana, L. Monconduit, D. Larcher, M.R. Palacín, Beyond intercalation-based Li-ion batteries: The state of the art and challenges of electrode materials reacting through conversion reactions, Adv. Mater. 22 (2010) 170–192. doi:10.1002/adma.201000717.
[6]
H. Senoh, T. Kiyobayashi, N. Kuriyama, Hydrogen electrode reaction of lithium and sodium aluminum
hydrides,
Int.
J.
Hydrogen
Energy.
33
(2008)
3178–3181.
doi:10.1016/j.ijhydene.2008.01.019. [7]
J. Manzi, S. Brutti, Surface chemistry on LiCoPO4 electrodes in lithium cells: SEI formation and
self-discharge,
Electrochim.
Acta.
222
(2016)
1839–1846.
doi:10.1016/j.electacta.2016.11.175. [8]
M. Holzapfel, A. Würsig, W. Scheifele, J. Vetter, P. Novák, Oxygen, hydrogen, ethylene and CO2 development in lithium-ion batteries, J. Power Sources. 174 (2007) 1156–1160.
doi:10.1016/j.jpowsour.2007.06.182. [9]
B. Sakintuna, F. Lamari-Darkrim, M. Hirscher, Metal hydride materials for solid hydrogen storage:
A
review,
Int.
J.
Hydrogen
Energy.
32
(2007)
1121–1140.
doi:10.1016/j.ijhydene.2006.11.022. [10] L. Silvestri, A. Paolone, L. Cirrincione, P.E. Stallworth, S. Greenbaum, S. Panero, S. Brutti, P. Reale, NaAlH4 nanoconfinement in mesoporous carbon for application in lithium ion batteries, J. Electrochem. Soc. 164 (2017) A1120–A1125. [11] C. Qiu, S.M. Opalka, G.B. Olson, D.L. Anton, Thermodynamic modeling of the sodium alanates and the Na–Al–H system, Int. J. Mater. Res. 97 (2006) 1484–1494. doi:10.3139/146.101410. [12] J. Graetz, J. Reilly, G. Sandrock, J. JOHNSON, W.M. Zhou, J. Wegrzyn, Aluminum hydride AlH3, as a hydrogen storage compound., Brookhaven National Laboratory, Upton, NY, 2006. doi:10.2172/899889. [13]
J.O. (John O.. Bockris, A.K.N. Reddy, M. Gamboa-Aldeco, Modern electrochemistry. Volume 1, Ionics, Kluwer Academic, 2002.
[14]
B.C. Smith, Infrared spectral interpretation : a systematic approach, CRC Press, 1999.
[15]
J.-C. Bureau, Z. Amri, P. Claudy, J.-M. Letoffe, Etude comparative des hexahydrido-et des hexadeuteridoaluminates de lithium et de sodium. I - Spectres raman et infrarouge de Li3-et Na3AlH6, et Li3AlD6, Mater. Res. Bull. 24 (n.d.) 23–31.
[16]
T. Kawamura, S. Okada, J. Yamaki, Decomposition reaction of LiPF6-based electrolytes for lithium
ion
cells,
J.
Power
Sources.
156
(2006)
547–554.
doi:10.1016/j.jpowsour.2005.05.084. [17]
D.R. MacFarlane, N. Tachikawa, M. Forsyth, J.M. Pringle, P.C. Howlett, G.D. Elliott, J.H. Davis, M. Watanabe, P. Simon, C.A. Angell, Energy applications of ionic liquids, Energy Environ. Sci. 7 (2013) 232–250. doi:10.1039/C3EE42099J.
[18]
A. Tsurumaki, M.A. Navarra, S. Panero, B. Scrosati, H. Ohno, N-n-Butyl-Nmethylpyrrolidinium hexafluorophosphate-added electrolyte solutions and membranes for lithium-secondary
batteries,
J.
Power
Sources.
233
(2013)
104–109.
doi:10.1016/j.jpowsour.2013.01.131. [19]
M. Agostini, A. Matic, S. Panero, F. Croce, R. Gunnella, P. Reale, S. Brutti, A mixed mechanochemical-ceramic solid-state synthesis as simple and cost effective route to high performing LNMO spinels, Electrochim. Acta. 235 (2017) 262–269.
Figure captions
Figure 1. (a) Galvanostatic EPTs of a Li/LP30/NaAlH4 (black) and a reference Li/LP30/SS (blue) cell in the 0.01-2.5V voltage range. The Li/LP30/NaAlH4 measure has been repeated in the 0.011.2V voltage range (red). C/20 current rate; (b) results of gas chromatographic analysis of the internal gases released within pouch cells discharged at 10 mV and then charged at 1.2 and 2.0 V, respectively.
Figure 2. Galvanostatic cyclation of Li/LP30/NaAlH4 cells in the 0.01-2.50V (black) and in the 0.01-1.20V (red) voltage range
Figure 3. Nyquist Plot (a) and OCP (b) of NaAlH4 in a lithium cell containing LP30 as electrolyte
Figure 4. FTIR-ATR spectra of NaAlH4 electrode upon aging in LP30 electrolyte
Figure 5. a) First discharge/charge voltage profiles obtained in galvanostatic mode as function of amount of ionic liquid into the electrolyte. Results obtained with LP30 are used as reference and b) Efficiency and Specific Capacity as function of electrolyte solution plot.
Figure 6. (a) Nyquist plot of NaAlH4 sample using IL30 as electrolyte; (b) RSEI as function of time at the NaAlH4/IL30 interface in comparison to that developed at the NaAlH4/LP30 interface.
Figure 7. FTIR-ATR spectra of NaAlH4 electrode upon aging in IL30 electrolyte
Figure 8. (a) Cell voltage profiles and (b) anode potential in a NaAlH4/LiFePO4 full cell.
Table 1. Composition of the electrolytes. Label
LP30 wt.%
IL wt.% LiPF6 Molality Ionic Conductivity i/S*cm-1
LP30
100
/
0.87
1.2 10-2
IL30
70
30
0.58
2.8 10-2
IL50
50
50
0.41
3.0 10-2
IL70
30
70
0.24
1.9 10-2