Article pubs.acs.org/cm
New Tailored Sodium Salts for Battery Applications Anna Plewa-Marczewska,*,†,‡ Tomasz Trzeciak,† Anna Bitner,† Leszek Niedzicki,† Maciej Dranka,† Grazẏ na Z. Ż ukowska,† Marek Marcinek,† and Władysław Wieczorek† †
Warsaw University of Technology, Faculty of Chemistry, Noakowskiego 3, 00640 Warsaw, Poland ALISTORE-European Research Institute, Fédération de Recherche CNRS no. 310433, rue Saint Leu80039 Amiens, CEDEX, France
‡
S Supporting Information *
ABSTRACT: This article describes synthesis and basic electrochemical and structural properties of newly designed sodium salts for application in liquid nonaqueous sodium electrolytes. There has been two imidazole fluorine derivative sodium salts synthesized: sodium 4,5-dicyano-2-(trifluoromethyl)imidazolate (NaTDI) and sodium 4,5-dicyano-2-(pentafluoroethyl)imidazolate (NaPDI). The structure of the salts has been confirmed by means of Raman spectroscopy, nuclear magnetic resonance (13C NMR and 19F NMR), X-ray diffraction, thermogravimetry (TGA), and differential scanning calorimetry (DSC). Electrochemical characterization included ionic conductivity measurements, dynamic viscosity, and electrochemical stability of solutions of the salts in propylene carbonate (PC) at different temperatures. Raman spectra of the electrolytes have been performed to carefully monitor the degree of ionic associations specially ion pairing tendencies.
1. INTRODUCTION The first successful attempt of a sodium battery was undertaken in 1967 by Ford Motor Company (USA) in the sodium−sulfur battery. This innovation was possible from the discovery by Ford researchers of the favorable Na+ ionic conduction properties in β-alumina. Sodium ion conduction reaches about 10 S·cm−1 at 300 °C and was sufficient for a ceramic electrolyte. In such a kind of batteries, liquid sodium worked as an anode vs liquid sulfur as a cathode with a voltage of 2.1 V. Because of high operating temperatures, corrosion problems, and poor overcharge properties, practical market applications have become extremely specific and limited to stationary energy storage.1 The second advance was the replacement of the sulfur cathode with a metal chloride in leading to the sodium/nickel chloride (also known as Zebra) cells with a potential of 2.58 V. NaAlCl4 was introduced as intermediate electrolyte with facilitated contact between solid cathode and real electrolyte. Operating temperatures were in that case considerably lower than sodium−sulfur batteries at 330 °C. Na is deposited on the anodic side during the first charge of the cell. This avoids the handling of metallic Na during cell assembly.2 Growing application of Li-ion and Li-polymer batteries for hybrid electric cars, electric vehicles, mobile devices, etc., requires a large amount of lithium in the form of Li2CO3 every year. High price and relatively small world reserves of 13 million tons forced scientists to look for new materials possible for use in battery technology.3 That is a clear reason for research effort to find new wellperforming types of batteries of high energy density, low cost, increased safety, and tolerated by the environment. The logical way is to look for lithium analogues with similar chemical © 2014 American Chemical Society
properties and higher resources on the planet (sixth most abundant element and the most abundant alkali metal on Earth). Sodium has only slightly lower its red-ox potential (2.72 V) than lithium. Low cost (about 7 times lower than lithium) and enormously easier to obtain and almost unlimited worldwide reserves make sodium a really promising candidate. Metallic sodium does not form dendrites being a key issue if lithium metallic anodes (with gigantic theoretical capacity of 1190 mAh·g−1) were tested in the present batteries. Several secondary advantages only promote the sodium concept, e.g, cheaper than aluminum current collectors.4 Surprisingly, while there is a numerous list of publications on the new lithium-ion cell electrode materials (or additions to those), with also big amount of those on electrolyte additives, there has been very little stress given to new salts used in lithium electrolytes themselves. Moreover, a limited but constantly being increased number of publications are devoted to sodium salts for electrolytes, and most of them are based on the lithium analogues studied in lithium electrolytes:5 NaClO46 and NaPF67 Apart from anions known before 1990 (ClO4−, AsF6−, PF6−, BF4−, or and CF3SO3−), so far there were very few promising introductions of new anions for nonaqueous electrolytes: starting with TFSI− (N(SO2CF3)2−),8 then methide ones, C(SO2CF3)3−,9 C(SO2CF3)2(RCO)−,10 and N(SO2C2F5)2− (BETI−).11 Unfortunately, all of them, including TFSI−, BETI−, and methide anions, had the major drawback of being Received: October 11, 2013 Revised: May 9, 2014 Published: May 23, 2014 4908
dx.doi.org/10.1021/cm403349t | Chem. Mater. 2014, 26, 4908−4914
Chemistry of Materials
Article
unable to form a passivation layer on Al current collectors.12 Meanwhile, a wide class of sulfone-imide and methide-imide anions was designed and synthesized13 without huge success in application in polymer electrolytes. The most successful among imide class conducting salts is LiTFSI within ionic conductivity of about 4.25 × 10−3 S·cm−1 in PC at 30 °C14 (at 0.6 mol· dm−3). The most important and successful of the imide salts class were LiTFSI and LiBETI. Their failure to be predominant on the batteries market was due to , among other mentioned problems, their synthesis. The novel, promising concept of the application of new anions is based on the application of “Hückel anions”. The name came from the transposition of the Hückel rule predicting the stability of the aromatic systems. One of the most common examples of this type of anion is 4,5-dicyanotriazole (DCTA). This particular structure is completely covalently bonded and shows very stable 6π (or 10π electron if CN bonds are involved in calculations) configuration. It can be produced from commercially available precursor and even more importantly does not comprise fluorine atoms. Salts of this type of anion were found to exhibit high (∼300 °C) thermal stability. LiDCTA was successfully tested in PEO matrices systems as a promising, improved electrolyte for rechargeable lithium batteries.15 Unfortunately DCTA failed as a component of the EC/DMC (1:1) battery electrolyte. The list of requirements for lithium salts to fulfill in order to become new predominant salt on market of lithium ion cells is not very long, but no existing salt fulfills it. Transference number above 0.5 (or at least better than LiPF6 in optimized carbonate solvent mixtures, which has a transference number of 0.3−0.41617), conductivity higher than 1 mS·cm−1 (10−3 S· cm−1), no decomposition in range of 0 4.5 V vs Li, no aluminum corrosion in this range, low price (at least lower than LiPF6, but the lower, the better), nontoxicity, moisture-proof (and air proof, stability in room atmosphere, easiness of handling), thermal stability up to at least 100 °C, and low association rate (lower than LiPF6 or very weakly associating LiClO418) is what is necessary to obtain by researchers.19 Several patents and papers published before described synthesis, production, and structural and electrochemical properties of pentafluorinated anions based lithium salts, which showed superior properties to the presently used lithium salts in electrolytes.20 With this in mind, in this work the new “tailored” anions especially for application as sodium electrolytes in sodium ion cells have been designed and investigated. This article presents the synthesis route and initial properties of newly designed sodium salts, e.g., sodium 4,5-dicyano-2(trifluoromethyl)imidazolate (NaTDI) and sodium 4,5-dicyano-2-(pentafluoroethyl)imidazolate (NaPDI) for application in liquid nonaqueous sodium electrolytes.
2.2. Salt Synthesis. Applied synthesis was based on the work previously reported by us.21 The synthesis scheme is shown in Figure 1.
Figure 1. Synthesis route of sodium salts (R = −CF3 or −CF2CF3).
2.2.1. Sodium 4,5-Dicyano-2-(trifluoromethyl)imidazolate (NaTDI). Trifluoroacetic anhydride (38.7 mL, 0.274 mol) was poured into a solution of diaminomaleonitrile (24.65 g, 0.228 mol) in 1,4dioxane (250 mL). The mixture was refluxed under inert argon atmosphere until complete conversion (TLC, silica gel, benzene−ethyl acetate 1:1 v/v). The resulting mixture was evaporated under vacuum to remove both solvent and acid. The viscous liquid residue was dissolved in diethyl ether (200 mL), and the resulting solution was extracted with aqueous solution of sodium carbonate (49.5 g, 0.399 mol in 300 mL water). The aqueous phase was washed with diethyl ether (2 × 100 mL). Then activated carbon was added to the water solution, and the slurry was heated (5 h, 50 °C). After filtering off the charcoal, water was evaporated under vacuum. In order to remove traces of water, the residue was evaporated with absolute ethanol (2 × 50 mL). Then the residue was dissolved in anhydrous acetonitrile (250 mL), and the precipitate was filtered off. Filtrate was fractionally crystallized. The crude product was purified by column chromatography (aluminum oxide neutral 100 g; acetonitrile 500 mL as eluent). The evaporation of the solvent and drying under vacuum (10 mbar, 90 °C, 72h) gave the product (37.9 g, 79.9% yield) as a white powder. 13C NMR (125 MHz; DMSO-d6)/ppm: 115.5 (s, −CN), 118.8 (d, J = 2 Hz, CC), 120.1 (q, J = 269 Hz, −CF3), 147.4 (q, J = 37 Hz, C− CF3). 19F NMR (470 MHz; DMSO-d6)/ppm: −60.3 (s, −CF3). Raman (780 nm)/cm−1: 2256 (vs, CN stretching), 1499 (vs, ring stretching), 1458 (s), 1321 (s, CN ring stretching), 994 (δ ring deformation NCN), 342, 161. 2.2.2. Sodium 4,5-Dicyano-2-(pentafluoroethyl)imidazolate (NaPDI). Pentafluoropropionic anhydride (25.0 mL, 0.127 mol) was poured into a solution of diaminomaleonitrile (11.41 g, 0.106 mol) in 1,4-dioxane (220 mL). The mixture was refluxed under argon atmosphere until complete conversion (TLC, silica gel, benzene− ethyl acetate 1:1 v/v). The resulting mixture was evaporated under vacuum. The viscous liquid residue was dissolved in diethyl ether (150 mL), and the solution was extracted with a solution of sodium carbonate in water (22.9 g, 0.185 mol, in 150 mL of water). The aqueous phase was washed with diethyl ether (2 × 100 mL). Then activated carbon was added to the water solution, and the slurry was heated (5 h, 50 °C). After filtering of the charcoal, water was evaporated under vacuum, and the residue was evaporated with absolute ethanol (2 × 50 mL). Then the residue was dissolved in anhydrous acetonitrile (250 mL), and the precipitate was filtered off. Filtrate was fractionally crystallized. The crude product was purified by column chromatography (aluminum oxide neutral 100 g; acetonitrile 500 mL as eluent). The evaporation of the solvent and drying under vacuum (10 mbar, 90 °C, 72 h) gave the product (21.15 g, 77.6% yield) as a white powder. 13C NMR (125 MHz; DMSO-d6)/ppm: 109.7 (td, J = 37 and 250 Hz, −CF2−), 115.0 (s, −CN), 118.8 (tq, J = 38 and 286 Hz, −CF3), 119.3 (s, CC), 146.4 (t, J = 27 Hz, C− CF2).19F NMR (470 MHz; DMSO-d6)/ppm: −110.1 (tm, J = 2 Hz, −CF3), −82.2 (qm, J = 2 Hz, −CF2−). Raman (532 nm)/cm−1: 2259 (vs, CN stretching), 1495 (vs, CC ring stretching 1453 (s), 1325 (vs, CN ring stretching), 1311, 956 (δ ring deformation NCN), 536, 316, 136. 2.3. NMR. 13C and 19F NMR spectra were recorded on a computer interfaced Varian VNMRS 500 MHz spectrometer with VnmrJ 2.2c Software at 125.72 and 470.4 MHz, respectively. Samples were dissolved in dimethyl sulfoxide-d6. All the measurements were
2. EXPERIMENTAL SECTION 2.1. Materials and Reagents. Diaminomaleonitrile (98%), 1,4dioxane (anhydrous, 99.8%), and acetonitrile (anhydrous 99.8%) were purchesed from Aldrich. Trifluoroacetic anhydride (>99%), sodium carbonate monohydrate (99.5%), propylene carbonate (anhydrous 99.7%), and metal sodium (ACS reagent, stick, dry) were purchased from Sigma−Aldrich. Diethyl ether (>99.5%) and ethanol (anhydrous, 99.5%) were purchased from POCh. Dimethyl sulfoxide (DMSOd6,“100%” atom D) for NMR analyses was purchased from Armar Chemicals. All reagents and solvents were used as received. 4909
dx.doi.org/10.1021/cm403349t | Chem. Mater. 2014, 26, 4908−4914
Chemistry of Materials
Article
performed at 25 °C. The spectra was analyzed using MestReNova 5.2.5 software. 2.4. Thermal Measurements. Thermal measurements (thermogravimetric analysis and differential scanning calorimetry) were performed using SDT Q600 by TA Instruments. Samples were heated from room temperature to 450 °C in open ceramic pans in the air atmosphere. Heating velocity was 10 °C·min−1. 2.5. Sample Preparation. All samples were prepared in argon glow-box with moisture level below 5 ppm. Liquid samples were obtained by dissolving dry salts (NaTDI or NaPDI, respectively) in propylene carbonate (PC). To grow crystals, concentrated solutions of NaTDI and NaPDI in PC were heated up to 40 °C in a hermetic container. Dissolved samples lived in a glovebox for crystallization. 2.6. Crystallography. Selected single crystals of NaTDI in PC (1) and NaPDI in PC (2) were mounted in inert oil and transferred to the cold gas stream of the diffractometer. Diffraction data was measured at 100(2) K with graphite-monochromated Mo−Kα radiation on the Oxford Diffraction Gemini A Ultra diffractometer. Cell refinement and data collection as well as data reduction and analysis were performed with the CrysAlisPRO.22 The structures were solved by direct methods and subsequent Fourier-difference synthesis with SHELXS-97.23 Fullmatrix least-squares refinement method against F2 values were carried out by using the SHELXL-9724 and OLEX225 programs. All nonhydrogen atoms were refined with anisotropic displacement parameters. Hydrogen atoms were added to the structure model at geometrically idealized coordinates and refined as riding atoms. Both propylene carbonate molecules in 1 were found to be disordered between two positions with occupancies 0.76:0.24 and 0.89:0.11. Furthermore, one of the CF3 groups in 1 was refined as disordered over two sets of sites with occupancies of 0.93 and 0.07 (Table 1).
2.7. Raman Spectroscopy. Raman spectra were recorded on a Nicolet Almega dispersive spectrometer. Diode lasers with an excitation line at 780 and 532 nm were used. The spectral resolution was about 2 cm−1 for each measurement. Raman experiments were carried out at room temperature in argon atmosphere. 2.8. Ionic Conductivity. The ionic conductivity was measured using electrochemical impedance spectroscopy (EIS) by VMP 3 Multichannel electrochemical analyzer. Frequency range from 500 kHz to 1 Hz with 5 mV a.c. signal was used. The cell was immersed in a HAAKE DC 50 cryostat to control the temperature in the range −10 to 50 °C. The temperature was changed in 10 °C increments, and the samples were equilibrated at each temperature for approximately 90 min. A Bernard Boukamp EQsoftware26 was used for analyzing the obtained impedance data. 2.9. Cyclic Voltammetry. Cyclic voltammetry (CV) for measuring electrochemical stability window measurements were performed on a Bio-Logic Science Instruments VMP3 multichannel potentiostat. Samples were cycled in Na/electrolyte/SS (stainless steel) system at ambient temperature. CV scan rate was 1 mV s−1. 2.10. Viscosity of Electrolytes. Viscosity of electrolytes was measured at rotational viscometer Physica MCR301 (Anton Paar) in inert argon atmosphere. Shear rate was from the range 10 to 1000 s−1. The average values were calculated for share rate from 100 to 1000 s−1.
3. RESULTS AND DISCUSSION 3.1. Thermal Measurements. Thermal analysis showed good thermal stability of salts. The mass loss is less than 1% while heating from room temperature to 300 °C. Both sodium salts are stable up to 330 °C (Figure 2). Above this temperature thermal decomposition starts. This is two-step process for NaTDI and one step for NaPDI.
Table 1. Crystal Data for the Single Crystal X-ray Structures of 1 and 2 compound reference chemical formula M (g·mol−1) crystal size (mm3) crystal system a (Å) b (Å) c (Å) α (deg) β (deg) γ (deg) unit cell volume (Å3) temperature (K) space group Z Dcalc (g·cm−1) radiation, λ (Å) absorption coefficient (μ/mm−1) F(000) Θ range (deg) no. of reflections measured no. of independent reflections Rint parameters/restraints R1, wR2 (I > 2σ(I))b R1, wR2 (all data goodness of fit on F2a Δρmin/max (e·Å−3)
1 C20H12F6N8Na2O6 620.36 0.48 × 0.46 × 0.25 orthorhombic 24.3324(9) 9.1536(2) 11.2469(9) 90.00 90.00 90.00 2505.0(2) 100.0(2) Pna21 4 1.645 Mo Kα, 0.7107 0.180 1248 3.3−32.8 247 691 4777 0.0417 474/83 0.0283, 0.0744 0.0304, 0.0756 1.071 +0.41/−0.26
2 C15H12F5N4NaO6 462.28 0.45 × 0.32 × 0.30 triclinic 9.4529(15) 9.5236(17) 12.302(2) 91.044(15) 108.200(15) 115.199(16) 937.2(3) 100.0(2) P1̅ 2 1.638 Mo Kα, 0.7107 0.176 468 3.4−31.5 62 124 5462 0.0371 282/0 0.0298, 0.0794 0.0320, 0.0812 1.028 +0.56/−0.32
Figure 2. TG curves with derivative signal of (a) NaTDI and (b) NaPDI. Bold line, weight lost; thin line, derivative weight.
3.2. Crystallography. The crystal structure of two sodium salts, 1 and 2, crystallizing from propylene carbonate solutions was studied by single crystal XRD method. Compound 1 crystallizes from the PC solution in the orthorhombic space group Pna21 as colorless plates. X-ray crystal structure determination reveals an ionic salt having the composition of Na+/TDI−/PC = 1:1:1. Coordination sphere around the Na+
Goodness-of-fit S = {Σ[w(F02 − Fc2)2]/(n − p)}1/2 where n is the reflections number and p is the parameters number. bR1 = Σ||F0| − | Fc||/Σ|F0|, wR2 = {Σ[w(F02 − Fc2)2]/Σ[w(F02)2]}1/2. a
4910
dx.doi.org/10.1021/cm403349t | Chem. Mater. 2014, 26, 4908−4914
Chemistry of Materials
Article
Compound 2, similarly to 1 crystallizes from the PC solution as a solvate but comprises two PC molecules per sodium cation (Na+/PDI−/PC = 1:1:2). Thus, in a crystal of 2, each Na+ cation is surrounded by two propylene carbonate molecules and three PDI ligands (Figure 5). Two of the PDI anions are
centers comprising four TDI anions and one PC molecule is depicted in Figure 3. Two of the dicyanoimidazolato ligands are
Figure 3. Extended view of coordination environment of the sodium cation in 1. Displacement ellipsoids are drawn at 50% probability level. Hydrogen atoms are omitted for clarity. Selected bond lengths [Å]: Na(1)−N(1) 2.5085(13), Na(1)−O(1) 2.347(3), Na(1)−N(4) 2.5377(12), Na(1)−N(8)1, 2.4964(13), Na(1)−N(9)2 2.5421(13), Na(1)−F(5A) 2.9941(16), and Na(1)−F(1) 2.9076(11). Symmetry codes: (1) 3/2 − x, −1/2 + y, −1/2 + z; (2) 1 − x, 1 − y, −1/2 + z.
coordinated through the nitrogen atom of the imidazole ring, and the remaining two are bound by the cyano groups. Furthermore, the coordination sphere on sodium cations is completed with one coordinated PC molecule. Additional weak interactions with fluorine atoms were observed with Na−F distances ranging from 2.9076(11) to 3.1810(13) Å. In salt 1 TDI anions act as a four dentate N-donor bridging ligands linking adjacent metal ions, which resulting in a metal−organic framework as shown in Figure 4. Electroneutral square grid is composed of Na+ cations and TDI anions, while PC molecules occupy cavities of 3D framework.
Figure 5. Extended view of coordination sphere of the Na+ cation in 2. Selected bond lengths [Å]: Na(1)−O(1) 2.3936(9), Na(1)−O(4) 2.3219(9), Na(1)−N(1) 2.4568(9), Na(1)−N(3)2 2.4234(11), Na(1)−N(4)1 2.4469(9), and Na(1)−F(1) 2.8042(9). Symmetry codes: (1) +x, 1 + y, +z; (2) 1 − x, 1 − y, 1 − z.
coordinated with the cyano groups, and the third one chelates metal with the imidazole nitrogen and fluorine atoms. The Na(1)−F(1) distance is equal to 2.8042(9) Å and is noticeably shorter than Na−F contacts observed in 1. PC molecules capture two coordination sites of the Na+ cation; therefore, one of four imidazole donor centers remains uncoordinated, and dicyanoimidazolate anions in crystal lattice of 2 acts as tridentate ligand. This results in formation of ladder-like coordination polymer propagating in the direction of the Y axis, as depicted in Figure 6. PC molecules are located above and under the plain of the ladder constituting isolated rods arranged in the form of close-packed columnar structure. 3.3. Raman Spectroscopy. The assignment of TDI− anion Raman spectra was supported by DFT modeling and was given after Scheers et al.27 for LiTDI.
Figure 4. View of 3D square grid supramolecular network of 1 projected onto the (1 0 0) plane. Hydrogen atoms are omitted for clarity, and propylene carbonate molecules are shown in wire-frame mode.
Figure 6. View of the 1D ladder-like coordination polymer running along [010] direction in 2. Propylene carbonate molecules are presented as wire-frame. 4911
dx.doi.org/10.1021/cm403349t | Chem. Mater. 2014, 26, 4908−4914
Chemistry of Materials
Article
Similar dependence was found for NaPDI based systems, in the LiTDI and PC based electrolytes, studied intensively by Scheers et al.27,28 formation of this shoulder was ascribed to ion pairs formation. In the most stable configuration for ion pair proposed on the basis of DFT calculations, the cation was coordinated by one of the ring nitrogen and fluorine from CF3 substituent. For concentrated solutions of solvated ionic pairs, however, a good agreement with experimental data was obtained for dimers, where each of the lithium cations was coordinated by ring nitrogen from the first anion and a nitrogen from cyano group of the second anion. In the case of PC based complexes, each of the possible coordination sites of TDI is linked to sodium cations, forming several substructures: (i) the first similar to described above, with sodium cations coordinated by cyano groups and ring nitrogens, (ii) another, where two cations are coordinated by two cyano groups from two different anions, and (iii) with sodium cation caught between fluorine and ring nitrogen. The behavior of the analyzed bands allows to conclude that the studied NaTDI systems at concentrated solutions behave similarly, as in the complexes, forming ions pairs or dimers through both available sites, i.e., cyano and ring nitrogens. In NaPDI based systems, the similarities between liquids and crystalline complexes are less pronounced, in particular for the ring deformation vibration. In contrast to NaTDI complex, in NaPDI complex only one of the ring nitrogens is coordinated by sodium, which indicates that in ionic pairs formation the cyano group may play a more significant role, than for NaTDI. 3.4. Ionic Conductivity of Electrolytes. Electrolytes with new sodium salts represent good ionic conductivity. All solutions are liquid in the tested range of temperature for concentrations of 0.1−1.5 M (Figure 7). Both NaTDI and NaPDI electrolytes exhibit the best conductivity at concentration 0.5 and 1 M in the tested range of temperature. Ionic conductivity is equal to 4 mS·cm−1 at 20 °C for concentrations 0.5 and 1 M for electrolytes with NaTDI (0.5 M, 3.71 mS·cm−1; 1 M, 3.78 mS·cm−1) and NaPDI (0.5 M, 3.79 mS·cm−1; 1 M, 3.83 mS·cm−1). Conductivity of NaTDI in PC excides conductivity for the most conductive concentration (0.5 and 1 M) at temperature higher than 40 °C. For both salts, 1.5 M samples exhibit the poorest conductivity. Moreover, ionic conductivity of the most conducting samples based on sodium salts at 20 °C is higher than for lithium salts with the same anions, where conductivity of LiTDI in PC is equal to 2 and 3mS·cm−1 for 1 and 0.5 M samples, respectively, and conductivity of LiPDI in PC equals 2 mS·cm−1 at 20 °C for concentration 1 M.29 The measured conductivity for NaTDI and NaPDI salts in PC is slightly lower than for marketavailable salts reported in the literature,30 where conductivity of NaPF6, NaClO4, and NaTFSI in PC is between 6 and 8 mS· cm−1 at condition defined as room temperature. 3.5. Cyclic Voltammetry. The PC 1 molar electrolyte solution was characterized by cyclic voltammetry (CV) against Na/Na+ electrodes at 1 mV s−1 scan rate (Figure 8). The CV profile is flat and did not show any signal of decomposition or impurities of the salt itself or electrolyte. The tiny peak at 1.5 V might be attributed to Na electrode impurities. This peak becomes smaller while cycling. The most important information is that the two salts show superior stability up to ∼4 V, which is definitely a promising value when considering the salts as sodium battery candidates.
Additionally, in order to better understand dependence between the structure of the complex and its spectral characteristics, Raman spectra of crystalline NaTDI-PC and NaPDI-PC complexes were compared with those of anhydrous salts and liquid electrolytes. XRD studies of crystalline structure of NaTDI-PC complex revealed that in this structure there are three types of sodium cations: two of them are six-coordinated: by two cyano groups, two ring nitrogens, fluorine and carbonyl group oxygen, and the third one is five-coordinated, lacking coordination through fluorine. In contrast to that, in NaPDI-PC complex, all sodium cations have the same coordination number and are linked with two cyano groups, one ring nitrogen and a fluorine from the same anion, and two carbonyl groups of PC. Anhydrous single crystals of NaTDI were not obtained, and in its polycrystalline hydrated form, sodium cations are coordinated solely by cyano groups. Raman spectra of solid anhydrous salts NaTDI and NaPDI and their complexes with PC exhibit differences in the position of bands corresponding to most significant and substitutionsensitive vibration modes, e.g., CN cyano stretching, ring stretching, and ring deformation. In the Raman spectral region of CN triple bond, stretching vibrations of TDI− anion of two bands are observed, of CN symmetric and asymmetric stretching modes, at 2257 and 2269 cm−1, respectively. In the same region for NaTDI-PC complex also two bands are found, a weak one at 2265 cm−1 and a much stronger one with maximum at 2248 cm−1, with two distinct shoulders at 2241 and 2251 cm−1. The observed split is due to the different geometry of imidazolium anions coordinated to the same cation, resulting in different CN bond strength. The position of the strongest of ring stretching vibrations, ascribed to C2−C2* stretch, is changed only slightly, while band ascribed to C2−N1 stretching mode shifts from 1322 to 1313 cm−1, reflecting participation of this site in the complex formation. Another band that may serve as a probe of anion−cation coordination is due to NCN bending vibrations. In the spectrum of the salt, it is centered at 995 cm−1, while for the complex the band is much broader and its maximum is found at 990 cm−1 (Table 2). The shift is more pronounced in spectra of the PC based electrolytes, and the maximum is seen at 980 cm−1 , accompanied, at salt concentration higher than 0.25 M, by a shoulder at 990 cm−1. Table 2. Most significant bands form Raman spectrum for dry native salts and solid complexes with PC peak position (cm−1) sample NaTDI NaTDI-PC 1M PC-NaTDI complex NaPDI NaPDI-PC 1M PC-NaPDI complex
CN stretching 2257, 2269 2227, 2238 (shoulder)
ring stretching
NCN bending
2265, 2248 2241, 2251 (shoulders) 2258, 2262 2227, 2240 (shoulder)
1322 1313 (shoulder) 1313, 1318
995 990, 982 (shoulder) 990, 981
1325 1300
2251, 2246, 2227 (shoulder)
1317, 1309, 1303
955 950 (shoulder), 941 958, 947
4912
dx.doi.org/10.1021/cm403349t | Chem. Mater. 2014, 26, 4908−4914
Chemistry of Materials
Article
Figure 7. Ionic conductivity at room temperature for various salt concentration in PC solvent: (a) NaTDI and (b) NaPDI.
3.6. Viscosity of Electrolytes. The viscosities of both NaTDI and NaPDI systems rise dramatically as salt concentration increases over value 0.5 M (Figure 9). At concentration 0.5 M, an ion pair appears. This observation stays in agreement with analysis of cyano group band positions in Raman spectra (not shown in this article). In addition, the difference in viscosity between the systems at a given temperature increases as salt content increases. The dependence of the salt content on the viscosity is significantly larger for the NaPDI-PC system than for the NaTDI-PC system especially with increasing concentration. Noted difference decreases with increasing temperature. This temperature dependence of the viscosity is connected with the mobility of charge carriers present in the electrolyte and is widely described in the literature. It is worth mentioning that viscosity of 1 M NaTDI and NaPDI in PC is 6.3 and 6.6 mPa·s, respectively, while viscosity of the NaClO4 in PC (1 M) is very similar (about 6.8 mPa·s) at 20 °C.30 It is known from previous work that, even for viscous systems like PEGDME500, a 1 M sample with NaTDI conducts better than other sodium salts, e.g., 1 M NaBF4.31
Figure 8. Cyclic voltammograms of PC electrolytes vs Na/Na+ electrodes. Scan rate was 1 mV/s: (a) NaTDI and (b) NaPDI.
to more than 300 °C (TGA). Moreover, interesting properties of synthesized salts and their solutions in the PC solvent were proven by structural characteristics, electrochemical testing, and Raman spectroscopy. NaTDI and NaPDI based electrolytes in PC show good conductivity with the values about 4mS·cm−1 at 20 °C for salt concentration 0.5 and 1 M for both salts. Ionic conductivities collected for both salts are almost the same considering slope over temperatures and pure values. Interestingly the high values of the conductivities are reached for 0.5 M salt concentrations, which is an extreme advantage if we consider material savings. Superior properties were also confirmed by Raman spectroscopy used for the ionic association monitoring. Raman spectroscopy showed excellent salt dissociation in carbonate as PC and absence of ion pairs for concentration up to 0.5 M. A long distance framework type ordering of both TDI− and PDI− anions were also observed via extensive X-ray studies (data not presented here). According to previous computional simulations, such superfine structures are suspect to play the crucial role of sodium coordination and, by that, facilitating substantially cation transport properties. Combining this with thermal stability over 300 °C plus the elevated electrochemical stability over 4.5 V for NaTDI and 4.2
4. CONCLUSIONS Novel sodium salts, with five-membered ring imidazolium anion, were synthesized. Salts were found extremely stable up 4913
dx.doi.org/10.1021/cm403349t | Chem. Mater. 2014, 26, 4908−4914
Chemistry of Materials
Article
(trifluoromethyl)imidazolate; NaPDI, sodium 4,5-dicyano-2(pentafluoroethyl)imidazolate; NaTDI, sodium 4,5-dicyano-2(trifluoromethyl)imidazolate; PC, propylene carbonate; PEGDME500, poly(ethylene glycol) dimethyl ether Mn = 500; TGA, thermogravimetry; TLC, thin layer chromatography
■
Figure 9. Viscosity for various salt concentration in PC solvent: (a) NaTDI and (b) NaPDI.
V NaPDI vs Na/Na+ makes these salts interesting as a candidate for electrolyte’s sodium salt carrier.
■
ASSOCIATED CONTENT
S Supporting Information *
Coordination environments and atomic coordinates. This material is available free of charge via the Internet at http:// pubs.acs.org.
■
REFERENCES
(1) Linden, D.; Reddy, T.B. Handbook of Batteries; McGraw-Hill: New York, 2003. (2) Dell, R. M. Solid State Ionics 2000, 134, 139−158. (3) US Departament of the Interior, US Geological Survey, Mineral Commodity Summaries 2011. (4) Eftekhari, A. J. Power Sources 2004, 126, 221−228. (5) Yamadaa, Y.; Doi, T.; Tanakab, I.; Okada, S.; Yamaki, J. J. Power Sources 2011, 196, 4837−4841 and references therein. (6) Nishijima, M.; Gocheva, I. D.; Okada, S.; Doi, T.; Yamaki, J.; Nishida, T. J. Power Sources 2009, 190 (2), 558−562. (7) Vidal-Abarca, C.; Lavela, P.; Tirado, J. L.; Chadwick, A. V.; Alfredsson, M.; Kelder, E. J. Power Sources 2012, 197, 314−318. (8) Alloin, F.; Sanchez, J.-Y.; Armand, M. B. Electrochim. Acta 1992, 37, 1729. (9) Dominey, L. A.; Koch, V. R.; Blakley, T. J. Electrochim. Acta 1992, 37, 1551. (10) Benrabah, D.; Sanchez, J.-Y.; Armand, M. Solid State Ionics 1993, 60, 87. (11) Krause, L. J.; Lamanna, W.; Summerfield, J.; Engle, M.; Korba, G.; Loch, R.; Atanasoski, R. J. Power Sources 1997, 68, 320. (12) Aravindan, V.; Vickraman, P. Eur. Polym. J. 2007, 43, 5121. (13) Kita, F.; Kawakami, A.; Nie, J.; Sonoda, T.; Kobayashi, H. J. Power Sources 1997, 68, 307. (14) Whittingham, M. S. Chem. Rev. 2004, 104, 4271. (15) Egashira, M.; Scrosati, B.; Armand, M.; Béranger, S.; Michot, C. Electrochem. Solid-State Lett. 2003, 6, A71. (16) Capiglia, C.; Saito, Y.; Kageyama, H.; Mustarelli, P.; Iwamoto, T.; Tabuchi, T.; Tukamoto, H. J. Power Sources 1999, 81−82. (17) Valøen, L. O.; Reimers, J. N. J. Electrochem. Soc. 2005, 152, A882. (18) MacCallum, J.; Tomlin, A.; Vincent, C. Eur. Polym. J. 1986, 22, 787. (19) Nazri, G.-A.; Pistoia, G., Eds. Lithium Batteries. In Science and Technology; Kluwer Academic Publisher: Norwell, MA, 2004. (20) Niedzicki, L.; Kasprzyk, M.; Kuziak, K.; Ż ukowska, G. Z.; Armand, M.; Bukowska, M.; Marcinek, M.; Szczeciński, P.; Wieczorek, W. J. Power Sources 2009, 192, 612. (21) Niedzicki, L.; Zukowska, G. Z.; Bukowska, M.; Szczeciński, P.; Grugeon, S.; Laruelle, S.; Armand, M.; Panero, S.; Scrosati, B.; Marcinek, M.; Wieczorek, W. Electrochim. Acta 2010, 55, 1450−1454. (22) CRYSALISPRO software system; Agilent Technologies UK Ltd.: Oxford, U.K., 2012. (23) Sheldrick, G. M. SHELXS−97, program for solution of crystal structures; University of Göttingen: Göttingen, Germany, 1997. (24) Sheldrick, G. M. Acta Crystallogr., Sect. A: Found. Crystallogr. 2008, 64, 112−122. (25) Dolomanov, O. V.; Bourhis, L. J.; Gildea, R. J.; Howard, J. A. K.; Puschmann, H. J. Appl. Crystallogr. 2009, 42, 339−341. (26) Boukamp, B. A. Solid State Ionics 1986, 10, 31. (27) Scheers, J.; Niedzicki, L.; Ż ukowska, G. Z.; Johansson, P.; Wieczorek, W.; Jacobsson, P. Phys. Chem. Chem. Phys. 2011, 13, 11136−11147. (28) Scheers, J.; Johansson, P.; Szczecinski, P.; Wieczorek, W.; Armand, M.; Jacobsson, P. J. Power Sources 2010, 195, 6081. (29) Niedzicki, L.; Kasprzyka, M.; Kuziaka, K.; Ż ukowskaa, G. Z.; Marcinek, M.; Wieczorek, W.; Armand, M. J. Power Sources 2011, 196, 1386−1391. (30) Ponrouch, A.; Marchante, E.; Courty, M.; Tarascon, J.-M.; Palac, M. R. Energy Environ. Sci. 2012, 5, 8572. (31) Gumienniczuk, M. MSc thesis, Warsaw University of Technolgy, Faculty of Chemistry, 2011.
AUTHOR INFORMATION
Corresponding Author
*(A.P.-M.)
[email protected]. Funding
Authors would like to thank European Research Institute Alistore ERI for financial support. Notes
The authors declare no competing financial interest.
■ ■
ACKNOWLEDGMENTS Authors would like to gratefully acknowledge Dr. Laurence Hardwick for his English quality assistance and corrections. ABBREVIATIONS CV, cyclic voltammetry; DCTA, 5-dicyano-triazole; DSC, differential scanning calorimetry; LiPDI, lithium 4,5-dicyano2-(pentafluoroethyl)imidazolate; LiTDI, lithium 4,5-dicyano-24914
dx.doi.org/10.1021/cm403349t | Chem. Mater. 2014, 26, 4908−4914
