Article pubs.acs.org/JPCC
Solvate Structures and Computational/Spectroscopic Characterization of LiBF4 Electrolytes Daniel M. Seo,† Paul D. Boyle,‡ Joshua L. Allen,† Sang-Don Han,† Erlendur Jónsson,§ Patrik Johansson,§ and Wesley A. Henderson*,†,∥ †
Ionic Liquids & Electrolytes for Energy Technologies (ILEET) Laboratory, Department of Chemical & Biomolecular Engineering, North Carolina State University, Raleigh, North Carolina 27695, United States ‡ X-ray Structural Facility, Department of Chemistry, North Carolina State University, Raleigh, North Carolina 27695, United States § Department of Applied Physics, Chalmers University of Technology, SE-412 96, Gothenburg, Sweden ∥ Electrochemical Materials & Systems Group, Energy & Environment Directorate, Pacific Northwest National Laboratory (PNNL), Richland, Washington 99352, United States S Supporting Information *
ABSTRACT: Crystal structures have been determined for both LiBF4 and HBF4 solvates: (acetonitrile)2:LiBF4, (ethylene glycol diethyl ether)1:LiBF4, (diethylene glycol diethyl ether)1:LiBF4, (tetrahydrofuran)1:LiBF4, (methyl methoxyacetate)1:LiBF4, (succinonitrile)1:LiBF4, (N,N,N′,N″,N″-pentamethyldiethylenetriamine)1:HBF4, (N,N,N′,N′-tetramethylethylenediamine)3/2:HBF4, and (phenanthroline)2:HBF4. These, as well as other known LiBF4 solvate structures, have been characterized by Raman vibrational spectroscopy to unambiguously assign the anion Raman band positions to specific forms of BF4−···Li+ cation coordination. In addition, complementary DFT calculations of BF4−···Li+ cation complexes have provided additional insight into the challenges associated with accurately interpreting the anion interactions from experimental Raman spectra. This information provides a crucial tool for the characterization of the ionic association interactions within electrolytes.
Chart 1. Examples of BF4−···Li+ Cation Coordination: (a) SSIP, (b) CIP-I, (c) CIP-II, (d) AGG-I, (e) AGG-II, and (f) AGG-III
1. INTRODUCTION Understanding the structure of concentrated liquid electrolytes remains a key challenge for linking structure with electrolyte properties. Vibrational spectroscopic analyses (Raman, FTIR) of electrolyte mixtures provide crucial insight regarding the molecular-level interactions between the anions and cations (i.e., ionic association),1−11 but the interpretation of the results is often challenging. Typically, the solvates which form are classified as either solvent-separated ion pairs (SSIPs), contact ion pairs (CIPs), or aggregates (AGGs) if the anions are coordinated to zero, one, or more than one Li+ cations, respectively (Chart 1). Which solvate species form and their distribution/population are key determinants for the wide variability in electrolyte properties. A knowledge of this is therefore necessary to provide mechanistic explanations for properties such as ionic conductivity, viscosity, oxidative/ reductive stability, volatility, etc.a key challenge for many electrochemical applications. When an anion coordinates one or more Li+ cations, the anion bond lengths/angles change as a result of variations in the electron density distribution. This results in the shifting of the vibrational bands for the anion. These shifts can be correlated with specific forms of anion···Li+ cation coordination. Information regarding the correct assignment of the anion bands to these specific forms of coordination, however, is generally unavailable. Assignments are often made based upon © 2014 American Chemical Society
computational analyses (DFT calculations); however, such calculations rarely directly match experimental results, and there is limited information available to validate such calculations. Thus, significant guesswork remains for the Raman band assignments to determine the ionic association Received: May 12, 2014 Revised: July 7, 2014 Published: July 21, 2014 18377
dx.doi.org/10.1021/jp5046782 | J. Phys. Chem. C 2014, 118, 18377−18386
The Journal of Physical Chemistry C
Article
interactions in solvent−lithium salt mixtures. If the band assignments are misconstrued, then highly misleading conclusions can result from the analysis of electrolytes. Although LiPF6 remains the dominant salt for current stateof-the-art lithium battery electrolytes,12 this salt has limitations which hinder its use at both high temperature (i.e., 60 °C) due to salt degradation and low temperature (i.e., −20 °C) due to increased electrolyte viscosity and solvate crystallization. Under similar extreme conditions, LiBF4-based electrolytes have improved performance at both high and low temperature relative to those with LiPF6.13−19 Further, LiBF4 is a highly useful salt for probing electrolyte interactions as the anion can be classified as intermediate in its association tendency with Li+ cations when the salt is dissolved in aprotic solvents.1 To fully utilize this salt and understand its molecular-level interactions, however, it must be possible to deconvolute the complex Raman spectra obtained for liquid electrolyte mixtures which consist of overlapping Raman bands due to a wide distribution of solvate species.1 In the present study, therefore, the Raman band positions for various forms of BF4−···Li+ cation coordination have been unambiguously assigned by correlating the Raman bands with solvate crystal structures. The latter consist of both known LiBF4 solvate structures previously reported, many quite recently, or new solvate structures with LiBF4 and HBF4 determined as part of the present study and reported here. These solvate structures, in and of themselves, provide detailed insight into the solvation and ionic association interactions which govern solvate formation. The results suggest that the conclusions drawn about BF4− anion interactions from previously reported LiBF4 electrolyte studies may be either incorrect or oversimplifications of the actual complex equilibrium of the solvate species present and thus may require reevaluation.1,20−30
Chart 2. Structures and Acronyms of the Solvents Used
Sample Preparation. Mixtures were prepared in the glovebox by combining appropriate amounts of LiBF4 and the desired solvents in hermetically sealed glass vials. The mixtures were heated and stirred on a hot plate until the salt fully dissolved. To obtain single crystals, mixtures more dilute than the desired solvate stoichiometry were typically prepared. Single crystals grew from the mixtures upon standing either at room temperature, at 4 °C (in a refrigerator), or at −23 °C (in a freezer) (Supporting Information). The single crystals were used both for the determination of the solvate structures via single-crystal X-ray diffraction and for the Raman analysis. For the latter, the crystals were ground in the glovebox into a fine powder using a mortar and pestle (if the Tm of the solvate is greater than room temperature). For the solvates which have a Tm at or below room temperature, mixtures were instead prepared with the stoichiometric amount of solvent and salt for the solvates, and the samples were crystallized directly in the Linkam heating/cooling stage just prior to the Raman measurements (note that this sometimes produced large crystals for which orientational effects impact the Raman peak intensities). Elemental Analysis. Elemental analysis of crystals for the (12C4)2:LiBF4 solvate was performed by Atlantic Microlab, Inc.C18H32O12F2BLi for (12C4)2:LiBF4, calcd: C 43.07, F 17.03, H 7.23; found: C 44.08, F 16.81, H 7.51confirming this as the correct solvate composition. Solvate Crystal Structure Determination. Samples were typically mounted on nylon loops with a small amount of Paratone N oil. X-ray measurements were made on a BrukerNonius Kappa Axis X8 Apex2 diffractometer. The unit cell dimensions were determined from symmetry constrained fits of the reflections. The frame integrations were performed using SAINT.31 The resulting raw data were scaled and absorption corrected using a multiscan averaging of symmetry equivalent data using SADABS.31 The structures were solved by direct
2. EXPERIMENTAL AND COMPUTATIONAL METHODS Materials. Anhydrous LiBF4 (Sigma-Aldrich, 99.998%, trace metal basis and Novolyte, electrolyte-grade) was used asreceived. Chart 2 shows the solvents used in this study and their acronyms: ethylene glycol dimethyl ether or 1,2dimethoxyethane (monoglyme or G1, anhydrous, 99.5%), diethylene glycol dimethyl ether or 2-methoxyethyl ether (diglyme or G2, anhydrous, 99.5%), triethylene glycol dimethyl ether (triglyme or G3, 99%), 1,2-dimethoxypropane (DMP, ≥99%), ethylene glycol diethyl ether or 1,2-diethoxyethane (EtG1, 98%), diethylene glycol diethyl ether (Et-G2, reagent grade, ≥98%), methyl acetate (MA, ≥98%), ethyl acetate (EA, anhydrous, 99.8%), tetrahydrofuran (THF, anhydrous, ≥99.9%), methyl methoxyacetate (MMA, 99%), ethylene carbonate (EC, 99%), γ-butyrolactone (GBL, ≥99%), acetonitrile (AN, anhydrous, 99.9%), succinonitrile (SN, 99%), 1,4,7,10-tetraoxacyclododecane or 12-crown-4 (12C4, 98%), N,N,N′,N″,N″-pentamethyldiethylenetriamine (PMDETA, 99%), N,N,N′,N′-tetramethylethylenediamine (TMEDA, purified by redistillation, ≥99.5%), phenanthroline (PHEN, ≥99%), and methanol (anhydrous, 99.8%). The solvents were purchased from either Fisher Scientific or Sigma-Aldrich. These solvents were dried over 3 Å molecular sieves, and the water content of the solvents was verified to be negligible (
