Further technological advancement of both lithium-ion and emerging battery technologies can be catalysed by an improved understanding of the chemistry and working mechanisms of the solid electrolyte interphases (SEIs) that form at electrochemically active battery interfaces.  However, collecting and interpreting spectra of SEIs is difficult for several reasons, including the chemically diverse composition of SEIs. To address this challenge, we herein present a vibrational spectroscopy and X-ray diffraction data library of ten suggested SEI chemical constituents relevant to both lithium-ion and emerging battery chemistries. The data library includes attenuated total reflectance Fourier transform infrared spectroscopy, Raman spectroscopy, and X-ray diffraction data, collected in inert atmospheres afforded by custom designed sample holders. The data library presented in this work (and online repository) alleviates challenges with locating related work that is either diffusely spread throughout the literature, or is non-existent, and provides energy storage researchers streamlined access to vital SEI-relevant data that can catalyse future battery research efforts.

Data Collection

Prior to any characterization, all pristine compounds were stored in an argon glovebox with base oxygen and water concentrations of ~0.1 ppm and ~0.5 ppm, respectively. The sources and purities of the studied chemicals are provided in Table 1 of the associated paper.

ATR-FTIR Spectroscopy. 

ATR-FTIR spectra were collected from 370 to 4000 cm^-1 at a spectral resolution of 2 cm^-1 using a Shimadzu IRTracer-100 instrument with an IRIS single reflection diamond accessory. Herein, we generally report data in the mid-IR range, with a low-energy cutoff of about 500 cm^-1. This approach was taken because the mid-IR range is arguably the most commonly explored and accessible to experimentalists, but also because we use some data below 500 cm^-1 to aid in the generation of a baseline for subtraction (see Data Processing for further details). We encourage those particularly interested in data around and below ca. 450 cm^-1 to consult the raw data available on the online repository.¹⁷

The ATR-FTIR instrument was housed in a nitrogen-filled glovebox with an oxygen concentration below 20 ppm. After being transferred into the ATR-FTIR enclosure in sealed vials, compounds were immediately placed on a clean diamond crystal for the ATR-FTIR measurement. This transfer approach was effective at minimizing unwanted reactions, as described in detail in the Technical Validation section below. Most compounds were measured using an average of 512 individual spectra (CH₃COOLi, Li₂CO₃, ⁷LiF, ⁶LiF, Li₂O, PEO) to maximize the signal to noise ratio, while only 50 spectra were accumulated for some of the more reactive compounds (LiH, LiPF₆, MnF₂, NiF₂) to minimize acquisition time and thereby reduce the likelihood of undesired reactions.

Raman Spectroscopy.  

Raman spectra were collected using a 2 cm-square and 5 mm thick custom-made polyether ketone (PEEK) cell with an optical window (2.5 cm-square and 1 mm thick glass microscope slide). The cell, which kept samples in an inert argon environment during measurement, is illustrated in Fig. 1a. Prior to cell assembly, PEEK cell bodies and glass slides were sequentially sonicated with acetone and isopropyl alcohol and baked at 40 ^oC for at least 4 hours before being transferred into the glovebox for assembly. After a cell was assembled, it was enclosed within a heat-sealed bag before being transferred to a Renishaw Qontor Raman microscope where Raman spectroscopy was conducted. A 488 nm excitation laser was used at a power ranging from 1 to 10 mW to collect data over 25 acquisitions from 100 to 3200 cm^-1. An additional measurement of the lithium oxide sample was obtained on the same instrument using a 633 nm laser (see Table 6 of the associated paper). Unwanted contributions to the Raman spectra from the glass optical window were avoided by focusing the laser on the surface of the compounds.

X-Ray Diffraction. Samples for XRD measurements were similarly assembled in the argon glovebox. Each compound was placed on a clean 2.5 cm-square and 1 mm thick glass microscope slide (cleaned and dried using the method described above) and covered with several sealing overlayers of polyimide tape (Kapton, Ted Pella, silicone adhesive, 70 µm thick) before being heat-sealed in individual plastic bags. These sealed samples were then transferred to a Bruker Phaser D2 instrument to collect X-ray diffraction patterns over a 2θ range of 10 to 90 degrees using an acquisition time of 0.2 seconds per step and a step size of 0.02 degrees per step. All samples remained in their sealed bags until right before the start of measurement. XRD patterns were collected through the tape, rather than through glass, to prevent significant XRD contributions from the glass crystal. The amorphous background from the tape was removed via processing as described below in the Data Processing subsection. Comparison of our measurement of Li₂O to the results of Weber et al.¹⁸ provides strong evidence that this approach successfully minimized unwanted reactions (see the Technical Validation section of the associated paper).

Data Processing

The collected raw data was processed to isolate features and patterns of the spectra that can be used to identify interphase constituents. Unwanted instrumental and/or background contributions were removed from all raw data. Furthermore, all FTIR measurements of inorganic compounds – and some organic ones – contained strong and broad absorption features, below 600 cm^-1, with increasing intensity as wavenumber decreased. These features were broad enough that the decreasing intensity side of the feature (e.g. see Palik and Hunter’s data on LiF¹⁹) could not be observed due to the instrument’s low-energy detection limit of 370 cm^-1. Because most researchers have detectors with similar limits (or even higher in energy), we processed our FTIR data to focus on the mid-IR regions that are most commonly accessible, nominally above ca. 500 cm^-1. As a result, when defining baseline profiles across the entire bandwidth to be subtracted from the raw data, we fit these low-wavenumber features (below our reporting window) as part of the baseline, even though they are technically not part of the background. Raw, unsubtracted data spanning the entire collected bandwidth is available in the online repository.¹⁷ A representative baseline fitting of the raw data (which will result in the removal of the downward-sloping part of the spectral feature that is observed below 600 cm^-1) and subtracted data are provided in Figure 2a and 2b, respectively. Spikes in the Raman spectra attributable to cosmic ray excitation were removed and are not included in the raw data. Raman spectra were subsequently processed through the subtraction of Gaussian and/or polynomial fits to eliminate background contributions arising from a number of phenomena (e.g. fluorescence, glass effects, surface roughness). Our instrument generated raw Raman data with unevenly spaced wavenumber values, so an interpolation was performed to transform the data. Gaussian fits to determine peak positions from the data before and after interpolation confirmed that this transformation did not affect the location or shape of spectral features. Gaussian fits were used to subtract out an amorphous background in XRD measurements generated by the Kapton tape. This background between 10 and 30 degrees appeared in all measurements through Kapton tape (but not in control measurements of bare metal foils) and was at lower 2θ than the first diffraction peak of most compounds, allowing for a consistent background subtraction for measurements of the few compounds (like lithium acetate) where the first diffraction peak was found below 30 degrees. For all data types, Fourier filtering was applied to reduce high frequency noise (with care taken to avoid distorting spectral features) and small vertical offsets were used in some cases to align the high-2θ baseline near zero. All data was normalized to take on values from 0 to 1.

The data files are in a .xlsx format which can be opened using Excel (including the free Excel Viewer) and Google Sheets among other applications. These files can then be exported as a .csv or in other format as desired for use in other software (including data processing and plotting programs).