Revealing lithium distribution at the microscale with Cipher

Introduction

Lithium-containing compounds and alloys are critical to many key technologies of the twenty-first century, from Li-ion batteries used to power mobile electronic devices and cars to lightweight structural alloys. Progress in these fields has been remarkable, given the lack of a method to determine lithium content at the microscale. Commonly, energy dispersive X-ray spectroscopy (EDS) in the scanning electron microscope (SEM) is employed for microanalysis. However, this has not been possible for elements with atomic number (Z) < 4 as the characteristic X-rays emitted (e.g., Li K at ~55 eV) are easily attenuated by the sample or the presence of an oxide layer or contamination and require the use of highly specialized detectors. Even so, a limit of detection of ~20 wt. % Li and the inability to perform quantitative measurements due to the dependence of the fluorescence yield on the Li bonding state present significant issues [1]. More recently, it has been suggested that elemental Li maps captured with these detectors are unreliable [2].

However, quantification of Li in the SEM was demonstrated recently by researchers at LKR/AIT using a composition by difference method using EDS and quantitative backscattered electron imaging (qBEI) [3, 4]. EDS analysis was used to quantify elements Z = 4 – 94, while qBEI was used to determine the mean atomic mass (the qBEI signal being a function of atomic number for Z = 1 – 94). The fraction of light elements (Z = 1 – 3) was calculated and assumed to be Li, given the MgLi alloy analyzed. Using this lithium by composition by difference method (Li-CDM), detection of < 5 wt. % Li was demonstrated with acceptable accuracy (~1 wt. %).

Gatan and EDAX announced the Cipher^™ System that integrates the EDAX Octane Elite or Elect Super EDS Detectors with the Gatan OnPoint^™ backscatter electron detector and a composition-by-difference module for DigitalMicrograph^® software to quantitatively measure the lithium composition of a sample. In this article, we assess the accuracy of Li-CDM in a range of non-metallic materials and describe the latest results using Cipher to analyze the lithium content in a stoichiometric lithium aluminate and a lithium nickel manganese cobalt oxide powder commonly employed in cathode materials of Li-ion batteries.

Materials and methods

Quantitative backscattered electron measurements were recorded from 55 samples (Micro-Analysis Consultants Ltd) using Cipher. The samples included elemental, mineral, semiconductor, and alloy materials and ranged in atomic number from 4 – 83. The samples were mechanically polished and coated with a 2.0 nm thick carbon layer to avoid charging in the SEM using a PECS™ II system.

Quantitative lithium analysis using Cipher was applied to two samples that are available commercially—a high purity (>99.99 %) LiAlO₂ (100) crystal substrate and a powder form of Lithium Nickel Manganese Cobalt Oxide (NMC) with nominal Ni:Mn:Co ratio 8:1:1 and 5.7 wt. % Li (approximately 25 at. %).

The samples were prepared by broad beam argon milling using a Gatan Ilion^® II or PECS^™ II polisher and coated with a 2.0 nm thick carbon layer to avoid charging. Before sample preparation, the NMC 811 powder sample was embedded in epoxy to form a solid block. A field emission SEM was used to collect EDS and qBEI data at 20 and 25 kV, respectively, selected to ensure that all X-ray lines were excited efficiently while also providing comparable sampling volumes of the signals. Electron backscatter diffraction (EBSD) using the Clarity^™ detector was also performed on a commercial NMC sample to reveal the crystal structure.

 

Results and discussion

Assessing the applicability of Li-CDM to compound materials

The qBSE signal as a function of mean atomic number, $\bar{Z}$, is plotted in Figure 2 with $\bar{Z}$ calculated using the modified electron approach of equation (1) (after [5]):

$\bar{Z} = \frac{\sum_{i=1}^n a_i Z_i^{1+x}}{\sum_{i=1}^n a_i Z_i^x}$  ... (1)
Equation 1.

 

where a_i represents the atomic fraction of element i and x ≃ 0.7. In line with other publications (e.g., [6]), the qBSE signal was fitted to the function:

$f(\overline{Z}) = C(1 - e^{-\overline{Z}/q})$ ... (2)
Equation 2.

 

where C and q are constants related to the SEM and detector settings.

For compounds with $\bar{Z}$ < 40, an excellent fit of the experimental data to the exponential function of equation 2 was observed with few—if any—outliers. However, for materials of $\bar{Z}$ > 40, although the experimental data continues to follow the general trendline, the increased scatter of the experimental data indicates that a large uncertainty would be expected in the Li-CDM calculation. Notwithstanding, it was confirmed that the Li-CDM is suitable for a wide range of metallic and non-metallic samples of $\bar{Z}$ < 40.

 

Evaluating the lithium content of lithium aluminate

Cipher was used to determine the lithium content in a lithium aluminate sample. The compositional analysis results are summarized in Table 1. The lithium content was determined to be 22.6 ± 3.5 at. % (9.5 ± 1.7 wt. %); within 2.4 at. % and only 0.9 wt. % of the nominal composition 25.0 at. % (10.5 wt. %).

	Li	Al	O
By stoichiometry
At. %	25.0	25.0	50.0
By EDS
At.%	-	29.6	70.4
Std. dev.	-	1.6	5.1
Li composition-by-difference
At. %	22.6	22.9	54.6
Std. dev.	3.5	1.0	4.0

Table 1. Elemental quantification results of LiAlO₂ sample.

Evaluating the lithium content of Li-ion battery cathode materials

The NMC 811 powder analyzed consisted of approximately spherical 'secondary' particles of 5 – 30 µm in diameter. These secondary particles are agglomerates of smaller 'primary' particles that can be observed in a crystal orientation map collected by electron backscatter diffraction (EBSD) (Figure 3). The orientation map captured by the EDAX Clarity EBSD Detector reveals primary particles 200 – 2,000 nm in size with random crystal orientations and low/no gaps between primary particles.

 

Quantitative EDS analysis at select locations within NMC particles was performed and revealed O, Ni, Mn, and Co with little-to-no variation within or between particles that were analyzed (Table 2). No other elements were found to be present above the minimum detection level. The Ni:Mn:Co ratio of 8.07:1.00:1.01 was determined experimentally and was found to be consistent with the nominal 8:1:1 ratio.

Expectation per nominal (at. %)
Li	O	Ni	Mn	Co		Ni	Mn	Co
-	66.7	26.7	3.3	3.3		8.00	1.00	1.00
Experimental (at. %)
	O	Ni	Mn	Co		Ni	Mn	Co
Spot 1	73.3	21.3	2.7	2.7		7.89	1.00	1.00
Spot 2	74.4	20.3	2.6	2.6		7.81	1.00	1.00
Spot 3	72.8	21.8	2.7	2.7		8.07	1.00	1.00
Area 1	73.0	21.4	2.7	2.9		7.93	1.00	1.07
Area 2	71.7	22.9	2.7	2.7		8.48	1.00	1.00
Area 3	73.2	21.5	2.6	2.6		8.27	1.00	1.00
Mean						8.07	1.00	1.01

Table 2. Quantitative EDS analysis of six NMC particles. Analysis positions as shown in Figure 4.

The lithium content from six different NMC particles was determined using Cipher; analysis locations are shown in Figure 4. The mean lithium concentration was determined to be 22.5 at. % (5.7 wt. %) and within ~1.5 wt. % of the nominal composition value of 7.3 ± 0.3 wt. %.

This is a significant step forward in the analysis of battery materials as, for the first time, the charge state of a cathode material was determined in a conventional SEM. Here the ~25 at. % Li corresponds to the uncharged battery state.

Summary

The lithium by composition by difference method was demonstrated in stoichiometric compound samples. The lithium content of a high-purity lithium aluminate crystal substrate was determined to be 9.5 wt. % within ~ 1 wt. % of the stoichiometric value, an accuracy similar to previous reports for metallic samples [2].

For the first time, the lithium content of an NMC cathode material was determined quantitatively in a conventional scanning electron microscope. A mean lithium content of 22.5 at. % was measured experimentally, corresponding to the uncharged battery state in this material. These results validate Li-CDM for a wider range of materials, opening exciting characterization possibilities in lithiated battery materials using Cipher.

References

1. P. Hovington et al., Scanning 38 (2016) p. 571–578
2. R. Gauvin and N. Brodusch, Microsc. Microanal. 28 (Suppl 1), 2022. doi:10.1017/S1431927622002847
3. J.A. Österreicher et al., Scripta Materialia 194 (2021) 113664
4. Austrian Patent Application A 50783/2020
5. J. Donovan et al., (2003) Microsc. Microanal. 9, p. 202. DOI: 10.1017/S143192760030137
6. A. Garitagoitia Cid et al., Ultramicroscopy 195 (2018) 47
7. D. Kaczmarek, Scanning Microscopy 12 (1) (1998) p. 161-169