Introduction

Spinel-type LMO and its derivatives are promising cathode materials for the development of batteries suitable for electric vehicles due to their high operating voltage, low cost, and safe performance1. Although LMO operates at high voltage, it has its own disadvantage of structural instability and capacity fading when operated at potentials >4.3 V. The reason for capacity fading is (i) electrolyte decomposition when operated at high potential, (ii) phase change (Jahn-Teller distortion/structural instability), and (iii) dissolution of manganese (Mn) to the electrolyte through hunter’s disproportionation reaction [2Mn3+(s) →Mn4+(s) + Mn2+(l)]2,3,4. Among these, active metal dissolution (Mn) is the most serious problem for long-time applications such as electric vehicles and smart grid applications, where slow deterioration of battery performance causes maintenance problems and increases the lifetime cost.

For over a decade, several reports were published on different methods to suppress Mn dissolution and develop a highly stable LMO. Also, few reports are found that quantitatively measure dissoluted Mn ion concentration in the electrolyte5,6,7,8,9,10. To understand the metal ion dissolution, it is important to detect when, where, and how much the dissolution happens in the lithium batteries. Therefore, developing a new analytical technique sensitive to the metal ions in electrode materials of battery systems is warranted. Moreover, developing in-situ techniques will help in future efforts to improve battery design and performance.

Nuclear magnetic resonance imaging (MRI) is a non-invasive technique, which combines spectroscopy and imaging, makes it a powerful tool and is widely used in various applications such as batteries, fuel cells, corrosion cells, and metallurgy11,12,13,14,15,16,17,18,19,20. However, no reports have focused on identifying the metal ion dissolution in lithium batteries using in-situ MRI [see Supplementary Table 1 for past reports on application of MRI in battery research]. Also, our group first reported that proton (1H) MRI is a suitable technique to visualize the degradation of battery electrolytes in a model lithium battery21. In the present work, we demonstrate the direct observation and quantification of Mn ion dissolution from LMO cathode using in-situ 1H MRI. Further, we identified an electrolyte system that suppresses the metal ion dissolution from LMO.

Results and discussion

Theory and scope of MRI in lithium battery electrolyte

In principle, the MRI signal intensity, when obtained using the spin-echo technique, is given by

$$I(x,y)\propto {\rho }_{(x,y)}\left\{1-\exp \left(\frac{-{TR}}{{T}_{1}}\right)\right\}\exp \left(\frac{-{TE}}{{T}_{2}}\right)$$
(1)

where TR and TE are repetition and echo time, which are experimental parameters. T1 and T2 are spin-lattice relaxation, and spin-spin relaxation time, which are material properties. Therefore, from Eq. 1, MRI signal intensity is clearly affected when T1 and T2 relaxation time changes. By varying the experimental parameters, we can obtain contrast-enhanced images that are either T1 weighted (longitudinal relaxation), T2 weighted (transverse relaxation), or proton density weighted images. A comparative image intensity plot generated by varying the parameters used in Eq. 1 is shown in Fig. 1a. However, there is no exact reference range. In this study, we have used the change in the relaxation time of the solvent protons in the electrolyte ethylene carbonate (EC), and dimethyl carbonate (DMC) in the presence of dissoluted Mn2+, a paramagnetic nuclei to indirectly observe and quantify the Mn amount dissoluted into the electrolyte.

Fig. 1: Evolution of MRI signal intensity.
figure 1

a MRI intensity change calculated according to Eq. 1, with respect to change in relaxation time, b MR images of standard MnCl2 in 1 M LiPF6 EC:DMC gel electrolyte (i) 0 (ii) 24 and (iii) 48 μM Mn2+, c Mn2+ concentration dependence of average MRI signal intensity extracted from (b).

Full size image

Calibration of 1H MRI signal intensity with manganese (Mn2+) concentration

The changes in 1H MRI signal intensity in the presence of paramagnetic Mn2+ were validated using gel electrolyte (1 M lithium hexafluorophosphate (LiPF6) in EC:DMC + poly(vinylidene fluoride-co-hexafluoropropylene) (PVdF-HFP)) with different concentrations of MnCl2 (0, 24, 48 μM). The results are shown in Fig. 1b, c, where the image contrast increases linearly with an increase in Mn2+ concentration. This is due to the decrease in the T1 relaxation time of the protons in EC/DMC with increasing Mn concentration [relaxation time of the electrolyte is given in Supplementary Table 2]. A calibration was determined by linear least square fit, giving the following equation

$${{{\rm{Y}}}}=3.6\,({{\pm }}0.23)\,{{{\rm{x}}}}\,{10}^{6}\,{{{\rm{X}}}}+53.547\,({{\pm }}7)$$
(2)

The result indicated that we could detect manganese ions in the order of μM.

Earlier reports have shown that MRI images are strongly affected by the presence of local magnetism from the electrode materials and metals used in batteries16,22. Therefore, we have investigated the effect of current collectors and cathode material on the MR images. The schematic diagram of the homemade cell used for the in-situ 1H MRI measurement is shown in Supplementary Fig. S1a, b. The parameters and conditions were optimized so that the images were not affected/distorted by the magnetic property of the materials used, as shown in Supplementary Fig. S2a–c. Moreover, the effect of the current under the magnetic field during in-situ measurement was studied briefly, and the results are shown in Supplementary Figs. S3 and 4).

Dissolution of Mn2+ in conventional electrolyte

The in-situ 1H MRI measurement results for Li/LMO cell using 1 M LiPF6 EC:DMC electrolyte is shown in Fig. 2(a–c). The charge–discharge profile is shown in Fig. 2a, and MR images acquired at regular intervals during charging and discharging is shown in Fig. 2b. The image acquired before applying current (Fig. 2b(A)) shows an even distribution of the signal intensity. During charging between 4.1–4.7 V, signal intensity near the LMO cathode increases significantly, and beyond 4.7 V, a strong increase in the intensity is observed (Fig. 2b(G)). The gradual increase in signal intensity during charging indicates the presence of manganese in the electrolyte, which affects the T1 relaxation of the solvent protons in the electrolyte (EC/DMC), leading to the increase in signal intensity (Supplementary Table 2). During discharge, the intensity near the LMO cathode spreads through the cell.

Fig. 2: Observation of dissoluted manganese.
figure 2

a Charge–discharge profile for LiMn2O4/1 M LiPF6 EC:DMC/Li cell, b 1H MR images acquired at potentials mentioned in (a), and (c) 1H MRI signal intensity corresponding to the active cell area (the square region between the electrodes marked in the blue frame in (b).

Full size image

For qualitative analysis, the average MRI signal intensity was extracted and is shown in Fig. 2c. The intensity gradually increased from ~4.1 V where a plateau begins to appear (Fig. 2a). This increase in signal intensity is attributed to the presence of dissoluted manganese in the electrolyte from the LMO cathode. Although the experiments were performed in controlled atmosphere, trace amount of H2O in the electrolyte is unavoidable, resulting in the formation of HF. During the charging process at ~4.1 V, lithium is gradually deintercalated from the LMO cathode, and a lithium-deficient phase is formed on the surface locally, which reacts with HF to form λ-Mn2O4 phase according to the following reaction23 process

$${{{{\rm{Li}}}}}_{1-{{{\rm{x}}}}}{{{{\rm{Mn}}}}}_{2}{{{{\rm{O}}}}}_{4}+{{{\rm{HF}}}}\to {{{\rm{\lambda }}}}-{{{{\rm{Mn}}}}}_{2}{{{{\rm{O}}}}}_{4}+{{{\rm{LiF}}}}+{{{{\rm{Mn}}}}}^{2+}+{{{{\rm{H}}}}}_{2}{{{\rm{O}}}}$$
(3)

where the Mn2+ ions get dissoluted into the electrolyte. This result corroborates with Dong et al., who reported a similar process to occur above 4.1 V7.

Beyond 4.7 V, a rapid rise in the MRI signal intensity is observed. As charging proceeds, oxygen is released from the LiyMn2O4 cathode (Eq. 3), leading to the oxidation of the electrolyte24.

$${{{{\rm{Li}}}}}_{{{{\rm{y}}}}}{{{{\rm{Mn}}}}}_{2}{{{{\rm{O}}}}}_{4}+{2{{{\rm{\delta }}}}{{{\rm{e}}}}}^{-}\to {{{{\rm{Li}}}}}_{{{{\rm{y}}}}}{{{{\rm{Mn}}}}}_{2}{{{{\rm{O}}}}}_{4-{{{\rm{\delta }}}}}+{{{{\rm{\delta }}}}{{{\rm{O}}}}}^{2-}$$
(4)

When oxygen is released from LiyMn2O4 cathode, the concentration of unstable Mn3+ ions increase, leading to faster and larger disproportionation of manganese ions from the LMO electrode (2Mn3+ → Mn4+ + Mn2+)25,26,27. Further, the MRI signal intensity increases continually and extends up to 4.1 V during discharge. This is because the manganese ions dissoluted from the surface of LMO cathode tend to accumulate in the electrolyte. Moreover, due to hydrodynamic flow of the electrolyte, the intensity spreads throughout the entire cell. After discharging the cell to lower voltages, no considerable change in the signal intensity was observed (Supplementary Movie 1).

In addition, a circular flow pattern appears near the Li anode throughout the charge–discharge cycle (Fig. 2b), which is due to the electrochemical convection in the cell19,28. We skip the discussion about convection phenomena here because the present article focuses on the visualization and quantification of Mn ion dissoluted from LMO cathode. The electrochemical convection affects the observed intensity profile of the manganese distribution significantly, restricting our quantitative analysis of manganese dissolution. To avoid this problem, we used a gel electrolyte consisting of 1 M LiPF6 EC: DMC with PVdF-HFP polymer.

Quantification of Mn2+ concentration

The charge–discharge profile and MR images for the gel electrolyte system is shown in Fig. 3a, b, which shows similar behavior to that of the liquid electrolyte system. The MR images of the gel electrolyte indicate that the electrochemical convection effect observed near the lithium metal anode in the liquid electrolyte system is suppressed, which enables us to analyze the manganese dissolution into the electrolyte quantitatively. The MR images obtained at much shorter intervals are shown in Supplementary Fig. S6 and Supplementary Movie 2.

Fig. 3: Mapping of manganese concentration.
figure 3

a Charge–discharge profile for LMO cell with gel electrolyte. b MR images acquired at potential marked in red in (a) and (c) MRI signal intensity change with respect to time extracted from the active cell region (region enclosed in the blue frame in ((b)(A)) and near the LMO cathode (region enclosed in the red frame in ((b) (B)). d Mn2+ concentration mapped images acquired at potentials indicated.

Full size image

The average MRI signal intensity change with respect to charging and discharging is shown in Fig. 3c. The average intensity increases during the charging process (Fig. 3c blue circles). Like the liquid electrolyte system, above 4.1 V gradual increase in the signal intensity is observed due to the presence of dissoluted manganese8. Contrary to the liquid electrolyte system, during the discharging process, the MRI signal intensity does not decrease. Also, we extracted the image intensity near the cathode and is shown in Fig. 3c (red circles), where the MRI signal intensity increases till the completion of the charging process and remains the same during discharging. In the absence of the electroconvection effect, the dissoluted manganese remains within the cell throughout the discharge process (Fig. 3b, c). Furthermore, manganese may be dissoluted into the electrolyte during discharge where the disproportionation reaction occurs faster. However, we cannot distinguish the contribution of manganese dissoluted during the discharge process at present.

We have converted the MR images into Mn2+ concentration map using the calibration curve (Fig. 1c) and shown in Fig. 3d. These maps show distribution of dissoluted Mn2+ ions from cathode into the electrolyte, and its propagation towards the anode. The maximum Mn2+ concentration estimated from the concentration map is 36 μM. To estimate the total Mn2+ dissoluted, we acquired MR images in the horizontal plane (XY) near the cathode, center of the cell and near the anode (Supplementary Fig. S6c, d). The amount of Mn2+ near the cathode, center of cell and near the anode was 28.6, 22.1, and 6.26 μM, respectively. Further, the actual volume of the region used to estimate the manganese concentration (region enclosed in blue in Supplementary Fig. S6c) near the cathode, center of the cell and near the anode was calculated to be 2.94 × 10−5 L, 2.94 × 10−5 L and 13.5 × 10−5 L respectively. Considering the actual volume estimated, the amount of manganese estimated near the cathode, center of the cell, and near the anode was found to be 8.42 ×10−10 mol, 6.51 × 10−10 mol, and 8.45 × 10-10 mol, respectively. Therefore, the total estimated amount of manganese in the cell using MRI was found to be 2.34 × 10−9 mol. Further, the amount of manganese dissoluted into the electrolyte was estimated to be 20.6 × 10−9 mol using ICP-AES measurements. The observed difference in the concentration estimated from MRI and ICP techniques is because the manganese concentration was estimated after 3 cycles using ICP, whereas the concentration from MRI is from one cycle only. Besides, the concentration estimated from both techniques falls in the same order. This demonstrates that using MRI technique metal ion dissolution in the order of few μM could be detected efficiently.

Suppression of Mn2+ dissolution

For comparison, the same experiment was carried out on a single solvent ether-based electrolyte methyl-3-cyanopropanoate (MCP) with lithium bis(trifluoromethane)-sulfonylimide (LiTFSI) as conductive salt29. The choice of new solvent and salt based on cyanoesters and LiTFSI, respectively, is to inherent high electrochemical stability and suppress HF formation29,30. This was expected to suppress the Mn2+ dissolution during the electrochemical charge–discharge process. The charge–discharge profile for LiMn2O4/1 M LiTFSI MCP/Li cell system is shown in Fig. 4a. The charge–discharge profile appears similar to that of 1 M LiPF6 EC: DMC (Fig. 3a). The acquired 1H MR images and the average MRI signal intensity for LiMn2O4/1 M LiTFSI MCP/Li cell system are shown in Fig. 4(b, c). In the case of 1 M LiPF6 EC: DMC, image intensity gradually increased near the surface of LiMn2O4, indicating the dissolution of Mn2+ ion into the electrolyte during charging. However, in the case of 1 M LiTFSI MCP during charging, no considerable increase in image intensity was observed, indicating that Mn dissolution was suppressed in 1 M LiTFSI MCP electrolyte system (Fig. 4b, c). During discharge when the current is reversed, a slight increase in image intensity was observed, which might be due to manganese dissoluted into the electrolyte during discharge following the disproportionation of Mn3+ to Mn4+ and Mn2+ and the intensity being spread out of the cell due to hydrodynamic flow of the electrolyte. The image intensity change is small when compared to the gel electrolyte results, indicating that manganese dissolution is suppressed in the LiTFSI MCP system (Supplementary Movie 3).

Fig. 4: Suppression of manganese dissolution.
figure 4

a Charge–discharge profile for LiMn2O4/1 M LiTFSI MCP/Li cell, b 1H MR images acquired at potentials mentioned in (a), and (c) 1H MRI signal intensity change during charging and discharging.

Full size image

Conclusions

We have illustrated a simple spin-echo technique with appreciable acquisition time to identify dissoluted Mn2+ in the electrolyte by 1H MRI. Our results prove in-situ MRI to be a promising tool in visualizing when, where, and how Mn2+ dissolution occurs from LiMn2O4 electrode into the electrolyte. This approach helps in exploring the metal ion dissolution in any electrochemical systems under different electrochemical conditions, such as changing the electrolyte solution, salt, electrodes, and additives. This identification method helps to design lithium battery materials and improve their performance. In the present case, we identified an electrolyte (1 M LiTFSI MCP) that suppresses Mn2+ dissolution due to electrolyte oxidation during the charging process. Furthermore, MRI technique provides advantages over other techniques because it can estimate even extremely small concentration (μM) of dissoluted metal ions.

Methods

Electrode

The positive electrode was prepared by mixing LiMn2O4 (Aldrich) as an active material, PVdF (Wako Pure Chemical Industries, Ltd.) as a binder, and acetylene carbon black (Wako Pure Chemical Industries, Ltd.) at a mass ratio of 85:9:6 using N-Methyl-2-Pyrrolidone NMP (Kishida Kagaku) as a solvent. The prepared slurry was coated on the aluminum foil with a thickness of 50 μm. Care was taken to obtain uniform active material of thickness 75 μm. The active electrode material was dried at 80 oC and pressed at 40 MPa for 5 min, and the total thickness of the electrode was 59 μm.

Liquid electrolyte

In the present study, two types of electrolytes were investigated (1) 1 M LiPF6 EC:DMC (1:1 v/v) (Kishida chemical co.) and (2) 1 M LiTFSI MCP. The MCP sample was received from MEET Battery Research Center, Münster, Germany.

Gel electrolyte

The gel electrolyte was prepared by dissolving PVdF-HFP (KYNAR FLEX 2801-00) in 1 M LiPF6 EC:DMC (1:1) with a stoichiometry of 30 mg/L at 353 K.

In-situ MRI cell

The specially designed homemade experimental cell used for in-situ MRI measurements is shown in Supplementary Fig. S1. The cells for the in-situ MRI measurements were assembled in a controlled argon atmosphere with the dewpoint of −88 °C. The LMO cathode and the Li metal anode (0.1 mm thickness and 10 mm diameter) were separated by a hollow cylindrical spacer made of PEEK with an inner diameter and thickness of 5 mm. For safety concerns of the battery operation, a groove was introduced in the spacer, which acts as a vent for gas/bubbles to escape. For electrical contact, Pt and teflon-coated Cu wire were used on the LMO positive and Li-negative electrodes, respectively.

Electrochemical measurement

The galvanostatic charging and discharging was carried out using Toho Giken PS–08 potentiostat/galvanostat instrument at a constant current of ±25 µA in both the liquid and gel electrolyte systems.

Nuclear magnetic resonance imaging (MRI)

The 1H NMR imaging measurement was carried out using Bruker Avance–400 NMR spectrometer operating at 9.4 T. A triple-axis gradient probe with vertical sample loading is used for the 1H NMR imaging. The two-dimensional 1H MR images were acquired using spin-echo pulse sequence. In-situ MR images were acquired with a slice thickness of 1.48 mm, size 30 × 30 mm and, a spatial resolution of 234 µm. The acquisition parameters such as echo time and repetition time were fixed as 9.0 and 500 ms for the 1 M LiPF6 EC:DMC system. For the gel electrolyte system TE and TR were fixed as 9.0 and 1000 ms respectively. For 1 M LiTFSI MCP electrolyte system TE and TR were fixed as 10.4 and 500 ms, respectively. The number of integration times (NS) was four. The minimum acquisition time to obtain a single image was 4 min 16 s. The open-source ImageJ software was used for further analysis of the images. All the measurements were performed at 20 °C.

Inductively coupled plasma atomic emission spectroscopic (ICP-AES) measurements

To quantitatively determine the dissoluted manganese concentration, a cell similar to that for in-situ MRI measurement was constructed and galvanostatiscally charged and discharged under similar conditions. After completing three cycles, the electrolyte was collected and the dissoluted manganese concentration was examined using Perkin Elmer’s Optima 3300XL inductively coupled plasma (ICP) emission spectrometer.