Marcel Diehla,
Marco Evertza,
Martin Winterab and
Sascha Nowak*a
aUniversity of Münster, MEET Battery Research Center, Institute of Physical Chemistry, Corrensstraße 46, 48149 Münster, Germany. E-mail: Sascha.Nowak@uni-muenster.de
bHelmholtz-Institute Münster, IEK 12, Forschungszentrum Jülich GmbH, Corrensstraße 46, 48149 Münster, Germany
First published on 16th April 2019
Lithium ion batteries (LIBs) are the energy storage technology of choice in the context of renewable energies and electro-mobility. It is imperative to get a thorough understanding of the aging mechanisms to achieve a prolonged cycle and calendar life. One major drawback of the technology is continuous capacity fading during operation, which is partly attributed to the loss of active lithium, the object of this work's analysis. While lithium ion battery aging is an intensively researched topic, there is still the need to determine the origin of the lost lithium and the lithium migration into the different cell components over time. To achieve this goal, different plasma-based mass spectrometric techniques in combination with isotope analysis are applied to obtain bulk as well as depth-resolved information about lithium ion movement and distribution of lithium in aged LIB cells. Different aging experiments are performed on NCM622/graphite cells with a 6Li-enriched electrolyte with subsequent Li analysis of the cell components. The results show that the charging rate, as well as the cycle number, has an impact on the 6Li/7Li-abundances and that the overall abundances show a rapid mixing of the isotopic species already after the first charge/discharge cycle for all cell components.
On the cathode side, another interphase forms during cycling, the cathode/electrolyte interphase (CEI).23–25 However, the CEI is more than one order of magnitude thinner than the SEI14,26 and is much less understood, but it still attributes to the capacity fading and loss of active lithium.{Börner, 2016 #378} In addition to that, there are more reported aging effects that attribute to the fading capacity of the positive electrodes including an increasing impedance, irreversible phase changes of the layered material and the release of oxygen at elevated cut-off potentials.14,23,27–29
Although there is a sophisticated repertoire of electrochemical methods for monitoring, the need for additional analytical methods is indispensable.30–34 For direct solid analysis, methods like glow discharge-mass spectrometry (GD-MS), laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS), time of flight-secondary ion mass spectrometry (TOF-SIMS) and X-ray photoelectron spectroscopy (XPS).35–42 Due to their capabilities of directly measuring the surface components, they represent promising approaches for analyzing new materials as well as the degradation of aged cells. Especially the GD-MS benefits from unique traits such as a fast analysis time while maintaining a nanometric sputter-rate and the ability to perform an isotope analysis throughout the profile of an electrode. Moreover, it is possible to detect a wide range of concentrations simultaneously, as well as a semi-quantitative evaluation without the need for standardized matrices.43–45
Nonetheless, the loss of lithium was already reported extensively in the literature, while the origin is still not fully clarified. Here, the use of isotope-labeled materials would help to better understand these losses and tailor future LIB for their needed desires.14,16,29,46
In this work, the lithium migration through the cell and the changes during cycling by means of different plasma-based mass spectrometric techniques in combination with isotope analysis are explored by using a 6Li-enriched electrolyte. The cell system is based on a NCM622/MCMB(HC) Li ion full cell operated at nominal voltages of 4.2 V using a EMC/EC (1:1 by weight) based carbonate mixture as electrolyte solvent. This method in combination with the mass spectrometric analytical techniques opens up manifold possibilities because the two main isotopes of lithium – 6Li and 7Li – have different abundances depending on the component and changes of these abundances can be measured after cycling. On the one hand, bulk information are obtained for the anode, cathode, and electrolyte using ICP-MS. On the other hand, depth profile analysis is applied to gather sophisticated information about the depth-resolved isotopic abundances of lithium to get a better understanding of the decomposition mechanisms.
The constant current cycling investigations were performed using a Maccor battery test system (Maccor series 4000, Maccor Inc.) and varying charging rates (C-rates) of C/5 (32 mA g−1) and C/10 (16 mA g−1) (charges per hour) for the formation steps and C/1 (160 mA g−1) for the following cycles. Combining the formation and the subsequent cycles, procedures between 1 and 13 cycles were applied. Above that, storage tests for two weeks at room temperature were performed between the lower and upper cut-off voltage of 3.0 V and 4.3 V.
For the ICP-MS measurements of the aged electrodes, the samples were dissolved. This was done by microwave-assisted acid digestion using a Multiwave Pro (Anton Paar GmbH). Polytetrafluoroethylene (PTFE) liner were filled each with 3 mL nitric acid (65 vol%, Suprapur®, Merck KGaA) and hydrochloric acid (32 vol%, Suprapur®, Merck KGaA) followed by a 7 min power ramp to 1400 W, a hold period of 42 min and a cool down step until 55 °C was reached. Afterward, the samples were diluted using deionized water (2 μg L−1 TOC, 25 MΩ, MilliQ, Merck KGaA).
y% = x × 95% + (1 − x) × 7.59% | (1) |
(2) |
In these equations, y% represents the measured 6Li-abundance of the component, x stands for the electrolytic and (1 − x) for the cathodic lithium fractions. The percentages 95% and 7.59% in the equations are the enriched and natural abundance of 6Li, respectively.
Fig. 1 Determined isotope abundances for the different pristine materials including the two used electrolytes, a cathode, (N = 3). |
Component | 6Li-abundance | 7Li-abundance |
---|---|---|
Electrolyte (ref.) | 5.79% ± 0.05% | 94.21% ± 0.71% |
Cathode | 8.40% ± 0.17% | 91.60% ± 2.33% |
Enriched electrolyte | 93.25% ± 2.12% | 6.75% ± 0.33% |
As depicted in Fig. 1 and Table 1 the abundances are matching well with the theoretical values, however, the deviations, especially for the cathode and the enriched electrolyte part, are significant. However, due to the differences between the natural and enriched abundances, small changes in the isotope ratios can still be resolved and qualitative analyses can be performed with these techniques.
Component | Total amount of 6Li/7Li/mg | Storage (2 weeks)/% | 1 Form. step/% | Formation C/5 + 10 cycles/% | Formation C/10 + 10 cycles/% |
---|---|---|---|---|---|
Anode | <LOD | <LOD | 43.67 ± 1.04 | 45.02 ± 1.91 | 36.13 ± 1.60 |
Cathode | 0.079/0.866 | 6.77 ± 0.30 | 31.04 ± 0.85 | 37.31 ± 2.01 | 35.61 ± 0.35 |
Electrolyte | 0.567/0.041 | 91.09 ± 0.55 | 49.56 ± 0.64 | 43.42 ± 2.30 | 50.20 ± 1.57 |
The results show that the storage tests had only a minor influence on the measured isotope ratios. There was no lithium detectable on the digested anodes which could be explained due to the applied washing and drying processes removing most of the electrolyte in the wetted pores, as well as, no potential was applied to the tested cells that could cause a decomposition of the electrolyte followed by the formation of an SEI and CEI. Additionally, small changes of the 6Li-abundances were detected for the electrolyte and the cathode in comparison to the benchmark measurements.
However, this trend changes after the first charge/discharge cycle, where the abundances of the bulk materials showed a significant alteration compared to the benchmark. In detail, the 6Li abundance of the cathode rose to 31.04% from 8.40% and decreased from 93.25% to 49.56% for the electrolyte. Moreover, the first measurable 6Li-abundance on the anode started at 43.67%. Considering all these data, there is a strong tendency towards the assumption that the electrolyte plays an active role in the intercalation and de-intercalation processes in lithium ion batteries because these measurements showed that lithium from the electrolyte gets incorporated into the structure of the cathode and vice versa yet after the first formation step.
Furthermore, after ten cycles using a rate of C/5, the 6Li-abundances did not show significant changes compared to cells cycled for one formation step. Especially for the anode, the differences were only 1.35%. This implied that the formed SEI during the first cycle is effective, hence, not much additional lithium from cathode or electrolyte is lost during subsequent cycling. The 6Li-abundance on the cathode rose to 37.31% and decreased to 43.42% in the electrolyte, which could be explained by the increased probability of 6Li-ions intercalation into the lattice structure due to prolonged cycling (intercalation and de-intercalation).
By using a lower C-rate during the formation, the 6Li-abundance lowers especially on the anode to 36.13% and increased to 50.20% for the electrolyte, explained by longer diffusion times – 10 hours compared to 5 hours – of lithium from the cathode towards the anode, hence, leading to the formation of the anode SEI that contains higher lithium fractions.
Fig. 2 shows similar trends as already discussed for Table 2 regarding the bulk isotope abundances. However, this technique has the additional benefit of revealing the distribution throughout the profile of the electrode and in this case an agglomeration in the first minutes of the measurement of 6Li was detected to a certain degree for all cells using the enriched electrolyte.
In detail, the overall abundance of 6Li is the highest after the first formation step and atop on that, the 6Li-abundance started at 57.45% and has a plateau-value of 43.88%. This is in good correlation with the findings from ICP-MS bulk measurements and the difference of the ratio is less than 0.2% compared to the bulk material. Using the same C-rate for the formation but adding ten more charge/discharge cycles leads to an average 6Li-abundance of 41.11% during the plateau from 6 minutes onward. Comparing this decrease of the abundances with the increase in Table 2, shows a difference between the two measuring methods, but may be explained to the fact that different cells had to be used for the measurements and the deviations could be caused due to the cycling of the cells.
On top of that, the differences for the C-rate C/5 and C/10 are in a comparable extent, with the cell formed with C/10 having a lower overall 6Li-abundance of 35.03% at the plateau, leading to comparable findings as described in Section 3.2 that the slower C-rate during the formation promoted the incorporation of 7Li (cathodic) into the SEI of the discharged anodes.
In Fig. 2, a general trend is apparent in the first 5 minutes of each measurement in the form of an agglomeration of lithium on the surface of the anodes which may be due to lithium from the electrolyte that deposits directly after applying a voltage because of the direct contact with the negative electrode. While the data in Fig. 2 presented only qualitative considerations, it is also of utmost importance to compare the lithium signals semi-quantitatively. For this, the concentrations of the 6Li and 7Li isotopes were determined using the ion beam ratios (IBRs) with C, Li, and Cu as matrix elements and compared in Fig. 3 for three different cells (a) cycled for one formation step with a C-rate of C/5, (b) cycled for three formation steps at C/5 and ten subsequent cycles at C/1, and (c) cycled with the reference electrolyte for one formation step with a C-rate of C/5.
Fig. 3(c) shows a low bulk concentration of 0.53 mg g−1 for the total amount of lithium. The measurements were stopped before a complete depth profile through the active material was achieved. The reasoning is the constitution of the sample itself because the porous nature of the anodes leads to the detachment of parts of the carbonaceous material, revealing spots of the copper current collector which changes the sputter-rates and measured intensities and leads consequently to wrong results. Therefore, the measurements were stopped after 30 minutes as the intensities were almost constant.
The comparison of the two cells in Fig. 3(a) and (b) reveal on the one hand that the total lithium content in the bulk material is 1.39 mg g−1 for (a) and 1.41 mg g−1 for (b) if the averages of the combined contents of the 6Li and 7Li isotopes between 6 and 30 minutes are added together. This means that almost the same amount of lithium is lost after one formation step, as well as after the longer cycling procedure. However, during the first 5 minutes of sputtering-equals a depth of 2–3 μm – the 6Li signals are almost identical regarding the peak-, as well as plateau-values. But, the amount of 7Li on the surface of the anode, increased dramatically during the longer cycling procedure. While the highest measured concentration after one formation step is 1.53 wt%, it increased to 2.31 wt% after ten cycles, which equals an increase of 50.98% of the total lithium amount. Subsequently, this led to the assumption that lithium is accumulated predominantly on the surface of the anodes during prolonged charge/discharge cycling.
As concluded in Fig. 4, both methods show the same trends for anode and cathode, while the deviation for the cathode is significantly higher.
Fig. 4 Trends showed after cycling for the discharged anode, cathode, and electrolyte for the ICP-MS, as well as the GD-SF-MS measurements. |
From Fig. 4(a) a trend is apparent that the 6Li-abundance is slightly declining after the first formation step. The cathodes in Fig. 4(b) shows an opposite trend with an increase in the abundance, but deviations between the two analytical methods are over 10% of the lithium abundance for the first formation step, while the abundances for the pristine electrodes have a good fit towards each other.
Anode | Cathode | Electrolyte | ||||
---|---|---|---|---|---|---|
Electrolytic | Cathodic | Electrolytic | Cathodic | Electrolytic | Cathodic | |
Storage | <LOD | <LOD | 0.00 | 1.00 | 0.96 | 0.04 |
1 form | 0.41 | 0.59 | 0.27 | 0.73 | 0.48 | 0.52 |
10 Cy. C/5 | 0.43 | 0.57 | 0.34 | 0.66 | 0.41 | 0.59 |
10 Cy. C/10 | 0.33 | 0.67 | 0.32 | 0.68 | 0.49 | 0.51 |
Fig. 5 Lithium abundances from Fig. 2(a) and (b) transformed into the electrolytic and cathodic fractions for the different components and aging conditions. |
In Table 3 two tendencies can be observed. First, all anodes show the trend that more than half of the deposited lithium originated from the cathode and the cells cycled at C/5 during the formation have almost the same electrolytic lithium fractions. A significantly lower fraction was achieved using C/10 during formation. Thus, the origin of the decomposed lithium is dependent on the charging rate, precisely, slower rates favor the deposition of cathodic lithium on the anode. Moreover, roughly two-thirds of the total lithium originated from the cathode for the cells cycled with C/10 during formation, assuring the presumption that a slower C-rate leads to a higher cathodic lithium fraction. The second trend was noticed for the cathode. In detail, a correlation between the electrolytic lithium fraction and the cycle number is shown in the data, leading to a higher fraction of Li after longer cycling procedures independent of the applied C-rates. For the electrolyte, no clear trend could be detected so far, which could be explained due to the surplus of electrolyte in coin cells compared to other cell setups such as cylindrical 18650 cells.
By converting the concentration-dependent depth profiles in the same way as it was done for Table 3, the differences between the formation process in Fig. 5(a) and the longer cycling procedure in Fig. 5(b) gets even more apparent. While Fig. 5(b) shows an almost constant distribution of the electrolytic and cathodic lithium through the whole sputtered profile, a strong accumulation of electrolytic lithium on the surface of the measured anode in Fig. 5(a) was detected. Furthermore, this agrees with the previously discussed results from Fig. 3 where an agglomeration of 7Li on prolonged cycled anodes was detected.
The transformed results from the ICP-MS and GD-SF-MS measurements elucidate the importance of the applied isotope analysis with regard to migration and decomposition processes in LIBs, revealing that the electrolytic lithium plays a most active part in the migration. In detail, a fast mixing of cathodic and electrolytic lithium is happening during the first charge/discharge cycle. Moreover, it is also shown that a higher amount of lithium from the electrolyte is decomposed on the surface of the negative electrode, which could not be determined without the use of isotopic-labeled components.
Since the mixing of the two different isotopic species happened to the biggest extent already after the first charge/discharge cycle and only slight changes could be observed after 13 cycles, further cycling experiments with longer cycling procedures were neglected in this work. Still, an influence of the different charge/discharge conditions could be determined with more cathodic lithium being incorporated into the SEI of the anode when a slower charging rate was applied during the formation which increased the time period in which the isotopic species could be mixed.
Moreover, by determining the fractions between lithium from the electrolyte and the cathode, so-called electrolytic and cathodic lithium fractions, it could be shown that the trends and tendencies are different for the three measured cell components. For the anodes, the fractions are particularly influenced by the charging rate during the formation because they show only minor differences if the same charging rate is applied but change between 13.6% and 17.5% for the cathodic fraction by bisecting the C-rate. The fractions for the cathode are influenced by the number of charge/discharge cycles leading to a higher electrolytic lithium fraction after a higher number of cycles. For the electrolyte, no clear trend is apparent which could be explained due to the surplus of electrolyte in coin cells in comparison to other cell setups. Above that, the GD depth profiles reveal that a strong accumulation of electrolytic lithium is detected on the surface of the anode after the first formation step. With a longer cycling procedure, this trend fades away and leads to almost constant lithium fractions throughout the electrode. This may be due to lithium from the electrolyte that decomposes directly after applying a voltage because of the direct contact with the negative electrode. All these findings considered, the conversion method, described in Section 2.6 to transfer the 6Li-abundances into the respective lithium fractions, is a strong technique to gather insights into the migration and distribution of lithium during cycling of LIB cells. As a follow-up to this work, these measurements should be extended using a SIMS system to gather additional information about the isotopic distribution on a particular scale.
This journal is © The Royal Society of Chemistry 2019 |