Monitoring Changes in Electrolyte Composition of Commercial Li-Ion Cells after Cycling using NMR Spectroscopy and Differential Thermal Analysis
使用核磁共振波谱和差热分析监测循环后商用锂离子电池电解质组成的变化

All cells tested herein were commercial LG 18650 high power cells in which their capacity was checked prior to the electrolyte extraction. Deuterated chloroform (CDCl₃) (>= 99% purity) and Tetramethylsilane (TMS) (ACS reagent) NMR grade, (>=99.9% purity) were purchased from Sigma-Aldrich. Wilmad (R) low pressure NMR tubes (300 MHz frequency) were used that were also purchased from Sigma-Aldrich.
本文测试的所有电池均为商用 LG 18650 高功率电池，在提取电解液之前检查其容量。氘代氯仿（CDCl ₃ ）（>=99%纯度）和四甲基硅烷（TMS）（ACS试剂）NMR级（>=99.9%纯度）购自Sigma-Aldrich。使用了Wilmad（R）低压NMR管（300 MHz频率），这些管也是从Sigma-Aldrich购买的。

Cells were cycled using Maccor S4000 in a BINDER climate chamber set to 40 °C. Cells were charged and discharged at 40 °C between 2.5 − 4.2 V with a multistage constant current (MCC) fast charging protocol. This consisted of 1.45 C charging to 40% state of charge (SOC), 1.2 C to 60% SOC and 0.7 C to 80% SOC based on coulomb counting, followed by a C/3 charge to 4.2 V. This was then followed by a constant voltage (CV) step with current tapered to C/10. The discharge was carried out at C/3 to 0% SOC. After every 23 cycles of fast charging, 2 benchmark cycles were performed at constant current of C/5 for charge and discharge with a CV step until the current is tapered to C/10. The benchmark cycles were used to adjust the capacity rating of the cell in order to maintain the fast charge SOC steps via coulomb counting.
使用Maccor S4000在设置为40°C的BINDER气候室中循环细胞。 电池在40°C的2.5 − 4.2 V之间使用多级恒流（MCC）快速充电协议进行充电和放电。这包括 1.45 C 充电至 40% 充电状态 （SOC）、1.2 C 充电至 60% SOC 和 0.7 C 充电至 80% SOC（基于库仑计数），然后 C/3 充电至 4.2 V。然后是恒压 （CV） 步进，电流逐渐减少到 C/10。放电在 C/3 至 0% SOC 下进行。在每 23 个周期的快速充电后，以 C/5 的恒定电流执行 2 个基准循环，以 CV 步长进行充电和放电，直到电流逐渐减小到 C/10。基准循环用于调整电池的额定容量，以便通过库仑计数保持快速充电 SOC 步长。

Electrolyte was extracted from fresh (as received - uncycled) and cycled cells that have gone through different cycling durations: 100, 200, 300, 600, 900, 1200 and 1600 cycles. For each cycle number, two cells were opened. All samples were prepared in an argon atmosphere glovebox (<5 ppm H₂O/O₂). The top cap of the battery cell was opened using a pipe cutting tool, allowing the exposure of the top surface of the cell jelly roll shown in Fig. 1A. The opened cell was then immediately placed in an HDPE (high density polyethylene) vial containing 10 g CDCl₃ with the exposed top part facing down to extract as much of the electrolyte and avoid any evaporation. CDCl₃ was used as the extraction solvent to acquire only the soluble organic components of the electrolyte and leave out its insoluble salt constituents. Note, monitoring both the solvent and salt consumption over cycle life would be of great value. Therefore, DMSO was also tested but showed to react with components within the cell resulting in electrolyte discolouration, thus CDCl₃ was the only solvent used through this study. The advantage of extracting only the organic components is that it allows the electrolyte to be further analysed if required with other complementary systems such as GC-MS that is susceptible to damage by corrosive fluoride-based species. The cell was allowed to mix continuously for 24 h in a tubular mixer making sure the cell is positioned vertically along its axis as shown in Fig. 1B, to maximise interaction between the solvent and the jelly roll. It is crucial to maintain a suitable mixing time as it can influence the extracted amount. As the basis of our methodology, we have chosen the duration of mixing to be 24 h as no significant change was observed when mixed for 48 h, shown in the Supporting Information (SI), Fig. 1S. This works towards an efficient and effective mixing time whereby the solvent is able to reach all the pores resulting in a homogeneous mixture. Following the mixing procedure, the solution was filtered using Whatman Spartan syringe filters (SPARTAN 30/0.45 RC filter unit). 1 g of reference standard (consisting of TMS: CDCl3 in 1:5 ratio) was added to 5 g of the filtrate, as shown in Fig. 1C. As both TMS and CDCl₃ are volatile solvents, the reference standard was freshly prepared each time. The final extracted solution containing the reference standard was then directly transferred into an NMR tube (Fig. 1D). All ¹H-NMR analysis were conducted on a Bruker Avance III HD 300 MHz spectrometer. The NMR data processing was done using TopSpinTM (Bruker, USA). Each spectrum was shifted against TMS peak at 0 ppm.
电解质是从新鲜的（接收的 - 未循环的）和循环电池中提取的，这些电池经历了不同的循环持续时间：100、200、300、600、900、1200 和 1600 次循环。对于每个周期数，打开两个单元格。所有样品均在氩气气氛手套箱（<5 ppm H ₂ O / O ₂ ）中制备。使用管道切割工具打开电池的顶盖，允许暴露图1A所示的电池果冻卷的顶面。然后立即将打开的电池放入含有10g CDCl ₃ 的HDPE（高密度聚乙烯）小瓶中，暴露的顶部朝下，以提取尽可能多的电解质并避免任何蒸发。使用CDCl ₃ 作为萃取溶剂，仅获取电解质的可溶性有机组分，而忽略其不溶性盐成分。请注意，在循环寿命期间监测溶剂和盐的消耗量将具有重要价值。因此，DMSO也进行了测试，但显示与电池内的成分发生反应，导致电解质变色，因此CDCl ₃ 是本研究使用的唯一溶剂。仅提取有机成分的优点是，如果需要，可以与其他互补系统（如易受腐蚀性氟化物基物质损害的GC-MS）一起进一步分析电解质。允许将细胞在管式混合器中连续混合24小时，确保细胞沿其轴垂直定位，如图1B所示，以最大限度地提高溶剂和果冻卷之间的相互作用。保持合适的混合时间至关重要，因为它会影响提取量。 作为我们方法的基础，我们选择混合的持续时间为24小时，因为混合48小时时没有观察到显着变化，如支持信息（SI）所示，图1S。这有助于实现高效和有效的混合时间，从而使溶剂能够到达所有孔隙，从而形成均匀的混合物。按照混合程序，使用Whatman Spartan注射器过滤器（SPARTAN 30/0.45 RC过滤装置）过滤溶液。将1 g参比标准品（由TMS：CDCl3组成，比例为1：5）加入到5 g滤液中，如图1C所示。由于TMS和CDCl ₃ 都是挥发性溶剂，因此每次都要新鲜制备参考标准品。然后将含有参考标准品的最终萃取溶液直接转移到NMR管中（图1D）。所有 ¹ H-NMR 分析均在 Bruker Avance III HD 300 MHz 波谱仪上进行。使用TopSpinTM（美国布鲁克）进行NMR数据处理。每个光谱在0 ppm时与TMS峰相比偏移。

Zoom In  放大Zoom Out  缩小Reset image size  重置图像大小

DTA measurements were performed on 8 cylindrical cells (1 fresh and 7 cycled cells of which gone through: 100, 200, 300, 600, 900, 1200 and 1600 cycles), with a dry Novonix cell used as a reference. OCV of all cells was close to 3.53 V +/− 0.01 V). All the cells were subjected to the following three testing conditions: (i) from room temperature, cool to −150 °C at a rate of 3 °C min⁻¹; (ii) hold isothermally at −150 ^oC for 20 min and (iii) heat to 45 ^oC at a rate of 0.5 ^oC min⁻¹. Testing was carried out onsite at Novonix Testing Services, Bedford, Canada. ⁴² The DTA signal originates from the difference in measured temperature during phase transition between the sample test cell at each cycle number and a reference uncycled fresh cell of the same type (T_sample – T_reference). ⁴¹
对 8 个圆柱形电池（1 个新鲜电池和 7 个循环电池，其中经历了：100、200、300、600、900、1200 和 1600 次循环）进行了 DTA 测量，以干燥的 Novonix 电池为参考。所有电池的OCV接近3.53 V +/− 0.01 V）。所有细胞都经受以下三种测试条件：（i）从室温，以3°Cmin的速率冷却至-150°C ⁻¹ ;（ii）在-150 ^o °C下保持等温20分钟，（iii）以0.5 ^o °C的速率加热至45 ^o ° ⁻¹ C。测试在加拿大贝德福德的 Novonix 测试服务公司现场进行。 ⁴² DTA信号来源于每个循环次数的样品测试池与相同类型（T-T _{sample reference} ）的参考未循环新鲜池之间相变期间的测量温度差。 ⁴¹

To investigate the robustness of the extraction method, electrolyte was extracted from five fresh cells (A-E) of the same type (LG 18650). Both GCMS (in SI, Fig. 2S) and NMR were used to obtain the composition of the electrolyte. Figure 2 shows the acquired NMR spectra. The major solvent peaks were found to be DMC, EC, PC, EMC, DEC with succinonitrile (SN) and tris(trimethylsilyl)phosphate (TMPS) as additives. SN acts as a corrosion inhibitor of the copper current collector as well as improving thermal stability in EC-based electrolytes. ^43,44 SN has also been shown to improve cell performance significantly at high voltages by forming a more stable and efficient SEI film. ⁴⁵ TMPS has been used in Li-rich cathodes to stabilise the electrode/electrolyte interface resulting in improved cycling performance. ⁴⁶ TMPS oxidises more readily as compared to other electrolyte components. In addition to this, it is highly reactive towards HF and acts as an HF scavenger, thereby improving the electrochemical performance of LIBs. ⁴⁷ A singlet peak at ∼3.53 ppm was also observed consisting of lithium ethylene dicarbonate (LEDC) resulting from EC reduction illustrated in Fig. 2D, as previously reported. ⁴⁸ NMR signals derived from EMC and DEC are difficult to separate as they result in overlapping chemical shifts. Additionally, as shown in Fig. 2B, the triplet peak overlaps with a singlet peak derived from CDCl₃ impurities making it difficult to quantify. Moreover, the quartet peaks of EMC and DEC shown in Fig. 2G were very small in nature (∼0.2% weight relative to other solvents) and require a higher resolution NMR to obtain sharper signals for quantification. Thus, the relative peaks of DMC (singlet at ∼3.78 ppm), PC (doublet at ∼1.5 ppm), EC (singlet at ∼4.5 ppm), SN (singlet at ∼2.78 ppm), TMPS (singlet at ∼0.19 ppm) and LEDC (singlet at ∼3.5 ppm) were the ones that were quantified and compared throughout. Table I illustrates the relative percentage weights present in the overall electrolyte extracted from 5 fresh cells (see SI3 for detailed information of the associated peaks). Similar compositions were observed for each cell with the electrolyte being dominated by DMC highlighting the reliability of the extraction method used.
为了研究提取方法的稳健性，从五个相同类型的新鲜细胞 （AE） （LG 18650） 中提取电解质。GCMS（SI中，图2S）和NMR均用于获得电解质的组成。图2显示了采集的NMR波谱。主要溶剂峰为DMC、EC、PC、EMC、DEC，以丁二腈（SN）和磷酸三（三甲基硅基）酯（TMPS）为助剂。SN可作为铜集流体的缓蚀剂，并改善EC基电解质的热稳定性。 ^43,44 SN还被证明可以通过形成更稳定和高效的SEI薄膜来显着提高电池在高压下的性能。 ⁴⁵ TMPS已用于富锂阴极，以稳定电极/电解质界面，从而改善循环性能。 ⁴⁶ 与其他电解质成分相比，TMPS更容易氧化。除此之外，它还对 HF 具有高度反应性，并充当 HF 清除剂，从而提高 LIB 的电化学性能。 ⁴⁷ 如前所述，还观察到 ∼3.53 ppm 的单线态峰，由图 2D 所示的 EC 还原产生的碳酸乙烯酸锂 （LEDC） 组成。 ⁴⁸ 来自EMC和DEC的NMR信号很难分离，因为它们会导致重叠的化学位移。此外，如图2B所示，三重态峰与来自CDCl ₃ 杂质的单重态峰重叠，因此难以定量。此外，图2G所示的EMC和DEC四重峰在性质上非常小（相对于其他溶剂的重量约为0.2%），需要更高分辨率的NMR才能获得更清晰的定量信号。因此，DMC（单重态，∼3.78 ppm）、PC（双重态，∼1。5 ppm）、EC（单线态 ∼4.5 ppm）、SN（单线态 ∼2.78 ppm）、TMPS（单线态 ∼0.19 ppm）和 LEDC（单线态 ∼3.5 ppm）是量化和比较的。表I说明了从5个新鲜电池中提取的总电解质中存在的相对百分比权重（有关相关峰的详细信息，请参见SI3）。观察到每个电池的类似成分，电解质以DMC为主，突出了所用提取方法的可靠性。

Zoom In  放大Zoom Out  缩小Reset image size  重置图像大小

Ordinarily, Li-ion batteries exhibit a slow capacity degradation, until they reach an inflection point - whereby the capacity fade becomes much more rapid and drops off quickly as shown in the capacity retention and the associated cycle efficiency plots illustrated in Fig. 3i. This is commonly referred to as a knee-point. ⁴⁹ It has been shown that a knee-point marks the rapid capacity fade trend to the end-of-life (EOL). ⁵⁰ Herein, electrolyte was extracted from various cells at different stages of the cell's cycle life before and after the knee-point. Figure 3i illustrates the multiple points at which electrolyte was extracted, at cycle numbers 100, 200, 300, 600 and 900 being above the 80% capacity retention mark and 1200 and 1600 falling below that mark. Cells were then disassembled and as illustrated in Fig. 3ii, the presence of dense lithium plating on the anode electrode surface was observed. Accordingly, this could indicate its contribution to the poor cycling efficiency seen from ∼1000 cycles as it continues to form throughout cycling.
通常，锂离子电池表现出缓慢的容量下降，直到它们达到拐点——此时容量衰减变得更快并迅速下降，如图 3i 所示的容量保持和相关循环效率图所示。这通常被称为膝盖点。 ⁴⁹ 研究表明，拐点标志着产能快速衰减趋势，直至寿命终结 （EOL）。 ⁵⁰ 本文中，电解质是从处于细胞循环寿命的不同阶段的各个细胞中提取的，在膝盖点之前和之后。图3i显示了提取电解液的多个点，循环编号100、200、300、600和900高于80%容量保持标记，1200和1600低于该标记。然后拆卸电池，如图3ii所示，观察到阳极电极表面存在致密的锂镀层。因此，这可能表明它导致了 ∼1000 次循环中观察到的不良循环效率，因为它在整个循环过程中继续形成。

Zoom In  放大Zoom Out  缩小Reset image size  重置图像大小

To observe and quantify the consumption of individual components, the peaks for DMC, EC, PC, LEDC, as well as SN and TMPS additives were analysed against the standard TMS peak, as shown in Fig. 4i. Hence, an initial increase in the peak areas was seen for DMC, PC, EC and LEDC relative to the TMS as illustrated in Fig. 4ii. This is likely due to the use of cells of a different batch number relative to the fresh cells, highlighting the sensitivity of the measurements. This was then followed by a gradual decrease over cycle number. TMPS showed a steady decrease over cycle number and stabilises after 800 cycles, whereas SN indicated a much more evident decrease at an exponential rate throughout the cycle life and was completely used up at cycle number 1600. The observed evidence of the constant decrease of SN through the cycles, can therefore be deduced that SN has a significant role in improving the cell's overall cycle life. Total disappearance of SN coincides with the occurrence of a knee-point – this further supports the significance of its role. Interestingly, the peak for LEDC stayed consistent throughout cycle number. LEDC was shown to be one of the main compositions of the SEI layer formed on graphite anodes in EC and EMC based electrolytes and have experienced no change in its composition upon cycling. ⁴⁸ This illustrates the effectiveness of the SEI layer containing LEDC by preventing further electrolyte reduction. In addition, we observed the increase of NMR peaks in the region around ∼5.77 and 5.64 ppm as illustrated in Fig. 5. This is a representation of degradation products as the cells were cycled originating from the decomposition products of DMC attributed to acetal species, such as methanediol and methoxymethanol. ⁵¹
为了观察和量化单个组分的消耗量，根据标准TMS峰分析了DMC、EC、PC、LEDC以及SN和TMPS添加剂的峰值，如图4i所示。因此，DMC、PC、EC 和 LEDC 相对于 TMS 的峰面积最初增加，如图 4ii 所示。这可能是由于使用了与新鲜细胞不同批号的细胞，突出了测量的灵敏度。然后，循环次数逐渐减少。TMPS在循环次数上稳步下降，并在800次循环后稳定下来，而SN在整个循环寿命中以指数速率下降更为明显，并在循环次数1600时完全用完。因此，观察到SN在循环过程中不断减少的证据可以推断出SN在改善细胞的整体循环寿命方面具有重要作用。SN的完全消失与膝点的发生相吻合，这进一步支持了其作用的重要性。有趣的是，LEDC的峰值在整个周期数中保持一致。LEDC被证明是EC和EMC基电解质中在石墨阳极上形成的SEI层的主要成分之一，并且在循环时其成分没有变化。 ⁴⁸ 这说明了含有LEDC的SEI层通过防止电解质进一步还原的有效性。此外，我们观察到该区域的NMR峰增加，约为∼5.77和5.64 ppm，如图5所示。这是降解产物的表示，因为细胞的循环起源于归因于缩醛物种的DMC分解产物，例如甲烷二醇和甲氧基甲醇。 ⁵¹

Zoom In  放大Zoom Out  缩小Reset image size  重置图像大小

Zoom In  放大Zoom Out  缩小Reset image size  重置图像大小

The compositional changes of each component were then compared and shown in Fig. 6. An overall decrease in % weight was seen in EC, PC, SN and TMPS whereas DMC and LEDC has showed an increasing trend over cycle life. This demonstrates the different rates at which each component is used up during cycling.
然后比较每种组分的组成变化，如图6所示。EC、PC、SN 和 TMPS 的重量百分比总体下降，而 DMC 和 LEDC 在循环寿命期间呈上升趋势。这显示了在循环过程中每个组件的不同消耗率。

Zoom In  放大Zoom Out  缩小Reset image size  重置图像大小

In addition, as a complementary technique DTA was implemented. It operates by studying the thermal behaviour of a sample test cell as it heats through the phase transitions of the electrolyte relative to the behaviour of a reference cell whose reference electrolyte does not go through any phase transitions in that same temperature range. The difference between these temperature signals is plotted against the test cell temperature, as illustrated in Fig. 7i which represents the DTA results performed on each cell at the various cycles. As battery liquid electrolytes contain a mixture of components comprising of organic solvents and salts, a typical DTA signal has multiple features resulting in a unique thermal fingerprint of the test electrolyte. Ding et al., illustrated the phase diagram of DMC:EC solvent mix in which similar temperatures are obtained herein. ⁵² They highlighted how the addition of small amounts of salt can impact the phase transition temperature of the binary mixture. This helps to explain the slight difference in temperatures obtained herein which can be attributed to the presence of LiPF₆ salt in the cell electrolyte. Generally, the position of the peak temperature as well as peak area are influenced by the solvent composition and salt concentration which can be used to track electrolyte changes. ⁴¹ Fig. 7ii, represents the peak temperatures obtained from the DTA signal of each cell. The peak temperatures gradually increase as the cycle number increases. This is attributed to be caused by electrolyte compositional changes. In addition, similar DTA observation was shown by Day et al. associated particularly with the consumption of LiPF₆ salt during cycling. ⁴¹ Additionally, the peak areas of the DTA signals were extracted, shown in Fig. 7iii. This can indicate the changes in the overall electrolyte mass, as solvent consumption leads to a reduction in peak area. An increase in the peak area is observed with cycle life up to 900 cycles. This can be attributed more towards the changes in salt concentration and the formation of degradation products as little variation was seen in solvent composition. However, after 900 cycles the peak area decreases significantly as we approach the knee-point which can be attributed to the changes in the overall electrolyte solvent ratio as a result of the complete consumption of SN component. This agrees well with the poor cycle efficiency observed in Fig. 3i as well as the compositional changes seen in the NMR observations, in which EC and SN resulted in greater reduction over the cycle life. The consumption of such components can trigger lithium plating as observed herein (see Fig. 3ii), this can be exacerbated specially if electrolyte components are being used up at different rates which can lead to localised plating. Furthermore, the cell which has undergone 1600 cycles (after the knee-point) resulted in a distinguished DTA signal, emphasizing the robustness of the method to determine aged cells in a non-destructive manner efficiently.
此外，作为补充技术，还实施了DTA。它通过研究样品测试池的热行为来工作，因为它通过电解质的相变加热，相对于参比电池的行为，参比电解质在相同的温度范围内不经历任何相变。这些温度信号之间的差值与测试电池温度作图，如图7i所示，它表示在不同周期下对每个电池执行的DTA结果。由于电池液体电解质含有由有机溶剂和盐组成的成分混合物，因此典型的 DTA 信号具有多种特征，从而产生测试电解质的独特热指纹。Ding等人说明了DMC：EC溶剂混合物的相图，其中获得了相似的温度。 ⁵² 他们强调了添加少量盐如何影响二元混合物的相变温度。这有助于解释本文获得的温度的微小差异，这可归因于电池电解质中存在 ₆ LiPF盐。通常，峰值温度的位置和峰面积受溶剂组成和盐浓度的影响，可用于跟踪电解质的变化。 ⁴¹ 图7ii表示从每个电池的DTA信号获得的峰值温度。峰值温度随着循环次数的增加而逐渐升高。这归因于电解质组成变化。此外，Day等人也观察到了类似的DTA，特别是与循环过程中 ₆ LiPF盐的消耗有关。 ⁴¹ 此外，还提取了DTA信号的峰面积，如图7iii所示。这可以表明总电解液质量的变化，因为溶剂消耗会导致峰面积的减少。观察到峰面积增加，循环寿命可达 900 次循环。这更多地归因于盐浓度的变化和降解产物的形成，因为溶剂成分的变化很小。然而，在900次循环后，当我们接近拐点时，峰面积显着减小，这可归因于SN组分完全消耗导致整体电解质溶剂比的变化。这与图3i中观察到的较差的循环效率以及NMR观测中看到的成分变化非常吻合，其中EC和SN导致循环寿命的更大降低。如本文所观察到的，此类组件的消耗会触发锂电镀（见图3ii），如果电解质组件以不同的速率消耗，这可能导致局部电镀，这种情况尤其会加剧。此外，经过 1600 次循环（在拐点之后）的细胞产生了显着的 DTA 信号，强调了该方法以非破坏性方式有效确定老化细胞的稳健性。

Zoom In  放大Zoom Out  缩小Reset image size  重置图像大小