Moisture behavior of lithium-ion battery components along the production process

The need of high quality lithium-ion batteries continuously grows since their first commercial usage. The enormous market for LIB give it a key role in modern day society: Mobile devices, temporary storage for renewable energies or transportation are just a few of the many fields of application. Therefore a safe and reliable product with a high capacity, cycle life and stability against aging is therefore obligatory. Strict quality control along the entire production process is necessary to ensure these properties and in consequence a high-quality product. Edge et al. give a good overview about lithium-ion battery degradation. [1]

The fact that moisture can have an impact directly on components of the LIB or the entire cell is widely known and scope of research for many years. Small amounts of water are inevitable to occur during the production of LIB, due to the hygroscopic behaviors of the LiPF₆ within the electrolyte [2], [3] and electrode materials [3], [4]. On the one hand, water within the cell can have a positive effect. Small amounts of water in the cell are proven to be beneficial for the solid electrolyte interface (SEI) formation on the anode [5], [6], [7]. On the other hand, there are many negative impacts that are linked to the presence of moisture [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23]. Moist commonly known is the creation of LiOH by the presence of H₂O either within electrolyte [8] or cathode material [9], [10]. Adding CO₂ to the hydroxide reaction to the more stable Li₂CO₃ [9], [10] occurs, e.g. in air with high humidity [10]. The reacted lithium is bound within the hydroxide and carbonate and not available for the usage in the LIB. Therefore the cell capacity and also the cell performance in general are therefore impaired [10], [11], [12], [13], [14], [15], [16]. Cathode material with a high nickel content is more sensible against H₂O and CO₂, because nickel tends to create carbonates. Unfortunately they are known for their high capacity and therefore very popular. Nickel rich cathodes for instance are also prone to create basic hydrated nickel carbonate-hydroxides in moist air, which also leads to capacity loss [9], [10]. Zou et al. were able to prove that the amount of LiOH and Li₂CO₃ directly depends on the amount of nickel within NMC (nickel, manganese, cobalt) cathodes [17]. Langklotz et al. and Meinl et al. observed a capacity loss for LFP or LiO material with increasing moisture content [14], [15]. A significant capacity loss and swelling within the cell above 1000 ppm_w water content was observed in multiple studies after analyzing cells with various moist electrolytes [3], [16], [18]. The passivation of lithium breaks above 1000 ppm_w as shown by Langklotz et al. in their work and is suggested to be the reason for lower cell capacity [14].

The highly reactive fluorine within the electrolyte is known to create hydrofluoric acid which can lead into additional difficulties [14], [19], [20], [21], [22], [23], [24]. Amatucci et al. state that already trace amounts of water are sufficient to create hydrofluoric acid (HF) [19]. Since the electrolyte reaches all components in the cell, the amount of water in all components of the final cell is relevant for the aging effect. Lux et al. could prove that besides the humidity also the environmental temperature influence the formation of HF. They could observe a supporting effect in the creation of the acid of 50 °C in contrast to regular ambient temperatures [24]. The HF can attack the SEI layer by dissolving Li-ions, which also impairs the cell performance [22], [24]. In addition to that, the active material of the cathode can also interact with the acid. In multiple works manganese dissociation was observed for lithium manganese spinel cathodes, which contain Mn(iii) [19], [22], [23]. Here it should be mentioned that mostly commercial NMCs contain Mn(iv), which tends much less to dissolution and even stabilizes the structure [25], [26], [27]. The reaction between HF and LiCoO₂ is described as a somewhat autocatalytic reaction. The acid attacks the LiCoO₂, which results in the creation of additional H₂O, which leads to more HF [23]. There are also suggested solutions to minimize water absorption by processing under dry room conditions. In their work Zaghib et al. proved this by handling LiFePO₄ at a relative humidity (rh) under 5 % [8]. Also the use of various materials like sodium can lead to a lower sensitivity against water [28], [29].

Since avoiding water within the LIB components and the cell is inevitable, the moisture behavior of all components needs to be fully understood. Depending on the relative humidity around the material and characteristics of the material itself, sorption equilibria of moisture within each component are established. The composition of the material takes an important part in the overall moisture behavior, whether the overall water uptake or the kinetics. Langklotz et al. compared water adsorption kinetics for different cathode material with NMC (Li(Mn0.37Co0.35Ni0.35)O2.07) or LFP, where the LFP cathodes adsorbed around 2000 ppm_w and the NMC cathodes 700 ppm_w. The overall water uptake on electrode surfaces can be described by the adsorption model of Brunauer, Emmet and Teller (BET) [14]. Eser et al. investigated the adsorption behavior of anode material in detail. They found out that the amount of the single components approximately add up to the total amount of the whole electrode. They could also prove that the used binder material carboxymethyl cellulose (CMC), which is also used in the anode for this study, carries the most amount of water within the anode material. While the moisture is located adsorbed in the porous structures of the components and the binder material of the battery, it can also be absorbed by the CMC binder. They could also observe a hysteresis behavior for anode material. Once the water content of anode material increased due to humidity, the water content does not drop down to the original value, if the dew point temperature decreases back to the origin. In this case an active drying process (also known as baking, secondary drying or post-drying) is necessary for their investigated electrode [4], [30].

Parts of the adsorbed water can be driven out of the battery cell and its components by this commonly used baking process [4], [14], [31], [32], [33], [34], [35], [36], [37]. There is a variety of secondary drying procedures for different production scales. In laboratory scale, an established method for electrode baking is the single sheet drying in a vacuum oven. In case of electrodes coated one-sided, some sort of fixture is advisable to avoid curling effects. If separator material is even post-dried in lab scale, it is usually done as an entire coil in a vacuum oven. Also cell stack drying or cell drying is used in laboratory scale, where the dried product is also placed into a vacuum oven. In pilot line scale, secondary drying of electrodes occurs either as a pack of multiple sheets or as a coil in a vacuum oven. Also, roll-to-roll baking is executed. Separator material, cells or cell stacks are similarly baked as in laboratory scale, but in larger sizes. In production scale electrode material is either post-dried as a coil in a vacuum oven or in a roll-to-roll process. The vacuum oven can work either as a stand-alone type, where the three typical stages of vacuum baking occur in one chamber: 1) product heat up, 2) vacuum drying and 3) product cooling. The other type of vacuum oven is the tunnel type, where multiple vacuum ovens and their chambers are directly connected. Here the three stages of vacuum baking are split, so that only the first section of the tunnel heats up the product, the middle section continuously holds vacuum and the last section cools down the dried component. The roll-to-roll drying can also be running as a stand-alone solution or it can be integrated in other roll-to-roll processes. The separator material by itself is usually not dried in production scale and only stored under dry room conditions of around 0,5 % rh. Cell baking is a commonly used method to ensure the least possible moisture within the final cell. This drying often takes place in vacuum ovens, also either as a stand-alone process or as tunnel type. Here multiple cells are stacked together on a tray and the moisture is only able to evaporate out of the filling holes.

Drying parameters vary a lot for all procedures, since they depend on the dried material, its size and morphology, the baking process, the initial water content and individual permitted limits after baking and inside the cell. Electrodes are usually thermally more stable than the separator, so they can handle higher temperatures without being damaged. In the literature temperature in vacuum ovens vary between 70 °C and 200 °C, the drying time between 2 and 24 h and the operating pressure goes down to 0.01 mbar. With rising temperature the final water content of the dried component drops, but too much drying can also be harmful to the product. For instance Huttner et al. analyzed different post-drying parameters for vacuum drying. In their investigations the most intensive drying caused the lowest water content but also the weakest cell performances in contrast to milder drying conditions. They suggest that too intense drying deteriorates the anode binder, which leads to poorer cell performances [31]. Eser et al. investigate very nicely the diffusion kinetics while post-drying of graphite anodes. They describe five different resistances within the porous structure of the LIB components, which directly influences the drying time. 1) external resistance, 2) resistance in macropores, 3) resistance in micropores, 4) resistance in the binder and 5) heat transfer resistance. Additionally, the concentration and temperature of the binder affects the drying process. They also state, that in a drying process with freely accessible coating surface like in a single sheet or roll-to-roll baking, the mass transport of water out of the coating into the gas phase depends more on the dryer itself, than on the diffusion through the coating [4].

Further processing after baking is usually under dry room conditions of at least −40 °C at ambient temperatures to minimize remoistening within the LIB components [36], [38], [39]. There is still some remoistening happening after the secondary drying, which Huttner et al. described in their work. They state that the relative humidity after baking outside the drying device is higher than it is inside. Therefore, the dried component inevitably begins to take up moisture again to a certain degree [40]. This effect needs to be considered while designing the baking process and also measuring the water content.

It has already been demonstrated in detail that even the pure NMC as initial material, the cathode and the electrolyte are affected by aging due to moisture. Consequently the best possible understanding of moisture behavior of all lithium-ion battery components is already necessary from the first process step, along the entire electrode production up to the cell itself. The authors are not aware of such an summarizing and extensive elaboration, which is the main motivation for this work. Busà et al. for instance investigated the effects of moisture exposure on raw NMC-811 material and their impact on the cell performances. [41] Other research like Geng et al. or Senthil et al. focus more on the improvement of the raw material to stabilize it against the potentially harmful exposure of moisture along the manufacturing process. [42], [43] Therefore this paper will start to describe how much moisture can be adsorbed by the individual components depending on the humidity. Then it will be evaluated how fast these values can be achieved under typical production conditions, which is a decisive criterion for the design of the production process of LIB. To give the reader an application-oriented and good overview, the detected water content along the entire production chain is then examined using a production campaign at a pilot line. As literature and own experiments proved that moisture in the final cell cannot be completely avoided, a total of 5 methods are compared to reduce the water content of the components in the LIB. Finally, a strategy is proposed to achieve this goal as effectively as possible for the material and methods considered.