3D-Printed Testing Plate for the Optimization of High C-Rates Cycling Performance of Lithium-Ion Cells

The 3D plate was designed with free CAD software (Blender) and printed with a Fused Deposition Modeling (FDM) 3D printer (Ultimaker3, Ultimaker). Aside from a good extrusion accuracy, the choice of the material is fundamental since it needs to be electrolyte resistant. The most used in commercially available test cells is the expensive polyether ether ketone (PEEK) thanks to its superior mechanical and chemical properties. Since the extrusion process of this polymer requires high temperatures and high-end 3D printers, we have opted for a cheaper and easy to print polymer such as polypropylene (PP). Figure 1a shows its simple design. The electrodes´ tabs are fixed in two guides (see Fig. 1a—I) allowing the electrodes stack to stay in the desired position. The reference electrode (RE) position is well defined by design: to avoid short circuits and limit measurement's artifacts, the Li-RE (0.2 mm² surface) is placed at a horizontal distance of 300 μm from the stack (see Fig. S1). Single-layer pouch cells assembled in the 3D plate setup will be named PP3D cells (Fig. 1c).

All materials were used as received. Anode and cathode slurries were prepared in a planetary mixer where active materials portions were kept as high as possible to match industrial targets and good coating performance in terms of mechanical and electrochemical stability. The used materials were Graphite (Hitachi Chemical), LiNi_0.8Mn_0.1Co_0.1O₂ (NMC 811, Targray), Na-CMC (Sunrose), PVdF (Solvay Solexis), SBR emulsion (Zeon), Super C65 (SC65, Imerys) and SFG6L graphite (Imerys). Distilled water and N-methylpyrrolidone (NMP, Sigma Aldrich) were used as solvents respectively for anode and cathodes. The slurries were then cast in a pilot line (LACOM Gmbh, Germany) equipped with a comma bar system, four drying ovens (total length: 8 m), and a Kr online mass scanner to ensure a uniform loading throughout the coatings. All electrodes were prepared by single-side coating and calendering to a specific density. The most important parameters of the electrodes are shown in Table I.

Where not specified, all the cells were assembled with the materials and components listed in Table I. For single-layer pouch cells (SLPCs), electrodes were punched with a customized mechanical puncher (anode area: 26 cm²; cathode area: 23.94 cm²), dried overnight at 130 °C under vacuum and stored in a dry cabinet. A tri-layered polymeric separator (Celgard 2325) was used and Al and Ni tab collectors were ultrasound-welded respectively to cathode and anode. Before electrolyte filling, the cell was furthermore dried at 80 °C overnight under vacuum.

Unlike regular single-layer pouch cells, when the PP3D plate is used, Celgard 2325 separator (26 cm²) was stacked between anode and cathode and not rolled around them as in the SLPCs. A small piece of freshly cleaned metallic lithium (PI-KEM) was filled inside the channel where a thin copper wire was inserted. Anode and cathode were aligned on the reference electrode side and held in position by Kapton tape. Both cell types were assembled in a dry room (dew point < −65 °C), then they were vacuum-dried overnight at 80 °C. SLPC and PP3D cells were filled respectively with 0.9 ml and 1.2 ml of electrolyte and vacuum-sealed in a glovebox (MBraun, O_2, and H₂O levels < 0.5 ppm).

All the electrochemical procedures listed in Table SI (Supporting Info available online at stacks.iop.org/JES/168/050508/mmedia) were carried out on a VMP-3 (Biologic) system connected to a climate chamber (VB-4004,Vötsch) where the cells were tested at controlled temperatures (25 °C and 40 °C). The resulting capacities are always normalized with respect to the first discharge value after the formation or the cathode active material weight. SLPCs and PP3D cells were cycled using an in-house designed pouch cell holder (Fig. 1b) equipped with pogo pins to test cells in a 4-points configuration. Thanks to the knowledge gained from previous tests on this anode/cathode combination, the bolts of the holder were tightened applying a defined torque momentum resulting in an average force of 100 N on the surface of the pouch cell. The holder assures a homogenous pressure on the cells and it is minimizing the electrical contact resistance between VMP probes and the tabs of the cells.

Aged cells were opened in an Ar-filled glovebox (MBraun, O_2, and H₂O levels < 0.5 ppm), and pictures of the graphite anodes were taken. The absence or presence of lithium deposition was further confirmed by GD-OES analysis according to the method published earlier. ²⁹ lithium deposition quantification was conducted using reference samples.

Aged anode samples were washed three times in Dimethyl Carbonate (DMC), left to dry, and eventually mounted on airtight sample holders. An additional test was performed on the pristine anode coating to evaluate its binder distribution following the sodium relative concentration (from the Na-CMC binder) throughout the coating thickness. GD-OES analyses were conducted using a GDA750 device (Spectruma). The measurements on the anodes were performed in RF mode at a discharge voltage of 550 V, a gas pressure of 2 hPa, and a pulse rate of 50% at a frequency of 2500 Hz. A mixture of H₂ 6.0 and Ar 6.0 (1:99 vol. %) was used as a discharge gas. The following emission lines were used for detection: Cu (327.4 nm), C (156.1 nm), O (130.2 nm), Li (670.7 nm), P (178.3 nm), H (121.6 nm), Na (589.0 nm). The sputtered craters in the samples produced by the measurement have a diameter of 2.5 mm.

The formation protocol in Table SI (supporting Info) was applied to SLPCs and PP3D cells. Figure 2a shows the three formation cycles for the two different setups. Charge/discharge profiles are overlapped leading to the same capacities. The use of the PP3D plate did not influence the electrochemical performance of the stack tested in a state-of-the-art SLPC. Firstly, both showed an initial irreversible capacity loss (ICL) of 11.5%, a value in perfect agreement with the ICL values found in the literature (10% < ICL < 20%). ^30,31

Zoom In Zoom Out Reset image size

The presence of the Li-RE in PP3D cells affords us to separate anode and cathode contributions. Figures 2b–2c report the third formation cycle for the anode and cathode, respectively. In Fig. 2b it is possible to distinguish the lithium intercalation process in the graphite anode. The dQ dE⁻¹ vs. E plot in the inset allows for better visualization of the graphite staging. The redox peaks around 230 (III), 120 (II), and 80 mV (I) correspond to the formation of LiC₃₆, LiC₁₂, and LiC₆ graphite intercalation compounds (GICs), respectively. ^32–35 In Fig. 2c, it is possible to observe the respective dQ×dE⁻¹ vs. E plot for the cathode side. The typical redox peak at 3.76 V is present together with three other oxidation peaks at 3.64 V, 4.02 V, and 4.20 V corresponding to the multiphase transition from hexagonal to monoclinic (H1 → M), monoclinic to hexagonal (M → H2) and hexagonal to hexagonal (H2 → H3). ^7,36,37

Graphite is certainly the most used anode material in LIBs and the lithium is inserted in its structure by an intercalation reaction. Because of several factors such as temperature, electrodes structure, electrolyte, charging current, etc, lithium plating can compete with the main Li storage mechanism. ³⁸ This alternative faradaic reaction becomes thermodynamically possible when the anode potential falls below the Li/Li⁺ couple potential (−3.04 V), where Li⁺ ions get reduced preferably on the graphitic surface leading to poor performance and safety issues. This phenomenon is exacerbated especially in the case of low temperatures, electrolytes with poor Li transport properties, ³⁹ high C-rates, and not optimized anode structures. Since the N/P ratio changes with the applied current depending on the capacity retention and the different kinetics of anode and cathode, it would be faster to detect single electrode potentials in a full-cell at different C-rates. This is done by introducing a RE that allows monitoring the anode and cathode potentials individually. Its positioning can be internal or external relative to the electrode stack and despite the investigation of both geometries by several groups, ^21,22,40,41 it is still hard to select the best setup. ¹⁹ To minimize the measurements artifacts, we deliberately used an external reference electrode to avoid shielding errors due to WE surface area coverage by RE ²⁶ and application of inhomogeneous stack pressure. The ohmic drop due to relatively poor conductive organic electrolytes was minimized by placing the RE as close as possible to the stack edge previously aligned to avoid misalignment. Thanks to the near point-like design, the copper wire used as an electric connection to the RE was completely isolated from the electrolyte avoiding mixed surface potentials.

Good cell performance and reproducibility were achieved by the application of an external pressure onto the electrode stack since it can greatly influence the cell performance during formation and cycling steps. ⁴² Materials undergoing big volume expansions during cycling (e.g. silicon) ^43,44 can benefit from the application of specific and tailored pressure. This is also true in the case of traditional materials such as graphite, where a better electrical contact can be obtained. Nevertheless, if excessive mechanical compression is applied especially to anodes and separators, their pores can close ⁴⁵ leading to a decrease of active area and consequently an increase of the current density on the anode surface. A pouch cell holder was designed and used to ensure homogeneous pressure distribution all over the cell surface providing better and more reproducible results. The holder can squeeze out all the gas bubbles potentially formed during electrolyte degradation in the first cycles. When pressure is not applied, gas bubbles can prevent full utilization of the active material, which often leads to lithium deposition in its surrounding. ^28,46 Figure 1b shows its design where a different pressure can be applied simply using a torque wrench to tighten the bolts. In the case of small pouch cells (single- and bi-layer), pressure differences play a bigger role than in the stacked pouch cells. In the latter case, the higher number of layers attenuates the pressure inhomogeneities experienced by single electrodes acting as a buffer. In the following measurements, the pressure distribution is furtherly ensured by soft polymeric pads on both lateral sides of the cells.

The electrode specifications (AM type, AM content, loading, density) were chosen for achieving good performance with high discharge currents (high power). As can be seen from Fig. 3, single-layer pouch cells were discharged up to a rate of 7 C (≈ 19 mAcm⁻²). The discharge steps were conducted in constant current mode leading to remarkable capacity retentions (see Table II). The final discharge step at C/2 (Fig. 3a) confirmed that the high current discharge did not damage the cells allowing for a full recovery of the initial discharge capacity. Electrodes loadings and charge/discharge conditions greatly affect cell performance. Nevertheless, the obtained results proved to be noteworthy, especially, when taking into consideration full- and half-cells studies with the same electrode materials. ^9,12,47

Zoom In Zoom Out Reset image size

To investigate the charging behavior PP3D cells were utilized, where anode and cathode potentials were monitored separately. High-charging current densities can greatly influence battery behavior since different phenomena can occur mainly at the graphite/electrolyte interface. Because of high charging currents and low temperatures, the anode surface potential can drop below 0 V (vs Li⁺/Li) and lithium plating becomes the thermodynamically favorable process over the intercalation. This can often lead to important safety concerns since lithium dendrites can pierce the separator and cause internal short circuits ³⁸ and deposited lithium can react exothermically with electrolyte. ^48–50 Using the PP3D plate, we investigated how our cathode/anode couple behaves in different charging conditions. The charge C-rate protocols denoted in Table SI (Supporting Info) were applied.

Figure 4 shows anode and cathode potentials at different charge C-rates as a function of the state of charge (SOC) at 25 °C and 40 °C, respectively. Most battery producers recommend an operating charging temperature ranging from 0 °C to 40 °C. These values depend on several factors such as electrodes' chemistry, electrolyte type, and cell design. Based on literature data and personal experience, it was decided to use a controlled cycling temperature of 40 °C to stress the cells and partially simulate the thermal behavior of bigger cell formats (stacked pouch cells, prismatic cells, cylindrical cells). ⁵¹ As can be seen from Figs. 4a–4b, at both temperatures the anode end-of-charge potential is progressively shifted toward zero because of the increasing polarization influencing the N/P ratio. When cells were cycled at 25 °C the anode potential reached the lithium deposition limit only when a 2 C current was applied in the CC charging step. After the application of the CV step, the potential increased to around 70 mV as in the other C-rates. This behavior was present in each 2C-charging cycle in different cells. This leads to the hypothesis that despite the negative potential at the end of the CC step, lithium re-intercalation occurred during the CV step. For the sake of clarity, we cannot exclude that extended cycling in these conditions could lead to heavy a more pronounced lithium deposition. As already mentioned, the temperature during charge and discharge plays a fundamental role in the electrodes kinetics. If the electrolyte and the electrodes are able to withstand and do not degrade at such a temperature, then a beneficial effect on the cell performance should be observed.

Zoom In Zoom Out Reset image size

When the same charging protocol was applied at 40 °C, the rate capability improved, and as expected the temperature contribution was particularly relevant at higher C-rates. The increase in Li⁺ ions mobility due to the electrolyte lower viscosity had a beneficial effect on the cell performance. The SOC reached right after the CC step in charge (Table III) increased from 86.2% to 88% at 1 C (+ 2%) and from 74.6% to 81.3% at 2 C (+ 9%). Potentials shifts were observed at the cathode side as well (Figs. 4c–4d). As it can be seen clearly from Fig. 5, the anode potentials at the end of the CC charge steps at 40 °C (Fig. 5b) are higher than those at 25 °C (Fig. 5a). These more positive potential values are the result of the smaller polarization associated with the temperature increase.

Zoom In Zoom Out Reset image size

To confirm the obtained results, we tested single-layer pouch cells. Long-term cycling tests were carried out at different combinations of temperature and charging C-rate. Their results are shown in Fig. 6. As already mentioned above, the cycling temperature is important since it can directly affect the lithiation/delithiation kinetics, and based on the previous result, a symmetrical 1 C/1 C cycling at 25 °C should have not showed a fast capacity decay due to Li plating. 2 C charging at 25 °C showed an anode potential slightly below 0 V, followed by an anode potential increase during the CV step (Fig. 5a). Lithium deposition does not lead always to dendrites and electrically isolated lithium ("dead" lithium) formation. Depending on the conditions, these lithium deposits can re-intercalate in the graphite. ^52–55 This reversible Li deposition usually does not lead to a large amount of capacity loss through cycling, however, it can be anyway a safety concern ⁴⁸ and should be avoided when possible. The results obtained in Fig. 4a suggest that perhaps we are dealing mostly with a reversible form of lithium deposition since no plateau is observed below 0 V and in the CV step the potential increases rapidly to ≈70 mV.

Zoom In Zoom Out Reset image size

At the same time, other cells were cycled at a rate of 2 C at 40 °C. As shown in Figs. 4b and 5b, the anode potential in these pouch cells never dropped below 0 V. If the electrolyte and the NMC811 cathode would not degrade because of the cycling temperature, better results could have been expected compared to the pouch cells cycled at 25 °C. Several studies point out that temperatures higher than 30 °C can accelerate NMC-based cathodes degradation. ⁷ Moreover, electrolyte decomposition products such as HF can trigger NMC dissolution. ⁵⁶ The dissolution of transition metals can pose a problem also to the anode since they can be included in the SEI structure worsening cell stability and cycling life.

After long-term cycling, the SLPCs were discharged at 2.7 V and disassembled in a glove box. In Fig. 7, it is possible to observe anodes and separators appearances after they reached 80% state of health (SOH). As predicted by the previous experiments (Figs. 4 and 5), pouch cells cycled with a charge rate of 1 C at 25 °C (Fig. 7a) and with a charge rate of 2 C at 40 °C (Fig. 7c) did not show any presence of dead lithium on their surface. When the temperature is raised to 40 °C better performance was expected due to an improvement of the intercalation kinetics, however the thermal stability of the electrolyte must be taken into consideration as well. Since the temperature was applied continuously throughout the cycling procedure, this can have affected the electrolyte stability resulting in faster cell aging. SLPCs cycled at 40 °C (Fig. 6) reached 80% SOH earlier than the cells cycled at 25 °C and no clear sign of gas evolution was observed. A sign of the electrolyte degradation can be seen in Fig. 7c where the polymeric separator showed local black areas and brownish spots. Taking a closer look at Fig. 7b, it is possible to notice a small "dead" lithium spot near the tabs. During the CV charging step, this region provides the shortest path for charge transport phenomena, therefore is more prone to be overcharged. This behavior was already detected in the PP3D cells where the anode potential was going slightly below 0 V. Nevertheless, the pouch cells cycled in these conditions showed comparable performance to those cycled with a 1 C charge current (see Fig. 6). The negative potential in the first cycles probably led to "dead" lithium that remained in that position without affecting further cycles.

Zoom In Zoom Out Reset image size

In order to assess the presence of deposited lithium by GD-OES depth profiling, it is assumed that the measured amount of lithium is composed of lithium in the SEI and metallic lithium, as defined by Ghanbari et al. ²⁹ The cell has been completely discharged, therefore the amount of intercalated lithium is negligible.

$$\begin{eqnarray}&&L{i}_{detected}=L{i}_{metallic}+L{i}_{SEI}\end{eqnarray} \tag{ 2 }$$

Furthermore, it is presumed, that the SEI consists only of lithium oxide (Li₂O). Li₂O has the highest Li-content (46.5 wt %) of all known SEI species, therefore this assumption gives the minimum amount of metallic lithium. The detected oxygen is used to calculate the amount of Li₂O in the SEI. If Eq. 2 gives a positive value for Li_metallic, Li deposition has been detected. A negative value means that all Li is bound in the SEI. ²⁹ Figure 7 shows all the GD-OES results performed at the center of the anode for each cell (white dotted circle in Fig. 7). Using this calculation, none of the investigated samples have shown lithium deposition. In Fig. 7a, the cell, which has been charged with a 1 C current at 25 °C shows a lithium gradient throughout the depth of the electrodes. This result is consistent with previous findings ⁵⁷ showing that the SEI is more pronounced at the electrode surface. In Fig. 7b is shown the GD-OES result from the area highlighted by the white dotted circle as in the other two anodes. Nevertheless, since a small lithium deposition was visible, a sample was harvested from the red square and analyzed to confirm the presence of metallic lithium. The GD OES results from this area can be seen in Fig. 8a, where the Li distribution shows a plateau at the anode surface with a higher mass percentage than the other analyzed samples in Fig. 7. Furthermore, Eq. 2 was applied and the calculated Li_metallic weight percentages have shown positive values only in this sample. The comparison between the calculated Li_metallic weight percentages of these different samples is presented in Fig. S2 (Supporting Info). The cell charged with a 2 C current at 40 °C (Fig. 7c) shows a steady Li content, which could originate from the electrolyte decomposition at elevated temperatures that leads to an enhanced SEI formation. ^58,59

Zoom In Zoom Out Reset image size

GD-OES analysis was also used to assess the binder distribution. Slurry preparation, electrodes coating, and the drying step greatly influence the electrode homogeneity. In particular, binder migration toward the electrode surface can occur when high thermal gradients are present during electrode drying. ⁶⁰ This phenomenon can lead to poor mechanical stability together with poor electrochemical performance of the electrode. By the detection of sodium which is present in the Na-CMC, the binder distribution was monitored. The results can be seen in Fig. 8b. The homogeneous sodium distribution throughout the whole pristine anode coating is shown by the flat course of the Na intensity. For the sake of comparison, an aged anode coating was analyzed as well. The Na distribution resulted to be the same as in the pristine anode. Since it was not detected a binder gradient, these results are the index of a successful anode drying step that resulted in a homogeneous binder distribution. The good performance in the C-rate discharge tests is also supporting these findings.