Influence of Low Pressures on the Performance of Lithium Ion Batteries for Airplane Applications

Different kinds of lithium ion batteries exist which differ in the materials used. The battery behavior is mainly dependent on the positive electrode (often referred to as cathode material). For aviation NMC (lithium nickel manganese cobalt oxide) and NCA (lithium nickel cobalt aluminum oxide) batteries are promising due to their high specific energy. ²⁵ On the other hand, LFP (lithium iron phosphate) cells are promising because of their high operational safety and high power capability. ²⁶

Not only the materials but also the design defines the battery behavior. For aviation cylindrical cells are considered suitable since they have a commercial standard and the highest energy density on cell level compared to others. ²⁷ However, on pack-level they have a high volume need. Heat dissipation within the cell is low due to the compactness but the space in between the cells can be used for cooling. The separation of the cells and the space in between also means that thermal runaway cannot spread easily. Pouch cells are also a promising cell format since they have the highest specific energy, are very flexible in form and have a high mechanical safety. ²⁷ However, preventing the spread of thermal runaways from one cell to the other is difficult. Pouch cells also require robust pack housing which lowers the specific energy on pack level. Prismatic cells are not considered useful for aviation since the specific energy and energy density are lower than for any of the other cell format types. ²⁷

The battery types examined in this project are presented in Table I. These batteries were chosen since they are specifically made for aerospace applications and have the NCA, NMC and LFP cathode materials. Since there are two cylindrical and one pouch cell comparing cell formats is also possible.

To examine the dependency of battery behavior on pressure, cycling tests were performed as well as impedance measurements by electrochemical impedance spectroscopy (EIS), which is a typical battery measurement and has been used and improved over decades. ²⁸ Charging and discharging of the battery was performed using a BasyTec XCTS cycler with a current range of +/− 25 A and 12 channels (accuracy +/− 2.5 mV and 50 mA). Due to the 12 channels, all battery cells can be measured in parallel. Impedance measurements were performed using a Zahner Zennium Pro device (Zahner-Elektrik GmbH, GER). Cycler and impedance spectroscope were connected to the battery via a BC-Mux multiplexer with a switch box. Each battery was connected with a four-point measurement by four twisted cables directly in case of the cylindrical cells or over a Gamry cell holder in case of the pouch cells. A dll (dynamic link library) driver was programmed to be able to combine cycling and electrochemical impedance spectroscopy and automatize the tests as much as possible. A schematic of this measurement set-up can be seen in Fig. 1.

Zoom In Zoom Out Reset image size

To control temperature and pressure, the batteries were placed in a combined climate and low-pressure chamber. ³⁰ Each pressure level was reached slowly in the range of half an hour to a few hours to be in a stable pressure state and let the batteries adapt to each pressure before the electrical tests started. The chamber temperature was set to 23 °C. The batteries were placed within two safety boxes inside the low-pressure chamber. There were no restraints around the cells. There was no cooling control on the cells but the low-pressure chamber generated some airflow. All electric connections (current and voltage cables as well as temperature, humidity and pressure sensors) were fed through chamber inlets into the safety boxes. Cell temperature was measured by an NTC thermistor of type B57861S502F40 (TDK electronics, GER) fixed on the upper surface of each battery. Environmental humidity and temperature were measured by external sensors type KS-CAN03 (Zila GmbH, GER) and the pressure with sensors type PT0517 (ifm electronic GmbH, GER). The data of the environmental conditions was transmitted via CAN and logged by a Data logger of type GL2400 (Vector Informatik GmbH, GER).

To ensure that all cells can be operated in low pressure, a withstand test was performed first where the pressure was lowered in steps from ambient pressure (95 kPa) over 75 and 50 kPa to 25 kPa. At 25 kPa the batteries rested for around 1 h before slowly increasing the pressure again to ambient pressure. After the test, the cells were examined for any visible damage or changes. Since none of the cells showed any leakages or sign of change, the electric behavior tests were performed afterwards on all cells.

To investigate the influence of the pressure on the DC (direct current) behavior of each battery, constant current (CC) discharges were performed in the low-pressure chamber. The cycler discharged each battery with a constant current corresponding to a specific C-rate which describes a current rating in reference to the rated cell capacity.

The tests were performed at 1C, 0.5C and 2C (1.25C for the LFP cells since 25 A is the maximum current for the cycler) within the specified voltage limits for each battery for atmospheric pressure (around 95 kPa), 75, 50 and 25 kPa at 23 °C. Charging was performed at 0.5C for all tests.

In order to obtain additional data on cell behavior at different pressures the impedances were measured at 23 °C for the different pressures. The impedance measurements were performed in galvanostatic mode with 0 A DC-current in 10% SOC (State of Charge) steps from 90% to 10% SOC. The SOC was determined using Eq. (1). The SOC boundaries of 90% and 10% were chosen to avoid overcharging or deep discharging the batteries during the low frequency impedance measurement. The discharges between the impedance measurements were performed at 1C. The frequency for each impedance spectroscopy ranged from 10 kHz to 10 mHz in single sine mode with 4 steps per decade and 4 measurement periods per frequency. The AC current was set according to the behavior of each battery type. The AC current was chosen as low as possible to assure low harmonics but high enough to measure the voltage response. If both requirements are fulfilled, the battery behavior can be assumed to be linear, time-invariant and causal. The AC current determined for the NMC battery for impedance spectroscopy was 200 mA, for the LFP 700 mA and for NCA 300 mA.

$$\begin{eqnarray}&&SOC=\left(1-\displaystyle \frac{{C}_{discharged}}{{C}_{rated}}\right)* 100 \% \end{eqnarray} \tag{ 1 }$$

As reference, the pressure was set to atmospheric pressure first (Atm. 1). The pressure was then lowered stepwise to 75 kPa, 50 kPa and 25 kPa. Afterwards, one or even two reference measurements were performed at atmospheric pressure again (later on called Atm. 2 and Atm. 3) to check for any degradation. For each pressure the tests were performed in the following sequence: discharge curves were taken for 1C, 0.5C and 2C (1.25C for LFP). Afterwards impedance measurements were done starting at 90% SOC and subsequently reducing the SOC to 10%. Each pressure level was kept constant for the complete measurement set time of 2 d. After those tests, the pressure was changed to the next pressure level and the electric measurements were repeated at this new pressure as can be seen in Table II.

The discharge curves for one measured NCA cell for a C-rate of 1 at atmospheric pressure as well as for 75, 50 and 25 kPa are shown in Fig. 2. It can be seen there is hardly any difference in the discharge behavior in the linear region around 20% to 95% SOC. At high and medium state of charge (SOC) a slight variation in voltage can be seen and at the end of discharge a difference in discharge capacity was observed. The maximum difference in voltage at high and medium SOC between the different pressure levels that was observed was 13 mV. The observed total discharged capacity (considered when the cut-off-voltage of 3.0 V is reached) differs only by 0.14 Ah between the atmospheric and low-pressure curves. Moreover, the maximum variation in voltage measured between atmospheric and low-pressure curves among all NCA cells during different C-rate tests was 15 mV. The maximum variation in total discharge capacity was 0.24 Ah.

Zoom In Zoom Out Reset image size

The observed deviation in voltage value might be attributed to differences in cell temperature as can be seen from the bottom plot in Fig. 2. A higher cell temperature decreases the overpotential resistance and therefore a higher temperature leads to a higher cell voltage. ^33,34 Despite the cell temperature being allowed to stabilize before the measurements were started, the starting temperatures for the measurements varied, due to the experimental setup. The starting temperature for the first atmospheric measurement was 20.6 °C, while all consequent measurements were between 22.5 °C and 24.2 °C. This temperature difference plus a possible formation of the fresh cells might explain the slightly lower voltage of the first atmospheric curve. The formation can also be seen if the two reference curves are compared. The cell voltage and capacity increase between the two reference measurements afterwards although the temperature and the pressure do not change. The differences in total discharge capacity and increase in voltage after the first cycle at atmospheric pressure also indicate formation of the cell since the discharge capacity increases slightly with the order of measurements. None of the differences can therefore be attributed to a pressure effect. This behavior was found for the discharge curves at 0.5C, 1C and 2C for all three NCA cells.

Figure 3 shows the total discharge capacity for the measurements at different pressures plotted against the C-rate with error bars reflecting three measurements of NCA cells. It can be seen, that the discharge capacity slightly increases with the order of the measurements from the first measurement at atmospheric pressure to the reference measurements at atmospheric pressure, indicating a formation of the cell. It can also be seen that the influence of the C-rate on the discharge capacity is much more pronounced than the variation between the measurements at different pressures.

Zoom In Zoom Out Reset image size

To further examine the pressure related behavior, impedance spectroscopy was performed on the NCA cells in 10% SOC steps from 90 to 10%. The Nyquist plot for one NCA cell at 10% SOC can be seen exemplarily in Fig. 4 on the left-hand side. The negative imaginary part of the impedance, calculated from the excitation sinusoidal current and the sinusoidal voltage response, is plotted on the y-axis and the real part of the so calculated impedance is plotted on the x-axis. Therefore, the high frequencies are on the left hand-side, the low frequencies on the right hand-side.

Zoom In Zoom Out Reset image size

In the Nyquist plot, it can be seen that the ohmic resistance at the x-axis intersection hardly varies. The difference of the ohmic resistance is less than 10 μΩ and does not show a pressure dependency.

In the medium to low frequency range, which includes the two semicircles and the diffusion branch, a difference between the measurements can be observed. The right semicircle varies between each measurement. The peak of the right semicircle decreases from (2.58–0.80j) mΩ (Atm. 1) to (2.17–0.61j) mΩ (Atm. 2). The semicircle decrease also leads to a shift of the diffusion branch to the left hand-side. This means, the real part impedance values measured at the same frequency decrease with each measurement. This area typically shows changes in the charge transfer and double layer capacitance and is strongly affected by temperature and SOC. ^35–38

The cell temperatures during the impedance measurements can be seen in the legend. Even though all measurements were within 2 degrees Celsius of each other, the atmospheric measurements before and after the low-pressure tests had the lowest and the highest temperature. It can be seen in Fig. 4 on the left-hand side that the curves for all low-pressure tests are located between the two atmospheric curves and they are in order of temperature. It can therefore be deduced that the observed variation is dominated by temperature and not pressure. To assess the influence that a variation of temperature of 2 degrees Celsius has on the impedance spectrum, a measurement at 22.9 and 21.3 °C is shown on the right-hand side of Fig. 4 for atmospheric pressure. This shows that a temperature difference of about 2 degrees Celsius affects the right semicircle in the range that was observed for the impedance measurements at 75 kPa, 50 kPa, 25 kPa and the second atmospheric measurement. The first atmospheric measurement (Atm. 1) showed a higher impedance. Since this was the first measurement taken it can be attributed to a formation of the cell. This was already observed in the discharge curves. All NCA cells showed this general behavior for all SOC levels.

Neither the discharge curves nor the impedance measurements showed any pressure dependency of the battery performance for the cylindrical cell with NCA cathode.

The discharge curves for one measured LFP cell for a C-rate of 1 at atmospheric pressure as well as for 50 and 25 kPa are shown in Fig. 5. It can be seen there is hardly any difference in the discharge behavior in the linear region around 20% to 95% SOC. At high and medium state of charge (SOC) a slight variation in voltage can be seen and at the end of discharge a difference in discharge capacity was observed. The maximum difference in voltage at high and medium SOC between the different pressure levels that was observed was 11 mV, which was also the maximum among all LFP measurements. The observed total discharge capacity (considered when the cut-off-voltage of 2.5 V is reached) differs only by 0.13 Ah between the atmospheric and low-pressure curves.

Zoom In Zoom Out Reset image size

The starting temperature for the first atmospheric measurement was 24.1 °C, while all consequent measurements were between 24.9 °C and 25.5 °C. This might explain the slightly lower voltage of the first atmospheric discharge curve that is shown in the zoomed part in Fig. 5. At the first glance the variation among different discharge curves seems to show a slight pressure dependency. However, the temperature differences during the measurements that can be seen in the lower part of Fig. 5, indicate that these differences are caused by temperature rather than pressure. The lowest discharge capacity was measured for atmospheric conditions which also had the lower temperatures, whereas the highest discharge capacity was measured at 50 kPa which had comparatively high temperature. The maximum difference in discharge capacity for all tests at low and atmospheric pressure was about 130 mAh, which is slightly higher than the system accuracy of ± 50 mAh considering ± 50 mA current accuracy and a measurement time of around 1 h.

Figure 6 shows the medium total discharge capacities for the C-rates of 0.5C, 1C and 1.25C of the LFP cells with error bars reflecting three measurements of three LFP cells. However, it has to be considered that the temperatures were slightly different for the three batteries due to the measurement set-up which leads to large error bars that could be over-interpreted. The maximum variation in discharge capacity is 0.34 Ah at 1.25C discharge.

Zoom In Zoom Out Reset image size

The results with respect to the capacity showed a big variation between cells, especially if compared to the NCA cells. In addition, the order of the variations differed for the different cells, so no clear dependencies could be shown. Therefore, a larger database would be required and the possibility of production variations should be examined as well. Despite of this, the measurements show that pressure does not have a significant influence on the cell capacity of the LFP cells observed in this study.

Figure 7 shows the Nyquist plot of the impedance measurements of an LFP cell at an SOC of 10% at the different pressures. The cell temperatures during the impedance measurements were between 22.2 °C and 23.5 °C which can be seen in the legend.

Zoom In Zoom Out Reset image size

From Fig. 7, it can be seen that the impedance has only one visible semicircle, which is according to literature. ³⁹ This semicircle does not change significantly with pressure. The maximum deviation in the imaginary part is below 20 μΩ and does not show any pressure dependency. The intersection with the x-axis, which defines the ohmic resistance, shows a maximum deviation of 10 μΩ and does also not show any pressure dependency either. The impedance results of the other LFP cells and at the other SOCs showed a similar behavior and therefore no pressure influence was observed.

Neither the discharge curves nor the impedance measurements showed any pressure dependency of the battery performance for the cylindrical cell with LFP cathode.

Figure 8 shows the discharge curves for one of the NMC pouch cells at different pressures for a C-rate of 1. The measurements show a difference in voltage at high and medium SOC.

Zoom In Zoom Out Reset image size

The maximum difference in voltage observed for the 1C-rate was 39 mV, which is about three times higher than for the other cell types. Since the voltages for the atmospheric reference curves that were taken after the pressure variations are higher than the voltages for the initial atmospheric measurement, some formation of the cell can be assumed. The highest voltage was observed for the lowest pressure. This is probably due to the slightly higher temperature combined with the ongoing formation of the cell, rather than pressure.

It can also be observed from Fig. 8 that the total discharge capacity (when the cut-off-voltage of 2.7 V is reached) differs by 0.26 Ah between the atmospheric and low-pressure curves. This value is about twice as high as for the cylindrical cells. This difference can in part be attributed to a formation of the cell (when comparing the atmospheric measurements before and after the low-pressure tests) or temperature, since the temperature for the low-pressure tests was slightly higher (about 3 K on average) than for the atmospheric measurements.

The same basic behavior was observed for the other NMC cells and the other C-rates. Figure 9 shows that neither the total discharge capacity for different C-rates, nor the variation in capacity among the different pressure tests, changed significantly. It has to be considered that the temperatures were slightly different for the three batteries due to the measurement set-up which leads to large error bars that could be over-interpreted. The NMC cell capacity is almost constant over the different C-rates for all pressures. The NMC cells heated most during the tests, especially at higher currents. The higher temperature at higher C-rates cancels out a current dependent capacity decrease. A direct pressure dependency is not visible in the capacity of the NMC cells. The higher capacities at lower pressures are determined by higher temperatures.

Zoom In Zoom Out Reset image size

Figure 10 shows the Nyquist plots of the impedance measurements at 10% SOC (above) and 20% SOC (middle) of the NMC cell at different pressures since those two SOC levels show pressure influence. At higher SOC the influence of pressure was less visible in the impedance spectra which can be seen in the bottom graph of Fig. 10 for 90% SOC. The temperatures during those measurements varied between 21.2 and 23.4 for different pressure levels but were similar for both SOCs with a maximum difference of 0.2 °C at each pressure level as shown in Table III.

Zoom In Zoom Out Reset image size

It can be seen in Fig. 10 that the ohmic resistance (given by the intersection of the impedance curve with the x-axis) changes. The ohmic resistance increases with decreasing pressure. The measurement at ambient pressure after the low-pressure test shows a decreased ohmic resistance compared to the 25 kPa measurement but the resistance is still higher than for the previous measurements. This trend was observed for all SOCs and is shown exemplarily for 10, 50 and 70% SOC in Fig. 11.

Zoom In Zoom Out Reset image size

Figure 11 also shows that temperature is not the reason for this change in ohmic resistance, since ohmic resistance would be expected to drop with increasing temperatures. The ohmic resistance increased on average by around 0.1 mΩ per 25 kPa for the pressure decrease to 75, 50 and 25 kPa. The reference measurement at atmospheric conditions that was taken after the low-pressure tests showed that this change in ohmic resistance was at least to some extend permanent. That the cycling of the cell alone was the reason for the increase in ohmic resistance can be excluded, since the atmospheric measurement after the low-pressure tests showed again a slightly lower resistance for all cells and SOCs examined. On top of that, measurements on an additional cell showed no clear increase pattern when three consecutive charge discharge impedance measurements at 23 °C were performed.

All three measured NMC cells showed the same increase in ohmic resistance during the low-pressure tests. It is possible that the low pressure causes the pouch cell to swell, which could lead to some delamination and worsening of contacts within the cell itself. Swelling of the examined cells was not visually observed with the naked eye. However, Wang et al. ⁴⁰ and Uno et al. ⁴¹ found swelling of pouch cells at very low pressures of 20 Pa. It is therefore possible that some swelling starts at the examined pressures. Since the possibility to swell is the main difference between the cylindrical and the pouch cells, it could explain the differences observed between pouch and cylindrical cells. To examine and prove this effect more research is needed with a high precise displacement measurement set-up.

Figure 10 also shows a variation in the two semicircles that can be seen in the Nyquist plot. They are usually associated with charge transfer processes and the double layer capacitance. ³⁵ The left semicircle in Fig. 10 decreases from the first atmospheric measurement over the low pressures to the second reference curve at atmospheric pressure. It therefore shows a dependence on number of cycles rather than pressure itself.

The change in the right semicircle is more pronounced for the lower SOC (10%) than for 20% SOC. This right semicircle is typically strongly temperature dependent and decreases in size for higher temperatures. ^36–38 However, the changes seen in Fig. 10 cannot be attributed to temperature since the temperature was highest for the measurement at 25 kPa, which shows the largest semicircle. It can be seen from Fig. 10 that the size of this right semicircle decreases when pressure is lowered from ambient to 75 kPa. There is hardly any change between the semicircle at 75 kPa and 50 kPa, but the semi-circle then increases strongly when the pressure is lowered further to 25 kPa. The right semicircle at 25 kPa has its peak at (9.50–2.89j) mΩ compared to Atm.1 with (9.21–2.55j) mΩ and all other measurements with around (8.50–2.30j) mΩ. Between the three cells the 25 kPa highest point changes most with 1.8 mΩ for the real part and 1.3 mΩ for the imaginary part. The other pressure values varied by less than 1 mΩ. Figure 10 shows the cell with the median behavior. The semicircle increase during 25 kPa also leads to a shift of the diffusion branch to the right hand-side. The right semicircle of the second atmospheric measurement, which was taken after the low-pressure tests (Atm. 2), is lower than measured at 25 kPa. This indicates that there is a reversible pressure influence at 25 kPa. This semicircle of the second atmospheric measurement is also similar to 75 kPa and 50 kPa, but lower than the initial measurement at atmospheric pressure, which further indicates a permanent change to the cell due to the low pressures.

In the impedance spectra for SOC of 20%, the increase in the right semicircle at 25 kPa is not as pronounced as for 10% SOC, but still visible. The described behavior of the ohmic resistance and the semicircles in the impedance spectra at low SOC was seen for all three NMC cells.

The Pouch cells with an NMC cathode were the only batteries tested whose performance showed a slight pressure dependency. The changes observed in the total discharge capacity (variation by 0.26 Ah) were about twice as high as for the cylindrical cells with different cathodes. The impedance measurement showed an increase of ohmic resistance with pressure and for low SOC (10 and 20%) also an influence on the right semicircle in the Nyquist plot.