Introduction

Reliability and safety problems arising from excessive heat generation are one of the most critical challenges inherent to lithium-ion batteries [1,2,3]. However, striking progress has been made concerning thermal issues in lithium-ion batteries based on comprehensive scientific efforts and a deep understanding of heat generation mechanisms [4,5,6,7,8]. In recent years, the scientific and economic interest in sodium-ion batteries has continually increased because the limited and unevenly distributed lithium resources may become a critical issue for future lithium-ion batteries [9,10,11,12]. Furthermore, sodium-ion batteries are potentially cheaper and more environmentally friendly compared to lithium-ion batteries. The tremendous scientific insights that led to significant progress in lithium-ion technology in the past may be highly useful in accelerating the development of sodium-ion batteries, especially concerning safety and thermal issues. Consequently, comparative studies of Na/Li-analogous materials and electrodes are urgently needed to explore their similarities and differences, and to identify approaches for optimization [13].

In this study, the heat generation in NaFePO4 (NFP) electrodes for future sodium-ion batteries and LiFePO4 (LFP) electrodes for lithium-ion batteries is investigated. Assuming negligible heat of mixing [14] and no phase changes except crystalline phase transitions, the local volumetric heat generation rate, \( \dot{q} \), is the sum of irreversible and reversible heats [15, 16]:

$$ \dot{q}={\dot{q}}_{\mathrm{irr}}+{\dot{q}}_{\mathrm{rev}}=\frac{I\eta}{V}-\frac{I T}{V}\frac{\partial {E}_{\mathrm{eq}}}{\partial T} $$
(1)

where I, η , T, and ∂E eq /∂T being the electric current related to the volume V, the overpotential, the absolute temperature, and the temperature coefficient.

The magnitude of irreversible heat generation is connected to limitations of the insertion reactions, which arise from ohmic resistance and the charge transfer at the solid/electrolyte interface as well as diffusion of the electroactive species in the solid host material and the liquid electrolyte. Under load, these limitations cause a potential drop (overpotential) across the electrode, resulting in Joule heating of the cell.

The reversible heat is determined by the entropy change due to the electrode reaction. Charge transfer and diffusion kinetics as well as thermodynamics depend on the design parameters of the electrode, as, e.g., particle size (distribution) [17,18,19], electrode design [20, 21], and active materials coating [22, 23]. Consequently, the identification of intrinsic similarities and differences of the heat generation rates in NFP and LFP electrodes requires identical design parameters.

In this study, NFP electrodes are prepared by an electrochemical cation displacement technique using LFP electrodes as the starting material. Consequently, the NFP and LFP electrodes are identical regarding particle size (distribution), active material loading, binder, conductive additives, carbon coating, etc. In this case, all differences in the heat generation rates are exclusively determined by the substitution of the alkali metal cation. Irreversible heat generation rates of NFP and LFP electrodes are computed from galvanostatic intermittent titration technique (GITT) measurements at different C-rates. Reversible heat generation rates are determined from the temperature coefficient, ∂E eq/∂T, related to the entropy of the electrode reactions at different compositions. Finally, total heat generation rates are computed. The results are discussed regarding qualitative and quantitative differences in the heat generation in sodium- and lithium-ion batteries.

Experimental and methods

Commercial electrodes were received from MTI Corp. and consist of carbon-coated active material (LiFePO4 = LFP) admixed with binder (polyvinylidene fluoride) and conductive additives (carbon black) on a 15-μm aluminum foil. Pristine electrodes were carefully examined by scanning electron microscopy. The mean particle size, the total thickness, and the porosity are determined to approximately 400 nm, 100 μm, and 0.3. Electrochemical experiments were carried out in a three-electrode arrangement using T-type Swagelok® cells assembled in an argon-filled glove box (< 5 ppm H2O, 10 ppm O2). In each case, a pristine LFP foil was the working electrode. The counter electrode consisted of either metallic sodium or lithium (99.9% trace metal basis ALDRICH ® Chemistry). The reference electrode was a wire of either metallic sodium or lithium with a diameter of approximately 0.5 mm. The reference electrode was placed between two layers of a porous borosilicate glass fiber (Whatman, GF/C) acting as the separator. The electrolyte was either high purity 1 M NaPF6 or LiPF6 (Alfa Aesar) both solved in a 1:1 wt% ratio mixture of ethylene carbonate and diethyl carbonate (ALDRICH ® Chemistry).

NaFePO4 electrodes were prepared with an electrochemical displacement method described in detail in Ref [24]. In brief, LiFePO4 electrodes were charged to remove the lithium and to create a vacant host structure for sodium insertion/extraction. Lithium extraction and subsequent sodium insertion were carried out by cyclic voltammetry between 2.0 and 4.0 V at a low scan rate of 0.05 mV s−1. The formation of NaFePO4 was verified by X-ray diffraction (XRD) [24].

NFP and LFP electrodes were cycled at a rate of C/10 until the discharge capacity was stable in a range of ± 2%. Subsequently, the electrodes were investigated by galvanostatic intermittent titration technique (GITT). Starting from NFP or LFP electrodes of known composition and in electrochemical equilibrium, a constant current is applied to the cell and the potential response is measured. After a certain time, t on, the current is interrupted, and concentration gradients within the solid electrode and the liquid electrolyte relax by the diffusion of the mobile species. This process is associated with a drift of the electrode potential towards a new steady state value. The procedure is repeated until a defined cutoff potential is reached. GITT measurements were performed at different C-rates varying from 0.1 to 2.0 C. The t on-time of a single current impulse in the GITT procedure was adjusted to the C-rate with the objective of maintaining the same change in composition during each pulse and to improve the comparability between the different C-rates. The relaxation periods, t off, between the current impulses lasted 3 h. To determine the temperature coefficient, ∂E eq/∂T, connected to the reversible heat (cf. Eq. 1), a further GITT measurement was carried out using a C-rate of 0.1 C. After each charge/discharge step, the equilibrium potential was measured at four temperatures until stable conditions ensued (∂E eq/∂T < 2 mV/h).

Results and discussion

Reversible heat

The reversible heat results from the entropy change due to the electrode reaction. It has been shown to significantly contribute to the overall heat generation in lithium-ion batteries [25, 26]. If the entropy increases (positive temperature coefficient), the corresponding amount of energy is consumed from the thermal energy of the system resulting in a cooling effect. Vice versa, a decrease in entropy (negative temperature coefficient) causes an exothermic heat effect.

Figure 1 compares typical temperature dependent measurements of the equilibrium potential, E eq, of the NFP and LFP electrodes. For both materials, E eq increases with decreasing temperature at the composition, Na0.8FePO4/Li0.8FePO4. This behavior indicates negative temperature coefficients, ∂E eq/∂T. The equilibrium potential slightly decreases at a constant temperature of 10 °C. This behavior might be attributed to slightly unintended variations of the ambient temperature or self-discharge of the electrode. However, the magnitude of this effect is very small in comparison with the impact of the intended temperature variations and does not influence the determination of the temperature coefficient. Steady state values of E eq are extracted for the respective temperatures and are also plotted in Fig. 1. As expected, a clearly linear relationship is revealed. At this specific composition, the temperature coefficient, ∣∂E eq/∂T∣, is larger for NFP compared to LFP.

Fig. 1
figure 1

Typical temperature dependent measurements of the equilibrium potential at a composition of x = 0.8 in NaxFePO4 and LixFePO4 and steady state data extracted from the measurements

Full size image

Figure 2 shows the composition dependence of the temperature coefficient of the NFP and LFP electrodes obtained after charging and discharging steps. The temperature coefficients of both materials show some identical features. At low x, ∂E eq/∂T is positive and decreases with increasing x, passing into a flat composition dependence around medium x. At high x, ∂E eq/∂T is clearly negative. The composition dependence of ∂E eq/∂T is caused by changes of the entropy of intercalation with the sodium/lithium concentration x in NaxFePO4/LixFePO4. For example, at high and low x, the change in configurational entropy by putting an additional amount of sodium/lithium in the host lattice is much larger compared to medium compositions [27]. Beside these common features, the temperature coefficients of NFP and LFP electrodes differ in their absolute values, composition dependence, and charge/discharge symmetry. Assuming reversible electrode reactions, the temperature coefficients obtained after charging and discharging steps at the same composition were expected to be identical [28]. This is in good accordance with the behavior observed for LFP. However, in the case of NFP, distinct differences are found at the same formal composition, depending on whether NFP was charged or discharged previously to the measurement. Particularly, in a range of approximately 0.1 < x < 0.6, the temperature coefficient is larger for previous charging steps but significantly lower around x = 0.7. This behavior is attributed to asymmetric phase transformation dynamics resulting in a path dependence of the sodium insertion-extraction kinetics recently reported in the literature [29,30,31], see Ref [32] for details.

Fig. 2
figure 2

Composition dependence of the temperature coefficient, ∂E eq/∂T, of (a) NFP and (b) LFP electrodes obtained after charging and discharging steps

Full size image

Reversible heat generation rates are computed from the determined temperature coefficients according to:

$$ {\dot{q}}_{\mathrm{rev}}=-\frac{IT}{V}\frac{\partial {E}_{\mathrm{eq}}}{\partial T}. $$
(2)

Figure 3 shows color-coded contour plots comparing the C-rate and composition dependence of the reversible heat generation rates of NFP and LFP electrodes. The contour plots here and hereinafter are based on 270 data points (9 different C-rates, 30 equidistant SoCs). The C-rate is positive and negative for charging and discharging, respectively. As predefined by Eq. 2, the reversible heat generation rates increase linearly with increasing C-rate at a given SoC. Both NFP and LFP show an endothermic effect at the beginning of charging, passing into an exothermic heat effect at high SoC, via a flat composition dependence at medium compositions. During discharging, the reverse reaction takes place accompanied by an endothermic effect at high SoC passing into an exothermic heat effect that gradually increases toward low SoC. For both materials, the largest values of \( {\dot{q}}_{\mathrm{rev}} \) are found at the end of discharge. The lowest values of \( {\dot{q}}_{\mathrm{rev}} \) are observed at the beginning of charging whereby the endothermic effect spans a wider SoC range in the case of LFP. A distinct symmetry between the charging and discharging is found in the case of LFP, whereas \( \mid {\dot{q}}_{\mathrm{rev}}\mid \) is partly different for the charging and discharging of NFP at the same SoC. This is reasonable according to the composition dependence of ∂E eq/∂T obtained after charging and discharging steps (cf. Fig. 2 and Eq. 2).

Fig. 3
figure 3

Color-coded contour plots comparing the C-rate and composition dependence of the reversible heat generation rates of NFP and LFP electrodes

Full size image

Table 1 compares the mean values and extrema of the reversible heat generation rates of NFP and LFP at the 1.0 C rate. The results indicate that the average reversible heat generation in NFP is smaller in comparison with LFP. Particularly, during discharging, the average \( {\dot{q}}_{\mathrm{rev},\mathrm{LFP}} \) is four times larger compared to the average \( {\dot{q}}_{\mathrm{rev},\mathrm{NFP}} \).

Table 1 Key data of reversible, irreversible, and total heat generation rates of NFP and LFP electrodes at the 1.0 C rate
Full size table

Irreversible heat

Figure 4 presents typical time-dependent potential responses within the GITT measurements obtained from the NFP electrode. An abrupt increase of the potential is observed at the beginning of the t on-time, caused by the ohmic resistance of the electrode and the charge transfer resistance at the solid/electrolyte interface. Subsequently, a further increase of the electrode potential is observed that is typically ascribed to solid-state diffusion in the host lattice [33]. When the current is interrupted, an abrupt decrease of the potential is observed followed by an exponential decay towards the new equilibrium potential, E eq(x).

Fig. 4
figure 4

Typical time dependent potential responses within the GITT measurements obtained from the NFP electrode. Pulse and relaxation periods are indicated as well as the data that are extracted to calculate the overpotential

Full size image

Equilibrium potentials, E eq(x), and potentials under load, E(x), are extracted from the GITT measurements as a function of the composition parameter, x, at different C-rates. From these data, the overpotential is calculated to:

$$ \eta \left(x,I\right)=E\left(x,I\right)-{E}_{\mathrm{eq}}(x). $$
(3)

Figure 5 shows the overpotential of NFP and LFP electrodes as a function of the composition for different C-rates. In both cases, the overpotential shows significant composition dependence and increases with increasing C-rate. The overpotential is largest at the end of charging and discharging where the host materials become completely depleted of vacancies and sodium/lithium-ions, respectively. An additional local maximum is observed in the case of NFP during charging at approximately x = 0.7. The differences in the composition dependence of the overpotential of NFP and LFP are in very good accordance with the results reported by Zhu et al. [34]. The additional features of NFP are again attributed to the path dependence of the sodium insertion-extraction kinetics [32]. The comparison between NFP and LFP further reveals substantially larger overpotentials for NFP. This is in very good agreement to a previous study concerning the charge transfer and diffusion kinetics of sodiation and lithiation of FePO4 [24]. Therein, it was found, that both the exchange current density and the solid-state diffusion coefficient are significantly lower in the case of sodiation. As expected from the previous results, the slower interfacial kinetics and more sluggish diffusivity cause larger overpotential for NFP in comparison with LFP.

Fig. 5
figure 5

Composition dependence of the overpotential at different C-rates for (a) NFP and (b) LFP electrodes

Full size image

Overpotential vs. composition data obtained from the GITT measurements are used to calculate irreversible heat generation rates of the NFP and LFP electrodes according to:

$$ {\dot{q}}_{\mathrm{irr}}\left(x,I\right)=\frac{I}{V}\eta \left(x,I\right). $$
(4)

Figure 6 shows color-coded contour plots comparing the C-rate and composition dependence of the irreversible heat generation rates of NFP and LFP electrodes. The irreversible heat generation rates of both materials show some identical features. For both materials, \( {\dot{q}}_{\mathrm{irr}} \) increases disproportionally with the C-rate, \( {\dot{q}}_{\mathrm{irr}}\sim {I}^a \) with a > 1, at a given SoC. Furthermore, during charging, \( {\dot{q}}_{\mathrm{irr}} \) is lowest at low SoC and rises towards the end of the charging process. During discharging, \( {\dot{q}}_{\mathrm{irr}} \) is lowest at low and mid SoC and increases at the end of discharging. The composition dependence of \( {\dot{q}}_{\mathrm{irr}} \) is caused by changes of the kinetic parameters that govern ohmic, interfacial, and diffusion limitations. For example, a decrease in the solid-state diffusion coefficient of sodium/lithium in the host lattice will cause an increase of the diffusion overvoltage and a rise of the irreversible heat generation. Besides this common behavior, the irreversible heat generation rates of the NFP and LFP electrodes differ in their absolute values at a given current and SoC as well as in the composition dependence. In the case of NFP, an additional maximum of \( {\dot{q}}_{\mathrm{irr}} \) is observed at approximately x = 0.7. Furthermore, the increase of \( {\dot{q}}_{\mathrm{irr}} \) at high and low SoC is much more pronounced for NFP in comparison with LFP. Consequently, NFP exhibits much larger irreversible heat generation rates at the end of charging and discharging. For example, at x = 0.9 and the 1.0 C discharge rate, \( {\dot{q}}_{\mathrm{irr},\mathrm{NFP}} \) is 90 W L−1, whereas \( {\dot{q}}_{\mathrm{irr},\mathrm{LFP}} \) is 23 W L−1. These differences are related to larger charge transfer and diffusion limitations of sodiation compared to lithiation of the FePO4 host lattice as described above. The larger limitations cause larger overvoltage and corresponding Joule heating of NFP compared to LFP at the same C-rate.

Fig. 6
figure 6

Color-coded contour plots comparing the C-rate and composition dependence of the irreversible heat generation rates of NFP and LFP electrodes

Full size image

Table 1 compares the determined mean values and extrema of the irreversible heat generation rates of NFP and LFP at the 1.0 C rate. The results indicate that the average irreversible heat generation as well as minimum and maximum values of NFP are much larger in comparison with LFP.

Total heat

Figure 7 compares color-coded contour plots showing the C-rate and composition dependence of reversible, irreversible, and total heat generation rates of NFP and LFP electrodes.

Fig. 7
figure 7

Color-coded contour plots comparing the C-rate and composition dependence of reversible, irreversible, and total heat generation rates of NFP and LFP electrodes

Full size image

The total heat generation rate of the electrodes results from the superposition of reversible and irreversible heats:

$$ {\dot{q}}_{\mathrm{tot}}={\dot{q}}_{\mathrm{irr}}+{\dot{q}}_{\mathrm{rev}}. $$
(5)

As the reversible heat generation rate is directly proportional to the applied C-rate, \( {\dot{q}}_{\mathrm{rev}}\sim I \), and the irreversible heat generation rate shows a disproportional current dependence, \( {\dot{q}}_{\mathrm{irr}}\sim {I}^{\mathrm{a}} \) with a > 1, the ratio between reversible and irreversible contributions is a function of the C-rate [6, 8]. According to Fig. 7, the total heat generation rate of the NFP electrode is slightly negative at the beginning of charging but rises rapidly with increasing SoC. The cooling effect is caused by the endothermic contribution of the reversible heat, which compensates the irreversible heat at this composition. During discharging, the total heat generation rate of the NFP electrode is strictly exothermic as both reversible and irreversible contributions are exothermic along almost the entire composition range. The largest heat generation rates are found at high currents at the end of charging and discharging.

The magnitudes of the determined volumetric heat generation rates of the LFP electrodes are consistent with previously measured values [35]. The behavior observed in Fig. 7 shows identical tendencies in comparison with the NFP electrode. However, as expected from the individual contributions of the reversible and irreversible heat, differences between NFP and LFP exist, regarding the absolute values of \( {\dot{q}}_{\mathrm{tot}} \) at a given C-rate and SoC. For example, at the 1.0 C rate, the heat generation rates of NFP and LFP at the beginning of charging are almost identical. However, with the progression of the charging process, the heat generation in the NFP electrode increases much stronger compared to LFP. Finally, at the end of the charging process, the heat produced by the NFP electrode more than doubles the value found for NFP. From the contributions of the reversible and irreversible heat, it can be concluded that these differences are mainly caused by differences in the irreversible heat generation rates. The same is valid for the discharging process, aside from the fact that the differences in the total heat generation rates are smaller compared to charging. Key data of the total heat generation rates of NFP and LFP electrodes at the 1.0 C rate are shown in Table 1. The results indicate that both average and extrema of the total heat generation are significantly larger for NFP in comparison with LFP.

To further compare the heat generation in NFP and LFP electrodes, the heat dissipated during complete charge and discharge cycles is determined by integrating the estimated heat generation rates over the respective charge/discharge times, τ:

$$ {Q}_{\mathrm{n}}={\int}_0^{\tau }{\dot{q}}_{\mathrm{n}}\ \mathrm{d}t $$
(6)

where Q n is the respective heat dissipated from the electrodes (Q rev, Q irr, or Q tot). Reversible and irreversible contributions as well as the total heat dissipated from the NFP and LFP electrodes are comparatively shown in Fig. 8. In both cases, the irreversible contribution, Q irr, increases with increasing C-rate. Assuming purely ohmic behavior, Q irr would be given by:

$$ {Q}_{\mathrm{irr}}=\frac{C^2{R}_{\Omega}}{V\tau} $$
(7)

with C and R Ω being the capacity of the electrode and the ohmic resistance. From Eq. 7 it can be concluded, that the heat dissipated during operation is inversely proportional to the charge/discharge time, τ. The experimentally determined dependence shown in Fig. 8 is more complicated, due to the non-ohmic contributions of charge transfer and mass transport limitations, but basically follows this relationship. At any C-rate, Q irr is larger for NFP compared to LFP.

Fig. 8
figure 8

Reversible and irreversible contributions and total heat dissipated during a complete charge-discharge cycle of NFP and LFP electrodes at different C-rates

Full size image

According to Fig. 8, for both materials, the reversible contribution, Q rev, is independent from the C-rate. In contrast to Q irr (Eq. 7), Q rev did not depend on the charge/discharge time [8]:

$$ {Q}_{\mathrm{rev}}=-\frac{CT\Delta S}{zFV}. $$
(8)

In Eq. 8, ΔS is the entropy change during a complete charge/discharge cycle. In both cases, Q rev is endothermic during charging cycles (desodiation/delithiation) and exothermic during discharging cycles (sodiation/lithiation). Thereby, the contribution of the reversible heat is larger for LFP in comparison to NFP. The total dissipated heat results from the superposition of reversible and irreversible contributions. The contribution from the reversible heat is found to be most significant at low C-rate, whereas the irreversible heat dominates at high C-rate. Furthermore, it turns out, that the dissipated heat is larger during discharging compared to charging. The differences between charging and discharging are mainly caused by the contribution of the reversible heat. Q rev is endothermic during charging and partially compensates or even overcompensates Joule heating. In contrast to that, Q rev is exothermic during discharging resulting in larger values of Q tot in comparison with charging.

In the case of LFP, the total dissipated heat is slightly negative at very small charging rates. In contrast to that, for NFP, Q tot is always positive. This is caused by both the smaller contribution of Q rev and larger irreversible heat generation. During discharging at small rates, Q tot is larger for LFP due to the larger contribution of the reversible heat. However, at higher C-rates, for both charging and discharging, the dissipated heat is larger for NFP compared to LFP especially at the ± 2.0 C rate where the differences in the irreversible heats become most significant and reversible heat effects have a minor impact.

Discussion

It should be noted that the contributions of irreversible and reversible heat and the resulting total heat generation are generally affected by the choice of the active material and the electrode design. For example, the particle size can be varied in order to affect the absolute charge transfer resistance and the diffusion path for lithium/sodium in the solid. Typically, electrode designs intended for high power density have thin electrodes, high porosity, and small particles. In this case, the charge transfer and diffusion limitations may have a minor impact, and the reversible heat may dominate the heat generation for a wider range of C-rates than in this study. In contrast to that, for electrode designs intended for high energy density applications, charge transfer and diffusion limitations can be significant even at small currents, and irreversible heat effects may dominate the heat generation at any C-rate. However, in the present study, NFP electrodes are prepared from electrochemical cation displacement using LFP electrodes as starting material. The investigated NFP and LFP electrodes are identical regarding particle size (distribution), active material loading, binder, conductive additives, carbon coating, etc., and all differences in the heat generation rates are exclusively determined by the substitution of the alkali metal cation. Consequently, we hypothesize that the differences in the heat generation rates revealed in this study are a general feature of Na/Li-analogous insertion materials.

Due to faster charge transfer and diffusion kinetics of LFP, the corresponding irreversible heat generation is lower compared to the more rate-limited NFP. In a previous study, the exchange current density of sodiation of FePO4 was found to be significantly smaller compared to lithiation [24]. Furthermore, the solid-state diffusion coefficients were found to differ almost one order of magnitude. The corresponding activation energies of the charge transfer reaction and solid-state diffusion are found to be larger in the case of sodiation as proposed by DFT calculations [36]. These differences in charge transfer and diffusion limitations and the corresponding irreversible heat generation are most likely attributed to the larger ionic radius and the higher molar mass of sodium (1.06 Å, 23 g mol−1) in comparison to lithium (0.76 Å, 6.9 g mol−1) [37], suggesting that these differences may be intrinsic for Na/Li-analogous insertion materials.

To reduce the irreversible heat generation rates of NaFePO4-based positive electrodes for future sodium-ion batteries, significant improvements may be necessary, including targeted nanostructuring, highly conductive catalytic coatings, and doping the material with supervalent cations to increase the rate constant and the diffusivity. This would decrease charge transfer and diffusion limitations, lower the overvoltage at a specific C-rate, and diminish the corresponding Joule heating.

In contrast to LFP, the insertion/extraction kinetics and thermodynamics of NFP exhibit a distinct asymmetry between charging and discharging [32], affecting the corresponding heat generation mechanisms. This effect is related to a path dependence of the sodium insertion-extraction kinetics resulting from increased interaction of sodium-ions compared to lithium-ions with the host lattice [29]. To design optimized and specifically adapted thermal management systems for future sodium-ion batteries employing NaFePO4 electrodes, this asymmetry needs to be considered.

It should be noted that the heat generation in a battery results from the superposition of the heat generation in the anode, the separator/electrolyte, and the cathode. Therefore, the final evaluation of differences in lithium/sodium analogous batteries requires further studies on the thermal behavior of lithium/sodium analogous anodes and the separator/electrolyte compartment. For this purpose, the presented experimental approach appears eminently suitable.

Conclusion

Heat generation rates connected to charging and discharging NaFePO4 (NFP) electrodes in sodium-ion batteries and LiFePO4 (LFP) electrodes in lithium-ion batteries are comparatively investigated. Significant differences between sodiation/desodiation and lithiation/delithiation kinetics and thermodynamics are found, resulting in different heat generation rates of NFP- and LFP-based electrodes. From the results of this study, the following main conclusions can be drawn:

  • The heat generation rates in both NFP and LFP electrodes increase with increasing C-rate.

  • In both types of electrodes, reversible heats are significant at low C-rates, whereas irreversible heats (Joule heating) dominate at large C-rates.

  • The heat generation is larger for discharging compared to charging due to the exothermic and endothermic contribution of the reversible heat, respectively.

  • The heat generation in NaFePO4-based electrodes is significantly larger compared to LiFePO4 mainly because of larger Joule heating.

The larger heat generation rates may complicate thermal management of NFP-based sodium-ion battery systems for high power applications. The differences between the heat generation rates of NFP and LFP electrodes, revealed in this study, exclusively result from the substitution of the alkali metal cation. Consequently, this may be a general feature of Na/Li-analogous materials and should be accurately considered for the development of novel active materials as well as advanced particle and electrode designs for future sodium-ion batteries. Fundamental insights regarding the impact of the electrochemical active alkali metal cation in the active material may significantly accelerate the development of sodium-ion batteries and other “beyond lithium-ion” batteries based on the comprehensive scientific and technical knowledge existing in lithium-ion technology.