Study of the Operation of Lead–Acid Battery Electrodes Under Hybrid Battery–Electrolyzer Cycling Profiles

Abstract

Flooded lead–acid batteries start producing oxygen and hydrogen during the final stages of charge and subsequent overcharge. The collection of the hydrogen produced allows for an increase in overall energy efficiency and transforms the system into a hybrid device typically referred to as a “Battolyzer” (battery electrolyzer). The present work explores the feasibility of the above approach through a detailed study of the long-term ageing process of flooded tubular lead–acid cells subjected to various rates of discharge and overcharge, emulating four different scenarios of Battolyzer use, starting from 70% depth of discharge cycling to nearly continuous water electrolysis. The combined results from the electrochemical and corrosion studies showed that the Battolyzer cells’ degradation was driven by the corrosion of the positive current collectors. The progress of the corrosion process was strongly correlated with the amount of hydrogen produced. The increase in the depth of discharge resulted in minor decreases in the corrosion current, indicating that the battery functionality of the Battolyzer was more advantageous than the continuous water electrolysis.

1. Introduction

The ongoing global decarbonization of sectors like transport and the production of electric power and heat relies strongly on battery and electrochemical electrolyzer technologies for the storage and conversion of electricity from intermittent renewable energy sources. This energy transition process has resulted in an emerging shift in raw material demand from fossil fuels like coal, oil, and natural gas to critical minerals like lithium, cobalt, nickel, copper, rare earths, and precious metals. As a consequence, the development and optimization of electrochemical technologies using noncritical natural resources remain tasks of great importance. Lead is a noncritical natural resource with almost exclusive use in electrochemical applications and, more specifically, lead–acid battery manufacturing. The use of lead outside energy storage applications is continuously declining due to the toxicity of this element. In contrast, in the current state of the art in lead–acid battery manufacturing, use and recycling are considered safe and durable, with collection and recycling efficiency exceeding 99% globally [1]. Lead–acid batteries can be divided into two main categories: “flooded” batteries with free-flowing, liquid sulfuric acid electrolytes and “sealed” or “valve-regulated” batteries with immobilized electrolytes in the form of a thixotropic gel or infused inside an absorptive glass mat (AGM) separator [2]. When the state of charge (SOC) of any lead–acid battery exceeds 70–75%, a process of oxygen evolution starts at the positive electrode. This oxygen can recombine back into water molecules at the negative electrodes of the valve-regulated cells or it can be vented outside in the case of cells with a “flooded” design. When the SOC of flooded cells exceeds 90–95%, a process of hydrogen evolution starts at the negative electrode. Thus, at the end of the charge and during its overcharge, a flooded lead–acid cell operates as an electrolyzer, producing gaseous hydrogen and oxygen and losing water from the electrolyte. Water electrolysis is considered a “parasitic” process, which decreases the energy efficiency of the battery, causing a need for maintenance (periodic top-up with deionized water) and accelerating the ageing of the positive electrodes through active material shedding and current collector corrosion [3]. The collection of the produced hydrogen can be an opportunity to turn some of the above drawbacks into an advantage. The approach is known in the literature as a battery electrolyzer hybrid cell or simply a “Battolyzer”. Up to now, the concept has mostly been explored by alkaline electrochemical conditions using nickel-based positive electrodes and a variety of negative electrodes [4,5].

The development of a lead–sulfuric acid Battolyzer system is a challenging task due to several particularities of lead electrochemistry. The first one is the poor electrocatalytic properties of lead and lead dioxide electrodes in sulfuric acid solutions, resulting in high hydrogen and oxygen overvoltage and water electrolysis onset above 2.3–2.4 V. The second disadvantage is the durability of the positive electrodes, which is limited by the current collector corrosion and the lead dioxide integrity, which are both accelerated by the process of oxygen evolution [6].

The aim of the present work is to present a detailed study of the lead–acid Battolyzer concept and its feasibility using state-of-the-art “deep-cycling” flooded cells with a tubular positive electrode construction [7]. This offers a combination of confined lead dioxide positive active material (PAM) inside woven or nonwoven textile gauntlets (tubes) and much thicker current collectors. PAM confinement significantly delays the loss of lead dioxide integrity, denoted as “softening and shedding”, while the corrosion of the thicker current collectors takes a longer time. As a consequence, the life cycle of the tubular cells can be very long, from 1000–2000 deep cycles to 6000–7000 full equivalent cycles in a partial state-of-charge mode [8]. According to Pavlov et al., the anodic corrosion process proceeds via a solid-state mechanism of oxygen vacancy insertion into the lead host matrix [9], with kinetics following the Tafel equation [10,11]. The presence of positive active material on the surface of the current collector slows down the corrosion process significantly, while the impact of the polarization mode is more complex [12].

2. Materials and Methods

The experiments were carried out using the flooded tubular cell model “2 HPZS 120”, manufactured by Hoppecke Batteries (Brilon, Germany,), using a positive current collector of lead alloy containing 4.2% Sb, 0.09% Sn, and 0.012% Se. Cells with a nominal capacity of 120 Ah and a nominal voltage of 2 V were received in a “dry charged” condition and filled with battery-grade H₂SO₄ electrolyte with a specific gravity of 1.26 g·cm⁻³. The ageing tests were carried out on four six-cell strings (12 V/120 Ah battery modules) using Digatron UBT 100-020 ME (Aachen, Germany) battery testing equipment with individual cell voltage monitoring. The experiments were performed at 25 °C in a thermostated water bath with the water level adjusted slightly above the maximum height of the cell electrolyte level, in accordance with aqueous battery testing standards [13]. Prior to the start of the ageing tests, the battery modules were subjected to a “reception test” comprised of 3 cycles with a 10 h rated charge/discharge current (12 A) and 3 more cycles with a 5 h rated current (24 A) in order to estimate the initial capacity and verify the homogeneity of the cell-to-cell performance. The discharge process was terminated by a cut-off voltage value of 1.75 V per cell. The “IUi” type of charge process was comprised of three steps, starting with a constant current phase (10 h or 5 h rated), followed by a constant voltage stage at 2.4 V per cell and an end of charge step with a constant current of 4 A and a duration of 4.5 h. A measurement of the cell impedance at 1 kHz was performed at the end of this sequence using an HIOKI BT3554 milliohm meter. Four cycling test sequences emulating the hybrid battery–electrolyzer operation under different scenarios are presented in

Table 1. Battery–electrolyzer hybrid cycling test protocols employing a combination of different depths of discharge (DOD) and charge factor (FC) values.

Test Protocol	DOD (%)	FC (%)	Charge Input (Ah/Cycle)	H2 Production (Ah/Cycle)
MOD1: Deep cycling	70	143	120	36
MOD2: Moderate DOD cycling	50	200	120	60
MOD3: Shallow cycling	20	500	120	96
MOD4: Continuous electrolysis	1	10,000	120	118.8

. They allow the evaluation of the impact of the ratio between the battery and electrolyzer functionalities on the energy efficiency and the electrode degradation mechanism. The cycling was applied during a four-week or eight-week-long period, followed by 3 capacity check-up cycles with 5 h rated current and an overcharge (Tafel) test comprised of constant current staircase increasing from 0 to 20 A. The ageing tests were renewed for another period after the removal of one cell from each battery module. The extracted cells were subjected to supplementary electrochemical tests. They include 3 charge/discharge cycles with 10 h rated current, overcharge test, and impedance spectroscopy tests in two-electrode and three-electrode modes using Ag/Ag₄SO₄/H₂SO₄ reference electrode (+0.65 V vs. SHE). The extracted cells were further subjected to teardown analysis, preceded by an impedance measurement at 1 kHz and electrolyte-specific gravity estimation. The teardown analysis included a visual examination of the cell components, followed by water rinsing and air drying at 60 °C for 48 h of all electrodes. Smaller samples of the extracted negative electrodes were rinsed with ethanol and acetone after the water rinsing in order to prevent the oxidation of the metallic lead from the negative active material during the subsequent drying.

The crystallographic composition of the dried new and aged positive and negative active materials has been characterized with powder X-ray diffraction using a Cu K-alpha ray source.

The corrosion process of the positive current collectors has been studied by the weight loss method [11,12,13] and by metallographic microscopy observations. Supplementary Information S1 presents a description of the preparation of the samples for the corrosion studies.

3. Results and Discussion

3.1. Energy Efficiency During the Lead–Acid Cells’ Operation Under Battery–Electrolyzer Cycle Ageing

Energy efficiency is one of the critical parameters for the evaluation of any kind of system for energy conversion and storage. In the case of a battery–electrolyzer hybrid cell, this parameter is more complex and it depends on the way of using the produced hydrogen (as a heat source via burner device or as an electric power source via fuel cell). Since the present research is part of the project Horizon Europe LoCEL-H2 [14], the hydrogen produced during the operation of the lead–acid cells will be considered as a source of heat for a “clean” cooking process [15]. Thus, a parameter denoted as a “combined energy efficiency” can be defined using the following expression:

$${\eta }_E=100{\displaystyle \frac{E_{discharge}+{\mathsf{\Delta }H}_{H2}}{E_{charge}}}$$

(1)

where E_discharge and E_charge are the energies accounted in discharge and charge mode and measured by the battery test equipment, and ∆H_H2 is the enthalpy of combustion of the produced hydrogen. The latter is calculated as the product between the number of moles of H₂ produced in the overcharge mode (calculated by Faraday’s law) and the specific enthalpy of combustion of the hydrogen (285.8 kJ·mol⁻¹ [16]).

Figure 1 presents the cell voltage profiles during the test cycles following the experimental protocols described in Table 1. All four cycling schedules share the same number of Amp-hours applied to the cells in charge mode and different discharge durations corresponding to different values of the depth of discharge (DOD). In this way, we can emulate different scenarios of battery–electrolyzer operation with various shares of hydrogen and electricity demands and a fixed production of electric energy from a renewable source (the latter corresponds to the fixed number of Amp-hours applied during the charge). It should be noted that, in the present work, the parameters DOD and SOC are referenced versus the nominal capacity of the cells according to the following expressions:

DOD = 100 (C_n − Q_discharge)/C_n

(2a)

SOC = 100 − DOD

(2b)

where Q_discharge is the number of Amp-hours obtained in discharge mode.

The first profile (also denoted as “MOD1”) shares some similarity with the so-called “battery deep cycling” operation typical for industrial battery applications like forklifts or golf-carts [17]. There are two main differences between the present case and the typical deep cycling. The first one is the use of higher end-of-charge current corresponding to the water electrolysis process. It is equal to C_n/10h, where C_n is the nominal capacity of the cell, while, in the “traditional” case, it corresponds to C_n/20h–C_n/40h, i.e., two to four times lower. The second different point is the set of a higher charge factor (FC) corresponding to the percentage ratio between the number of Amp-hours in discharge mode and the Amp-hours injected during the subsequent charge (FC = 100 Q_charge/Q_discharge). The charge factor is a parameter representing the overcharge applied to the battery cell. The optimal range of the charge factor during the deep cycling of a flooded lead–acid cell is 110–115%, corresponding to 10–15% of overcharge applied, which is necessary for the complete recharge of the cell and the elimination of the electrolyte stratification [17]. In the case of the cycling profile “MOD1”, presented in Figure 1a, the charge factor is equal to 143%, i.e., the applied overcharge is 3–4 times more than usual. The calculation of the combined energy efficiency in this case remains practically constant, close to 75%, almost for the whole duration of the ageing test (9 months corresponding to 584 cycles, without taking into account the periodic check-up tests). This result is in good agreement with the little difference observed between the voltage data.

Figure 1b presents the voltage transients recorded during the tests under the experimental protocol “MOD2”, which combines slightly lower depth of discharge (DOD = 50%) with higher charge factor (FC = 200%), corresponding to more hydrogen produced. The combined energy efficiency under this test profile decreases to 67.5% and does not degrade much despite the excessive overcharge, which is a well-known battery electrode degradation root cause. This lower energy efficiency is related to the high oxygen and hydrogen overvoltage at the positive and negative electrodes of the lead–acid battery [6].

The third battery–electrolyzer operation scenario (denoted also as “MOD3”) is presented in Figure 1c. It starts with a shallow discharge with DOD = 20% and continues with further increase of the hydrogen production during the overcharge stage. The latter is terminated when the charge factor reaches 500%. The energy efficiency observed during this test mode decreased further down to 57%; however, it also remains rather constant until the end of the tests terminated after 6 months of cycling.

The last cycling mode corresponds to nearly continuous water electrolysis with constant overcharge current equal to C_n/10h, interrupted by an hour-long rest period including a short discharge corresponding to DOD = 1%. The combined energy efficiency during this operation scenario is about 51%. It should be noted that the latter exceeds the voltage efficiency considering the theoretical water splitting voltage of 1.23 V and the overcharge voltage plateau of 2.9 V at a current equal to C_n/10h. This difference can be explained by the accounting of the hydrogen combustion enthalpy, which sums an entropic and a Gibbs free energy component.

Last but not least, it is important to note that the above-mentioned combined energy efficiency values should be considered as results measured in “ideal” conditions. The latter includes minimum Ohmic resistance of the cells delivered by a very narrow gap between the positive and the negative electrodes, large relative size of the terminals due to the small nominal capacity (the cell model “2 HPZS 120” corresponds to the smallest one in this product range), and a four-point connection scheme excluding the losses due to the cables.

3.2. Electrochemical Kinetics of the Water Electrolysis During the Lead–Acid Cells’ Operation Under Battery–Electrolyzer Cycle Ageing

The operation of the lead–acid cells in hybrid battery–electrolyzer mode and the electrodes’ ageing mechanism were also studied by periodic performance check-up procedures, including capacity measurement and overcharge test. Figure 2 presents a summary of the results of such overcharge tests carried out in single-cell mode right after three capacity estimation cycles, which left the cells completely charged. The overcharge is carried out using an exponential-like, constant-current, staircase profile, presented in Figure 2a. The duration of the overcharge current steps is decreased progressively in order to limit the overall overcharge applied to the cell to 17 Ah or 15% of the nominal capacity and yet to achieve nearly steady-state voltage values. The cell voltage as well as the positive and negative half-cell potentials are further analyzed using semi-logarithmic (Tafel) plots. An Ohmic drop correction was applied to all positive half-cell potential data, knowing that practically all cell degradation phenomena occur at this electrode.

The comparison of data presented in Figure 2b–f shows that all Tafel plots contain two regions—a linear high-current domain corresponding to an overcharge current higher than ≈ C_n/100h (1A) and a low-current domain, which does not correlate well with the Tafel equation. The estimated values of the Tafel slope at the negative electrodes remain in the range of 130–160 mV·dec⁻¹, which is slightly higher than the typical value of 120 mV·dec⁻¹ [6]. The observed effect of increasing Tafel slopes with the progression of the ageing process can be related to a redistribution of the poorly soluble expander molecules in the porous negative active material. The ageing causes a decrease in the Tafel equation intercept, related to the exchange current density. This effect can be ascribed to the process of transfer of antimony from the corroding positive current collectors to the surface of the negative active material, denoted in the literature as “antimony poisoning” [18]. The main consequence of this process is a decrease in the hydrogen overvoltage, i.e., enhanced hydrogen production. It should be noted that, in the present work, this effect is observed mostly at lower overcharge currents, explaining the small changes of the overcharge voltage plateau observed in Figure 1. The absence of significant antimony poisoning effects (which are much smaller in comparison with the previous experience of our research team) can be related to the “boost charge” effect of the hydrogen production at C_n/10h. The latter corresponds to a process of formation of gaseous SbH₃ at the negative electrode, which is readily vented together with the produced hydrogen [18].

The analysis of the Tafel plots of the positive electrodes indicates that the test mode impacts the oxygen evolution kinetics in a different way. The application of cycling profiles with high DOD (Figure 2c,d) results in an increase in the Tafel slope from 70 mV·dec⁻¹ to 90 mV·dec⁻¹. The latter corresponds to decreased reversibility of the oxygen evolution, and it can be related to progressive accumulation of PbO₂ with higher crystallinity and lower electrochemical activity [19]. On the other hand, the ageing of the cells under conditions of more intense electrolysis results in a shift in the Tafel lines towards lower overvoltage values without any significant change in the Tafel slope. We can suggest that this effect is due to the increased doping of PbO₂ with Sb(III) and Sb(V) species coming from the accelerated positive grid corrosion, which are causing increased PbO₂ electrochemical activity throughout increased hydration of the PbO₂ particles [20]. It is important to mention that observed oxygen evolution Tafel slopes are fairly close to the typical value of 80 mV·dec⁻¹ [6].

3.3. Evolution of the Discharge Performance Under Battery–Electrolyzer Cycle Ageing Conditions

Figure 3 presents a comparison of the discharge behavior of cells subjected to different periods of ageing using the cycling schedule “MOD4” (“electrolyzer” mode). The discharge tests with a current equal to C_n/5h have been carried out on the entire string, while those with 10 h rated current have been performed in single-cell mode, using a reference electrode for half-cell potential monitoring. Figure 3a presents the discharge performance of Cell#6 from the string 25MOD4. It can be seen that the initial 5 h rated capacity is slightly above 100 Ah, while, with a current equal to C_n/10h, this parameter exceeds 120 Ah. The comparison of this result with the data in Figure 3b, presenting the discharge performance of another new cell filled with excess of electrolyte, indicates that the new cells are limited by the positive electrodes, as well as by the electrolyte quantity (the overfilling was due to the absence of a compression jig, which can limit the appearance of a cell wall “dimple”). Figure 3c indicates that, after two months of operation in nearly continuous overcharge mode, the discharge capacity increases above 120 Ah and 140 Ah, respectively, at C_n/5h and C_n/10h. The position of both cell voltage transients suggests that there is no significant change in the cell’s internal resistance despite the aggressive overcharge. The analysis of the half-cell potential data shown in Figure 3d indicates that, after the period of 2 months under the profile “MOD4”, the positive electrode is no longer limiting the discharge process and the electrochemical behavior of the cell resembles a battery with a “balanced” ratio between the positive and the negative active material content.

The results presented in Figure 3e show that the cells subjected to 4 months of ageing under the test profile “MOD4” start experiencing a significant loss of discharge performance at 5 h rated current, keeping nearly the same 10 h rated discharge capacity as at the beginning of the tests. The above difference suggests considerable “resistive” ageing [21]. On the other hand, there is no “visible” ageing-related voltage drop increase at the start of the discharge process. The corresponding half-cell potential transients recorded during the C_n/10h discharge test show that the electrochemical behavior is still typical for a battery with a “balanced” ratio between the positive and the negative electrodes. This result is rather unexpected because it is very well known that the ageing mode “MOD4” causes severe and selective degradation of the positive plates.

The data in Figure 3g show that, after five months of operation in electrolyzer mode, the cell lost more than 60% of its nominal capacity at both discharge rates applied. The analysis of the half-cell potential transients in Figure 3h indicates that, at the very end of cell lifetime, the negative electrodes become the capacity-limiting battery component, following the trend observed in Figure 3b,d,f.

Figure 4 presents the evolution of the discharge performance during the check-up cycling of cells from the battery string subjected to the ageing profile “MOD1” (or “deep cycling”), combining a DOD = 70% and FC = 143%. The discharge performance data corresponding to the beginning of this test are identic with those already discussed above (Figure 3a,b). It can be seen that, in this operation mode, the discharge performance at both currents (C_n/5h and C_n/10h) remains nearly constant at the cell and half-cell level, resembling the curves shown in Figure 3c,d.

The comparison between the data in Figure 3 and Figure 4 shows that the impact of the overcharge (i.e., the hydrogen production function of the hypothetical battery–electrolyzer hybrid system) on the cell performance degradation rate is much stronger than the use in battery mode with longer discharge periods. Thus, it is necessary to define a parameter accounting for the overcharge applied to the battery cells, starting from the beginning of their operation.

Such a parameter can be the cumulative number of Amp-hours overcharge (Q_ovch), which is equal to the difference between the cumulative number of Amp-hours in charge/overcharge mode and those in discharge mode:

$$Q_{ovch}\left(t\right)={\int }_0^t{i}_{ch}\left(\tau \right)d\tau -{\int }_0^t{i}_{dsch}\left(\tau \right)d\tau$$

(3)

where t is the global time of battery cell operation, τ is a local time variable, and i_ch(τ) and i_dsch(τ) represent the charge and the discharge current profiles. If Q_ovch is divided by the nominal capacity of the cell C_n, we obtain a dimensionless parameter very similar to the number of equivalent cycles characterizing the battery ageing in partial state of charge applications [21]. Since the overcharge process coincides with the production of hydrogen, we can reference this new parameter as “equivalent cycles of hydrogen production” (H₂ equivalent cycles or n_H2).

A summary of the periodic performance check-up tests carried out on all four battery strings is shown in Figure 5. It allows an easier overall assessment of the evolution of discharge performance as a function of the above-defined number of H₂ equivalent cycles. In addition, the DC internal resistance (R_DC) at SOC = 100% is also presented. The last parameter is calculated using Ohm’s law, taking into account the applied discharge current and the difference between the open-circuit voltage prior to the start of the discharge and the cell voltage measured after 300 s (U_300s) and 600 s (U_600s) of discharge:

R_DC = (OCV − U_300s)(C_n/5h)⁻¹

(4a)

R_DC = (OCV − U_600s)(C_n/10h)⁻¹

(4b)

The voltage sampling selection at 300 s and 600 s from the beginning of the discharge corresponds to the point of 1.7% depth of discharge. The latter is positioned on the cell voltage plateau following the “coup de fouet” region related to the nucleation of PbSO₄ at the beginning of almost any completely charged lead–acid battery [22]. R_DC is a complex parameter with at least two components. Its first component is the Ohmic resistance of the cell, and the second one is the sum of the charge transfer resistance of the discharge reactions at both electrodes.

Figure 5a and 5b show that the lead–acid cell operation under cycling protocol “MOD1” (deep cycling) proceeds with rather small variation in the discharge capacity and an initial decrease in the DC internal resistance to steady-state values. The observed difference between R_DC at both discharge rates can be explained by the nonlinear behavior of the charge transfer resistance component, which tends to lower values when the applied current is increased following the Butler–Volmer equation [23]. On the other hand, the initial decrease in R_DC can be related to the initial development of a more porous structure of the active materials, which decreases the local current density.

The increase in the hydrogen production share during the operation of the cells under the test protocol “MOD2” results in the appearance of a local minimum and subsequent recovery of the discharge capacity data presented in Figure 5c. A similar kind of irregularity is also observed in the evolution of the R_DC shown in Figure 5d. A further increase in the share of overcharge operation during the cycling mode “MOD3” results also in a similar type of “irregular” evolution of the capacity and the internal resistance caused by the joint effect of the active materials cycling and the process of positive current collector corrosion.

The overview of the results from the check-up of the cells subjected to the last test protocol “MOD4” presented in Figure 5g resembles, to a certain extent, the behavior of the lead–acid batteries operating in stand-by mode [24]. On the other hand, the evolution of the internal resistance values shown in Figure 5h is much slower than expected, no matter that it matches qualitatively the behavior of the discharge capacity presented in Figure 3g.

3.4. Evolution of the Electrochemical Impedance During the Battery–Electrolyzer Cycle Ageing

Figure 6 presents the evolution of the electrochemical impedance during the battery–electrolyzer cycle ageing of cells from the strings subjected to the cycling schedules “MOD2” (medium DOD cycling) and “MOD4” (continuous electrolysis mode). The measurements are carried out consecutively in “cell mode” (Figure 6a,b), “positive half-cell mode” (Figure 6c,d), and “negative half-cell mode” (Figure 6e,f) at open circuit, after a 24 h long rest period in an air-conditioned laboratory (T = 23 °C ± 2 °C). It was also verified that the sum of the positive and negative half-cell impedance matches the impedance measured directly in cell mode.

The data in Figure 6a,b show that, at the beginning of the ageing under both types of test profiles, there is an initial increase in the cells’ Ohmic resistance (i.e., the spectra shift on the right in the Nyquist plots) without any other significant change in the curves’ pattern. The latter remains very similar to the cases discussed in the literature featuring a high-frequency inductive part related to all metallic pieces in the system (including the cables of the four-point measurement connection to the testing equipment), medium-frequency capacity loop, and low-frequency diffusion (or Warburg) “tail” [25]. The comparison of the different spectra indicates that the gain of the Ohmic resistance is not monotonic and it can feature a local maximum followed by a short decrease before going up again at the stage when the cell capacity starts decaying very fast (the case presented in Figure 6b).

The closer inspection of the positive half-cell impedance spectra presented in Figure 6c,d also indicates an initial increase in the Ohmic resistance related to this electrode and further growth only for the case of the cells from the string subjected to the test protocol “MOD4”. The progression of ageing is also accompanied by the appearance of high/medium-frequency “pseudo-inductive” loop, which can be related to the dynamics of the PbO₂ hydration/dehydration phenomenon and its link with the antimony doping [26].

The analysis of the negative half-cell impedance data indicates that this electrode contributes mostly to the overall cell impedance (it can be seen that the axis of the plots shown in Figure 6a,b,e,f remains the same range, while the positive half-cell impedance is roughly one order of magnitude lower). This particularity is because the specific surface area of the positive active material (porous, nano-sized PbO₂) is about 10 times higher than its counterpart (porous Pb micro-particles) [2]. The data in Figure 6f indicate a significant increase in the negative half-cell impedance, coinciding well with the result presented in Figure 3h (discharge performance limited by the negative electrodes). This rather unexpected result will be further discussed in the next paragraphs, a priori knowing that the severe degradation phenomena are concentrated at positive electrodes.

Considering that ageing under the proposed battery–electrolyzer test scenarios impacts mostly the Ohmic resistance of the cells (apart from a relatively short period preceding the practical end-of-life of the positive electrodes), a more detailed analysis of this parameter has been carried out and presented in Figure 7. The plots in Figure 7 contain the Ohmic resistance of the cell (the intercept of the impedance spectrum and the real impedance axis of the Nyquist plot, i.e., Z’ values corresponding to Z″ = 0), as well as the 1 kHz impedance measured by HIOKI BT3554 apparatus at SOC = 100% before the launch and after the termination of the ageing tests of the corresponding cell.

The data in Figure 7 show clearly that the Ohmic resistance is very close to the readings of the HIOKI, and both types of data are practically interchangeable. Using this approach, allowing the elimination of the cell-to-cell variations in the Ohmic resistance, one can see the dynamics of the evolution of this parameter. All four studied cases start with a gain in the Ohmic resistance in the range of 0.5 mOhm (with a variation of about ±5%). This result can be explained easily with the initial growth of the corrosion layer on the surface of the positive current collectors. The Ohmic resistance passes further a local maximum, corresponding to the above-mentioned gain of 0.5 mOhm. The observed maximum becomes broader under profiles including more intense hydrogen production. The resistance starts increasing again when the end-of-life of the cells approaches (e.g., in Figure 7c,d). The effect observed in Figure 7 is rather unusual for the common lead–acid battery health monitoring practice, but it will be explained using the results from the teardown analysis of the cells presented in the next paragraph.

All results of the electric and the electrochemical tests indicate that the dynamics of the ageing processes may remain “hidden” over periods of significant duration. On one hand, this is an advantage corresponding to a stable operation but, on the other hand, it can be a drawback requiring more sophisticated continuous battery monitoring, including almost mandatory Amp-hour counting in both charge and discharge modes. The latter will permit the practical implementation of the parameter “number of H₂ equivalent cycles” as a precise lead–acid battery or battery–electrolyzer ageing indicator [27].

3.5. Visual Examination of the Cells’ Components and X-Ray Diffraction Characterization of the Active Materials

The study of the mechanisms of operation and degradation of lead–acid battery electrodes under battery–electrolyzer hybrid operational conditions continued with a teardown analysis of the cells subjected to testing with different durations. The first step of this characterization was a visual examination of the completely charged cell’s components. The latter includes the appearance and the texture of the positive and negative electrodes, the state of the positive and negative cast-on-strap welding, as well as the separators. The specific gravity of the electrolyte was measured using a digital hydrometer, together with a visual check for the appearance of an opacity. The results from these tests showed that all visual degradation indicators are concentrated at the positive electrodes. These degradations include swelling, deformations and rupture of the porous gauntlets (appearance of fissures running along the gauntlet length), minor softening of the PbO₂ in the bottom section of the electrodes, and accumulation of very fine PbO₂ particles (referred to as “mud”, which is collected in the so-called “mud-space” on the bottom of the cell casing). There was also a scaling-off of corrosion fragments from the positive current collector sections exposed directly to contact with the electrolyte. The degradation effect of the latter was considered rather marginal due to the considerable thickness of these electrode fragments (about 9–10 mm, in comparison to the current collector spines, which have a diameter of 3 mm). Two cells from the string “MOD3” and two cells from the string “MOD4” failed during the testing due to the appearance of internal short-circuits, which can be associated with the local swelling of some gauntlets, resulting in points of separator squashing. One cell from the string “MOD1” and one cell from the string “MOD2” failed earlier due to a loss of discharge capacity performance. The latter was considered rather as a random anomaly. There was no significant positive current collector height growth at any of the examined cells. The visual effects of the ageing of the electrodes from a cell subjected to five months of ageing under the continuous electrolysis profile “MOD4” can be seen in the Supplementary Information S2.

X-ray diffraction (XRD) was used to evaluate the crystallographic composition of the positive and negative active materials at SOC = 100% using a semi-quantitative phase analysis based on the height of selected diffraction peaks. The positive active materials were prepared from plates washed with tap water until neutral pH, followed by 48 h drying in air at 60 °C. The negative active materials were prepared from disk fragments cut from negative plates washed with tap water until neutral pH. The wet disks were immersed in ethanol for 1 h and, next, in acetone for another hour before putting them to drying in air at 60 °C. The above treatment replaces the water, catalyzing the oxidation of the metallic lead with inert and highly volatile solvents, which are rapidly evaporated from the pores of the samples. Our practice from earlier projects showed that negative active material samples from state-of-the-art new lead–acid batteries contain about 2–3% of tetragonal lead oxide after the above-mentioned washing and drying routine.

The negative active material (NAM) samples from the new cell, subjected to three capacity measurement cycles and a short Tafel test, showed the presence of considerable amounts of orthorhombic and tetragonal lead oxides (o-PbO and tet-PbO) together with the metallic lead, which is the main electrode constituent. One can suggest that the detected lead oxides are an artefact from the dry-charge cell manufacturing process, which remains even after activation of the cell and initial cycling. At the same time, the data shown in Figure 3b indicate that the detected lead oxide content does not affect the electrochemical performance of the negative plates.

The X-ray diffraction analysis of the NAM samples from the cells analyzed after four months of ageing under the four applied test profiles showed a phase composition similar to that observed for the new cell. After a careful examination of the XRD background signal, it was concluded that there are no detectable traces of PbSO₄. The XRD analysis of several samples subjected to 5 and 7 months of ageing showed that the amount of o-PbO and tet-PbO decreases progressively. Again, no PbSO₄ was detected.

The positive active material (PAM) sample from the new cell was composed of beta-PbO₂ with minor traces of alpha-PbO₂ (<4% of volume). The PAM samples from cells operated 4 months under the test profiles “MOD1”, “MOD2”, and “MOD3” were found to be composed of more than 99% of the volume of beta-PbO₂. On the other hand, about 4% of alpha-PbO₂ was found in the PAM sample from the cell operated for four months under the profile “MOD4”. It has to be noted that the latter sample was actually a mixture of PAM and a considerable fraction of the thick corrosion layer detached readily from the current collector spines.

PAM samples from the cells cycled during 7 months under the test protocols “MOD1” and “MOD2” were XRD analyzed as well. Again, over 99% of the PAM was composed of beta-PbO₂. The current collector spines from both cells contained a dense and adherent corrosion layer, which was removed after the collection of the main part of PAM by multiple bending of the spines (the corrosion layer fell off the spine surface as scales). The XRD analysis of this mixture of PAM and corrosion layer (there is a significant amount of PAM stuck on the corrosion layer scales) showed the presence of alpha-PbO₂ in volume fractions as high as 45%. This finding is in excellent agreement with the work of Pavlov et al., and it shows that the anodic corrosion process under the hybrid battery–electrolyzer operation modes follows a solid-state mechanism of oxygen species insertion into the lead alloy host matrix [9,10,11,12].

All X-ray diffraction data commented on in this paragraph can be found in the Supplementary Information S3–S12.

3.6. Anodic Corrosion of the Positive Current Collectors

The results from the visual examination of the cells’ components and X-ray diffraction characterization of the active materials showed that the corrosion of positive current collector spines is the root cause of the cells’ degradation under hybrid battery–electrolyzer operation scenarios. Figure 8 illustrates the evolution of the corrosion process at cells subjected to the test protocol “MOD4” or nearly continuous water electrolysis. The samples are prepared from the same electrode locations at three levels of the positive plate height. The sampling method is illustrated in the Supplementary Information S1b. The observation of the samples from the new cell subjected to the short initial cycling mentioned previously showed the presence of a relatively thin corrosion layer, with a thickness varying between 15 and 30 µm (the thickness was observed under higher magnification using a metallographic microscope). It can be seen that, after 2 months of ageing, the thickness of the corrosion layer increases to 0.7–0.9 mm. The latter values are doubled after another 2 months of ageing. It should be mentioned that, after 5 months of ageing under the same test protocol, the remaining fragments of the current collector spines were too small, and it was not possible to prepare samples for microscopy observations. The comparison of the micrographs corresponding to the different positive plate heights shows that the corrosion process remains rather homogeneous during the majority of the cells’ lifetime. Taking into account the discharge performance data in Figure 3g, it can be suggested that, after 4 months of ageing, the spine diameter is close to the critical threshold. The micrographs corresponding to 4 months of testing also indicate that, beyond this point, the corrosion mechanism changes from a homogeneous process towards “pitting”, which may cut the current collector spines horizontally.

The quantitative estimation of the corrosion rate was based on the measurement of the surface area of the metallic part of the spine’s cross-section, instead of using the corrosion layer thickness, which may vary or even be transformed into more porous PbO₂ active material (especially under the other test profiles employing high depth of discharge). The measurement was based on the estimation of four characteristic diameters (long, short, and two diagonals) and the use of the mean diameter to calculate the surface of the perpendicular cross-section of a cylindrical wire.

The resulting change in the spine diameter was converted into a volume loss of metallic lead per centimeter of spine length and, next, into a weight loss per square centimeter of initial spine area (using mg·cm⁻² units) by simplifying the alloy composition to pure lead.

The summary of the current collector weight loss measurements based on the microscopy observations is shown in Figure 9. The corrosion data from all four test profiles can be compared by using the quantity of the produced hydrogen (the applied overcharge) as the x-axis. It can be seen that there is a nearly linear correlation between the weight loss and the number of H₂ equivalent cycles. The results also indicate that the test mode exerts rather a small impact on the corrosion rate. The data also show that the corrosion is slightly faster in the bottom section of the electrodes and slightly lower in the middle section. The mechanism behind this minor spatial distribution of the corrosion process is complex. It is hard to relate it to the electrolyte stratification, for example, especially in the case of the data in Figure 9d, due to the negligible share of the discharge process and the constant gassing. The observation of slightly faster corrosion in the bottom and the top section of the electrode is also in good agreement with the work of Guo et al., describing a local increase in the tubular electrode’s resistance in their top and bottom sections [28].

Figure 10 presents the results from the corrosion rate measurements using the weight loss method. The estimation is carried out using four positive current collector spines cut to a length of 223 mm ± 1 mm. The results shown in Figure 10a are very similar to the data plotted in Figure 9. It can be seen that the corrosion process under the test profile “MOD4” is slightly faster than the other three cases. The measurements presented in Figure 10a have been used to estimate the partial current corresponding to the electrochemical process of corrosion defined with the reaction:

Pb + 2H₂O → PbO₂ + 4H⁺ + 4e⁻

(5)

The application of Faraday’s law for the above reaction can be expressed as:

$$i_{corr}={\displaystyle \frac{4∆m_{spine}F}{∆t_{electrolysis}{A}_{Pb}}}$$

(6)

where i_corr is the corrosion current, ∆m_spine is the current collector weight loss, F is Faraday’s constant, ∆t_electrolysis is the duration of the H₂O electrolysis, and A_Pb is the atomic weight of the lead. The term ∆t_electrolysis is calculated as the ratio between the number of amp-hours of overcharge and the applied overcharge current, equal to C_n/10h.

Figure 10b shows that the corrosion current remains close to 0.1 mA·cm⁻². This result allows modeling of the corrosion process as independent of the depth of discharge. However, it is expected to vary strongly with the change in the temperature and the water electrolysis current (which, in its turn, defines the electrode potential following the Tafel plots shown in Figure 2b–f) [10,11,12].

3.7. Evolution of the Positive Electrode Deformation During the Ageing Process

The process of positive electrode “growth” is an important and very adverse effect caused by the positive current collector corrosion [24]. It is driven by the difference between the density of the lead (11.35 g·cm⁻³ or 54.8 mmol·cm⁻³) and that of the lead dioxide (9.4 g·cm⁻³ or 39.3 mmol·cm⁻³). The tubular plate construction allows rather clear tracking down of this process using microscopy observations, such as the ones presented in Figure 11 for the case of the cycling profile “MOD4”. In the present work, the term “thickness” coincides fairly well with the internal diameter of the gauntlet, which was measured once directly and twice using a three-point circle fitting function of the digital microscope software. The results from these measurements can be further transformed into strain (or deformation) using the samples from the new cell as a reference. The evolution of the gauntlet strain (corresponding also to the electrode thickness growth) is summarized in Figure 12. Starting “back-forward” from the case corresponding to the profile with DOD = 1%, shown in Figure 12d, one can see that the corrosion process causes progressive thickness growth, which is more significant in the middle and bottom sections of the positive electrode. The data indicate that there is a critical strain threshold of the gauntlets, being in the range of 11–12%. The increase in the electrode thickness beyond this point results in gauntlet tissue rupture running along the seams. The result from this observation is marked as a dotted asymptote sketching a rapid increase in the strain (Figure 12c,d).

The evolution of the gauntlet deformation becomes more complex with the increase in the depth of discharge. This complexity can be related to the cyclic change in the electrodes’ thickness due to the periodic formation and depletion of PbSO₄, which is less dense (6.29 g·cm⁻³ or 20.8 mmol·cm⁻³) than PbO₂. The latter phenomenon is “balanced” by the tensile strength and the elasticity of the gauntlet material, which applies a sort of “dynamic” compression, preventing (or at least delaying) the occurrence of PAM softening and shedding [29,30]. As a result of this “balance”, one can see even a gauntlet shrinkage phenomenon (negative strain) located in the top section of the electrodes. The increase in the DOD to 70% results in homogenization of the strain distribution along the electrode height. Despite the observed variations, it is clear that the electrode thickness increases due to the corrosion process, resulting in a decreased gap between the positive and negative plates. The latter explains fairly the observed variations in the ohmic resistance of the cells along the ageing (Figure 7a–c). Last, but not least, it should be mentioned that the massive gauntlets rupture matches rather well the moment of discharge performance failure related to current collector spine ultimate corrosion damage.

The data in Figure 12 show clearly that the hybrid type of operation, combining higher DOD values and a smaller share of hydrogen production, will be much more beneficial in terms of cells’ lifetime in comparison with the regimes of continuous water electrolysis. Apart from this, the positive electrode’s durability can be improved by using alloys with higher corrosion resistance [2,3,6], tubular plates with thicker spines, as well as woven gauntlets with superior mechanical performance [29].

4. Conclusions

The operation of the flooded traction lead–acid battery cells with tubular positive electrodes has been studied under test protocols combining charge/discharge operation and excessive overcharge corresponding to the use of the cells in water electrolyzer mode. The whole ensemble of the results obtained under four operation scenarios showed that the lead-based battery–electrolyzer hybrid electrochemical system could be an interesting candidate for a variety of renewable energy conversion and storage applications. The latter includes emergent use cases like long-duration energy storage and opportunity purchase of cheap (or even negative-priced) electricity and its conversion into much more pricy hydrogen.

The electrochemical and electric tests showed that the efficiency of the hybrid operation increases with the increased share of the “battery function”, as it might exceed 75% under profiles combining deep cycling and intense overcharge. The monitoring of the cells’ operation indicated that the ageing process in battery–electrolyzer modes proceeds without any significant degradation of the electric performance over significant periods, covering more than 80% of the total cell lifetime. This feature makes the use of Amp-hour counting (battery history bookkeeping) a critical part of the battery–electrolyzer monitoring.

The teardown analysis of the cells subjected to hybrid cycling profiles showed that the positive grid corrosion is the main degradation mechanism, while the negative electrodes remain practically intact. The corrosion process evolution follows closely the quantity of the applied total overcharge (i.e., the quantity of the produced H₂). The increase in the depth of discharge (i.e., the “battery function”) results in a minor decrease in the corrosion rate, indicating that the hybrid lead-based battery–electrolyzer cells’ use with H₂ production as a “by-product” can be profitable, especially in emerging applications of renewable energy storage and conversion.