In smart building environments, especially those relying on distributed wireless sensor networks for HVAC and air quality control, battery selection plays a pivotal role in ensuring system reliability and minimizing maintenance overhead. Many devices require short-duration high pulses for wireless data transmission and two-way communications (Guida et al., 2023; Google Scholar, 2024). LiSOCl₂ batteries are the ideal choice for powering wireless sensors within smart infrastructure—including wireless HVAC sensors, high-voltage direct current (HVDC) sensing units (Islam et al., 2016), low-power IoT nodes, security and occupancy detectors, structural health monitors (Jams et al., 2022), and adaptive lighting or ventilation controllers (Gu et al., 2020). LiSOCl₂ batteries exhibit exceptionally low self-discharge characteristics, which is fundamental to their suitability for long-term sensor deployments. Quality LiSOCl₂ batteries typically maintain self-discharge rates of less than 1% per year (T. Batteries GmbH, 2025; Hagedorn 2020), allowing them to retain their charge for extended periods and ensuring reliable operation even after years of inactivity or intermittent use. This characteristic is particularly critical for applications where sensors may remain dormant for extended intervals or must function reliably throughout long deployment cycles (International Energy Agency, 2024).

These batteries are available in both spiral-wound and bobbin-type configurations, with the latter design achieving even lower self-discharge rates—as low as 0.7% annually—due to the beneficial passivation effect (Islam et al., 2016). However, while passivation significantly enhances shelf life, its cumulative effects combined with exposure to extreme temperatures can develop gradually over years, potentially leading to underestimated battery life in predictive models if not properly accounted for. It is crucial to recognize that battery quality varies significantly among manufacturers. Lower-grade LiSOCl₂ cells may exhibit annual self-discharge rates of up to 3% —three times higher than quality variants (Ave, 2001). At this elevated self-discharge rate, batteries can lose up to 30% of their total capacity over a 10-year period (Google Scholar, 2025), making the advertised 40-year operational lifespans unrealistic and emphasizing the importance of selecting high-quality cells for critical long-term applications. Building on these considerations, this paper will shed more light in this direction by providing a detailed comparison of LiSOCl₂ batteries from different manufacturers, by evaluating key performance metrics such as constant discharge rates, voltage stability, and capacity retention. Beyond their excellent charge retention, LiSOCl₂ batteries demonstrate remarkable temperature tolerance, functioning effectively across extreme temperature ranges from −55°C to +85°C, with specialized variants capable of operating at temperatures up to 125°C or below −80°C (DTIC, 1991; Schlaikjer, 1985). This wide operational envelope makes them exceptionally robust against the environmental fluctuations commonly encountered in building infrastructure and HVAC systems. The batteries also provide consistent voltage output throughout most of their discharge cycle, which is essential for the stable operation of sensitive wireless sensors and ensures accurate data transmission and reliable device performance over time (Roth et al., 2025). Combined with their high energy density, these characteristics enable compact sensor designs—a crucial advantage in modern smart buildings where space constraints and aesthetic integration are paramount considerations.

Manufacturers specify battery performance under ideal conditions, but actual usage environments often differ. These batteries should have the ability to handle both low-power standby modes and occasional high-current pulses, which is ideal for long-term IoT sensor applications with wireless data transmission. The profile in Figure 1 demonstrates a real current profile, with initial low-power standby mode (5 µA), followed by a moderate activation phase (10 mA), a brief high-current pulse (100 mA), and a return to ultra-low standby current (5 µA). Such discharge behavior is crucial for devices requiring periodic data transmission, sensor activation, or wireless communication.

However, this work focuses exclusively on constant discharge currents (1 mA, 10 mA, 30 mA, and 100 mA) to ensure a controlled and repeatable evaluation of battery performance. Working with constant currents allows us to directly compare different battery brands under standardized discharge conditions, validate manufacturer specifications, and assess efficiency, capacity, and degradation trends systematically. Perhaps, once these baseline characteristics are well understood, future studies can extend the analysis to variable current profiles to simulate real-world usage more accurately and investigate dynamic load effects on battery performance, which the authors intend to present in their future works.

Within the lithium family there are a number of primary (non-rechargeable) chemistries shown in Table 1, including iron disulfate (LiFeS₂), lithium manganese dioxide (LiMnO₂), lithium thionyl chloride (LiSOCl₂), and lithium metal-oxide (Morrison and Marincic, 1993; Katırcı et al., 2024; Stern, 2021). Of all these choices, lithium thionyl chloride (LiSOCl₂) batteries are overwhelmingly chosen for long-term deployments because they deliver the highest capacity and highest energy density of all lithium cells to support product miniaturization. LiSOCl₂ batteries can be manufactured in two ways, using a spiral wound or bobbin-type construction. Spiral wound cells have a larger surface area for higher rate energy flow, while bobbin-type cells have less surface area to maximize the passivation effect (Guida et al., 2023; Google Scholar, 2024; Horn and Shao-Horn, 2003). Bobbin-type LiSOCl₂ cells also feature an incredibly low self-discharge rate as low as 0.7% per year, largely due to harnessing the passivation effect, enabling certain low-power devices to work for up to 40 years on the original battery.

In recent years, a lot of research on the long-term storage of batteries has been mostly aimed at lithium-ion secondary batteries rather than primary batteries (Miles, 1997; Zabara and Ulgut, 2020; Ave, 2001), and the existing life prediction models are mostly developed based on lithium-ion secondary batteries. There are few studies (Zabara et al., 2021; Gabano, 1980) on the life prediction model of lithium primary batteries. At present, research on the anode of a LiSOCl₂ battery is mainly aimed at the voltage lag phenomenon of the battery and the lithium chloride passivation film that appears on the anode lithium, which leads to the low initial discharge voltage platform of the battery (Morrison and Marincic, 1993; Katırcı et al., 2024; Stern, 2021). During the storage of the battery, lithium metal will contact SOCl₂, and a redox reaction will occur, resulting in a lithium chloride passivation film insoluble in electrolytes. The transfer of electrons and the flow of electrolytes will be hindered by this passivation film (Zabara and Ulgut, 2020; Wang et al., 2023; Stern, 2021).

When an ultra-long-life power source is essential it is important to conduct thorough diligence when comparing competing battery brands. This evaluation process includes the need for all prospective battery manufacturers to provide fully documented test results, along with in-field performance data under similar loads and environmental conditions along with multiple customer references.

Among the prominent manufacturers of lithium thionyl chloride (Li-SOCl₂) batteries, EVE (ER14250), SAFT (LS 14250), TEKCELL (SB-AA02), and TADIRAN (SL-750) are chosen for this study and their respective characteristics (Evemall, 2025; Saft4U, 2025; T. Batteries Gm bH, 2025; Vitzrocell, 2025) are presented in the table below. The selected batteries from EVE, SAFT, TEKCELL, and TADIRAN were chosen as they are cost effective and easily available in the European market (Wang et al., 2023; Jain, 1998). Additionally, they offer closely matched datasheet parameters, enabling a fair and controlled comparison under identical test conditions. This approach allows the authors to isolate and examine subtle yet significant performance differences that may not be apparent from manufacturer specifications alone. These brands are also widely used in real-world deployments, especially in wireless sensor networks for smart buildings (Saunders, 1998; Bryzgalov et al., 2013; Sun et al., 2021), ensuring that our findings are directly relevant to current industry practices. Additionally, by including manufacturers from different regions, the study provides a broader perspective on global battery performance standards and highlights potential quality variations across the international market.

While EVE, TEKCELL, and SAFT batteries demonstrated higher capacities (1.2 Ah) and high discharge currents (upto 100 mA), TADIRAN offered slightly lower capacity (1.1 Ah) with a maximum discharge current of 100 mA (Zabara et al., 2021).

Commercial batteries often specify their capacity and performance at standard discharge rates, such as 20 mA, 10 mA, and 1 mA. Figure 2 consists of the three graphs illustrating the voltage discharge profiles of four different LiSOCl₂ batteries (EVE, SAFT, TEKCELL, and TADIRAN) under varying discharge currents. The curves demonstrate voltage stability over time, followed by a steep voltage drop near the end of battery life, characteristic of lithium primary batteries. The differences between these batteries become evident in their lifespan, discharge stability, and overall performance at different loads.

In order to represent typical functional scenarios and to provide discharging curves comparable to datasheet values, four discharge currents are chosen, 100 mA and 30 mA for high-load scenarios, simulating relatively heavy usage or wireless transmission, 10 mA for medium-load scenario, representing moderate usage during active operation, and 1 mA for low-load scenario, simulating very light or idle usage but not deep sleep modes where the capacity might be as low as few µA. Also, manufacturers of these commercial batteries provide capacity value and performance characteristics at standard discharge rates, often around 20 mA, 10 mA, and 1 mA. Testing at these discharging currents values allows for direct comparison with manufacturer specifications estimated from datasheet. The experiment setup is repeated for 1 mA, 10 mA, 30 mA and 100 mA discharge current values. In order to get a credible reproducible result, five cells from each of the battery types were tested, so 25 cells were set up for every experiment. In total 100 batteries were used in this study.

Circuit assembly is done as shown in the schematic in Figure 3. Figure 3a shows the schematic used in the experiment and Figure 3b shows a photo of the experiment setup. An INA219 circuit is used to measure the current through resistor R1. The sensor is interfaced with an ESP32-S2 microcontroller for data acquisition. A MOSFET is employed to control the circuit’s operation, with its gate connected to an operational amplifier (op-amp) for signal amplification and driving. Resistors R3(27 kΩ) and R4 (62 kΩ) form a voltage divider to stabilize the voltage at the op-amp input to define discharging current. The circuit powers the op-amp with a 5 V source and the ESP32-S2 with a 3.3 V supply.

In the given circuit, the operational amplifier (op-amp) is configured to maintain a constant voltage of 1 V over the R₂ resistor. This is typically achieved using a feedback loop where the op-amp adjusts the output to ensure the voltage at the inverting input remains at 1 V. If a known resistance R₂ is used, maintaining 1 V across R₂ allows for precise current control using Ohm’s Law. The regulated discharging current is calculated using the following Equation 1. I = 3.3 V R 2 × 27 K 27 K + 62 K (1)

The specific required discharging current is set by the value of resistor R2. The specific R2 values for each discharging currents are given in Table 3 below. The precise value of the current is measured by voltage over shunt resistor R1 by INA219. The values of R1 and R2 are tabulated below.

Multiple identical circuits (as shown in the experimental photo in Figure 3b) were replicated on breadboards for testing. Each breadboard setup included the same components to ensure consistency across measurements. The ESP32-S2 microcontroller is programmed to communicate with the INA219 circuit via I2C, a code is written to log current and voltage data, which is crucial for evaluating battery performance and circuit behavior. The circuit configuration allows for real-time monitoring of battery discharge characteristics, including current capacity and voltage drops.

Since these batteries have a relatively flat discharge profile, their voltage drops sharply after 2.5 V, making operation below 2 V unreliable. This 2 V cut-off voltage for LiSOCl₂ batteries as a chosen threshold also aligns with manufacturer specifications and ensures compatibility with embedded systems and industrial applications. Additionally, setting 2 V as the limit prevents power instability, avoids unexpected shutdowns, and maintains safety. These configurations ensured accurate current regulation corresponding to each discharge rate.

The first discharging test is done for 1 mA which lasted for about 1,200 h. The second test is done for 10 mA constant current discharge, which needed around 90 h to discharge until cutoff voltage is reached, the third test with 30 mA lasted for 26 h duration and the last test with 100 mA lasted for 3 h duration. Figure 4 shows EVE, SAFT, TEKCELL, and TADIRAN—at four different discharge currents: 1 mA (plot a), 10 mA (plot b), 30 mA (plot c), and 100 mA (plot d).

After conducting extensive experiments and collecting data on the battery capacity under different conditions, box-whiskers plot for comparison of different capacities of selected types of batteries is plotted. Figure 5 visually represents the total energy capacity (in Wh) of four battery types (EVE, SAFT, TEKCELL, and TADIRAN) at different discharge currents: 1 mA, 10 mA, 30 mA, and 100 mA.

Further statistical evaluation is conducted and summarized in Table 4 to help in understanding the different battery types exhibiting unique characteristics in terms of their energy capacity and stability under different loads. By analyzing the mean, standard deviation, sample variance, and other key statistical metrics, we can determine which battery is the most efficient and reliable.

This analysis demonstrates that the experimental values often diverge considerably from the manufacturer’s stated values, indicating variations in performance under actual operating conditions as seen in Table 5. Total energy capacity is calculated for each curve from manufacturer data and is given by the following Equations 2, 3. E = ∫ P d t (2)

Since P = V × I (3)

Where V is the voltage at each time step and I is 0.1, 0.03, 0.010 and 0.001 A respectively, the total energy can be approximated numerically as shown in the Equation 4 given below. E ≈ ∑ V i + V i + 1 2 × I × t i + 1 − t i (4)

Once the total energy is found, then the average is calculated among the batteries of the same type and is tabulated in the table given below.

Table 5 compares the four batteries at different discharge currents, evaluating their actual energy output and percentage difference from the manufacturer’s claimed values.

The comparison of four leading LiSOCl₂ battery brands—EVE, Saft, TEKCELL, and TADIRAN—reveals critical insights for optimizing power circuit design in sensor networks within smart building environments. Notably, we observed significant discrepancies between manufacturer specifications and actual performance under real-world conditions, particularly at varying discharge currents. These deviations directly impact the reliability and efficiency of energy budgeting, necessitating a shift toward empirically informed design choices for sustainable building automation systems.

Performance at Low Discharge Current (1 mA): At a discharge current of 1 mA, the batteries operate close to their manufacturer’s rated capacity, with only slight deviations. EVE delivers 3.78 Wh against a claimed 3.8 Wh, showing a minor 0.53% deviation—making it highly efficient. Saft shows a larger deviation of 4.44%, producing 4.08 Wh instead of 4.27 Wh. TEKCELL drops significantly from its claimed 4.07 Wh to an actual 3.29 Wh, indicating a considerable performance gap. TADIRAN performs the worst, providing 2.77 Wh against a claimed 3.91 Wh, suggesting that it struggles even at low discharge rates. This alignment, especially in the case of EVE, indicates that the methodologies employed by both the manufacturer and our research are highly consistent, validating the accuracy and reliability of our testing approach. The close match between empirical results and manufacturer specifications (Evemall, 2025) confirms that our experimental design effectively mirrors real-world operating conditions and manufacturer testing protocols.

Design Considerations for Minimizing Power Peaks: To reduce energy consumption and extend operational life, it is important to focus on minimizing power peaks (Lallart and Guyomar, 2008), (Kaushalya et al., 2024) and decoupling battery management circuit design from transient current surges (Zhang et al., 2017). By targeting the root causes of these peaks within the power circuit, engineers can mitigate the stress on batteries, thereby enhancing overall system stability and lifespan. This approach is particularly effective when combined with low-current duty-cycling schemes that maintain discharge levels below manufacturer testing thresholds.

Medium-Low Discharge Current (10 mA): At 10 mA, the energy capacity begins to decline across all batteries. EVE performs slightly better than expected (2.94 Wh vs 2.90 Wh), showing a small 1.38% positive deviation. Saft sees a drop from 3.11 Wh to 2.85 Wh, indicating some early efficiency loss. TEKCELL again shows a major drop, from 3.55 Wh to 2.59 Wh, reinforcing its inefficiency at moderate loads. TADIRAN struggles with another steep decline from 2.84 Wh to 2.37 Wh, further confirming its lower energy retention. These observations highlight the importance of designing power circuits that can handle moderate loads efficiently, as even slight increases in discharge current can lead to noticeable capacity reductions.

Informed Battery Selection and Power Management Architecture: The observed variances in discharge capacity across different brands and current levels provide a robust dataset for guiding battery selection tailored to specific application needs. For ultra-low-power scenarios typical in smart buildings (Bhat and Sudharshana, 2024), such as occupancy detection and indoor air quality monitoring (Bhaskar et al., 2025), selecting batteries with strong performance at sub-milliampere discharge currents is essential. This selection process supports the development of robust power management architectures (Zhang et al., 2017) that go beyond idealized assumptions, offering real-world reliability.

Medium Discharge Current (30 mA): At 30 mA, the efficiency loss becomes more pronounced. EVE provides 2.17 Wh, indicating that it retains a relatively higher energy output under moderate loads. Saft drops to 1.81 Wh, a substantial decrease from its manufacturer’s expected 1.99 Wh. TEKCELL shows 1.12 Wh at 30 mA, with the manufacturer value expected at 2.4 Wh at 18 mA. TADIRAN performs at just 0.79 Wh at 30 mA, while the rated value was 1.98 Wh at 20 mA. These findings highlight the necessity for engineers to consider the specific discharge profiles of batteries when designing power management systems (Niharika, 2025), ensuring that the selected batteries can sustain the required loads without significant efficiency losses.

Integration with Energy Harvesting Systems: The superior low-current performance of specific LiSOCl₂ brands positions them as ideal components in hybrid energy systems (Malele et al., 2025), where primary batteries are supplemented by ambient energy sources such as light, vibration, or thermal gradients. In such configurations, these batteries act as reliable energy reservoirs during periods when harvesting conditions are unfavorable.

High Discharge Current (100 mA): At 100 mA, EVE still holds up relatively well at 0.61 Wh, showing that it can sustain high discharge rates better than competitors. Saft follows closely at 0.58 Wh, maintaining similar performance to EVE. TEKCELL and TADIRAN become completely inefficient, both dropping to 0.13 Wh, making them unsuitable for high-drain applications. This data is crucial for designing circuits capable of dynamically adjusting sampling rates and transmission intervals based on real-time energy availability, which is vital for predictive energy management strategies that extend network lifetime without compromising data quality.

Advancing Sustainable Building Design: The deployment of maintenance-free, long-lasting power sources within sensor networks aligns directly with the principles of sustainable building design. By enabling sensor nodes to operate autonomously for decades, our approach reduces both material waste and the operational burden of maintenance—particularly in hard-to-access infrastructure locations. These sensor systems enable real-time adaptive control of HVAC, lighting, and other building systems, resulting in measurable reductions in energy usage.

The paper provides a comprehensive comparative study of Lithium thionyl chloride (LiSOCl₂) batteries under controlled discharge conditions (1 mA, 10 mA, 30 mA, and 100 mA). The results presented from the tests show big differences among the batteries from different manufacturers, which would result in different service life if inbuilt in the same IoT device. Also, in the case of TEKCELL and TADIRAN batteries the tests show difference between manufacturers datasheet values and real discharging curves in terms of overall energy.

When comparing measured vs manufacturer data, across all currents, there are noticeable deviations between manufacturer-specified values and experimental results. SAFT and EVE maintain closer alignment with rated capacities, while TEKCELL and TADIRAN show significant discrepancies. TADIRAN consistently exhibits the lowest total energy consumption, with greater variation at lower currents. This aligns with the manufacturer’s specification, which states that TADIRAN has a lower total energy capacity of just 1.1 Ah compared to other batteries that have 1.2 Ah.

In the statistical evaluation EVE leads in terms of mean energy capacity (2.38 Wh), followed by SAFT (2.33 Wh). SAFT has the highest standard deviation (1.49 Wh), indicating more variability in performance. TEKCELL and TADIRAN exhibit lower efficiency and higher energy dissipation under load. The results underscore that SAFT and EVE are superior choices for high-drain applications, while TEKCELL and TADIRAN may be more suitable for low-drain, intermittent use.

Our findings directly support the development of smarter energy storage strategies and improved energy management design in the following ways:

• Informed Battery Selection: By empirically comparing four leading LiSOCl₂ battery brands under controlled conditions, we provide actionable data for selecting batteries that deliver optimal performance and longevity in real-world smart building applications.