Three different substrates were used in this experiment to test the EC_b response on the alkaline solution for different soil types. Two of the substrates are agricultural soils, and one substrate is pure quartz sand. The Carbon Drawdown Initiative provided the soils (Carbdown, 2022). The two soil samples originate from arable land in Bramstedt in Northern Germany and Fürth in Southern Germany. Those sites are being used for long-term EW field trials, and the soil samples have been taken before applying the silicate mineral powder on those sites. The soil samples were collected from the upper 30 cm of the soil, representing a typical deeper plowing depth (Baumhardt et al., 2008).

The Bramstedt test site is situated within a fen, an area characterized by wetlands. The field was initially used as grassland and cropland the years before the soil sampling took place. The soil in this location has some peat and a considerable organic matter content, resulting in a dark color. The soil is a strong loamy sand with a clay content of 12–17 %, as determined according to DIN 19682-2 (2014).

The Fürth test site was previously used for agriculture. It has a low inorganic carbon content and much lower organic carbon content than the soil from Bramstedt (Table 1). The soil type is identified as brown earth, which is also known as Cambisol or, for sandy soils, Brunic Arenosol (Amelung et al., 2018). Both types are characterized by little or no soil profile differentiation (IUSS Working Group WRB, 2022). The brown earth can be found in Europe, North- and South America, and southern parts of Siberia (Pehamberger and Gerzabek, 2009), making the findings comparable to other European agricultural sites.

To mimic non-reactive soil with minimal nutrients as well as low organic matter and clay content, resulting in low effective CEC, plain quartz sand (commercially available pool filter sand) was used (Table 1).

Soil samples were air-dried and sieved to 2 mm to break down the soil clods, remove bigger stones, and get a homogeneous mass, destroying any natural soil horizons. Therefore, the following experiments will treat the upper 30 cm soil sample as one homogeneous material.

The soil samples for the mineralogical analysis were dried at 40°C and wet-milled. The bulk composition of the soil was measured by X-ray diffraction quantitative phase analysis (QPA, Bruker D8 Advance system).

The total organic carbon (TOC) and total inorganic carbon (TIC) were measured by splitting the soil sample into two representative parts. One part was treated with HCl to remove all TIC from the total carbon content (TC). Both samples were fused by open-system pyrolysis at 1,600–1,800°C, in which the amount of CO2/TC produced was measured by chromatography with the Carlo Erba EA1108 elemental analyzer. The TIC was then determined by subtracting the total organic carbon from the total carbon (TIC = TC − TOC).

The soil samples for the cation exchange capacity (CEC) measurement were dried at 110°C to lose adsorbed water and ground to < 500 μm. The CEC analysis is based on the exchange of the sample with Co-hexamine trichloride [Co(NH₃)₆]Cl₃ (Co-Hex) (Orsini and Remy, 1976; Ciesielski et al., 1997) solution and the subsequent spectrophotometric analysis of the exchanged liquid using a Shimadzu UV-1280.

The Co-Hex solution is added to the dried soil sample. After the exchange reaction, the suspension is centrifuged and filtered, and absorption is measured at a wavelength of 475 nm to determine the concentration of the left Co-Hex ions. The calibration set is composed of different Co-Hex concentrations and blanks. The result gives the effective CEC at ambient pH.

For the pool filter quartz sand as a non-reactive soil example, the parameters were not measured. As the Bramstedt soil sample contains more organic carbon than the Fuerth soil, it is referred to as organic-rich in the following text. It also contains more clay minerals (Table 1), making the soil overall more reactive (Churchman, 2018).

All experiments were conducted in a controlled lab environment at T = 20 °C and pCO₂ = 450 ppm, slightly higher than the average global mean atmospheric pCO₂ of 418 ppm (Lan et al., 2023). The samples of each site were dug and mixed, then dried and sieved to 2 mm; thus, there will no longer be high pCO₂ in the soil pore space.

All the different soils were watered with sodium bicarbonate solutions (NaHCO₃) with ≥99% purity; for more details, see the datasheet in the supplement. This compound was chosen for carbonate alkalinity creation because of its high solubility [96 g L⁻¹ at 20 °C, (Haynes, 2015)]. One mole of HCO3- increases the TA of the solution by one acid equivalent (eq) that can be buffered; thus, 1 mol NaHCO₃ L⁻¹ equals 1 eq L⁻¹. The concentrations of the different solutions were chosen based on previous column EW experiments conducted using the same organic-rich soil (Li, 2023). They observed TA up to 10 meq L⁻¹ in the percolating soil water of soil columns treated with silicate minerals. NaHCO₃ should simulate the released weathering products to avoid the long weathering reaction times when applying silicate rock powder to soils. In this study, the kinetics of the weathering process are not analyzed. Instead, the solutions already contain the weathering products. The solutions were produced volumetrically by weighing in NaHCO₃ powder, dissolving, and filling up to the target volume. The solutions used in the experiment have NaHCO₃ concentrations of 0, 2.5, 5, 10, and 50 mmol L⁻¹.

The EC_b sensor (Dragino LSE01) used in this study is composed of a conductivity meter to get the bulk soil electrical conductivity EC_b, a resistance temperature detector (RTD) to measure the soil temperature, and a frequency domain reflectometry (FDR) probe to measure the volumetric water content (θ) with the compensation from soil temperature and electrical conductivity. It is designed to measure θ of saline-alkali soil and loamy soil (mineral soils) and is made for outdoor agricultural use (Dragino, 2022). The sensor is not strictly research-grade because there is just a single internal calibration for θ that could not be adjusted. Still, it was used because the same sensors were used on the test sites in the Carbdown project (Carbdown, 2022) to test them for the assessment of carbon sequestration. The sensor was calibrated for a mineral soil type and not organic-rich soils. Hence, the exact θ values might be biased, and θ will not be interpreted in-depth in this study. The EC_b and θ value pairs are not compared for different soils. Nevertheless, sensors should show differences in EC_b at constant volumetric water content θ for different NaHCO₃ treatments for the same soil.

The sensor converts the measured EC_b values to the reference temperature at 25 °C, suppressing the measured values' temperature dependence. It uses the same reference temperature as the EC_W measurement of the solutions and soil water, which increases with temperature by about 1–3 % per °C (Robinson and Stokes, 2002).

To determine how the EC_b reacts to the released weathering products in the soil water (Na⁺ and HCO3-), air-dried soil samples were watered with different NaHCO₃ concentrations and purified water as a non-alkaline reference (Figure 1). The EC_b, θ and the soil temperature were measured with the Dragino LSE01 probe. In addition, the EC_W and the pH of the input solution were measured with the WTW Multimeter 3630 IDS. For further details about the used probes, see Table 2.

For each concentration/experimental run, a container (see Figure 1) was filled with 1.5 L (around 1,700 g) of sieved air-dry soil to ensure a large enough volume for the sensor's electrical field penetration depth and physical size. This sensitive measurement volume of the sensor can be approximated by a cylinder with a height of 10 cm and a 7 cm diameter around the central probe (Dragino, 2022).

The containers were then filled with the NaHCO₃ solution in 75 g steps until reaching the soil saturation point, starting from 500 g of added solution for the used soils. After each irrigation step, the soil was mixed manually with a spatula to homogenize the soil water mix. Ten measurements in different sensor positions were taken to minimize the effects of soil heterogeneity and sensor-soil coupling.

The brown earth from the Fuerth test site was chosen to investigate the reactions of the alkaline NaHCO₃ solutions with the soil more closely. It is the soil with a much lower organic matter content; therefore, one would expect a weaker pH buffering reaction. The water of the saturated paste was extracted from this soil. The chemical analysis of the soil water extracts gives a better understanding of the observed EC_b signal, as the EC_b is mainly a function of the dissolved ions in the soil water (Hendrickx et al., 2002; Corwin and Lesch, 2005). Air-dried soil samples were saturated with the different NaHCO₃ solutions. The method used closely follows the method presented by Corwin and Yemoto (2020) for measuring soil salinity.

For each reaction setup, 1,700 g of air-dried soil was mixed with 485 mL of NaHCO₃ solution. This ratio was kept constant for all treatments. Therefore, the brown earth was watered with the produced NaHCO₃ solution until it reached the saturation point. Following Corwin and Yemoto (2020), it is the point where the soil water mix: (i) surface sparkles, (ii) can flow, (iii) slides freely off a spatula, (iv) is malleable so that when running a spatula through the mix the channel will hold shape and (v) when shaking the container the channel collapses. The soil water mix is called the saturated paste at this saturation point.

Even though it is faster and more effective to extract the soil water at higher water content like 1:1, 1:2, and 1:5 (soil to water ratio) in this experiment, the low saturation level extraction will be used. It is the standard for EC studies as it provides the lowest but sufficient water volume to be analyzed for its chemical composition (Corwin and Yemoto, 2020). It is closer to the natural moisture processes in actual soil. In our case, it represents the actual conditions in the soil of the (θ, EC_b) measurements of the EC_b experiment setup.

To measure the range of responses the brown earth was irrigated with the same solutions as in the previous experiment of 0, 2.5, 5, 10, and 50 mmol L⁻¹ of NaHCO₃.

There are three duplicates to capture the variability and get more reliable results for each NaHCO₃ treatment. In total, 15 saturated paste samples (5 NaHCO₃ solutions times 3 duplicates) except a third one for the high concentration of 50 mmol L⁻¹ due to the lack of further extraction equipment. After saturating the samples with the solutions, they were stored for 24 h in the fridge inside containers covered with snap-tight lids to avoid evaporation from the sample. In this period, the soil and the solution have time to react, and dissoluble compounds can dissolve in the soil water. After the time had passed, the saturated paste was placed outside the fridge until it reached room temperature again. Then, the soil water of the saturated paste, called saturation extract, was pumped out with vacuum pumps (see Figure 2). The extraction took 1 day, as the soil holds very little free water at the saturation point, and it takes longer to pump it out. After applying the vacuum for 1 day, the saturation extract was directly analyzed for the chemical composition; see Table 2 for further details.

All the soil water samples were filtered with 0.45 μm cellulose acetate (CA) dead-end filtration. All measurements except the EC_W and pH were measured in the filtered samples. The EC_W probe applies an internal linear temperature compensation to 25°C. All shown EC_W values thus are normalized to 25°C.

All three different substrates have in common that the EC_b increases with θ (Figures 3–5). Most EC_b(θ) plots have a curvilinear shape with the slope changing with θ. Considering the purified water treatments [NaHCO₃] = 0 mmol L⁻¹ together with the [NaHCO₃] = 50 mmol L⁻¹. The slope dECbdθ is increasing with higher solution concentrations since this also increases the electrical conductivity of the soil water (EC_W). All the characteristics mentioned above can be described with the EC_b equation from Rhoades et al. (1976):

With a and b as empirical fitting constants for the soil, surface electrical conductivity (EC_S), and the soil water electrical conductivity (EC_W).

When considering plausible TA levels of up to 10 meq L⁻¹, based on leachate water from soil EW experiments by Li (2023) for the same organic-rich soil (see Figure 5), a reliable separation of EC_b, in terms of the different NaHCO₃ treatments, is not possible. The initial concentration of the NaHCO₃ inputs cannot be derived anymore because the resulting EC_b(θ) curves are almost equal for low-concentrated NaHCO₃ solution additions. Additionally, the considered high-water contents θ are rare in natural soils and are probably just present after heavy rainfall. Outdoors, the soil moisture is mostly at field capacity (FC) (McCauley et al., 2005). That implies typically smaller θ, so a regression analysis's determination of the different NaHCO₃ treatments is even more limited. However, this experiment also showed distinguishable EC_b(θ) curves when using pure pool filter quartz sand as substrate (see Figure 3), as the sand conserves the water's chemical properties.

The brown earth and the organic-rich soil buffered a considerable part of the introduced TA (see Figure 9) as the soil reacted with the NaHCO₃ solution. The EC_b represents how the soil reduced the introduced TA, thereby lowering the EC_b signal. Apart from the initially different EC_W of the NaHCO₃ solutions, the EC_b(θ) curves look very similar. However, when the NaHCO₃ addition and thus the concentration in the soil water is big enough, it can be detectable with EC_b and θ.

The Dragino LSE01 sensors are not calibrated precisely enough to accurately measure soil volumetric water content (θ) for all soil types. A specific calibration needs to be done for each soil type to achieve a more precise θ reading. It's important to note that the mineral soil type calibration used may not be suitable for organic-rich soil samples.

Different soils and materials have varying abilities to absorb water. Coarse sand, for example, consists mainly of macropores, while other soils contain finely textured micropores. Additionally, saturation levels and dissolution reactions may change over time, so one needs to consider the time dynamic component for similar experiments. The soil needs sufficient time to absorb the added solution.

The biggest driver of fluctuations in the chemical composition of the soil solution samples is the low water content of the soil. The extracted soil solution sample volume can vary significantly. Therefore, it is recommended to use a different method for ongoing experiments. One approach involves increasing the water content significantly (1:1, 1:2, or 1:5 soil: water ratios) to homogenize everything. A centrifuge can then be used to extract the solution with better efficiency. The vacuum pumps in this experiment might have allowed for preferential pathways and led to different extraction sample volumes. Nevertheless, the low water content was chosen in this experiment to keep it close to the electrical conductivity observed outdoors in the field.