Introduction

Lithium has become a critical element in various industries, particularly in the production of rechargeable batteries for electric vehicles, portable electronics, and energy storage systems. In these applications, lithium in the form of a compound is deposited on an aluminum collector that serves as a cathode due to its ability to charge and discharge reversibly1,2,3,4,5. As the demand for cleaner, safer, and more efficient energy solutions increases, the need for reliable and accurate measurement of lithium in different sources has grown substantially. Moreover, the electric car industry is expected to continue expanding, and according to it, the demand for lithium6,7,8,9,10,11,12.

The main Li-bearing clay minerals in nature are lepidolite [K(Li, Al)3(Al, Si, Rb)4O10(F, OH)2]13 (polylithionite [KLi2AlSi4O10F2]14, trilithionite [KLi1.5Al1.5(Si3Al)O10F2]15), zinnwaldite [KAl(Fe, Li)(Si3Al)O10F2]16, masutomilite [KLiAlMn2+(Si3Al)O10(F,OH)2]17, swinefordite [Ca0.2(Li, Al, Mg, Fe3+, Fe2+)3(Si, Al)4O10(OH, F)2·nH2O or Ca0.2(Li, Al, Mg, Fe)3(Si, Al)4O10(OH, F)2·nH2O]18, cookeite [LiAl4(Si3Al)O10(OH)8 or (Al, Li)3Al2(Si, Al)4O10(OH)8]19, jadarite [LiNaB3SiO7(OH)]20 and hectorite [Na0.3(Mg, Li)3Si4O10(OH)2 or Na0.3(Mg, Li)3Si4O10(F,OH)2·nH2O]21,22,23. They have a well-known structure and are widely distributed in Asia and America, recently becoming a new Li resource22,24. The high Li concentrations are believed to result from hydrothermal alterations associated with saline lake waters (brine)25,26. This option has not yet been used for lithium extraction, but some countries are currently in various stages of research on its use27.

Lithium determination in soil samples has been an attractive challenge for analytical chemistry as well as geological and environmental sciences. The amount of lithium in batteries, soils, and pharmaceuticals is crucial for industries such as technology, extractive, and health care. Its importance can reach the climatologic disciplines since the isotopic composition of lithium in clay and limestone can be applied in paleoclimatologic research28. The determination of lithium in several geological samples, including granite, limestone, basalt, loess, and clay, has been reported, although studies on lithium-rich clays have scarcely been reported29. 7Li and 6Li isotope ratios have been used as tracers for weathering rocks and soils28. Zhang et al.30 evaluated the Li isotope behavior in granite from the eastern Tibetan Plateau, China, where lithium content gradually diminished with weathering. On the other hand, Tsai et al.31 used the Li isotopes as tracers to locate the origin of the loess and assess the local weathering in the southern Chinese Loess Plateau. In geochemical and paleoclimatological research, lithium isotope ratios are usually determined by Multi-collector ICP-mass spectrometry (MC-ICP-MS), a highly precise method for the isotopic analysis of lithium that requires a enough concentrated target element purified sample and free or very low matrix elements28,32. This method was applied to quantify lithium isotopic composition in granite, upper crustal composites, sedimentary rocks33, while a variation of MC-ICP-MS method was conducted by Li et al.32 and allowed the isolation of lithium from geological reference materials such as Limestone NIST SRM 1, Coral JCP-1, granodiorite GSP-2, continental flood basalt BCR-2, ocean island basalt BHVO-2, marine manganese nodule NOD-A-1, basalt JB-2, granite JG-2, and kaolinite KGa-2. However, the literature about the worldwide occurrence of lithium-rich clay deposits is scarce, as pointed out by Lit et al.29 in their study of clay deposits from the south of China, in contrast to the recent estimations of lithium-rich clay deposits in Nevada, United States and Sonora, Mexico.

Among extraction methods, the classical base-acid-carbonate method is considered the main technique used to extract lithium from minerals. Also, microwave-assisted acid digestion has recently been recognized as an effective technique for extracting lithium from low-content clay minerals, like hectorite clay, offering a complementary approach to the traditional method34,35.

Given the increasing importance of lithium, accurate chemical measurements are essential for promoting the economic development of mining activity. For effective quality control and quality assurance of lithium quantification methods, it is crucial to have a metrological reference material that provides traceability to the International System of Units (SI)36. This need underscores the importance of developing a certified reference material (CRM) for lithium, particularly in clay matrices, due to the diversity of lithium sources and the complexities associated with its exploration, extraction, and processing.

The development and certification of reference materials involve complying with the technical and production requirements in reference materials (RMs) in accordance with the ISO 17034:2016 Standard37, which establishes the general requirements for the competence of RM producers, as technically the most important are the production of CRM, homogeneity and stability studies, characterization, assignment of property values and their uncertainties. Among the requirements that CRMs must meet are the evaluation and monitoring of homogeneity and stability, which are properties that define the specific conditions to determine the useful life of CRMs and ensure the reliability of the certified values and associated uncertainties over time. To evaluate homogeneity and stability, the guidelines established in the ISO-Guide 35:201738 Guidance for characterization and assessment of homogeneity and stability.

Homogeneity refers to the consistency of a property value within and between separate material units, which is crucial for ensuring reproducible measurements. In this work, a subset of units (10 to 30) of the batch was selected for a homogeneity study using a stratified random sampling scheme. The property values for each unit were measured using a repeatable measurement procedure. The results were evaluated using appropriate statistical methods to obtain information on between-unit and within-unit homogeneity39,40.

Stability is the ability of an RM, when stored under specified conditions, to maintain a declared value of a property within specified limits for a given period. Stability is determined through short-term studies (isochronous) to establish transport conditions and long-term studies (classical) to determine storage conditions and validity periods41,42. Evaluating the stability of certified properties is essential because property values can change over time for various reasons, to different degrees, and at different rates. Thus, assessing the effects of storage and transport conditions on RMs is necessary to define the best conditions to ensure their stability and intended use43,44,45,46,47.

The studies carried out in this paper required the measurement of Li using FAAS, where other siliceous and calcareous clay materials were used to validate the analytical method. For validation of ISO guidelines, CENAM has been demonstrating their CMCs in category 13, which correspond to CRM of elements in ores in the frame of CIPM-MRA (see more information in the Additional information S4)48.

Previous studies have emphasized the importance of CRMs in analytical chemistry, but the lack of a CRM for lithium in clays has left a significant void. This study addresses this gap with the primary objective of developing and conducting rigorous testing to certify a traceable novel reference material (RM) for lithium in clays, specifically hectorite, in accordance with ISO Guide 35:2017 standards38. This RM aims to meet the needs of various industries by providing a reliable standard for lithium measurement, thereby enhancing the accuracy and reliability of chemical analyses supporting the efficient exploration, exploitation, and handling of lithium ore. To achieve this, we focused on evaluating the homogeneity and stability of lithium in clay RMs, which implies determining the minimum amount of sample necessary to ensure reproducible results of the lithium mass fraction w(Li), and establishing the best temperature conditions for the proper handling of RM in its transport, storage, and period of validity to ensure a sufficiently homogeneous and stable RM concerning specified properties, to provide traceability to the SI.

The integration of recent advancements in lithium measurement techniques and adherence to ISO 17034:201637 guidelines ensures that the developed CRM will address current limitations and set a new standard for quality control and quality assurance in lithium quantification. This comprehensive approach not only enhances the reliability of analytical results but also supports the sustainable and efficient management of lithium resources, thereby contributing to the broader goal of advancing cleaner energy solutions.

Materials and methods

Sample

The candidate CRM of lithium clay, referenced as CRM-6200752a, was provided by the National Metrology Center (CENAM)49.

Certified reference material (CRM)

For calibration purposes, the SRM-3129a lithium (Li) Standard Solution (80 mg/kg) was obtained from the National Institute of Standards and Technology (NIST)50.

Reagents

The following reagents were used in the study: Nitric acid (HNO3, 69–70%, J.T. Baker®), hydrochloric acid (HCl, 36.5–38%, J.T. Baker®), hydrofluoric acid (HF, 48–51%, J.T. Baker®), hydrogen peroxide (H2O2, 30%, J.T. Baker®), boric acid (H3BO3, granular, J.T. Baker®), magnesium perchlorate (Mg(ClO4)2, Sigma Aldrich®), and deionized water (18 µS/cm).

Instruments

The instruments utilized for sample preparation and analysis included: an analytical balance (Sartorius MSA225S), a hot plate, and a laboratory furnace (Thermo Scientific) were used for sample preparation. A microwave digestion system (MARS 6TM & MARSXpress, CEM Corporation) with Teflon (XP-1500) vessels was used for microwave closed-vessel digestion. A flame atomic absorption spectrometer (FAAS) (AAnalyst™ 800, Perking Elmer® & WinLab32™ control software) was used to measure Li in the clay samples.

Sample selection

Ten bottles of candidate material were selected using a randomly stratified sampling method for homogeneity studies, and an additional one bottle was taken for stability studies.

Sample preparation

Three subsamples with masses of 0.05, 0.1, and 0.25 g of the candidate material were weighed into a Teflon high-pressure digestion vessel (TFM) for homogeneity studies. For stability studies, 0.1 g subsamples were used.

The subsamples were digested using a microwave-assisted acid digestion method in a closed digestion system with temperature and pressure control using a microwave Mars system.

Firstly, 6 mL of HNO3, 3 mL of HCl, and 2 mL of HF were added to each sample. The digestion process was performed in three stages: (i) a pre-digestion, ii) a first digestion stage, and iii) a second digestion stage. Once the first digestion process was completed, 2 mL of H2O2 was added. Digestion specifications for homogeneity and stability tests are described in Table 1.

Table 1 Digestion specifications for homogeneity and stability tests.
Full size table

Finally, 1 mL of H3BO3 was added to the subsamples digested to neutralize excess HF. The subsamples were evaporated on a hot plate and redissolved to a total volume of 30 g using deionized water. Subsequently, a dilution process involved taking 5.3 g from each subsample, which were redissolved until 20 g of 2 vol.% of HNO3 solution was obtained.

Instrument conditions

The prepared subsamples were measured in duplicate to determine the lithium mass fraction w(Li) using an external calibration method with the FAAS at a wavelength of 670.8 nm.

Mathematical model for lithium mass fraction measurement

The lithium mass fraction w(Li) was measured by external calibration with the FAAS using the following mathematical model, as shown in Eq. (1).

$$w\left(Li, x\right)=\left[w\left(Li\right)-{w}_{ref}\right]\cdot \frac{{w\left(Li\right)}_{CRM, meas}}{{w\left(Li\right)}_{CRM, prep.}}\cdot \frac{{m}_{af1}}{{m}_{x}}\cdot \frac{{m}_{af2}}{{m}_{al}}$$
(1)

where, \(w\left(\text{Li},\text{ x}\right)\) is the mass fraction of lithium in-clay (mg/kg), \(w\left(Li\right)-{w}_{ref}\) is the mass fraction of lithium in the calibration curve (mg/kg), \({m}_{x}\) is the clay mass (g), \({m}_{af1}\) is the first dilution calibration mass (g), \({m}_{al}\) is the aliquot mass in the second dilution (g), \({m}_{af2}\) is the second dilution calibration mass (g), \({w\left(\text{Li}\right)}_{\text{CRM},\text{ meas}.}\) is the CRM lithium mass fraction measured, and \({w\left(\text{Li}\right)}_{C\text{RM},\text{ prep}}\) is the CRM lithium mass fraction prepared.

Homogeneity and stability studies

ISO 17034:201637 establishes that a reference material must be homogeneous and stable. However, the evaluation of homogeneity and stability is carried out based on the guidelines established in ISO-Guide 35:201738, so the tests carried out in the evaluation of homogeneity and stability, as well as their associated uncertainties, of the lithium clay reference material, were based on what is described in said guide:

Homogeneity test

The homogeneity test involved analyzing ten bottles (identified as B001, B042, B054, B076, B114, B153, B177, B208, B231, and B256) that were sampled randomly from the batch of clay. Two independent subsamples were taken from each bottle and three experiments were conducted on all bottles and subsamples. Each experiment involved measuring w(Li) using three different sample masses: 0.05, 0.10, and 0.25 g. Five instrumental replicates were made using FAAS for each subsample and specific mass of samples. In total, sixty subsamples were analyzed. Figure 1 shows the design of homogeneity studies for one bottle (B001) and one subsample mass.

Fig. 1
figure 1

Diagram showing the experimental design for the homogeneity studies.

Full size image

The w(Li) homogeneity test in clay was assessed using a one-way variance analysis (ANOVA) at a 95% confidence level. According to ISO 35:201738, the homogeneity of w(Li) in clay was evaluated with the F-statistic test, establishing whether Ftest < Fcritical value. If this condition was met, no statistically significant differences were found between and within bottles. Therefore, according to the trend of the data, the homogeneity of the RM batch is confirmed.

The uncertainty associated with the homogeneity of the w(Li) was estimated using Eq. (2).

$${u}_{bb}^{2}={s}_{bb}^{2}=\frac{{M}_{{between}}-{M}_{within}}{{n}_{0}}$$
(2)

where, Mbetween is the mean square between bottles, Mwithin is the mean square within bottles and n0 is number of observations per bottle. If the measurement method has limited repeatability, Eq. (3) will be used.

$$u_{bb}^{\prime}=\sqrt{\frac{{M}_{within}}{{n}_{0}}}*\sqrt[4]{\frac{2}{{\upnu }_{Mwithin}}}$$
(3)

Where, vMwithin is degrees of freedom of the mean of squares within bottles.

Stability test

Short-term isochronous stability study

The short-term lithium stability in clay CRM w(Li) was evaluated using an isochronous stability study (short-term stability), influenced by transport effects at temperatures of 20 °C, 40 °C, and 50 °C for 6 weeks. For the temperature of 20 °C, a desiccator with Mg(ClO4)2 as a drying agent was used, and for the temperature of 40 °C and 50 °C, a heating oven was used. For the experiment, 18 weighing bottles with a sample mass of approximately 0.2 g of clay CRM were placed for each temperature study. After 1 week, three samples were removed from each temperature and placed inside a desiccator. This activity was performed until the last sample was removed and 6 weeks of study were completed (Fig. 2).

Fig. 2
figure 2

Short-term isochronous stability study scheme.

Full size image

Subsequently, a 0.1 g mass of CRM sample was taken for measurement w(Li) microwave-assisted sample preparation using an acid digestion process, as described in section "Materials and methods". The w(Li) measurement in samples was carried out using FAAS at a wavelength of 670.8 nm. After performing the calibration curve, samples were measured in a random sequence. Also, the midpoint of the calibration solution was measured in successive block measurements to make instrumental memory effect correction drifting.

The ANOVA for regression was applied to assess the stability of w(Li), where two criteria were applied:

Criteria (a)

A test for a statistically significant gradient is a t-test for a slope significantly different from zero. This is conducted by calculating the t-statistic in Eq. (4) and comparing it with the two-tailed critical value of t-Student for n − 2 degrees of freedom at the 95% confidence level. If the calculated value of test statistic \({t}_{{b}_{1}}\) is less extreme than the critical value, the slope is ≅ 0 at the 95% confidence level, then the w(Li) is stable.

$${t}_{{b}_{1}}= \frac{\left|{b}_{1}\right|}{s\left({b}_{1}\right)}\text{ where}{ t}_{95 \%, n -2}$$
(4)

where, b1 is the slope of the linear regression model, s(b1) is the standard deviation of the slope of the linear regression model, and t-Student with a confidence level of 95% (section B.3.4, ISO 35:201738).

Criteria (b)

The statistical test of the p-value is used to ensure that w(Li) in the batch is stable. In a significance test, the null hypothesis H0 is rejected if the p-value is less than or equal to a predefined threshold value α, referred to as the alpha or significance level of α = 0.05. Then, if the p-value is above 0.05 as confidence level (p-value > 0.05), it is suggested that there are no significant differences in the data values of the w(Li), then the w(Li) in clay is stable.

The uncertainty associated with the stability of the w(Li) was estimated using Eq. (5).

$$u\left[w\left(Li\right)\right]=s({b}_{1})\cdot t$$
(5)

where: t: 6-week stability study time, \(s\left({b}_{1}\right):\) standard deviation of the slope of the linear regression model (section B.3.4, ISO 35:201738).

Classic long-term storage stability study

Prediction of shelf life in the case of a linear trend establishes the principle of estimating the 95% confidence interval (Eq. 6) for future values, taking account of the estimated degradation rate, and choosing the shortest time (tshelf) at which one of the 90% confidence limits (upper or lower Eq. (7)) intersects a specification limit (section B.4, ISO 35:201738).

$$ L_{90\% } = \hat{Y} \pm t_{{90\% /2,\left( {n - 2} \right)}} \cdot S_{YX} \sqrt {1 + \frac{1}{n} + \frac{{\left( {X - \overline{X}} \right)^{2} }}{{\sum {\left( {X - \overline{X}} \right)^{2} } }}} $$
(6)
$$ L_{95\% } = \hat{Y} \pm t_{{95\% /2,\left( {n - 2} \right)}} \cdot S_{YX} \sqrt {1 + \frac{1}{n} + \frac{{\left( {X - \overline{X}} \right)^{2} }}{{\sum {\left( {X - \overline{X}} \right)^{2} } }}} $$
(7)

Where, Ŷ is the fitted response for predictor value X, tα/2,n−2 is the t-value with n − 2 degrees of freedom, SYX is the standard error of the estimate, and n is the total data.

Time (tshelf) is determined using the second-order linear quadratic equation (Eq. 8) in its positive solution.

$$ x = \frac{{ - b \pm \sqrt {b^{2} - 4ac} }}{2a} $$
(8)

A second-order equation is written as ax2 + bx + c = 0, where a, b, and c are coefficients of real numbers and a ≠ 0.

Results and discussion

Homogeneity test

Figure 3 shows lithium mass fraction results for sample sizes of a) 0.05 g, b) 0.10 g, and c) 0.25 g. The error bars indicate the standard deviation of measurements for each bottle. The dashed line represents the mean lithium mass fraction w(Li).

Fig. 3
figure 3

Homogeneity studies of w(Li) for samples (a) 0.05 g, (b) 0.10 g, and (c) 0.25 g.

Full size image

Data was analyzed by one-way ANOVA for 0.10 g and 0.05 g samples. The statistical analysis results show that the Ftest was less than the Fcritical value (Table 2), indicating that no significant difference was found within/between bottles. However, the 0.25 g sample showed that the Ftest was higher than the Fcritical value, which indicated that a significant difference was found within/between bottles (see Additional Information 1: Tables S1.1, S1.2, and S1.3).

Table 2 One-way ANOVA results of samples for homogeneity test.
Full size table

Based on the ANOVA statistical analysis, results obtained from the mass fraction lithium w(Li) using a sample mass of 0.05 and 0.10 g show that the batch is homogeneous. Also, it is observed that the results of mass 0.05 g have less variability (RSD: 0.15%) than 0.1 g (RSD: 0.91%). This could be due to the lower value of w(Li), which causes a lower signal value in the instrument.

However, using a sample of 0.25 g, based on the ANOVA statistical analysis, it was observed that the batch was not homogeneous. This result is explained as follows:

  1. a.

    The samples were measured outside the working range of the measurement method to maintain the same measurement conditions. However, this implies that an extrapolation of the calibration curve was applied.

  2. b.

    The sample was not completely digested, possibly due to the higher sample mass. Therefore, a higher mass value is recommended to optimize the parameters of the digestion method, such as the amounts of acids, time, and temperatures.

The homogeneity uncertainty for the three different mass values is shown in Table 3, using Eq. (3).

Table 3 Uncertainty associated with homogeneity for three different masses.
Full size table

Stability study

Short-term isochronous stability study

The short-term stability test was conducted at three different temperatures: 20 °C, 40 °C, and 50 °C. The measurement results of lithium are described in Fig. 4. The error bar represents the measurement standard deviation for each bottle, and the dashed line represents the linear fitting of the data. The samples showed a linear behavior in all the temperatures. For each temperature, the data was tested by one-way ANOVA (see Additional information 1: Tables S1.4, S1.5, and S1.6). The Ftest found was less than the Fcritical value (Table 4), meaning no significant differences in the values of the w(Li) in clay, so w(Li) is stable for the study temperatures in 6 weeks was found in the samples at the same temperature (95% confidence level).

Fig. 4
figure 4

Short-term stability study at (a) 20 °C, (b) 40 °C, and (c) 50 °C.

Full size image
Table 4 One-way ANOVA results of samples for comparison of F values for stability test.
Full size table

On the other hand, the data were adjusted by linear regression, comparing tb1 versus t95%, n-2 (critical value46,51,52,53) (Table 5). At each temperature, it can be concluded that tb1 < t95%, n-2 with a confidence level of 95%, demonstrating that w(Li) is stable at the time of analysis.

Table 5 One-way ANOVA results for stability parameters, Student-t and pvalue.
Full size table

To establish whether there were differences in stability at each temperature, the three slopes were also compared by the ANOVA test (see Additional information 1, Fig. S1.1). The p-value was 0.7415 (see Additional information, Table S1.7), meaning the slopes were not statistically different at the 95% confidence level. In other words, this test also confirms that the samples were stable at the temperature range of 20–50 °C. The results of the stability uncertainty assessment for the three study temperatures are shown in Table 6, which was determined with Eq. (5).

Table 6 Uncertainty associated with stability for three temperatures.
Full size table

Classic long-term storage stability study

The results of the classic long-term storage study are shown in Fig. 5, where the prediction model showed the result of useful life. In this study, a linear trend model was used to determine the upper and lower intervals of the linear behavior of the values of the lithium mass fraction of the RM at confidence intervals of 90% and 95%. Also, the time at which a significant change was observed in the mass fraction of Li in clay (tshelf) was estimated. The crossing between 90 and 95% confidence intervals allowed to estimate the prediction of Li stability in the CRM corresponds to 155.3 weeks, equivalent to 3 years.

Fig. 5
figure 5

Classic stability study at 20 °C.

Full size image

In Fig. 5, the blue markers correspond to the w(Li) in clay during the 19-week study period and the green markers are values of w(Li) in clay estimated using the linear equation. This allowed the study period to be extended in the long term to estimate the useful life of the CRM.

The lithium clay certified reference material proved to be homogeneous and stable, fulfilling the objective of this study after the characterization of the w(Li) material was carried out using isotopic dilution with inductively coupled plasma mass spectrometry. Later on, the assigned value of w(Li) in the lithium clay material included the contribution of uncertainty of characterization, homogeneity, and stability. The budget of the uncertainty for the certified value, including their contributions, is shown in the Additional information in Tables and Figs. S2 and S3.

Conclusion

This work determined that the w(Li) in the lithium clay reference material is homogeneous and stable and was validated under the criteria established in ISO 17,034:2016 and ISO Guide 35:2017.

The rigorous homogeneity and stability testing, including both short-term and long-term studies, enhances the reliability and validity of the RM, providing traceability to the SI and offering essential data for ensuring accurate and reliable measurements in mining and supporting the economic development of Lithium-related industries. From results obtained in homogeneity testing, samples 0.10 g and 0.05 g showed that Ftest was less than the Fcritical value, indicating no significant difference was found within/between bottles. However, sample 0.25 g showed that Ftest was higher than the Fcritical value, which indicated that a significant difference was found within/between bottles. A minimum sample size of 0.05 g could be used to obtain results with acceptable precision in the Li mass fraction w(Li).

The short-term isochronous stability study determined that the mass fraction of lithium in the clay CRM is stable for temperatures of 20, 40, and 50 °C. These temperatures establish the transport conditions of the RM so that if the CRM is exposed to a temperature of 50 °C, the mass fraction of Li will not significantly change. Thus, the stable w(Li) values at various temperatures suggest robust transport and storage conditions, aligning with previous studies emphasizing the importance of certified reference materials in analytical chemistry.

The classic long-term storage stability study allowed the establishment of the storage conditions and validity period of clay CRM in 3 years. After this period, it is recommended that the mass fraction of Li be monitored to determine the feasibility of increasing the validity period of RM.

This CRM could be used in the validation of lithium measurement methods that allow obtaining reliable results in lithium measurements that help the exploration, exploitation, and management of lithium ore, in addition to the fact that the lithium CMR will be used for use in any analytical technique that requires lithium clay material for the assurance and quality control of analytical measurement processes, to obtain reliable measurements and traceability to the International System of Units (IS).