Silicon (Si) accumulation in plants increases plant resistance against abiotic and biotic stress with consequences for plant performance and ecosystem functioning (Cooke et al., 2016; Katz et al., 2021). Siliceous structures in plants consist of hydrated amorphous silica (SiO₂ · nH₂O), also known as phytogenic silica. In plants phytogenic silica can be found i) within cells (i.e., in the cell wall and the cell lumen) forming relatively stable, recognizable phytoliths, that can also be found in soils as plant microfossils or ii) in intercellular spaces and extracellular (cuticular) layers forming relatively fragile silica structures that are not preserved in soils (Hodson, 2016; Sangster et al., 2001). In agricultural soil-plant systems an accurate quantification of Si contents in plant materials is crucial for assessing crops’ need for Si supply and Si fertilization efficacy, because soil Si availability is often deduced from plant Si contents in agronomic studies (Korndörfer et al., 2001; Miles et al., 2014; Wu et al., 2020). In this context, a reliable, fast, and cost-effective method for the quantification of Si in soil amendments/fertilizers like, e.g., biochar or slags (Eltohamy et al., 2023; Haynes, 2017; Li et al., 2019) would be greatly appreciated by agricultural scientists as well as practitioners.

For the determination of Si contents in plants, Si is commonly extracted from plant materials using alkaline extractants like sodium carbonate (Na₂CO₃) or Tiron (disodium 4,5-dihydroxy-1,3-benzenedisulfonate, C₆H₄Na₂O₈S₂) and subsequently Si concentrations in the extracts are determined spectroscopically (Guntzer et al., 2010; Puppe et al., 2023; Schaller et al., 2022). Traditionally Si concentrations have been measured using molybdenum blue colorimetry (Isaacs, 1924; Nakamura et al., 2020; Volk & Weintraub, 1958). However, this method “only” measures the concentration of monomeric silicic acid (i.e., the plant available Si fraction) in samples as this is the only molybdate reactive form of silicic acid (polysilicic acid does not form a Si-molybdate blue complex, see our Discussion). Nowadays inductively coupled plasma optical emission spectrometry (ICP-OES) is widely used in environmental sciences for a fast routine determination of Si, because this technique is characterized by relatively high sensitivity and low expenditure of human labor (Delvigne et al., 2021). However, as an ICP-OES instrument is relatively expensive (including running costs), it is not readily available to most laboratories. Microwave plasma atomic emission spectroscopy (MP-AES) might represent a promising alternative to ICP-OES, because MP-AES generally has lower running costs. This is because MP-AES uses smaller torches and needs less gas compared to ICP-OES. Furthermore, MP-AES uses nitrogen, which is a comparably inexpensive gas. Nitrogen can even be obtained directly from the air by a nitrogen generator coupled to the MP-AES instrument, and thus no gas connection or supply is required. Since its commercial introduction in 2011, MP-AES has been successfully used as analytical tool in various research fields (Balaram, 2020). However, to the best of our knowledge no study of the aptitude of MP-AES for the determination of Si concentrations in alkaline plant extracts has been published until now.

To evaluate the capability of the MP-AES technique regarding the determination of Si contents in plant materials, we conducted a laboratory study using a relatively large sample set (n = 42) of retained husk and straw samples of winter wheat (Triticum aestivum) from a long-term field experiment (Puppe et al., 2021). Si was extracted from husk and straw samples using Tiron as recommended by Guntzer et al. (2010), who showed the general suitability of Tiron for the extraction of total Si from plant materials of different plant species (i.e., Triticum durum, Equisetum arvense, Dicksonia squarrosa, Ulmus laevis, and Larix gmelinii), and Puppe et al. (2023), who found Tiron to be more efficient in Si extraction compared to Na₂CO₃ using hydrofluoric acid digestion as a reference. Subsequently, Si concentrations in the Tiron extracts were measured via ICP-OES as well as MP-AES, corresponding Si contents were calculated (details in Material and Methods), and finally the results were compared. To evaluate the performance of MP-AES in biogeochemical Si plant-soil studies, we also correlated the results obtained from MP-AES measurements with data on plant available Si concentrations in corresponding soil samples as well as phytolith contents in the husk and straw samples (data from Puppe et al. (2021)) and compared these results to the ones obtained from ICP-OES analyses.

Si contents in plant materials obtained from MP-AES measurements were generally very similar to the ones obtained from ICP-OES. Regarding all tested samples (n = 42) we found a coefficient of determination (R²) of about 0.98 in the diagram of MP-AES results plotted against corresponding ICP-OES results (Figure 1A). The coefficients of determination for husk (n = 21, Figure 1B) and straw (n = 21, Figure 1C) samples were slightly lower (R² = 0.96 and 0.88, respectively). Our correlation analyses indicated a strong linear relationship between the results of MP-AES and ICP-OES measurements as Pearson’s correlation coefficients were generally higher than Spearman’s correlation coefficients (Table 1).

The differences (δSi) between Si contents in husk and straw samples determined by ICP-OES and MP-AES (δSi = Si (ICP-OES)—Si (MP-AES)) were more pronounced in husk samples ranging between −0.3 and 3.4 mg g⁻¹ (Figure 2). In straw samples δSi ranged from −1–2.1 mg g⁻¹. While in 6 out of 42 plant samples MP-AES results were slightly higher (indicated by negative δSi values) than ICP-OES results, in 36 out of 42 samples ICP-OES results were slightly higher compared to MP-AES results (δSi values positive). The mean level of achievement (in %, calculated as: Si (MP-AES)/ Si (ICP-OES) x 100%) was 95% for all tested samples (n = 42) as well as for husk (n = 21) and straw (n = 21) samples alone (Figure 2).

Si contents in husk samples were generally higher than in straw samples (Figure 3). While ICP-OES measurements revealed Si contents ranging from 8.5 to 19.5 mg g⁻¹ and 3.2–7.8 mg g⁻¹ (excluding outliers) for husk and straw samples, respectively, MP-AES measurements showed ranges of 8.2–19.9 mg Si g⁻¹ and 3.0–7.5 mg Si g⁻¹ for these husk and straw samples, respectively.

This was also reflected by phytolith contents, which were strongly correlated (p ≤ 0.001) to Si contents in husk and straw samples (Table 1). Phytolith contents in husk and straw samples ranged from 19.1 to 48.2 mg g⁻¹ and 6.0–17.9 mg g⁻¹ (excluding outliers), respectively (Figure 4). Plant available Si concentrations in corresponding soil samples ranged from 6.6 to 11.6 mg kg⁻¹ (Figure 4). While we found no correlations between plant available Si in soils and Si contents in straw samples at all, weak monotonic correlations (p ≤ 0.03) between plant available Si in soils and husk samples were revealed (Table 1).

In general, our results proved MP-AES to reliably measure Si concentrations in plant extracts. Regarding variations of MP-AES and ICP-OES results between husk and straw samples two potential causes can be discussed: i) sample-specific and ii) instrument-specific differences.

Sample-specific differences refer to potential Si fluctuations in husk and straw samples (i.e., Si content heterogeneity) and differences in the Si extraction efficacy between these samples caused by differences in phytolith compositions. Based on our results we can exclude a statistically significant Si content heterogeneity in husk and straw samples. In fact, MP-AES and ICP-OES results matched considerably better for husk than straw samples, with husk samples generally showing more pronounced fluctuations (indicated by a bigger range) in Si contents than straw samples. This is underlined by the fact that the ranges in Si contents of husk and straw samples measured by ICP-OES and MP-AES were almost identical. In a previous study Puppe et al. (2023) found differences in the Si extraction efficacy between husk and straw samples of winter wheat, which were also ascribed to differences in the proportion of cell wall and lumen phytoliths in these samples (Puppe et al., 2022). While cell wall phytoliths are associated with a carbohydrate matrix, lumen phytoliths seem to contain more proteins and glycoproteins than cell wall phytoliths making cell wall phytoliths more stable than lumen phytoliths (Hodson, 2016; 2019). As sample-specific differences in the Si extraction efficacy would have led to a stochastic distribution of δSi values, i.e., a roughly balanced relation of positive and negative δSi values, we exclude sample-specific differences as drivers of the observed phenomenon. However, our data rather indicated a general trend in δSi values, which were positive (i.e., ICP-OES results were slightly higher) in about 86% and negative (i.e., MP-AES results were slightly higher) in only about 14% of the plant samples. This general trend clearly indicates an instrument-specific cause for the obtained results.

In general, in emission spectroscopy atoms/molecules are excited in a high-temperature flame or plasma and photons emitted during transition from an excited state to a lower energy state are measured by use of an electronic detector, whereby emission intensity is directly proportional to the analyte concentration in the sample. An inductively-coupled plasma (ICP) is generated from argon gas reaching temperatures of about 10,000 K, which is hotter than the surface of the sun, providing a powerful source for atomizing, ionizing, and exciting analytes (Wilschefski & Baxter, 2019). In contrast, a microwave plasma (MP) is generally lower in temperature than an ICP making MP-AES measurements potentially more prone to matrix effects, which can distort the signal of the analyte. Matrix effects are caused by concomitant chemical species like easily ionizable elements (e.g., sodium (Na)) or highly concentrated inorganic acids (Santos et al., 2020; Todolí et al., 2002; Zhang & Wagatsuma, 2002). Thus, we ascribe the observed differences between ICP-OES and MP-AES measurements to matrix effects caused by the disodium salt Tiron (C₆H₄Na₂O₈S₂), which was used as extractant in our study. Some researchers used hydrogen peroxide (H₂O₂) to reduce potential matrix effects in ICP-induced emission spectroscopy measurements (Guntzer et al., 2010). However, Kodama and Ross (1991) found that the destruction of Tiron prior to spectroscopic analyses is not necessary to obtain reliable results, a finding that is underpinned by our long-term laboratory experience with Tiron extractions and corresponding results (Puppe et al., 2023). In fact, the adjustable device settings (e.g., the nebulizer flow rate adjustment or the analytical wavelength selection, see Fontoura et al. (2022)) allowed us an improved MP-AES performance with negligible interferences, which is well-reflected in our results showing a mean level of achievement of 95% (cf. Figure 2).

Our correlation analyses showed that phytolith contents in plant samples were highly, linearly correlated to corresponding total Si contents in these samples determined by MP-AES (cf. Table 1). In this context, it should be noted that the detection of relationships between phytolith contents in plant samples and corresponding total Si contents is hampered by the fact that Si in plants is not only represented by extractable phytoliths, but also by fragile silica structures (Meunier et al., 2017; Puppe et al., 2017), which are not covered by phytolith extraction techniques. As the majority of phytogenic silica might be stored in these fragile structures, current phytolith extraction techniques (ashing, acid digestion) thus might strongly underestimate the ‘real’ Si content of plants, which in turn hampers the proper interpretation of the role of phytoliths in plant-soil systems (Kaczorek et al., 2019; Puppe et al., 2017). In this context, the simultaneous quantification of total Si contents and extraction of phytoliths can be used to quantify both the fragile silica and the extractable phytolith fractions in plant samples. For the grasses Calamagrostis epigejos and Phragmites australis, for example, Puppe et al. (2017) showed that 16.4% and 15.9% of total Si were stored in extractable phytoliths >5 μm, respectively. For American beech (Fagus grandifolia) Wilding and Drees (1971) reported that about 72% of extractable leaf phytoliths were smaller than 5 μm. We clearly need more research on fragile silica structures and extractable phytoliths to understand their roles in biogeochemical Si cycling. However, in this context it should be kept in mind that there are also silica structures on a nanometer scale in plants that might be relatively stable in soils (Watteau & Villemin, 2001).

The difficulty in interpreting phytogenic silica contents and corresponding relationships with soil Si fractions is also indicated by the weak and insignificant correlations between phytoliths extracted from plant samples and plant available Si concentrations in corresponding soil samples in our study. Indeed, the correlation was stronger when plant available Si concentrations in soils were correlated to (total) Si contents determined spectroscopically, even if the correlations were statistically significant only for husk samples (Table 1). For straw samples, which showed lower Si contents than husk samples, no statistically significant correlations were detectable at all in our study. In general, relationships between plant available Si concentrations in soils and Si contents in plants are quite complex, because plant-specific Si uptake rates and plant available Si concentrations in soils can be highly variable as they are controlled by several factors. While plant Si uptake rates are related to phylogeny, growth stage, and Si availability in soils (Hodson et al., 2005; Ma et al., 1989; Schaller et al., 2022), Si concentrations in soils are controlled by physicochemical soil properties (e.g., soil pH, texture, or adsorption capacity) and biological factors like vegetation and biogenic silica contents in soils (Cornelis & Delvaux, 2016; Haynes, 2014; Schaller et al., 2021). Due to this complexity, studies of soil-Si-plant-Si relationships often show inconsistent results. Generally, studies that cover several study sites with different soil types and widely varying physicochemical soil properties report close relationships between plant Si contents and plant available Si concentrations in soils (Korndörfer et al., 2001; Miles et al., 2014). In contrast, studies limited on few or only one study site often show ambiguous results (Keeping, 2017; Klotzbücher et al., 2017; Puppe et al., 2023).

Moreover, the method of determination of plant available Si concentrations in soils can affect the interpretation of corresponding results. In fact, plant available Si (i.e., monosilicic acid, H₄SiO₄) has been traditionally determined by molybdenum blue colorimetry (Babu et al., 2016; Jones & Handreck, 1965; Winslow et al., 1997), which is based on the observation that only monosilicic acid reacts immediately (in contrast to polysilicic acids) with molybdic acid forming a blue Si-molybdate complex (Govett, 1961). Contrary, emission spectroscopy like ICP-OES or MP-AES might overestimate plant available Si concentrations in soils, because this technique also measures polysilicic acid. However, such an overestimation seems to be negligible in most extractants indicating that emission spectroscopy is generally well-suited for the determination of plant available Si concentrations in soils (Wang et al., 2004; Zellner et al., 2015).

In summary, we conclude that MP-AES is a suitable atomic emission spectrometry technique for the reliable determination of Si concentrations in Tiron extracts. Since ICP-OES and MP-AES results in our study showed only slight, statistically insignificant differences, we found matrix effects in MP-AES measurements caused by Tiron to be generally negligible. As MP-AES is more cost-effective compared to the de facto standard technique ICP-OES regarding investment and running costs, MP-AES represents a promising alternative for all researchers with a focus on biogeochemical Si cycling in general. However, as our study was limited to plant materials with Si contents of about 0.3%–2%, future research should confirm the suitability of MP-AES for Si determinations of plant materials with Si contents beyond this range.