2.1. Increased H:C_org ratio despite similar aromaticity: Where does the hydrogen come from?

Biochars were produced from pelleted, low-ash wood shavings (biomass composition S1 Table) at 450 °C and 750 °C in a continuously operated auger reactor [27]. These showed well-expected properties, mass yield (y_m), and y_c as shown in Table 1 [28–30]. Based on the y_m of the pure biochars, we mixed basanite rock powder (elemental and mineral composition, S2 Table) to the wood shavings prior to pelleting and pyrolysis to achieve the desired rock powder content in the RE-biochar.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Basic physico-chemical properties of biochar and rock-enhanced biochar (BC) produced at 450 °C (450) or 750 °C (750), containing 0 (-), 10% (9:1), 50% (1:1) or 90% (1:9) basanite rock powder. Data on main- and trace element content are presented in S7 Table and basic properties of basanite in S2 Table).

https://doi.org/10.1371/journal.pone.0355781.t001

Rock addition increased the ash content of the biochar in agreement with the rock amendment rate. Dry and ash-free (daf) organic carbon (C_org) content and y_c remained largely constant for RE-biochars produced at the same HTT (Table 1). The most remarkable effect of rock addition was the extremely high H:C_org molar ratio of 0.87 observed for the RE-biochar produced at 750 °C with 90% rock content (BC750–1:9). As pure basanite showed 0.3% hydrogen content in elemental analysis (0% C), we used this data to calculate a H:C_org molar ratio of the biogenic fraction, which was still considerably higher than expected (H_bio:C_org = 0.54). This calculation assumes that the rock, or at least its H content, remained unchanged by co-pyrolysis, which we examine in more detail below.

Lower dosages of rock powder had a limited effect on this biochar property and rather decreased the H:C_org molar ratio at 750 °C. At 450 °C, H:C_org molar ratio increased only slightly with increasing rock content. To verify these results, we reproduced all samples (S3 Table) with the difference that pelleting was not carried out in the lab but by a service provider and thus presumably under different pressure conditions (higher pellet density in 2^nd batch). In this 2^nd batch, the BC750–1:9 sample had a H:C_org molar ratio of 0.67 (H_bio:C_org = 0.37) (S4 Table), which is lower than in the first experiments but still much higher than the control and generally higher than expected for a biochar made at this HTT. Such high H:C_org molar ratios are almost unprecedented at a HTT of 750 °C [29,30]. Yang et al. [13] have documented an increase of H:Corg molar ratios in biochar produced at 500 °C from 0.2-0.3 to 0.3-1.3 for the addition of various minerals and salts, including Ca(H₂PO₄)₂, which caused the maximum values.

Still, a quantification of BPCA markers (%BPCA-6) [31] and hydropyrolysis (HyPy) [32,33] did not reveal any considerable changes due to rock enhancement, which strongly suggests that changes in H:C_org ratio (Table 1) are not the result of changes in biochar’s aromaticity. For HTT 450 °C, BPCA-6 – the hexacarboxylated benzene marker indicative of highly condensed aromatic domains in biochar, accounted for 23.3%−27.6% of total BCPAs and 42.7%−58.4% of C_org resisted HyPy (BC_HyPy), representing the aromatic carbon moieties of eight or more condensed rings. For HTT 750 °C, BPCA-6 presented 69.6%−73.4% of total BCPAs and BC_HyPy 95.6%−97.4% of C_org. Also, absorption spectra obtained by Fourier-transformed infrared spectroscopy (FTIR) and Raman spectroscopy revealed no difference in bulk carbon speciation between (RE-)biochars produced at the same HTT. Fourier transformed infrared spectroscopy showed the presence of aromatic C = C bonds at 1589 cm^-1 and absence of symmetric- and asymmetric aliphatic carbon structures (2840-2890 cm^-1 and 2920−2990 cm^-1, respectively) in all samples (S5 Fig). Raman spectroscopy revealed that ratios of D/G intensity- and area ratios were indifferent between samples obtained at the same HTT (S6 Fig). The G band at 1600 cm^-1 relates to C-C linkages in aromatic moieties, i.e., the degree of graphitization. The D band at 1345 cm^-1 relates to structural defects of the latter [34]. As the aromatic backbone remained largely unchanged compared to the control biochar, the increase in the H:C_org molar ratio can only be explained by an increase of the H content of the mineral phase and/or an increase of the H:C_org molar ratio in the non-BC_HyPy fraction. This may result from aliphatic compounds (e.g., alkanes, H:C ≈ 2) in the pyrogenic coating of the rock surface [24]. This aliphatic- or small aromatic systems would be included in measured non-BC_HyPy fraction or the S1&S2 fraction of sequential HAWK pyrolysis (c.f. section 2.1.1).

2.1.1. Hydrogen bound in semi-volatile organic compounds.

An increase in non-BC_HyPy should correspond to a decrease in thermal stability and an increase in thermally labile, i.e., volatile organic compounds. We quantified the daf mass loss (m_vol,TGA) of 5 mg (RE-)biochar in the range of 50−1000 °C at 10 °C min^-1 thermo-gravimetric analysis (TGA). In parallel, a HAWK-pyrolizer with coupled C_org-analyzer quantified carbon loss (C_vol,HAWK) from 10 mg sample in the range of 300−650 °C at 25 °C min^-1 [35]. Both analyses ran under inert atmosphere. The content of C_vol,HAWK (Table 1) increased with rock powder content from 10.9 and 0.3 mg C_vol,HAWK per gC_org at 0% rock to 12.9 and 3.3 mg C_vol,HAWK per g^-C_org for biochars containing 90% rock that were produced at HTT of 450 °C and 750 °C, respectively. The increase was small at 10% and 50% rock but pronounced at 90% rock content. Likewise, m_{vol,TGA50−350°C} was 2−4% of daf biochar for samples with 0−50% rock content but 9% from BC450–1:9 and 21% from BC750–1:9 (Fig 1). The same trend is observed for m_{vol,TGA350−650°C} (corresponding to the HTT of the HAWK pyrolizer), except for BC450. Still, the C_vol,HAWK fraction of daf biochar is < 1% and cannot solely explain the reduced thermal stability observed in equivalent temperature ranges. The fraction of daf biochar ultimately resisting to thermal degradation at the end of the TGA temperature ramp is referred to as BC_1000C (%). The strongly increased BC_1000C fraction in RE-biochars produced at HTT 450 °C, same as the complete loss of BC_1000C from the BC750–1:9 sample cannot rationally be explained by their initial PyC speciation (Fig 1). We elucidate potential limitations of methods, which were initially developed for carbonaceous materials, in the thermal analysis of RE-biochars in section 2.5.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Top: Thermogravimetric analysis of rock-enhanced biochars: Mass loss (%) of dry and ash free (daf) biochar in specific temperature ranges (in gray scale) and fraction retained at 1000 °C (BC_1000C).

Bottom: Solid-state electrical conductivity (SEC) in mS cm^-1. (A + C): Rock-enhanced biochars pyrolyzed at 450 °C (BC450) containing 10-90% basanite (BC450-9:1, BC450-1:1 and BC450-1:9), (B + D): Rock-enhanced biochars pyrolyzed at 750 °C (BC750), containing 10-90% basanite (BC750-9:1, BC750-1:1 and BC750-1:9). Orange shaded bars with BC + R notation denote equivalent post pyrolysis mixtures of biochar and rock powder. LOD = Sample specific limit of detection, calculated from biochar pellets height after compression at 160 MPa (4.5-6.0 mm = distance between anode and cathode) and maximum resistivity measurable with the setup (100 x 10⁶ Ohm).

https://doi.org/10.1371/journal.pone.0355781.g001

The solid-state electrical conductivity (SEC) of RE-biochar decreased compared to pure biochar, which was expected because silicate rock is an electric insulator. However, SEC was higher than in the equivalent post-pyrolysis mixtures (PPMs) of pure biochar and rock powder (Fig 1C + D). These results are consistent with Meyer zu Drewer et al. [24], who co-pyrolyzed the same basanite with landscaping wood on the same pyrolysis unit at 650 °C. They attributed the increased SEC to pyrogenic coatings of rock particles with conductive, soot-like secondary char formed from gaseous pyrolysis products [36,37]. Such secondary char contains condensed aromatic structures and condensates of tar [38]. Condensates are typically more hydrogenated than a primary aromatic lattice [39], which may partially explain the elevated H:C_org molar ratio of RE-biochars. Sorption of volatile organic carbon onto SiO₂-rich mineral surfaces [40] can provide nucleation sites for secondary char. Both in Meyer zu Drewer et al. [24] and in the present study, rock enhancement did not change y_c. Thus, the amount of carbon from secondary char coating is quantitatively minor compared to the total carbon yield. It can be concluded that small, hydrogenated carbon species, as approximated by C_vol.HAWK content (dummy H:C_org ratio = 2; aliphatic carbon) and non-BC_HyPy content (dummy H:C_org ratio = 0.625; pyrene) contribute only to a negligible extend to the total H content of the biochar (S8 Table).

2.1.2. Hydrogen associating or reacting with minerals.

Gaseous hydrogen (H₂) is an inherent by-product of biomass pyrolysis [41–43]. Its production increases with HTT and may exceed 50 vol% of the pyrolysis gas depending on feedstock selection and process conditions [44,45]. Physi- and chemisorption of H₂ to silicate minerals can occur in consequence of a solid-gas exchange where H₂ or other H-bearing volatiles are present. The sorption of H₂ to minerals is dominated by physical adsorption [46]. Silicate minerals can be polar due to isomorphic substitution, charged surfaces and edges or functional groups such as silane groups (Si-OH). Hydrogen can easily be polarized and bind to charged mineral surfaces by van der Waals forces [47]. The adsorption capacity increases with micro porosity and specific surface area (SSA) [46]. Consequently, H₂ sorption to dense minerals such as pyroxene or plagioclase is likely limited but considerable for amorphous Fe-oxides and layered silicates. Especially defect sites and the interlayer space (>2.9 Å) in 2:1-layer silicates are relevant for H₂ adsorption [43–45,48]. Catalytic metals can further enable hydrogen spillover to surfaces that otherwise do not bind H under identical conditions [49].

Lastly, H (respectively H₂O) can also be structurally incorporated in nominally anhydrous minerals and isomorphically substitute geogenic metal ions such as Mg²⁺. This was demonstrated for olivine under extremely high pressure and temperature regimes. At 1 atm, this process is much more limited and likely restricted to surface sites. Natural, nominally anhydrous olivine contains < 0.05 mol H kg^-1 [50,51]. Sundvall & Skogby (2011) [52] investigated the protonation of pyroxene under varying conditions. Under 1 atm up to 700 °C they found that most pyroxenes lost H while hydration was only observed for pyroxenes rich in iron (Fe). They explained this observation by a surface redox reaction. Olivine and pyroxene are both present in the basanite employed here, which also contains 7.8% Fe (S2 Table).

We used X-ray diffraction (XRD) to detect possible changes in the mineralogy of basanite thermally treated in the absence of biomass and basanite co-pyrolyzed to form RE-biochars (9:1). The heating of pure rock powder had no effect on the speciation of silicate minerals (R450 and R750 in S2 Table). The content of all silicate minerals in the 450 °C RE-biochars was in good agreement with expected, calculated concentrations. In RE-biochars produced at 750 °C, the content of all silicate minerals fell below the expected values, while the amorphous fraction increased (Table 2); for example, the detected content of the potassium (K)-bearing mineral leucite was 50% below the expected value. We assume that X-ray diffraction was not affected by the presence of PyC as illustrated by the successful detection of calcite in pure BC750, despite a total inorganic carbon (TIC) content of < 1%. In summary, the assessment of bulk mineralogy confirmed changes induced by co-pyrolysis that were absent in thermal treatment without biomass, but did not directly prove H uptake. Further research should employ X-ray photoelectron spectroscopy (XPS) to investigate changes in binding energy in surface minerals only.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Concentration of mineral species in pristine basanite rock powder (R), pure biochar (BC450, BC750) and selected rock-enhanced (RE-) biochars produced at 450 °C containing 10% rock (RE-BC450-9:1) and produced at 750 °C containing 10% rock (RE-BC750-9:1) based on X-ray diffraction (XRD) measurements. The data is presented in mass-%.

https://doi.org/10.1371/journal.pone.0355781.t002

The inorganic phase is generally a more relevant contributor to elevated H:C_org ratios than low molecular weight hydrocarbons. The contribution of geogenic hydrogen can explain 50% of the increase in H:C_org observed in BC750–1:9 compared to BC750 (H:C_org versus H_Bio:C_org). The sorption of a dense monolayer of H₂ to the rock surfaces (BET-surface, section 4.2) present in BC750–1:9 could explain about 5% of the remaining delta between the H_Bio:C_org of BC750 and BC750–1:9. However, this can be both an over- and underestimation. On the one hand, not all mineral surfaces are exposed (pyrogenic coating) and not all sorb H₂ well. On the other hand, small interlayer spaces of phyllosilicates (0.3-0.4 nm; most relevant for H₂ sorption) are not accounted for in BET analysis. Hydrogen can enter pores inaccessible for N₂. We acknowledge that there must be additional, unknown drivers contributing to elevated H_Bio:C_org which were not quantified here. One possible scenario is the “lock-in” of water that may have formed as a product of redox reactions or in the syngas. Water may be structurally integrated in minerals (hydration) or retained in obstructed parts of the biochar pore system (section 2.3)

In summary, the elevated H:C_org ratio in BC750–1:9 are attributed to multiple processes that differ in their degree of evidential support. First, the contribution of geogenic hydrogen is directly quantified from bulk elemental data and accounts for 50% of the observed increase relative to BC750. Second, physical adsorption of molecular hydrogen to mineral surfaces was estimated at 5% based on the BET SSA, which carries considerable uncertainties. Third, aliphatic and less condensed (lower-aromaticity) carbon species in secondary char have much higher H:C_org molar ratio than the bulk biochar, though their minor mass fraction limits their quantitative contribution to the elevated H:C_org. Fourth, the retention of water formed during redox reactions or trapped in obstructed pore domains remains a speculative and hardly quantifiable mechanism, whose contribution is, however, considered marginal. Collectively, the processes identified so far do not fully account for the elevation in H:C_org, suggesting that additional mechanisms are at play.

2.2. Reduction of iron minerals during co-pyrolysis

The pyrolysis gas contains reducing agents such as H₂ or carbon monoxide, which could react with mineral components and are known to reduce iron at high temperatures [53–56]. X-ray diffraction showed that the content of the ferric iron mineral hematite (Fe₂O₃) fell below the limit of detection as the result of co-pyrolysis, in contrast to the thermal treatment of the rock powder without biomass (S2 Table). This can be explained by the reduction of Fe(III) to Fe(II), forming, e.g., magnetite or other reduced minerals such as siderite or vivianite. The disappearance of hematite is concomitant with an increase in magnetic susceptibility (χ), which quantitatively describes the content of ferrimagnetic compounds in a sample. It is determined by measuring the extent to which a sample of known volume and mass alters a magnetic field and is expressed in m³ kg ⁻ ¹ [57]. Pure biochars contained only trace amounts of ferrimagnetic minerals (3x10^-8 m³ kg ⁻ ¹), whereas basanite rock powder showed more than 70 times higher levels (2.25x10^-6 m³ kg ⁻ ¹) that further increased to 2.77x10^-6 m³ kg ⁻ ¹ upon thermal treatment at 750 °C. In RE-biochars, ferrimagnetic contents exceeded the expected mixture values, particularly at 750 °C (Fig 2). To the best of our knowledge, this is the first evidence that not only pure Fe minerals (e.g., hematite) can be reduced during the production of biochar [58], but also Fe minerals embedded in a silicate rock matrix.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Measured content of bulk ferrimagnetic minerals in m³ kg^-1 sample and expected content of bulk ferrimagnetic minerals considering their respective biogenic and geogenic fraction and ferrimagnetic properties of the latter (grey).

On the X-axis: Rock-enhanced biochars produced at 450 °C (BC450) containing 10−90 wt% basanite (BC450-9:1, BC450-1:1 and BC450-1:9) and rock-enhanced biochars produced at 750 °C (BC750), containing 10−90 wt% basanite (BC750-9:1, BC750-1:1 and BC750-1:9).

https://doi.org/10.1371/journal.pone.0355781.g002

2.3. ¹H FFC-Relaxometry: Effects on water–surface interactions and pore domain connectivity

Nitrogen sorption isotherms and the BET method showed that RE-biochars produced at 450 °C displayed higher surface areas than expected from the arithmetic combination of biochar and rock powder surface areas (S9A,B Fig). At 750 °C, surface areas decreased progressively with increasing rock content but remained within the expected range for mixtures containing up to 50% rock, whereas the 90% rock treatment showed a marked decline (56% below the expected value). This behavior likely reflects a substantial reduction in accessible pore domains due to obstruction of pores by minerals, as suggested by Meyer zu Drewer et al. [24]. To complement these structural measurements with a functional characterization of water-accessible domains, we employed fast field cycling (FFC) ¹H-NMR relaxometry. This technique probes the spin–lattice relaxation of water protons within porous matrices [59]. Relaxometry provides information on water–surface interactions, thereby allowing the assessment of pore-domain accessibility and connectivity in porous materials such as biochars.

The relaxation patterns in Fig 3 reveal three effects of co-pyrolysis with rock powder. First, the shift of T₁ distributions reflects HTT-dependent changes in the balance between water–surface and water–water relaxation. At 450 °C, rock addition shifts T₁ toward shorter values, indicating that water is more strongly bound to biochar and mineral surfaces. In contrast, at 750 °C, rock addition shifts T₁ toward longer values. This in turn is indicative for increasingly bulk-like relaxation environments where water interacts less strongly with biochar and mineral surfaces (i.e., more bi-polar water-water connections in larger pores). The observed opposite trends align with the HTT-dependent restructuring of accessible pore domains inferred from BET measurements, where the SSA increased with rock content at 450 °C but decreased at 750 °C. Second, the marked decrease in the pore connectivity index (PCI) indicates reduced exchange between relaxation domains, implying that water mobility becomes increasingly confined to domains that exchange less efficiently with each other. This observation implies a generally reduced mobility of liquids and potentially gases in RE-biochar. Third, in RE-biochars produced at 750 °C, the alteration of T₁ distributions is consistent with the formation of mixed-valence Fe(II)/Fe(III) oxides such as magnetite, generated through the reductive transformation of Fe-oxides during pyrolysis. Such paramagnetic phases can contribute to the overall relaxation behavior, and may locally increase relaxation efficiency (i.e., shorter T₁) Accordingly, the net shift toward longer T₁ at 750 °C is interpreted primarily as a pore-domain/ accessibility effect (more bulk-like water environments), with magnetic phases acting as an additional, superimposed contribution [57,60,61].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. (A + B): Distribution histogram of measured longitudinal relaxation times (T₁) of hydrogen during fast field cycling nuclear magnetic resonance (FFC-NMR) relaxometry.

Logarithmic scaled X-axis in milliseconds and in-figure labels in seconds. (C + D): Dimensionless pore connectivity index (PCI) based on T₁ relaxation times, presented as mean ± standard deviation. Left: (Rock-enhanced) biochars pyrolyzed at 450 °C (BC450) containing 10-90% basanite (BC450-9:1, BC450-1:1 and BC450_1:9). Right: (Rock-enhanced) biochars pyrolyzed at 750 °C (BC750), containing 10-90% basanite (BC750-9:1, BC750-1:1 and BC750-1:9). Measurements on pure basanite rock powder, after thermal treatment at 450 °C (R450) and 750 °C (R750) respectively are indicated as yellow dashed lines in panel A and B.

https://doi.org/10.1371/journal.pone.0355781.g003

Altogether, these effects highlight a chemically active interfacial environment during the co-pyrolysis, leading to the restructuring of pore domains compared to pure biochar and mineralogical transformations that shape porosity and surface properties. Given these effects, it is plausible that, at sites enriched in reactive or partially reduced metals, a fraction of interfacial carbon may experience interactions stronger than those expected from simple physisorption, potentially including the formation of metal–carbon (M–C) linkages and complexes. Possible examples include carbides such as Fe₃C [62]. While such interactions could in principle contribute to the modified relaxation behavior, they cannot be detected by FFC-NMR alone and would require complementary high-resolution spectroscopic or microscopic analyses for verification. At this stage, M–C linkages can therefore be regarded as a possible, yet unconfirmed, additional contribution to the overall interfacial relaxation environment.

2.4. Carbon speciation at the biochar-mineral interface

Scanning transmission electron microscopy coupled to electron energy loss spectroscopy (STEM-EELS) was used to resolve elemental composition and carbon speciation in BC750–1:1 at the nanometer scale (Fig 4, S10 Fig). Six EELS transects crossing from biogenic components of RE-biochar into adjacent rock consistently showed the K edge of aromatic C at 284.9–285.5 eV [63] transitioning to the characteristic Si-O peak at 540 eV [64] within the mineral phase (Si-O peak cropped from Fig 4). These measurements revealed a distinct carbon speciation at the biochar-mineral interface. Aliphatic carbon features (σ(C–H) and σ(C–C) at 287–288 eV [65,66] were detected in 4 of 6 scans, while carboxylic features at 288.6 eV [63,67] occurred in 2 of 6 scans. Such aliphatic and oxygenated signals align with the known composition of volatile pyrolysis intermediates like furans, phenols, alkanes, alkenes or alkynes, and support the idea of a pyrogenic coating of mineral surfaces. Basanite is a non-acidic aluminosilicate rock, composed mainly of plagioclase, pyroxene, and nepheline. These minerals provide negligible Brønsted acidity and therefore behave similarly to other non-acidic silicate surfaces, which are known to sorb and stabilize radicals and oxygenated vapors rather than drive catalytic transformations [68,69]. Accordingly, the interfacial carbon detected here is most likely derived from adsorption and consistent with the thin pyrogenic coating [24]. The presence of organo-mineral bonds as suggested above could not be confirmed, but the limited number of STEM-EELS scans can neither refute their existence. The sorption of such low-molecular compounds with a high H content (e.g., H:C ≈ 2 in alkanes) – or more complex pyrogenic hydrocarbons is a minor contributor to elevated H:C_org ratios.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Left panels (A,C,E): Scanning transmission electron micrographs obtained in high-angle annular dark field mode (HAADF), showing ultrathin cross section of a BC750-1:1 pellet.

Mineral particles (white) are embedded in the surrounding biochar matrix (dark) with an indicated measurement transect (color-coded) spanning the pure biochar (blue to purple) over the mineral-biochar interface (green to turquoise) into the rock particle (yellow to light green). Right panels (B,D,F): Corresponding electron energy loss spectroscopy (EELS) spectrograms obtained along the measurement transect. Note that the Si-O peaks are cropped from the spectra. Three additional spectrograms are shown in S10 Fig.

https://doi.org/10.1371/journal.pone.0355781.g004

Elevated concentrations of K were identified at the biochar-mineral interface in 3/6 scans (297–300 eV) [70] and a 2D EELS mapping revealed a potassium shell at the mineral surface (Fig 5). Beam-induced K migration [71] was excluded experimentally (Inversed beam trajectory; S10B Fig). Enrichment of biogenic ash around rock particles could in principle produce such a shell, but this would also result in proportional enrichment of other cations, which was not observed. Instead, pyrogenic free radicals [72,73] may mobilize K from the outer layers of K-rich basanite minerals such as nepheline or leucite, whose heterogeneous distribution at the micrometer scale – as confirmed by scanning electron microscopy coupled with energy dispersive X-ray spectroscopy (SEM-EDX, S11 Fig) – likely explains the occurrence of K enrichment in only 3 of 6 STEM transects. This process may lead to enhanced alkalinity production due to a stronger release of basic cations (such as K⁺) in short-term weathering experiments using RE-biochar.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. (A): Micrograph showing an ultrathin cross section of a BC750-1:1 pellet, with clipping (B) subjected to Electron Energy Loss Spectroscopy mapping of elemental distributions.

(C): Mapping of the distribution of the dominating elements: silicon-dominated rock particle (light-blue), carbon-dominated biochar matrix (dark-blue), potassium-rich shell (yellow).

https://doi.org/10.1371/journal.pone.0355781.g005

2.5. Matrix-effects of inorganic phases on biochar analysis

The inorganic constituents of RE-biochar differ fundamentally from aromatic carbon moieties. Where mineral enrichment is high (e.g., 90% rock), inorganic constituents contribute a relevant volume-fraction to the RE-biochar. In such a case, interface regions are widely distributed and may become the spatially dominating surface in the system. Accordingly, it is necessary to critically examine whether established methods of biochar analysis can be applied to RE-biochar. It has been shown that the main and trace elements in RE biochar can be reliably quantified after chemical digestion, even though X-ray fluorescence is commonly used for rock [24]. The present study, however, revealed distortions in thermal and optical analytical techniques.

Following TGA (50−1000 °C), RE-biochars produced at 450 °C showed higher BC_1000C fractions than BC450, while BC_1000C for BC750–1:9 was zero. These findings cannot be reconciled with the results of HyPy and BCPA, which in turn provided a consistent picture. A likely source of error is the challenge to obtain a representative sample at the milligram-scale from a material with large density contrasts (biochar: ~ 0.2 kg L^-1; basanite: ~ 2.6 kg L^-1). If the actual ash content in the 5 mg subsample for TGA deviates strongly from the nominal value used for daf correction, mass loss and BC_1000C fraction may be over- or underestimated [24].

Matrix effects from the higher thermal conductivity and heat capacity of basanite compared to biomass and biochar may further distort apparent thermal stability [24]. Equivalent PPMs could not be reliably analyzed due to subsampling uncertainty at the milligram scale [24], and there was not enough biochar available for a mixture in the gram range followed by grinding in the vibratory ball mill. Therefore, the results of the thermal analysis of the RE-biochars can be unreliable and must be interpreted with caution. Demineralization of high-ash biochar samples, e.g., hydrofluoric acid treatment, is one option to be considered for the removal of silicates. It was shown that pre-analysis demineralization of samples can increase the predictive power of, e.g., elemental analysis and H:C_org molar ratio to forecast actual aromaticity [74].

Random reflectance is increasingly used in biochar characterization [35,75–77]. For BC450, a mean Ro of 2.1% was recorded, which decreased to 1.7-1.9% in RE-biochars (Fig 6) and Ro% distributions were increasingly leptokurtic at 50−90% rock (narrowing of the curve = clustering of data around mean) with kurtosis of 1.7-3.0 (gaussian distribution = 3). This likely reflects a more homogeneous heating due to improved thermal conductivity of the rock–biomass mixture. For BC750 mean Ro was 4.3%, which increased to 4.5% in BC750–9:1 and BC750–1:1 and decreased to 3.6% in BC750–1:9 with increasingly platykurtic distribution (spreading of the distribution curve = larger standard deviation) with increasing rock content. Generally, HTT 450 °C samples were right skewed, i.e., the population had a lower median than mean, while HTT 750 °C samples were rather left skewed. These strong effects of rock-enhancement on Ro% distributions, despite nearly unchanged aromaticity (c.f. BC_HyPy, BPCA-6, FTIR, Raman) highlights that a high ash content can considerably impact Ro determination. It remains speculative whether this is due to the influence of the rock on the reflective properties or whether a proportion of the aromatic carbon now interacts with the rock in such a way that these structures can no longer be identified as macerals under a light microscope and thus not contribute to Ro. Therefore, Ro-derived metrics (e.g., persistence indices or carbonization temperatures) and mean Ro values of skewed distributions must be interpreted with caution. Both TGA and Ro highlight limitations of biochar analysis when applied to RE-biochars or high ash biochars in general. Future research but also routine analysis, such as for the European Biochar Certificate (EBC) [78], should explicitly account for these matrix effects in experimental design, sample preparation and data interpretation.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Histograms of random reflectance (Ro%) distributions obtained from n = 500 measurements per sample.

Left column: Rock-enhanced (RE-)biochars produced at 450 °C, containing (A): 0% (BC450), (B):10% (BC450-9:1), (C): 50% (BC450-1:1) or (D): 90% basanite rock powder (BC450-1:9). Right column: RE-biochars produced at 750 °C, containing (E): 0% (BC750), (F):10% (BC750-9:1), (G): 50% (BC750-1:1) or (H): 90% basanite rock powder (BC750-1:9). Indications in green: Mean Ro%, Kurtosis (Gaussian distribution = 3, leptokurtic >3 and platykurtic <3) and Skewness (Gaussian distribution = 0, right skewed >0 and left skewed <0).

https://doi.org/10.1371/journal.pone.0355781.g006

4.1. Biomass preparation and pyrolysis

Spruce wood shavings (Picea abies L.) were sourced from Allspan German Horse GmbH, Karlsruhe, Germany (50.7% C, 0.5% ash; S1 Table). Basanite rock powder (“Eifelgold”; particle size range 0–250 µm) was obtained from Provinzial-Basalt- u. Lavawerke, Sinzig, Germany and characterized by X-ray diffraction, X-ray fluorescence, laser granulometry and extraction in ultra-pure water as described elsewhere [24]. The basanite’s SSA was measured as described in section 4.2. Both, pure biomass and homogenized mixtures of biomass with fine rock powder were pelleted at 6 mm without additional binders using a WK230 pellet press (Evertec, Dieburg, Germany). Afterwards, the pellets were dried (12 h at 60 °C), and fine material (< 6 mm) was removed by sieving.

The required feedstock mixing ratios to obtain RE-biochars with nominal 10%, 50% or 90% rock content, were calculated based on mass-yield (y_m) measured during production of pure biochar at 450 °C (BC450; y_m = 25.9%) and at 750 °C (BC750; y_m = 18.9%). The mass of the basanite was constant during pyrolysis, which was also verified by thermo-gravimetric analysis (TGA) of pure basanite. As some rock powder might got lost during pelleting and sieving, the actual biogenic (f_bio) and rock mass fraction (f_rock) of the feedstock was re-calculated following equation 1 and 2 (Table 3).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Composition of feedstock to produce rock-enhanced biochar (BC) at 450 °C (450) or 750 °C (750), containing 10% (9:1), 50% (1:1) or 90% (1:9) basanite rock powder after pyrolysis.

https://doi.org/10.1371/journal.pone.0355781.t003

(1)

(2)

With ash (%) being the ash content and C_org (%) the organic carbon content of the feedstock mix. The factor r describes the ratio of C_org to ash in the pure biomass (s1). The same procedure was applied to calculate f_bio and f_rock for RE-biochars, assuming no mass loss of rock powder during pyrolysis and no impact of rock powder on biochar dry and ash free (daf) yield.

Feedstock transport-rates (t_f in g min^-1) through the non-heated pyrolysis reactor were determined for each treatment by measuring material output in 10 min intervals (n = 6). In the same manner, biochar transport-rates (t_b in g min^-1) through the heated pyrolysis reactor were determined in 30 min intervals (n = 3). Pyrolysis was performed at 450 °C and 750 °C, respectively, with a residence time of 15 min using a PYREKA research pyrolysis unit equipped with a continuously operating auger reactor (PYREG GmbH, Dörth, Germany) under Nitrogen flow of 2 L min^-1 [24,27]. The PYREKA operated in one continuous process per treatment.

The ratio of feedstock to biochar output during the first hour (equivalent to four reactor residence times) and after the feeder ran empty, was excluded from transport rate calculations and therefore from calculation of y_m and y_c. The biochar mass yield was calculated as follows:

(3)

With y_m being the pyrolysis mass yield as % of dry weight feedstock converted to dry weight biochar, t_f the mean transport-rate of the feedstock (g min^-1) and t_b the mean (RE-)biochar transport-rate (g min^-1). Exceptions were BC450–1:9 and BC750–1:9, here, due to the limited material volume, no transport-rate was determined but the total mass before and after pyrolysis measured. Further, the coefficient of variation (%CV) in t_b was calculated as a proxy for the variability of the continuous production process and resulting y_c.

The y_c was calculated as follows:

(4)

With y_m being the mass yield of pyrolysis process (%), C_org(f) the C_org content of the feedstock and C_org(BC) the C_org content of the produced (RE-)biochar.

The (RE-)biochar production during the ramp-up/down time as described above was discarded not to be included in any analytical sample. The remaining production was pooled and thoroughly mixed. While many of the pellets were broken into smaller fragments by the moving parts in the reactor, ten intact pellets were randomly selected and set aside for microscopy. The remaining (RE-)biochar was milled in an impact mill with 2 mm sieve. Due to the brittle nature of the material, a fine powder was produced this way. The particle size distribution of comparable biochars after milling was quantified by laser granulometry (Sympatec Helos KFMagic, Sympatec GmbH, Clausthal-Zellerfeld, Germany) and is typically one order of magnitude smaller than the sieve size [27] (> 90 mass-% at < 200 µm for a 2 mm sieve).

To supply larger quantities of (RE-)biochar to other follow-up experiments (not subject to this publication), a second batch of all treatments was produced. To this end, the same raw materials were used (S3 Table), pelleted by Nature Power Pellets (Wolferstadt, Germany) and pyrolyzed on the PYREKA under same process conditions as described above. The pellets produced by Nature Power Pellets had the same dimensions but a higher density. Biochar and RE-biochars from the first production batch were subject to all analytics described in section 4.2 while materials from the second batch were only subject to basic characterization (S4 Table) with additional calculation of y_m, y_c, f_bio and f_rock.

4.2 Biochar analysis

All (RE-)biochars were analyzed following the analytical guidelines of the European Biochar Certificate [78] by Eurofins Umwelt-Ost GmbH (Bobritzsch-Hilbersdorf, Germany). The C_org content of the daf biochar (C_{org_daf}) was calculated as follows

(5)

With C_org(%) being the organic carbon content and ash (%) the total ash content of the (RE-)biochar.

The hydrogen to organic carbon molar ratio (H:C_org) of the bulk sample was corrected for geogenic hydrogen using equation 6.

(6)

With H:C_orgBio being the hydrogen of the biogenic fraction to C_org molar ratio, i.e., the bulk ratio corrected for geogenic hydrogen, %H_Bulk and C_org content of the bulk sample, f_rock the rock fraction of the sample, H%_rock the hydrogen content of the basanite and M_C and M_H the molar mass of carbon and hydrogen. The carbon content of the rock was below the detection limit, and we assume that all rocks relevant to ERW do not contain organic carbon.

Solid state electrical conductivity (SEC) was measured using a Black Gauss I conductivity meter [79] while applying 5 t pressure (equivalent to 160 MPa) to the sample. The mineralogical composition of selected RE-biochars was measured by XRD on a D8 Advance (Bruker, Billerica USA) performed by QMineral (Heverlee, Belgium).

Thermogravimetric analysis (TGA) was conducted using an STD 650 TG-DSC system (Waters, New Castle, United States of America). Dry samples (5 mg) were heated at open, pre-weighed alumina crucibles. The temperature was then increased from 50 °C to 1000 °C at a rate of 10 °C min ⁻ ¹ under N₂ atmosphere with a flow of 50 mL min ⁻ ¹. The recorded total weight loss was adjusted for the sample’s total ash content (including rock content) to represent the weight loss of daf biochar. The fraction of biochar that remained resistant to thermal degradation up to 1000 °C is referred to as BC_1000C.

Benzenepolycarboxylic acids marker (BPCA) were extracted and analyzed in duplicate following Glaser et al. [80] and Brodowski et al. [81]. In brief, the (RE)-biochar samples were digested using trifluoroacetic acid to remove metal ions and organic matter including carbonates and C_org other than those in aromatic moieties. Then the samples were oxidized in a pressurized vessel using nitric-acid. The digests were purified from remaining polyvalent cations using exchange resins. After freeze-drying, BPCAs were trimethylsilyl derivatized and analyzed by gas chromatography connected with a flame ionization detector (GC-FID) (GC-2010 Shimadzu, Kyoto, Japan).

Two types of analytical pyrolysis were performed. Firstly, hydropyrolysis (HyPy) was performed according to Meredith et al. [32]. In brief, 300 mg were mixed with 10 wt% ammonium molybdate-tetrahydrate as a catalyst. The samples were then heated in a hydrogen purged reactor under 150 bar pressure, first from ambient temperature to 250 °C at a rate of 300 °C min ⁻ ¹, and subsequently from 250 °C to 550 °C at 8°C min ⁻ ¹, where samples were held for 2 minutes. The residual sample was weighed and analyzed for its C_org content, referred to as BC_HyPy and presented in percent of the sample’s initial C_org. Secondly, using a HAWK sequential-pyrolyzer with coupled TOC analyzer (Wildcat Technologies, Humble, Texas USA). Here, 10 mg were inserted into the pyrolyzer preheated to 300 °C and held for 3 min under helium atmosphere. The mass of the carbon released as CO and CO₂ was continuously measured and is referred to as S1 and relates to free hydrocarbons such as oils [76]. Then the sample was heated at 25 °C min^-1 to 650 °C with the mass of volatilized carbon summarized as S2. After cooling to 150 °C, the atmosphere was changed to oxygen, and the sample was re-heated to 850 °C at 25 °C min^-1 oxidizing the residual carbon [82]. S1 + S2 was defined as the reactive, labile carbon and the residual carbon was considered unreactive under the conditions applied [35,76].

Fourier transformed infrared spectroscopy was conducted using attenuated total reflectance (ATR) mode with an Invenio X FT-IR Spectrometer (Bruker Corporation, Billerica, USA). Infrared absorbance was determined from 60 consecutive scans, covering a wavenumber range of 400–4000 cm ⁻ ¹. The absorbance spectrum was then smoothed and baseline-corrected using the software Spectragryph v.1.2.26.1.

Raman spectroscopy was conducted under a LabRAM HR 800 Raman microscope (HORIBA Europe GmbH, Oberursel, Germany), which was coupled to an BX41 light microscope (Olympus Corporation, Tokyo, Japan). All measurements used a 532 nm laser (5 mW), which scanned a 100x100 µm area. Spectral acquisition time was 10 s. Five measurements were conducted per treatment and spectra converted D/G intensity- and area ratios. The baseline correction was done using the LabSpec 6 software (Horiba) and consecutive band-fitting according to Ferrari & Robertson (2000) [83] using a Lorentzian line shape (D-band) and a Breit-Wigner-Fano line shape (G-band) with a self-made Matlab-script.

Fast field cycling ¹H nuclear magnetic resonance relaxometry (¹H FFC-NMR, i.e., ¹H Relaxometry) was performed following Conte (2021) [59]. Briefly, samples of 1 g sample was weighed in a 10 mm NMR tubes, then water was added in 2.5:1 ratio. The mixture was vortexed for 5 minutes and left to settle for 24 h. Measurements were conducted on a Stelar Smartracer Fast-Field-Cycling Relaxometer (Stelar SRL, Mede, Italy) set to 25 °C. The proton spins were polarized at a polarization field (B_POL) corresponding to a proton Larmor frequency (n_L) of 10 MHz for a period (T_POL) of about five times the T₁ estimated at this frequency. After each B_POL, the magnetic field intensity (B_RLX) was set at the proton Larmor frequency, n_L, of 1 MHz. The period τ, during which B_RLX was applied, has been varied on 32 logarithmic spaced time sets. Free induction decays (FID) at each τ value were recorded following a single ¹H 90 ° pulse applied at an acquisition field (B_ACQ) corresponding to n_L of 7.2 MHz. A time domain of 100 μs sampled with 512 points was applied. Field-switching time was 3 ms, while spectrometer dead time was 15 μs. For all the experiments, a recycle delay of 2 s was used. All decay curves acquired were analyzed by the UPEN algorithm to obtain the relaxograms [84,85]. Relaxograms were finally used to calculate the Pore Connectivity Index (PCI) [86] (additional S12 Text).

For SEM, RE-biochar pellets were mounted on an aluminum sample holder using carbon pads (Plano GmbH, Wetzlar, Germany) and sputter-coating them with gold (3–5 nm) using a Cressington 108auto (TESCAN GmbH, Dortmund, Germany). Analysis was conducted with a JSM-6610 LV (JEOL Ltd., Tokyo, Japan) at a working distance of 11 mm and acceleration voltage of 15 kV. Backscattered electron images (BES) were acquired, and Energy-dispersive X-ray (EDX) analysis of the same clippings was carried out using a 20 mm² Oxford X-max detector (Oxford Instruments, Abingdon, United Kingdom).

For Scanning Transmission Electron Microscopy (STEM) with Electron Energy Loss Spectroscopy (EELS) specimens, Re-biochar samples (< 3 mm), were embedded in epoxy resin (Spezifix 40, Struers, Willich, Germany). Ultrathin sections of 50 nm were prepared using a Leica EM UC6-NT ultramicrotome (Leica Microsystems, Vienna, Austria) equipped with a 35° ultra diamond knife (Diatome, Biel, Switzerland). The sections were transferred onto lacey carbon-coated copper grids using a Perfect Loop (Diatome). The measurements were performed using a monochromated probe corrected FEI Titan G2 60–300 STEM-microscope with a Schottky-type X-FEG electron source operated at 300 kV. The microscope was equipped with a Gatan Imaging Filter, Quantum including a direct electron detection camera (K2 Summit) and a CCD camera (Ultrascan1000) with dual-EELS capability. The microscope was operated in monochromated mode with a semi-convergence angle of 15.6 mrad and a beam current of about 200 pA, resulting in a spatial resolution of 0.12 nm and an energy resolution at full-width at half maximum of 0.3 eV. The measurements were performed at a temperature of −160 °C using a cryo TEM-holder (Mel-Build, Fukuoka, Japan). The high angular annual dark field (HAADF) detector was used for micrographs acquisition.

Gas adsorption was performed using nitrogen isotherms at 77 K on an AUTOSORB-IQ volumetric sorption device (3P-Instruments GmbH & Co KG, Odelzhausen, Germany). Samples of 0.1-0.2 g biochar were degassed at 150 °C for 12 h with a final vacuum of <10 ⁻ ² Pa applied before analysis. Surface area (Brunauer-Emmett-Teller method) was calculated following the guidelines of DIN ISO 9277:2014. The pore volume distribution is inferred from the total pore volume determined at P/P₀ = 0.99 (Gurvich’s law) and measured adsorption isotherms according to the density functional theory (DFT) [87].

The magnetic susceptibility was measured on 6.4x10^-6 m³ cubes of manually compressed material. Measurements were conducted in a 200 A m^-1 field at low (976 Hz) and high (15616 Hz) frequency (χlf and χhf) using a MFK1-FA Kappabridge (AGICO Inc., Brno, Czech Republic). The frequency dependent magnetic susceptibility was calculated as [(χlf—χhf)/χlf] × 100 [57,60].

Random Reflectance (Ro%) was measured on (RE)-biochar fragments between 63 and 1000 µm, that were embedded in an epoxy-resin and polished (ISO 7404−2 2009). White light reflectance measurements (ISO 7404−5, 2009) were carried out using a Zeiss Axio Imager II microscope (Zeiss, Oberkochen, Germany) with integrated Discus-Fossil system (Hilgers Technisches Buero, Königswinter, Germany). Measurements were conducted manually on n = 500 randomly selected, exposed macerals, with each measurement covering an area of 0.3 µm². Reflectance measurements were calibrated using KB N Lasf1.317 and KB cubic zirconia 3.11 standards.

4.3. Statistical considerations

Rock-enhanced biochars were produced in a single continuous batch, with RE-biochar yield recorded at regular intervals and the coefficient of variation of f_b and y_c calculated. In general, analyses were conducted on representative samples without replication (n = 1). Depending on the analytical method varying numbers of repeated measures were conducted per sample (e.g., Raman spectroscopy with n = 5, FTIR with n = 60, and random reflectance with n = 500). Mean, median, skewness and kurtosis of each Ro distribution were calculated using Microsoft Excel.