Phosphorus chemistry in plant charcoal: interplay between biomass composition and thermal condition

Sampling

Plant fruit (i.e. cone), wood (i.e. twig and bark) and leaf (i.e. needle) parts of Pinus rigida (pitch pine) and Pinus strobus (white pine) were collected from the Albany Pine Bush Preserve, and Picea abies (Norway spruce) cone and needle were collected from the Five Rivers Environmental Education Center, both located in Albany, New York. Leaf, wood and bark of deciduous trees, including Quercus rubra (red oak), Quercus alba (white oak), Acer rubrum (red maple) and Fagus (beech), were collected from a residential neighbourhood, also in Albany. Specifically, wood parts (bark, twig and hardwood) were collected from living plants during tree cutting. The biomass was dried in an oven at 70°C until there was no further weight change, then pulverised using a blender. Detailed summary of all biomass types used in this work is included in Table 1.

Preparation of charcoal and natural fire residue samples

The pulverised biomass was compacted in glass containers, wrapped with aluminium foil and buried in a sand bucket (at least 10 cm below the surface), which was heated in a benchtop muffle furnace (Thermolyne, Thermos Scientific). Because the focus of this work is to compare charcoal P chemistry across a wide range of plant species and parts, charcoal samples (n = 22) were produced consistently under controlled thermal conditions (pyrolysis temperatures 300, 450, and 600 °C for 6–8 h). The samples were labelled as: biomass abbreviation + pyrolysis temperature. For example, SC600 was for charcoal from pyrolysis of spruce cone at 600°C. Details of sample labelling can be found in Table 1. In addition, in order to demonstrate the relationship between combustion completeness and P mobility, three sets of burning residues were also included: (1) complete burning products – ash samples (n = 2) that were produced from spruce cone and red oak wood in air at 550 °C for ~6 h (Wu et al. 2023); (2) mixed fire residue samples (n = 2) that were produced from igniting spruce cone and red oak wood in air, simulating natural fire conditions; and (3) burning residues (n = 5) from generators that burn woods (provided by Resource Management, Inc.). These samples represent realistic fire burning products from varying burning conditions and consist of both charcoal and ash. Combustion completeness of these samples was evaluated based on mass recovery and total Carbon (C) content of residues.

Fig. 1.

(a) ³¹P liquid NMR spectra of the extracts from Norway spruce cone (SC-R) and needle (SN-R), red oak wood (ROW-R), spruce cone sample spiked with phytic acid (SC-R w/Phytate). (b) ³¹P NMR spectra of the extracts from charcoal samples of Norway spruce cone and red oak wood produced at 300–600 °C and residue samples from simulated fire condition (SC-N and ROW-N). (c) Relative abundance of different P species in the extracts of raw biomass based on ³¹P NMR peak density. (d) Relative abundance of different P species in charcoal extracts. Description of the charcoal samples can be found in Table 1.

Characterisation of physicochemical properties of charcoal

Biomass elemental composition (C, H, N, S) was determined by an CHNS Elemental Analyser (2400 Series II, PerkinElmer Inc, USA). Analysis of total P and other macronutrients (i.e. K, Ca, Na, Mg, Mn, Al, and Fe) in biomass samples was performed by the Cornell Nutrient Analysis Laboratory. The elements were solubilised using a hotplate digestion method and determined by inductively coupled plasma–optical emission spectrometry (ICP-OES) following the EPA Method No. 6010-B. Elemental contents (P and metals) in charcoal were calculated based on the mass recovery (i.e. char yield) of each biomass type, because these macronutrients are assumed to completely remain in charcoal due to their high volatilisation temperatures (>750 °C) (Bodí et al. 2014). Thus, total P in charcoal was calculated using the following equation: Total P in charcoal (mg kg⁻¹) = Total P in Biomass/char yield. Contents of K, Ca, Na and Mg in charcoal were also calculated similarly. Brunauer, Emmett and Teller (BET) specific surface area (SSA, N₂ adsorption) of selected charcoal samples (Table 1) from different biomass samples yet the same production temperature (e.g., 600°C) was measured using a gas adsorption analyser (3Flex, Micromeritics, Inc.).

Solution extraction of P

Characterisation of P speciation

Molecular moieties and complexation states of P in biomass and charcoal samples were characterised using ³¹P liquid NMR (n = 6 for biomass samples, n = 14 for charcoal samples) and P K-edge XANES (n = 1 for biomass samples, n = 12 for charcoal samples).

Physical and chemical properties of biomass and charcoal

Parts of representative coniferous and deciduous trees were selected to produce charcoal at temperature from 300 to 600 °C, resulting in charcoals with a range of physicochemical properties (Table 1 and Supplementary Table S2). This enabled the testing of our hypothesis that biomass composition (specifically starting P species and elemental stoichiometry) and pyrolysis temperature control P speciation and mobility of charcoals. For example, P content in the selected biomass ranges from 135 to 1000 mg kg⁻¹, and that of other macro-nutrients (e.g. Ca, K, Mg and Na) also varies for more than an order of magnitude. The varying elemental contents resulted in a big range of metal-to-P molar ratios, with Ca/P ranging from 1.8 to 160, K/P from 0.6 to 5.6 and Mg/P from 0.8 to 3.1 (Table 1).

P speciation in biomass and charcoal

Detailed P speciation, including molecular structure and complexation environment, was characterised for biomass and charcoal, to explore the thermochemical mechanisms of P transformation during incomplete combustion and the impacts of fuel composition and thermal condition. Overall, all biomass types consist of diverse molecular structures, whose relative abundance varies among different plant species (for the same compartment) and different compartments (for the same species) (Fig. 1). The P molecules identified (by ³¹P liquid NMR) include orthophosphate, phosphate-monoester and -diester, pyrophosphate and polyphosphates (Fig. 1), which are typical P species presented in plants (Noack et al. 2014). In the analysed samples, organophosphates (mono- and di-esters) account for about 40–50% of total P. The main difference is the relative abundance of orthophosphate and condensed phosphates: (1) polyphosphate is relatively abundant in the cones and needles of pine trees; and (2) spruce cone and oak wood contain pyrophosphate that is negligible in other biomass samples. Relative abundance of organophosphates and condensed phosphates depends on many factors, such as: (1) physiological functionality of specific P species in the plant parts, e.g. phospholipids, adenosine triphosphate and nucleic acids in cell structure, metabolism and reproduction, respectively; (2) plant growth stage, e.g. leaf senescence (Smith et al. 2015); and (3) nutrient status (Hidaka and Kitayama 2011).

Following pyrolysis, only orthophosphate and pyrophosphate were identified, and abundance of pyrophosphate varies among charcoal samples (Fig. 1b, d). For example, abundant pyrophosphate (~40–50%) was found in the extracts of chars derived from Norway spruce cone, regardless of pyrolysis temperature (300–600°C). However, pyrophosphate abundance is relatively small (even absent) in all other tested chars. Similarly, only orthophosphate and pyrophosphate were identified in residues from simulated burning (SC-N and ROW-N), with pyrophsophate in SC-N more abundant than that in ROW-N. Because ³¹P liquid NMR only identifies and quantifies P structures in liquid extracts, which is a destructive analysis and does not reveal complexation environments of the P structures, P K-edge XANES analysis was used to estimate in situ the P species in chars. It is worth noting that LCF of P K-edge XANES data is relatively insensitive (particularly to organophosphates), with fitting errors at ~5–10% for even spectrally distinct mineral species (Werner and Prietzel 2015). Therefore, the XANES data was interpreted semi-quantitatively only. First, XANES data shows a gradual reduction in organophosphate abundance (fitted as phytate) and formation of abundant crystalline calcium phosphates (fitted as hydroxyapatite) in chars (Fig. 2 and Supplementary Table S2). For example, an increasing amount of hydroxyapatite (from 0 to 20%) and a decreasing amount of phytate (from 50 to 30%) were found in spruce cone chars (from 300 to 600 °C). The result is consistent with the ³¹P liquid NMR result showing that organophosphates are mineralised during pyrolysis. Second, although abundant Ca and Mg phosphates were similarly identified in all char samples that agree with the stoichiometric abundance of these elements, their abundance differs among chars. For example, spruce cone chars have relatively low contents of Ca phosphates (~10–50%) compared with other types of charcoals (e.g. from red oak and pitch pine biomass), corresponding to its small Ca/P molar ratio (1.8) (Table 1). Variations in pyrophosphate abundance and crystalline Ca phosphates in chars can be related to initial P speciation in their biomass, which will be discussed in the last Section: Mechanisms of P transformation during burning and impacts on P mobility.

Fig. 2.

(a) A series of P K-edge XANES spectra for samples from spruce cone. (b) Relative abundance of P species estimated by LCF of the XANES data of spruce cone samples (raw biomass, chars and 600 °C chars after EDTA or HCl extraction). (c) Relative abundance of P species of charcoal samples produced at 600 °C, including spruce needle (SN), red oak wood (ROW), red oak leaf (ROL), red oak bark (ROB), pitch pine bark (PPB) and pitch pine cone (PPC). Samples with ‘-EDTA’ and ‘-HCl’ represent samples after EDTA and HCl extraction, respectively. Description of the charcoal samples can be found in Table 1.

Solubility and dissolution kinetics of charcoal P

P solubility estimated by sequential extraction

Phosphorus speciation and solubility in the chars were also estimated by the modified Hedley extraction, with P of different chemical environments being sequentially extracted by water, NaHCO₃, NaOH and HCl. The residual P possibly represents P in the C matrix that is not accessible or extractable by chemical solutions. Following pyrolysis, P in charcoal was greatly immobilised, as 70–90% are residual P and the labile P (H₂O and NaHCO₃ extractable P) accounts for only 2–15% (Fig. 3). In general, P becomes less available as temperature increases, except for chars of spruce cone, where the labile P increases with temperature (8–15%). Phosphorus extracted in the labile pool is mostly orthophosphate, with <1% non-orthophosphates (the difference between total P and orthophosphate) in most charcoal. The P_o is likely pyrophosphate that was identified in ³¹P NMR, becasue it does not react with molybdate reagent before persulfate digestion (Doolette and Smernik 2011). Only a small P fraction (1–5%) partitions in the NaOH pool, which is consistent with the relatively low Fe and Al abundance (compared with P) and low Fe and Al bound P in the chars. The HCl P pool is relatively abundant (up to 20%), but its pool size varies broadly among chars. As shown in chars of spruce cone, pitch pine cone and red oak bark, the HCl P pool decreases as pyrolysis temperature increases. This is contradictory to the fact that P is increasingly mineralised at increased temperatures, forming more crystalline Ca phosphates as shown by XANES data (Huang et al. 2017). The HCl P pool also varies among 600 °C chars of different biomass types, ranging between 2 and 17%. The large residual P pool and great variations in P partitioning among the sequential extraction pools in chars suggest that compared with chemical speciation, matrix effect is likely to play a more important role in controlling P solubility.

Fig. 3.

Partitioning of P in different extraction pools determined by the modified Hedley fractionation method for charcoal samples. Description of the charcoal samples can be found in Table 1.

Release kinetics of Olsen-P from charcoal

To further explore the mechanism of P mobilisation from chars, rate and extent of P release during NaHCO₃ extraction were measured and correlated with physicochemical properties (i.e. specific surface area and elemental stoichiometry) of the chars (Fig. 4). This NaHCO₃ extractable P is exchangeable P (P_ex) that is considered to be directly bioavailable in soils, also referred to as mobile P or labile P herein (Olsen 1954). As shown in the release rate data, the Olsen-P was rapidly released within the first few hours and stabilised afterwards (Fig. 4a). This suggests similar chemical nature of the P_ex among different chars. Three kinetic models (i.e. first order, second order, and simple Elovich equations) were applied to understand the possible release mechanisms. The simple Elovich equation is a best fit because it yields better s.e. and R² values (Supplementary Table S3). The simple Elovich equation has been extensively used to describe the desorption/release system, especially when first order equation is insufficient to describe the heterogeneous and multiple desorption/release processes (Chien and Clayton 1980). This Olsen-P pool (approximate to sum of the H₂O and NaHCO₃ pools) varies among chars and was found to correlate linearly with the total P (Fig. 4b). However, the percentage of Olsen-P does not correlate with either surface P concentration or the Ca/P molar ratios (or (Mg + K)/Ca), which represent physical and chemical factors, respectively (Fig. 4c, d).

Fig. 4.

(a) Temporal release of exchangeable P for selected plant charcoal and its fitting by the simple Elovich equation. (b) Correlation between Olsen-P and total P content. (c) Correlation between Olsen-P and surface P concentration (total P/surface area). (d) Correlation between Olsen-P and (Mg + K)/Ca molar ratio of charcoal samples. Description of the charcoal samples can be found in Table 1.

Impacts of combustion completeness on P chemistry and mobility

Fire varies in intensity and duration, resulting in the formation of fire residues with different properties. One of these is burning completeness, which affects the relative portions of ash and charcoal (Úbeda et al. 2009; Pereira et al. 2012). A high degree of combustion completeness is characterised by heavy white ash deposition and substantial loss of organic matter, always correlating with a high degree of fire intensity or severity. To evaluate the effect of burning completeness on P mobility, three types of samples representing increasing degrees of combustion completeness produced from the same biomass were compared: (1) charcoal; (2) residues from simulated open burning (i.e. natural fire residue samples); and (3) ash (i.e. complete combustion product) (Fig. 5). The results showed that P becomes more mobile as burning completeness increases. For example, the fraction of Olsen-P increases from 10 to 70% for samples of spruce cone and from 10 to 40% for those of red oak wood at pH 8 (Fig. 5a). A similar trend was observed for pH 4.0 (Fig. 5b). We further correlated the Olsen-P with the C contents of a set of combustion samples (from simulated burning in the lab and industrial boilers that burn entirely woods), and the result showed that the Olsen-P decreases with increasing C content (Fig. 5c). The result clearly shows that P in charcoal is less soluble than P in ash, and increasing combustion completeness may enhance P solubility and mobility.

Fig. 5.

Phosphate release kinetics for fire residues of different burning completeness at (a) pH 8.0 (0.5 M bicarbonate buffer) and (b) pH 4.0 (0.2 M acetate buffer). (c) Correlation between Olsen-P and total C content (wt%) of fire residues. SC-Natural and ROW-natural refer to residue samples from simulated burning.

Mechanisms of P transformation during burning and impacts on P mobility

Charcoal is the incomplete burning product of fires; it is a significant nutrient pool and different from the complete burning product, ash, in terms of chemical composition and property. This study advances the current understanding of P thermochemistry during charcoal formation and mechanisms of P mobilisation from charcoal, via thorough characterisation of P speciation and solution extractability across a wide range of charcoal types.

In terms of P speciation, our results show that although P is similarly mineralised among different samples during charcoal formation, the resulting P speciation (i.e. molecular structure and complexation state) varies greatly among plant species and parts, which pinpoint the roles of starting P speciation and elemental stoichiometry. First, as shown in ³¹P NMR data, orthophosphate and pyrophosphate are the sole structures identified in the charcoal extracts, despite the diverse P structures (such as different organophosphates and polyphosphates) in the biomass feedstock (Fig. 1). The data first suggested that phosphate ester (O–R) and phosphoanhydride (P–O–P) bonds (existing in organophosphates and polyphosphates) are cleaved during pyrolysis, even at temperature as low as 300 °C. Varying abundances of pyrophosphate were identified (in spruce cone chars and white pine cone chars) and remained stable at the tested temperature range (300–600 °C). Pyrophosphate was also inconsistently detected in natural fire residues. For example, pyrophosphate accounts for about 0.2–2.8% of total extractable P in fire residues from a pine forest fire but is absent in the residue from a eucalypt forest fire (García-Oliva et al. 2018; Santín et al. 2018). Without information on the biomass composition and fire conditions, causes of this variation were unknown. Pyrophosphate could be potentially inherited from the starting biomass or formed from the polymerisation of phosphate molecules. Phytate is considered the precursor for pyrophosphate, as shown in studies on phytate-rich organic wastes such as manure and nut shell (Uchimiya et al. 2015). Our data suggested that phytate is not present in the starting biomass (Fig. 1a) and is unlikely to be the source of pyrophosphate in spruce cone charcoals. Instead, it is more likely to inherit from the biomass that contains pyrophosphate. Second, our result also suggests that the complexation state of P (complexation with metals such as Ca, Mg, K and Na, as well as mineralogy) is likely determined by both the starting P speciation and thermochemical reactions during biomass burning. Calcium and Mg phosphates were the main minerals identified by XANES, and their abundance increases with increasing metal to P molar ratios in the starting biomass (Fig. 2). With Ca and Mg being the most abundant metals in biomass, they are the most likely ones to form complex various P structures and remain unchanged during heating. However, cleavages of phosphate ester (O–R) and phosphoanhydride (P–O–P) bonds during heating induce changes to the complexation state, because the organic moieties and adjacent P will be replaced by a metal cation. The metal to replace organic moieties during organophosphate mineralisation may depend on its localised abundance and P-binding affinity. Therefore, variations in elemental stoichiometry and the abundance of organophosphates and condensed phosphates in biomass, as well as thermochemical reactions during heating, likely determine P speciation in charcoal.

Regarding P solubility, our results show that most P are likely entrapped physically in the C matrix, and the soluble P fraction is mostly inorganic phosphates that are mobilised through dissolution. Although P in charcoal exists mostly as inorganic phosphates (as summarised above), only a small portion (less than 20% for most charcoals) was extractable by the sequential extraction solutions that include concentrated HCl. Charcoal is a carbonaceous porous material and a fraction of P is likely bound within the C matrix and not exposed to extraction solution, despite the soluble nature of the P species. The extractable P partitions are found mainly among H₂O, NaHCO₃ and HCl, along with the fast release kinetics of Olsen-P, suggesting that charcoal P is available mainly through solubilisation of inorganic phosphates on charcoal surface (Figs 3 and 4). However, the relative abundance of these P pools varies among charcoals, which is possibly controlled by their different P chemical forms and surface exposure. For example, spruce chars tend to have more soluble P than other chars formed at the same thermal condition, because of their low Ca/P ratios and potentially more soluble Mg and K phosphates. The matrix effect is reflected in varying P solubility among different T chars from the same biomass (Fig. 3). In general, charcoal becomes more aromatic and porous as heating temperature increases, whose effects on P solubility are counteractive. As a result of both chemical and physical factors, P solubility does not correlate simply to either surface P concentration or metal ratios (Fig. 4). The matrix effect is also manifested in enhanced P mobilisation as a result of complete burning, with more P being freed from the C matrix and its solubility depending primarily on the P chemical forms (Fig. 5).