Phosphorus transformations in plant-based and bio-waste materials induced by pyrolysis

The reeds, rice husks, and cocoa prunings have mineral contents substantially higher than those of the woody acacia material (Table 1). The relatively sparse mineral concentrations in the acacia bark are typical of woody feedstock. Total P, by weight, in the plant-based materials ranges from 0.3 (acacia bark) to 1.1 g kg⁻¹ (cocoa prunings). The P contents of the reed, rice, and cocoa samples are similar to previous reports (e.g. Zheng et al. 2013; Gao et al. 2015); the relatively high P content of the cocoa prunings possibly reflects recent P fertiliser history. Compared with the plant materials, the lower C and predominantly higher mineral contents of the bio-waste feedstocks reflect the high-mineral diets of animals and humans. Total P ranges almost by an order of magnitude, from 13 to 104 g kg⁻¹, decreasing in the order: bone > BIOS sewage sludge > pig slurry > SLDG sewage sludge > poultry manure (Table 1).

Generally, the C and mineral contents of the biochars follow approximately the same orders as those for the feedstocks (Table 1); this common trend reflects the close dependency of biochar properties on the composition of the feedstock (Verheijen et al. 2010). Relatively high-mineral containing grass, grain husks, straw residues and manures generally produce biochar with high ash contents, in contrast to that from woody feedstocks (Demirbas 2004). For instance, some manure (e.g. chicken litter) and sewage sludge (e.g. Al-treated biosolids) biochars can contain as much as 50 and 5% (by weight) ash and P, respectively (Amonette and Joseph 2009; Wang et al. 2012). For all of the plant-based materials, the C content of biochar is higher than that of the source feedstock (Table 1). This increase in concentration is attributed to pyrolysis removing a considerable mass of material in the form of volatiles (e.g. hydrocarbons, H₂, CO and CO₂), and leaving behind a stable matrix of largely aromatic C (Baldock and Smernik 2002; Demirbas 2004). During thermal degradation of the plant biomass, N content vaporises at relatively low temperatures, after peaking at around 300 °C (Baldock and Smernik 2002). Conversely, due to increased stability, P, S and most metals vaporise at temperatures that are considerably higher; e.g. P at around 1000 °C (Zhang et al. 2012). Similar processes occur during the pyrolysis of the bio-waste materials; however, the higher proportions of labile C and N in bio-wastes relative to those in plants explain the thermally induced decreases in these parameters (Table 1). For all materials, the differential vaporisation of elements during pyrolysis explains why P is retained and becomes concentrated in the biochars. Indeed, such was the concentrating effect in the reed, rice and cocoa biochars, that the total P contents of these three materials reached values (1.6–3 g kg⁻¹) comparable with those of green waste compost (Chan and Xu 2009).

In general, the combination of P standards yielding the best LCFs for the plant materials were hydroxyapatite (HAP), P sorbed to Fe oxides, and ATP (Na salt) (Table 2). Hydroxyapatite was identified by LCF as the dominant P form in the reed. The spectra of the reed feedstock and corresponding biochar displayed a prominent post-edge shoulder at around 2154.5 eV [Fig. 1a: line (i)], characterising calcium phosphates. Whilst evident, the equivalent feature is weaker for the other plant spectra, reflecting lower contents of HAP (Table 2). The prominence of the HAP reflects the fundamental importance of Ca in plant cellular metabolism, whereby Ca-bearing minerals comprise as much as 50% of known biominerals, with approximately half in the form of calcium phosphates (Weiner and Dove 2003). Weiner and Dove discuss in detail the abundance of carbonate-substituted HAP in biological material, but note that there is no evidence for the biological formation of the non-carbonated form. On the contrary, a combination of X-ray powder diffraction (XRD), thermal analysis and Fourier transform infrared spectroscopy (FITR) has identified non-substituted HAP as a significant inorganic constituent of some plant species (Shaltout et al. 2011). The LCF was unable to distinguish between substituted and non-substituted forms of HAP in the current study.

Among the plant-based feedstocks and biochars, there was no clear suggestion of a pre-edge peak (around 2149 eV) in the P K-edge spectra (Fig. 1a); this points towards a lack of evidence for precipitation of non-crystalline or crystalline Fe phosphate minerals (Hesterberg et al. 1999; Shober et al. 2006). Nonetheless, Fe-bearing minerals comprise approximately 40% of biominerals (Weiner and Dove 2003); many of these exist as a variety of oxides, hydroxides and hydrous oxides—all characterised by a strong affinity to sorb P. A further possible explanation for the Fe–P phases is contamination of the plant materials with attached soil or dust particles.

The LCF indicated that ATP (Na salt) was the only organic P form in the plant-based materials and their biochars (Table 2). Whilst it is likely that ATP does exist in the feedstocks, it is well known that the XANES spectra of organic P reference compounds lack distinguishable pre- or post-edge features (Peak et al. 2002; Shober et al. 2006); this renders organic P speciation difficult (Toor et al. 2006). As such, it is likely that the ATP is represented in unrealistically high quantities in the feedstocks and probably comprises additional organic polyphosphates also, such as phytic and nucleic acids.

As expected, the proportion of total P as organic P (albeit poorly characterised) in all four plant-based samples decreased following pyrolysis. This loss at moderate pyrolysis temperatures can be attributed to the cleaving of organic P bonds in the feedstocks, and has been demonstrated by others, using ³¹P-NMR spectroscopy; for example, Uchimiya and Hiradate (2014) reported a major decline in phytate from residues of cotton seed hull upon pyrolysis at around 350 °C. It is speculated that the persistence of the ATP signal following pyrolysis is due in part to the formation of inorganic polyphosphates. For three of the four plant materials (reed, rice, and acacia), the proportions of HAP increased markedly with pyrolysis; indeed, according to the LCF results, estimates in the biochars of rice and acacia doubled those in the corresponding feedstocks (Table 2). While some workers have claimed that the removal of organic matter by combustion (e.g. calcination) is necessary to thermally form crystalline HAP in plant biomass, the current findings appear to agree with Zheng et al. (2013) who reported increases in crystalline forms of Ca- and Mg phosphates following pyrolysis.

In the case of the cocoa, pyrolysis appears to shift the P chemistry towards pyrophosphates (Mg₂O₇P₂), at the expense of all other dominant forms (Table 2). A smaller increase is observed for the rice. Previous studies on the pyrolysis of plant biomass at temperatures between 350 and 650 °C have observed a dominance of pyrophosphate in the biochar; this is thought to originate from the degradation of phytates (Uchimiya 2014; Degenstein et al. 2015).

The LCF results indicated that HAP, P sorbed to Fe oxides, and organic P (represented by ATP) were dominant in all but one (bone chips) of the five bio-waste feedstocks (Table 2). In addition, the pig slurry recorded a substantial proportion (30%) of hydrated soluble P salt (Na₂HPO₄·xH₂O) (Table 2).

According to the LCF, as much as 98% of the P in the bone chips exists as HAP (Table 2), and is characterised by the permanent post-edge shoulder around 2154.5 eV (line i), as shown in Fig. 1b. An additional, well-defined calcium phosphate feature appears around 2163 eV (line ii). Pyrolysis of the bone appeared to transform almost half of the HAP to metal pyrophosphate. Given the dominance of Ca in the bone, it is possible that a portion of total P in the bone char that was fit as Mg₂O₇P₂ (44%) was actually Ca₂O₇P₂. Based on the LCF of P K-edge XANES spectroscopic characterisation of rendered bones sampled from the poultry industry, Zwetsloot et al. (2014) reported lower proportions of HAP (60% of total P) than those estimated here. Moreover, these workers observed marked increases in HAP (to as much as 93%) following pyrolysis at temperatures (350–550 °C) similar to those applied in the current work. Further comparisons between the studies are difficult, primarily due to the absence of pyrophosphate standards in Zwetsloot et al. (2014), and also the possibility that the crystallinity of the HAP standards differed between the studies. The LCF indicated that ATP accounted for as much as 52% of total P in the poultry manure (Table 2). As we discussed previously, it is likely that the P estimated as ATP includes contributions from a broad range of P polymers. In the case of non-ruminant (e.g. poultry and pigs) manures, the dominant form is probably phytate. Phytic acid composes 60–80% of total P in cereals, oilseeds and legumes that are commonly used in livestock feed. As such, owing to its poor digestibility, phytic acid can account for as much as 60% of total P in manures, with smaller amounts of diester P compounds; e.g. phospholipids and DNA (Shober et al. 2006; Uchimiya and Hiradate 2014). Similar to its decline in the plant-based materials, the size of the organic P pool in the poultry manure decreased markedly with pyrolysis (Table 2), with the reduced ATP signal (19% of total P) probably comprising inorganic forms of polyphosphate. Consistent with the ³¹P-NMR findings of Uchimiya and Hiradate (2014), the majority of the mineralised P at relatively low pyrolysis temperatures transformed into inorganic orthophosphate and pyrophosphate; dominant inorganic compounds appearing in the poultry manure biochar included HAP and inorganic Mg₂O₇P₂ (Table 2). According to the LCF, P sorbed to Fe oxides maintained a significant presence, increasing slightly from 14 to 16% of total P following pyrolysis (Table 2). However, such small variations could be assessments of noise in the analysis rather than real changes, and should be regarded with caution, especially in view of the relatively poor fit (R-factor = 0.006) for the poultry manure biochar compared with its feedstock (R-factor = 0.003) (Table 2; Fig. S1).

In spite of net increases in aromatic C at the expense of aliphatic groups, this region of pyrolysis temperature is known to produce labile organic matter that can condense on the biochar surface (Lu et al. 2013); this organic matter contains certain hydroxyl groups that may coordinate with Ca and favour the formation of HAP. It is possible that the aromatic C fraction stabilises the HAP on the biochar surface as nano-crystals (Uchimiya 2014). Possible mechanisms for the appearance of pyrophosphate in the poultry manure biochar include the general fragmentation of long-chain polyphosphates into shorter chains, or the thermal degradation of phytate to produce orthophosphate that thermally dehydrates to produce pyrophosphate. Regardless of the source, it is possible that the pyrophosphate becomes stabilised by forming complexes on the biochar surface (Uchimiya and Hiradate 2014).

In addition to the substantial amounts of HAP (23%) and polymerised (17%) P in the pig slurry, soluble Na₂HPO₄·xH₂O accounted for as much as 30% of total P (Table 2). Similarly, previous workers have reported relatively high concentrations of water extractable P in pig slurry (Leinweber 1996). Compared with all of the other materials studied, the pig slurry exhibited relatively little change in P chemical forms due to pyrolysis. The minor, apparent decreases in the proportions of sorbed P (2%) and organically bound P (4%) were matched almost by increases in HAP and Na₂HPO₄·xH₂O (3%). As noted earlier, such small shifts are probably assessments of noise, in spite of the relatively confident values for goodness-of-fit (R-factor = 0.001) (Table 2; Fig. S1). As previously speculated, the ATP fraction possibly conceals thermal transformations from organic to inorganic polyphosphates. Moreover, it is also possible that drying the pig slurry at 60 °C transformed the P into forms that remained relatively resistant to the subsequent pyrolysis.

The two sewage sludges (SLDG and BIOS) were dominated, but to different extents, by HAP, P sorbed to Fe oxides and ATP. A minority (< 5%) of P in the SLDG was precipitated as hydrated AlPO₄ (Table 2). These dissimilarities occur in spite of the biosolids having received similar sludge treatment processes, and possibly reflect differences in composition between the two source wastewaters. The presence of polyphosphates in both sludges is explained by the primary process for removing soluble P (and N and K) from municipal sewage; prokaryotic accumulation is a microbiological process that accumulates the P in organic polyphosphate compounds as intracellular granules (Mehta et al. 2015). Pyrolysis of both sludges yields an overall increase in metal-associated P at the expense of organic P, but the transformations differ between the two materials. The large increase in Ca-P in the BIOS biochar, relative to that in the SLDG (Table 2) is evident in the spectrum at around 2163 eV (line ii), as shown in Fig. 1b. According to the LCF analysis, there is no longer a presence of adsorbed P on Fe oxides but, instead a marked appearance of FePO₄·xH₂O (Table 2). However, the absence of a clear, pre-edge peak in the XANES spectra questions the validity of this analysis. The apparent total loss of organic P (from 24% of total P) in the BIOS feedstock is almost matched by the appearance of pyrophosphate in the biochar (20%), indicating that the Mg₂O₇P₂ was derived mostly from the transformation of intracellular, long-chain polyphosphates. In the case of the SLDG, no new P compounds are introduced by pyrolysis; instead, as reflected in the similarity between the two spectra, all three mineral P forms simply increase, albeit by different extents (Table 2, Fig. 1b).