Differentiating Inorganics in Biochars Produced at Commercial Scale Using Principal Component Analysis.

Characterizing the inorganic phase of biochar, beyond determining element concentration, is needed for appropriate application of these materials because mineral forms also influence element availability and behavior. Inorganics in 13 biochars (produced from Poultry litter, switchgrass, and different types of wood) were characterized by proximate analysis, chemical analysis, powder X-ray diffraction (XRD), and scanning electron microscopy with energy-dispersive X-ray (SEM-EDX) spectroscopy. Principal component analysis (PCA) was used to compare biochars and characterize associations between elements. The biochars were produced using commercial-scale reactors and represent materials with properties relevant to field application. Bulk inorganic concentration and composition were responsible for differentiating biochars after PCA of chemical data. In comparison, differentiation based on PCA of diffractogram fingerprints was more nuanced. Here, contributions from cellulose and turbostratic crystalline C influenced separation between samples. It was also sensitive to mineral forms of Ca (whewellite and calcite). Differences in crystalline C and Ca minerals separated two biochars generated from the same willow feedstock using the same pyrolysis conditions at different temperatures. PCA of 606 SEM-EDX point scans revealed that inorganics belong to four main clusters containing Ca, Fe, [Al, Si], and [Cl, K, Mg, Na, P, S] consistent with XRD identification of calcite, magnetic Fe-oxide, silicates, and sylvite. It further suggested that amorphous P-containing minerals associated with Ca (not identified through XRD) were constituents of willow and poultry litter-derived biochars. However, unlike PCA of XRD, it was not able to differentiate the two biochars derived from willow. The three analysis methods provided different perspectives on the properties of the biochar inorganic phase. Combining information from multiple methods is needed to better understand the inorganic composition of biochars.

Introduction

Biochars are carbonaceous residues produced by pyrolysis of biomass and are known to increase moisture retention, improve air permeability, and improve soil structure.[1] Biochars contain an organic C (OC) and an inorganic (ash) component. Biochar OC is defined by its condensed aromatic structure content, and this property is rightly the subject of extensive study.[2,3] However, there have been fewer studies dealing with biochar ash and it is important to further understand the inorganic components of biomass pyrolysis products as these materials become more widely used.

Biochar ash is important because of its acid neutralization potential and because it may influence agrochemical retention, nutrient levels, and microbial populations in soils.[4] Inorganics in biochars may come from the feedstock or may be introduced during collection, processing, pyrolysis, or postproduction.[5] During pyrolysis, ash concentrations are enriched up to ∼600 °C, above which compounds such as CaCO3 begin to decompose.[6] Biochars derived from woody material usually have low ash content (between 0.9 and 4.4 wt % ash), but high ash content of up to 70 wt % has been reported in biochars from cow manure.[6−8]

It has been demonstrated that considering ash content alone is not sufficient at explaining the behavior of biochars. The concentration and element composition of inorganics in biochars influence their potential to immobilize environmental contaminants. For example, although removal of Cr(III) was related to ash content, improved removal of Cr(VI) was more specifically attributed to Fe-oxide concentrations rather than total ash.[9] In addition, phosphates and carbonates in biochar produced from manure precipitated Pb from solution.[10] In a recent study, Clemente et al. found that at acidic pH, removal of cations, such as Cd and Ni, from solution was facilitated by the presence of compounds containing Ca, Fe, Mg, P, and S, whereas Mn interfered with Cd and Ni removal.[11] The removal of Pb and Cd from solution by manure-derived biochar modified with MnO2 was shown to be pH-dependent and due to different adsorption mechanisms that included electrostatic interactions, binding to −O and −OH groups, and ion exchange.[12] Furthermore, a recent study by Limwikran et al. demonstrated variations in the water solubility and soil diffusivity of K and P, which the authors attributed to variations in the crystallinity of K- and P-bearing phases.[13] Therefore, it appears that mineral forms and not just element concentrations of biochar-associated inorganics should also be considered to fully appreciate the impact of these materials in the receiving environment. As a step toward this understanding, determining variations in the chemical and mineralogical compositions of a range of biochars is needed.

Biochars are often characterized by measuring element concentrations to determine their potential as a nutrient or contaminant source.[6−8,14,15,13] Understanding element behavior is improved by identifying specific minerals and their abundance.[5] To this end, powder X-ray diffraction (XRD) is used to identify minerals associated with pyrolysis residues.[6,7,10,13,16] Combining chemical and mineralogical data has allowed investigators to describe nutrient release and interactions between biochars and environmental contaminants.[7,10,13] Scanning electron microscopy-energy-dispersive X-ray spectroscopy (SEM-EDX) provides spatially resolved element composition of amorphous and crystalline phases and often complements chemical and XRD data. SEM-EDX is used to characterize biochar physical structure, as well as the location and element composition of associated minerals.[7,17−19,13,16] However, with SEM-EDX, it may be difficult to represent the diversity in particle composition in a sample, which makes intersample comparison challenging. One possible approach was demonstrated in a study, which subjected SEM-EDX spectra to principal component analysis (PCA) to understand the fossilization of algae.[20] Because PCA can be used to unearth latent patterns and reveal the main features of large data sets, the PCA of EDX spectra has the potential to uncover the main chemical features of inorganic particles in biochars.

The main objective of this study was to evaluate conventional chemical or mineralogical information to determine the extent of their utility at characterizing the biochar inorganic phase. PCA of bulk chemical and XRD data can differentiate biochars according to their main chemical or mineralogical features, which can highlight similarities and differences between the materials. In this study, PCA of EDX spectra will be used to differentiate inorganic particles within the same biochar. It will also be used to determine whether element distribution at the micron scale is similar across different biochars. Although this approach may not identify specific mineral phases, it may reveal the distribution and frequency of association between elements, complement chemical and XRD data, and may later be linked to differences in element availability and contaminant interactions observed by others. The information obtained using these three methods will be compared to evaluate their strengths and weaknesses and to demonstrate that multiple methods are needed to fully understand the nature of the inorganic phase in biochars. This would in turn help orient their application in agriculture and land reclamation. Biochars produced at large scales were chosen because these represent products encountered in real-world scenarios. It is known that feedstock, pyrolysis, and post-production conditions influence biochar properties.[2,8] Here, 13 biochars produced from different feedstock under different large-scale pyrolysis conditions were analyzed using chemical methods and XRD. Of the 13 samples, 8 were produced using 4 feedstock (basswood sawdust, willow, switchgrass, and poultry litter) that were pyrolyzed using the same method operated at two different temperatures; 10 biochars and 1 wood ash were further characterized by SEM-EDX. A second objective was to determine whether inorganics in biochars have a distinct signature compared to combustion ash because the latter are known to contain high ash concentrations, and there have been extensive studies on combustion ash.[5,21−23] The biochars were compared to two wood ash from co-generation power plants.

Results and Discussion

Feedstock and Biochar Production

Descriptions of the feedstock and methods used to produce the biochars and wood ash are summarized in Table . The method used to produce the Pyrovac biochar is described in detail by Roy et al.,[24] and the pyrolysis units used to produce the BassWd, Willow, SGrass, and Poultry litter biochars are described in detail by Fransham.[25] The BassWd, Willow, SGrass, and Poultry litter biochars were produced from four feedstock pyrolyzed using the same method adjusted to two different temperatures. The bulk chemical compositions of feedstock used to produce Airex 2, Pyrovac, Willow 450, Willow 500, SGrass 450, SGrass 500, Poultry litter 450, and Poultry litter 500 are summarized in Table . The feedstock composed of poultry manure and wood chips used to produce Poultry litter biochars (Table ) contained higher concentrations of Ca, K, Mg, N, Na, and P, as well as lower total OC levels compared to plant-derived feedstock (Table ).

Table 1

Summary of Feedstock and Conditions Used To Produce Biochars and Wood Ash

		Feedstock	Pyrolysis and combustion conditions
	Softwood
1	Airex 1	softwood sawdust	fast pyrolysis at 370–425 °C, cooled by water atomization
2	Airex 2	spruce sawdust	fast pyrolysis at 425 °C, cooled by water atomization
3	Pyrovac	softwood bark	fast pyrolysis at 475 °C, 20 kPa reactor pressure[24]
	Hardwood
4	BassWd 400	basswood sawdust	fast pyrolysis at 400 °C, steel shot heat carrier, <2 s reaction time, <2 s gas residence time, 10 min char residence time[25]
5	BassWd 450	basswood sawdust	fast pyrolysis at 450 °C, otherwise similar to BassWd 400
6	Willow 450	willow biomass	fast pyrolysis at 450 °C, otherwise similar to BassWd 400
7	Willow 500	willow biomass	fast pyrolysis at 500 °C, otherwise similar to BassWd 400
8	BasquesC	maple trunks and branchesa	slow pyrolysis at 500–700 °C, low O2 atmosphere, traditional concrete oven, product sieved to <0.64 cm
9	BasquesF	maple trunks and branchesa	similar to BasquesC, product sieved to <0.16 cm
	Other
10	SGrass 450	switchgrass biomass	fast pyrolysis at 450 °C, otherwise similar to BassWd 400
11	SGrass 500	switchgrass biomass	fast pyrolysis at 500 °C, otherwise similar to BassWd 400
12	Poultry litter 450	poultry litter (manure and wood chips)	fast pyrolysis at 450 °C, otherwise similar to BassWd 400
13	Poultry litter 500	poultry litter (manure and wood chips)	fast pyrolysis at 500 °C, otherwise similar to BassWd 400
	Wood ash
14	WdAsh A	unknown	co-generation power plant
15	WdAsh B	unknown	co-generation power plant

These do not meet lumber specifications.

Table 2

Concentrations of Selected Elements, Total Inorganic C (IC), and Total Organic C (OC) in the Feedstock Used To Produce Eight Biochars, with the Standard Error (S.E.) of Duplicates

	mg kg–1 dry weight (S.E.)									% dry weight (S.E.)
	Al	Ca	Fe	K	Mg	N	Na	P	Si	total IC	total OC
Airex 2	933.0 (75.0)	10 402.6 (153.2)	1500.4 (63.4)	910.8 (40.2)	818.4 (15.3)	8100	651 (15)	<150	998 (53)	0.1 (0.00)	45.8 (0.5)
Pyrovac	183.9 (4.3)	8472.2 (0.0)	330.0 (11.9)	1357.9 (16.2)	576.2 (10.3)	2700	71 (a)	277.0 (14.1)	<720	0.2 (a)	51.1 (0.3)
Willow	60.6 (9.7)	4836.6 (86.6)	110.9 (4.9)	2288.5 (135.3)	498.3 (1.6)	5700	<59	570.2 (2.2)	<720	bdl	48.8 (0.8)
SGrass	161.1 (4.9)	3566.0 (0.0)	301.7 (6.5)	1026.2 (19.9)	934.6 (5.9)	4800	<59	444.4 (2.7)	<720	0.8 (a)	47.5 (0.5)
Poultry litter	306.2 (2.1)	24 627.9 (428.3)	1033.8 (0.5)	19 541.7 (53.5)	5921.4 (21.4)	32 600	4556 (80)	12 795.8 (53.5)	<720	0.2 (0.05)	43.0 (0.5)

Duplicate analysis was below detection limit; bdl = below detection limit.

Then class="Chemical">se do not meet lumn class="Chemical">ber specifications.

Dupln class="Chemical">icate ann class="Chemical">alysis was below detection limit; bdl = below detection limit.

General Properties of Biochars

Organic Phase

Figure compares the levels of Ca, Fe, K, Mg, N, Na, P, and total OC in feedstock and their corresponding eight biochar products. There was some evidence of OC loss from poultry litter during pyrolysis since the total OC of Poultry litter biochars did not significantly differ from its feedstock, unlike the total OC enrichment observed with biochars produced from wood and switchgrass (Figure ). Furthermore, the total OC content of Poultry litter biochar produced at 500 °C was lower than the total OC in Poultry litter biochar produced at 450 °C, in contrast to the increased total OC with temperature observed in Airex, BassWd, and SGrass biochars (Table ). Similarly, Singh et al. observed increased total OC with pyrolysis temperature for biochars produced using wood and leaves, whereas total OC decreased with increasing pyrolysis temperature for biochars produced using poultry litter and cow manure.[6] Poultry litter and manure feedstock contain high protein concentrations, which decompose at lower temperatures than cellulose and lignin in plant material, with consequent partial volatilization of C.[5,17]

Figure 1

Relative changes in elemental content and total organic C (OC) from feedstock to biochar with standard error of the mean (n = 2). “Poultry” refers to biochar produced from poultry litter. Values >0 indicate enrichment, whereas values <0 indicate depletion of the element. Two-tailed t-tests on log10 transformed data were performed to identify significant changes in element concentrations by comparing feedstock and biochar concentrations at p < 0.05 (**) and p < 0.1 (*). The N measurements had <1% difference between three and four replicates, and the average was reported by the contracted lab; t-tests were not performed for these data because the value of replicates were unknown. Enrichment of selected elements and total OC in selected biochars was calculated using the following equation: (concentration in biochar – concentration in feedstock)/concentration in feedstock. Fifty percent of the detection limit was used when the concentration of an element in the feedstock was below detection limit.

Table 3

pH; Electrical Conductivity (EC); N, P, K; Total Organic C (OC), and Total Inorganic C (IC) of Biochars and Wood Ash, with Standard Error (S.E.) of Duplicates

			dS m–1	mg kg–1 dry weight (S.E.)			% dry weight (S.E.)
		pH	EC	N	P	K	total OC	total IC
1	Airex 1	5.8	0.14	2000	<260	668.9 (32.4)	60.0 (0.1)	0.2 (0.06)
2	Airex 2	7.9	0.10	2200	92.0 (a)	1465.1 (5.3)	79.6 (0.1)	bdl
3	Pyrovac	8.7	0.29	6500	853.9 (53.3)	2858.6 (162.0)	76.6 (1.2)	0.7 (0.08)
4	BassWd 400	6.5	0.16	4400	543.7 (30.6)	3650.7 (36.4)	72.4 (0.5)	0.3 (0.01)
5	BassWd 450	8.4	0.14	5200	690.6 (30.1)	5005.6 (0.0)	74.7 (0.3)	0.4 (0.01)
6	Willow 450	9.1	0.65	7300	3262.7 (79.3)	10 062.9 (312.0)	66.6 (0.4)	1.5 (0.01)
7	Willow 500	9.4	0.47	7500	3566.9 (110.6)	11 433.1 (158.1)	65.7 (0.4)	1.8 (0.04)
8	BasquesC	9.4	0.74	8300	1067.4 (269.4)	8218.0 (1751.4)	63.9 (2.9)	0.9 (0.08)
9	BasquesF	9.4	0.81	7800	1311.4 (10.3)	9546.7 (191.0)	58.4 (1.2)	1.1 (0.01)
10	SGrass 450	9.4	0.46	6900	2300.7 (31.7)	4875.7 (63.3)	68.5 (2.4)	1.0 (0.12)
11	SGrass 500	10.0	0.55	7800	2997.4 (26.5)	6657.8 (111.4)	70.0 (2.0)	1.0 (0.04)
12	Poultry litter 450	10.6	5.66	22 400	43 802.6 (634.8)	66 391.6 (899.3)	53.0 (0.1)	1.6 (0.06)
13	Poultry litter 500	10.5	6.01	22 400	44 046.1 (52.9)	66 412.9 (423.0)	46.0 (1.6)	2.3 (0.26)
14	WdAsh A	10.0	0.85	1900	1257.5 (98.3)	1213.0 (0.0)	10.2 (0.7)	1.5 (0.01)
15	WdAsh B	10.8	0.58	2300	1355.3 (58.5)	8094.8 (207.3)	17.0 (0.5)	0.8 (0.06)

Duplicate was below detection limit; bdl = below detection limit.

Dupln class="Chemical">icate wn class="Chemical">as below detection limit; bdl = below detection limit.

Biochar total OC ranged from 46.0 to 79.6 wt % and were greater than the WdAsh total OC of 10.2–17.0 wt % (Table ). These values are in line with previous studies that report biochar total C yields of 55.0–94.0% in plant-derived biochars and 16.5–74.0% in manure-derived biochars.[2,6,26] Low total OC (<50 wt %) in residual ash from power-generating plants (e.g., WdAsh) is also expected.[22]

The proximate analysis results are illustrated in Figure . Volatile matter concentration was highest in Airex 1, which was also pyrolyzed at lower temperatures compared to other biochars (Table ), whereas Airex 2 and Pyrovac had the highest fixed matter content. Airex 1 and BassWd 400, which were pyrolyzed at lower temperatures, had slightly acidic pH and significantly more volatile matter compared to Airex 2 and BassWd 450 (Table and Figure ). This trend was similar to those reported by Singh et al. in that biochars produced at 400 °C were generally more acidic than those produced at 550 °C.[6] The acidity can be attributed to acidic organic compounds that remain after pyrolysis at lower temperatures, such as anhydrides, esters, and carboxylic acids. These compounds are more abundant in the parent material, and their concentrations are hypothesized to decrease with increased pyrolysis temperatures.[2]

Figure 2

Ash, fixed, and volatile matter in biochars and WdAsh determined using ASTM D1762-84. “Poultry” refers to biochar produced from poultry litter. Error bars show ±standard error of the mean (n = 2). Different letters or numbers denote significant difference between samples (Tukey’s honestly significant difference (HSD) α = 0.05).

Ash and Electrical Conductivity

From proximate analysis (Figure ), ash content was significantly higher in wood ash (WdAsh A and WdAsh B), resulting in significantly lower fixed and volatile matter content compared to biochars. Among biochars, those derived from poultry litter (Poultry 450 and Poultry 500) had the highest ash content, followed by switchgrass-derived biochars (SGrass 450 and SGrass 500). Compared to WdAsh, Poultry litter biochar had 5–34% less ash, but 7–10 times greater electrical conductivity (EC) values (Table and Figure ). EC is related to the amount of readily soluble ions and therefore, compared to WdAsh, a much greater proportion of the inorganic material in Poultry litter biochar was likely soluble. In this study, EC was influenced by feedstock more than pyrolysis temperature. The high EC values for Poultry litter are typical of biochars derived from manure (1.65–2.77 dS m–1).[4,26]

Potential as Nutrient Source

From Table , Poultry litter biochars had much higher concentrations of N, P, and K than WdAsh, with estimated fertilizing N/P2O5/K2O values of 2.2:10.0:8.0. WdAsh A had the second highest value with 0.2:0.3:1.5, followed by Willow 500 with 0.9:0.8:1.4. The higher fertilizing value of poultry litter compared to wood-derived biochars is typical and consistent with studies reporting improved wheat and corn growth after addition of poultry litter-derived biochar to soils.[6,27] Except for Poultry litter biochars and WdAsh A, total OC/N ratios of the samples were >60 (Table S1). At these high ratios, N can be immobilized by microorganisms, which limits its availability to crops.[28] However, it should be noted that more stable OC structures are formed during pyrolysis and that the proportion of bioavailable OC and N will determine the availability of N to crops.[29] As shown in Figure , volatile matter concentrations varied from 19 to 58 wt % in biochars and 13 to 16 wt % in WdAsh. Volatile matter content represents easily degradable OC and is positively correlated to biochar biodegradability in soils.[30,31] For example, high volatile matter concentrations of 22.5 wt % stimulated microbial respiration and led to decreased N, P, K uptake and plant growth for up to 5 weeks after biochar addition.[30] Of the materials evaluated in this study, Airex 1, Airex 2, BassWd 400, BassWd 450, Willow 450, and SGrass 450 had volatile matter content >23 wt % and may limit N availability to crops. Biochars could also become potential sources of other elements (Table S2) at levels that may not always be desired, as discussed in Section 2, Elements of Interest, in the Supporting Information.

Bulk and Microscale Inorganic Composition

Differentiation Based on Chemical Analysis

In this study, samples were digested with aqua regia to determine the pseudo-total concentrations of elements in relatively accessible domains. From Table and Figure , it appears that although poultry litter-derived biochars had higher inorganic content compared to the plant-derived biochars, there was relatively greater enrichment of Ca, N, Fe, Na, K, P, and Mg during pyrolysis of plant compared to poultry litter feedstock. Similar poor inorganic enrichment was observed in Airex 2 (Figure ). Although the Airex 2 feedstock contained higher Al, Fe, Mg, N, Na, and Si concentrations than other plant-derived feedstock (Table ), Airex biochars had significantly lower ash and inorganic content compared to other plant-derived biochars (Figures and 3). Loss of H2O-soluble minerals may have occurred when Airex biochars were cooled by water atomization, and this mode of inorganic loss requires further study.

Figure 3

(a) Average (n = 2, with error bars showing standard error of the mean) pseudo-total element concentrations (mmol kg–1) in biochars and WdAsh (∑[Al, As, Ba, Ca, Cd, Cr, Cu, Fe, K, Mg, Mn, Mo, N, Na, Ni, P, Pb, S, Sb, Se, Si, Sr, Ti, Zn, total IC]) and (b) mol % composition of the pseudo-total element concentrations of biochars and WdAsh. The 11 elements that make up 89–93% of total inorganics in each sample are listed and the sum of the remaining elements is included as “Others”. “Poultry” refers to biochar produced from poultry litter.

(a) Average (n = 2, with error bars showinpan>g standard error of the mean) pseudo-total element concentrations (mmol kg–1) in biochars and WdAsh (∑[Al, As, Ba, Ca, Cd, Cr, Cu, Fe, K, Mg, Mn, Mo, N, Na, Ni, P, Pb, S, Sb, Se, Si, Sr, Ti, Zn, total IC]) and (b) mol % composition of the pseudo-total element concentrations of biochars and WdAsh. The 11 elements that make up 89–93% of total inorganics in each sample are listed and the sum of the remaining elements is included as “Others”. “Poultry” refers to biochar produced from poultry litter.

The pseudo-total concentration of inorganics and their composition are shown in Figure . There were similarities in element composition among samples, and 19 elements made up >95 mol % of total inorganics in the biochars and WdAsh. N, P, K, Ca, Mg, and S, which are the main nutrients in plants, along with total inorganic C, Al, Fe, Na, and Si were proportionally the most abundant inorganic elements detected in the biochars and WdAsh (Figure ). Nitrogen, total inorganic C, and Ca made up 48–76 mol % of the total inorganics in the biochars and WdAsh (Figure ), similar to the study of Singh et al.,[6] which identified N and CaCO3 as the main inorganics in biochars. WdAsh were combusted at higher temperatures (and in an oxygenated atmosphere) than biochar, and the lower proportion of N in WdAsh is consistent with other studies that show an exponential increase in volatilization of N compounds between 500 and 800 °C.[19,32] Poultry litter had higher N concentrations than vegetation-type feedstock (Table ), and a significant amount of this N is known to be in proteins that decompose at >320 °C.[17] Although N concentrations in poultry litter-derived biochars were higher than those in plant-derived biochars (Table ), comparison between feedstock and pyrolysis products (Figure ) suggests that some N may have been lost from poultry litter during pyrolysis through partial volatilization. Pyrolysis products identified by others include NH3(g), HCN(g), N2, NO, and HNCO, and the mechanism responsible for N loss may be similar to what occurs during sewage sludge pyrolysis discussed by Wei et al.[32] Na and K were also among the most abundant inorganic elements in Poultry litter biochar (Figure ), which contained 3–362 times more Na and 5–99 times more K than the other samples (Table ). Poultry litter-derived biochars may contribute to soil salinity if used as an amendment, with saline soils defined as having EC > 4 dS m–1.[33]

Convenpan>tional chemical data were subjected to PCA to distinguish biochars based on bulk inorganic element composition (Figure ). This multivariate technique summarizes complex data sets by combining correlated variables into a new vector, thereby revealing relationships between samples or between variables.[34] In PCA, the amount of variation explained by the uncorrelated principal components (PC)s decreases in the order of PC1 > PC2 > PC3. With this method, vector and loadings plots display relationships between variables and PCs. In this study, the vector and loadings plots were used to visualize how chemical, diffraction, and spectral information were related to the PCs in terms of magnitude and direction. From chemical analysis of the biochars and wood ash, the first two PCs accounted for 74.0% of the sample variance. That is, they explain 74% of the differences in the bulk chemical composition among the 15 samples. Airex 1 and 2, which had the lowest total element concentrations, were plotted at negative PC1 and PC2 (Figures and 4). The samples with high Si and Al concentrations (WdAsh and Basques) occupied the positive PC1 and negative PC2 regions, whereas the samples with high Fe and P (SGrass and Poultry litter) occupied the positive PC1 and PC2 regions. The samples at negative PC1 and positive PC2 regions contained similar pseudo-total inorganic concentrations as those plotted at positive PC1 (Figures and 4): Willow and WdAsh A; BasquesC, BasquesF, SGrass, and WdAsh B. This was because in addition to pseudo-total element concentrations, concentrations of specific elements also played a role in differentiating the samples. Biochars and WdAsh were differentiated based on inorganic content, followed by composition. Biochars from the same feedstock and fast-pyrolyzed using steel shots with a difference of 50 °C (BassWd, Willow, SGrass, Poultry litter) had similar bulk inorganic composition despite the 50 °C difference in pyrolysis temperature (Tables , S1, and S2 and Figures and 4). These pairs of biochars were not differentiated from each other by PCA of traditional chemical data. That is, feedstock had a greater influence in differentiating between these biochar pairs.

Figure 4

Scores plots and vectors (length represents relative influence) of element concentrations, considering the 19 most abundant elements, in biochars and WdAsh. Total IC refers to total inorganic C. Data points correspond to the following: 1 = Airex 1; 2 = Airex 2; 3 = Pyrovac; 4 = BassWd 400; 5 = BassWd 450; 6 = Willow 450; 7 = Willow 500; 8 = BasquesC; 9 = BasquesF; 10 = SGrass 400; 11 = SGrass 500; 12 = Poultry litter 450; 13 = Poultry litter 500; 14 = WdAsh A; 15 = WdAsh B. For samples digested using two different methods, A are samples acid-digested using aqua regia without hydrofluoric acid (HF) and B are samples acid-digested with HF, as published by Beauchemin et al.[7] Loadings plots are shown in the Supporting Information (Figure S1).

For Airex 1, BassWd 400, BassWd 450, WdAsh A, and WdAsh B, the pseudo-total (aqua regia) concentrations in this study were compared to those published by Beauchemin et al.,[7] which were prepared with an additional HF acid digestion step (Figure ). Digestion with HF acid resulted in (to different extents) higher inorganic concentrations, especially for samples richer in quartz or feldspar (i.e., WdAsh; Figure ). The extent to which digestion method influenced the observed inorganic concentrations was reflected in the level of separation along PC1 and was greatest for WdAsh A.

Figure 5

X-ray powder diffraction patterns of 13 biochars and 2 WdAsh. Vertical lines correspond to regions attributed to broad peaks from cellulose and turbostratic crystalline C (tcc). Mineral phases are labeled based on database matches. “Poultry” refers to biochar produced from poultry litter.

Differentiation Based on X-ray Diffraction

XRD patterns have beenpan> extenpan>sively used to determine the level of pyrolysis and to characterize the inorganic composition of biochars.[2,6,7,10,13] In this study, semicrystalline minerals were not identified in the diffractograms of Airex 1 and Airex 2 (Figure ), likely because of severe dilution by the organic phase, which is consistent with their low inorganic content (Figure a). Quartz, calcite, and magnetic iron oxide (e.g., magnetite) were common features in the remaining samples (Figure ). Contributions from weak but distinct signals attributed to magnetic iron oxide were detected in BassWd biochars, despite their relatively low inorganic content (Figures a and 5). As expected, intense signals attributed to minerals were observed in WdAsh and biochars that also contained higher inorganic content (Poultry litter, Willow, SGrass).

The XRD patterns of Pyrovac, Basques, Poultry litter, SGrass biochars, and WdAsh were complex, containing broad regions of semicrystalline C and overlapping peaks (Figure ). These features made it difficult to assign minor peaks to specific minerals, particularly in poultry litter-derived biochars, where several different types of silicates were tentatively identified (labeled as various silicates in Figure ). HyperSpec[35] processing coupled to PCA of the diffractograms allowed for comparison of the full diffraction fingerprints. Comparisons therefore did not necessitate exhaustive peak identification and assignment. PCA was conducted to determine the main features responsible for differences between XRD patterns. Although diffractogram fingerprints were considered, the PCs were still found to be related to specific signals that can be attributed to known features in pyrolysis products (cellulose, turbostratic crystalline carbon) and minerals (e.g., calcite, magnetic Fe-oxide, quartz, whewellite), as indicated by the vector and loadings plots (Figures and S2).

Figure 6

Scores and vector plots (length reflects relative influence) of X-ray diffraction patterns of biochars and WdAsh for the first three principal components (PCs) with % of explained variance in parenthesis: (a) PC1 vs PC2 and (b) PC3 vs PC2. Data points correspond to the following: 1 = Airex 1; 2 = Airex 2; 3 = Pyrovac; 4 = BassWd 400; 5 = BassWd 450; 6 = Willow 450; 7 = Willow 500; 8 = BasquesC; 9 = BasquesF; 10 = SGrass 450; 11 = SGrass 500; 12 = Poultry litter 450; 13 = Poultry litter 500; 14 = WdAsh A; 15 = WdAsh B; tcc = turbostratic crystalline carbon. Loadings plots are shown in the Supporting Information (Figure S2).

The first three PCs captured 95.3% of the variance in the XRD patterns (Figure ). According to the vector and loadings plots, PC1 mainly corresponded to contributions from turbostratic crystalline C (Figures a and S2). PC1 separated Airex and BassWd from other samples in the PC1 versus PC2 scores plot (Figures and S2). Distinction of Airex and BassWd from the other materials was consistent with their low ash and inorganic concentrations (Table , Figures , 3, and S2). PC2 and PC3 were related to contributions from minerals, mainly quartz, calcite, sylvite, and magnetite (Figures b and S2). The PC3 versus PC2 scores plot distinguished both SGrass from other biochars because of intense signals from magnetic Fe oxide (e.g., magnetite) in their diffractograms (Figures , 6b, and S2). Quartz and calcite were major mineral phases in WdAsh and were observed in biochars to varying degrees; thus, WdAsh samples were not separated from biochars (Figures and 6). Poultry litter biochars were differentiated from other samples in both XRD scores plots because their diffractograms contained intense signals from inorganics, particularly sylvite, and no significant signal from quartz was identified (Figures and 6). Previous studies have also reported that properties of manure-based biochars differed from plant-based biochars.[6,17] That Airex and BassWd biochars were separated from the other samples on the basis of no or low signals from minerals, WdAsh was grouped with biochars containing quartz and calcite, and Poultry litter (and to some extent SGrass) were distinguished from other samples were reminiscent of groups observed after PCA of chemical data (Figures –6).

Distinctions between biochars based on XRD data were not always similar to distinctions based on chemical data. According to chemical analysis, Airex 1 was closely related to other biochars, which was not the case according to XRD (Figures and 6). Airex 1 was pyrolyzed at lower temperatures than other biochars (Table ) and contained residual cellulose, which resulted in an XRD pattern characterized by two broad peaks with maxima at 2θ = 16 and 22° (Table and Figure ). These features were captured by PC2, and Airex 1 was separated from other samples in both PC1 versus PC2 and PC3 versus PC2 scores plots (Figures and S2). The scores plot of XRD patterns were sensitive to the semicrystalline C regions (cellulose and turbostratic crystalline carbon). In addition, although chemical analysis did not distinguish biochar pairs produced using the same feedstock and the same pyrolysis method with a difference of 50 °C, Willow 450 was distinguished from Willow 500 because the former contained Ca oxalate (whewellite) and more intense crystalline C signals (Figures and 6). Singh et al. also reported the transformation of Ca oxalate to calcite when pyrolysis temperature of eucalyptus char was increased from 400 to 550 °C.[6] Using SEM-EDX and XRD, Rees et al. also demonstrated that Ca oxalate crystals in young poplar bark were transformed to calcite.[16] The presence of different Ca-containing minerals may affect the ability of biochars to neutralize acidic soils. With pKa values of 1.27 and 4.28, oxalate may be less effective at neutralizing acidic soils compared to carbonate, which has a pKa of 6.4 and 10.3.[36] Although there were agreements on how biochars were sorted using chemical and XRD data, PCA of XRD patterns provided additional information by differentiating biochars using properties that were not apparent from chemical analysis.

Microscale Frequency and Distribution of Elements

PCA of EDX spectra was conducted to fill in the gaps between chemical and XRD analyses. For example, Poultry litter biochars contained ∼44 g kg–1 P (Table ), yet a P-containing mineral was not identified by XRD (Figure ). This inconsistency is likely because XRD is not sensitive to amorphous phases. Associations between elements at the micrometer scale that were identified through EDX point scans may occur through reaction, crystallization, co-precipitation, sorption, or co-localization of inorganics. PCA of the EDX spectra (representative spectra in Figure ) identified common associations between elements. The vector plot in Figure and the loadings plots for PC2, PC3, and PC5 (Figure S3) illustrate how each element influenced the PCs in terms of magnitude and direction. It should be noted that although the association between different elements in an EDX spectrum may be consistent with the presence of minerals identified by XRD, the application of PCA to EDX spectra does not necessarily identify specific mineral phases, and the signal from an element must be sufficiently intense to influence the score of a particular spectrum. N-containing inorganics were not identified in the EDX spectra (Figure ). It is possible that N did not form particles that were distinct from the biochar organic phase. Further studies are needed to determine how N is associated with biochars.

Figure 7

Scanning electron microscopy with backscattered electron images of representative biochars and WdAsh A and selected energy-dispersive X-ray spectra corresponding to labeled point scans for (a) Airex 2, (b) Pyrovac, (c) BassWd 450, (d) Willow 500, (e) BasquesC, (f) SGrass 450, (g) Poultry 450 (produced from poultry litter), biochars, and (h) WdAsh A.

Figure 8

Principal component analysis of scanning electron microscopy-energy-dispersive X-ray spectra of 10 biochars and WdAsh A. Scores and vector (length represents relative influence) plots are for (a) PC1 vs PC2 and (b) PC3 vs PC2. (c) EDX spectra were filtered using PC3 to plot data that did not contain signals from Cl, K, Mg, Na, and P at PC5 = 0, and the resulting filtered PC5 vs PC2 scores plot illustrates the distribution of Cl, K, Mg, P, and S in the samples; the positions for Si, Al, C, and Fe along PC2 are also illustrated. Loadings plots represented by the principal components are shown in Figure S3 (Supporting Information), and scores plots for individual samples are plotted in Figure S4 (Supporting Information) for clarity. Particles also consisted of mixtures of these elements as indicated by the double-headed pink arrows in (a). “Poultry” refers to biochar produced from poultry litter.

Scanninpan>g electron microscopy with backscattered electron images of representative biochars and WdAsh A and selected energy-dispersive X-ray spectra corresponding to labeled point scans for (a) Airex 2, (b) Pyrovac, (c) BassWd 450, (d) Willow 500, (e) BasquesC, (f) SGrass 450, (g) Poultry 450 (produced from poultry litter), biochars, and (h) WdAsh A.

EDX Spectra Pattern Distribution

The first three PCs accounted for 77.2% of the total variance in EDX spectra (Figure ). PC1 and PC2 differentiated the spectra into four main groups based on contributions from the following elements: (i) Ca, (ii) Fe, (iii) [Al, Si], and (iv) [Cl, K, Mg, Na, P, S] (Figure a). These elements are also components of mineral phases identified using XRD: Ca in calcite; Fe in magnetic Fe oxide; Si in quartz; K, Na, Ca, Al, and Si in feldspar; and K and Cl in sylvite (Figure ). Spectra with signals from Ca were obtained from all biochars, consistent with the ubiquity and abundance of Ca and calcite in the samples (Figures , 5, 7, and 8). PCA of EDX spectra generated a discrete Fe-rich cluster because of high Fe concentrations in some samples. For biochars produced with steel pellets as heat carriers (i.e., BassWd 400, 450; Willow 450, 500; SGrass 450; and Poultry litter 450), elevated Fe was likely due to contamination during pyrolysis. Steel from pyrolysis equipment and materials (e.g., steel pellets) are known to be a possible source of contaminants during biochar production.[37,38]

Particles in the region between the four main groups contained mixtures of Ca, Fe, [Al, Si], and [Cl, K, Mg, Na, P, S] (Figure a). Mixtures of Ca and [Al, Si] were related to minerals in the plagioclase feldspar group. Ca and [Al, Si] mixtures were more common in Basques, WdAsh, and Pyrovac, which was consistent with the presence of feldspar in these samples (Figures , 7, 8, and S4). Furthermore, PCA of the EDX spectra places the inorganic particles in Pyrovac, BasquesC, and WdAsh A along a similar trajectory (Figures and S4). These materials were abundant in calcite, quartz, and plagioclase feldspar (Table , Figures , and 8). Ca silicates are known to form during combustion, and Basques, Pyrovac, and WdAsh were produced at higher temperatures compared to the other biochars (Table ). In addition, soil contamination[5] and concrete in the traditional oven used to produce Basques biochar may be additional sources of these minerals.

Because PCA of EDX spectra targeted inorganic particles and excluded C, dilution and influence of the organic phase were avoided unlike chemical and XRD analyses. SEM-EDX allowed for analysis of elements in biochars with low inorganic concentrations: Airex 1 and Airex 2. Unfortunately, the PCA of EDX spectra also showed overlapping element distributions for Willow 450 and 500 (Figures and S4). Although EDX analysis avoids the dilution effect expected in bulk chemical analysis, both methods similarly classified these biochars in the same group (Figures and 8). Unlike XRD, EDX did not distinguish Ca in calcite from that in whewellite, and failed to differentiate Willow 450 from Willow 500 biochar (Figures and 6). In Poultry litter biochars, sufficiently intense EDX signals from [Al, Si] were rarely observed, although silicates were thought to be among the main signals in their diffractograms (Figures , 7, and 8). The EDX analysis was consistent with chemical information because although Poultry litter biochars contained between 1534 and 1851 mg kg–1 Si (Table S1), it was not among the 11 most abundant elements in these biochars (Figure ). The results suggest that amorphous nonsilicate minerals that are not amenable to XRD analysis were abundant in Poultry litter biochars (Figures and S4).

Elements in Amorphous Phases

PCA of EDX spectra can help further characterize elements of interest in amorphous phases. P and K are of interest as nutrients and P can precipitate cations from solution or compete with oxyanions for sorption sites.[11,14] The filtered PC5 versus PC2 scores plot was generated to clarify the distribution of [Cl, K, Mg, P, S]. From the loadings plot (Figure S3), particles that contain [Cl, K, Mg, P, S] have positive PC3 values and were simultaneously correlated to PC5. Therefore, a filtered PC5 was generated to plot spectra without [Cl, K, Mg, P, S] along PC5 = 0 (i.e., PC3 ≤ 0) (Figure c). Because P was negatively correlated to PC5, it was also separated from [Cl, K, Mg, S] (Figures c and S3). This mathematical treatment was useful for surveying P (and to some extent K), particularly in Poultry litter and Willow biochars.

For Poultry litter biochars, spectra with signals from [Cl, K, Mg, S] also contained signals from Fe (Figures g and 8c). Fe is known to form sulfates, carbonates, and oxides with K and Mg, although concentrations of these compounds in combustion products are typically <0.1%.[5] Coating of Fe particles by water-soluble elements can also explain these associations. Biomass dehydration is known to occur during the initial stages of pyrolysis[2] and may promote coating of Fe particles by water-soluble elements, including sylvite. This hypothesis is consistent with the intensity of signals from K, Cl, and Fe in Poultry litter biochar EDX spectra (Figure ). It also agrees with a previous study, which demonstrated that water-soluble Cl, K, Pb, S, and Zn coated more stable silicates in ash from municipal waste combustion.[39] In Pyrovac, BasquesC, and WdAsh A, association of [Cl, K, Mg, S] and [Al, Si] and/or Ca likely represents feldspar (Figures , 7, 8, and S4). In contrast, distinct K-containing phases were not identified in Willow XRD patterns although K concentrations in Willow (10.0–11.4 g kg–1) were comparable to BasquesC (8.2 g kg–1) (Table and Figure ). The K signal in Willow biochars were localized and indicates association with Ca (Figures , 8c, and S4). Similar associations of [Cl, K, Mg, S] were observed in Poultry litter biochar (Figures c and S4). From Figure c, the distribution of P was constrained and was often associated with Ca. The likely formation of Ca phosphate in Willow is supported by the EDX spectrum in Figure d. Ca phosphate may also be present in other biochars (including Poultry litter) based on data distribution in the scores plots (Figures and S4). Ca, K, Mg, and their mixtures have been known to form phosphates and carbonates during combustion.[5] In addition, both K and P can co-precipitate with CaCO3 and P can sorb to calcite without necessarily forming a distinct mineral phase.[40−42]

PCA of EDX spectra revealed that associations between P or [Cl, K, Mg, S] and Ca, Fe, or [Al, Si] differed between samples (Figure ). These associations may have implications on sample-specific element availability and acid-neutralizing ability. For example, sylvite in Poultry litter biochar is water-soluble, which along with K associated with Ca carbonates may be more readily available compared to K in feldspar from WdAsh. In acidic soils, silicates (e.g., albite and anorthite) can be less reactive than carbonates and phosphates (e.g., calcite and hydroxyapatite). Literature values estimate the dissolution rate of albite to be 6 orders of magnitude lower than that of calcite at neutral pH.[43,44]

Inorganic Composition of Biochars and Wood Ash

A total of 13 biochars were analyzed and compared to 2 wood ash samples. The properties of these materials were typical of those from the literature, considering their feedstock and pyrolysis temperature. The known strengths and limitations of the three methods used in this study are summarized in Table . PCA of bulk chemical data separated the materials based on inorganic content and concentrations of specific elements. This method may not be sufficient at predicting element availability because it does not provide information on element form, and complementary analyses are required. The intensity of cellulose and turbostratic crystalline C relative to the intensity of minerals influenced the classification of biochars by PCA of XRD data. PCA of chemical, XRD, and EDX data suggest similarities between WdAsh and Basques biochars because silicates and carbonates were common in these samples. Among the 15 samples, WdAsh and Basques were produced at higher temperatures in the presence of abundant (WdAsh) or low (Basques) O2 concentrations. Poultry litter-derived biochar and material pyrolyzed at lower temperature resulting in high cellulose content (Airex 1) were separated from other biochars. PCA of EDX spectra discriminated four main clusters corresponding to particles rich in silicates, calcite, and magnetic Fe oxide, and nutrients including P and K. It showed similar element association in materials produced at higher temperatures: BasquesC, Pyrovac, and WdAsh. It uncovered element association in amorphous phases. It also revealed the association of P or K with [Al, Si], Ca, or Fe and the likely formation of Ca phosphates. Unfortunately, PCA of EDX spectra failed to distinguish biochars that contained minerals with the same inorganic element component, but different mineralogy (e.g., Willow 450 and Willow 500). Nevertheless, these results can help clarify the behavior of inorganics in biochars and wood ash used in water treatment, agriculture, or land reclamation. For example, nutrients (P, K, Mg, and S) associated with Ca, as observed in Poultry litter biochar, are expected to be more available than those associated with aluminosilicates [Al and Si], as observed in BasquesC (hardwood) biochar or WdAsh A because calcite would be more soluble than silicate in acidic soils. By evaluating data from three methods, it was demonstrated that multiple techniques are needed to appreciate the inorganic composition of pyrolysis products.

Table 4

Strengths and Limitations of the Three Methods Discussed in This Study

Strengths	Limitations
Bulk Chemical Analysis
widely used, available, and affordable	inorganic elements are diluted by the abundant OC
relative ease of sample preparation and analysis	detected element concentration may be influenced by digestion method
information on pseudo-total bulk element concentration	complementary method(s) required for mineralogical data
X-ray Diffraction
provides information on bulk mineral composition	not sensitive to amorphous phases
provides additional information on semicrystalline organic phase	dilution of inorganic elements by semicrystalline organic phase
relative ease of sample preparation and analysis	difficulty in assigning low-intensity and overlapping signals, require complementary method(s) to confirm assigned mineralogy
SEM-EDX (mounted loose powder)
ease of sample preparation	analysis is time-consuming
targeted element content at micron scale means element ratios from EDX more closely resemble ratios in minerals than that obtained through bulk chemical analysis	intersample comparison may be challenging in the absence of multivariate data analysis
analysis of elements in amorphous phases	sample topology may influence accuracy of detecting low-energy elements
avoids dilution effect from biochar organic phase	inability to determine mineralogy, and requires complementary approach(es) for confirmation
also provides morphological information

Using pyrolysis residue as soil amendments may be a way to recycle minerals taken up by biomass for growth back into the environment. But this requires understanding the makeup of the inorganic phase in these materials. Classification schemes are potential tools for screening biochars for specific applications. Tests under laboratory and field conditions is beyond the scope of this paper, but are needed to confirm the relationship between these classifications, environmental parameters, and availability of elements of interest.

Materials and Methods

Biochars and Wood Ash

Naminpan>g conventions used in this paper, available descriptions of feedstock, and conditions used in the production of the 13 biochars and 2 wood ash are described in Table . Biochars were acquired from the following industrial partners and commercial producers: Pyrovac (St. Lambert de Lauzon, Quebec); Airex Industries Inc. (Laval, Quebec; details for the Airex production were kept confidential); Charbon de bois franc Basques (St. Mathieu de Rioux, Quebec); and Abri-Tech Inc. (Namur, Quebec) provided the BassWd, Willow, SGrass, and Poultry litter biochars. The WdAsh were obtained from anonymous producers associated with co-generation power plants. Organic C functional groups in Airex 2, BasquesF, Poultry litter 450, Pyrovac, SGrass 450, and Willow 450 were previously characterized.[11] Airex 1, BassWd 400, BassWd 450, WdAsh A, and WdAsh B were used in a study by Beauchemin et al. on the impact of biochar on metal leaching from mine tailings;[7] Airex 2, BasquesF, Poultry litter 450, Pyrovac, SGrass 450, and Willow 450 were used in a study by Clemente et al. on metal removal from solution.[11] The feedstock for Willow, SGrass, Poultry, Airex 2, and Pyrovac were chemically characterized to evaluate differences in element composition between feedstock and the resulting biochars.

Characterization

The pH and EC of biochars and wood ash were measured after suspending 1 g of sample in 20 mL of distilled water, with shaking for 1.5 h.[45] The properties of 13 biochars, 2 wood ash, and 5 feedstock used to produce 8 of the biochars were evaluated. The following standard methods were used: ASTM D7582 for moisture content, ASTM D1762-84 for proximate analysis, and ASTM D5373 for total N. Total C was determined by combustion using an Eltra CS-2000. The total inorganic C was determined using the same method after removal of OC by combustion at 400–450 °C for 12 h, and the total OC was obtained by difference. Pseudo-total concentrations of the following elements were obtained in duplicate by microwave digestion in a mixture of HCl and HNO3 (aqua regia): Ag, Al, As, B, Ba, Be, Bi, Ca, Cd, Co, Cr, Cu, Fe, K, Li, Mg, Mn, Mo, Na, Ni, P, Pb, S, Sb, Se, Si, Sr, Te, Ti, Tl, V, Zn, and Zr. Digests were analyzed by inductively coupled plasma (ICP)-atomic emission spectroscopy or ICP-mass spectrometry. Elements detected in the digestion extract, together with total N and total inorganic C, make up the pseudo-total inorganic element concentration.

Powder XRD patterns were obtained on a Rigaku D/MAX 2500 rotating anode powder diffractometer using monochromatic Cu Kα radiation at 50 kV, 260 mA, a 0.020° step scan size, and a scan speed of 1° min–1 with 2θ between 5 and 70°. JADE v9.3 (Materials Data Inc., Livermore, CA), the Inorganic Crystal Structure Database, and the International Centre for Diffraction Data were used to identify mineral phases. Commercially available cellulose (Sigmacell type 101; Sigma Aldrich, St. Louis MO) was used to assign 2θ = 22.0, 15.8, and 34.3° to semicrystalline cellulose, and regions between 2θ = 22.7–23.9 and 43.3–43.7° were assigned to turbostratic crystalline C according to Keiluweit et al.[2] A total of 10 representative biochars (Airex 1, Airex 2, Pyrovac, BassWd 400, BassWd 450, Willow 450, Willow 500, BasquesC, SGrass 450, and Poultry litter 450) and WdAsh A were prepared for SEM-EDX by mounting loose powder on double-stick C tape attached to an Al stub. The samples were then coated with a thin layer of C. Secondary and backscattered electron imaging as well as EDX were performed at an accelerating voltage of 20 kV using a Hitachi S-3200N variable-pressure scanning electron microscope equipped with an XFlash Bruker silicon drift detector (SDD). Semiquantitative elemental composition of inorganic particles was determined by conducting point analyses with the XFlash SDD, and the resulting spectra were background-corrected using DTSA v.II.[46]

Statistical Analysis

R v.3.1.1 and the hyperSpec and factoextra packages were used for statistical analyses.[35,47] The F-test two sample for variance function in Excel indicated that the variances of the element concentrations in the parent material and biochar were unequal. Therefore, concentrations were log10-transformed prior to conducting two-tailed, independent sample t-tests to determine whether differences in element concentration between feedstock and biochars were significant. ANOVA followed by Tukey’s honestly significant difference (HSD) were used to determine whether differences between means of proximate analysis were significant (α = 0.05).

PCA of inorganic concentrations, XRD, and EDX were calculated separately. Prior to PCA of chemical data (Al, Ba, Ca, Cr, Cu, Fe, K, Mg, Mn, N, Na, Ni, P, S, Si, Sr, Ti, Zn, and total inorganic C concentrations), values below detection limit were coded as 0.5 × detection limit. All values were then log-transformed and centered to stabilize variance. The hyperSpec package was used to process X-ray diffractograms and EDX spectra, which allows comparison of diffraction and spectral fingerprints. XRD and EDX data were normalized to the highest count prior to PCA. The EDX of 606 point scans from 10 representative biochars and WdAsh A were prepared by truncating the spectra to include signals between 0.80 and 8.0 keV to avoid the overwhelming C signal, minimize the effects of topography (expected to compromise the detection of signals at <0.80 keV), and because only one particle had a signal at >8.0 keV. Therefore, the PCA of EDX spectra did not take into account C, N, and O contributions.