Demineralization of biomassNaturally-occurring potassium was successfully removed by water demineralization of the ground biomass (Table S2). Potassium decreased by more than 89%, and sodium was almost completely removed. Water demineralization was less successful with Ca and Mg, removing up to 39% and 45%, respectively.During pyrolysis of woody biomass, potassium lowers the activation energy for thermal decomposition of plant biopolymers and catalyzes condensation reactions critical to solid char formation, resulting in enhanced production of gaseous and solid products38,39,40,41,42. A reduction in naturally-occurring potassium, either by acid- or water-washing, was expected to reduce biochar yield in favor of liquid products and volatiles, with the latter related to the formation of interconnected, open porosity in pyrolytic carbons14,22,38,42,53. Demineralization of feedstocks has also been observed to increase the oxygen content of biochars14. These expected effects were tracked in the appropriate sections below as a function of potassium content to validate experimental methods in gauging the effects of lignin S/G, as well as investigating any synergistic effects between lignin S/G and naturally-occurring potassium content.Thermal gravimetric analysisThermal gravimetric studies provided insight into the moisture, ash, volatile carbon, and recalcitrant carbon content, as well as decomposition behavior, and chemistry of biochars. Characteristic differences were observed based on S/G and demineralization (Supplemental Information Table S3).Demineralization reduced the ash content (Table S3) in BESC-316-based biochars, BC-B and BC-BL40, by approximately half (1.91% wt. to 1.02% wt., respectively). Similarly, the reduction for CHWH-27-2-based biochars, BC-C and BC-CL40, was ⁓60% wt. (2.44% wt. to 0.94% wt., respectively). BC-C contained more inorganics than BC-B, while for BC-CL40 and BC-BL40 inorganics content was roughly equivalent. Demineralized biochars lost more weight between 450 and 1000 °C (in inert environment) compared to untreated biochars (Fig. 1), which can be accounted for by the catalytic action of potassium in untreated feedstock promoting char-forming reactions. Moisture content was lower for demineralized biochars, but no trend for moisture was observed for S/G. The relative quantity of recalcitrant carbon increased with demineralization alone, however no trends for volatile carbon were observed.Figure 1TGA and DTG results for biochars heated in argon indicate the strong effects of demineralization on thermal decomposition. Demineralized biochars lost more weight at temperatures above 450 °C than untreated biochars.DTG curves for all biochars, shown in Fig. 1, reveal that above ⁓450 °C, the rate of mass loss for demineralized chars increased significantly compared to untreated chars. This result will be discussed in the following section.Effusion of CO2 and H2
Gaseous species are produced during thermal decomposition as carbonization progresses. CO2 and H2 are produced from oxygenated and hydrogenated surface species, consistent with the hydrogen- and oxygen-rich functional groups of biochar, thus can be used to interpret biochar composition. Assignment of oxygenated functional group origins for CO2 was made according to Szymanski et al.54, validated by diffuse reflectance infrared Fourier-transform spectroscopy (DRIFTS) spectra. CO2 and H2 MS ion currents and DTG curves are shown in Fig. 2 for each biochar. Effusion of CO2 at approximately 120 °C was attributed to surface-adsorbed atmospheric CO2. Production began in earnest at approximately 450 °C, and between 450 and 750 °C the first of two maxima in CO2 effusion was observed, attributed to decomposition of carboxylic anhydride and lactone groups on the primary char. The second maximum, observed between 750 and 950 °C, is attributed to ether and pyrone structures on the primary char. BC-CL40 produced the most intense CO2 ion current, indicating high S/G may enhance oxygen-content. The greater CO2 effusion of demineralized chars compared to untreated chars is consistent with higher O-content14,42. Even naturally-occurring K can lower the activation energy for thermal decomposition of lignocellulosic feedstock, catalyze condensation reactions, and increase the production of gaseous products, particularly CO2, possibly through catalysis of gasification reactions38,39,40,41,42. The overall effect is reduction of oxygenated species in the solid biochar from untreated feedstock.Figure 2Background-subtracted MS signals for H2 (top) and CO2 (center) with the corresponding DTG signals (bottom—also shown in this figure). The dashed vertical line indicates the first H2 evolution peak.H2 effusion began above 700 °C in all materials. Untreated biochars showed a broad peak at 800 °C, while this peak was much lower intensity for demineralized chars. All biochars exhibited a sharp increase in H2 production above 900 °C. Demineralized biochars produced a greater increase in intensity above 900 °C compared to untreated materials, consistent with differences in H-mobility during carbonization or in the chemical environment of hydrogen31.In amorphous carbons such as biochars, interpretation of hydrogen effusion is complicated by its relationship to carbon structure, to catalytic action of inorganics, particularly K and Ca, and to gasification reactions which may produce hydrogen31,55,56. Hydrogen originates from C–H in both sp2 and sp3 hydrocarbons, and during pyrolysis it migrates along the carbonaceous skeleton as carbonization and reorganization processes proceed31. BC-CL40 shows the highest H2 production near 1000 °C, much higher than the original pyrolysis temperature (700 °C), which suggests a possible link between high S/G, demineralization, and high-temperature carbonization behavior.In Fig. 2, the DTG curves from Fig. 1 are reproduced to aid in interpretation of ion current data. Demineralized biochars showed higher rates of thermal decomposition above 450 °C. Between 450 and 800 °C this increase roughly corresponds to an increase in CO2 effusion. Above 800 °C, however, CO2 effusion drops off. H2 production alone is unlikely to account for the increased rate of mass loss. Instead, this increase at high temperatures is attributed to products of thermal cracking processes with increased carbonization and production of heavier polyaromatic hydrocarbons57,58.H/C as an indicator of aromaticityAtomic hydrogen to carbon ratio (H/C) has been identified by Xiao et al. as a simple indicator of biochar aromaticity13. H/C for all biochars (Table S3) ranged from 0.4 to 0.5, which are somewhat high compared to other 700 °C biochars13. This result is attributed to the very short dwell time at HTT. Within the error of the H/C calculation procedure, there was no apparent link to H/C based on S/G or demineralization. This result is supported by the Raman results discussed further on. It is likely, then, that H/C, or aromaticity, is not dependent on lignin S/G or K-content of feedstocks but may have variation based on other feedstock parameters.Diffuse reflectance infrared Fourier-transform spectroscopy (DRIFTS)DRIFTS was used to semi-quantitatively validate the presence of oxygen- and hydrogen-containing functional groups. Raw, vertically-shifted DRIFTS spectra for all biochars and the reference activated carbon are shown in Fig. 3a, while the background-subtracted, normalized fingerprint region for only the biochars are shown in Fig. 3b to highlight the presence of oxygenated functional groups. In Fig. 3b, all spectra were normalized to the maximum at 1585 cm−1, the aromatic ring C=C stretch, based on the observation that all materials contain roughly the same H/C and the same density of 6-member rings59, as evidenced by identical I(D)/I(G) reported in Table 2. The Norit RB3 measurement was used to determine how its oxygen-content compared to that of the biochars, aiding in the interpretation of ammonia breakthrough curves.Figure 3(a) Raw, vertically shifted DRIFTS spectra for the four biochars and Norit RB3; (b) Normalized, background-subtracted DRIFTS fingerprint regions of all four biochars. Curves in b correspond to the legend in a.Table 2 Results of peak fitting D and G bands of Raman spectra (N = 3; \(\overline{X }\)± σs) and volumetric N2 adsorption.The normalized spectra from the four biochars were quite similar, most notably regarding intensity related to carbonyl groups (1690 cm−1) and C–O groups (1300–1000 cm−1).The low intensity in the O–H stretch region (3000–4000 cm−1) in Fig. 3a for all materials indicates little to no physically adsorbed water and an absence of –O–H functional groups, ruling out the presence of phenolic and carboxylic acid groups. This observation does not, however, rule out anhydrides that will participate in ammonia chemisorption9,10,48.The fingerprint region in Fig. 3a shows important characteristic differences in oxygen-containing surface chemistry between the biochars and the activated carbon (Norit RB3) with implications for ammonia adsorption. In Fig. 3b, the C=O conjugated lactone/quinone stretch at 1690 cm−1, the aliphatic ether C–O–C mode at 1050–1000 cm−1 and aromatic ether mode 1300–1210 cm−1 are observed in the spectra of all biochars. These groups were not observed in the spectrum of Norit RB. The normalized C=O peak at 1690 cm−1 is most intense for BC-CL40.Raman spectroscopyThe dominant features of the Raman spectra (Fig. 4) for all biochars, the so-called D and G bands, occur in the first-order region between 1000 and 1800 cm−1. D and G band integral sub-peaks have been previously fitted using a number of physically meaningful models59,60,61. Comparative Raman curves for biochars and Norit RB3 are shown in Fig. S3. The low, broad second-order region, characteristic of amorphous carbons, implies limited and highly disordered 3D stacking of aromatic ring structures62,63. In this present work, the concepts of Ferrari and Robertson59 have been employed to fit the first-order band between 1000 and 2000 cm−1 with two functions and a linear background64,65,66. Resulting fitted parameters, listed in Table 2, were used comparatively. More information on fitting parameters is found in the Supplemental Information.Figure 4Raw, vertically shifted Raman spectra for all biochars. The D and G bands are the most prominent features.Ratios of the peak heights from each of the fitted D and G bands were used to infer information about the structural arrangements (relative degree of amorphous character) of the biochar carbon skeletons. The relative intensity of the D and G bands, I(D)/I(G), is an indicator of aromatic ring clustering in amorphous carbons59. I(D)/I(G) was equivalent for all biochars, indicating neither S/G nor K-content is determinant of the size of aromatic ring clusters. It is more likely that HTT is deterministic of I(D)/I(G), and I(D)/I(G) was much lower for biochar than for the activated carbon Norit RB3, consistent with a more highly carbonized structure expected in the latter (Fig. S3).The position of the G-band (Pos(G)) was found to be consistent for all materials, including Norit RB3. A Pos(G) near or above 1600 cm−1 indicates linear sp2 chains are present in the material59.The full-width at half-maximum (FWHM) of the G-band is also relevant to biochar structure. The narrower FWHM(G) for BC-C (75 cm−1 compared to 81 cm−1 for other biochars) indicates in increase in order of sp2 carbon phases which could manifest as either larger or less defected clusters of aromatic rings, or more order in linear sp2 chains63. BC-C was prepared from the feedstock with high K-content and high S/G, implicating some synergy between the two feedstock parameters in determining carbonization behavior.Specific surface area and porosityThe results of volumetric N2 adsorption were used to assess the role of S/G in the development of porosity and nitrogen-accessible surface area, and also to understand the potentially competing roles of oxygen-content and pore volume and surface area on dynamic ammonia adsorption in biochars. Results are listed in Table 2.SBET also showed a strong dependence on naturally-occurring K-content, consistent with the findings of others14,40. Demineralized chars have approximately four hundred times greater SBET compared to the untreated biochars.Lignin S/G trended negatively with SBET in demineralized chars, with BC-BL40 showing 11% greater SBET than BC-CL40. According to the fundamentals of pore formation in pyrolytic carbons, and consistent with the work of others26, it is possible this result could be attributed to enhanced formation of volatiles favored by low S/G feedstock.Pore formation is initiated during pyrolysis as gases and volatiles escape the bulk, leaving behind voids. Open pores are formed when volatiles and gases escape efficiently from the bulk, exposing the spaces between aromatic layers (micropores) to larger connecting channels (mesopores) which subsequently open to macropores or to the surface67. If volatiles are not able to efficiently exit the bulk, they can clog pathways out of the structure, resulting in closed porosity and very low accessible surface area68.Somewhat counterintuitively, for demineralized chars, high methoxy group content should result in more cross-linking among intermediate products following homolysis of methoxy groups, which may reduce volatiles production by way of enhanced secondary reactions between volatile species and the char skeleton26,69. The result should be a reduction in N2-accessible surface area for high S/G materials, which was indeed observed in this present work. Based on the high significance of this result, more studies with P. trichocarpa variants are merited to determine the magnitude of any correlation between S/G and pore volume and SBET over a range of HTT.Dynamic ammonia adsorptionBreakthrough curves for ammonia uptake by the four biochars and Norit RB3 are shown in Fig. 5. Full breakthrough curves for representative experiments are shown in the Supplemental Information, Fig. S6. The curves in Fig. 5 represent the average of three breakthrough trials for each material, having a width of two standard deviations on the mean. Mean breakthrough times for C/C0 = 0.10 and C/C0 = 0.01 for all materials are reported in Table 3.Figure 5Averaged breakthrough curves (mean of 3 repetitions) for all four biochars and Norit RB3. The colored bands represent ± σ. Horizontal dashed lines are included to guide the eye to C/C0 = 0.01 and 0.1Table 3 Mean breakthrough times (N = 3; \(\overline{X }\)± σ).Biochar performed better than Norit RB3 in ammonia breakthrough tests. BC-B, BC-BL40, BC-C and BC-CL40 had 78%, 110%, 97%, and 115% and 131%, 185%, 146%, and 177% increase in mean breakthrough time compared to Norit RB3 at C/C0 = 0.10 and C/C0 = 0.01, respectively.Among biochars alone, demineralization and lignin S/G both correlated strongly with ammonia breakthrough. Mean breakthrough time for BC-BL40 was 18% and 23% longer than BC-B at C/C0 = 0.10 and C/C0 = 0.01, respectively, and mean breakthrough time for BC-CL40 was 9% and 13% longer than BC-C at C/C0 = 0.10 and C/C0 = 0.01, respectively. For untreated biochars, BC-C had 10% longer mean breakthrough time compared to BC-B at C/C0 = 0.10, but only 7% increase in mean breakthrough time at C/C0 = 0.01. BC-B and BC-C showed similar specific surface areas (Table 2) but had very different lignin S/G contents (1.67 and 3.88, respectively). It is proposed that these differences in breakthrough time are attributed to different oxygen chemistries at the particle surfaces and macropores.Breakthrough time did not correlate with SBET when all materials were considered collectively (R2 = 0.535 and 0.546 for C/C0 = 0.01 and 0.10, respectively), shown graphically in Fig. 6a. For instance, SBET for Norit RB3 was approximately 103 times that of BC-C, however the breakthrough time for Norit RB3 (6.0 s) was approximately half that of BC-C (11.8 s) at C/C0 = 0.10. Norit RB3, BC-BL40, and BC-CL40 are microporous (Supplemental Information Fig S5). The poor correlation between SBET and breakthrough time suggests that high micropore surface area alone is not deterministic of ammonia breakthrough performance. However, when only the biochars were considered, breakthrough time showed a positive correlation with SBET at both C/C0 = 0.01 and 0.10 (R2 = 0.7540 and 0.9366, respectively). This relationship is shown in Fig. 6b.Figure 6(a) Breakthrough time vs. BET surface area for all five materials for C/C0 = 0.01 and C/C0 = 0.1; (b) Breakthrough time vs. BET surface area for biochars only.Microporosity in carbon-based adsorbents is typically attributed to slit-like pores associated with misaligned clusters of aromatic rings. The weak correlation between SBET and breakthrough time indicates that high N2-accessible micropore surface area, and thus carbon structure, is not responsible for adsorption up to breakthrough.The poor correlation with micropore surface area indicates that another factor, most likely surface chemistry, is driving breakthrough time, and specifically, the chemistry of outer surfaces and macropores. Lodewyckx and Wood noted that for chemisorption, micropore volume may be less important than chemical interactions between the contaminant and active sites in activated carbons49. Huang et al. and Domingo-Garcia et al. found that surface area, particularly that associated with micropores, was not the governing factor in determining breakthrough behavior of ammonia on oxidized activated carbons due to the prevalence of acidic oxygen-containing groups at the entrances to micropores45,48.For large particles, such as granules or rods, breakthrough is dominated by boundary layer diffusion70,71. For particles on the order of ⁓150 µm, as in this study, intraparticle diffusion is expected to determine the overall dynamic adsorption rate71. Because microporosity was not correlated to breakthrough time, differences in surface chemistry can be interpreted as driving the observed behavior.The presence of surface acid sites, evidenced by DRIFTS band between 1750 and 1650 cm−1 (Fig. 3b) for all biochars, indicate carboxylic anhydrides, pyrone, quinone, or lactone species on the surface of biochar particles are likely to be involved in adsorption up to breakthrough. These bands are noticeably absent from or have very low intensity in the DRIFTS spectrum of Norit RB3 (Fig. 3a). Weak acid groups would readily take up ammonia via chemisorption during the initial stages of the dynamic adsorption process, but restrict further access to micropore surface area if located at the edges to micropores. Domingo-Garcia et al. observed this latter phenomenon for oxidized activated carbons48. Thus, for oxygenated species located within pores, access would be limited on short time scales associated with breakthrough. Because microporosity did not correlate to breakthrough time, differences in surface chemistry can be interpreted as driving the observed behavior.
Effects of lignin syringyl to guaiacyl ratio on cottonwood biochar adsorbent properties and performance
