Structural Transformation of Biochar Black Carbon by C60 Superstructure: Environmental Implications

Pyrogenic carbon is widespread in soil due to wildfires, soot deposition, and intentional amendment of pyrolyzed waste biomass (biochar). Interactions between engineered carbon nanoparticles and natural pyrogenic carbon (char) are unknown. This study first employed transmission electron microscopy (TEM) and X-ray diffraction (XRD) to interpret the superstructure composing aqueous fullerene C60 nanoparticles prepared by prolonged stirring of commercial fullerite in water (nC60-stir). The nC60-stir was a superstructure composed of face-centered cubic (fcc) close-packing of near-spherical C60 superatoms. The nC60-stir superstructure (≈100 nm) reproducibly disintegrated pecan shell biochar pellets (2 mm) made at 700 °C into a stable and homogeneous aqueous colloidal (<100 nm) suspension. The amorphous carbon structure of biochar was preserved after the disintegration, which only occurred above the weight ratio of 30,000 biochar to nC60-stir. Favorable hydrophobic surface interactions between nC60-stir and 700 °C biochar likely disrupted van der Waals forces holding together the amorphous carbon units of biochar and C60 packing in the nC60 superstructure.


Results and Discussion
TEM imaging of biochar nanostructures. Biochars are hereby denoted by the feedstock acronym and pyrolysis temperature, e.g., pecan shell feedstock (PS25) and biochar produced at 700 °C (PS700). Grand Canonical Monte Carlo (GCMC) density functional theory (DFT) analysis of CO 2 isotherm indicated a progressive increase in the surface area of biochars from 271 to 542 m 2 g −1 as a function of pyrolysis temperature (400-700 °C, Table S1). Carbon dioxide measures surface area originating primarily in micropores <1.5 nm in aperture. Because PS700 showed the highest total surface area, TEM was employed to image the nanostructure of PS700.
To probe variations in mass/thickness, high angular annular dark field (HAADF) scanning TEM (STEM) was used, as it is sensitive to the atomic number (∝Z 2 ) 32 . In this fashion, regions of PS700 having high Z are brighter under the HAADF STEM contrast, while regions with low Z are darker. Because carbon has a low Z value, an accelerating voltage of 120 keV was employed to increase the scattering cross-section and improve image contrast. As shown in Fig. 1, ball-milled PS700 particles of 1-2 µm in size had regions with darker and brighter contrasts. In Fig. 1, yellow squares indicate two locations where EDS signals were obtained. Spectrum 1 (obtained from a particle without brighter contrast) shows carbon as the only element. Spectrum 2 (obtained from a particle with brighter contrast) is dominated by Ca, C, and O signals. Copper peaks in both spectra originate from the Cu grid. Thus, calcium is the cause for the brighter contrast in the STEM-HAADF image. Phase contrast TEM images confirmed that the presence of Ca is related with crystalline CaCO 3 phase 33 .
The HAADF STEM is a mass-thickness contrast technique, which is sensitive to variations in thickness, and is used to identify meso-(2-50 nm) and macropores (>50 nm) 32 . Fig. 2 shows a bright-field STEM image of PS700. Because the carbon is amorphous (shown by Fast Fourier Transform (FFT) in the inset of Fig. 2a), the contrast primarily results from differences in mass/thickness throughout the amorphous carbon. Pores within the amorphous carbon sample would create a darker contrast in Fig. 2. These dark spots shown in Fig. 2b are likely the locations of micropores measurable by CO 2 GCMC (Table S1). Based on the intensity profile across these regions (described in detail in Section VIII of Supporting Information), the pore diameter was estimated to be approximately 0.39 ± 0.02 nm (Fig. 2c). For the first time, bright-field STEM was employed to visually estimate the size of pores open to the surface of a black carbon material (amorphous carbon of PS700) to be approximately 0.39 ± 0.02 nm.
TEM imaging of nC 60 -stir superstructure. Bright-field TEM images of nC 60 -stir show lattice fringes, which indicate the presence of crystallinity (Fig. 3a,b); higher magnification images are provided in Figures S4-S5, Supporting Information. The FFT obtained from Fig. 3b (displayed in the inset) reveals that the nC 60 -stir exhibits a face-centered-cubic structure with a lattice parameter, a = 1.356 nm. To better illustrate this configuration, a model 34 is shown in Fig. 3d. Clearly, the nC 60 -stir is a superstructure formed by C 60 near-spheres in a fcc configuration with a lattice parameter of 1.356 nm. To further confirm the superstructure of nC 60 -stir, both nC 60 -stir (Fig. 3c, top spectrum) and fullerite (Fig. 3c, bottom spectrum) were analyzed by XRD. The XRD of nC 60 -stir shown in Fig. 3c (top spectrum) matched that of fullerite in the XRD database (bottom spectrum). In conclusion, both TEM and XRD analyses indicate that nC 60 -stir is a superstructure self-assembly of near-spherical C 60 molecules in a fcc configuration (Fig. 3d), much like the parent fullerite 16 . Although this fcc tendency has been observed in nC 60 -stir 12 , the structural origin has not been explained or interpreted, and was often assumed to be the homo-aggregate of individual C 60 molecules 1-7 . The polydispersity of nC 60 superstructure likely controls its dissolution into the toluene extraction fluid used to determine the aqueous "dissolved" C 60 concentration, [nC 60 ] stock , as illustrated in the next section. Characterization of aqueous nC 60 stock solutions. Prolonged stirring of fullerite in water (to prepare nC 60 -stir) is intended to simulate the mixing process 8 taking place in the environment following an accidental spill of fullerite. In previous studies, as summarized in Section VI of Supporting Information, the reproducibility of nC 60 -stir preparation method had either not been addressed or was called into question 8,35 . Of four separate batches of nC 60 -stir prepared to test the reproducibility of production in the present study, only three produced HPLC-detectable nC 60 -stir ( Fig. 4c-e). Although absorbance was low because of the low nC 60 -stir concentration, two broad peaks were observed at 260-274 and 350 nm (Fig. 4c-e), in agreement with the literature; 35,36 a baseline shift at 800 nm 36 in some prior reports (which is absent in filtered samples) indicates the presence of unfiltered fullerite. In Fig. 4c-e, nC 60 -stir concentration (by HPLC) and size (by DLS) varied for 3 different batches from 0.10-0.26 mg L −1 and 124-223 nm, respectively. As a reference, Fig. 4a shows the UV/visible spectrum of 20 mg L −1 C 60 in toluene corresponding to its HPLC peak.
Because filtration is often used to fractionate nC 60 into different nm ranges (based on the filter's pore diameter) before experiments 3,37 , filtered nC 60 stock solutions were characterized. The thin lines in Fig. 4b (nC 60 -sonicate), and 4c-4e (separate batches of nC 60 -stir) show decreased UV/visible absorbance upon syringe filtration (0.45 µm PTFE). In Fig. 4c-e (nC 60 -stir), the characteristic peaks at 260 and 350 nm disappeared upon filtration. For nC 60 -sonicate (Fig. 4b), the peaks at 260 and 344 nm remained but were at lower intensity after filtration; the concentration of nC 60 -sonicate decreased by an order of magnitude, and resulting low count rate (14.3 kcps at 11 attenuation) did not permit a size measurement by DLS. Other syringe filters (0.45 µm cellulose acetate and nylon, VWR) lead to similar reduction in the absorbance of nC 60 -sonicate, even though all stock solutions in Fig. 4 had been previously vacuum-filtered through 0.45 µm cellulose acetate membrane filter, as described in the Methods section. Based on above-described influence of syringe filtration, only toluene layer (and not aqueous layer) was filtered prior to the HPLC quantification of C 60 in the subsequent sections. Above observations suggest that intended size fractionation of nC 60 via filtration, often employed in the prior reports 3,37 , could filter out the particles smaller than the filter's pore diameter. Figure 5 presents the mass distribution (in μg) of nC 60 -sonicate (a-b) and nC 60 -stir (c) in mean ± s.d. from duplicate retention experiments. In Fig. 5, nC 60 mass fractions were determined based on [nC 60 ] dissolved for "dissolved" portion of nC 60 in water at the sampling time, and [nC 60 ] retained for biochar-associated "retained" fraction, as described in detail in Methods section. Mass balance was determined as the sum of dissolved and retained fractions. The "dissolved at t = 0" fraction was based on [nC 60 ] stock measured at each sampling time using the control reactors containing nC 60 -stir or nC 60 -sonicate alone, without biochar. Figure 5 in the unit concentration and [nC 60 ] retained,calc (Equation 1) are provided in Figure S6, Supporting Information.

Surface interactions between nC 60 supercrystals and 2-mm biochar pellets.
Kinetic experiments were first performed at 5 g L −1 PS300 using nC 60 -sonicate that is an order of magnitude more concentrated than nC 60 -stir (Fig. 4). As shown in Fig. 5a, dissolved mass of nC 60 -sonicate remained constant over 1 wk period in the presence of low pyrolysis temperature (300 °C) biochar. Hot toluene extraction of oven-dried biochar after the retention experiment did not recover nC 60 -sonicate during the timecourse of the experiment (Fig. 5a). As a result, the C 60 mass balance (dissolved+retained) equaled the dissolved fraction over 1 wk period. To investigate the influence of pyrolysis temperature and biochar loading, 1-day equilibration experiment was conducted at a higher loading (20 g L −1 ) of 300, 500, and 700 °C biochars (Fig. 5b). Higher biochar loading did not significantly decrease the dissolved nC 60 -sonicate fraction in the presence of PS300 (5-20 g L −1 in Fig. 5a,b). Because of negligibly low solid-associated "retained" fraction, the mass balance of nC 60 -sonicate was within the error range of the dissolved fraction (PS700 < PS300 < PS500). In conclusion, the retention of nC 60 -sonicate by 5-20 g L −1 of 300-700 °C biochars was negligibly low. The results in Fig. 5a,b suggest charge repulsion between hydroxyl-enriched nC 60 -sonicate and carboxyl-enriched PS300; both are negatively charged 8,38 . Retained fraction of nC 60 -sonicate was observed only on PS700 (Fig. 5b), which relative to PS300 is more hydrophobic and contains fewer oxygen-containing functional groups 38,39 . Figure 5c presents the retention of nC 60 -stir on 5 g L −1 of 300, 500, and 700 °C biochars. At an order of magnitude lower concentration of nC 60 -stir than nC 60 -sonicate, a significant fraction of nC 60 -stir was retained by biochars in the order, PS300 ≈ PS700 < PS500 ("retained" in Fig. 5c). Dissolved nC 60 -stir fraction followed the order, PS500 ≈ PS700 < PS300. The mass balance did not show a clear temperature trend, and exceeded the dissolved fraction at t = 0 for PS500 having the highest retained fraction. This could originate from the greater extraction efficiency of nC 60 -stir from dried biochar (using hot toluene method) than from water. In conclusion, the recovery of nC 60 -stir retained by the biochar was greater than nC 60 -sonicate, despite an order of magnitude lower [nC 60 ] stock of nC 60 -stir than nC 60 -sonicate. The nC 60 -sonicate is likely to contain higher amounts of hydroxyl substituent than nC 60 -stir, because of the sonication process 22 incurring radical formation 40 . Favorable hydrophobic interactions between nC 60 -stir and biochars likely drove the formation of biochar-associated nC 60 -stir in Fig. 5c.
When high temperature (700 °C) biochar was equilibrated with nC 60 -stir at a sufficiently high biochar loading (≥20 g biochar L −1 ), 2-mm biochar pellets reproducibly disintegrated to form a homogeneous, stable, black-colored aqueous colloidal suspension (Fig. 6e). The PS700+nC 60 -stir suspension (Fig. 6e, far right) was produced at >30,000 biochar/nC 60 ratio by weight, and contained measurable [nC 60 ] dissolved . The suspension was stable after the supernatant containing nC 60 -stir was replaced by water ( Figure S7 top far right, Supporting Information). The disintegration of biochar pellets was not observed at lower biochar loadings ( Figure S7 top), when low pyrolysis temperature biochar was employed (PS350 in Figure S7, bottom), or when nC 60 -sonicate was employed instead of nC 60 -stir. Higher temperature biochars have higher attrition 41 to form smaller particles by mechanical forces 42,43 . Higher biochar loading could enhance the mechanical crushing of biochar pellets during the end-over-end rotation in the presence of nC 60 -stir. Hydrophobic interactions between PS700 and nC 60 -stir (but not nC 60 -sonicate) lead to the retention of nC 60 -stir (Fig. 5c) accompanied by the disintegration of biochar (Fig. 6). Collectively, hydrophobic interactions drove the retention of nC 60 -stir (but not nC 60 -sonicate having   Figure 6a,b indicate the formation of amorphous carbon (FFT shown in inset of Fig. 6b) nanoparticles from PS700 pellets, in the presence of nC 60 -stir. Spaghetti-like amorphous carbon is composed of random carbon domains with defects linked by the aliphatic carbon chains 44 . On the other hand, Fig. 6c,d show a polycrystalline structure composed by nanocrystals having different orientations, as confirmed by multiple diffraction spots (FFT inset of Fig. 6d). However, the lattice spacing was not consistent with the C 60 superstructure (Fig. 3), and thus could originate from CaCO 3 in pecan shell biochar (Fig. 1) 41 . It is inherently challenging to distinguish two carbon materials (char and C 60 ) by TEM because of low contrast and overlapping projection 30,45 . However, Fig. 6d shows graphitic structures (arrow) on the edge of amorphous carbon that could originate from the decomposition of nC 60 -stir superstructure to form C 60 molecules. This graphitic structure was not observed in biochar alone (Figs 1, 2) or nC 60 -stir alone (Fig. 3). Figure 7a focuses on the polycrystalline structure within the phase contrast TEM image of PS700+nC 60 -stir. Nanocrystals having different orientations are confirmed by the multiple diffraction spots and lattice fringes in different directions (Fig. 7b is the FFT image of Fig. 7a). By indexing the FFT image, all diffraction rings matched CaCO 3 planes: (102), (104), (113), and (018) towards outer rings in Fig. 7b. However, both yellow and red diffraction spots in Fig. 7b matched the lattice spacing of C 60 superstructure: 0.49 nm of (220) plane (red spots in Fig. 7b), and 0.795 nm of (200) plane (yellow spots). In Figures 7c (for yellow spots in Figure 7b) and 7d (for red spots in Figure 7b), an inverse FFT technique was employed to visualize the retained fraction of nC 60 -stir in PS700. As shown in Fig. 7, the inverse FFT on the red and yellow diffraction spots revealed the C 60 superstructures embedded within PS700, i.e., [nC 60 ] retained .
In conclusion, Figs 4-7 re-emphasize that nC 60 is a superstructure, rather than the homo-aggregate of C 60 molecules. If nC 60 was the aggregate of individual C 60 molecules, penetration into microporous networks is expected. Such reaction is expected to be highly irreversible, and will be controlled by the abundance of micropores (<2 nm) 46,47 , which progressively increases from 400 to 700 °C (Table S1). In contrast, van der Waals and hydrophobic interactions involving the polyaromatic surface of PS700 will favor the heteroaggregation of nC 60 superstructure. Biochar's hydrophobicity progressively increases as a function of pyrolysis temperature, resulting in the lower H/C atomic ratio (Table S1). Heteroaggregation of nC 60 -stir with PS700 (Fig. 5c) and associated hydrophobic interactions could disrupt the relatively weak van der Waals forces holding together (i) amorphous carbon units 44   Physical disintegration of biochar particle by the engineered carbon nanomaterial (nC 60 , Fig. 6) without sonication 49 , will pose a number of environmental consequences. Environmental transport of pyrogenic carbon is strongly size-dependent 50 , and constitutes a significant proportion of the global carbon cycle 51 . The presence of hydrophobic carbon nanoparticles, like nC 60 -stir occurring from accidental spill, will promote the transport of biochar soil amendment 52 by producing biochar nanoparticles (Fig. 6) and composites (Fig. 7). Produced biochar nanoparticles will have additional environmental consequences 53 , including the off-site migration of sorbed pollutants 25 .

Methods
Distilled, deionized water (DDW) with a resistivity of 18 MΩ cm (APS Water Services, Van Nuys, CA) was used in all procedures. Unless otherwise noted, all chemical reagents were obtained from Sigma-Aldrich (Milwaukee, WI) at the highest purity available.
Pecan shell biochars. As described in detail previously 49,54 , pecan shell feedstock (PS25) was ground (SM 2000 cutting mill, Retsch Gmbh, Haan, Germany) and sieved (<2 mm) prior to pyrolysis at 300, 350, 400, 500, 600, or 700 °C under a flow rate of 1,600 mL min −1 N 2 for 4 h using a laboratory scale box furnace (22 L void volume) with a retort (Lindberg, Type 51662-HR, Watertown, WI). Biochar products were allowed to cool to room temperature overnight under the N 2 atmosphere. Proximate and ultimate analysis results 54 and N 2 and CO 2 isotherms-based surface area and porosity are summarized in Table S1 of Supporting Information. Aqueous nC 60 stock solutions. Two published methods were used. In the first 8,55,56 , bulk fullerene C 60 powder (99.9% purity fullerite; Materials and Electrochemical Research, Tucson, AZ) was magnetically stirred in DDW (1.0 g L −1 ) in the dark for 40 d. The suspension was initially black and gradually turned brown over the 40 d stirring period. The suspended particles were removed by the vacuum filtration (DDW pre-rinsed 0.2 μm cellulose acetate membrane; Sartorius, Bohemia, NY) to produce a clear stock solution having a light yellow/ brown hue and a pH of 6.0 (Sartorius Professional meter PP-15). The resulting stock solution is denoted nC 60 -stir. The second method 57 employed a sonication probe (450 Sonifier, Branson Ultrasonics, Danbury, CT) to increase the nC 60 concentration by the oxidative formation of hydroxyl substituents 22,37 . Five mL of C 60 powder dissolved in toluene (1.2 g L −1 clear purple solution) was added to a solution composed of 50 mL DDW and 1.5 mL ethanol. Figure S1 of Supporting Information shows the resulting solution having toluene (top) and aqueous (bottom) layers 58 . This solution was sonicated by directly inserting the probe for 3 h while periodically adding DDW to replace the water evaporated as a result of the exothermic sonication process. Sonication caused the solution to develop a cloudy yellow-white color ( Figure S2), and the final solution (after toluene was driven off by the heating) was clear yellow. This solution was vacuum filtered through 0.45 μm (cellulose acetate) and then 0.2 μm (cellulose nitrate) membrane filters. The resulting stock solution is denoted nC 60 -sonicate ( Figure S3). An analogous procedure was followed to produce nC 70 stock solution by sonication (nC 70 -sonicate characterized in the Section VII of Supporting Information) to use as the internal standard in HPLC quantification of nC 60 . All aqueous fullerene stock solutions (nC 60 -stir, nC 60 -sonicate, and nC 70 -sonicate) were stored at 25 °C in the dark, and were stable for several months, as reported in the literature 59 . The stock solutions were characterized by UV/visible spectrophotometry (HP8452A, Hewlett-Packard, Palo Alto, CA) with DDW as the blank. Hydrodynamic diameter was determined by dynamic light scattering (DLS; Zetasizer NanoZS, Malvern, Westborough, MA). All DLS analyses were performed in triplicate using the disposable sizing cuvette at the material RI of 2.20, attenuation of 11, water as the dispersant, and by the default general method algorithm; count rate (in kcps) and polydispersity index (PdI) were recorded in addition to the hydrodynamic diameter.

Surface interaction of nC 60 with biochars.
Batch experiments were conducted in duplicate using amber glass vials with Teflon-lined screw caps (40 mL nominal volume, Thermo Fisher Scientific, Waltham, MA) at 5-20 g biochar L −1 ; 30 mL of undiluted nC 60 stock solution was added directly to dry 2-mm biochar pellets. Reactors were equilibrated by shaking end-over-end at 70 rpm. Control experiments were conducted for the nC 60 stock solution without biochar, and biochar without nC 60 stock solution, each in duplicate. At each sampling time, biochar was allowed to settle for 1 h, and then the supernatant was carefully decanted into a new glass vial. The supernatant was mixed with 200 g L −1 NaCl stock solution to yield 1 wt% NaCl. The NaCl was used to facilitate the transfer of nC 60 from the aqueous to toluene phase, and to prevent emulsion 9 . After vigorous shaking by hand, 4 mL toluene (HPLC grade) was added, and the reactor was vortexed and then rotated at 70 rpm overnight. After allowing the two (water and toluene) layers to separate, only the toluene layer was syringe filtered (0.45 μm Millipore Millex-GS; Millipore, Billerica, MA), and 200 µL filtrate was injected into HPLC system with diode array detector (Agilent Technologies, Santa Clara, CA) and Cosmosil Buckyprep-M Packed column (4.6 × 2500 mm; SES Research, Houston, TX). The HPLC column was designed to separate C 60 from C 70 in pure toluene mobile phase at 1.0 mL min −1 flow rate, and C 60 was quantified at λ max of 336 nm in toluene (Fig. 4a) and the retention time of 8 min 60 . Determined nC 60 concentration is hereby denoted [nC 60 ] dissolved . The same procedure was used to determine C 60 concentrations in nC 60 -stir and nC 60 -sonicate stock solutions hereby termed [nC 60 ] stock .
The portion of nC 60 retained by biochar (hereby termed [nC 60 ] retained ) was independently quantified by a hot toluene extraction method 61 . Biochar remaining in each reactor (after decanting supernatant; residual supernatant was determined gravimetrically) was transferred to a clean vial using DDW, and then oven-dried at 45 °C overnight. After recording the weight of oven-dried biochar in the new vial, 2 mL toluene was added, and the resulting biochar suspension in toluene was immersed in 65 °C water bath for 6 h. The reactor was vortexed and rotated at 70 rpm overnight, and then syringe filtered (0.45 μm) for the HPLC analysis of C 60 , as described above. Solid-associated nC 60  where V s is the total volume (30 mL), and m (in g) is the dry weight of 2-mm biochar pellets.
To determine the portion of nC 60 retained by the reactor (hereby termed [nC 60 ] vial ), both the vial and cap of the amber glass reactor were washed thoroughly 3 times with DDW to remove residual supernatant containing nC 60 . Washed reactors were air dried, and then 2 mL toluene was added. The capped reactor was then vortexed and rotated at 70 rpm overnight, and then syringe filtered (0.45 μm) for the HPLC analysis. The resulting [nC 60 ] vial was determined to be negligible for all experiments presented in this study.
In each experiment, mass balance (in μg C 60 ) was calculated based on [nC 60 ] dissolved (solution-phase nC 60 concentration at the sampling time) and [nC 60 ] retained . The mass balance was compared with μg of nC 60 added to each reactor at t = 0, which was calculated based on the reactor volume (30 mL) and [nC 60 ] stock determined at each sampling time using the controls containing nC 60 stock solution without biochar.
TEM imaging of nC 60 -stir and biochar before and after the reaction. Ball-milled and sieved (400 mesh, <37 µm) PS700 was sonicated in ethanol for 15 min. One drop of the resulting suspension was deposited on a 200 mesh carbon-lacey Cu grid. To prepare samples containing nC 60 -stir, two drops of nC 60 -stir before and after the reaction with PS700 were deposited on the grids. TEM images were obtained using a JEOL 2010F TEM (JEOL USA, Peabody, MA) operated at 120 kV, coupled with an energy-dispersive x-ray spectroscopy (EDS). As widely described in the literature, TEM observations could modify the structure of aqueous colloids if exposed to high acceleration voltages 10 , drying or addition of surfactants/solvents during sample preparation 63 , or by freezing employed during the cryogenic TEM. With this in mind, the present study employed 120 kV and short exposure times to minimize these experimental artifacts.
X-ray diffraction. The crystalline structures of nC60-stir stock solution and the bulk fullerite powder were characterized by X-ray diffraction (XRD) with a Philips X'pert diffractometer (with Cu Kα radiation) using a step-scan mode in the range of 10° to 110° with intervals of 0.03° and wavelength of 1.5406 Å (Cu Kα). XRD computer simulations were carried out using a Diamond 3.2e2 software.