Re-recognizing micro locations of nanoscale zero-valent iron in biochar using C-TEM technique

Biochar supported nanoscale zero-valent iron (NZVI/BC), prepared commonly by liquid reduction using sodium borohydride (NaBH4), exhibits better reduction performance for contaminants than bare NZVI. The better reducing ability was attributed to attachment of nanoscale zero-valent iron (NZVI) on biochar (BC) surface or into the interior pores of BC particles due to observations by scanning electron microscopy (SEM) and plan transmission electron microscopy (P-TEM) techniques in previous studies. In this study, cross-sectional TEM (C-TEM) technique was employed firstly to explore location of NZVI in NZVI/BC. It was observed that NZVI is isolated from BC particles, but not located on the surface or in the interior pores of BC particles. This observation was also supported by negligible adsorption and precipitation of Fe2+/Fe3+ and iron hydroxides on BC surface or into interior pores of BC particles respectively. Precipitation of Fe2+ and Fe3+, rather than adsorption, is responsible for the removal of Fe2+ and Fe3+ by BC. Moreover, precipitates of iron hydroxides cannot be reduced to NZVI by NaBH4. In addition to SEM or P-TEM, therefore, C-TEM is a potential technique to characterize the interior morphology of NZVI/BC for better understanding the improved reduction performance of contaminants by NZVI/BC than bare NZVI.

, is conducted through directly observing the inner ultra-thin slices (with thickness less than 100 nm) of samples using P-TEM technique. This technique can eliminate the problem of superimposition of P-TEM technique by avoiding the interfering of the sample surface signals, and thus, obtain the pure signals of interior structure of the sample [33][34][35] . Therefore, C-TEM technique has been applied widely in characterizing of solid materials such as bulk heterojunction films 36,37 , biochar 38 , and activated carbons 39 in previous studies. In this study, therefore, C-TEM technique was firstly employed to directly identify the interior structures of NZVI/BC, especially the micro locations of NZVI in NZVI/BC particles. The comparison of the pH-dependent percent removal curves of metal cations under the presence of solid materials with the pH-dependent precipitation curves of metal cations without solid materials, a successful method established to identify whether the removal of metal cations with solid materials is adsorption or not in our previous studies 40,41 , was employed here to explore whether the removal of Fe 2+ or Fe 3+ ions with BC is adsorption or not. If the two curves overlapped, the removal of Fe 2+ or Fe 3+ in the presence of BC should be attributed primarily to precipitation but not adsorption 40 . If the removal of Fe 2+ or Fe 3+ with BC is by adsorption, the pH-dependent percent removal curve should be over the precipitation curve at some given pH 41 . Reduction experiments of Fe(OH) 2 and Fe(OH) 3 by NaBH 4 was also conducted here to explore whether the adsorbed Fe(OH) 2 or Fe(OH) 3 on BC can be reduced by NaBH 4 . These experiments and the results could be helpful for exploring the underlying mechanisms to explain the better reduction performance of NZVI/BC than bare NZVI for contaminants.

Results and discussion
The surface morphology of BC particles observed by SEM technique is relatively smooth in lamellar structure (Fig. 1a), with abundant irregular pores (Fig. 1b). However, in C-TEM images (Fig. 1c,d), the slices of BC particles are in the sheet-like shape, showing abundant slit pores in uniform pore size. Bare NZVI particles in SEM (Fig. 2a) and C-TEM (Fig. 2b) images are roughly spherical and aggregated significantly to a chain-like structure, which should be attributed to the intrinsic magnetic attraction between the NZVI particles 4,5,12,42,43 . There are a lot of chain-like clusters and roughly spherical particles dispersed on BC surface in SEM images of NZVI/ BC (Fig. 3a,b). These chain-like clusters and roughly spherical particles on the surface of NZVI/BC are NZVI particles, as was also observed by the SEM images of bare NZVI particles (Fig. 2a) and NZVI/BC in previous studies 11,16,18,42 . EDX spectrum of NZVI/BC shows that the main elemental composition of NZVI/BC is C, Fe and O (Fig. 3c). The homogeneous distribution of C and Fe was also verified in SEM elemental mapping of NZVI/ BC ( Figure S1). NZVI particles formed in NZVI/BC are identified by the characteristic peak at 44.7° (Fig. 3d) of XRD pattern of NZVI/BC. NZVI particles formed and dispersed on the surface of NZVI/BC were also observed in P-TEM images of NZVI/BC in previous studies 19,22 . However, in C-TEM images of NZVI/BC ( Fig. 4a and Figure S1), NZVI particles are existed isolated from BC particles rather than attached on the surface of BC particles. The distinguished observations in C-TEM images of NZVI/BC from that in SEM and P-TEM images of NZVI/BC, could be attributed to the top surface view of SEM technique 27,29 and the superimposition problem of   30,31 . Thus, both of SEM and P-TEM techniques cannot explore the interior structure signals of NZVI/BC particles. Moreover, NZVI particles are not observed in interior slit pores of BC particles ( Fig. 4b and Figure S1), i.e., NZVI particles cannot be embedded into the interior pores of BC particles. As reported in the previous literatures 11,44,45 , the average particle size of NZVI on NZVI/BC is about 30-100 nm. In this study, the particle size of NZVI in NZVI/BC is in the range of 40-180 nm (Fig. 3b). However, BC is basically composed of micropores and mesopores with average diameter less than 10 nm 13,38,46 . The N 2 adsorption-desorption isotherm and pore size distribution of BC in this study are shown in Figure S3. The specific surface and average pore size of BC are 579.15 m 2 /g and 2.486 nm, respectively. Therefore, NZVI particles cannot enter into interior pores of BC particles, showing the clean and slit pores without any iron particles in C-TEM images of NZVI/BC (Fig. 4).
In addition, the surface morphology of NZVI/BC in SEM images (Fig. 3a,b) and inner cross section structure    15,[20][21][22] . However, we observed that the removal curves of Fe 3+ in the presence of BC under various pH overlapped with the precipitation curve of Fe 3+ without BC (Fig. 6a,b). Moreover, removal curves of Fe 3+ were independent of the dosage of BC (20 and 200 mg) in Fig. 6a or initial concentration of Fe 3+ (50 and 150 mg/L) in Fig. 6b. Therefore, the removal of Fe 3+ ions in the presence of BC should be attributed primarily to precipitation by formation of Fe(OH) 3 rather than adsorption on BC 40 . Although the removal curve of Fe 2+ didn't overlapped with the precipitation curve of Fe 2+   Table S1. Oxidation of Fe 2+ to Fe 3+ with or without BC were observed (Table S1). In the presence of BC and higher pH, more Fe 2+ were oxidized to Fe 3+ (Table S1). Therefore, the oxidation of Fe 2+ to Fe 3+ occurred in batch experiment for Fe 2+ by BC. Characteristic peak of NZVI was not observed in XRD diffraction pattern of NZVI/BC*, a sample prepared by removing excessive free Fe 2+ in Fe 2+ /BC system before adding NaBH 4 (Fig. 3d) 3 precipitates, deposited on BC and then reduced to NZVI by NaBH 4 was proposed as another possible way to load NZVI on BC surface or in BC interior pores 21,23 . However, depositing of Fe(OH) 2 or Fe(OH) 3 precipitates on the surface or in the interior pores of BC particles were not observed in C-TEM images of NZVI/BC* ( Figure S4), showing that the surface and pores of NZVI/BC* are clean and similar to that of BC ( Figure S4). Moreover, NZVI can be obtained by reducing iron salt with NaBH 4 rather than hydroxide 4,47 , i.e., Fe(OH) 2 or Fe(OH) 3 cannot be reduced to NZVI by NaBH 4 . For example, the suspensions of dark green Fe(OH) 2 precipitates and reddish brown Fe(OH) 3 precipitates (Fig. 7a) , prepared from Fe 2+ and Fe 3+ respectively via chemical precipitation, are not changed in color after dropping NaBH 4 (Fig. 7b) under nitrogen atmosphere, i.e., No black ZVI particles formed after reduction treatment of Fe(OH) 2 and Fe(OH) 3 precipitates with NaBH 4 . The characteristic peak of NZVI was not observed in XRD diffraction patterns of the dried solid samples collected from NaBH 4 treated Fe(OH) 2 and NaBH 4 treated Fe(OH) 3 (Fig. 3d), indicating NZVI cannot be obtained from Fe(OH) 2 and Fe(OH) 3 precipitates by reduction using NaBH 4 . Therefore, NZVI cannot be loaded on BC via depositing of Fe(OH) 2 /Fe(OH) 3 precipitates on surface or into interior pores of BC particles as well as reduction of Fe(OH) 2 /Fe(OH) 3 precipitates to NZVI by NaBH 4. www.nature.com/scientificreports/ Except for loading of NZVI on the surface or into the interior pores of BC, smaller size of NZVI in NZVI/BC was proposed to interpret the better reduction performance of contaminants by NZVI/BC than bare NZVI 5,22,42 . Smaller-sized NZVI is more reactive, which can be attributed to its higher surface area-to-volume ration and stronger quantum effects 4,6,11 . In this study, bare NZVI particles are aggregated into micron-size particles (Fig. 2), which is consistent with the previous reports 6,42,48 . However, the particle size of NZVI in NZVI/BC is in the range of 40-180 nm (Figs. 3b and 4), much smaller than that of bare NZVI (Fig. 2), implying that the agglomeration of NZVI particles is effectively alleviated by BC 10,22 . The alleviation of NZVI agglomeration was also observed in sample NZVI + BC, showing that the particle size of NZVI in NZVI + BC is in the range of 50-220 nm (Fig. 5). Therefore, the smaller size of NZVI in NZVI/BC than that of bare NZVI could be possibly responsible for the better reduction performance of contaminants by NZVI/BC than bare NZVI [10][11][12] . In previous studies 10,13,42 , the smaller size of NZVI in NZVI/BC was attributed to the attachment of NZVI on BC surface or into the interior pores of BC particles that inhibit the aggregation of NZVI. However, in this study, the observed smaller size of NZVI in NZVI/BC should be attributed to the mixing of NZVI with BC that inhibit the aggregation of NZVI, but not the attachment of NZVI on/in BC particles.

Conclusion and outlook
Employing C-TEM technique for liquid reduction method prepared NZVI/BC, it was observed in this study that NZVI particles are existed isolated from BC particles in NZVI/BC, rather than attached on surface or in interior pores of BC particles that was observed in previous study by SEM and P-TEM techniques. This observation was also supported by the negligible adsorption of Fe 2+ /Fe 3+ and the negligible attachment of Fe(OH) 2 / Fe(OH) 3 precipitates on BC surface or interior pore structure of BC particles, as well as the negligible reduction of Fe(OH) 2 /Fe(OH) 3 to NZVI by NaBH 4 . Therefore, both of the widely used SEM and P-TEM techniques have difficulty in observing the inner structure of NZVI/BC. In addition to SEM and P-TEM, C-TEM is a useful technique to characterize the interior morphology of NZVI/BC and other porous materials supported NZVI for better exploring their underlying improved reduction performance of contaminants. Besides, C-TEM can also be employed to characterize the interior morphology and configuration of other composite materials. The better reduction performance of NZVI/BC than bare NZVI could be attributed to the smaller size of NZVI particles in NZVI/BC, due to mixing NZVI with BC particles, which inhibit the aggregation of NZVI, but could not be explained by the attachment of NZVI particles on surface or into interior pores of BC particles in NZVI/BC.

Materials and methods
Chemicals. Ferrous  Preparation of BC and NZVI materials. BC was prepared from wood chips (Ningbo, Zhejiang, China) by oxygen-limited pyrolysis method 49 . Briefly, dried wood chips, grounded to pass through a 0.15 mm sieve, were packed in crucibles, covered with lids, and then, placed in a muffle furnace, pyrolyzed at 700 °C for 6 h. The prepared BC was washed with 1 mol/L HCl (m/v = 0.1 g/mL) by stirring for 24 h to remove impurities and then rinsed using deionized (DI) water to neutral pH. The obtained BC sample was dried overnight at 80 °C in an oven for usage.
Bare NZVI was synthesized using the liquid-phase reduction method 11,12,50 . Briefly, under nitrogen atmosphere, 100 mL NaBH 4 solution (1.2 mol/L) was dropped (1-2 drops/s) into 100 mL FeSO 4 solution (0.44 mol/L) with stirring. The black NZVI particles were separated by centrifugation for 5 min at 3000 rpm. Then, NZVI www.nature.com/scientificreports/ particles were washed with deoxygenated DI water, followed by ethanol, and finally dried in a vacuum oven at 60 °C for 12 h. NZVI/BC was prepared according to the procedures reported in the previous literature 51 . Under nitrogen atmosphere, 100 mL of 0.44 mol/L FeSO 4 solution was mixed with 5.0 g BC for 3 h. Then, 100 mL of 1.2 mol/L NaBH 4 was added dropwise into the mixture. After reaction, NZVI/BC was collected by centrifugation at 3000 rpm for 5 min, washed with deoxygenated DI water and ethanol, and dried in a vacuum oven (60 °C, 12 h).
NZVI/BC* was prepared as follows. 100 mL of 0.44 mol/L FeSO 4 solution was mixed with 5.0 g BC under nitrogen atmosphere for 24 h. Then the solid was collected by centrifugation at 3000 rpm for 5 min and followed by washing with deoxygenated DI water for three times to remove excess free Fe 2+ that not adsorbed on BC. Under nitrogen atmosphere, 100 mL of 1.2 mol/L NaBH 4 was added dropwise into the suspension of BC which was pretreated with FeSO 4 solutions. Then, by centrifugation at 3000 rpm for 5 min, washed with deoxygenated DI water and ethanol, and dried in a vacuum oven at 60 °C for 12 h, nZVI/BC* was collected.
NZVI + BC was prepared directly by mixing BC and NZVI particles. Briefly, 1.0 g BC and 0.5 g NZVI were mixed in 30 mL ethanol in a beaker and dispersed by ultrasonic cell breaker (SCIENTZ-II D) for 10 min (20 °C, 570 W). Then, the mixed NZVI + BC sample was separated by centrifugation at 3000 rpm for 5 min and dried in a vacuum oven (60 °C, 12 h).

Characterization of BC and NZVI materials.
The inner structure information of BC and NZVI materials was observed by C-TEM technique directly 38,39 . These samples were firstly embedded into embedding agent, i.e., spurr resin, which is a mixture of 10 g vinyl cyclohexene dioxide (ERL 4221), 8 g diglycidyl ether of polypropylene glycol (DER 736), 25 g nonenyl succinic anhydride (NSA) and 0.3 g 2-dimethylaminoethanol (DMAE), purchased from SPI-Chem Supplies. Then, spurr resin embedded samples were heated at 70 °C for 12 h, and cut into ultra-thin slices with thickness less than 90 nm by ultramicrotome (Leica, EMUC7, Germany) equipped with diamond knife (Fig. 8). The slices were finally collected on copper mesh grids for observation of TEM at 200 kV (JEM1200, JEOL, Japan).
X-ray diffraction patterns of samples were obtained by an X-ray diffractometer (XRD, Philips, Netherlands) equipped with a CuKα radiation source and scanned at a speed of 2° per min. Surface morphologies of samples were examined by using a field emission SEM (FE-SEM) (SIRON, FEI, Netherlands) equipped with an energy dispersive spectroscopy (Oxford Inca EDS) at a voltage of 25.0 kV.  (25 °C). Briefly, 20 mg or 200 mg BC was added into vials having 18 mL background solution with various pH value, prepared from DI water with 0.01 mol/L HCl and NaOH. Then, 2 mL Fe 2+ or Fe 3+ solution (500 mg/L or 1500 mg/L), was added into the vials to give the initial Fe 2+ or Fe 3+ concentration of 50 mg/L or 150 mg/L, respectively. The vials were capped and transferred to a horizontal shaker (DHZ-D) at 150 rpm for 48 h. After centrifugation at 3000 rpm for 5 min, supernatant was taken from vials, and filtered with a 0.22 μm needle filter to remove the solid. The pH value of supernatant was measured using a glass pH electrode (Mettler Toledo). The concentration of Fe 2+ iron in supernatant was determined by UV-spectrophotometer (UV-2450, Shimadzu, Japan) with 1,10-phenanthroline spectrophotometric method at a maximum wavelength of 510 nm 52 . For measurement of Fe 3+ , Fe 3+ in supernatant was reduced to Fe 2+ firstly by NH 2 OH·HCl 53 , and then, determined by UV-spectrophotometer with 1,10-phenanthroline spectrophotometric method 52 .

Batch experiments for Fe
The removal efficiency of Fe 2+ or Fe 3+ was calculated by Eq. (1): where C 0 (mg/L) is the initial concentration of Fe 2+ or Fe 3+ in solution, Ce (mg/L) is the equilibrium concentration of Fe 2+ or Fe 3+ in solution.
(1) R% = (C 0 − Ce)/C 0 * 100%  Reduction of Fe(OH) 2 and Fe(OH) 3 by NaBH 4 . For reduction experiments of Fe(OH) 2 or Fe(OH) 3 precipitates by NaBH 4, 10 mL 0.25 M Fe 2+ or Fe 3+ solution and 10 mL 1.0 M NaOH solution were mixed and then centrifuged at 3000 rpm for 5 min to collect Fe(OH) 2 or Fe(OH) 3 precipitates. The collected Fe(OH) 2 or Fe(OH) 3 precipitate was washed with deoxidizing DI water for three times to remove excessive Fe 2+ or Fe 3+ ions. Before reduction by NaBH 4, another 20 mL deoxidizing DI water was added to redisperse Fe(OH) 2 or Fe(OH) 3 precipitates. Then, 20 mL 1.0 M NaBH 4 was added dropwise under nitrogen atmosphere to reduce the precipitates. After reduction reaction, solid product was collected by centrifugation at 3000 rpm for 5 min, washed with deoxidizing DI water and ethanol, and finally dried in a vacuum oven at 60 °C for 12 h. License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.