Figure 1A shows the variation of the coupling substances of OC (here referred to as A-HA) and dissolved Fe (hydr)oxides at different A-HA concentrations and temperatures. The variations in the colors of Fe (hydr)oxides are considered to reflect the interfacial reaction between A-HA and iron, even as a basis for a preliminary assessment of the mineral phases by visual observation (Munsell Color Classification System, Tables S1 and S2), such as yellow for goethite, blood red for hematite, and reddish-brown for ferrihydrite. The presence of Fe, O, and C elements in Fe (hydr)oxides further validates the similarity with natural OC–mineral dynamic processes (Fig. 1B). The Fe and C elements are uniformly distributed, as well as present with substantial overlap [19]. As depicted in Fig. 1C, the C content in Fe (hydr)oxides without A-HA ranged from 0.08% to 0.27%. After adding A-HA, C content increased to 2.89%–3.87% (2 g/l A-HA), 7.63%–9.18% (5 g/l A-HA), and 13.78%–16.32% (10 g/l A-HA). This suggests that more A-HA can be sequestered by Fe (hydr)oxides with higher concentrations of A-HA, providing direct mineral protection through adsorption, complexation, coprecipitation, and intercalation complex formation. Increasing temperatures also positively affected carbon fixation in neutral reaction systems, which is necessary to enhance the activity and rate of the reaction system [20].

Thermogravimetric mass loss was noted over the range of temperatures corresponding to the initial A-HA concentration in the adsorption and coprecipitation system (Fig. 1D). Further differential thermogravimetric and differential thermal analyses were conducted, with each peak representing the mass loss of a specific molecule or component (Fig. S1) [21]. Thermogravimetric data shows that the total weight loss visibly increases with the increase of coprecipitated A-HA concentrations (0, 2, 5, and 10 g/l) between 550 and 650 °C, ranging from 23.4% to 35.8%, 36.6%, and 53.5%, which was attributed to decomposition of recalcitrant carbon, such as aromatic compounds of lignin or polyphenols or polycondensed aromatic carbons [22]. Besides, differential thermogravimetric peak maxima are slightly shifted to lower temperatures (Fig. S1). In accordance with the research from Sodano et al., lower temperatures is mainly attributed to the presence of the A-HA/Fe (hydr)oxide, although the catalytic effect of the Fe on the decarboxylation reaction cannot be excluded [23,24]. Coprecipitation of A-HA to Fe (hydr)oxide showed strong selective retention of aromatic components, and initial complexation of Fe ions by aromatic carboxylic moieties and precipitation as carbon-rich A-HA/Fe (hydr)oxide associations contributed to the total carbon retention, especially at higher A-HA solution [23].

As shown in Fig. 1E, the addition of A-HA results in a considerable increase of Fe_p (organically complexed Fe oxides) contents in A-HA/Fe (hydr)oxide of 0.841 ± 0.094 to 2.203 ± 0.747, 1.167 ± 0.061 to 3.142 ± 0.754, and 1.743 ± 0.072 to 5.337 ± 0.687 mg/l with 2 to 10 g A-HA/l, as compared to that of Fe (hydr)oxide (0.591 ± 0.055 to 1.741 ± 0.195 mg/l). Meanwhile, there is no marked correlation between Fe_d (total reactive Fe oxides) and A-HA (Fig. 1F), supporting the hypothesis that A-HA/Fe (hydr)oxide composites can stabilize more “new carbon”. There are marked negative correlations of Fe_HCl and TOC_HCl (Fig. 1G and H), suggesting that the higher amounts of the A-HA associated OM and the precipitated insoluble Fe(III)−organic complexes are to be expected in which coprecipitation occurs at high C/Fe ratios. This is closely related to the chemically stabilized of Fe-bound A-HA functional groups, and precipitated insoluble A-HA/Fe complexes represent an apparently overlooked terrestrial carbon pool, and act as an “iron gate” and “carbon latch” in regulating OC dynamics [25–27].

Based on the above results, it can be predicted that A-HA may play an essential function in mediating carbon stability processes and promoting the formation of organo-metallic complexes, apart from acting as the carbon source. This supports the notion that Fe (hydr)oxides forms in solution and that organic ligands may be associated with iron (complexation) or present on the surface of nuclei or crystals. Thus, the binding energies of Fe2p1/2 and Fe2p3/2 in “organically complexed iron” was found to markedly decrease by 0.3 to 724.7/711.0 eV and by 0.7 to 724.5/710.8 eV under different aging crystallization conditions (pH = 7, T = 50 or 75 °C, and C_A-HA = 10 g/l) (Fig. 2A). The decreased binding energies could not be clearly explained by the electronegativity difference between A-HA and iron [28]. Instead, it was inferred that the Fe center of Fe (hydr)oxides undergoes a phase transition upon complexation and other reactions with A-HA [29]. X-ray diffraction (XRD) and Fourier transform infrared spectroscopy (FTIR) results confirmed that the introduction of A-HAs facilitated the crystallization of more stable Fe(III) crystalline forms such as goethite and hematite. In Figs. S2 and S3, 2 broad peaks of the XRD pattern appeared when the reaction temperature was adjusted to 50 °C. These peaks, combined with the characteristic peaks, indicate the presence of 2-line ferrihydrite. Upon the addition of a definite amount of 10 g/l A-HA, the main XRD diffraction is ascribed to hematite (PDF#98-000-0240), goethite (PDF#98-000-0229), and 2-line ferrihydrite. The FTIR absorption peaks are attributed to goethite bending vibrations and the hematite Fe-O absorption peaks. By analyzing fitted peak areas from x-ray photoelectron spectroscopy (XPS) data, the relative proportions of surface species were quantified. Deconvolution of the XPS C 1s profile shows the presence of carbon-related functional groups, such as C-C/C-H, aliphatic carbon (284.80 eV), O-C-O, aromatic C (286.11 to 286.66 eV), and O-C=O, carboxylic C (288.50 to 288.85 eV) (Fig. 2B and Table S3) [30,31]. The binding energies of the O-C-O and O-C=O groups in the A-HA/Fe coprecipitates shifted to lower values, and their proportions decreased markedly. This shift reveals the formation of precipitate and bridging complexes, which alters the electronic environment of the carbon atoms. Due to the greater electronegativity of hydrogen, fewer valence electrons of the carbon in the group are abstracted by iron after complexation, and the nuclear electrons experience a lower binding energy [32]. This result also suggests that aromatic C and carboxylic C groups interact with Fe (hydr)oxides through interfacial C-O-Fe bonds, which is in line with the FTIR results [31]. The increased or decreased intensity at ca. 1,716, 1,375 cm⁻¹ provided evidence for the formation of complexes between A-HA and Fe (hydr)oxides (Fig. S3) [33,34].

Results of high-angle annular dark-field scanning transmission electron microscopy (STEM) surface scan at the nanoscale indicated that Fe, C, and O mass fraction in the complexing compound are (67.86 ± 11.31)%, (11.15 ± 1.30)%, and (37.97 ± 4.73)% (Fig. 2C). Substantially different spatial distribution patterns were observed for Fe and C elements (Fig. 2D and E), implying that the spatial heterogeneity of the Fe is associated with the C. In order to characterize the spatial heterogeneity and chemical features of organic C species at organo-mineral interfaces, EELS line scans of the spatial distribution of organic compounds near the surfaces of A-HA₁₀/Fe-50 were performed as a function of distance [35] (Fig. 2F). The EELS line scan analysis suggests that the spatial distribution of C species is rather heterogeneous, with the C signal peaking close to the outer layer. In addition, similar to those in previous studies, distribution patterns of OC in the porous and sparse areas of Fe (hydr)oxides were also observed, which indirectly demonstrates the important role of nanopores or mineral interstitials in carbon sequestration [36,37]. To further investigate the spatial heterogeneity of oxidized and reduced carbon in A-HA₁₀/Fe (hydr)oxide-50, we probed 3 specific areas to compare their H/L ratios [38]. Specifically, the A-HA₁₀/Fe (hydr)oxide-50 interfacial region can be divided into 3 areas, the outer layer (red line, area 1), the intermediate layer (blue line, area 2), and the inner layer (green line, area 3), as denoted in Fig. 2F. The H/L ratios for area 2 and area 3 are similar, with H/L ratios ranging from 1.81 to 1.85, and slightly higher than that of area 3 (with an H/L ratio of 1.75). From the outer layer to the inner layer of the A-HA₁₀/Fe (hydr)oxide-50 interface, the relative abundance of oxidized carbon increases and the relative abundance of reduced carbon decreases. This pattern suggests that carboxyl C is the most dominant organic C species in the inner layer, while aromatic C is mainly attached to the outer layer of the organic-mineral interface [35,38,39].

As shown in Fig. 3A and B, in a system where the color developer coexists with the A-HA/Fe (hydr)oxide complex, the processes of complex formation, rapid reduction of solid Fe to Fe²⁺ (510 nm), and maintenance of dynamic desorption can be observed in situ. The shape of the ultraviolet-visible absorption spectra increases linearly with wavelength in the range from 360 to 400 nm of A-HA/Fe (hydr)oxide precipitates (Fig. 3B). The emergence of these bands is postulated to be a result of modifications in interchromophore interactions and conformations within the A-HA molecules, which are caused by the accumulation of electric charge and/or other nonspecific effects related to deprotonation and Fe binding [40]. Moore et al. proposed that simple organic molecules can be geopolymerized (here refers to A-HA/Fe (hydr)oxide) into recalcitrant forms by the Maillard reaction [41]. As illustrated in Fig. 3C, we similarly attribute the catalytic effect of dissolved iron to a complexation mechanism to cation bridging. This creates a basis for the generation of more favorable free energy reactions. Firstly, the adsorption effect facilitates the clusters and orients of the reactants on the Fe (hydr)oxide surface, while the redox reaction between A-HA and the Fe (hydr)oxide generates dissolved Fe(II), which creates a bridging effect to form A-HA/Fe (hydr)oxide. Finally, dissolved Fe(II) may reoxidize and precipitate to A-HA/Fe (hydr)oxide for further catalytic reactions.

Figure 3E and Fig. S4 show the characteristic fluorescence spectra of the samples with the 4 fluorescent components identified by PARAFAC analysis. Among the transitions associated with fluorescence, quinones are characterized by excitation maxima in the range of 250 to 300 nm and distinct excitation peaks in the range of 330 to 400 nm due to the π→π* transitions of quinones and benzenes [42,43]. The excitation wavelength of the fluorescence component 3 of samples exactly matches that of the quinone component. Field and laboratory studies have shown that the quinone-like component is considered to be the most important redox-active functional group in humic substances, i.e. the electron chelating ability of A-HA/Fe (hydr)oxide is attributed to the quinone molecule of A-HA. The quinone fraction in (2 to 10) g/l A-HA accounted for 25.82% to 29.22% of the fluorescence of component 3. With the formation of the A-HA/Fe complex, the quinone fraction in the reaction solution decreases to (6.23 to 0)% and (9.82 to 0)% after 5 min and 7 d. The quinone-like components in the desorbed solution of A-HA/Fe complex accounted for 20.38% (10 g/l A-HA), 19.12% (5 g/l A-HA), and 0% (2 g/l A-HA) (2 g/l A-HA) of the fluorescence of component 3, respectively. That is, biomass humification provides a heterogeneous pool of quinones that produces “quinone boosting”, a high concentration of A-HA with a high quinone content and electron accepting capacity that achieves the reduction of Fe(III) to Fe(II). As shown in Fig. 3D, The Fe²⁺ content in A-HA/Fe (hydr)oxide was maintained at 12.076 ± 0.731 to 9.929 ± 0.263%, which was 84.250 to 6.595 times higher than the Fe²⁺ content in Fe (hydr)oxide. Our findings indicate that most of the reduced Fe remained in the solid phase as easily reoxidizable authigenic A-HA/Fe²⁺ (hydr)oxide. The solid-phase Fe³⁺/Fe²⁺ (hydr)oxide can serve as an immobile and rechargeable “redox battery”. The Fe “redox battery” is a characteristic feature of permeable environments or sediments with varying redox conditions, making Fe a key determinant of carbon turnover [44].

The crystalline phases of A-HA/Fe (hydr)oxides incubated with neutral pH, 4 °C/25 °C/50 °C/75 °C temperatures, and varying A-HA concentrations (0 to 10 mg/l A-HA) were characterized with XRD patterns (Fig. S2). Notably, A-HA particularly influenced the crystallization of Fe (hydr)oxides. As shown in Fig. 4A and B, the initial ferrihydrite mainly appeared as dispersed primary particles (Fe (hydr)oxides system, T = 50 °C). After A-HA introduction, primary particles aggregated to form agglomerates, displaying needle-like, rounded clusters, and dispersed particles, which indicated the presence of oriented assembly (OA) in the crystallization process. A-HA induced A-HA/Fe(III) coprecipitates consisted of 2-line ferrihydrite, rounded cluster-like hematite, and needle-like goethite/hematite. When the aging temperature increased to 75 °C (Fig. 4C and D), coprecipitates even comprised pure hematite monomorph (50 to 100 nm). This demonstrates specificity of A-HA in allowing regulation of Fe (hydr)oxides. As mentioned earlier, the addition of A-HAs facilitated the crystallization of more stable Fe crystalline forms, such as goethite and hematite, which was accurately confirmed through room temperature Mössbauer data (Fig. 4E to H and Table S4) [45]. Mössbauer spectra allow us to characterize the crystallinity-continuum of Fe (hydr)oxides solid phases. The pathway of A-HA-regulated Fe (hydr)oxides was accurately determined, showing the crystallization progression from ferrihydrite (100%) to mixed phases of ferrihydrite (15.9%), goethite (14.3%), and hematite (44.4%), with a substantial increase in hematite fraction (60.7%) during crystallization, ultimately almost completely transitioning to hematite (94.7%). A-HA facilitated Fe (hydr)oxides transformation from superparamagnetic to magnetically ordered state [46]. The QS values of Fe (hydr)oxides varied with increasing C/Fe, indicating that A-HA affected the spatial coordination of central Fe²⁺/Fe³⁺ and octahedral structure around the central Fe atoms, which is essential for surface properties, morphology, and crystallization of Fe (hydr)oxides [47]. The Mössbauer spectra of the A-HA/ Fe (hydr)oxides yielded sextets, indicating electron transfer reactions occurred between the Fe²⁺ and native solution Fe³⁺ atoms during the 50 °C anoxic incubation [48]. Our results proved that coexisting A-HA/Fe (hydr)oxide precipitate promotes formation of more goethite or hematite via Fe(II)-precipitates electron transfer and template-directed nucleation and growth [49].

The XRD result shows that, anthropogenic soil fraction was mainly SiO₂ (Fig. S5), considering the low binding capacity of A-HA to quartz minerals (Fig. 5A). Across over long timescales, Fe (hydr)oxide can strongly affect on C levels through a variety of mechanisms, including adsorption of large amounts of dissolved OC (Fig. 5A), maintenance of OC stability, and decomposition and synthesis of complex OC [50]. Besides, colloidal Fe (hydr)oxide or A-HA-Fe (hydr)oxide associations is important building unit of the aggregates system, actively promoting the formation and stabilization of soil aggregates (Fig. 5B) [7].

As shown in Fig. 5C, Fe (hydr)oxide acts as a flocculant for the quartz sand and A-HA, and the microstructure (A-HA/Fe (hydr)oxide) is precipitated on the surface of the quartz sand as a “binder”, acting as “bridges”. This suggests that the microstructure-induced aggregate formation can connect particles of all sizes and underlines that also sand grains can be found within aggregates, contributing to larger aggregate sizes [51]. The cementing by A-HA/Fe (hydr)oxide can increase the diameters of the aggregates and the microscopic pore structure by connecting and gluing more subunits (Fig. 5B, D, and E). Asano and Wagai [52] demonstrated that micron/submicron-sized aggregates are formed from “presumed building blocks” consisting of organomineral-metal as effective binding agents, at least in siliceous soils. After 3 to 15 d of anoxic incubation, the aggregate structure of sterilization soil in related to the distribution of different soil aggregate sizes was altered obviously by the addition of A-HA (Fig. 5G). For the formation of larger microaggregates, A-HA as a gluing agent and more Fe (hydr)oxides as cementing agents are more essential to glue together building units composed of smaller microaggregates [9]. The percentage of micro-aggregates (< 0.25 mm) was greatly reduced, which decreased from 28.38% (AS0-15d) in the control group to 25.94% (AS2-15d), 11.69% (AS5-15d), and 10.23% (AS10-15d). Correspondingly, the proportion of >2 mm was increased markedly from 53.16 ± 6.21% in the control group to 63.06 ± 5.07%, 84.28 ± 6.03%, and 84.96 ± 7.83% for the same treatments. As shown in Fig. 5F, in consistent with the generally proven fact that incorporation into organomineral complexes and occlusion within microaggregates play a critical role in OC stability against biodegradation [38,50,53].

All chemicals used in this study were of ACS-grade quality. High purity water (with resistivity > 18 MΩ·cm) from a Millipore Milli-Q purification system was used to prepare solutions. Iron nitrate nonahydrate (Fe(NO₃)₃·9H₂O), hydroxylamine hydrochloride, sodium polyphosphate decahydrate, ammonium oxalate, dithionite, ammonium fluoride, 1,10-phenanthroline, acetic acid, sodium acetate, potassium hydroxide (KOH), and hydrochloric acid (HCl) were acquired from Sinopharm Chemical Reagent Co., Ltd. The pure sand particles were quartz with a particle size of <0.0450 mm was obtained from Tianjin Damao Chemical Reagent Co., Ltd.

A-HA was prepared using the hydrothermal humification method from our previous studies [59,60], with details provided in Text S1. Then, 1 g of A-HA was dissolved in 1 l of 6% ammonia solution to create a 10 g/l A-HA stock solution, which was later diluted to the desired concentration using high-purity water. Fe (hydr)oxides were synthesized based on the modified process from Schwertmann and Cornell [61,62], and the synthesis was repeated using various A-HA concentrations and different aging conditions. Specifically, 2 g of Fe(NO₃)₃·9H₂O were dissolved in 20 ml of high purity water, and the pH was adjusted to approximately 8 using 0.1 to 1 mol l⁻¹ NaOH or HCl to form flocculent substances. Then, 14 ml of 0, 2, 5, and 10 g/l A-HA solution was added, and the pH was further adjusted to 7 ± 0.05. The mixtures were aged for 7 d in water baths set at 4 ± 0.5, 25 ± 0.5, 50 ± 0.5, and 75 ± 0.5 °C, after vigorous shaking, abbreviated as A-HA_X/Fe (hydr)oxide-Y (X = 0 to 10 g/l A-HA, Y = 4 to 75 °C). The temperature settings were obtained from Chen and Das et al. to increase the reaction rate of the system, validate experimental results over a wide temperature range, and reduce the interference of thermal transformations in the system [20,63]. The detailed experimental design can be found in Table S1. Anthropogenic soil (AS) was prepared by mixing 20 g the pure sand with the prepared A-HA/Fe (hydr)oxides, abbreviated as AS_X –Y (X = 0 to 10 g/l A-HA, Y = 3 to 15 d anthropogenic maturation). It should be noted that the construction of anthropogenic soil was carried out under aseptic conditions. Both the ferric ion and A-HA reaction systems were protected by nitrogen. Samples were sterilized in autoclaves before sampling and subsequent testing.

TOC was assessed using a TOC analyzer (TOC-L CPN). The mineralogy of Fe (hydr)oxides and A-HA/Fe (hydr)oxides coprecipitates phase was characterized by XRD (Miniflex, Rigaku), and peak identification was performed using Jade 9.0 software. The molecular structures of Fe (hydr)oxides were analyzed by FTIR (Nicolet 460, Thermo Fisher). Particle morphology and size were characterized by TEM (HT7800, HITACHI). XPS (Escalab 250Xi, Thermo Fisher) was applied to measure the chemical composition on sample surfaces, and the data were analyzed using XPSPEAK software. The thermal stability of Fe (hydr)oxides was investigated through TGA (TGA8000, PerkinElmer) under a nitrogen atmosphere, heating from 25 to 800 °C at a rate of 20 °C/min. The ⁵⁷Fe Mössbauer spectra of Fe (hydr)oxide was recorded on an SEE Co W304 Mössbauer spectrometer, using a ⁵⁷Co/Rh source in transmission geometry. A fluorescence spectrometer (Shimadzu F-7000, Hitachi) was employed with a 150-W xenon lamp for the acquisition of EEM fluorescence spectrum. Direct visualization of the interfacial behavior between A-HA and Fe (hydr)oxide at the (sub)nanoscale using an advanced Cs-STEM [38]. Particle size distribution was obtained using the laser diffraction method on the Mastersizer 2000 with the Hydro G attachment. The quantitative analyses methods for Fe(III) and Fe(II) were referred to Sun, Jeewani, and Liao et al [50,54,64]. Detailed procedures for these instrumentation methods can be found in the Text S2. Real-time measurement of Fe(II) was performed using the modified 1,10-phenanthroline method. Fluoride was added to remove interference from Fe(III) in aqueous solution. The color developer was added to a closed nitrogen-filled reaction vessel of A-HA/Fe. A blank sample, consisting of the color developer and the reaction solution without A-HA, was used to reduce error.

To determine the effects of A-HA/Fe (hydr)oxides on aggregates stability, ultrapure water was used to rapidly wet the AS aggregates. The following steps were taken: 20 g of AS samples were gently immersed in ultrapure water for 15 min, and excess liquid was then removed. The completely moistened AS was then transferred to sieves with pore sizes of 0.054, 0.15, 0.25, 0.5, 1, and 2 mm to separate aggregate fragments through wet sieving. The resistant soils on each sieve were recovered, dried at 60 °C, and weighed to obtain aggregate fractions of <0.054 mm, 0.054 to 0.15 mm, 0.15 to 0.55 mm, 0.25 to 0.5 mm, 0.5 to 1 mm, 1 to 2 mm, and >2 mm. All batches were performed in triplicate for further data analysis (unless otherwise indicated).