Structural Evolution of Nanoscale Zero-Valent Iron (nZVI) in Anoxic Co2+ Solution: Interactional Performance and Mechanism

The structures of nanoscale zero-valent iron (nZVI) particles evolving during reactions, and the reactions are influenced by the evolved structures. To understand the removal process in detail, it is important to investigate the relationships between the reactions and structural evolution. Using high resolution-transmission electron microscopy (HR-TEM), typical evolved structures (sheet coprecipitation and cavity corrosion) of nZVI in anoxic Co2+ solutions were revealed. The system pH (pH measured in mixture), which controls the stability of coprecipitation and the nZVI corrosion rate, were found to be the determining factors of structural evolutions. X-ray photoelectron spectroscopy (XPS) results indicated that the formation and dissolution of sheet structure impacts on the ratio of Fe(0) on the nZVI surface and the surface Co2+ reduction. The cavity structure provides the possibility of Co migration from the surface to the bulk of nZVI, leading to continuous removal. Subacidity conditions could accelerate the evolution and improve the removal; the results of structurally controlled reactions further indicated that the removal was suspended by the sheet structure and enhanced by cavity structure. The results and discussion in this paper revealed the “structural influence” crucial for the full and dynamical understanding of nZVI reactions.


Removal Results.
The compare of long term and short term removal kinetics with 1 g/L nZVI are presented in Figure S1. The short term removal kinetics were similar to the results of Uzum, rapid removal stopped in a period of 30 min [1] . However, the long term removal results show differences. The long term removal rates achieved100%, 74.7%, 45.2% at initial Co 2+ concentration of 50 mg/L, 500 mg/L and 1000 mg/L, in a period of 10 days, respectively. Figure S1. Comparison of removal kinetics between long term and short term experiments with 1g/L nZVI.
The compare of long term and short term removal capacity with 1 g/L nZVI are presented in Figure S2. The highest capacity of long term is up to 452.2 mg/g, which is tripled to the highest capacity of short term (148.8 mg/L). Also the short term removal capacity is similar to the previous research (172 mg/g) [1] . The compare of long term and short term Fe 2+ releasing with 1 g/L nZVI are presented in Figure S3. Higher initial Co 2+ concentration resulted in more Fe 2+ releasing; Fe 2+ releasing and Co 2+ removal always have similar tendency, Fe 2+ continuously released when the long term removal lasted, and kept invariant when the rapid short term removal stopped. The excessive Fe 2+ releasing may attribute to the further reaction between Co 2+ and nZVI, as Li and Zhang reported, Fe 2+ will release when Ni 2+ reduced by nZVI [2] . Figure S3. Comparison of Fe 2+ releasing between long term to short term experiments with 1g/L nZVI.
Desorption experiments were also conducted. The samples after various removal times were desorbing at 50 ml ultra-pure water for 2 hours. As shown in Figure S4, the higher initial Co 2+ concentration may lead to more desorption, however, both the quantity and the ratio of desorption were decrease with reaction proceed, indicating more stable combination of Co 2+ and nZVI.  Figure S4. Desorption results of samples after various removal times.

Kinetics Fitting.
The pseudo first-order and pseudo second-order model were used for removal kinetics fitting. It should be noticed that it is not precise for using sorption kinetics models for this research because the reaction between nZVI and Co 2+ is not exactly sorption. However, the kinetics models could indicate the tendency of the reactions to exam whether the short term kinetics could accord with long term kinetics.
The pseudo first-order equation (Lagergren's equation) describes adsorption in solid-liquid systems based on the sorption capacity of solids [3] . It is assumed that one cobalt ion is sobbed onto one sorption site on the n-ZVI surface: Where A represents an unoccupied sorption site on the n-ZVI and k1 is the pseudo first order rate constant (h -1 ). The linear form of pseudo first order model is: Where qe and qt (mg/g) are the adsorption capacities at equilibrium and at time t (h), respectively.
Where k2 is the rate constant for pseudo second-order adsorption (g•mg −1 h −1 ) and k2qe 2 (mg•g −1 h −1 ) is the initial adsorption rate. This model assumed that one cobalt ion is sobbed onto two sorption site on the n-ZVI surface: The pseudo second-order rate expression, which has been applied for analyzing chemisorption kinetics from liquid solutions [4][5] , the linear form is: The result of Kinetic model fitting is shown in Table S1. The high r 2 values of short term pseudo second-order fitting indicating that the removal of Co 2+ by nZVI perfectly fellow a second-order sorption model in short term reaction. However, the low r 2 values of long term fitting indicates that long term removal kinetics did not exactly fellow either sorption model because the long term removal including many other reactions beside sorption. However, we could found distinct difference between short term and long term kinetics. The short term reaction kinetics did not follow the long term reaction equations; the removal amounts in short term reactions are quite higher than the values which were calculated by the long term equations: 1) theoretical removal values calculated by long term pseudo first-order equations were 1.73, 9.48, 7.312 mg/g for 50, 500, 1000 mg/L Co 2+ ；2) theoretical removal values calculated by long term pseudo second-order equations were 0.87,5.76,4.81 mg/g for 50, 500, 1000 mg/L Co 2+ ; 3) experimental values 14.88, 109.57, 134.23 mg/g. The results indicate that rapid and efficient equilibriums had been achieved in short term reactions, however then, altered by long term reactions. As discussed in manuscript Sheet Structure, the short term equilibriums were caused by the rapid formation of sheet structure and then altered by the dissolution of sheet structure in long term reaction.

Evolution Images
These additional images ( Figure S5- Figure S11) provide an integrate vision of nZVI structural evolutions in various conditions. It should be noticed that, in Figure S11, hollows on the particles in red circles were obtained by SEM, which just according with the TEM images, indicating that the cavity should be bowl shaped.
As discussed in manuscript Sheet Structure, system pH only implies the macroscopical pH influenced by both phases. Actually, solid phase may have more influence to the interface, so sheet structure could still be slightly determined at very few part of nZVI at higher magnification ( Figure S12).

EDS Analysis
The results of EDS analysis is shown in Figure S16. The ratio of Fe to O on core structure is 2:1, indicating that the ratio of Fe metal to Fe oxides may be 5:1. And the ratio of Fe to Co to O on cavity structure is 3.5:1:1.5 also showing a dominance of zero valent metal. The ratio of metal to O on sheet structure is approximate to 1:2, indicating a dominance of metal hydroxides or metal oxides. However, Co could be hardly determined on sphere structure, which indicated that sphere was an iron oxide or hydroxide. It should be noticed that the actual quantity of atoms cannot be reflect by the EDS analysis due to the material density of different parts maybe different. That is, Co atom enrichment may be quite different although the ratio of Co on sheet and cavity is approximate.

XPS analysis
The results of XPS analysis of Fe 2p of nZVI reacted with 1000 mg/L Co 2+ at different times are presented in Figure S17. As the results shown, a small quantity of zero valent iron with the Fe 2p3/2 line centered at 707.0 ± 0.1 eV binding energy and the Fe 2p 1/2 line centered at 720.1 ± 0.1 eV and a large quantity of ferric iron with the Fe 2p3/2 line centered at 711.6 ± 0.1 eV binding energy and the Fe 2p 1/2 line centered at 724.6 ± 0.1 eV could be determined on fresh nZVI surface due to the covering FeOOH shell structure leading few exposure of zero valent iron [6][7][8][9] . Accordingly, zero valent iron could be scarcely determined on the sheet wrapped structure which formed after 1 hour reacting. However, the quantity of the zero valent iron on particle surface obviously increased with the diminishment of the sheet structures. After 10 days reacting, the proportions of zero valent iron on the pure cavity structures even Elements surpassed the fresh nZVI. That is, the surface zero valent iron was re-exposed with the dissolution of the sheet structures. Figure S17. XPS spectra of Fe 2p of nZVI reacting with 1000 mg/L Co 2+ at different times.
The results of XPS spectra of Co 2p 3/2 of nZVI surface are presented in Figure S18: only Co (II) can be determined on nZVI surface after 1 hour, which was in accordance with Uzum [1] ; about 10% Co (II) was reduced by nZVI to Co (0) after 5 days reacting, and finally about 52% Co (0) was formed after 10 days reacting, the photoelectron peaks at 786.2, 782.6 and 778.1 eV are assigned to the 2p 3/2 binding energies of Fe auger, Co (II) and Co (0), respectively [11][12][13] . Obviously, a further reduction could be determined by XPS analysis. The association between the portion of zero valent iron and the reduction of Co (II) on nZVI surface could be prefect established, which means the quantity of zero valent iron on nZVI surface played a decisive role to the reduction of Co (II).  Figure S18. XPS spectra of Co 2p 3/2 of nZVI surface reacting with 1000 mg/L Co 2+ at different times.
The results of XPS spectra of O 1s of nZVI surface are presented in Figure S19. Oxygen formed H2O with the O 1s line centered at 532.9±0.1 eV binding energy [14] , formed both iron and cobalt oxides with the O 1s line centered at 530.0±0.1 eV binding energy, formed FeOOH with the O 1s line centered at 531.4±0.1 eV binding energy [6][7][8][9] . However, the peak centered at 532.0±0.1 eV of this study cannot be determined by pervious researches, it may be the peak of Fe / Co coprecipitation due to the peak excursion.
On fresh nZVI, FeOOH and iron oxides may be the dominance, then converted to the Fe / Co coprecipitation and oxides due to the formation of the sheet structure, finally back to FeOOH due to the dissolution of the sheet structure. On the other hand, the ratio of OHand O 2to H2O decreased with the reaction proceed indicating an increase of zero valent metal, which is according with the results of XPS analysis of Fe and Co.

Precipitation pH
The theoretical pH leading the dissolution of the precipitation was deduced by the solubility product as equation (5) (6), supposing all precipitation was bivalent metal hydroxide as a simplification:  respectively. The Ksp of Fe(OH)2 and Co(OH)2 is 4.87×10 -17 and 5.92×10 -15 at 25 ℃, respectively [10] .
The theoretical pH leading the dissolution of the Fe(OH)2 and Co(OH)2 at different times are calculated in Table S2.

System pH Controlled Experiments
The TEM images of structural evolution of nZVI in system pH controlled removal are presented as follow. The TEM images of structural evolution of nZVI in system pH controlled removal of 50 mg/L Co 2+ are present in Figure S20. In 50 mg/L Co 2+ , the sheet dissolved in pH of 6.5 was earlier than which in pH of 7.5 ( Figure S20 b and c). The TEM images of structural evolution of nZVI in system pH at 7.6-7.8 of 1000 mg/L Co 2+ are present in Figure S21. The TEM images of structural evolution of nZVI in system pH at 6.5-6.7 of 1000 mg/L Co 2+ are present in Figure S22. As Figure S21 shown, the sheet structure would grow bigger with the removal process in high system pH and high initial Co 2+ concentration. As shown in Figure S22, the sheet structure may rapidly dissolved and form cavity structure in a lower system pH, even it would take several days' evolution in pH uncontrolled system.   The XPS analysis results of nZVI reacted 3 hours with 1000 mg/L Co 2+ at different system pH are shown in Figure S23. In high system pH, Co mainly formed divalent, however, about 33% was reduced to zero valent in low system pH which achieved after 8 days in uncontrolled system.  Figure S23. XPS spectra of Co 2p 3/2 of nZVI surface reacting with 1000 mg/L Co 2 + for 3hours at different system pH.

Structure Pre-control
The results of acid-pretreatment and base-pretreatment is present in Figure S24 a and i, cavity structure and sheet wrapping structure was perfectly obtained. And these pretreated particles were immediately used for further reaction. The structural evolution of pretreated particles in 1000 mg/L Co 2+ were presented in Figure S24. The pretreatment change the stage of structural evolution: the structure of acid-pretreatment after 3 days' reaction is similar to the structure of no pretreated after 5 days' reaction, also similar to the structure of base-pretreatment after 8 days' reaction; the structure of acid-pretreatment after 5 days' reaction is similar to the structure of no pretreated after 10 days' reaction; the structure of no pretreated after 3 days' reaction is similar to the structure of base-pretreatment after 5 days' reaction. Figure S24. TEM images of the structural evolution of nZVI in reaction of 1000 mg/L Co 2+ with acid-pretreatment (a-d), no pretreatment (e-h) and base-pretreatment (i-l).

XRD analysis and discussion
XRD analysis was carried out on a Bruker D8Advance X-ray diffraction instrument (Cu Kα), the diffraction angle (2θ) from 10 to 90° was scanned. The samples were centrifuged and then vacuum drying at 40℃ or freezing drying at -10℃ for XRD analysis. The XRD analysis results are shown in Figure S25.
The peak reveals the existence of iron on the basis of the Jade pdf-# 65-4899(Fe), the peaks at the 2θ of 44.9°, 64.9°, 82.2° indicated the presence of iron. However, the results were significantly influenced by the pretreatment of samples in this research, XRD analysis to the same sample with vacuum drying or freezing drying was different: no obvious peak could be found when he samples were freezing dried ( Figure S25a); peaks could be distinct when he samples were vacuum drying at 40℃ ( Figure S25b).
Therefore, the prepared nZVI presented an amorphous phase of iron, and would not change to crystal in reactions. And heating of nZVI would change the amorphous phase iron to crystal as shown in Figure   S25b.