Versatile Functionalization of Carbon Nanomaterials by Ferrate(VI)

Various forms of carbon nanomaterials are selected as substrates to clear the mist in understanding the reactivity/utility of ferrate(VI) in oxidizing carbon nanomaterials. It unravels a modest reactivity of ferrate(VI) in liquid phase that only oxidizes the active defects on carbon surface and a powerful oxidizing ability in solid state that can open the inert C=C bonds in carbon lattice. Respective benefit and limitation of the wet and dry approaches using ferrate(VI) in functionalizing carbon nanomaterials are discussed. Various forms of carbon nanomaterials are selected as substrates to clear the mist in understanding the reactivity/utility of ferrate(VI) in oxidizing carbon nanomaterials. It unravels a modest reactivity of ferrate(VI) in liquid phase that only oxidizes the active defects on carbon surface and a powerful oxidizing ability in solid state that can open the inert C=C bonds in carbon lattice. Respective benefit and limitation of the wet and dry approaches using ferrate(VI) in functionalizing carbon nanomaterials are discussed. As a high-valent iron compound with Fe in the highest accessible oxidation state, ferrate(VI) brings unique opportunities for a number of areas where chemical oxidation is essential. Recently, it is emerging as a novel oxidizing agent for materials chemistry, especially for the oxidation of carbon materials. However, the reported reactivity in liquid phase (H2SO4 medium) is confusing, which ranges from aggressive to moderate, and even incompetent. Meanwhile, the solid-state reactivity underlying the “dry” chemistry of ferrate(VI) remains poorly understood. Herein, we scrutinize the reactivity of K2FeO4 using fullerene C60 and various nanocarbons as substrates. The results unravel a modest reactivity in liquid phase that only oxidizes the active defects on carbon surface and a powerful oxidizing ability in solid state that can open the inert C=C bonds in carbon lattice. We also discuss respective benefit and limitation of the wet and dry approaches. Our work provides a rational understanding on the oxidizing ability of ferrate(VI) and can guide its application in functionalization/transformation of carbons and also other kinds of materials.

Recent years have witnessed an emerging role of ferrate(VI) in materials science [30][31][32][33][34], where ferrate(VI)-enabled oxidative functionalization/transformation of a target material is a key step toward functional applications. Of particular interest is the oxidation treatment of carbon materials [35][36][37][38][39][40][41]. Peng et al. [35] reported the first example of applying K 2 FeO 4 for the oxidation/ exfoliation of graphite in concentrated H 2 SO 4 . Highly water-soluble single-layer graphene oxide was produced in 1 h at room temperature, which indicated the extraordinary oxidizing ability of K 2 FeO 4 in H 2 SO 4 medium. But soon after, Sofer and coworkers [36] provided completely opposite results, showing such liquid-phase oxidation was impossible for graphite, attributed to the extreme instability of K 2 FeO 4 in acidic environments. Nevertheless, some other studies suggested the moderate oxidation effects of K 2 FeO 4 by the production of graphite/graphene oxides with relatively low degrees of oxidation [37,38] and by the result of nondestructive oxidation of carbon nanotubes (CNTs) [39]. Therefore, the reactivity of K 2 FeO 4 in liquid phase (i.e., H 2 SO 4 medium) remains confusing in oxidizing carbon materials, which ranges from aggressive to moderate, and even incompetent (see Table S1 for a comparison of the literature results).
Oxidations by ferrate(VI) under solvent-free condition provide an alternative and green way for harnessing its oxidizing power. Very recently, our group discovered that the intrinsic high reactivity of K 2 FeO 4 was accessible in the "dry" solid state by oxidizing small-molecule substrates [42]. On this basis, K 2 FeO 4 solids were applied for the oxidation of CNTs under mechanical milling, and effective surface functionalization was achieved. However, the dry chemistry of ferrate(VI) is still poorly understood and the mechanism of K 2 FeO 4 oxidation on carbon materials remains unclear. A critical question is whether K 2 FeO 4 is reactive enough to open the inert C=C bonds in carbon lattices.
Herein, we scrutinize the reactivity of ferrate(VI) in liquid phase and solid state using various forms of carbon materials as substrates, as depicted in Fig. 1. Fullerene C 60 is selected as a probe molecule, and four typical nanocarbons with diverse physical structures and different defect degrees are further used to test the reactivity. Our systematic results provide a rational understanding on the performance of this attractive oxidizer in materials chemistry.

Preparation of K 2 FeO 4
The purity of K 2 FeO 4 should be examined before it is used for oxidizing carbon materials. The commercially supplied K 2 FeO 4 products from several manufactures only have actual purities of 20% or below, although the declared purities are > 90%. Therefore, we synthesized K 2 FeO 4 according to the literature [43] and purified them by recrystallization as described in our previous work [42]. Preparation details were provided in Supporting Information (SI).

Liquid-Phase K 2 FeO 4 Oxidation
Oxidation of C 60 . 30 mg of C 60 was slowly added to 20.0 mL of sulfuric acid (95-98%) in a 50-mL two-necked flask under argon atmosphere, and the dispersion was sonicated for 30 min. Then 2.5 g of K 2 FeO 4 was slowly added to the flask under argon flow at 0 °C, and the reaction mixture was stirred at 60 °C for 12 h. The resulting dispersion was diluted in 500 mL of cold water and stirred for 30 min.

3
The solid was obtained by centrifugation (12,000 rpm), followed by washing in sequence with 2 M HCl (several times to remove Fe 3+ ), ultrapure water (18.2 MΩ cm) and ethanol. The product was finally dried at 60 °C in a vacuum oven.
Liquid-phase oxidation treatments of other carbon materials were described in SI.

Solid-State K 2 FeO 4 Oxidation
Oxidation of C 60 . 100 mg of C 60 and 2.5 g of K 2 FeO 4 were mixed together by brief grinding in an agate mortar. The mixture was then introduced into a 50-mL stainless milling jar together with 26 g of 5-mm-diameter stainless steel balls (ball-to-powder weight ratio 10:1). Ball milling was performed at a rotational speed of 250 rpm for 6 h or 180 rpm for 2 h in a horizontal planetary ball milling (WXQM-2L, Tecan Powder). The jar was opened every 30 min to break up the mixture materials if they were agglomerated or adhered to the sidewall during milling process. The obtained solid was washed by centrifugation (12,000 rpm) in sequence with 2 M HCl (several times to remove Fe 3+ ), ultrapure water (18.2 MΩ cm) and ethanol. The product was finally dried at 60 °C in a vacuum oven.
Solid-state oxidation treatments of other carbon materials were described in SI.

Characterization
The purity of K 2 FeO 4 sample was tested by spectrophotometry, X-ray diffraction (XRD) and 57 Fe Mössbauer spectroscopy ( 57 Co(Pd) source). Chemical structure of C 60 samples was determined using matrix-assisted laser desorption/ionization Fourier transform ion cyclotron resonance mass spectrum (MALDI-FTICR MS) and 1 H NMR (400 MHz). Oxidation degrees of carbon materials were analyzed using X-ray photoelectron spectroscopy (XPS, binding energies were calibrated with respect to C 1s peak at 284.6 eV) and thermogravimetry (TG, 10 °C min −1 , N 2 ). Defect degree and morphology were characterized by Raman spectra, scanning electron microscopy and transmission electron microscopy (SEM and TEM). Details of instruments and test conditions were described in SI. . c Structural models of the carbon nanomaterials used in this study: C 60 , defective CNT (DCNT), graphitized CNT (GCNT), defective and graphitized carbon nanofibers (DCNF and GCNF). I D /I G refers to the relative intensity of D to G band in Raman spectrum, taken as a measure of defect degree

Purity Analysis of K 2 FeO 4
K 2 FeO 4 oxidizer used in our experiments has a purity of 95%, as determined by spectrophotometry. XRD pattern confirmed single-phase character of the solid (Fig. S1), and the 57 Fe Mössbauer spectrum proved 97.6% relative content of ferrate(VI) (Fig. S2, Table S2).

Probing the Reactivity Using C 60 Molecules
We first probed the reactivity of K 2 FeO 4 using C 60 . As a special kind of carbon material with defined molecular structure, C 60 can provide directive and reliable assessment results. The spherical cage of C 60 confers an excess of strain to C=C bonds, inducing a unique sp 2.28 hybridization of the carbon atoms with a pyramidalization angle θ P of 11.6° [44,45]. This endows C 60 with a moderate reactivity: inerter than that of the sp 3 -C atoms (θ P = 19.5°) appearing as defects on carbon surface and more active than that of the sp 2 -C atoms with smaller θ P s in graphite (θ P = 0°) and most other carbon materials (e.g., CNTs and CNFs) [46]. In addition, the smallmolecule property of C 60 allows its product structure to be easily determined by standard organic analytical methods such as mass and NMR spectroscopy.
After the liquid-phase oxidation, C 60 products showed scarcely any changes with respect to the pristine sample, even with a large excess of K 2 FeO 4 (ca. 300 mol equivalent) and long reaction time (up to 12 h). They displayed black color, poor water dispersibility and good solubility in toluene (Fig. 2b inset). The product structure was determined by MALDI-FTICR mass spectrometry (MS). Except from the prominent peak at m/z 720.000 (intact C 60 ions), no newly produced ion peaks were observed (Fig. S3), which unambiguously showed K 2 FeO 4 in liquid phase was not reactive enough to oxidize C 60 .
In contrast, K 2 FeO 4 efficiently oxidized C 60 in the "dry" way. By only 1 ~ 2-h treatment, the majority of products became hydrophilic, yielding a thick dark-brown aqueous layer after phase separation with toluene ( Fig. 2b inset), which was ascribed to the generation of oxygen-containing groups. With prolonged reaction time, the purple toluene layer faded and the aqueous phase gradually became yellowish transparent, indicating continuous oxidative transformation and increasing oxidation degree [47]. The product MS showed many ion clusters above m/z 720.004, clearly separated by multiples of 16 (Fig. S4).
The oxidation treatment not only introduced oxygen and hydroxy entities to C 60 cages, but also gave rise to cageopened products such as C 59 O 9 H 13 (865.057), C 53  Based on the results provided by C 60 probe, it is clear that: (1) K 2 FeO 4 in H 2 SO 4 environment can hardly attack the sp 2.28 -C of strained C=C bonds. As a reasonable inference, it would not be able to oxidize the C=C bonds that are inerter in most carbon materials (such as the graphite in debate); (2) K 2 FeO 4 in solid state can open the strained C=C bonds and even consume the C atoms in the skeleton. Therefore, the active defective sites on carbon surface could be readily oxidized by such dry chemistry, as has been observed in our previous work on DCNT functionalization [42].

Further Testing the Reactivity Using Nanocarbons
It is still uncertain whether (1) the liquid-phase K 2 FeO 4 takes effect in the oxidation of active defects and (2) solid-state K 2 FeO 4 is oxidizing enough for less strained C=C bonds. To address these issues, the reactivity of K 2 FeO 4 was further determined using nanocarbons including CNTs and CNFs, in both defective and graphitized types (Figs. 1c, S5 and S12). Specifically, DCNT contains rich numbers of surface defects (adatoms, vacancies, cracks, etc.) [48], exhibiting an I D /I G ratio of up to 1.84, while GCNT has a well-graphitized, nearly defectless surface [49][50][51] with a very low I D /I G of 0.028. DCNF is made of stacked graphene "cups" exposing large amounts of edge sites on the outer shells [52,53] giving a high I D /I G of 1.84, while GCNF, constructed by close packing of carbon rods, features edge-closed loops formed by high-temperature graphitization [54] with a much lower I D /I G of 0.33.

K 2 FeO 4 in Liquid Phase Could Only Oxidize Surface Defects
The oxidizing ability of K 2 FeO 4 in liquid phase was found to be modest: the oxidation was efficient to the defective nanocarbons while inoperative to the graphitized ones. To be specific, DCNTs treated by K 2 FeO 4 /H 2 SO 4 for 2 h showed good aqueous dispersibility, in contrast to the insoluble raw material (Fig. 3a). The surface O/C ratio (detected by XPS) showed an increase from 3.5% (raw) to 9.1% (2 h) (Fig. 3b).
In addition, TG weight loss (Fig. 3c), originated from thermolysis of functional groups on surfaces, also supported the increased oxygen content on 2-h-treated DCNTs. However, GCNTs after treatment (up to 8 h) showed properties that were substantially unchanged compared to the raw material, including poor water dispersibility, few contents of surface oxygen and low levels of TG weight loss (Fig. 3a-c). The results of CNFs treated by K 2 FeO 4 followed a similar trend with those of CNTs (Fig. 3a, e, f). Note that 8-h-treated GCNFs displayed distinguishable oxidation effects, since the so-called graphitized CNFs actually contained a certain number of defects in view of the I D /I G of 0.33. The above results suggested that liquid-phase oxidation by K 2 FeO 4 could only occur at the original defects on carbon materials. Raman spectra further confirmed that additional defects (which would arise from the reactions on C=C bonds) were not produced during the reaction process, as reflected by the almost unchanged I D /I G values for both unoxidized nanocarbons and the oxidatively modified ones (Fig. 3d, g). As a result, carbon nanomaterials were protected from structural damage during the oxidation treatment, as shown by SEM and TEM images in Figs. 4 and S5. These results supported the nondestructive oxidation of CNTs reported by Zhang and Xu [39].
According to the reactivity clarified above, K 2 FeO 4 in H 2 SO 4 medium can only provide a slight oxidizing effect on graphite, which arises from reactions at the edge sites of graphene sheets, and the basal planes are unlikely to be affected. This explains why K 2 FeO 4 /H 2 SO 4 method is unsuitable for preparing graphene oxide as pointed out by Sofer et al. [36]. On the other hand, when bulk graphite sample was replaced by nanographite platelets with high ratio of edge-to-plane sites (Fig. S6), distinct oxidizing effects were observed: the surface O/C ratio was increased to 6.8% with discernible -COOH peak in C 1s spectrum after 8-h treatment (Fig. S7), which manifested again the ability of K 2 FeO 4 to oxidize the defective sites on carbon surface.

K 2 FeO 4 in Solid State Could Open the Inert C=C Bonds
K 2 FeO 4 in solid state was able to oxidize not only the defective nanocarbons but also the graphitized ones. For nanocarbons which got negligible oxidation in liquid phase, effective oxidative modifications were achieved in the solid state; and in cases when liquid-phase oxidation was successful, the solid-state reactions would provide higher levels of oxidation (Fig. 3, Table S4). Such strong oxidizing effect can be  (Fig. 3d, g).
To figure out whether the bond breakage is merely a result of physical effect by ball milling or contributed from the chemical reactivity of K 2 FeO 4 solids, two isomorphous salts, non-oxidizing K 2 SO 4 and weakly oxidizing K 2 CrO 4 , were also used in the dry reaction system for comparison. We used a mild ball milling condition of 250 rpm as suggested by the C 60 experiments, and the nearly defectless GCNTs were taken as substrate. As displayed in Fig. 5, GCNTs during ball milling with K 2 SO 4 showed gentle increase in the defect content with time (I D /I G only increased 0.025 after 12 h), which was reasonably induced by mechanical force. K 2 CrO 4 treatment had a comparable effect with slightly  Fig. S13 more defects at long reaction times. In contrast, K 2 FeO 4 led to efficient introduction of defects, producing much higher defect degrees than the other two salts at the same condition. Therefore, K 2 FeO 4 solids possessed sufficient oxidizing ability to directly open the inert C=C bonds under the mechanochemical conditions. It was no surprise that C 60 cages were easily damaged by such dry oxidation.

Discussion
Our systematic studies using elaborately selected carbon materials clearly reveal two distinct oxidizing activities of ferrate(VI) depending on the reaction environment. Their different oxidation effects on four nanocarbons are depicted in Fig. 3h. Liquid-phase oxidation only works for the defective sites, which means the reactivity of ferrate(VI) is largely depressed in H 2 SO 4 medium and becomes much weaker than that of the commonly used oxidizers (e.g., KMnO 4 and HNO 3 ) for carbon materials. Such modest reactivity limits its scope of application, but may be desirable for functionalizing defective carbon materials, where effective surface oxidation can be achieved with no risk of disturbing the carbon structure or morphology.
In solid state, ferrate(VI) readily oxidizes the surface defects and its oxidizing power is strong enough to break the inert C=C bonds in carbon lattice. Solid-state ferrate(VI) oxidation is thus generally applicable for introducing oxygenated groups (e.g., −COOH) onto various carbon surface. In addition, the mechanical force involved in reaction can favor the oxidation performance by producing more defects. On the other hand, the mechanochemical conditions (e.g., rotation speed, ball type and time) must be optimized to avoid undesired structural damage, especially for fragile materials like CNFs (see Figs. 4 and S8-S11 for detailed studies on the effect of reaction condition on product structure).

Conclusions
By using molecular and nanoforms of carbon as substrates, we unraveled the reactivity of ferrate(VI) in oxidizing carbon materials. The theoretically strong oxidizing power of ferrate(VI) is largely depressed in H 2 SO 4 medium, yielding a modest reactivity that only oxidizes the active defects on carbon surface. This liquid-phase ferrate(VI) oxidation can be used as a gentle approach to functionalizing defect-rich carbon materials with the advantage of protecting structural integrity. Ferrate(VI) in solid state releases a high oxidizing power that is capable of opening the inert C=C bonds in carbon lattice, making it generally applicable to introduce oxygenated groups to various carbon materials. This intrinsic strong reactivity underlies the dry chemistry of ferrate(VI) and implies its wide scope of applications in green and powerful oxidative functionalization/ transformation.
These two distinct oxidizing abilities could also apply to other kinds of materials. Considering the emerging role of high-valent iron compounds (represented by ferrate(VI)) in materials science, understanding their reactivity in different conditions is of fundamental importance for guiding their applications.