1 Introduction

Nature utilizes FeIV =O and FeV=O complexes as the active centers for a number of important enzymatic oxidations, which motivates many fundamental studies on the properties of high-valent iron compounds, especially their chemical and biological reactivities [1,2,3,4,5,6,7,8]. FeVI, the highest accessible oxidation level of iron, generally exists in the form of ferrate(VI) ion (FeO42−) with four FeVI=O bonds. Ferrate(VI) possesses powerful oxidizing ability, as revealed by its higher redox potentials (up to + 2.2 V in acidic conditions) than those of most traditional oxidants [9,10,11,12]. Together with the environmentally benign nature, ferrate(VI) compounds (commonly K2FeO4) have been considered as promising oxidizing agents in several areas, including water remediation [13,14,15,16,17], organic synthesis [18,19,20,21], high-capacity battery [22,23,24] and O2 evolution [25,26,27,28,29].

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 K2FeO4 for the oxidation/exfoliation of graphite in concentrated H2SO4. Highly water-soluble single-layer graphene oxide was produced in 1 h at room temperature, which indicated the extraordinary oxidizing ability of K2FeO4 in H2SO4 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 K2FeO4 in acidic environments. Nevertheless, some other studies suggested the moderate oxidation effects of K2FeO4 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 K2FeO4 in liquid phase (i.e., H2SO4 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 K2FeO4 was accessible in the “dry” solid state by oxidizing small-molecule substrates [42]. On this basis, K2FeO4 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 K2FeO4 oxidation on carbon materials remains unclear. A critical question is whether K2FeO4 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 C60 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.

Fig. 1
figure 1

a A schematic representation of K2FeO4–carbon reaction system. b Depiction of two reaction conditions (in H2SO4 medium or by solid-state ball milling). c Structural models of the carbon nanomaterials used in this study: C60, defective CNT (DCNT), graphitized CNT (GCNT), defective and graphitized carbon nanofibers (DCNF and GCNF). ID/IG refers to the relative intensity of D to G band in Raman spectrum, taken as a measure of defect degree

Full size image

2 Experimental Section

2.1 Preparation of K2FeO4

The purity of K2FeO4 should be examined before it is used for oxidizing carbon materials. The commercially supplied K2FeO4 products from several manufactures only have actual purities of 20% or below, although the declared purities are > 90%. Therefore, we synthesized K2FeO4 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).

2.2 Oxidation of Carbon Materials

2.2.1 Liquid-Phase K2FeO4 Oxidation

Oxidation of C60. 30 mg of C60 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 K2FeO4 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. The solid was obtained by centrifugation (12,000 rpm), followed by washing in sequence with 2 M HCl (several times to remove Fe3+), 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.

2.2.2 Solid-State K2FeO4 Oxidation

Oxidation of C60. 100 mg of C60 and 2.5 g of K2FeO4 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 Fe3+), 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.

2.3 Characterization

The purity of K2FeO4 sample was tested by spectrophotometry, X-ray diffraction (XRD) and 57Fe Mössbauer spectroscopy (57Co(Pd) source). Chemical structure of C60 samples was determined using matrix-assisted laser desorption/ionization Fourier transform ion cyclotron resonance mass spectrum (MALDI-FTICR MS) and 1H 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, N2). 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.

3 Results and Discussion

3.1 Purity Analysis of K2FeO4

K2FeO4 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 57Fe Mössbauer spectrum proved 97.6% relative content of ferrate(VI) (Fig. S2, Table S2).

3.2 Probing the Reactivity Using C60 Molecules

We first probed the reactivity of K2FeO4 using C60. As a special kind of carbon material with defined molecular structure, C60 can provide directive and reliable assessment results. The spherical cage of C60 confers an excess of strain to C=C bonds, inducing a unique sp2.28 hybridization of the carbon atoms with a pyramidalization angle θP of 11.6° [44, 45]. This endows C60 with a moderate reactivity: inerter than that of the sp3-C atoms (θP = 19.5°) appearing as defects on carbon surface and more active than that of the sp2-C atoms with smaller θPs in graphite (θP = 0°) and most other carbon materials (e.g., CNTs and CNFs) [46]. In addition, the small-molecule property of C60 allows its product structure to be easily determined by standard organic analytical methods such as mass and NMR spectroscopy.

After the liquid-phase oxidation, C60 products showed scarcely any changes with respect to the pristine sample, even with a large excess of K2FeO4 (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 C60 ions), no newly produced ion peaks were observed (Fig. S3), which unambiguously showed K2FeO4 in liquid phase was not reactive enough to oxidize C60.

Fig. 2
figure 2

a Comparison of C60 oxidation results by K2FeO4 in liquid phase and solid state. b MS of oxidized C60 products by 6-h solid-state reaction (ball milling at 250 rpm). Inset shows photographs of C60 samples in toluene–water biphasic system. L12 refers to the product by 12-h liquid-phase treatment, and S1–6 refers to that by solid-state reaction for 1–6 h

Full size image

In contrast, K2FeO4 efficiently oxidized C60 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 (O) and 17 (OH) mass units. Various oxidized species were identified, including C60O (736.000), C60O2H (753.003), C60O3H (768.998) and C60O11H17 (913.078). The attachment of –OH groups onto C60 cage was also confirmed by 1H NMR (Fig. S4).

The oxidation treatment not only introduced oxygen and hydroxy entities to C60 cages, but also gave rise to cage-opened products such as C59O9H13 (865.057), C53O15H18 (894.063) and C41O25H49 (941.248). We suspected it was the mechanical force that broke the molecular cages, but the production of these broken cages was not remarkably depressed (62% vs. 50%, relative quantifications by MS) when the energy input was substantially decreased (from 250 rpm × 6 h to 180 rpm × 2 h). On the other hand, when K2FeO4 was replaced by a non-oxidizing isomorphous salt K2SO4, the amount of cage-opened products was dramatically reduced (20% by K2SO4 vs. 62% by K2FeO4, 250 rpm × 6 h). These results suggested that the intrinsic reactivity of K2FeO4 in solid state was strong enough to cleave the C=C bonds of C60 (presumably by the addition of FeVI=O with C=C bonds).

Based on the results provided by C60 probe, it is clear that: (1) K2FeO4 in H2SO4 environment can hardly attack the sp2.28-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) K2FeO4 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].

3.3 Further Testing the Reactivity Using Nanocarbons

It is still uncertain whether (1) the liquid-phase K2FeO4 takes effect in the oxidation of active defects and (2) solid-state K2FeO4 is oxidizing enough for less strained C=C bonds. To address these issues, the reactivity of K2FeO4 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 ID/IG ratio of up to 1.84, while GCNT has a well-graphitized, nearly defectless surface [49,50,51] with a very low ID/IG 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 ID/IG 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 ID/IG of 0.33.

3.3.1 K2FeO4 in Liquid Phase Could Only Oxidize Surface Defects

The oxidizing ability of K2FeO4 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 K2FeO4/H2SO4 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 K2FeO4 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 ID/IG of 0.33.

Fig. 3
figure 3

Oxidation results of various nanocarbons by K2FeO4 in liquid phase and solid state. If applicable, water-dispersible portions of the products were isolated for measurements to better show the oxidation effects. a Photographs of 0.1 mg mL−1 samples in water that had been sonicated for 5 min and settled for 3 days. b, e XPS O/C ratios (inset: representative C 1s spectra). c, f TG weight losses (inset: representative TG curves). d, g ID/IG ratios (inset: representative Raman spectra). h A depiction of the effects of K2FeO4 oxidation on nanocarbons in liquid phase and solid state (oxygen atoms are shown in red). (Color figure online)

Full size image

The above results suggested that liquid-phase oxidation by K2FeO4 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 ID/IG 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].

Fig. 4
figure 4

SEM images of carbon nanomaterials under different reaction conditions. a–d DCNTs: a raw, b L2, c S2 at 250 rpm, d S2 at 300 rpm; e–h GCNTs: e raw, f L8, g S8 at 250 rpm, h S8 at 300 rpm; i-l DCNFs: i raw, j L2, k S2 at 100 rpm, l S2 at 250 rpm; m-p GCNFs: m raw, n L8, o S2 at 100 rpm, p S12 at 100 rpm

Full size image

According to the reactivity clarified above, K2FeO4 in H2SO4 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 K2FeO4/H2SO4 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 K2FeO4 to oxidize the defective sites on carbon surface.

3.3.2 K2FeO4 in Solid State Could Open the Inert C=C Bonds

K2FeO4 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 interpreted as the ability to open the inert C=C bonds in carbon lattice, as indicated by the evidently increased ID/IG values after solid-state oxidations (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 K2FeO4 solids, two isomorphous salts, non-oxidizing K2SO4 and weakly oxidizing K2CrO4, were also used in the dry reaction system for comparison. We used a mild ball milling condition of 250 rpm as suggested by the C60 experiments, and the nearly defectless GCNTs were taken as substrate. As displayed in Fig. 5, GCNTs during ball milling with K2SO4 showed gentle increase in the defect content with time (ID/IG only increased 0.025 after 12 h), which was reasonably induced by mechanical force. K2CrO4 treatment had a comparable effect with slightly more defects at long reaction times. In contrast, K2FeO4 led to efficient introduction of defects, producing much higher defect degrees than the other two salts at the same condition. Therefore, K2FeO4 solids possessed sufficient oxidizing ability to directly open the inert C=C bonds under the mechanochemical conditions. It was no surprise that C60 cages were easily damaged by such dry oxidation.

Fig. 5
figure 5

ID/IG ratio results of GCNTs by solid-state milling (250 rpm) with three isomorphous salts. A comparison of Raman spectra is shown in Fig. S13

Full size image

3.4 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 H2SO4 medium and becomes much weaker than that of the commonly used oxidizers (e.g., KMnO4 and HNO3) 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).

4 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 H2SO4 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.