Microwave-assisted synthesis of iron sulfide motifs for electrochemical applications

To attach FeS₂ octahedra onto MWNTs (SES Research, >95 weight %) through an in-situ growth process, the method described previously for creating pristine FeS₂ was utilized. The as-obtained samples of commercial MWNTs were characterized by lengths of several microns and average outer diameters, measuring in the range of 40–100 nm; Figure S1(A) highlights a transmission electron microscopy (TEM) image of the as-obtained MWNTs. In terms of purity, from the SES research company website, these MWNTs possess >95% nanotubes vs. amorphous carbon (<2%) with an ash content of <0.2%. In the first step, set quantities of these MWNTs (∼1 mg) were put into the H₂O:EG 2:1 mixture, prior to the addition of iron and sulfur precursors. These were dispersed through the mediation of ultrasonication. After the relative iron and sulfur precursors were added (2:3 FeCl₃:TAA), the solution was stirred for 30 min followed by our as-developed microwave protocol, conducted at 180 °C for 30 min under otherwise identical conditions. The as-prepared FeS₂/MWNT composites were ultimately separated by centrifugation, following by washing with ethanol, prior to characterization.

Pre-formed Fe₃S₄ nanosheets were synthesized, as stated above. Then 1 mg of MWNTs was dispersed in 5 ml ethanol using sonication to ensure a final mass loading of 10% MWNTs. Next the Fe₃S₄ material was added in with the MWNT mixture and stirred for 30 min. After stirring, the solution was centrifuged and washed with ethanol. FeS₂/MWNT composites were similarly constructed, by means of comparison.

Powder XRD has been utilized to analyze the crystallographic information and confirm the chemical composition of as-prepared samples. To prepare the XRD sample for analysis, as-prepared nanomaterials were dispersed in ethanol and then drop cast onto a zero-background holder (MTI Corporation, Zero diffraction plate for XRD, B-doped, p-type Si, 23.6 mm in diameter by 2 mm in thickness). The Rigaku Miniflex diffractometer, operating in the Bragg configuration with Cu Kα radiation (λ = 1.5418 Å), was used to obtain sample powder XRD data over the 2θ range of 20°–60°, with a scan rate of 10° min⁻¹.

Electron microscopy techniques that we employed included scanning electron microscopy (SEM), TEM, and high-resolution transmission electron microscopy (HRTEM). These were all used to obtain key information about sample morphology, sample size, and crystal lattice identification. Specifically, SEM data were acquired using a Hitachi 4800S instrument under 10 kV voltage conditions. To prepare the SEM sample for characterization, our Fe_x S_y nanomaterials were dispersed in ethanol prior to drop-casting onto a silicon wafer followed by air drying. The TEM and HR-TEM samples were prepared by drop casting aliquots of the Fe_x S_y nanomaterials solution in ethanol onto a lacey carbon-coated 300 mesh copper grid. The TEM images were collected using a JEOL JEM 1400 LaB6 TEM, equipped with a 2048 × 2048 Gatan CCD camera, operating at an accelerating voltage of 120 kV. HRTEM analysis was performed on a JEOL 2100F, run with an accelerating voltage of 200 kV.

XPS was used to confirm elemental composition and derive oxidation state information about the various different elements within our samples. Specifically, XPS sample preparation was similar to that used for SEM. Fe_x S_y nanomaterials were dispersed in ethanol and then drop cast onto a Si wafer (1 cm × 1 cm). The XPS characterization was conducted using a home-made system, with a model SPECS Phoibos 100 electron energy analyzer for electron detection. Al K_α radiation (1486.6 eV) (model XR 50) was used as the x-ray source for the characterization process. Data collected in the Fe 2p and S 2p regions were interpreted using commercial CasaXPS software; we used well-established background correction and curve fitting algorithms for our subsequent data processing.

Working electrodes were developed by combining an active material—Fe₃S₄, Fe₃S₄/MWNTs, or Fe₃S₄/graphene—with conductive carbon black, and polyvinylidene difluoride, all of which were suspended in N-methylpyrrolidinone to generate electrodes with an 8:1:1 mass ratio. Within an Ar environment, coin cells were assembled by pairing working electrodes with lithium metal anodes, separated by a polypropylene film, in the presence of a 1.0 M lithium bis(trifluoromethanesulfonyl)imide and 0.2 M LiNO₃ in a dipropylene glycol dimethyl ether (DPGDME) electrolyte. Cyclic voltammetry (CV) measurements were collected using a Biologic VSP multichannel potentiostat at a scan rate of 0.1 mV s⁻¹ between a 3 V to 1 V potential range, for five cycles. The resulting current values were normalized with respect to the active mass of each working electrode.

A recently published paper discussed the significance of anion identity on the resulting iron oxide nanoparticle size, using a microwave-assisted method [48]. Generally, it was observed that the type of anion used in the reaction affected the corresponding size of the iron oxide nanoparticle products [48]. In particular, it was theorized that whereas the use of sulfate anions impacted the nucleation process through steric effects, the introduction of chloride ions favored lower bonding coordination modes, all of which led to the formation of variously sized particles. Another study suggested that the smaller Cl⁻ anion could be incorporated into the structure of the iron oxyhydroxide precursor, β-FeOOH, thereby stabilizing the FeO₆ complex, whereas the larger SO₄ ²⁻ anion distorted the FeO₆ octahedral unit, thereby assisting in the creation of a different iron hydroxide, namely α-FeOOH [49].

Given the clear importance of the anion, we compared the behavior of chloride versus sulfate salts under reaction conditions known to successfully produce FeS₂ octahedra. Specifically, we ran 2:3 M:S ratios in the presence of Fe³⁺ within a 2:1 H₂O:EG mixture at 180 °C for 30 min. The use of sulfate yielded not only compositionally undesirable marcasite impurities (figure S2(A), orange) but also a polydisperse mixture of morphologies, including unwanted particles (figure S3(B)). This finding correlates with the prior reported study on iron oxides, wherein the SO₄ ²⁻ anion appeared to induce the formation of the orthorhombic precursor, which in our case was marcasite. By extension, the chloride anion tended to favor the production of the preferred cubic material. Hence, we were able to highlight the importance of the choice of iron precursor salt type, in this case, ferric chloride.

The precursors' relative concentrations were analyzed within a precise and limited range of relatively small increments, comprised of 2:3, 2:4, and 2:5 M:S molar ratios, respectively (figure S2(B)). It was experimentally deduced that a 2:3 M:S molar ratio represented an optimized quantity, because in this case, XRD was consistent with a pure pyrite composition whereas the corresponding SEM data highlighted the formation of homogeneously dispersed octahedra (figures S2(B), yellow, and S3(A)). Increasing the M:S molar ratios to 2:4 and 2:5 worsened the sample quality in that we observed increasingly greater distortions of the octahedral shape coupled with a rise in the amount of marcasite impurity detected (figure S2(B) orange and yellow, respectively, and figures S3(C) and (D), respectively).

Since, as previously noted, marcasite formation is favored at lower pH [34], the appearance of marcasite at higher M:S ratios may indicate that the solution pH becomes lower with greater S precursor content, due to the increased hydrolysis of TAA and the concomitant formation of ammonium ions. Therefore, 2:3 was determined to be the optimized M:S ratio, when using TAA. Reflecting the trajectory of our experiments, in the presence of TAA, the optimized reaction conditions consisted of a 2:3 M:S molar ratio in which ferric chloride was utilized as the iron precursor. This sample was subsequently used for additional experimentation with other reaction parameters.

The role of solvent was consequently explored. For microwave-assisted reactions, solvent choice is significant in the context of its intrinsic tangent factor, a reflection of how much microwave energy it can absorb. Because EG maintains a higher tangent factor (and hence an increased absorption of microwave energy) as compared with water, the reaction kinetics in this medium will be noticeably faster than in a pure aqueous solvent. Furthermore, the reducing properties of EG render it possible for ferric chloride to be used as a precursor, because this solvent can convert Fe³⁺ to Fe²⁺. In prior studies, adjusting the relative volumes of H₂O to EG could promote the formation of either magnetite or hematite, respectively [50]. Hence, EG played a dual role as an effective reducing agent and a microwave heat absorber.

Therefore, not surprisingly, the amount of EG incorporation impacts upon the resulting chemistry and morphology of as-prepared samples, a finding that was apparent upon systematically increasing relative EG content in the context of H₂O:EG volume ratios of 1:0, 2:1, and 1:1, respectfully. Specifically, using a 1:0 H₂O: EG ratio, we isolated not only slight marcasite impurities (figure S2(C), red) but also a mixture of nanoflowers and octahedra (figure S3(E)). With a 2:1 H₂O:EG volume, we observed uniform octahedra (figure S3(A)) that consisted of pure pyrite (figure S2(C), orange). However, increasing the amount of EG still further to a 1:2 H₂O:EG volume ratio led not only to a greater size polydispersity of the octahedra but also a 'roughened' surface (figure S3(F)). Furthermore, although the associated XRD was indicative of FeS₂, it was still of low enough crystallinity that we could not conclusively rule out the presence of other detrimental impurities, especially amorphous ones (figure S2(C), yellow).

Hence, our data implied the presence of a 'sweet spot' in which the use of a 2:1 H₂O:EG solvent ratio by volume enabled us to achieve the targeted chemical composition and morphology. Our observations reflect similar findings in a previous report, which analyzed the effect of varying the relative components of a water:diethylene glycol mixture [51].

The effect of reaction time was similarly investigated. Specifically, the reaction was tested using a 2:1 H₂O:EG ratio with a 2:3 M:S precursor ratio, run at 180 °C after 10, 15, 30, and 45 min, respectively. At lower reaction times, i.e. at 10 min, the solution yielded no perceptible precipitate. By increasing the microwave irradiation time to 15 min, octahedra and particles were observed (figure S3(G)). However, the amount of as-prepared precipitate was minimal and insufficient for an interpretable XRD scan. Raising the irradiation time further to 30 min solved that latter issue. Whereas the measured XRD pattern (figure S2(D), yellow) was consistent with pure pyrite formation, the associated morphology consisted of homogeneously dispersed octahedra (figure S3(A)) with average diameters of 1.1 ± 0.5 μm. Increasing the reaction time still further to 45 min was less helpful in terms of sample quality, because we observed the formation of distorted octahedra (figure S3(H)) with a composition comprised of pyrite with marcasite impurities (figure S2(D), orange). Again, because marcasite formation is favored under lower pH conditions [34], the appearance of marcasite at longer reaction times may be an indication that the solution pH decreases over time with increasing TAA hydrolysis and corresponding production of H⁺ as the reaction proceeds.

To summarize, in the presence of TAA, our collective data supported the notion that the optimized reaction conditions could be described as follows. In particular, the use of a 2:3 M:S molar ratio in the presence of a 2:1 H₂O:EG solvent volume ratio at 180 °C for 30 min, while utilizing ferric chloride as the iron precursor [37], was optimal in terms of forming the desired motif consisting of uniform pyrite FeS₂ octahedra, measuring ∼1 μm in diameter. Simulations, probing the production of FeS₂ pyrite polyhedral motifs using the METADISE program [39], have suggested that the primary facet, characteristic of FeS₂ pyrite octahedra, consists of the (111) plane. In particular, theory implied that the (111) plane is highly reactive and could only form either within a highly S-rich environment or in the presence of additives as diverse as arsenic, an ethanolamine:H₂O mixture [52], or PVP [37, 39]. As such, our work was able to experimentally demonstrate that octahedral pyrite FeS₂ possessing the desired (111) plane could be produced in the absence of not only toxic additives but also surfactants, thereby reinforcing the novelty of our contributions.

Although the above-mentioned optimized synthesis of FeS₂ is a novel endeavor in and of itself, the disadvantages of FeS₂ and other metal sulfides for electrochemical performance needed to be addressed. In particular, as mentioned, this material is limited by an undesirable volume expansion that occurs under realistic cycling conditions. Previously, in a similar study, our group mitigated for this effect through the microwave-assisted addition of conductive MWNTs in an effort to boost the intrinsic electrochemical behavior of vanadium sulfide motifs [43].

Therefore, by analogy, we hypothesized a similar outcome in terms of improving the reversibility and conductivity of our as-produced FeS₂ materials through a simple addition of conductive MWNTs into the reaction medium, prior to metal precursor and sulfur precursor addition. As with our previously published strategy on vanadium sulfides [43], an in-situ method was initially tried. In particular, we added in 1 mg of MWNTs to the mixed solvent solution, before placing in iron and sulfur-containing precursors. The solution then underwent microwave irradiation under the optimized reaction conditions described previously, consisting of using a 2:3 M:S precursor ratio in a 2:1 H₂O:EG volume solvent ratio at 180 °C for 30 min, with ferric chloride as the iron-containing precursor.

However, unlike what we had been able to implement with VS₄, the addition of MWNTs in this case appeared to noticeably affect both the chemical composition and the morphology of the isolated products. Specifically, with the in-situ reaction protocol, octahedra still formed, but the size polydispersity perceptibly increased (figure S3(K)) coupled with the appearance of undesirable marcasite impurities, as seen in the associated XRD pattern (figure S2(E), orange).

To overcome this issue, an equally promising ex-situ reaction process was explored, wherein the same quantity of MWNTs was sonicated with pre-formed FeS₂ octahedra for 30 min. This procedure resulted in the formation of a relatively homogeneous dispersion of MWNTs, which were relatively uniformly distributed onto the surfaces of the iron sulfide octahedra (figure S3(J)). The resulting XRD pattern evinced the presence of FeS₂ pyrite with a slight amorphous hump, located at ∼25° and indicative of the presence of carbon. Therefore, the ex-situ method represented an effective means with which to enable conductive carbon attachment without initiating the formation of either morphological or compositional impurities.

Similar with our efforts on TAA, as described previously, we compared and contrasted the use of a chloride versus a sulfate precursor to determine the effect of the choice of the anion, associated with the iron precursor, when using reaction conditions of 180° for 10 min with 1:2 M:S ratio in the presence of a 2:1 H₂O:EG volume solvent mixture. By analogy with TAA, the use of sulfate gave rise to not only undesirable marcasite impurities (figure S4(A), orange) but also a morphological mixture, incorporating unwanted particles (figure S5(B)). This was a much more heterogeneous sample, as compared with the use of the chloride precursor (figures S3(A), red and S4(A)). This finding supports the idea that the use of the sulfate anion led to the preferred production of orthorhombic marcasite versus cubic pyrite. Hence, the optimized iron precursor salt type incorporated chloride.

By analogy with TAA, the M:S molar ratio incorporating LC was systematically varied from 1:1 to 1:4. It was observed that at 1:1 M:S, there was no evident precipitate. Upon increasing the relative amount of LC to 1:2 M:S, we were able to generate the desired Fe₃S₄ (figure S4(B), red). However, if we continued to raise the relative quantity of LC within the context of 1:3 M:S, we produced unwanted pyrite impurities (figure S4(B), orange). We confirmed this trend of diminishing returns in that the use of a 1:4 M:S molar ratio also led to the formation of pyrite impurities (figure S4(B), yellow).

With respect to the isolated morphology, with a 1:2 M:S molar ratio, the sample was mainly comprised of nanosheet aggregates (figure S5(A)). The lengths of these sheets appeared to increase with correspondingly higher LC content for the 1:3 and 1:4 M:S molar ratio samples (figures S5(C) and (D), respectively). Hence, based on all of these collective results, the 1:2 M:S molar ratio was found to be an optimal compromise in terms of enabling the generation of simultaneously chemically pure and reasonably monodisperse nanosheet aggregates.

The solvent ratio was subsequently optimized. It was observed that a pure aqueous solution yielded a mixture of Fe_x S_y materials, including pyrite, marcasite, greigite, and FeS (figure S4(C), red) with a resulting morphology, consisting of nanosheet aggregates with octahedral impurities (figure S5(E)). The addition of EG within a 2:1 volume ratio gave rise to a comparative increase in Fe₃S₄ production, characterized by fewer pyrite impurities (figure S4(C), orange) coupled with a nanosheet morphology possessing smaller dimensions than that noted using a pure aqueous solution (figure S5(A)). Further increasing the relative amount of EG within a 1:1 volume ratio was less beneficial in that it yielded very little precipitate. This was found to consist of Fe₃S₄ and pyrite (figure S4(C), yellow) which had formed as an aggregated nanosheet morphology (figure S5(F)). Hence, by analogy with TAA, whereas the 1:1 H₂O:EG volume ratio could indeed generate the correct composition to some degree, the lack of crystallinity of that sample did not rule out the possibility of other compositional impurities, including amorphous ones. Moreover, analysis of the associated morphology implied that whereas the constituent nanosheet size had reduced in magnitude, the corresponding size of the overall aggregate had actually enlarged. Therefore, based on all of these data, we determined that the optimized solvent ratio consisted of a volume ratio of 2:1 H₂O:EG.

Finally, the reaction time was explored after 5, 10, 30, and 45 min, respectively. It was observed that after 5 min, there was insufficient precipitate for XRD characterization, though SEM analysis revealed a distinctive nanoflower assembly (figure S5(G)). Raising the reaction time to 10 min yielded monodisperse nanosheet aggregates (figure S5(A)), composed of pure Fe₃S₄ greigite (figure S4(D), red). Increasing the time further to 30 min led to the appearance of not only pyrite impurities (figure S4(D), orange) but also nanosheet aggregates (figure 4(H)). At 45 min, we could not isolate enough precipitate for XRD analysis; the sample that we produced was comprised of nanoflowers, hierarchically and unusually arranged in a linear fashion (figure S5(I)). As such, based on all of these data, we concluded that 10 min of reaction time generated the best results in terms of promoting the desired Fe₃S₄ growth. To summarize, the optimized conditions for Fe₃S₄ nanosheet growth can be described, as follows: a 1:2 M:S molar ratio of FeCl₂:LC within a 2:1 H₂O:EG solution by volume, heated via a microwave-assisted reaction protocol for 10 min at 180 °C.

Similar to what we had undertaken with the TAA-based system, the method of addition of the MWNTs was also investigated in the presence of LC. An in-situ procedure was first explored, wherein the MWNTs were dispersed in a mixed solvent solution, prior to Fe and S precursor addition, before subjecting the reaction medium to microwave irradiation at 180 °C for 10 min, as per the optimized conditions we had finalized previously. Unlike the in situ FeS₂ reaction above wherein the introduction of MWNTs led to compositional impurities, in the case of Fe₃S₄, there were the added complications of the appearance of serious morphological inhomogeneities upon MWNT addition. Specifically, the latter manifested themselves as not only the inhomogeneous growth of nanosheet aggregates but also the apparent growth of Fe–S structures on the MWNTs themselves (figure S5(J)).

Therefore, by analogy with TAA, an ex-situ reaction was explored, wherein the same amount of MWNTs as mentioned above was sonicated with preformed Fe₃S₄ nanosheets for 30 min. The resulting SEM image (figure S5(K)) highlights the attachment of MWNTs onto Fe₃S₄ nanosheets with no apparent morphological impurities, thereby reinforcing the effectiveness of applying the ex-situ method to functionalizing these metal sulfides with conductive CNTs. Furthermore, the addition of graphene onto the Fe₃S₄ nanosheets was performed in an analogous manner (figure S5(L)); the corresponding as-generated sample did not give rise to any obvious morphological impurities.

Upon optimizing the synthesis conditions for generating FeS₂ nanomaterials, we collected XRD data, which corroborated the production of cubic FeS₂ (figure 1(C)). In particular, the experimental pattern (black) clearly matches that of the predicted cubic pattern expected for pyrite FeS₂ (blue, JCPDS 42-1430) with no obvious marcasite impurities (red, JCPDS 37-0475). The HRTEM images (figures 1(A) and (B)) emphasize the crystalline nature of this material; the complementary SEM image (figure 1(D)) highlights the fabrication of octahedra with edge lengths of ∼220 ± 54 nm. Furthermore, HRTEM analysis of these features evinced a measured d-spacing value of 2.4 ± 0.1 Å, consistent with the (210) FeS₂ plane (figure 1(B)). Additional HRTEM-EDS results were utilized to determine the relative elemental content and distribution of Fe and S (figure 2). Specifically, Fe and S elements were found to be homogenously dispersed within the octahedral motif, as noted by figures 2(B) and (C), respectively. From the accompanying EDS spectra, the relative ratio between the Fe and S elements was determined to be 1:2, respectively, again confirming the desired FeS₂ composition (figure S6(A)).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

However, further experimentation, including XPS and electrochemical characterization and testing, could not be conducted on pristine FeS₂ alone, because of the low yield (∼2 mg) of the original reaction. This situation could be ascribed to our inability, despite significant time expended and numerous iterations attempted, to upscale this synthesis to hundreds of mg, a necessity for ensuring the interpretability of the resulting electrochemical analysis. In fact, the quality of the products of these larger-scale syntheses was characterized not only by the appearance of compositional impurities such as marcasite but also morphological contaminants such as spheres [40]. Therefore, moving forward, only Fe₃S₄ nanosheets and their corresponding MWNT-based composites were subjected to electrochemical testing.

Optimized Fe₃S₄ nanomaterials were characterized with XRD to ascertain the compositional purity of spinel Fe₃S₄ (figure 3(C)). As observed from the XRD data, the experimental pattern (black) clearly matches that of the database spinel pattern, expected for greigite Fe₃S4 (red, JCPDS 89-1998) with no apparent impurities. HRTEM images (figures 3(A) and (B)) show consistent features with that noted in the SEM image (figure 3(D)). Specifically, we noted the appearance of nanosheets, measuring ∼35 ± 15 nm in width and characterized by crimped edges that were roughly microns in length. Furthermore, HRTEM analysis of these nanosheets suggested a d spacing value of 2.9 ± 0.1 Å, consistent with the expected (311) Fe₃S₄ plane (figure 3(B)). Moreover, the corresponding Fe₃S₄/MWNT composite was also analyzed using HRTEM (figure S7(A)); the blue arrows indicate the locations of adjoining MWNTs. In the corresponding HRTEM image of the Fe₃S₄/graphene composite, the blue arrows point to the presence of graphene (figure S7(B)). Complementary HRTEM-EDS was utilized to determine the elemental distribution of Fe and S (figure 4). Both Fe and S elements were perceived to be homogeneously dispersed within the nanosheets, as noted in figures 4(B) and (C), respectively. From the EDS spectra, the relative ratio between the Fe and S elements was found to be 1:1.3 respectively, thereby verifying the anticipated Fe₃S₄ composition (figure S6(B)).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

By contrast with FeS₂, we were able to generate sufficient quantities of Fe₃S₄ for further analysis. Specifically, the surface chemistry of these nanosheets was probed using XPS, wherein we collected data on the Fe 2p and S 2p regions. The S 2p spectra consisted of both S²⁻ and S⁻ species, due to the spin–orbit coupling of the S orbitals (figure 5(A)) [59]. In particular, the S²⁻ peaks were located at both 163.0 and 164.2 eV, corresponding to the 2p_3/2 and 2p_1/2 peaks, respectively, whereas the S⁻ 2p_1/2 peak was positioned at 161.2 eV. The Fe 2p region was analyzed, incorporating the presence of two oxidation states of iron, namely Fe²⁺ and Fe³⁺ (figure 5(B)). The associated peaks for Fe²⁺ and Fe³⁺ were found not only at 709.6 and 712.2 eV for the 2p_3/2 peaks but also at 723.2 and 725.8 eV for the 2p_1/2 peaks, respectively, all of which agreed with the previous literature [59]. Because of the observation of the mixed oxidation states of both Fe and S, these findings further validated our assertion that Fe₃S₄ had indeed been produced.

Zoom In Zoom Out Reset image size

The Fe₃S₄/MWNT composite was also analyzed using XPS with little to no changes noted in the profiles of the S 2p and Fe 2p spectra (figures S8(A) and (B) respectively). Within the S 2p region, the S²⁻ signal indicated the presence of not only peaks located at 163.2 and 164.4 eV for the 2p_3/2 and 2p_1/2 peaks respectively but also a S⁻ 2p_1/2 peak situated at 160.9 eV (figure S8(A)). With respect to the Fe 2p region, by analogy with pristine Fe₃S₄, we observed the presence of two oxidation states, namely Fe²⁺ and Fe³⁺ species (figure S8(B)). In particular, Fe²⁺ peaks were observed at 709.5 and 723.2 eV for the 2p_3/2 and 2p_1/2 signals, respectively, whereas Fe³⁺ manifested itself as peaks present at 712.0 and 725.7 eV for the 2p_3/2 and 2p_1/2 data, respectively. Furthermore, the C 1s region exhibited peaks, indicative of C–C and C–O bonds at 284.4 and 286.5 eV respectively, comparable to previously published literature on metal sulfide/MWNT composites (figure S8(C)) [43].

The analogous Fe₃S_4/graphene composite was also probed using XPS with little to no alterations observed at the profiles of the Fe 2p and S 2p spectra, associated with either the Fe₃S₄ or the corresponding MWNT composite (figures S8(E) and (D) respectively). With respect to the S 2p region, a S²⁻ signal consisted of features, situated at 163.0 and 164.1 eV, and could be attributed to the 2p_3/2 and 2p_1/2 peaks respectively (figure S8(D)). In this case, we did not observe any extraneous S⁻ peaks. Regarding the Fe 2p region, by analogy with our prior discussion of similar samples, two oxidation states were noted that could be assigned to Fe²⁺ and Fe³⁺ species (figure S8(E)). In particular, Fe²⁺ peaks were observed at 710.3 and 723.8 eV, associated with the 2p_3/2 and 2p_1/2 signals, respectively, whereas Fe³⁺ peaks located at 712.4 and 726.4 eV could be assigned as 2p_3/2 and 2p_1/2 data, respectively. Next, the C 1s region was found to exhibit 2 peaks, similar to what we had found for the MWNT composites above, and these signals could be attributed to the presence of C–C and C–O bonds at 284.2 and 286.0 eV, respectively. Our findings are comparable to previously published literature on related metal sulfide/graphene composites (figure S8(F)) [8]. Collectively, this combination of diffraction, microscopy, and spectroscopy data indicates that we were able to fulfill our stated objective herein of optimizing sample uniformity, homogeneity, and monodispersity in terms of the isolated chemical composition and morphology of as-generated Fe₃S₄ nanosheets and their associated Fe₃S₄/graphene and Fe₃S₄/MWNT composites.

The CV experiments were conducted to probe the electrochemistry of the Fe₃S₄ structure and associated composites within lithium half cells. We note that a non-toxic ether, DPGDME, was selected as the electrolyte to provide for a benign battery system. The results illustrated in figure 6 display cyclic voltammograms for Fe₃S_4, Fe₃S₄/MWNT, and Fe₃S₄/graphene, respectively. The overall reactions taking place can be represented by the following: Fe₃S₄ + 8Li⁺ + 8e⁻ → 4Fe + 3Li₂S [1]. The initial CV curves show three distinctive peaks in the reduction region at approximately 1.37 V, 1.58 V, and 1.87 V. These peaks have been previously assigned to the formation of Li_x Fe₃S₄ at ∼1.8 V, the reduction of Fe³⁺ to Fe²⁺ at ∼1.6 V, and a two-phase reaction to form Fe⁰ and Li₂S at ∼1.37 V, respectively [1]. Two distinctive peaks are observed in the charge region at approximately 1.9 V and 2.5 V, corresponding to the oxidation of Fe into Li₂FeS₂ and the delithiation of Li₂FeS₂ into either Li_2−x FeS₂ or FeS, respectively [1, 4, 5]. Interestingly, figure 6 demonstrates differences among the sample types. The delivered current density for Fe₃S₄/graphene is smaller than that of the pure Fe₃S₄ and that of the Fe₃S₄/MWNT composite; properly understanding these observations will be the basis of more extensive characterization studies in the future. Nevertheless, these preliminary findings suggest that MWNTs are a more effective electrode additive for these sulfide-based materials as compared with graphene.

Zoom In Zoom Out Reset image size