**2. Results and Discussion**

With the goal of generating hybrid Carbon nanostructure/IF-WS2 we conducted the steps summarized in the experimental section and produced samples whose qualities are described below. We have divided the results and discussion segment into three parts; the first one presenting the microstructural characteristics and stability of samples based on CNF/IF-WS2, the second one including the study of Graphene/IF-WS2 and the third showing the mechanical properties of both CNF/IF-WS2 and Graphene/IF-WS2 epoxy composites. The samples made in the absence of polymer

were denominated hybrids and the ones with nanostructures embedded in epoxy are referred as composites as a way to distinguish them.

#### *2.1. Carbon Nanofiber/IF-WS2 Hybrids*

The combination of existing 3D CNF structures with IF-WS2 in the absence of a polymeric matrix, as mentioned in the introduction section, results in inhomogeneous solids. For example, the addition of IF-WS2 to already existing CNF generates samples where diffusion paths dominate the final structure: the surface of the carbon fibers tends to contain larger amounts of the IF-WS2 phase than sections of the sample not exposed to the surface. The use of solvents and sonication to disperse the IF or its precursors, is not enough to produce a homogeneous product. A second example; The addition of metal catalyst particles (to promote the carbon fiber growth) to existing IF-WS2 followed by thermal treatments to proceed with the carbon nanofibers growth in the presence of a carbon source, produces a thin layer of metal sulfides in the surface of the metal particles. This phenomenon is observed also in the absence of carbon sources: at moderate temperatures (below 350 °C according to our findings) the metal reacts with the tungsten sulfide producing thin films of secondary phases in the metal particle surface, which in practice poison the catalyst, capping it and hindering the growth of the carbon fibers. Hence, this method also fails to produce the expected product. Variations in the experimental parameters, such as gas flows, placement of precursors, size and nature of metal catalyst, amid others, generates, in the best case scenario, amorphous carbonaceous byproducts of different sizes and morphologies, far from the desired fiber or porous structures [47].

Diverse precursor options could be used to generate IF-WS2 structures; (NH4)2WS4 thermal decomposition and WO3 reaction with sulfur-containing compounds are among the most common routes employed [1,48–56]. The generation of IF-WS2 carried out in our laboratories using such precursors resulted in samples with different levels of long range order and particle size, as can be inferred from its XRD pattern and SEM analysis. Figure 1 presents the WS2 patterns obtained when decomposing (NH4)2WS4 and from the reaction of commercial WO3 nanoparticles with H2S atmospheres. It is worth noting that the temperatures needed for such reactions to occur differ, the former was performed at moderate temperatures (*ca.* 500 °C) while the latter required higher temperatures to be carried out to completion (*ca.* 800 °C). The diffraction peaks for IF-WS2 generated from WO3 are narrower and the reflections (00l) with l = 2*n* more evident and well defined than the ones obtained from (NH4)2WS4. SEM observation of both products confirmed the presence of the hollow cage structures, the so-called 3R phase. Moreover, the former tetrathiotungstate precursor reacts with metal particles when it decomposes, which makes WO3 the best choice to be combined with metal catalyst for fiber growth.

The two stage *in situ* synthesis of homogeneous CNF/IF-WS2 required diverse atmospheres and temperature steps to be accomplished successfully. The first stage had the objective to grow CNF from a metal catalyst intermixed with WO3 and the second stage consisted on sulfurizing the mixture of CNF/WO3 to transform the tungsten oxide into IF-WS2.

The dispersion of the metal particle (nickel in this case) with the WO3 nanoparticles was performed using a solvent and sonication, followed by evaporating the mixture until dry since it rendered a more homogenous precursor than simply grinding the solids. A first temperature step was

carried out at 350 °C in reducing atmospheres as a precautionary measure to assure that the metal catalyst surface was free of oxides. A second step at the same temperature including ethylene (as carbon source) and small amounts of oxygen diluted in inert gas (as reaction initiator and radical originator) was performed to cover the metal particle with an initial thin layer of carbon (which prevents the metal gross agglomeration and sintering at higher temperatures). The growth step was made at 550 °C, following protocols previously developed [57]. A final temperature stage at 900 °C in H2S containing environment was included to transform the WO3, now dispersed in between carbon nanofibers, into IF-WS2. The sample was then allowed to cool to room temperature using an inert atmosphere.

**Figure 1.** XRD patterns of IF-WS2 particles generated from different precursors. IF-WS2 generated from (**a**) commercial WO3 particles; (**b**) from ammonium tetrathiotungstate. All peaks were identified as IF-WS2 [12].

The electron micrographs of the nickel nanoparticles employed, the carbon nanotube intertwined nanofibers generated and the hybrid CNF/IF-WS2 are presented in Figure 2. The ratios of catalyst to WO3 used were designed to have nominal loading values of 0.5, 1.5, and 5% of IF-WS2 with respect to the total weight of the CNF. The distribution of the IF particles was studied using SEM in Secondary Electron (SE) and Backscattered Electron (BSE) modes along with Energy Dispersive Spectroscopy (EDS) mapping and can be described as discrete pockets of cage like particulates of approximately 1–2 microns dispersed into the CNF threads (Figure 2c). This finding is consistent across different sample locations.

The catalyst used for the carbon nanofibers was not removed and represents close to 3% of the total weight of the sample. Its effects on the properties of the epoxy composites are discussed in Section 2.3.

The last stage of the *in situ* protocol, the sulfurization, transforms not only the WO3 into WS2 but the nickel catalyst into nickel sulfide. Indeed, the hybrid CNF/IF-WS2 samples are composed by the solids mentioned in Table 1 (experimental section), where nominal values and elemental analysis by ICP methods are contrasted. The inductive plasma emission spectroscopic data shows that the final composition attained is within 0.2% of the targeted values.

As a means to understand the characteristics of the samples and the different phases formed at the diverse synthesis steps, some of the runs where halted before completion and the products at that point, were analyzed by diverse techniques. As part of those trials, experiments of fiber growth in the absence of WO3 with and without sulfurization treatment were performed. The X-ray diffraction analysis of the latter, containing only CNF samples, is presented in Figure 3. The main reflection, close to 26°, corresponds to the (002) peak of graphite, characteristic of many carbon products that include a crystalline component with various degrees of basal plane alignment [58]. In the present case the peak is associated to the CNF presence. The diffraction pattern of samples after the carbon fiber growth step but before sulfurization (red line in Figure 3), show only the peaks of graphite and a couple of reflections, close to 45 and 52°, that correspond to the nickel particles used as catalyst. The phases encountered after sulfurization (blue line) corroborate the presence of two extra crystalline structures, nickel sulfide: Ni3S2 [59,60] and Ni17S18 [61].

**Figure 2.** Microstructural Analysis of CNF and WS2 precursors and CNF/IF-WS2 hybrid**.**  Scanning Electron Microscopy secondary electron images of (**a**) starting nickel nanoparticles; (**b**) carbon nanofibers grown from Ni catalyst; (**c**) nanoparticle dispersion and cluster of WS2 nanoparticles within carbon nanofibers (CNF/IF-WS2 hybrid).

**Figure 3.** X-ray diffraction pattern of carbon nanofibers. (**a**) Typical peaks of graphite and nickel metal are identified in samples as prepared; (**b**) XRD pattern after sulfurization step shows graphite and nickel sulfide reflections.

The XRD patterns of CNF/IF-WS2 hybrids at the three different levels of loadings prepared are shown in Figure 4. The same phases recognized in Figure 3 after sulfurization; graphitic peaks, Ni3S2 and Ni17S18, along with a comparatively weak peak corresponding to IF-WS2, can be identified.

**Figure 4.** X-ray diffraction patterns for hybrid CNF/IF-WS2 samples. The main difference between XRD patterns in samples at diverse loadings is the relative intensity for the nickel sulfide *vs.* graphitic peaks.

SEM in BSE mode studies along elemental mapping by EDX confirmed the presence of IF-WS2 as highly dispersed phase. Figure 5 shows the metal distribution as bright spots (top left), including both nickel and tungsten phases. The IF-WS2 distribution can be inferred as the positions where W elemental mapping is found (top right). The W mapping shows the presence of mainly small (nm) particulates homogeneously distributed in the sample. In contrast, the Ni particles used as catalyst, now converted into sulfides, appear both as nanometer particulates and micron size agglomerates; with dimensions that could be correlated with the different fiber diameters (bottom right). The fact that the nickel elemental distribution shows micron size clusters is a sign that some agglomeration of some of the initial particles might have occurred during the fiber growth, since the original catalyst had a size distribution in the nm range, as shown in Figure 2a. The sulfur map (bottom left) encompasses both, the IF-WS2 and Ni sulfide components.

The samples' thermal analysis under oxygen containing atmospheres show that all specimens; CNF (catalyst included), CNF sulfurized and hybrid CNF/IF-WS2 are stable at least up to a temperature of 500 °C (Figure 6). At higher temperatures the carbonaceous component burns off to produce CO2, the tungsten sulfide reverts to its oxidized form and the metal catalyst oxidizes. Experiments conducted in inert atmosphere showed that hybrid CNF/IF-WS2 samples do not present weight changes in the window of conditions used (RT to 950 °C) and can be considered thermally stable. Indeed, the hybrid samples stability observed surpasses the ones observed for polymer products without WS2, which in either atmosphere start decomposing at much lower temperatures, and are in agreement with stability observed in air for IF-WS2 based composites [19,20,62]. The

reported thermal stability of WS2 fullerene-like particles alone under oxidizing atmospheres indicates that, depending on particle size, oxidation might begin close to 290 °C (100 nm) or delay up to close to 440 °C (3 m), while exposed to inert gases the phase is stable up to 1200 °C [63,64]. In our case, the use of CNF without polymeric component seems to increase the thermal stability of the hybrid, given that the average IF-WS2 particle size produced was in the nanometer scale (expected to oxidize close to 300 °C) and changes in weight were not evident until above ~500 °C. The thermal stability observed for IF-WS2 might be related to the particulates being embedded in the CNF intertwined fibers, which might delay their reaction. Further studies to fully explain these phenomena are currently under way.

**Figure 5.** Phase distribution studied by SEM/EDX analysis. **Top left**: Backscattered electron image showing the carbon fibers in grey and the metal containing phases as bright spots in the sample CNF/IF-WS2 with the lowest IF-WS2 loading. Elemental mapping by EDX analysis confirmed the presence of IF-WS2 mainly as highly dispersed phase (**top right**) with small (nm) particle size. The elemental Ni map shows the presence of nm and micron size particulates (**bottom right**). The sulfur map (**bottom left**) includes both, the IF-WS2 and Ni sulfide components.

**Figure 6.** Thermogravimetric Analysis. Temperature programmed oxidation (burn off process) for (**a**) carbon nanofibers-including nickel catalyst; (**b**) sulfurized carbon nanofibers -including nickel sulfide and (**c**) CNF/IF-WS2 hybrid with 3% tungsten sulfide loading. All of the samples are stable in oxygen containing environment up to 500 °C.

#### *2.2. IF-WS2/Graphene Hybrids*

The complications introduced by the use of metal catalyst for nanofibers growth were not encountered during the synthesis of Graphene/IF-WS2 hybrids, since no extra metal components were needed to generate them. Only small amounts of urea were used as an expansion agent. The production of graphene was accomplished by the reduction-expansion of graphite oxide (GO), which was generated from graphite flakes by the process described in the experimental section. The GO mediated process generates disordered graphene; where graphene sheets tend to entangle with each other but remain separated enough to maintain a relative high surface area (*ca.* 600 m<sup>2</sup> /g as per BET analysis). It is worth noting that the oxygen content in the synthesis environment when using GO as a carbon precursor is sufficient to oxidize some of the IF-WS2 if the latter is used since the beginning of the protocol, forcing the use of WO3 as the tungsten source. Thus, WO3 nanoparticles were subject to H2S treatments to sulfurize them after its mixtures with graphene were obtained.

The two variations on the synthesis conditions consisted of: (i) dispersing the WO3 nanoparticles with GO and then performing the exfoliation at high temperature (to generate Graphene/WO3) followed by sulfurization (to obtain Graphene/IF-WS2) or (ii) perform the GO exfoliation first and then directly mixing the resulting graphene with WO3 and then sulfurizing the product (rendered Graphene/IF-WS2). The microstructures created are presented in Figure 7.

The first synthetic approach, GO plus WO3 exfoliation followed by sulfurization, resulted in hybrid Graphene/IF-WS2 where the IF structures were located on the surface of disordered graphene sheets (marked by arrows, Figure 7a). Due to the thermal exfoliation and urea reduction process, when volatile groups leave the graphite oxide structure at temperatures close to 200 °C, they carry along the WO3, which ends in top of the layers, where they remain through the sulfurization step that converts them into IF-WS2 cage structures. The exfoliation process separates the graphene sheets but also promotes the separation of IF particles, which were found in all cases as individual particles.

The position of the IF particulates following the second synthetic approach, is quite distinct. In this protocol, GO is first thermally exfoliated and reduced and the subsequent graphene is mixed with WO3 (with the aid of solvents and sonication, followed by solvent evaporation until dry) and the mixture is then sulfurized. The IF particulates are found between the graphene layers, intimately embedded into the sheets structure (Figure 7b). Small (few nm diameter) IF particulate agglomerates, marked in the image with a blue arrow, were observed along individual particles, indicated by red arrows. The IF-WS2 observed by Transmission Electron Microscopy presents the characteristic hollow cage, partially faceted structure and an interlayer spacing of 0.62 nm.

For comparison, samples in which Graphene and IF-WS2 were formed as individual phases and then added as a physical mixture using solvents and sonication (not grown *in situ*), present similar structures to the ones in Figure 7b, demonstrating that the protocols developed to introduce IF-WS2 into carbon structures are required only when tridimensional architectures of the carbonaceous component are needed but not if a layered, two dimensional structure is used.

**Figure 7.** Microstructural Analysis of Graphene/IF-WS2 hybrids and IF-WS2*.* Scanning Electron Microscopy images of (**a**) Graphene/IF-WS2 hybrid generated from GO and WO3; (**b**) Graphene/IF-WS2 hybrid made directly from Graphene and WO3; and (**c**) Transmission electron micrograph showing the characteristic hollow core and interlayer spacing for IF-WS2 particles.

#### *2.3. Epoxy Composites*

The use of epoxy as polymeric matrix was intended (independent of its properties as a composite for use in protection systems) to determine the mechanical properties observed when using carbon nanomaterials with 3D *vs.* 2D structures (CNF *vs.* Graphene) that included low loadings of IF-WS2. Other instances of WS2 nanostructures embedded in epoxy can be found in [14,23], where WS2 nanotubes were used as filler instead of the IF particulates used in this study. To the best of our knowledge, the only existing references including IF-WS2 particles in epoxy systems do not involve a carbonaceous component [21–24].

Experimental testing in our laboratory and previous epoxy resin composite research data [47,65] were used to select 1% as targeted loading of filler material into epoxy matrix. Bare epoxy resin, produced in identical conditions to the ones containing diverse fillers, was used as reference. Table 1 presents the values for the different filler components used to prepare the composites, all

cases include only 1% of total filler and 99% of epoxy, by weight. On the table, the first column contains the nominal values of IF-WS2 component targeted during synthesis, the rest of the columns are the values encountered when analyzing the filler samples by ICP Emission Spectroscopy.

Two types of nanoindentation measurements showed clear increases in the Young's modulus of the epoxy-CNF composites. As described in the experimental section, dynamic mechanical analysis was performed on epoxy-CNF composites to ascertain the degree of viscoelastic deformation of these materials at room temperature. At loading frequencies of both 1 Hz and 45 Hz, the elastic modulus response was dominated by the storage (elastic) modulus with a very minor contribution, less than 5%, from the loss (viscous) modulus. As such, quasi-static indentation was deemed appropriate for these materials. We used quasi-static nanoindention to measure the Young's modulus and hardness for each of the composites using the conditions listed in Table 2. The inclusion of carbon nanofibers into the epoxy matrix increases the epoxy modulus by 29%. CNF sulfurized, with no IF-WS2 but subject to sulfurization process with H2S, improves it nearly by 52% (Figure 8a). A possible explanation to the higher modulus values for the later might be related to the presence of sulfur and its effects on the epoxy-CNF interface adhesion. The three hybrid samples tested, CNF containing IF-WS2 particulates with 0.5, 1.5, and 3%, show much higher modulus than the rest, with values that almost double the one for bare epoxy (for hybrid CNF/IF-WS2 0.5%). While we observed a clear variation in sample performance for hybrids containing CNF and diverse IF-WS2 contents, the data does not show a clear correlation with the actual amount of IF added.

For all samples that included fillers we observed a more dramatic change in hardness than for the elastic modulus; the initial addition of carbon nanofibers had a large effect and increased the hardness by more than 114% (Figure 8b). The addition of sulfurized fibers further raised the hardness by another 38%. The sulfurization of the fibers appears to improve the ability of the fibers to bond with the epoxy matrix, further improving the interfacial strength. The increase in hardness could also be partially due to the inclusion of nickel sulfide particles. As shown in Figures 3 and 4, after the sulfurization step is carried out all nickel used as catalyst for the CNF growth transforms into Ni3S2 or Ni17S18. While these particles contribute to an increase in performance in this case, it is possible that other effects could appear under different situations.

Nickel sulfide contamination is an ongoing concern for the tempered glass industry. High temperature structures of nickel sulfide can develop in the glass manufacturing process and become included in tempered glass. This inclusion will result in a stress concentration and if the particle is of a certain size, the pane of glass will shatter under loading far below what is expected of the material [66]. Depending on the future inclusions of these materials into other matrixes, the possible impact of sulfide particles as stress concentrators should be considered and further studied. In the case of polymeric matrices, like the ones studied herein, no failure modes related to the existence of nickel sulfide are expected, since the working and processing temperatures do not reach those where the nickel sulfide high temperature phase appears (715 °C). In the case of hybrids in the absence of polymers, the porous carbon fiber structure is presumed to accommodate the volume expansion of the phase change and does not constitute a concern.

All the hybrid CNF/IF-WS2 samples showed a significant increment in hardness values, being CNF/IF-WS2 0.5% the most remarkable, with a 247% improvement over the pure epoxy hardness.

**Figure 8.** Mechanical properties of epoxy composites*.* All composites contained 99% epoxy and 1% loading of filler nanostructures: CNF, sulfurized CNF or hybrid CNF/IF-WS2. For the later, the hybrids contained mostly carbon fibers, with only 0.5, 1.5, and 3% of IF-WS2. (**a**) Modulus data and (**b**) Hardness values.

In order to fully understand the CNF/IF-WS2 epoxy composite mechanical properties observed, further study of their interfaces by IR or other spectroscopic techniques is recommended. Such data will help characterize the changes introduced by the inorganic components in the oxirane ring bands, the overall epoxy resin structure and its degree of polymerization.

The inclusion of two-dimensional graphene into the epoxy matrix in equal loadings than the ones described above, 1wt% filler in epoxy, by itself or as Graphene/IF-WS2, resulted in increased values of modulus and hardness (Figure 9). However, the improvement when compared to the 3D nanofiber structures is modest; the hybrid Graphene/IF-WS2 with 0.5% of IF-WS2 showed a 9% modulus increment over bare epoxy and a 47% increase in hardness. Histograms for the composites that contain CNF have been highlighted in blue and the ones containing graphene in red. From the Figure is clear that the samples based in 3D CNF architectures containing IF-WS2 present the highest values. During the composite fabrication, the graphene based samples seemed to be more difficult to disperse into the epoxy matrices than the CNF ones and inconsistencies in the sample distribution into the epoxy puck might have a detrimental effect on the mechanical properties values. Moreover, the fact that the graphene sheets were distributed randomly in the sample and not oriented in the (002) direction, where the strong covalent bonds are located, might be another point to be considered to explain the observed mechanical behavior.

As mentioned in Section 2.2, the use of Graphene/IF-WS2 prepared *in situ vs*. the components physical mixtures do not seem to present an advantage in terms of phases distribution or their microstructural characteristics. In terms of mechanical properties, the *in situ* route and the physical mixtures of Graphene/IF-WS2 are comparable; the values for modulus are almost the same and the hardness is higher for the former. This result might be related to the fact that IF particulates might easily disperse in between the graphene sheets without the need of a multistep process. It is worth noting that in the sample created *in situ*, the graphene structure has been exposed to H2S atmospheres, while in the physical mixture the graphene was pristine and never in contact with sulfurizing environments. Given that the improvement in modulus and hardness over neat epoxy observed in those samples was minimal, the present study did not further investigate the interfacial effects created by diverse functionalities in the graphene surface.

**Figure 9.** Mechanical properties of epoxy composites. with 1% loading of CNF, sulfurized CNF, Graphene, IF-WS2, CNT/IF-WS2 hybrids or Graphene/IF-WS2 hybrids. (**a**) Normalized Modulus data; (**b**) Normalized Hardness values.

Overall, the epoxy composite samples containing 3D CNF structures along IF-WS2 created *in situ*, showed drastic improvements in the Young's modulus and hardness values by the use of only 1% hybrid weight loadings. The carbon nanofiber inclusions seem to have a much greater impact in the mechanical properties of the composite than the graphene based counterparts for similar IF loadings.

The values for modulus and hardness improvements over the bare polymer mentioned above for CNF/IF-WS2 composites surpass the ones encountered for other hybrid polymeric nanocomposites incorporating IF-WS2 nanoparticles and a carbon phase [43–45]. However, in order to strictly compare mechanical properties for diverse fillers and polymeric matrices, data gathering should be performed under the same conditions and similar experimental setup to be valid. In particular, the use of nanoindentation instead of DMA tends to produce different values (usually larger for the former technique), as recently corroborated by Flores *et al*. [67]. Thus, comparison between the two techniques outcomes is not adequate given that no correlation between them exist to date.
