Skip to main content
NCBI home page
UKPMC Funders Author Manuscripts logoLink to UKPMC Funders Author Manuscripts
. Author manuscript; available in PMC: 2018 Jun 1.
Published in final edited form as: J Phys Chem C Nanomater Interfaces. 2017 May 1;121(21):11747–11756. doi: 10.1021/acs.jpcc.7b00074

Hydrogen Chemical Configuration and Thermal Stability in Tungsten Disulfide Nanoparticles Exposed to Hydrogen Plasma

Alex Laikhtman 1,*, Gennady Makrinich 1, Meltem Sezen 2, Melike Mercan Yildizhan 2, Jose I Martinez 3, Doru Dinescu 4, Mariana Prodana 4, Marius Enachescu 4, Julio A Alonso 5, Alla Zak 1,*
PMCID: PMC5458142  EMSID: EMS72654  PMID: 28596812

Abstract

The chemical configuration and interaction mechanism of hydrogen adsorbed in inorganic nanoparticles of WS2 are investigated. Our recent approaches of using hydrogen activated by either microwave or radiofrequency plasma dramatically increased the efficiency of its adsorption on the nanoparticles surface. In the current work we make an emphasis on elucidation of the chemical configuration of the adsorbed hydrogen. This configuration is of primary importance as it affects its adsorption stability and possibility of release. To get insight on the chemical configuration, we combined the experimental analysis methods with theoretical modeling based on the density functional theory (DFT). Micro-Raman spectroscopy was used as a primary tool to elucidate chemical bonding of hydrogen and to distinguish between chemi- and physisorption. Hydrogen adsorbed in molecular form (H2) was clearly identified in all the plasma-hydrogenated WS2 nanoparticles samples. It was shown that the adsorbed hydrogen is generally stable under high vacuum conditions at room temperature, which implies its stability at the ambient atmosphere. A DFT model was developed to simulate the adsorption of hydrogen in the WS2 nanoparticles. This model considers various adsorption sites and identifies the preferential locations of the adsorbed hydrogen in several WS2 structures, demonstrating good concordance between theory and experiment and providing tools for optimizing of hydrogen exposure conditions and the type of substrate materials.

Introduction

The use of hydrogen in the renewable energy systems requires an effective, safe, and stable storage solution. Nanostructured materials such as inorganic nanotubes (INT) and inorganic fullerene-like (IF) nanoparticles (NP) are appealing because of their extremely high surface area and layered structure, where potentially many sites can either chemi- or physisorb hydrogen. A newly developed technology enables the synthesis and production of pure IF and INT phases of WS2 in commercial quantities.1,2 Therefore, WS2 INT and IF have been tested as possible candidates for hydrogen storage material with a significant industrial potential. These materials may allow hydrogen to be either chemi- or physisorbed inside their crystalline structure, inside hollow cores of IF/INT, on their surface or in the open interstitial pore spaces of IF/INT powder meshes. Recently, exposure to high pressure molecular hydrogen at 35-350 °C was found to have measurable but limited adsorption rate - up to 0.2 wt.% of hydrogen.3 This approach to use plasma activated hydrogen successfully resulted in significant increase of the hydrogen adsorption concentration. Indeed, treatment of the WS2-INT and WS2-IF by hydrogen activated in microwave (MW) plasma resulted in much higher values of the adsorbed hydrogen content, ~1 wt.%.3 This enhancement in the hydrogen adsorption concentration was attributed to more effective interaction of plasma-activated vs. high pressure molecular hydrogen with NP substrate surfaces. The increase of hydrogen adsorption was due to the stronger chemisorption of the atomic/ionized hydrogen formed in the MW plasma compared to weaker physisorption of molecular hydrogen, and to higher energy and momentum of the hydrogen molecules in the plasma environment. In addition, plasma-originated hydrogen ions and electrons may interact with the WS2 NP modifying existing defects and pores, or producing new ones, and so contribute to hydrogen binding, diffusion, and higher stability of its adsorption. An important consideration for choosing low energy hydrogen plasma as the activation method was in the high permeation of hydrogen/deuterium combined with the retention of the plasma properties along sample width. It was shown that hydrogen and deuterium molecules keep their highly vibrational excited states while passing from the outermost nanoparticles in the web to the innermost ones.4–6 This process should guarantee that most of the NP in the powder batch face similarly excited hydrogen molecules and undergo the same plasma treatment. This phenomenon is of particular importance ensuring homogeneous distribution of the adsorbed hydrogen through the exposed powder samples.

High hydrogen concentrations in the materials, proposed as media for hydrogen storage, is an important task, together with appropriate storage and release properties. However, these goals could be successfully accomplished only if the adsorption mechanism, chemical configuration and energy states of the adsorbed hydrogen are well understood. INT and IF of WS2 provide excellent tools for such a study owing to their simple structure, chemical and thermal stability, ease of modeling of their interaction with hydrogen molecules and ions, and possibility of precise calculation of the energy states of variously adsorbed species. The main novelty of this work is getting insight on the chemical configuration, by combination of experimental analysis methods and theoretical modeling. Our primary goal was to study the chemical configuration of the adsorbed hydrogen and to elaborate on its thermal stability inside the WS2 NP (either IF or INT) under ambient conditions. Whereas actual optimization of the technical parameters aiming to further increase the hydrogen adsorption rate in these specific materials is left out of the scope of the present work. A number of experimental techniques were applied for this purpose. Micro-Raman spectroscopy was extensively used to evaluate the chemical configuration of hydrogenated compounds due to its capability to detect chemical bonds involving hydrogen atoms.7–10 This method was highly sensitive and allowed to discover molecular hydrogen inside amorphous and crystalline solids.10 In addition, we used plasma of deuterium instead of hydrogen in several experiments to unambiguously prove that the measured features are due to the hydrogenation (deuteration) by plasma and not due to residual water or hydrogen retained from the synthesis of WS2 NP. Precise and well explained data for micro-Raman spectra of different types of WS2 NP has been recently reported.11,12 Transmission electron microscopy (TEM) analysis of NP after plasma treatment was used to accurately examine the pristine changes on the NPs’ surface and in their interlayer distances. To study the interaction of the INT-WS2 and IF-WS2 with activated H2 we also monitored them by electron diffraction (ED), and X-ray diffraction (XRD).

High thermal stability of the substrate materials (NP of WS2)13 allowed us to carry out thermogravimetric analysis (TGA) aiming to determine the weight percent and thermal stability of the adsorbed hydrogen.

A DFT model specifically developed for these WS2 NP materials was applied for the first time to calculate the energy of different adsorption sites and thus to determine the actual configuration of adsorbed hydrogen/deuterium.

Experimental

The IF-WS2 and INT-WS2 were synthesized from tungsten oxides by solid-gas reaction via the combined reduction-sulfidization process, at elevated temperatures of 750-900 °C, as it was described in detail elsewhere.2,14,15 This process, first, leads to the formation of the external cylindrical/spherical sulfide layer on the surface of the oxide whiskers/spherical NP, and then, to outwards-inwards gradual sulfurization of the oxide core resulting in hollow multiwall WS2-INT NP. The studied nanotubes were of 50-150 nm wide and ~20 μm long, and semi-spherical WS2-IF NP- of 50-250 nm in diameter.

Exposure of the WS2 NP to hydrogen/deuterium activated by radiofrequency (RF) plasma was performed in the plasma reactor dedicated to this purpose. The pure IF-, pure INT-WS2, and a ~1:1 mixture of INT with bulk WS2 platelets, in the form of either powder or a pressurized tablet ~70 mg each, were used as substrate materials for the RF hydrogen plasma treatment. Before the exposure to plasma, the substrate materials were in situ annealed at 300 °C for ~15 min (or until the pressure decreased to the initial value - prior to the annealing) in high vacuum (base pressure <2×10-5 Torr) to remove residual water and other contaminations. After cooling down to room temperature, the powders or the tablet were cleaned for 10 min by 200 W Ar plasma. This value of plasma power does not affect the structure of the WS2 NP, as it was previously reported for even higher Ar plasma power values (up to 400 W).16 After this cleaning procedure, the WS2 substrates were immersed into a RF plasma of either hydrogen or deuterium. The 400 W power RF plasma of hydrogen or deuterium was maintained during 30 min under gas pressure of 1 Torr, which resulted in minor temperature increase up to 35 °C. Later on the powders were evacuated from the plasma reactor chamber for different analyses. However, to evaluate the success of the hydrogenation process and to find the optimal conditions for the RF plasma, several sets of the hydrogenated/deuterated samples were in situ annealed and the relative pressure changes were measured as a function of the annealing temperature. This is unlike MW plasma hydrogenation previously reported,3 where the in situ pretreatment and post-treatment annealing were not possible.

To determine the chemical nature of hydrogen adsorbed in the NP as a result of the plasma treatment, micro-Raman measurements were performed in either Horiba LabRam 800 (CSSNT) or Renishaw InVia (SUNUM) microspectrometers. The Horiba instrument was particularly suitable as it is capable to measure spectral features in a range of up to 5000 cm-1. The green laser light of 532 nm was used for excitation being focused to a spot of 1-3 µm in diameter. Special precautions, like performing the measurements under continuous flow of nitrogen gas and limiting the laser power to <300 μW, were undertaken to prevent material heating under the laser beam,17 but it was not possible to fully avoid it. This could result in partial desoprtion of H2 under laser irradiation. Moreover, some oxidation of the substrate material was observed when irradiated for a longer time at higher laser intensities. For all substances the non-hydrogenated samples, used as references, were measured as well.

To elucidate the thermal stability of the adsorbed hydrogen and to determine its concentration, TGA measurements in a He atmosphere were performed in the temperature range of 25-500 °C using the SDT Q600 Simultaneous TGA/ DSC instrument.

To compare the morphological and structural quality of the materials before and after their treatment with hydrogen/deuterium plasma, TEM, and XRD measurements were carried out. ED pattern in the TEM was used to determine the evolution of the inter-planar spacing in the WS2 NP as a result of hydrogen/deuterium adsorption. For the TEM investigations, a JEOL ARM 200F TEM/STEM system was used.

With the exception of the in situ pressure measurements, all other analysis techniques were performed ex situ.

Results and Discussion

Plasma induced hydrogen adsorption and thermal stability

As explained above, to evaluate the success of the RF plasma hydrogenation/deuteration process one set of the samples was heated at a rate of ~20 °C/min to release the adsorbed gases at the end of the hydrogenation process and after gas evacuation. The pressure vs temperature dependence was recorded. In Figure 1 the results obtained for WS2-IF and WS2-INT, 30 min and 5 h after hydrogenation and deuteration, as well as the reference measurement of the empty plasma chamber are shown. The curves were normalized to the average sample mass of 60 g. Indeed, as seen from this figure the pressure increased even after deducting the pressure increase measured for the same experimental setup and after hydrogen plasma treatment, but without any substrate material (line 1). In such a way, the obtained pressure increase was entirely attributed to hydrogen release from the plasma-exposed WS2 NP and originates from this plasma-hydrogenation process only. For lines 2 and 6 two distinct peaks were observed for the samples measured within 30 min after plasma treatment, at 38 °C and around 80 °C. The pattern of the lines 4 and 5 is more complex as a result of lower intensity of the second (higher temperature) peak. The first feature (at the lower temperature) could be attributed to weekly bonded hydrogen, probably located on the surface, and the second one (at the higher temperature) is due to release of more strongly bound or deeper lying hydrogen. The relative amounts of released hydrogen were qualitatively compared by integration of the curves areas for the entire temperature range, estimated as 1:0.8:0.95:0.65:0.75for lines 2-6, respectively. It can be observed that the most substantial integral pressure increase was obtained for the deuterated and hydrogenated IF (lines 2 an 4). The curves for all the samples measured within 30 min revealed maximum hydrogen release at ~38-40 °C, while the peak obtained for IF is higher compared with the peak for INT. However, the release of deuterium from the INT was slower and the integrated area under the curve is not much slower than for IF. For all materials, the pressure started to grow (i.e., hydrogen started to desorb) at ~ 34 °C, while different temperatures as high as 300 °C were required to entirely release of all plasma-adsorbed hydrogen for various samples. It should be noted, however, that the measured temperature response to the heating was somewhat different between various samples due to some variations in their masses, grains sizes and the structures.

Figure 1.

Figure 1

Open in a new tab

Heating-induced pressure increase of the RF plasma hydrogenated and deuterated WS2 NP.

To elaborate on the stability of the plasma-adsorbed hydrogen or deuterium, the pressure increase upon heating was measured at the delay of 5 h after deuteration (Figure 1, line 3). The fact that most of D2 remained adsorbed and then released even after 5 h of vacuum storage, following the exposure process, is indicative on its high adsorption stability. The relative deuterium concentrations (total area under the curves) remained quite similar for the 30 min (line 2) as for 5 h delay (line 3): 1:0.8, respectively, although the onset of desorption shifted from 34 °C to 46 °C and the desorption rate was slower. In agreement with our previous assumption, this small but measurable hydrogen loss during the 5 h-delay vacuum storage, which is reflected in the shift in the onset temperature, can be attributed to desorption of non-bonded hydrogen molecules, probably on the very surface of the NP under continuous pumping in the chamber.

It is stressed that numerous adsorption-desorption cycles were performed and only negligible differences were recorded between the cycles.

One would like to integrate the pressure curves in Figure 1, by converting the pressure data to the mass flow, to determine absolute hydrogen concentrations in our WS2 materials. However, the large volume of the chamber (> 100 lit), constant flow of hydrogen and pumping during the measurements, the position of the pressure gauge far from the sample, along with unknown hydrogen desorption kinetics, and somewhat nonhomogeneous temperature response make these calculations inaccurate. As an alternative method for quantitative measurements of adsorbed hydrogen, the TGA was subsequently performed. The results of the TGA measurements for untreated (as synthesized) and hydrogenated INT-WS2 are presented in Figure 2 (lines 1,2). It can be observed that the reference sample lost more weight during the heating than the hydrogenated one. However, the desorption peak of the former appears just above 120 °C as seen from the derivative curve 2’ which, therefore, is attributed to evaporation of the residual water mostly and is consistent with previously reported TGA data.13 The maximum rate of the weight loss for the hydrogenated sample was obtained at 80 °C, as evident from the derivative curve 1’, which is at a similar position to the second peak (~80 °C) in the pressure measurements presented in Figure 1. It can be attributed to the adsorbed hydrogen. The absence of the peak associated with water desorption (120 °C) for the hydrogenated sample is not absolutely clear yet, since TGA measurements were performed ex situ, when both samples were exposed to air. We can only suggest at this point that WS2 NP gained hydrophobic properties by the fact that adsorbed hydrogen prevents water molecules from sticking the NP surface.

Figure 2.

Figure 2

Open in a new tab

TGA plots (lines 1,2) for reference (green line) and hydrogenated (blue line) WS2 NP. The appropriate derivative curves (lines 1’,2’) are plotted below.

The hydrogen concentrations estimated from TGA curves measured for all hydrogenated samples (not shown), were in the range of 0.4-0.55 wt. %. This value is more than twice the values measured for the WS2 materials exposed to unactivated high pressure hydrogen, as reported in our previous work, though lower than for the WS2 NP hydrogenated by MW plasma.3 We would like to stress again that the analysis methods of the MW plasma hydrogenated samples were completely different from those used in this study, which makes the comparison of the results not simple. In addition, in the MW plasma reactor the in situ measurements of hydrogen release were not technically possible.

For the deuterated samples, the TGA measurements showed nearly identical behavior as for hydrogenated WS2, and the estimated mass concentration of the plasma-adsorbed deuterium was about twice that of hydrogen, as the ratio of their molecular masses. This result confirms that released H2 is due to the initial hydrogenation and not because of the H2 residuals from the synthesis reaction, assuming that the amount of the adsorbed D2 and H2 molecules was the same under similar plasma conditions for the same type of WS2 NP.

It should be emphasized that the TGA measurements were performed ex situ, several days or weeks after the plasma-hydrogenation process, while the samples were stored at ambient atmosphere in closed vials. Therefore, we suppose that the real hydrogen concentration adsorbed by the samples immediately after plasma treatment is somewhat higher than the values listed above.

As evident from the results presented in Figure 1 and those of TGA, various types of WS2 NP adsorb somewhat different quantity of hydrogen or deuterium. We leave the explanation of this aspect for our future work and concentrate here on the chemical configuration of the adsorbed hydrogen/deuterium.

Determination of chemical configuration of plasma-adsorbed hydrogen/deuterium

To determine the chemical configuration of the adsorbed hydrogen, micro-Raman measurements were performed. The laser induced heating accounted for a significant problem, as it could result in immediate desorption of hydrogen from the irradiated area and oxidation of WS2. In the previous study of one of us,17 the temperatures in excess of 1000 °C were measured on the microcrystalline particles under the micro-Raman laser spot of the power similar to the values used in the current work. These results were obtained when heat dissipation mechanism was ineffective. Indeed using the laser power above 0.3 mW focused to the spot of ~1 µm resulted in oxidation of the studied materials as evidenced by strong W-O Raman peaks at ~900 cm-1 (data not shown),18 and visual color change (from brown-black to greenish-yellow) of the laser-irradiated area. It was shown previously that the oxidation process of WS2 NP occurs in air at the temperatures >300 °C.13 The oxidation of NP confirms that the temperature of the NP probed by the micro-Raman laser beam increased to >300 °C. At these conditions the hydrogen desorption is inevitable, as clear from Figure 1. In order to prevent or diminish these disturbances, a low laser power was used. However, this unfavorably affected the quality of the recorded spectra, bearing in mind a very low cross section of the Raman features related to molecular hydrogen.19,20

The micro-Raman scattering for all samples was measured in the spectral range of 0-4400 cm-1. The typical result obtained for both virgin and hydrogenated INT-WS2 in the window of 0-1200 cm-1 is depicted in Figure 3. Two characteristic modes of WS2 can be clearly observed: the E2g mode at 356 cm-1 and the A1g mode at 419 cm-1. This spectrum resembles those previously measured for our substrate materials and described in great detail elsewhere.13,14

Figure 3.

Figure 3

Open in a new tab

The typical micro-Raman spectrum measured for WS2 NP. No change following the hydrogenation was observed in this spectral region.

A narrower spectral region of 3400-4300 cm-1 for the hydrogenated NP of each type was measured and similar results were obtained for all types of the WS2 NP. As before, the micro-Raman spectra of reference, non hydrogenated materials were also recorded. In Figure 4 the spectra for the INT mixed with bulk WS2 are presented. A small peak centered at ~4150 cm-1 was observed for the hydrogenated samples only (line 2). This peak was the most prominent for the 1:1 mixture of the INT with bulk NP, the intermediate intensity was recorded for the IF-WS2, and the lowest though still measurable feature was observed for the pure INT-WS2. This feature was previously attributed to the H–H stretching mode in H2 molecules,10,19,20 and, therefore, its appearance in our spectra reveals that indeed hydrogen adsorption took place and is present in the molecular (H2) form. One should take into account a very small Raman cross section of this H-H mode as compared to the Raman peaks of WS2 and a strong fluorescence background above 3000 cm-1. We may suppose that H2 molecules were adsorbed on the surface of the NT, in the hollow core or being intercalated between the layers in the WS2 NP of all types.

Figure 4.

Figure 4

Open in a new tab

Micro-Raman spectra of WS2 NP zoomed to the 3400-4300 cm-1 region: (1) reference (non-hydrogenated), (2) RF-plasma hydrogenated, (3) RF-plasma hydrogenated, then vacuum annealed at 450 °C.

To finally prove that the Raman feature measured at 4150 cm-1 is indeed due to physisorbed H2 molecules, the sample was heated in vacuum to 450 ºC for several discrete periods lasting between 15 min to 2 h. The intensity of the H2 mode peak gradually decreased until nearly disappeared following the longest annealing time, as shown in the Figure 4 (line 3).

To determine the presence of chemisorbed hydrogen (H-S bonding) Raman scattering measurements were performed in the 1500–3000 cm-1 region (not shown). However, the features which could be attributed to H-S bonds21 never appeared, so indicating that no chemisorbed hydrogen was present in the WS2 NP, at least at the moment of performing the Raman measurements. Nevertheless, we suspect that atomic hydrogen or free hydrogen ions, which could adsorb or intercalate in between the layers of WS2 structure and create chemical bonds on the open edges or defects, desorbed as a result of environmental interaction, similar to that observed previously.22 Additional study is required to clarify this aspect.

Nevertheless, to unambiguously prove that micro-Raman spectroscopy is indeed capable to reveal molecular hydrogen in WS2, the deuterium gas in the RF plasma was used instead of hydrogen. To determine the chemical configuration of the adsorbed deuterium, micro-Raman measurements were performed in the spectral range of 2500-4000 cm-1, and the results are shown in Figure 5. Indeed, a small peak, now centered at ~2970 cm-1 was observed (Figure 5, line 1). This feature was previously attributed to the D–D stretching mode in deuterium (D2),19,23 and, therefore, its appearance in our spectra proves that deuterium is present in the molecular form as was in the case of hydrogen. No features related to D-S bonds were observed here too.24,25 Furthermore, RF plasma deuterated WS2 samples (IF and INT) were heated in vacuum to 450 ºC for up to 120 min, as for the hydrogenated samples. As seen from Figure 5 (line 2), the intensity of the ~2970 cm-1 peak decreased to nearly zero, supporting our assumptions that the observed mode corresponds to adsorbed D2 gas.

Figure 5.

Figure 5

Open in a new tab

Micro-Raman spectra of (1) deuterated and (2) subsequently vacuum-annealed WS2 NP.

The XRD spectra (data not shown), did not show measurable changes in the crystalline structure after RF hydrogen plasma treatments. This evidence is indicative of stability of the tubular and fullerene-like structures under plasma treatment. On the other hand, it is illustrative of the absence of systematic hydrogen intercalation in between the layers of WS2 samples. Nevertheless, the high resolution TEM image of the plasma-hydrogenated IF-NP depicted in Figure 6 shows that in some particles the interlayer distance was increased to more than 0.65 nm (up to 0.85 nm in a few isolated locations, shown by arrows) as compared to the 0.62 nm interlayer distance in the non-affected regions. This expansion was not observed in pristine NP but did occur in WS2 NP intercalated by foreign species.22 The ED pattern on the individual NP is also shown in Figure 7, confirming the expansion of the layers from 0.62 nm (3.22 1/nm) for the unexposed NP to 0.65 nm (3.09 1/nm) for the hydrogenated NP. The difference between XRD and TEM/ED data could be due to the sampling features used in both techniques, namely large amount of powder used in XRD gives average spectra from many NP, while TEM and ED analyze an individual NP. As a consequence, we suppose that not all NP or/and not all the layers were expanded. The intercalation of hydrogen between the WS2 layers can be the natural explanation for this expansion. The dynamics of layers expansion in intercalated WS2 NP stored in ambient conditions was previously discussed.22

Figure 6.

Figure 6

Open in a new tab

TEM image of the hydrogenated IF-WS2 NP. The arrows show the expansion of the interlayer spacing up to 0.85 nm.

Figure 7.

Figure 7

Open in a new tab

Electron diffraction pattern of the reference (left) and hydrogenated (right) IF-WS2 NP.

DFT modeling of the hydrogen configuration and adsorption sites

Density functional theory (DFT) calculations have been performed to model the adsorption of hydrogen on planar and curved layered WS2 nanostructures.26 These calculations help in the interpretation of the experimental results presented above. The Perdew-Burke-Ernzerhof (PBE) functional has been used for electronic exchange and correlation effects,27 and van der Waals interactions have been included through the perturbative DFT+D formalism of Grimme.28 Periodic boundary conditions were applied to model the infinite WS2 planar layer, the trilayer, the surface and the bulk crystal. The Brillouin zone for all the layered systems was sampled by means of a [4×4×1] Monkhorst-Pack grid,29 guaranteeing full convergence in energy and electronic density. In the (4×4) unit cell of all the layered systems considered, the unit cell in each WS2 layer, with a calculated size of 12.6×12.6 Å2, contains 48 atoms, showing perfect WS2 stoichiometry. The calculated inter-layer distances are 6.13 Å for the case of the trilayer, and 6.15 Å for the bulk. In the surface case, the optimized distance between layers near the surface is 6.14 Å. It should be noticed that the predicted interlayer distance in the bulk is in good agreement with the experimental value of 6.23 Å,30 the deviation being only 1.2 %.

We have also investigated a variety of WS2 NT: arm-chair and zig-zag single-walled NT, and (10,10)@(16,16) and (20,0)@(32,0) double-walled (DW) NT. The calculated distance between the inner and the outer walls in the DW-NT is 6.21 Å, slightly larger than the interlayer distances for planar multilayers (6.15 Å). This result is consistent with the experimental one,31 because the spacing between layers in multiwall NT was found to be 6.31 Å as compared to 6.23 Å in the bulk, and was attributed to the elastic strain. Comparison of the calculated interlayer distances in NT, 6.21 Å, with the experimental value of 6.31 Å shows a deviation of 1.6 %. Although the theoretical and experimental spacing values show small differences, the relative changes are in good agreement. To test the robustness of the structures obtained, and their thermal stability, DFT molecular-dynamics simulations were carried out by heating some NT up to 850 °C, and the structures were found to be stable.

The calculations of the adsorption of molecular hydrogen on the surface of different layered WS2 structures reveal two preferential adsorption configurations with similar adsorption energies: H2 molecules on top of a W atom, with the molecular axis lying parallel to the surface (configuration I) and perpendicular to the surface (configuration II). In these adsorption configurations, a triangular hollow formed by three S atoms permits the accommodation of physisorbed H2 molecules. The H2 adsorption energies and distances have been monitored as a function of H2 coverage, ranging between 0.0625 monolayers (ML) coverage (one H2 molecule per unit cell; the WS2 unit cell has 48 atoms or 16 molecules) up to full 1 ML coverage (16 H2 molecules per unit cell or one H2 molecule per WS2 molecule). The results for planar layered substrates are given in Table 1, where dads is the adsorption distance between the center of mass of the H2 molecule and the plane of S atoms. The results are the same for the WS2 layer, the trilayer, and the surface.

Table 1.

Adsorption energies Eads per H2 molecule, and adsorption distances dads, for adsorption configurations I and II. The results are the same for the WS2 layer, the trilayer, and the surface.

Configuration I Configuration II
H2 coverage Eads
(eV)		 dads
(Å)		 Eads
(eV)		 dads
(Å)
0.0625 ML 0.06 2.83 0.07 2.95
0.25 ML 0.06 2.84 0.07 2.96
0.5 ML 0.06 2.84 0.06 2.96
1 ML 0.05 2.86 0.06 2.98
Open in a new tab

The adsorption energies are positive; that is, molecular adsorption is favorable. As the H2 coverage increases, adsorption energies tend to decrease and adsorption distances tend to increase, although these changes are small. A relevant result is that the H2 adsorption on the topmost layer of the WS2 multilayers (trilayer and surface) induces a small increase of the distance between the topmost and the second WS2 layers, and this decoupling is enhanced by increasing the H2 coverage: the interlayer distances of 6.13 and 6.14 Å for the clean WS2 trilayer and the surface, respectively, increase up to values of 6.21 and 6.23 Å, respectively, for 1 ML coverage. Another interesting observation is that the strength of the adsorption energies is similar to that on graphene.32,33

Similar conclusions are obtained for DW-NT. The results reported in Table 2 correspond to low coverage of the outermost wall of the NT, one adsorbed molecule per NT unit cell. For H2 adsorption on the innermost wall of the DW-NT, that is, on the hollow internal surface, the adsorption energies are practically the same as for adsorption on the external walls. However, a slight increase in the adsorption distances is observed. By increasing the H2 content (up to 0.5 ML) for external and internal adsorption, no significant changes are found compared to the case of the planar layers.

Table 2.

Adsorption energies Eads per molecule and adsorption distances dads for the most stable H2 adsorption configurations (I and II, as in Table 1) on the external surface of WS2 DW-NT. The coverage is low, one molecule adsorbed per NT unit cell.

NT Configuration I Configuration II
Eads
(eV)		 dads
(Å)		 Eads
(eV)		 dads
(Å)
(10,10)@(16,16) 0.06 2.83 0.07 2.95
(20,0)@(32,0) 0.06 2.84 0.06 2.96
Open in a new tab

The calculations indicate that molecular hydrogen can be adsorbed on the surface of the NP and on the outermost and innermost walls of the NT. However, intercalation between the WS2 layers is unfavorable. This conclusion has been obtained by simulating the intercalation of the trilayer with increasing amounts of molecular hydrogen; more specifically, H2 molecules were placed in between the first and second layers of the trilayer (see Figure 8).

Figure 8.

Figure 8

Open in a new tab

Optimal structures for intercalation of molecular hydrogen between the first and the second layers of the WS2 trilayer: 0.0625 ML H2 content (left panel); 0.25 ML H2 content (right panel).

Table 3 summarizes the intercalation energy per molecule and the distance between the first and the second WS2 layers, as a function of intercalate concentration. The calculations revealed that the distance between the first and the second layers increases substantially as the hydrogen content increases, from 6.21 Å in the case of no intercalation to values close to 8.6 Å for intercalation of a full hydrogen monolayer. The energy of intercalation is negative in all cases; that is, intercalation is unfavorable due to the elastic energy needed to expand the interlayer distances. Interestingly, an interlayer expansion up to 9 Å in some regions was revealed in the TEM micrographs (Figure 6). Such an expansion was never observed in virgin WS2 NP. Consequently, the observed interlayer expansion can be interpreted as indicating that some molecular hydrogen intercalates in the NP under the plasma conditions, albeit inducing substantial local distortions on the IF structure. In fact, we find no other mechanism able to explain a local increase in the interlayer distances so large (more than 30%), and at the same time so close to the DFT prediction. Such a substantial expansion can only be due to stuffing some material between the WS2 layers. The plasma conditions provide the energy required for the H2 molecules to intercalate in the following way: In the plasma formed sheath on the surface of the sample, the plasma ions are accelerated by the potential drop of ~3Te, where Te is the electron temperature: ~2 eV in the conditions similar to those used by us.34 These accelerated ions collide with neutrals and form an energetic mixture of ions-neutrals flowing toward the sample surface.

Table 3.

Distance between the first and the second layers, dL1-L2, and intercalation energy per molecule, Eint/H2, for intercalation of molecular hydrogen between the first and the second layers of the WS2 trilayer. Results are given for several H2 concentrations.

H2 content dL1-L2
(Å)		 Eint /H2
(eV)
0.0625 ML 6.21 -2.34
0.25 ML 7.51 -1.01
0.5 ML 8.17 -0.48
1 ML 8.61 -0.19
Open in a new tab

The concentration of stored hydrogen achieved in the RF experiments, estimated from the TGA measurements, is 0.4-0.55 wt.%. How those concentrations can be understood with the help of the DFT model calculations, considering that only the outer layers and the hollow cores could adsorb molecular hydrogen? By taking the simplest model of a single WS2 layer with one side of the layer fully covered by H2 (1 ML), the concentration of hydrogen is 0.81 wt.%. But, for 1 ML of H2 on the upper surface of the WS2 bilayer the concentration decreases to 0.40 wt.%, to 0.27 wt.% for a trilayer, and so on. In other words, the hydrogen concentration decreases quickly for n-layer systems, and the same occurs for multiwall NT. Consequently, additional adsorption mechanisms are required to explain the measured adsorption.

One plausible mechanism is that the plasma conditions force the intercalation of some molecular hydrogen, locally distorting the structure of the nanoparticles, as discussed above. The effects of that forced intercalation become reflected in the regions showing increased interlayer distances detected by TEM (see Figure 6).

Another plausible method to obtain higher hydrogen concentration could be H2 adsorption on planar-like nanostructures and “small” INT (single- to five-layered nanotubes) formed by high power plasma treatment of “big” multiwall INT. Previous work by one of us showed that this occurs when thick WS2 NP and NT are subject to high-power irradiation.16 “Daughter” NT and platelets, one to three layers thick, were observed to form in those experiments. In the current study, however, TEM analysis of the powders after plasma treatment did not show evidences for the formation of new morphological structures. The reason can be the different plasma conditions used in the previous work: higher power of the plasma (600 W) and heavier (Ar) plasma indeed could result in single wall NT formation.

A third plausible mechanism is the condensation of more than one hydrogen monolayer on the surface of planar platelets, or on the external surfaces and hollow internal surfaces of NT. As a test, we performed simulations for the covering of a single WS2 layer with more than 1 ML of H2. The starting configuration was formed by H2 molecules on top of all W and S atoms (this corresponds to 2 ML coverage in the notation used above). The molecules were initially on the same plane. But this turns out to be an unfavorable configuration because the H2 molecules are too close to their neighbor molecules, and structural relaxation of the system allowed the adsorbed molecules to optimize their positions forming 2 ML, as shown in Figure 9. The H2 layer closest to the WS2 layer is at a distance of 2.84 Å from the plane of S atoms, and the top H2 layer is separated from the first one by a distance of 2.52 Å. The molecules of the first layer sit on top of the W atoms, while the positions of the molecules of the top H2 layer correspond to sites above S atoms. In all cases, the molecular axes are nearly perpendicular to the plane. The calculated adsorption energy, 0.06 eV per molecule, is nearly the same as the adsorption energy per molecule for a single H2 ML (see Table 1).

Figure 9.

Figure 9

Open in a new tab

Optimized structure for 2 ML of H2 molecules adsorbed on top of a WS2 layer.

The mechanisms discussed above are not mutually exclusive, and could add up to explain the amount of stored hydrogen observed in the experiments. In fact, an additional possible mechanism could also be offered. Neutral H atoms and H+ ions are generated in the RF hydrogen plasma, in addition to molecular hydrogen. Intercalation of atomic hydrogen leads to positive intercalation energies (see Table 4); that is, the intercalated systems are stable. The intercalated H atoms attach to S atoms and induce only a very small increase in the distance between adjacent WS2 layers, in contrast with the intercalation of molecular hydrogen (compare Tables 3 and 4). However, the Raman experiments performed did not detect H-S (or D-S) bonds, which would reveal the presence of adsorbed atomic hydrogen. So we have to admit that the amount of atomic hydrogen intercalated in between the layers of platelets, IF, and NT is non-significant during the Raman measurements, perhaps because the hydrogen atoms have recombined to form hydrogen molecules, which in turn do not intercalate. In addition, absence of the H-S and D-S bonds related features in the Raman measurements could be a result of desorption under Raman beam heating or even samples aging (time required to transformation of the sample after plasma treatment).

Table 4.

Distance between the first and second layers, dL1-L2, and intercalation energy per atom, Eint /H, for the intercalation of atomic hydrogen between the first and the second layers in the WS2 trilayer. Results are given as a function of increasing H content.

H content dL1-L2
(Å)		 Eint / H
(eV)
0 ML 6.14 -
0.25 ML 6.15 1.15
0.5 ML 6.17 1.32
0.75 ML 6.21 1.36
1 ML 6.25 1.38
Open in a new tab

Conclusions

In summary, the configuration and energy states of the hydrogen species adsorbed in the NP of WS2 were elucidated by using experimental analytical methods and a DFT theoretical model. The micro-Raman spectroscopy detected the H-H and D-D stretching modes at 4150 cm-1 and 2970 cm-1, respectively, indicating that the hydrogenation by the RF plasma results in physisorption of molecular hydrogen/deuterium. No evidences of H-S and D-S bonds were found by micro-Raman analysis. The adsorbed hydrogen is found to be quite stable at room temperature conditions, both in air and vacuum. The DFT calculations, for the first time applied to the studied system, show that, the physisorbed hydrogen molecules are adsorbed, probably as one or more layers, on the very surface of the NP (e.g., on the top-most layer and the innermost layer of the hollow NT). DFT calculations performed to analyze the possible intercalation of molecular hydrogen between WS2 layers delivered insertion energies indicating that such intercalation is unfavorable. However, we suppose that partial intercalation of molecular hydrogen could take place under the RF plasma conditions. The forced intercalation causes local damage and a substantial expansion of the interlayer distances, which we have observed by TEM.

On the other hand, DFT calculations for atomic and ionic hydrogen reveal that intercalation would be energetically favorable. No direct evidence was found for intercalation of atomic hydrogen since it may have followed either or both processes: de-intercalation and shrinking caused by environmental affects before being measured by micro-Raman. Additional studies are required for identification of possible intercalation of atomic hydrogen.

The results obtained by the DFT calculations can be used now for further optimization of hydrogen exposure conditions, e.g., using plasma with higher concentration of atomic and ionic hydrogen, and lowering the temperature of the RF-plasma enhanced process to intensify the adsorption process. In addition, as the adsorption of molecular hydrogen is found to be favorable on the very surface, single wall NT can be used instead to increase the hydrogen weight percent.

Acknowledgments

This work was partially supported by the Israel-Korea grant of Israeli and Korean governments, by ENIAC European grant “Lab4MEMS” Nr. ENIAC 3 /2013-UEFISCDI, and by European COST actions MP1103, MP1302, and CM1301. The work of JAA was supported by MINECO (grant MAT2014-54378-R) and Junta de Castilla y León (grant VA050U14). JIM acknowledges funding from the ERC Synergy Grant ERC-2013-SYG-610256 NANOCOSMOS, and computing resources from CTI-CSIC. AZ acknowledges the support provided by the INNI-FTA grant and Pazy foundation. The authors thank Amnon Fruchtman (Plasma Lab., HIT) for helpful discussion of the RF plasma exposure experiments and analysis of the pressure curves.

References

  • 1.Feldman Y, Zak A, Popovitz-Biro R, Tenne R. New Reactor for Production of Tungsten Disulfide Hollow Onion-Like (Inorganic Fullerene-Like) Nanoparticles. Solid State Sci. 2000;2:663–672. [Google Scholar]
  • 2.Zak A, Sallacan-Ecker L, Margolin A, Feldman Y, Popovitz-Biro R, Albu-Yaron A, Genut M, Tenne R. Scaling Up of the WS2 Nanotubes Synthesis. Fuller Nanotub Car N. 2010;19:18–26. [Google Scholar]
  • 3.Laikhtman A, Michaelson S, Hoffman A, Kim TK, Moon HR, Zak A. Using Hydrogen Activated by Microwave Plasma vs. Molecular Hydrogen for Hydrogen Storage in Tungsten Disulfide Inorganic Nanotubes. Int J Hydrogen Energ. 2014;39:9837–9841. [Google Scholar]
  • 4.Cacciatore M, Rutigliano M. Dynamics of Plasma–Surface Processes: E–R and L–H Atom Recombination Reactions. Plasma Sources Sci Technol. 2009;18:023002. [Google Scholar]
  • 5.Nishijima D, Sugimoto T, Iwakiri H, Ye MY, Ohno N, Yoshida N, Takamura S. Characteristic Changes of Deuterium Retention on Tungsten Surfaces due to Low-Energy Helium Plasma Pre-Exposure. J Nucl Mater. 2005;337–339:927–931. [Google Scholar]
  • 6.Glazunov GP, Andreev AA, Baron DI, Causey RA, Hassanein A, Kitayevskiy KM, Konotopskiy AL, Lapshin VI, Neklyudov IM, Patokin AP, et al. Hydrogen Permeability and Erosion of W-Pd Bimetallic Systems. Fusion Eng Des. 2006;81:375–380. [Google Scholar]
  • 7.Ross DJ, Halls MD, Nazri AG, Aroca RF. Raman Scattering of Complex Sodium Aluminum Hydride for Hydrogen Storage. Chem Phys Lett. 2004;388:430–435. [Google Scholar]
  • 8.Weselucha-Birczyńska A, Babel K, Jurewicz K. Carbonaceous Materials for Hydrogen Storage Investigated by 2D Raman Correlation Spectroscopy. Vib Spectrosc. 2012;60:206–211. [Google Scholar]
  • 9.Mosquer E, Diaz-Droguett DE, Carvajal N, Roble M, Morel M, Espinoza R. Characterization and Hydrogen Storage in Multi-Walled Carbon Nanotubes Grown by Aerosol-Assisted CVD Method. Diam Relat Mater. 2014;43:66–71. [Google Scholar]
  • 10.Panella B, Hirscher M. Raman Studies of Hydrogen Adsorbed on Nanostructured Porous Materials. Phys Chem Chem Phys. 2008;10:2910–2917. doi: 10.1039/b719678d. [DOI] [PubMed] [Google Scholar]
  • 11.Gartsman K, Kaplan-Ashiri I, Tenne R, Rafailov PM, Thomsen C. Micro Raman Investigation of WS2 Nanotubes. AIP Conf Proc. 2005;786:349–352. [Google Scholar]
  • 12.Rafailov PM, Thomsen C, Gartsman K, Kaplan-Ashiri I, Tenne R. Orientation Dependence of the Polarizability of an Individual WS2 Nanotube by Resonant Raman Spectroscopy. Phys Rev B. 2005;72:205436. [Google Scholar]
  • 13.Schuffenhauer C, Wildermuth G, Felsche J, Tenne R. How Stable are Inorganic Fullerene-Like Particles? Thermal Analysis (STA) of Inorganic Fullerene-Like NbS2, MoS2, and WS2 in Oxidizing and Inert Atmospheres in Comparison with the Bulk Material. Phys Chem Chem Phys. 2004;6:3991–4002. [Google Scholar]
  • 14.Feldman Y, Frey GL, Homyonfer M, Lyakhovitskaya V, Margulis L, Cohen H, Hodes G, Hutchison JL, Tenne R. Bulk Synthesis of Inorganic Fullerene-Like MS2 (M = Mo, W) from the Respective Trioxides and the Reaction Mechanism. J Am Chem Soc. 1996;118:5362–5367. [Google Scholar]
  • 15.Feldman Y, Lyakhovitskaya V, Tenne R. Kinetics of Nested Inorganic Fullerene-Like Nanoparticle Formation. J Am Chem Soc. 1998;120:4176–4183. [Google Scholar]
  • 16.Brüser V, Popovitz-Biro R, Albu-Yaron A, Lorenz T, Seifert G, Tenne R, Zak A. Single- to Triple-Wall WS2 Nanotubes Obtained by High-Power Plasma Ablation of WS2 Multiwall Nanotubes. Inorganics. 2014;2:177–190. [Google Scholar]
  • 17.Laikhtman A, Hoffman A. Laser Power Effects on the Raman Spectrum of Isolated Diamond Chemical Vapor Deposition Particles. J Appl Phys. 1997;82:243–248. [Google Scholar]
  • 18.Baserga A, Russo V, Di Fonzo F, Bailini A, Cattaneo D, Casari CS, Li Bassi A, Bottani CE. Nanostructured Tungsten Oxide with Controlled Properties: Synthesis and Raman Characterization. Thin Solid Films. 2007;515:6465–6469. [Google Scholar]
  • 19.Hartwig CM, Vitko J., Jr Raman Spectroscopy of Molecular Hydrogen and Deuterium Dissolved in Vitreous Silica. Phys Rev B. 1978;18:3006–3014. [Google Scholar]
  • 20.Schmidt BC, Holtz FM, Bény J–M. Incorporation of H2 in Vitreous Silica, Qualitative and Quantitative Determination from Raman and Infrared Spectroscopy. J Non-Cryst Solids. 1998;240:91–103. [Google Scholar]
  • 21.Thompson SD, Carroll DG, Watson F, O'Donnell M, McGlynn SP. Electronic Spectra and Structure of Sulfur Compounds. J Chem Phys. 1966;45:1367–1379. [Google Scholar]
  • 22.Zak A, Feldman Y, Lyakhovitskaya V, Leitus G, Popovitz-Biro R, Wachtel E, Cohen H, Reich S, Tenne R. Alkali Metal Intercalated Fullerene-Like MS(2) (M = W, Mo) Nanoparticles and Their Properties. J Am Chem Soc. 2002;124:4747–4758. doi: 10.1021/ja012060q. [DOI] [PubMed] [Google Scholar]
  • 23.Centrone A, Siberio-Perez DY, Millward AR, Yaghi OM, Matzger AJ, Zerbi G. Raman Spectra of Hydrogen and Deuterium Adsorbed on a Metal–Organic Framework. Chem Phys Lett. 2005;411:516–519. [Google Scholar]
  • 24.Anderson A, Binbrekt S, Tang HC. Raman and Infrared Study of the Low Temperature Phase of Solid H2S and D2S. J Raman Spectrosc. 1977;6:213–220. [Google Scholar]
  • 25.Sakashita M, Fujihisa H, Yamawaki H, Aoki K. Molecular Dissociation in Deuterium Sulfide under High Pressure: Infrared and Raman Study. J Phys Chem A. 2000;104:8838–8842. [Google Scholar]
  • 26.Baroni S, Corso AD, Gironcoli S, Giannozzi P. QUANTUM ESPRESSO Package. 2005 in www.quantum-espresso.org.
  • 27.Perdew JP, Burke K, Ernzerhof M. Generalized Gradient Approximation Made Simple. Phys Rev Lett. 1997;78:3865–3868. doi: 10.1103/PhysRevLett.77.3865. [DOI] [PubMed] [Google Scholar]
  • 28.Grimme SJ. Semiempirical GGA-Type Density Functional Constructed with a Long-Range Dispersion Correction. J Comput Chem. 2006;27:1787–1799. doi: 10.1002/jcc.20495. [DOI] [PubMed] [Google Scholar]
  • 29.Monkhorst HJ, Pack JD. Special Points for Brillouin-Zone Integrations. Phys Rev B. 1976;13:5188–5192. [Google Scholar]
  • 30.Wilson JA, Yoffe AD. The Transition Metal Dichalcogenides Discussion and Interpretation of the Observed Optical, Electrical and Structural Properties. Adv Phys. 1968;18:193–335. [Google Scholar]
  • 31.Zak A, Sallacan-Ecker L, Margolin A, Genut M, Tenne R. Insight into the Growth Mechanism of WS2 Nanotubes in the Scaled-Up Fluidized Bed Reactor. Nano. 2009;4:91–98. [Google Scholar]
  • 32.Cabria I, López MJ, Alonso JA. The Optimum Average Nanopore Size for Hydrogen Storage in Carbon Nanoporous Materials. Carbon. 2007;45:2649–2658. [Google Scholar]
  • 33.Cabria I, López MJ, Alonso JA. Hydrogen Storage Capacities of Nanoporous Carbon Calculated by Density Functional and Møller-Plesset Methods. Phys Rev B. 2008;78:075415. [Google Scholar]
  • 34.Geng ZC, Xu Y, Yang XF, Wang WG, Zhu AM. Atomic Hydrogen Determination in Medium-Pressure Microwave Discharge Hydrogen Plasmas via Emission Actinometry. Plasma Sources Sci Technol. 2005;14:76–82. [Google Scholar]

RESOURCES