Argon adsorption in open-ended single-wall carbon nanotubes

Thermodynamic and neutron-diffraction measurements combined with molecular dynamics simulation are used to determine the adsorption energies and the structure of argon condensed in the various adsorption sites of purified open-ended single-wall nanotube bundles. On the basis of these experiments and the simulation results, a consistent adsorption scenario has been derived. The adsorption proceeds first by the population of the walls inside the open nanotubes and the formation of one-dimensional Ar chains in the grooves at the outer surface of the bundles, followed by the filling of the remaining axial sites inside the nanotubes and the completion of a quasihexagonal monolayer on the outer surface of the bundle. The measurements also provide an estimate of the relative abundance of the various adsorption sites revealing that a major part of the adsorbed Ar is stored inside the open-ended nanotubes. Nanotube bundles generally show a certain degree of heterogeneity and some interstitial sites should be populated over a range of Ar chemical potential. However, for the sample used here, diffraction data and simulations suggest that heterogeneity is not a key feature of the bundles and there is little direct evidence of interstitial sites being populated.

Figure 1

Schematic section of a nanotube bundle containing $N=48$ nanotubes with an outer diameter $L=(17\pm 1)\mathrm{\AA }$. The bundle diameter is about 120 Å $(\approx \sqrt{N}L)$, the total number of hollows, grooves and interstitial channels is $48(N_H=N)$, $21(N_{\mathrm{G}}\approx 3\sqrt{N})$, and $73(N_{\mathrm{IC}}\approx 2N-N_{\mathrm{G}})$, respectively. Depending on the number of adsorbate rows that can be accommodated at the inner walls $\left(M_{\mathrm{T}}\right)$, as central rows $\left(M_{\mathrm{t}}\right)$, and on the outer convex surface of the peripheral nanotubes, $M_{\mathrm{S}}$, the maximum number of internal $(\mathrm{INT}=\mathrm{T},\mathrm{t})$ and surface $\left(\mathrm{S}\right)$ sites becomes $N_{\mathrm{INT}}=(M_{\mathrm{T}}+M_{\mathrm{t}})N_H$ and $N_{\mathrm{S}}=M_{\mathrm{S}}{N}_{\mathrm{G}}$, respectively. Note that not all tubes have to be open-ended and that only the widest interstitial channels may be accessible to the adsorbates (black circles on $\mathrm{G},\mathrm{S}$, IC, $\mathrm{T}$ sites, gray circle on an axial $\mathrm{t}$ site, drawn to the size of an Ar atom with 3.8 Å diameter).

Figure 2

Adsorbed amounts of Ar on the open-ended nanotube bundles as a function of equilibrium pressure. The adsorbed quantities were normalized to 1 g of the SWNT sample, which essentially contains carbon nanotube bundles but also amorphous and graphitized carbon. Two risers are observed between 0 and $5\mathrm{mmol}∕\mathrm{g}$ and between 5 and $7.5\mathrm{mmol}∕\mathrm{g}$, respectively. The first riser corresponds to the filling of high energy adsorption sites (internal sites, grooves, and the widest interstitial channels, see Fig. 1). The second riser (lower binding energy) is mainly assigned to adsorption in the $\mathrm{t}$ sites and on the curved graphene sheets forming the outer surface of the bundles.

Figure 3

Isosteric heat of adsorption $q_{\mathrm{st}}$ as a function of coverage (total adsorbed amount) obtained from Ar adsorption isotherms recorded at different temperatures. The two plateaus (indicated by the solid lines) correspond to the two risers in Fig. 2. For large doses $q_{\mathrm{st}}$ tends towards the latent heat of vaporization $q_{3\mathrm{D}}=6.5\mathrm{kJ}∕\mathrm{mol}$ (dashed line).

Figure 4

Neutron-diffraction spectra of the bare SWNT bundles (solid line) and upon adsorption of $5.1\mathrm{mmol}∕\mathrm{g}$ (dashed line, top of the first riser in Fig. 2) and $8.93\mathrm{mmol}∕\mathrm{g}$ (dotted line, top of the second riser in Fig. 2) of ${}^{36}\mathrm{Ar}$. The vertical lines indicate the positions of the Bragg diffraction peaks expected for a hexagonally close packing of the nanotubes with a bundle lattice spacing of 17 Å. The narrow peaks arise from the aluminum cell, from the carbon skeleton of the nanotube, and from graphitic carbon impurities.

Figure 5

Diffraction difference spectra (after subtraction of the bare SWNT spectrum) for several Ar doses (0.64, 1.27, 2.55, 3.83, 5.10, and $8.93\mathrm{mmol}∕\mathrm{g}$). (a) Small wave vector region $(Q⩽1.5{\mathrm{\AA }}^{-1})$. The intensity of the main (10) Bragg peak at $0.43{\mathrm{\AA }}^{-1}$ decreases monotonously with Ar coverage (curves from top to bottom). The dashed lines indicate the positions of the Bragg diffraction peaks expected for a hexagonally close packing of the nanotubes with a bundle lattice spacing of 17 Å. (b) Large wave vector region $(Q⩾1.5{\mathrm{\AA }}^{-1})$, showing the evolution of ordered Ar structures (increasing intensities with increasing Ar coverage, curves from bottom to top). The vertical lines indicate the peak positions expected for Ar condensed in linear chains (dotted lines) and in 2D hexagonal arrays (dashed lines). Sharp double-spiked features in the difference spectra, arising from the contributions of the cell and amorphous carbon, have been removed from the spectra.

Figure 6

Calculated diffraction spectra (after subtraction of the bare SWNT contribution) for several Ar doses adsorbed on a homogeneous 37 nanotube bundle (Mhomo2 model). (a) Small wave vector region $(Q⩽1.5{\mathrm{\AA }}^{-1})$, (b) large wave vector region $(Q⩾1.5{\mathrm{\AA }}^{-1})$. Solid line: after filling the $\mathrm{T}$ sites of 10 out of the 37 tubes; dashed line: all $\mathrm{T}$ and $\mathrm{G}$ sites are populated; dotted line: in addition, all the $\mathrm{S}$ sites and $\mathrm{t}$ sites are occupied. A good overall agreement between simulation and experiment (Fig. 5) is obtained.

Figure 7

Details of the evolution of the difference spectra in the region of the higher order bundle lattice peaks. (a) Experimental data, replotted from Fig. 5a with the solid, dashed, and dotted lines corresponding to Ar doses of 2.55, 5.10, and $8.93\mathrm{mmol}∕\mathrm{g}$, respectively. (b) Calculated diffraction difference spectra as in Fig. 6a with the solid, dashed, and dotted lines corresponding to the filling of the $\mathrm{T}$ sites of 10 out of the 37 tubes, all $\mathrm{T}$ and $\mathrm{G}$ sites, and all $\mathrm{T},\mathrm{G},\mathrm{S}$, and $\mathrm{t}$ sites, respectively. (c) Calculated diffraction difference spectra showing the influence of the adsorption on the IC sites on the diffraction pattern. Solid line: all IC sites filled; dashed line: all IC and $\mathrm{T}$ sites filled; dotted line: all IC, $\mathrm{T}$, and $\mathrm{G}$ sites filled; dash-dotted line: all IC, $\mathrm{T},\mathrm{G}$, and $\mathrm{S}$ sites filled.