Abstract
The storage of large quantities of hydrogen at safe pressures1 is a key factor in establishing a hydrogen-based economy. Previous strategies—where hydrogen has been bound chemically2, adsorbed in materials with permanent void space3 or stored in hybrid materials that combine these elements3—have problems arising from either technical considerations or materials cost2, 3, 4, 5. A recently reported6, 7, 8 clathrate hydrate of hydrogen exhibiting two different-sized cages does seem to meet the necessary storage requirements; however, the extreme pressures (
2 kbar) required to produce the material make it impractical. The synthesis pressure can be decreased by filling the larger cavity with tetrahydrofuran (THF) to stabilize the material9, but the potential storage capacity of the material is compromised with this approach. Here we report that hydrogen storage capacities in THF-containing binary-clathrate hydrates can be increased to 4 wt% at modest pressures by tuning their composition to allow the hydrogen guests to enter both the larger and the smaller cages, while retaining low-pressure stability. The tuning mechanism is quite general and convenient, using water-soluble hydrate promoters and various small gaseous guests.
The structure II (sII) hydrates constitute a large family of clathrates with an ideal unit cell 16S
8L136H2O, where the large (L) and small (S) cavities can be filled with guest molecules10. The 'solid solution' theory of van der Waals and Platteeuw, the classical approach to understanding clathrate behaviour, uses the basic expression:
which relates the free energy difference, 
w, between ice and a hypothetical empty hydrate framework to the minimum occupancy 
i of the hydrate cavities of type i required to render it stable; 
i are the number of cages of type i in the unit cell normalized by the number of water molecules. For sII hydrate, the 'best value' of
w of 884 J mol-1 requires the large cages to be filled to more than 99% for stability. This is consistent with the observation that many sII hydrates are known for which the stoichiometry is L17H2O within experimental error. The small cages can be occupied by a small guest, leading to a double hydrate, (2S)xL17H2O, generally stable to higher temperatures, where x is the fractional occupancy of the small cages, as recently illustrated9 for THF and H2 (x 
1, 1 wt% H2). The recently reported H2 clathrate is also sII with multiple occupancy of the cages (4 in L, 2 in S)6, 7, 8. Double hydrates represent an opportunity to engineer hybrid structures that combine H2 storage capacity with much less severe synthetic pressures.
We carried out initial studies on materials produced from 5.56 mol% THF solution in water, which gives a clathrate of composition THF17H2O when cooled below the melting point of 277.3 K. The THF hydrate was then pressurized with H2 gas at various pressures up to 120 bar, and examined for structure, cage populations and composition by powder X-ray diffraction (PXRD), Raman and NMR spectroscopy, and direct measurement of the H2 released on decomposition. From the PXRD results (Supplementary Fig. 1 and Supplementary Table 1), the material was confirmed to be a sII hydrate according to its phase behaviour (Supplementary Fig. 2). The Raman spectra (Fig. 1) show four transitions due to rotational fine structure associated with the high-pressure H2 gas, and a broad band that can be identified with H2 inside hydrate cages. The hydrate H2 line is shifted to lower frequency compared to the free gas11, as has also been observed for O2 and N2 hydrates12. 1H NMR spectroscopy was used to monitor H2 in the product of the reaction of H2 with perdeuterated THF hydrate (THF-d817D2O), so that only H2 signals and residual protons in water and THF would be observed. The spectra in Fig. 2 show a broad line at 4.3 p.p.m. that can be attributed to H2 in the small cavities of the double hydrate. Full loading of the small cavities of THF17H2O with hydrogen (2H2 per small cage) will result in a storage capacity of 2.1 wt% H2.
Figure 1: Raman spectra of the THF + H2 double hydrates.
![Figure 1 : Raman spectra of the THF |[plus]| H2 double hydrates. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact npg@nature.com](/nature/journal/v434/n7034/images/nature03457-f1.0.jpg)
Frozen 5.56 mol% THF solutions were stored in a refrigerator at 243 K for at least one day, and then ground to a fine powder ( 200 
m). The powdered solid was exposed to hydrogen gas (ultrapure carrier grade, 99.999%) at pressures from 56 to 120 bar and at a temperature of 270 K in a high-pressure Raman cell equipped with two sapphire windows and circular grooves for coolant. The Raman spectrum was obtained using a SPEX 1404p single grating Raman spectrometer, with a CCD detector cooled by liquid nitrogen, using the focused 488 nm line of an Ar-ion laser for excitation. The laser intensity was typically 300 mW. The H–H vibron peak at 4,128 cm-1 consists of both the sharp spectral line assigned to hydrogen gas and the broad peak due to hydrogen molecules stored in clathrate hydrate cages. The broad peak becomes higher when hydrate formation pressure increases in the stable region of the pressure–temperature diagram (Supplementary Fig. 2).
Figure 2: Magic angle spinning 1H NMR spectra of the THF + H2 double hydrates formed at 120 bar and 270 K as a function of concentration of THF.
![Figure 2 : Magic angle spinning 1H NMR spectra of the THF |[plus]| H2 double hydrates formed at 120|[thinsp]|bar and 270|[thinsp]|K as a function of concentration of THF. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact npg@nature.com](/nature/journal/v434/n7034/images/nature03457-f2.0.jpg)
The NMR samples were prepared from deuterated water (D2O, 99.9 at.% D) and THF (THF-d8, 99.5 at.% D) by the same procedure as for Raman experiments. The chemical shifts of D2O and THF-d8 were identified from the NMR spectrum of THF-d8 hydrate. Spectra were recorded using a Varian INOVA600 spectrometer, spin rate 12 kHz, pulse length 5 s, repetition time 15 s. The samples were transferred to an NMR rotor and analysed at 1 bar and 183 K. As the hydrate is not absolutely stable under these conditions, quantitative spectra cannot be obtained because of some decomposition, although a consistent picture can be constructed from the spectra that illustrate several features. The distribution of hydrogen molecules in both cages depends on temperature and pressure13.
Is it possible to obtain higher hydrogen loading, while keeping the H2 pressure at a reasonable value? Although double hydrates have been known for over a century, techniques for hydrate analysis in terms of guest distribution over the hydrate cage sites were developed only recently, so little effort has gone into attempts to tune hydrate compositions. In order to increase the hydrogen content of the hydrate, the hydrogen guest must also enter the large cavities of sII. The solid-solution model then requires that the large cage occupancies
L(THF) and
L(H2) become comparable, where the
L are given by Li = CLipi/(1 + CL1p1 + CL2p2); the C
Li are the Langmuir constants for the large cage, and the p
i are the partial pressures of the guest. As C
L(THF) is much larger than C
L(H2), this requires the partial pressure (concentration) of THF to be lowered considerably in order to place H2 clusters into the large cage.
Hence, a number of experiments were carried out at an H2 pressure of 120 bar where the THF concentration was decreased from 5.56 mol% down to 0.1 mol%. The compositions of the hydrates produced were obtained from measurements of the Raman band intensities. The H2 content of the clathrate produced at each THF loading was calculated from the integral intensity of its Raman band (after appropriate subtraction of intensity from the overlapping gas lines) relative to that of the 5.56 mol% sample and taking into account the different relative volumes of hydrate and ice. Extinction coefficients were assumed to be constant, based on the observation that the ratio of Raman intensity to volumetrically measured H2 content remains constant over the range of THF concentrations (Supplementary Material and Supplementary Fig. 3). The 5.56 mol% sample was taken to be fully loaded, with 2H2 per small cage and 2.09 wt% H2 content. Results are presented in Table 1 as H2:THF mole fractions and wt% H2 content of the hydrate phase (based on the model discussed below). The H2:THF mole ratio of the 'reference' 5.56 mol% sample is 4, whereas for samples with 4.0 and 2.0 mol% THF the mole ratio falls slightly below 4, indicating a small deficit of H2 in the small cages.
What is most interesting, however, is that at 1.0 mol% and down to 0.15 mol% the mole ratio increases above 4 to as high as 23 for the 0.15 mol% sample, indicating that large cages must also contain H2. The 1 mol% concentration of THF is approximately at the eutectic composition of THF hydrate and below this, solid ice Ih would be the first phase to appear, thus reflecting the fact that the concentration of THF has declined to the point where hydrate no longer forms (L = CLp/(1 + CLp) = <1). For sII hydrate to be stable relative to ice, the only condition that needs to be satisfied is that the occupancy of the large cage is close to 1, so
THF +
H2 1. This approach represents a general strategy for tuning hydrate compositions—at or below the eutectic composition of water-soluble sII hydrate formers, it is necessary to introduce other guests into the large cages of sII hydrate for stability.
At this stage we can consider the exact nature of the THF hydrate with high H2 content. We have concluded that some large cages contain THF and some have H2 clusters, while maintaining the double cage occupancy of the small cages. Thus the formation of double hydrate at low THF concentration (1.0–0.15 mol%) probably progresses as follows:
where (2H2)2 and (4H2)x represent clusters of two or four H2 molecules occupying small cages and large cages, respectively (as suggested from very-high-pressure studies of H2 hydrate6). THF hydrate formed in the first process (equation (2)) reacts with H2 to form a double hydrate (equation (3)), which is the kinetically limited product. Disproportionation of the double hydrate formed in equation (3) with the ice must then occur in order to allow H2 to occupy large cages and thus produce a thermodynamically stable double hydrate (equation (4)). 1H NMR spectroscopy was used to identify the distribution of guests over the cages in the double hydrate with high H2 content. When the concentration of THF reaches 1 mol% a new line appears at 0.15 p.p.m., which continues to grow with decreasing THF concentration. Considering the other experimental evidence for the increased loading of the sample, we assign this peak to H2 clusters in the large cages (Fig. 2).
We can then calculate the values of x in equation (4) and the effective wt% H2 for the clathrate formed from each THF concentration. Figure 3 shows a plot of wt% H2 stored in hydrate as a function of THF concentration—note that these represent a maximum wt% H2 content, based on the assumptions above. Cross-checks were made with a direct gas release measurement for the 5.56, 0.7 and 0.5 mol% samples, which gave respectively 2.1, 2.4 and 2.7 wt% H2 compared to 2.1, 2.6 and 3.0 wt% H2 from the Raman measurements. This can explain the increase in H2 content in the hydrate, and indeed we observe that the wt% H2 was 2.24 at 1.0 mol% increasing to 4.03 wt% at 0.15 mol% THF. At even lower THF concentration, 0.1 mol%, a H2-containing hydrate was no longer formed under the conditions used. We take this to mean that the combination of THF concentration and H2 pressure was insufficient to fill the large cages in the structure to produce a stable hydrate. When H2 was applied to frozen 0.2 mol% THF solutions at pressures between 90 and 140 bar, the H2 content reached a steady state for pressures higher than 100 bar. If the H2 in the large cages is present in the form of (H2)4 clusters, as has been proposed for pure H2 hydrate, about 60% of the large cages will be occupied by these, with the remainder expected to contain THF. Similar experiments have shown that, upon decreasing the THF concentration below 1.0 mol%, it is possible to introduce other guests such as CH4 into the large cages of sII hydrate. In recent 13C NMR studies of the sII double hydrate of CH4 and THF, two CH4 resonances at -4.5 p.p.m. (small cage) and -8.1 p.p.m. (large cage) were found, but the large cage resonance was only seen in solutions with concentrations below 1.0 mol% THF.
Figure 3: H2 gas content as a function of THF concentration, and a schematic diagram of H2 distribution in the cages of THF + H2 hydrate.
![Figure 3 : H2 gas content as a function of THF concentration, and a schematic diagram of H2 distribution in the cages of THF |[plus]| H2 hydrate. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact npg@nature.com](/nature/journal/v434/n7034/images/nature03457-f3.0.jpg)
(H2 gas content is calculated from g of H2 per g of hydrate, and expressed as wt%.) In region III, H2 molecules are only stored in small cages, while in region II both small and large cages can store H2 molecules. At the highly dilute THF concentrations of region I, H2 molecules can still be stored in both cages, but extreme pressures ( 2 kbar) are required to form the hydrates. Pure H2 clathrate (2H2)2(4H2)17H2O would have a 5.002 wt% H2 content.
Although the hydrate lattice appears to be tunable for higher levels of storage, up to 4.0 wt%, the reaction is so slow as to make the synthetic approach impractical. It is beyond the scope of this Letter to devise an industrial strategy for the efficient synthesis of H2-containing hydrate, but a promising approach is as follows: The reaction of the hydrate/ice mixture with H2 is limited by diffusion through a bulk solid phase. So, if the reacting phase can be dispersed so as to improve the surface area to volume ratio, improved kinetics should be possible. To demonstrate proof of principle, we have dispersed the aqueous phase on a silica bead support, and after exposure to H2 gas, followed the reaction kinetics by measuring the rate at which gas is consumed. Figure 4 shows that the reaction proceeds as expected and that the conversion is now complete in 1 h; this is a tremendous improvement from the bulk reaction, which took a week or more. We recognize that this is not a practical solution, as the product resides inside the silica gel, and the additional weight takes the material out of the required range as a hydrogen storage material. However, this does show that a dispersed phase will react in a reasonable time, giving a product that is a suitable storage material.
Figure 4: Formation/release kinetics of the H2 + THF double hydrate in the pores of silica beads.
![Figure 4 : Formation/release kinetics of the H2 |[plus]| THF double hydrate in the pores of silica beads. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact npg@nature.com](/nature/journal/v434/n7034/images/nature03457-f4.0.jpg)
At three THF concentrations (5.56, 0.7 and 0.2 mol%) the amount of H2 stored in the cages was measured using an isometric micro-syringe pump. Hydrates were maintained at 1 bar and 270 K for complete release of hydrogen after forming at 120 bar and 270 K. Hydrogen bubbles are observed during hydrate release (inset).
High resolution image and legend (58K)