Hydrogen storage in rippled graphene: perspectives from multi-scale simulations

Exploring new perspectives for green technologies is one of the challenges of the third millennium, in which the need for non-polluting and renewable powering has become primary. In this context, the use of hydrogen as a fuel is promising, since the energy released in its oxidation per unit mass (~142 MJ/kg) is three times that released, on average, by hydrocarbons, and the combustion product is water (Ramage, 1983). Being hydrogen a vector of chemical energy, efficient conservation, and non-dispersive transportation are the main goals. Three issues must be considered to this respect: (i) storage capacity, (ii) storage stability, and (iii) kinetics of loading/release. Commercial technologies are currently based on cryo-compression or liquefaction of H2 in tanks. These ensure quite a high gravimetric density [GD, point (i)], namely 8-13% in weight of stored hydrogen, and a relatively low cost (Züttel, 2003). However, concerning points (ii) and (iii), these technologies pose problems of safety, mainly due to explosive flammability of hydrogen, and consequent unpractical conditions for transportation and use (Mori and Hirose, 2009). Therefore, research efforts are directed toward solid-state based storage systems (energy.gov, Bonaccorso et al., 2015). Interactions of hydrogen with materials are classified as physisorption, occurring with H2 by means of van der Waals (vdW) forces, or chemisorption, i.e., chemical binding of H leading to the formation of hydrides (Mori and Hirose, 2009), requiring dissociative(associative) chemi(de)sorption of H2. Intermediate nature interactions, sometimes called "phenisorption," can also occur between hydrogen electrons and the electrons of external orbital of metals. Indeed, stable and robust (light) metal hydrides (Sakintuna et al., 2007; Harder et al., 2011) are currently considered an alternative to tanks. Their main drawback is their high chemisorption and chemidesorption barrier, both many times the typical thermal energy, implying slow operational kinetics, which becomes acceptable only at very high temperature. Physisorption, conversely, generally results in barrierless and weak binding. It was considered as a storage mechanism in layered (Zhirko et al., 2007) or porous (Sastre, 2010) materials, and shown to be effective at low temperatures and/or high pressure. Therefore, it generally seems that if storage stability (ii) is improved then the loading/release kinetics (iii) is worsened. Graphene shows good potential to be an efficient hydrogen-storage medium (Tozzini and Pellegrini, 2013): carbon is among the lightest elements forming layered and porous structures, and graphene is probably the material with the largest surface to mass ratio. These two conditions are in principle optimal to produce high GD [point (i)]. In addition, the chemical versatility of carbon allows it to interact with hydrogen both by physisorption (in sp2 hybridization) and chemisorption (Goler et al., 2013a) (in sp3 hybridization). ["Phenisorption is also obtained in graphene by functionalization with metals (Mashoff et al., 2013)]. On the other hand, concerning points (ii) and (iii), pure graphene does not perform dramatically better than other materials. H2 easily physisorbs onto graphene layers or within multilayers, but it was theoretically shown (Patchkovskiim et al., 2005) that large GD (6-8%) are reached within multi-layered graphene at cryogenic temperatures, while the room temperature value is at best ~2-3%. This was confirmed by measurements (Klechikov et al., private communication), which also indicate that graphene does not perform better than other carbon based bulk materials, such as nanoporous carbon or carbon nanotubes. In all cases, a key parameter determining GD is the specific surface to volume ratio. Theoretical works also show that stability can be improved (and GD optimized) at specific interlayer spacing (~7-8 Å), due to a cooperative effect of vdW forces (Patchkovskiim et al., 2005). A similar effect is responsible for the accumulation of physisorbed hydrogen within graphene troughs at low temperatures (~100 K) observed in simulations (Tozzini and Pellegrini, 2011). On the other hand, hydrogen chemisorption on graphene produces graphane (Sofo et al., 2007), its completely hydrated alkane counterpart, stable at room temperature [point (ii)] and with 8.2% GD [point (i)]. Graphane, however, shares with other hydrides high chemi(de)sorption barrier (~1.5 eV/atom). As in other materials, physisorption has good kinetics (iii), and bad storage capacity (i), and stability (ii), while chemisorption has good (i) and (ii), and bad (iii). Graphene, however, displays extremely peculiar properties: a unique combination of strength and flexibility, and its bidimensionality. Therefore, it can be buckled on different scales down to nanometer (Fasolino et al., 2007; Wang et al., 2011), either statically, i.e., forming stable ripples as an effect of external constraints [i.e., compression or interaction with a substrate (Goler et al., 2013b)], or dynamically, sustaining traveling ripples, i.e., coherent transverse out-of-plane acoustic modes, also called ZA or flexural phonons (Lindsay and Broido, 2010; Xu et al., 2013). We explored the possibility of exploiting these properties in the context of hydrogen storage. We first evaluated the dependence of chemisorbed H stability on the local curvature by means of a density functional theory (DFT) based study (Tozzini and Pellegrini, 2011). This revealed a linear dependency of the binding energy on the local curvature, leading to an over-stabilization of the adsorbate of up to 1-2 eV on crests and corresponding destabilization within troughs (Figure 1A), which is also modified by the presence of already chemisorbed hydrogen (Rossi et al., submitted). Theoretical evaluations were confirmed by Scanning Tunneling Microscopy performed on naturally corrugated graphene grown on SiC (Goler et al., 2013a), which demonstrated the presence of hydrogen prevalently on the convexities. These observations lead to the idea that an inversion of curvature (from concavity to convexity) could detach chemisorbed hydrogen. As a matter of fact, this mechanism was demonstrated in a simulation in which the curvature inversion was realized dynamically by the passage of a ZA coherent phonon of ~2 nm wavelength and ~THz frequency (Figure 1B). The simulation shows that hydrogen undergoes associative desorption at room temperature upon curvature inversion, which can be called a "mechanical catalysis": the energy barrier is overcome by means of the energy provided by the traveling phonon.