Isolation and evolution of labile sulfur allotropes via kinetic encapsulation in interactive porous networks

Reported here are the isolation and direct observation of extremely reactive S2 and its conversion into bent-S3 via a cyclo-S3 2+ intermediate on interactive sites in porous coordination networks.


Introduction
Cryogenic trapping methods, coupled with spectroscopy or crystallography, have been widely used to investigate transient chemical species (Whittle et al., 1954;Misochko et al., 2003;Kawano, 2014;Edman et al., 1999), but these methods do not always allow the observation of very labile reactive intermediates. To circumvent this problem, we propose encapsulation of transient species in an interactive porous network under non-equilibrium conditions. The kinetic trapping method is widely used in cryogenic trapping (Whittle et al., 1954;Mü ck et al., 2012), although there is no report using porous coordination networks and no direct X-ray observation has yet been achieved. The method using porous coordination networks might provide a unique way for in situ observation of very labile chemical intermediates and for the following of chemical reactions Many fascinating guest encapsulation studies have been performed in the past using porous coordination networks (Matsuda, 2013;Cook et al., 2013;Kitagawa & Uemura, 2005;Eddaoudi et al., 2001;Fé rey, 2008;Peterson et al., 2014;Ohmori et al., 2005). Most of these studies, however, have only been carried out in conditions of thermodynamic equilibrium, which makes the observation of transient intermediates hardly possible (Kubota et al., 2014;Ikemoto et al., 2014;Kawamichi et al., 2009). Although time-resolved techniques offer attrac-tive alternatives, they generally require the reactions to be reversible (Ohashi, 1998). Our approach involves the stabilization of transient species via the active sites located in the channels of porous coordination networks. Here we report the first direct X-ray observation of extremely reactive S 2 species and their conversion towards bent-S 3 via cyclo-S 3 2+ on an interactive site in a channel of a porous coordination network.
Sulfur has a very rich chemistry, with around 30 allotropes known to date, although the transient nature of the smaller allotropes makes their isolation and characterization very challenging (Peramunage & Licht, 1993;Evers & Nazar, 2013;Xin et al., 2012;Meyer, 1976;Steudel & Eckert, 2003;. In a recent study, we reported the direct observation of bent-S 3 (trisulfur or thiozone) through encapsulation in a Znbased porous coordination network . The crystal structure of the network-S 3 complex revealed important interactions between S 3 and the network iodides of such strength that release of the S 3 molecules only took place at high temperatures (500 K). We suspect that S 3 encapsulation might have taken place via a 'ship-in-a-bottle' type of mechanism (Ichikawa et al., 1991;Rau et al., 1973): first the smaller S 2 (disulfur) enters the pores of the network and then it converts to S 3 (trisulfur) because S 3 is more stable than S 2 ; however, this mechanism remains to be proven. We have also reported the synthesis of two porous coordination networks of CuI with tetra-4-(4-pyridyl)phenylmethane (TPPM) with fascinating properties (Kitagawa et al., 2013). The kinetic product of the synthesis is network 1 (Fig. 1a), [(CuI) 2 (TPPM)] n , which contains molecular-sized channels with accessible iodide sites. These iodide sites are highly interacting and can adsorb molecules such as I 2 through chemisorption (Kitagawa et al., 2013). The thermodynamic product of the synthesis is network 2 (Fig. 4a), [(Cu 2 I 2 )-(TPPM)] n , which contains smaller one-dimensional channels with no exposed iodide sites. In network 2, only physisorption of I 2 is possible within the hydrophobic one-dimensional channel, because the iodide sites are located in small cavities that are poorly accessible. The channels of networks 1 and 2 have the precise molecular dimensions needed for trapping small molecules, with the added advantage of the interacting iodide sites in network 1. Network 1 can accommodate S 2 (5.8 Â 3.6 Â 3.6 Å ) or S 3 (6.8 Â 4.6 Â 3.6 Å for the bent form, 5.8 Â 5.5 Â 3.6 Å for the cyclo form) because of its channel size of 5.8 Â 5.5 Å . In contrast, network 2 can accommodate only S 2 , because of its channel size of 4.0 Â 3.9 Å . Because we expect strong interactions between small sulfur allotropes and the iodide sites of the networks (see the supporting information), these materials may serve as traps for labile sulfur intermediates. In order to isolate the most reactive species, we consciously arrested the encapsulation process before it reached equilibrium via cooling (kinetic trapping); we aimed to observe the 'ship-in-a-bottle' conversion from reactive S 2 to the more dynamically stable S 3 by heating.

Kinetic trapping of sulfur gas
Sulfur gas was encapsulated in networks 1 and 2 under kinetic conditions; an excess amount of elemental sulfur and desolvated network 1 or 2 were placed at different sites of a zigzag shaped glass tube (see Fig. S1 in the supporting information). The glass tube was then sealed in a vacuum ($10 À6 Torr; 1 Torr = 133.322 Pa) and heated in a flame at the site containing the sulfur. The zigzag tube was sufficiently long, and the sulfur and the network thus sufficiently separated, that high temperatures could be reached at the sulfur site while the network was kept at room temperature, creating a sharp temperature gradient. Shortly after heating the sulfur powder, the yellow crystals of network 1 turned dark yellow, whereas the crystals of network 2 did not display a colour change.

Direct observation of transient small sulfur by X-ray diffraction
Within 5 min of the colour change, a single crystal of network 1 was mounted on a goniometer and X-ray diffraction data were collected at 250 K. Because diffuse scattering was observed at 250 K (see Fig. S4 in the supporting information), the crystal structure was solved making use of Bragg diffractions only (see the supporting information). On the basis of a Laue check, and after careful consideration of various crystal systems and space groups, the structure was solved in the tetragonal P4 space group.
The crystal structure analysis clearly revealed the existence of physisorbed S 2 and bent-S 3 species on the iodide sites of the framework channels (  for [(ZnI 2 ) 3 (TPT) 2 (S 3 )] n by structure solution from X-ray powder diffraction  and that of S 3 in the gas phase as observed by rotational spectroscopy (McCarthy et al., 2004). These physisorbed S 2 and bent-S 3 do not have any interaction with iodide; reactive S 2 can take part in subsequent reactions because it is not stabilized by the pores.
In order to investigate the transient nature of S 2 in the channels of network 1, we collected two additional sets of X-ray single-crystal diffraction data at 300 and 350 K using a heating rate of 10 K min À1 between measurements (see Fig. S3 in the supporting information). The diffraction data at 300 K showed a space-group change from P4 to I4, a sharpening of the diffraction spots and the almost complete disappearance of diffuse scattering, which indicates that successive reaction of the sulfur species had taken place on heating. Analysis of the 300 K structure revealed the formation of cyclo-S 3 chemisorbed on bridging iodide sites, and the presence of physisorbed bent-S 3 and physisorbed cyclo-S 3 in the network 1 channels (Fig. 1). The cyclo-S 3 allotrope has been predicted to be less stable than the bent-S 3 structure, but still energetically accessible, by theoretical calculations (Flemmig et al., 2005) but it had never been observed before. A theoretical investigation of the adsorption of cyclo-S 3 on the network iodide sites revealed that chemisorption is only possible if cyclo-S 3 is present as a dication, cyclo-S 3 2+ (see the supporting information). Even though we did not use any restraints for the bond lengths, the geometric parameters obtained from this X-ray analysis matched those obtained by theoretical calculation (Fig. 2). The cyclo-S 3 2+ state is isoelectric with a cyclo-SiS 2 molecule (Mü ck et al., 2012) isolated by matrix isolation, indicating the potential existence of a cyclic form. We could not determine the counter pair formed by oxidation, because of severe disorder of the physisorbed species for which restraints on bond length were used during the refinement. A structure redetermination of the single crystal at an even higher temperature, 350 K, revealed only bent-S 3 species in network 1, suggesting a complete transformation of chemisorbed cyclo-S 3 2+ (and physisorbed cyclo-S 3 species) to bent-S 3 species (see Fig. S6 in the supporting information). After the heating cycle, the same single crystal was cooled back to 250 K for a second structure redetermination at low temperature, but the diffraction data were not of sufficient quality to allow structure solution. Refinement using the initial P4 space group was unsuccessful, which indicates an irreversible P4 to I4 phase transformation.
From this series of X-ray diffraction experiments of sulfurencapsulating network 1, we propose one of the possible reaction pathways of small sulfur allotropes: first, S 2 was kinetically trapped by physisorption and partly transformed into physisorbed bent-S 3 ; second, on heating the S 2 converted to chemisorbed cyclo-S 3 2+ , and physisorbed cyclo-S 3 and bent-S 3 species; and third, the cyclo-S 3 species transformed to the more stable bent-S 3 species (Fig. 1e). Despite the kinetic nature of these experiments, we always found consistent results upon repetition of the diffraction measurements on different single crystals. We also observed chemisorbed S 2 molecules on the interactive iodide sites (see Fig. S7 in the supporting information). Our theoretical calculations predicted chemisorption of S 2 to be less favourable than physisorption, because S 2 needs to change its electronic spin (see the supporting information).

Spectroscopic confirmation of sulfur species
The trapping of sulfur in network 1 was also investigated at room temperature using microscopic Raman and IR spectroscopy. Raman spectra of the samples after sulfur encapsulation displayed new bands at 475 and 573 cm À1 (Fig. 3a). These bands have been assigned to chemisorbed cyclo-S 3 2+ (475 cm À1 ) and chemisorbed cyclo-S 3 2+ plus bent-S 3 species (573 cm À1 ) with the help of density functional theory (DFT) calculations (see the supporting information) (Picquenard et al., 1993). After 18 h, the intensity of the cyclo-S 3 2+ species band decreased significantly, which suggests that the cyclo-S 3 2+ species were consumed and converted into bent-S 3 (see Fig. S9 in the supporting information). Bent-S 3 species were clearly detected by IR spectroscopy (band at $680 cm À1 ; see  (a) Raman spectra of network 1, desolvated (black), solvated with DMSO (pink) and after sulfur encapsulation (red). (b) Raman spectra of network 2, desolvated (black), solvated with DMSO (pale blue) and after sulfur encapsulation (blue). The inner graph in part (b) shows a magnified view of the 690-750 cm À1 region for the sulfur-encapsulated network 2 sample. Black arrows highlight the bands appearing after sulfur encapsulation [475 and 573 cm À1 for cyclo-S 3 2+ and bent-S 3 in part (a), and 728 cm À1 for S 2 in part (b)]. The asterisks (*) indicate the effects of cosmic rays and the dagger ( †) shows cyclo-S 8 on the crystal surface.

Figure 2
Geometric parameters from X-ray diffraction and theoretical calculation for chemisorbed cyclo-S 3 2+ . Red numbers indicate values obtained from X-ray analysis and blue numbers refer to values obtained from calculation of I-cyclo-S 3 2+ . Atom colouring: Cu orange, I purple and S pink.
in the supporting information). There are two possibilities for the mechanism of the transformation of S 2 to cyclo-S 3 2+ to bent-S 3 : (i) direct conversion of S 2 to S 3 using catalytic iodide sites; or (ii) conversion including dimethylsulfoxide (DMSO) (see the supporting information). The oxidation of S 3 into cyclo-S 3 2+ might be preceded by other sulfur species accepting electrons and protons, resulting in H 2 S n species. We observed new bands in the IR spectra in the region of 2225 cm À1 , which are most likely due to S-H stretches (see Fig. S17 in the supporting information) (Marsden & Smith, 1988). Attempts to reveal the reaction mechanism by removing DMSO completely resulted in deterioration of the single crystals. A possible reaction mechanism is outlined in the supporting information. However, the reactions occurring are complex and, unless the intermediates are strongly adsorbed on the network (like the species identified by X-ray diffraction), they are difficult to characterize. In fact, although it is not trivial to reveal the reaction mechanism, we clearly observed the structural change in these small sulfur species on an interactive site using X-ray diffraction.

Sulfur species in network 2
Kinetic trapping of sulfur gas in network 2 resulted in physisorbed S 2 species only, with no evidence of S 3 . X-ray analysis at 30 K revealed that S 2 physisorbed on two different sites of network 2: (i) aligned in the one-dimensional channel of the structure and presenting severe disorder; and (ii) within small cavities adjacent to the Cu 2 I 2 units ( Fig. 4; see Fig. S8 in the supporting information for structure details). Only physisorption of S 2 was observed on iodide sites in this network, because of steric hindrance around the iodide sites. The smaller size and linear shape of the one-dimensional channels suppress the conversion of linear S 2 molecules into S 3 species. This is an example of shape-selective trapping of a linear reactive intermediate. However, the S 2 molecules existing adjacent to the Cu 2 I 2 units have a weak interaction with iodide. These interactions come from charge transfer from iodide to sulfur, as shown by calculation (see Table S5 in the supporting information). This type of interaction is different from the Lewis acid-sulfide interaction shown in a sulfide-encapsulating Ni-MOF system (MOF = metal-organic framework; Zheng et al., 2014). The existence of S 2 in network 2 after sulfur encapsulation was further confirmed with microscopic Raman spectroscopy at room temperature; a new band appeared at 728 cm À1 (Fig. 3b), which corresponds to S 2 symmetric stretching (Barletta, 1971). S 2 remained stable within network 2 up to 500 K (see Fig. S2 in the supporting information).

Conclusions
We observed labile sulfur allotropes reacting in an interactive pore using X-ray diffraction. We found unexpected reactions of S 2 on an interactive site: chemisorption, transformation of S 2 into cyclo-S 3 , and bent-S 3 species. On the basis of X-ray and vibrational analyses and theoretical calculations, we propose that the chemisorbed species is cyclo-S 3 2+ rather than neutral cyclo-S 3 . We also, for the first time, isolated S 2 in a onedimensional channel by kinetically suppressing further reactions. The method reported here provides a new means for future investigations of other labile reaction intermediates. Indeed, this method makes it possible to find out new reactions of sulfur allotropes.

Figure 4
Crystal structures for (a) desolvated network 2 and (b) network 2 after sulfur encapsulation at 30 K. The stoichiometry of the structure is {[(Cu 2 I 2 )(C 45 H 32 N 4 )]Á(S 2 ) 0.975 } n . Part (b) shows a different view of disordered S 2 in the one-dimensional channel with a ball-and-stick model: each coloured molecule corresponds to S 2 . Atom colouring: C grey, N blue, Cu orange, I purple, and S yellow, brown, red and green. H atoms have been omitted for clarity.