Intermolecular Interactions in Highly Disordered, Confined Dense N2

DOI: 10.1021/acs.jpclett.7b00902. Publication Date (Web): May 12, 2017. Copyright © 2017 American Chemical Society. *E-mail: [email protected]...
0 downloads 0 Views 1MB Size
Letter pubs.acs.org/JPCL

Intermolecular Interactions in Highly Disordered, Confined Dense N2 Mario Santoro,*,†,‡ Federico A. Gorelli,†,‡ Roberto Bini,‡,§ and Julien Haines∥ †

Istituto Nazionale di Ottica, CNR-INO, 50019 Sesto Fiorentino, Italy European Laboratory for Non Linear Spectroscopy (LENS), 50019 Sesto Fiorentino, Italy § Dipartimento di Chimica, Università degli Studi di Firenze, 50019 Sesto Fiorentino, Italy ∥ ICGM, CNRS, Univ. Montpellier, ENSCM, 34090 Montpellier, France ‡

ABSTRACT: Molecular nitrogen is a benchmark system for condensed matter and, in particular, for looking at universal properties of strongly confined dense systems. We conducted Raman and X-ray diffraction measurements on a dense and disordered form of molecular nitrogen subnanoconfined in a noncatalytic pure SiO2 zeolite under pressure, up to 50 GPa. In this form, N2−N2 interactions and, consequently, distances are found to be very close to those of bulk N2 and intramolecular interactions progressively weaken upon increasing pressure. Surprisingly, the filled zeolite is still crystalline at 50 GPa with silicon in tetrahedral coordination by oxygen, which is a record pressure for this type of coordination among all the known forms of silica. We have thus found a rationale for the polymerization of a number molecules occurring in the microchannels of noncatalytic zeolites under pressure, where the pressure threshold is found to be very similar to that observed in bulk samples.

T

interactions for molecular systems in this state with a reduced number of nearest neighbors and local dimensionality with respect to the parent bulk system. The elucidation of this fundamental aspect can provide a profound insight into chemical reactions that may occur in strongly confined systems in zeolites as the sole effect of pressure and confinement, which we largely investigated in the recent past.24−27 Indeed, this study is based on our large experience on noncatalytic purely siliceous zeolites filled by molecular systems at high pressures. We have first discovered that the insertion of simple molecules such as Ar and CO2 in these zeolites, in amounts that completely fill the pores, deactivates pressure-induced amorphization (PIA) of the host scaffold at least up to 25 GPa.28,29 Then, we obtained purely pressure-induced self-assembly of simple carbon-based molecules such as ethylene, acetylene, and carbon monoxide within the zeolite channels, in the gigapascal range, which led to the formation of novel nanocomposites made of single-chain polymers embedded in the zeolite.24−27 Xray diffraction (XRD) measurements on the ambient pressure recovered nanocomposites revealed that atoms forming the embedded polymers are distributed around well-defined sites with a large degree of disorder. We also found a similar type of disorder in preliminary studies on dense atomic Ne confined in zeolites under pressure. Interestingly, the threshold pressure for confined polymerization was found to be very similar to the polymerization threshold pressure in the same molecular systems in bulk samples. Indeed, the present work provides a rationale for this similarity. The zeolite we use in this work is

he high-pressure behavior of dense bulk nitrogen, which we use as a reference here, has been one of the most investigated among those of simple substances (see refs 1−13 and references therein). What triggered these studies was nitrogen being a model system for a number of phenomena in strongly condensed matter such as a rich variety of phase transitions between thermodynamically stable and also metastable phases, N2 orientational plastic disorder and order−disorder transformations in solid state, and molecular transformations in both solid and fluid states. The present highpressure phase diagram of solid nitrogen includes up to 12 to 13 distinct molecular phases in the 0−150 GPa pressure range.9,12,6 Interestingly, pressure shifts of the crystalline frequencies for the N2 stretching mode are strongly sublinear and ultimately tend to level off or even exhibit maxima at around 60−100 GPa. 1,3,5−10 This behavior signals the progressive weakening of the N−N triple bond upon increasing pressure, which, in turn, ends up in the formation of nonmolecular, polymeric nitrogen above 100 GPa. Polymeric nitrogen was indeed predicted by ab initio calculations14−23 and also found experimentally.8−11,13,12 In this work we aimed to investigate the high-pressure behavior of dense molecular nitrogen, confined on the Ångstrom scale in the microchannels of a noncatalytic pure siliceous zeolite. This type of confinement of dense guest systems in a host scaffold obviously prevents the full crystallization of the guests potentially ending up into highly disordered subnanophases of these systems, and N2 certainly represents a case system in this respect. The physical and chemical behavior of matter in such exotic states is rather unknown and intriguing and thereby worthy of investigation. In particular, it is very important to characterize intermolecular © XXXX American Chemical Society

Received: April 13, 2017 Accepted: May 12, 2017 Published: May 12, 2017 2406

DOI: 10.1021/acs.jpclett.7b00902 J. Phys. Chem. Lett. 2017, 8, 2406−2411

Letter

The Journal of Physical Chemistry Letters silicalite-1, which is available as single crystals with sizes of tens of micrometers and its pressure behavior up to ∼25 GPa is now well known.28,30,31 Silicalite-1 is characterized by a framework of 4-, 5-, 6-, and 10-membered rings of corner-sharing SiO4 tetrahedra forming interconnected, mutually orthogonal straight and sinusoidal 5.5 Å diameter channels (refs 32 and 33 and references therein). Here we show the high-pressure formation of a highly disordered state or nanophase of dense molecular N2 confined within the channels of silicalite-1 in the 0−50 GPa pressure range, observed by means of in situ Raman spectroscopy in diamond anvil cells (DACs). The pressure behavior of the N2 stretching frequencies of this disordered form provides physical insight into the relevant microscopic interactions and mechanisms. We also performed XRD measurements at the maximum pressure and on the ambient pressure recovered sample. These measurements serve as a monitor of the degree of crystallinity of the host silicalite-1 and, as a consequence, indirectly provide additional information on the confined disordered nanophase of N2. Dense nitrogen was loaded in DACs together with one to two crystals of hydrophobic hydrogen-free silicalite-1-F with their long axis corresponding to the crystallographic c direction parallel to diamond culet surface. In these samples, nitrogen was in two forms: confined N2 in silicalite-1 and bulk N2 surrounding the zeolite crystals. The gasket hole had an initial diameter and a thickness of about 100 and 40 μm, respectively. Silicalite-1 crystals have sizes of 80 × 40 × 40 μm3, and they were broken into several strongly oriented pieces after the loading pressure was applied and the sample chamber thickness and diameter were consequently reduced. Two types of Raman spectra were observed in these samples for points in pure bulk N2 and points in the silicalite-1 crystals, respectively. It was then clear that silicalite-1 was almost bridging the diamond anvils, except thin micron size layers of bulk nitrogen on the top and the bottom. In Figure 1, we report a selection of Raman spectra of the N2/silicalite-1 mixture, measured at the top of silicalite crystals upon increasing pressure up to ∼50 GPa in the frequency range of the N2 stretching. Here we observe several peaks for confined nitrogen, whose number, linewidth, and total spread increase upon increasing pressure, and also the frequencies exhibit positive pressure shifts. At the lowest pressure, 0.7 GPa, confined nitrogen exhibits three sharp peaks spaced by 1.2 to 1.6 cm−1. These can be reasonably assigned to N2 molecules in three different local environments: sites along the straight channels, along the zigzag channels, and at the intersection of the two types of channels. At least two observations support that this fine Raman structure is not specific to N2 but rather to the structure of the channels. First, we performed some measurements on O2/silicalite-1 mixtures, where at similar pressures confined oxygen also exhibited three finely spaced Raman peaks for the O2 stretching (Figure 1, inset). Then, preliminary Raman measurements on N2/TON and O2/TON mixtures at these pressures, where TON is a pure SiO2 zeolite with a 1D channel system with a single type of channels, show that confined N2 and O2 exhibited only a single peak instead of three. It is then highly probable that confined N2 molecules in silicalite-1 distribute on the three different types of site with some degree of disorder already at pressures of tenths of gigapascals, based on our experience of the polymer/zeolite composites.24−27 At 13.6 GPa we observe at least four poorly resolved peaks for confined N2 (see arrows in Figure 1), which split each other by 5−12 cm−1. The total

Figure 1. Selected Raman spectra of a N2/silicalite-1 sample measured upon increasing pressure. Vertical sticks: peaks of bulk solid N2 phases, indicated by Greek letters (f is the fluid phase). Arrows: peaks for confined N2 at 13.6 GPa. Inset: Raman spectrum of an O2/silicalite-1 sample at 0.6 GPa; vertical stick: peak of bulk fluid O2. The spectra have been normalized to the same peak intensity; then, they have been vertically shifted.

spread of these components and their individual linewidths are larger than those observed at 0.7 GPa by about one order of magnitude. While the increased total frequency spread is related to increased microscopic interactions in a sense that will be specified below, the increased number of peaks testifies for confined N2 being distributed among a larger number of average sites within the channels. Also, the strongly increased linewidth of the single peaks suggests that the molecules are distributed around the average sites with a significantly increased degree of disorder. Above 15−20 GPa the peaks merge into a single, broad, and featureless band, strongly suggesting that the distribution of molecules within the channels is now entirely statistical, and hence the disorder of confined N2 is increased further in these conditions. It is then reasonable to think that the broad, single band roughly resembles the density of intramolecular N−N stretching states, DOS, of this highly disordered form, as it was an amorphous material. The total width of this band increases up to the maximum investigated pressure, 50 GPa, and then reversely decreases upon reducing pressure; nevertheless, the fine structure observed in the upstroke run is never recovered. In Figure 2, we report the XRD pattern of the N2/silicalite-1 mixture measured at the maximum pressure of 50.3 GPa, where we observe two distinct Bragg reflections of silicalite-1 at low diffraction angles of 4.059 and 4.760°, respectively. These reflections are single crystal-like, meaning that the silicalite-1 chips are strongly oriented. The crystals are highly strained. We also see a very weak peak of silicalite-1 between 12.0 and 12.5°. The whole pattern is superimposed to an intense and broad background signal due to Compton scattering from diamonds 2407

DOI: 10.1021/acs.jpclett.7b00902 J. Phys. Chem. Lett. 2017, 8, 2406−2411

Letter

The Journal of Physical Chemistry Letters

silica. Silicalite-1 thus partially amorphized upon reducing pressure, which may be due either to loss of confined N2 from the zeolite or to the silicalite-1 being damaged by inserted N2 expanding to a larger extent than the host scaffold. The partial amorphization of silicalite-1 reasonably explains the observation that the fine structure of the N2 stretching Raman peak for confined nitrogen was not recovered upon reducing pressure. To provide a quantitative description of microscopic interactions probed by the N2 molecules in the confined state, we have calculated the spectral moments of the N2 stretching Raman band as a whole, which allowed us to describe essential aspects of these interactions not depending on the fine structure details. The moments have been calculated from the measured spectra after subtraction of the smooth diamond fluorescence background and of the sharp peaks of bulk nitrogen, which, in turn, involved a fitting procedure. In Figure 3 we report the pressure shift of the spectral centroid, that is, of

Figure 2. XRD pattern of a N2/silicalite sample at 50.3 GPa and of the ambient pressure-recovered sample. Insets: selected portion of the 2D XRD patterns. The white large central area is the shadow of the beam stop glued on a tungsten wire (vertical white narrow strip), and the black spots at the edge of this area, which have been masked before performing the 2D average, are due to double diffraction in the diamond anvils with two scattering events. The very weak peak at 12.0 to 12.5° at 50.3 GPa is also due to silicalite-1, whereas the broad, diffuse peak at ∼11.5° in the recovered sample is due to amorphous silica.

and air. Remarkably, we do not detect diffuse scattering from potential amorphous silica, which indicates that the majority of our silica sample is still crystalline silicalite-1, although we cannot exclude that a minor portion of silica is indeed amorphous with an XRD signal that is overwhelmed by the strong Compton background. PIA of silicalite-1 in nonpenetrating, pressure-transmitting media has been found to occur at several gigapascals, whereas the insertion of dense simple molecular systems deactivates PIA at least up to 25 GPa.28 Here we found that the insertion of dense N 2 deactivates PIA up to at least 50 GPa, which is a record pressure in this respect and it is also a record pressure for silicon being four-fold-coordinated by oxygen in any known form of silica. Very importantly here, our findings show that a guest highly disordered dense N2 form is hosted in a porous crystalline, although highly strained, scaffold. The increased disorder of confined nitrogen at high pressures is then likely to originate, at least partially, from the strain of the host system. It is also important to note that the deactivation of PIA in silicalite-1 is a strong piece of evidence that N2 completely fills the zeolite and, as a consequence, that confined highly disordered nitrogen is a true nanophase or -state as a whole rather than being a trivial ensemble of isolated impurities. The XRD pattern of the recovered sample at ambient pressure still confined by the gasket after it has been removed from the DAC is also shown in Figure 2. Nitrogen completely flowed out of the sample chamber under these conditions, and the sample is pure silica. In this pattern, we observe two weak and spotty peaks at 3.01 and 3.29° from a highly oriented and strained phase obtained from silicalite-1 and a broad, diffuse peak at ∼11.5°, that is, the first sharp diffraction peak of amorphous

Figure 3. Pressure shift of the frequency centroid of the Raman N2 stretching peak for confined nitrogen. Full and open circles and black, thick line: data points measured upon increasing and decreasing pressure, respectively, and a guide for the eye fitted through the points. Error bars originate from the peak fitting procedure and spatial variation of frequencies and pressures. Red1 and green3 lines: reproduced literature data for the N2 stretching frequencies of bulk crystalline N2, ν1, and ν2.

the average wavenumber vavg calculated as vavg = ∫ vI(v)dv/ ∫ I(v)dv, where v and I(v) are the wavenumber and the spectral intensity, respectively, together with the literature crystalline frequency components of bulk N2.1,3 Pressure behavior of the spectral centroid of confined disordered N2 compares very well with those of the ν2 manifold of bulk N2. In particular, this behavior for the confined form is markedly sublinear similar to the crystalline frequencies. These findings have one very important implication: the pressure dependence of microscopic interactions probed by N2 molecules in the confined form is dominated by that of the intermolecular N2−N2 interactions between nearest N2 neighbors, which are very similar, along with intermolecular distances to those in the molecular crystal at the same pressures. Specifically, the sublinear character of the frequency pressure shift of confined disordered nitrogen 2408

DOI: 10.1021/acs.jpclett.7b00902 J. Phys. Chem. Lett. 2017, 8, 2406−2411

Letter

The Journal of Physical Chemistry Letters suggests that in parallel with bulk N2 a progressive weakening of the N−N triple bond occurs in this form upon increasing pressure, which may end up in some form of polymeric, confined nitrogen at sufficiently high pressures. Interestingly, the result is that the interactions between N2 and the silicalite-1 pore walls, that is, N−O interactions, although strong in principle, may only weakly affect the pressure dependence of microscopic interactions experienced by N2. This point indicates that N2−N2 distances and molecular packing along the channels of the zeolite, that is, along the stacking directions of confined molecules, are more easily changed upon increasing pressure than the N−O distances, an argument that calls for future structural studies now beyond the scope of the work. Furthermore, the spectral centroid of confined N2 being very close to the ν2 rather than the ν1 frequency of bulk N2 suggests that a particular type of orientational disorder for confined nitrogen occurs. Because ν2 and ν1 are the stretching frequencies for disk-like and sphere-like orientationally disordered molecules in solid N2, respectively, our result suggests that disk-like orientational disorder prevails in confined disordered nitrogen. Confirmation of the scenario given above for microscopic interactions and additional information on these forces are then obtained from the calculation of the spectral standard deviation σ, as σ2 = ∫ (v − vavg)2I(v)dv/∫ I(v)dv. Indeed, 2σ is a reasonable measure of the total frequency spread of the N2 stretching DOS for disordered N2, and we can compare this quantity to the total spread observed in bulk N2 crystalline phases. Although for crystalline bulk N2 Raman peaks only sample the Brillouin zone center, the total spread of DOS in this case can be reasonably estimated as the total splitting between the different crystalline frequency components, that is, as the difference between ν1 and the lowest ν2 component, which we obtained from literature.1,3 In Figure 4, we compare the total spread of the N2 stretching DOS for confined disordered N2 and for crystalline bulk N2, respectively, as a function of pressure. It is very striking how consistent these two behaviors are with each other. This result is a strong confirmation of pressure dependence of microscopic interactions seen by N2 in the confined form being dominated by pressure dependence of nearest-neighbor intermolecular interactions. These interactions, together with intermolecular distances, are very close to those in the molecular crystal, and they are the main origin of the total spread in the Raman peaks of confined disordered N2. Also, the similarity of N2−N2 interactions and distances between confined nitrogen in silicalite-1 and the bulk crystal clearly show once again, in addition to PIA being deactivated up to at least 50 GPa, that confined nitrogen is a dense state of closely stacked molecules, which are indeed very far from being isolated guest impurities in the host scaffold. In conclusion, we have shown that high-pressure confinement of N2 on the Ångstrom scale in a purely siliceous zeolite, at tens of gigapascals, leads to a highly disordered dense form of molecular N2. On the contrary, the zeolite remains mostly crystalline up to 50 GPa, although the crystal is highly strained; this pressure is certainly a record pressure for silicon being fourfold coordinated in any known form of silica. The pressure dependence of microscopic interactions affecting N2 in this form is dominated by that of the N2−N2 intermolecular, vander-Waals-type interactions, which are very similar, along with intermolecular distances, to those in bulk, crystalline N2. From these findings, we can straightforwardly provide important clues

Figure 4. Pressure shift of 2σ, where σ is the frequency standard deviation with respect to the frequency centroid of the Raman N2 stretching band for confined nitrogen. Full and open circles and black, thick line: data points measured upon increasing and decreasing pressure, respectively, and a linear guide for the eye fitted through the points. Error bars originate from the peak fitting procedure and spatial variation of frequencies and pressures. Red1 and green3 lines: reproduced literature data for the difference between the maximum and the minimum N2 stretching frequencies of bulk, crystalline N2.

on high-pressure polymerization reactions that we observed in confined simple C-based molecular systems in purely siliceous zeolites.24−27 In these systems, we observed polymerization at pressures very close to those observed in bulk samples. This is now clearly understandable in terms of intermolecular interactions and, consequently, distances being very similar in the confined and in the bulk systems. This general result leads to foresee the pressure threshold for any other possible chemical reactions that may occur between molecules confined in noncatalytic zeolites once this threshold is known for the bulk system. In particular, on N2/siliceous zeolite systems there is a very intriguing perspective. Similarly to bulk N2, confined disordered N2 may transform into polymeric nitrogen forms above 1 Mbar because N2−N2 intermolecular interactions are similar in the two systems. These confined polymeric forms should be made of single chains of N atoms fitting the channels of the zeolite and would thereby differ from the known, 3D polymeric forms of bulk nitrogen.



METHODS High-purity nitrogen was cryoloaded in DACs at 77 K together with commercially available hydrophobic hydrogen-free silicalite-1-F (SOMEZ, France). DACs were equipped with 300 μm culet and ultralow fluorescence Ia diamonds, and we used Re as the gasket material. Pressure was measured by the ruby fluorescence technique.34 Raman spectra were performed using the 647.1 nm line of a Kr+ laser as the excitation source. Backscattering geometry was used with X20 Mitutoyo microobjective giving a laser spot of