Hydrazinium azide (HA) has been investigated at high pressures to 68 GPa using confocal micro-Raman spectroscopy and synchrotron powder x-ray diffraction. The results show that HA undergoes structural phase transitions from solid HA-I to HA-II at 13 GPa, associated with the strengthening of hydrogen bonding, and then to N8 at 40 GPa. The transformation of HA to recently predicted N8 (N≡N+—N—N=N—N—+N≡N) is evident by the emergence of new peaks at 2384 cm−1, 1665 cm−1, and 1165 cm−1, arising from the terminal N≡N stretching, the central N=N stretching, and the N—N stretching, respectively. However, upon decompression, N8 decomposes to ε-N2 below 25 GPa, but the remnant can be seen as low as 3 GPa.

Nitrogen-rich extended solids1 contain large chemical energy because of the high stability of nitrogen molecules at ambient conditions. For example, cubic gauche nitrogen (cg-N),2–4 a polymeric form of nitrogen molecules made of three-fold coordinated nitrogen single bonds in a three-dimensional (3D) network, contains large chemical energies,4 ∼33 kJ/cm3, three times that of HMX,5 11 kJ/cm3, one of the most powerful explosives used today. Under the right conditions, cg-N can exothermically disintegrate or depolymerize to stable nitrogen, rapidly releasing a large sum of energy. The specific impulse of cg-N is estimated to be about 400 s,6 ideally suited for novel propellants, for example. Therefore, the synthesis and recovery of novel extended forms of nitrogen molecules and nitrogen-rich materials have been a topic of interest to development of high energy density materials with fast energy release.7 

The transitions to single-bonded nitrogen polymers such as cg-N8 and LP-N,9 however, require not only the formidably high pressures and temperatures above 110 GPa and 2000 K but also these nitrogen polymers become unstable below 40-60 GPa upon the pressure unloading.9,10 Therefore, research efforts have been made for using various forms of nitrogen-rich compounds as the precursors to synthesize similar forms of nitrogen polymers. Those include nitrogen-rich molecules,11 ionic nitrogen salts,12 metal nitrides,13 and simple azides.14–16 The rationale for these efforts follows the concept that these precursors are considerably higher in density (1.5-2.0 g/cm3) than N2 molecules (∼1 g/cm3) and, thereby, can be considered as intermediates toward high density (>3.5 g/cm3) nitrogen-rich polymers.

Azides compounds with a general formula of XN3, where X = H, alkali metals, and nitrogen-rich ions, have been one of the most commonly used precursors for high-pressure synthesis of high density and singly bonded nitrogen polymers. A wide range of novel nitrogen products have been reported based on the computational structural search and optimization, which include N5 rings [Fig. 1(a)],16 N5+ chain [Fig. 1(b)],17 N6 chains [Fig. 1(c)],18 N6 rings,19 polymeric nitrogens,20 and N8 molecules [Fig. 1(d)].21 However, only a few of them including N5 rings16 and V-shaped N5+ (Ref. 17) have been experimentally synthesized. Among the predicted structures of nitrogen, N8 is particularly interesting because its covalently bonded molecular configuration [Fig. 1(d)] is similar to the nitrogen arrangement in hydrazinium azide (HA, [N2H5]+ [N3]) [Fig. 1(e)]. Note that a pair of N3 ions positioned on each end of two nitrogen atoms in N2H5+ along the c-axis resembles almost N8-like configuration. This N8 molecule is a molecular form of solid nitrogen that is predicted to be stable below 20 GPa.21 Yet, the synthesis of N8 has been challenging because of the intrinsic instability of polynitrogen compounds.22 While an experimental study has reported the existence of N8 species in carbon nanotube bundles using an electrochemical method,22 other techniques such as the use of high pressure to synthesize neutral N8 molecules remain unexplored yet.

FIG. 1.

Crystal structures of (a) P-1-CsN5 showing N5 rings,16 (b) V-shaped N5+.17 (The anions are not included for simplicity.) (c) N6 molecule,18 (d) N8 molecule21 showing the molecular arrangement similarity with (e) HA.23 The blue atoms correspond to the nitrogen atoms and the orange atom corresponds to the hydrogen atoms. The red dotted lines represent the hydrogen bonds.

FIG. 1.

Crystal structures of (a) P-1-CsN5 showing N5 rings,16 (b) V-shaped N5+.17 (The anions are not included for simplicity.) (c) N6 molecule,18 (d) N8 molecule21 showing the molecular arrangement similarity with (e) HA.23 The blue atoms correspond to the nitrogen atoms and the orange atom corresponds to the hydrogen atoms. The red dotted lines represent the hydrogen bonds.

Close modal

Considering the similar nitrogen arrangement and stoichiometric considerations, we have investigated HA at high pressures aimed at the synthesis of predicted stable molecular nitrogen allotropes (such as N8 or N6 in particular) or similar nitrogen dominant ionic products (such as N5+ ring or V-shaped N5+ ions) at high pressures. At ambient conditions, HA crystallizes to a monoclinic (P21/b, call this as HA-I) structure with four molecules per unit cell as shown in Fig. 1(e).23 In this structure of HA-I, N2H5+ cations are connected to each other through hydrogen bonds [red dashed line in Fig. 1(e)] forming infinite chains wound around the 21 axes. The remaining four hydrogen atoms are hydrogen bonded with four neighboring N3 ions [not shown in Fig. 1(e)]. One interesting thing about the way the nitrogen atoms are arranged in this HA structure is that two nearest N⋯N distances are separated by only 2.8 and 3.2 Å. So when atoms are pushed together, it is conceivable to form N8 (or N5) by fusing two (or one) N3 with N2H5 along the c-axis. Furthermore, as shown by previous studies,14 the linear azide anions (N3) can polymerize into single-bonded high energy density solids at high pressures. The presence of hydrogen, on the other hand, can provide chemical internal pressure and lower the transition pressure.24,25 This is exactly what we have found in the present study; that is, HA-I transforms to HA-II in a hydrogen-bonded network above 13 GPa and further to a linear nitrogen chain of N8 above 40 GPa. The formation of N8 is best characterized by the emergence of a strong Raman peak at 2384 cm−1 arising from the stretching mode of two terminal N≡N bonds.

Sodium azide (NaN3) and hydrazine sulfate (N2H4·H2SO4) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Hydrazine monohydrate (N2H4·H2O) was obtained from Tokyo Chemical Industry Co. (Tokyo, Japan) and n-butanol was purchased from Daejung (Seoul, Korea). They were used without further purification. Hydrazinium azide (HA) was obtained by the reaction of sodium azide and hydrazine. A three-neck round-bottom flask with a magnetic stirring bar was charged with sodium azide (1.690 g), hydrazine sulfate (1.692 g), and hydrazine monohydrate (0.65 ml), after which 250 ml of n-butanol was slowly added under N2. The mixture was reacted for 30 min at 120 °C. The solvent was decanted and was cooled to 0 °C. The resulting product was washed three times with ether and dried under vacuum to yield 1.440 g of crystalline HA powder.

Transparent crystals of HA were crushed into fine powders and loaded into a diamond anvil cell (DAC) in an Ar-gas purging dry box. The sample was loaded without any pressure medium to prevent any potential physical/chemical contamination. We used 0.3 mm flat type Ia diamonds and the pressure was measured by the Ruby luminescence.26 

A home-built confocal micro-Raman system was used to collect the vibrational spectra of HA and probe the pressure-induced phase changes. The Raman system is equipped with a monochromator (Acton SpectraPro, Princeton Instruments), a back-illuminated 2D charge coupled detector (SPEC-10, Princeton Instrument), a holographic diffractive bandpass filter (Kaiser Optics), and a Raman notch filter (OD = 6, Kaiser Optics). The system is typically operated in a back scattering geometry, using an apochromatic objective lens (20×, 32 mm working distance, infinity-corrected, Edmund) coupled with a pair of confocal lenses (f = 4, 100 mm focal length) and an adjustable rectangular aperture to collect the Raman scattered light only from the plane of sample. An Ar+ ion laser (λ = 514.5 nm) is used as an excitation source after expanding it to ∼10 times. The system yields, typically, a spatial resolution of ∼3 μm and a spectral resolution of 1.5 cm−1.

X-ray diffraction was performed using micro-focused (20 μm × 20 μm) synchrotron x-rays (λ = 0.3445 Å) at 16IDB of High Pressure Collaborating Access Team (HPCAT) at the Advanced Photon Source (APS). The x-ray diffraction intensities were recorded over a large range of 2θ between 3° and 40°, using a large 2D image plate detector. We used the Fit2D27 and DIOPTAS28 software to convert the 2D diffraction image to the 1D angle-resolved x-ray diffraction (ARXD) pattern and, then, employed various diffraction analysis software including GSAS,29 EXPGUI,30 and VESTA31 to analyze the ARXD data.

Figure 2 shows a series of micro-photographic images taken from HA in DAC at various pressures (the numbers in GPa). HA phase I exists as colorless crystals at ambient conditions, which becomes darker as pressure increases to 4-15 GPa and then becomes transparent again above 20-30 GPa and stays transparent to 68 GPa—the maximum pressure studied. These changes in visual appearance accompany with the spectral changes in Raman spectra of HA as discussed below.

FIG. 2.

Micro-photographs of HA showing the pressure-induced changes in visual appearance. The sample darkens initially to 15 GPa and then becomes transparent again at 25 GPa. It stays transparent up to 68 GPa.

FIG. 2.

Micro-photographs of HA showing the pressure-induced changes in visual appearance. The sample darkens initially to 15 GPa and then becomes transparent again at 25 GPa. It stays transparent up to 68 GPa.

Close modal

HA phase I in P21/b with Z = 4 allows a total of 120 normal modes (117 vibrational and 3 acoustic): Γvib = 30 Bu + 30 Au + 30 Ag + 30 Bg. Among these modes, the Ag and Bg are Raman active, whereas the Au and Bu modes are IR active.32 Figure 3 shows the Raman spectra of HA obtained as pressure increases to 54 GPa. The Raman spectrum at ambient pressure is consistent with the calculated spectrum of HA-I,32 consisting of (i) the external lattice modes of N3 at 93 and 116 cm−1 and of N2H5+ clustered at ∼150–300 cm−1; (ii) the internal vibrational modes of N3 stretching, ν1(N3) at 1340 cm−1 [only seen in Fig. 4(a) due to strong intensity] and the overtones of bending, 2ν2(N3) at 1249, 1260, and 1275 cm−1; and (iii) those of N2H5+—the stretching, νs(NH3) at 2796 cm−1, ν1(NH2) at 3164, 3190 cm−1 and, ν3(NH2) at 3260 cm−1; various bending modes νb(HNH) at 1417, 1507 cm−1 in NH2 and NH3, 1102, 1126 cm−1 for out of plane wagging between NH2 and NH3, and 1165 cm−1 for twisting between NH2 and NH3; and (iv) strong N—N single bond stretching mode of the N2H5 units at 994 cm−1.

FIG. 3.

Raman spectra of HA to 68 GPa, showing the spectral changes associated with the phase/chemical transformations of HA-I to HA-II at 13 GPa and further to N8 at 40 GPa.

FIG. 3.

Raman spectra of HA to 68 GPa, showing the spectral changes associated with the phase/chemical transformations of HA-I to HA-II at 13 GPa and further to N8 at 40 GPa.

Close modal
FIG. 4.

(a) Pressure evolution of the internal ν1(N3) mode. The intensity of the peak decreases with the increase of pressure and finally disappears above 54 GPa. (Inset) Pressure dependence of the ν1(N3) mode, showing an abrupt slope change in the pressure-dependent peak shift (Δν/ΔP) above 13 GPa. (b) Pressure-dependent Raman peak shifts of HA showing the phase/chemical transformation of HA-I to II at 13 GPa and to N8 at 40 GPa. The large symbols highlight the significant Raman modes.

FIG. 4.

(a) Pressure evolution of the internal ν1(N3) mode. The intensity of the peak decreases with the increase of pressure and finally disappears above 54 GPa. (Inset) Pressure dependence of the ν1(N3) mode, showing an abrupt slope change in the pressure-dependent peak shift (Δν/ΔP) above 13 GPa. (b) Pressure-dependent Raman peak shifts of HA showing the phase/chemical transformation of HA-I to II at 13 GPa and to N8 at 40 GPa. The large symbols highlight the significant Raman modes.

Close modal

Upon increasing pressure, all these Raman modes are shifted toward high frequencies (blue shifts), with exceptions of the νs(NH3) at 2988 cm−1 and ν1(NH2) at 3164 cm−1 [also, see Fig. 4(b)]. The red shifts in the νs(NH3) and ν1(NH3) indicate a pressure-induced weakening of N—H covalent bonds as a result of a strengthening of N⋯H hydrogen bonds. In addition, several spectral changes occur at 13 GPa: (i) the emergence of a few new peaks at 155 cm−1, 1140 cm−1, and 1654 cm−1, (ii) strong intensity drop of the νs(N—N) at 1000 cm−1, (iii) the disappearance or considerable weakening of several peaks including the νb(HNH) at 1417 cm−1 and the 2ν2(N3) mode at 1282 cm−1, and (iv) a sharp drop in the rate of pressure-dependent shift of the ν1(N3) [as highlighted in the inset of Fig. 4(a)]. We attribute these spectral changes to the transformation of HA-I to HA-II, which occurs at around 13 GPa.

Upon further increase in pressure above 40 GPa, several significant spectral changes occur, indicating the transformation of HA-II to N8. The spectral changes associated with this transformation are (i) the emergence of a new strong peak at 2476 cm−1 at 40 GPa, which dominates the Raman spectra at 54 GPa. It is most likely the νs(N≡N). (ii) The emergence of multiplet between 1000 and 1250 cm−1 into two peaks at 1150 and 1200 cm−1 at 40 GPa and further down to a single peak at 1170 cm−1 of the νs(N—N) at 54 GPa. This is likely from the νs(N—N). (iii) The weakening of νb(HNH) modes at 1537, 1613, and 1695 cm−1, which merge into a broad feature centered at ∼1670 cm−1 at 54 GPa. This peak compares with the νs(N=N) at ∼1500 cm−1 for azanes.33 (iv) The intensity enhancement of the νb(N3) modes at 640 and 658 cm−1, indicating the change in linear N=N=N to bent N=N—N. (v) The multiplets of νs(NH3) and νs(NH2) modes between 2700 and 3500 cm−1 become one broad peak centered at 3300 cm−1 at 54 GPa. These spectral changes result in a relatively simple Raman spectrum at 54 GPa, consisting of the νs(NH) peak at 3341 cm−1, the νs(N≡N) at 2384 cm−1, the νs(N=N) and/or νb(HNH) at 1670 cm−1, the νs(N—N) at 1165 cm−1, the νb(N=N—N) doublet at 675 and 635 cm−1, and several poorly resolved lattice modes below 600 cm−1. Hence, it clearly indicates that the HA-II to N8 transformation accompanies a significant modification in chemical bonding.

Figure 4(b) plots the pressure dependent Raman peak shifts, underscoring the discontinuous spectral changes in the peak positions and the rate of pressure shifts at around 6-13 GPa (for HA-I to II) and 30-40 GPa (for HA-II to N8).

Most Raman peaks of N8 are observed upon decompression to ∼25 GPa, below which several spectral changes occur as shown in Fig. 5. For example, the strong but relatively broad νs(N≡N) peak at 2348 cm−1 becomes a very sharp doublet characteristic to the νs(N≡N) of ε-N2. Therefore, it indicates that N8 decomposes to molecular N2 and other products at 25 GPa. This chemical change also accompanies several peak splittings of the νs(N—H) at 3300 cm−1, the νs(N=N)/νb(HNH) at 1600 cm−1, and the νs(N—N) at 1100 cm−1, as well as the weakening of the νb(NNN) at ∼670 cm−1. These spectral changes upon decompression result in the Raman spectra very similar to those of HA-II at ∼10 GPa and HA-I at 2.5 GPa.

FIG. 5.

Raman spectra of HA upon decompression, showing that HA-I is reversible with an exception of the N2 vibron at 2327 cm−1.

FIG. 5.

Raman spectra of HA upon decompression, showing that HA-I is reversible with an exception of the N2 vibron at 2327 cm−1.

Close modal

Figure 6(a) shows the powder x-ray diffraction patterns of HA at high pressures. The diffraction pattern at 2.8 GPa is collected at the initial loading with a minimum X-ray exposure (∼10 s). These powder diffraction patterns up to 4.7 GPa indexes are well within the reported P21/b structure of HA-I. At 14 GPa, a drastic change in the diffraction pattern is observed suggesting a structural phase transition. This structural phase transition is consistent with the spectral change observed at 13 GPa in Fig. 4. All of the major diffraction peaks congregate at around 2θ = 8°. With further compression to 19 GPa, the Braggs peaks at 7.6° and 8.7° become much weaker, with only one broad feature centered at 8.2°. This peak shows a blue shift with increasing pressure and becomes even broader at 39.5 GPa. On the other hand, it is important to note that the sample develops dark color upon x-ray exposures, indicating a possibility of photo-induced decomposition, especially in phase II. Nevertheless, upon the transformation to N8, we found that the sample is no longer photo-sensitive and remains transparent after x-ray exposures during decompression. In fact, the measured diffraction pattern of N8 at 35 GPa (at a new spot) is considerably sharper than those obtained during compression at 20-39.5 GPa in Fig. 6(a). A strong peak at 8.84° at 35 GPa is in sharp contrast to a weak and broad feature centered at the same angle at 40 GPa, for example. This sharp peak of N8 remains strong to ∼16 GPa, below which it becomes weaker and develops a few other weak features as N8 transforms to phase II. Yet, the remnant of N8 can be seen even at 3.6 GPa.

FIG. 6.

(a) Powder X-ray diffraction pattern of HA (λ = 0.344 Å) upon compression and decompression. HA-II shows photo-sensitivity upon x-ray exposures during compression, which deteriorates the diffraction pattern between 25 and 40 GPa, whereas N8 is not photo-sensitive, which yields considerably sharper diffraction patterns upon decompression. (b) (Bottom panel) The Le-Bail refinement of the initial diffraction pattern at 2.8 GPa. (Middle panel) The diffraction pattern collected during decompression from 35 GPa compared with ε-N2 at 33 GPa. (Top panel) Diffraction pattern at 3.6 GPa during decompression process compared with that of δ-N2 at 4 GPa, showing that the intense diffraction line is not caused due to the presence of molecular N2. The structural parameters of N2 phases were taken from Ref. 34. (c) The pressure dependence of d-spacing of N8.

FIG. 6.

(a) Powder X-ray diffraction pattern of HA (λ = 0.344 Å) upon compression and decompression. HA-II shows photo-sensitivity upon x-ray exposures during compression, which deteriorates the diffraction pattern between 25 and 40 GPa, whereas N8 is not photo-sensitive, which yields considerably sharper diffraction patterns upon decompression. (b) (Bottom panel) The Le-Bail refinement of the initial diffraction pattern at 2.8 GPa. (Middle panel) The diffraction pattern collected during decompression from 35 GPa compared with ε-N2 at 33 GPa. (Top panel) Diffraction pattern at 3.6 GPa during decompression process compared with that of δ-N2 at 4 GPa, showing that the intense diffraction line is not caused due to the presence of molecular N2. The structural parameters of N2 phases were taken from Ref. 34. (c) The pressure dependence of d-spacing of N8.

Close modal

The Lebail fitting performed using the structural model for the P21/b structure presented in Ref. 23 shows a reasonable agreement with the measured data [Fig. 6(b)]. The diffraction pattern of N8 (during decompression) is relatively simple, consisting of one relatively broad peak centered at 8.84°, which can be observed as low as 3.6 GPa. Note that this diffraction line does not correspond to that of ε-N2 or δ-N2 at their respective pressures [as shown in Fig. 6(b)]. This result indicates that a small amount of N8, or the remnant, is still recovered down to 3 GPa. The d-spacing of this intense peak is plotted in Fig. 6(c), which shows a strong pressure dependence to ∼2.7 Å at ambient pressure.

The present data indicate that HA-I transforms to HA-II at 6-13 GPa and further to N8 at 40 GPa. The former transition is accompanied by the pressure-induced red shift of the νs(NH3) modes, underscoring the strengthening of hydrogen bonding among the NH3, NH2, and N3 moieties in phase II. Upon the completion of HA-I to II transition at 13 GPa, all the N—H stretching modes show the pressure-induced blue shift, suggesting the strengthening of N—H covalent bonds and thereby the weakening of the hydrogen bonding above 13 GPa. A similar kind of pressure-dependent changes in bonding was also observed in phase I to II transition in ammonium azide.15 This phase II seems to remain stable up to 40 GPa.

The Raman spectrum of N8 becomes relatively simple at 54 GPa, consisting of the peaks at 3323, 2384, 1165, 675, and 640 cm−1, and a few lattice modes at 507, 456, 334, 272, and 149 cm−1 as seen in Fig. 3. The most remarkable spectral feature of N8 is the broad yet strong peak at 2384 cm−1, which dominates the Raman spectra above 40 GPa. This peak is clearly from the N≡N stretching vibration but is characterized from that of pure nitrogen in three important aspects as highlighted in Fig. 7: (i) it is a very broad peak in comparison with those of molecular N2. The FWHM of this newly observed peak from the current study is 50 cm−1 at 40 GPa, whereas the FWHM of the ν2 mode of ε-N2 at 40 GPa is less than 4 cm−1 as reported in Ref. 35. (ii) It is a singlet (very symmetric), in contrast to the multiplet (ν1, ν2a, ν2b, and ν2c in pure ε-N2 above 40 GPa).35 (iii) The peak position 2378 cm−1 at 40 GPa is significantly different from that of the highest intensity mode (ν2c) of ε-N2 at 40 GPa at 2393 cm−1 at the same pressure. Thus, we conclude that the 2378 cm−1 peak of N8 is not from ε-N2, but from the νs(N≡N) in the predicted C2v-N8.18 

FIG. 7.

Raman spectra at the N2 vibron region during decompression showing that the broad peak at 2386 cm−1 above 41 GPa is different from that of pure ε-N2.

FIG. 7.

Raman spectra at the N2 vibron region during decompression showing that the broad peak at 2386 cm−1 above 41 GPa is different from that of pure ε-N2.

Close modal

The observed Raman spectra at 54 GPa support the formation of molecular N8, rather than molecular N6 or V-shaped N5+ ions. Recall that the nearest N⋯N distance between two adjacent N3 ions in HA-I [Fig. 1(e)] at 3.5 Å is considerably greater than that between N3 and N2H5 ions at 2.8 Å. The hydrogen bonding between N2H5+ with N3 ions in HA-II further interferes the direct coupling between two adjacent N3 ions and thereby the formation of N6. Importantly, the observed νs(N=N) at 1665 cm−1 indicates the presence of N=N moiety, supporting the formation of N8 (N≡N—N—N=N—N—N≡N) than N6 (N≡N—N—N—N≡N).

The formation of covalently bonded N8 is also more plausible at high pressure than the formation of ionic N5+ and N8. The polymerization of alkali metal azides,9 for example, supports this conjecture. Even though the formation of N5+ and N8 have been experimentally reported17,22 at ambient conditions, it is not known if these species would be stable at high pressures. But, it seems unlikely, especially for the N5+ in the presence of another cation of N2H5+. On the other hand, the calculated Raman frequencies for the N8 anion22 do not agree at all with the present Raman spectra at 54 GPa.

The proposed N8 molecule is a new molecular nitrogen allotrope made of N≡N+—N—N=N—N—N+≡N chains.18 In fact, the strongest vibration for this C2v-N8 molecule has been calculated to be for the B2 mode at 2143 cm−1 corresponding to the terminal N≡N bonds,22 which is 241 cm−1 lower than our observed frequency at 2384 cm−1. The next strongest vibration has been reported for another B2 mode at 913 cm−1, which is also 254 cm−1 lower than the νs(N—N) observed at 1165 cm−1 in the present study. The difference (∼250 cm−1) in the wavenumber shift can be understood as a result of higher pressure in the present study, whereas those calculations are done for the ambient pressure. The intensity enhancement of the peaks at 675 and 640 cm−1 can be understood as the bending of linear N=N=N chain in N3 ions to bent N≡N+—N and/or N=N—N moiety. The peak at 1600 cm−1 signifies the νs(N=N). Based on these vibrational evidences, we argue that we have synthesized neutral N8 molecules by compressing HA above 40 GPa. Then, the broad vibrational band of the νs(NH) at 3323 cm−1 is from a byproduct of N8 molecules, such as, a saturated hydronitrogens like azanes in a general form of NnHn+2 (NH3 for n = 1, N2H4 for n = 2, etc.).

The formation of N8 molecules may involve, first, the charge redistribution in N3 ions from N=N+=N to N≡N+—N, assisted by the hydrogen bonding in HA-II; then, its reaction with N2H5+ to form N≡N+—N—NH2 and release NH3. Considering the crystal arrangement of HA [see Fig. 1(e)], it is likely that these processes occur concertedly between N2H5 and two adjacent N3 ions to form N≡N+N—N=N—N+N≡N. This reaction mechanism assumes that hydrogen atoms are captured by other N3 and/or N2H5+ ions to yield saturated hydronitrogens or azanes. Supporting the spectral evidence for this proposed mechanism is the intensity enhancement of the νb (NNN) at ∼670 cm−1 and the emergence of ν1(NH2) and ν3(NH3) into a single νs(NH) mode at 3323 cm−1. Upon decompression, N8 decomposes to N2 + 2N3, while azanes convert back to the counter ion, N2H5+.

The structural determination of phase II and N8 is quite challenging due to the photo-sensitivity of the sample, less number of resolvable peaks and inability to obtain any information from hydrogen atoms. However, the powder diffraction patterns obtained at 35 GPa are quite different from those of pure N2 phases [see Fig. 6(b)], whereas the Raman data clearly indicate the formation of ε-N2 below 25 GPa. Thus, we conjecture that only a small amount of N2 is formed upon decompression based on the absence of diffraction lines corresponding to pure N2. Note that our Raman system is capable of collecting the Raman signal of N2 less than 0.1%. Therefore, the intense diffraction line at 8.84° at 35 GPa most likely corresponds to N8, representing the packing of nitrogen atoms. In fact, we found that the observed d-spacing of 2.23 Å at 35 GPa is virtually comparable to the next nearest neighbor distance of cg-N (2.15 Å at 115 GPa)4 as well as the nearest N⋯N distance of 2.66 Å in ε-N2 at 35 GPa.34 Upon further decompression of the sample down to 3 GPa, the intensity of this intense peak decreases significantly; however, it still remains as the dominant peak. This suggests that some of the polymeric chains of N8 are metastable at low pressures down to 3 GPa.

We have presented the high-pressure Raman study of HA up to 68 GPa. Based on the present Raman data, it is most likely that HA-II transforms to N8 at 40 GPa. The Raman spectrum of N8 is characterized by a strong peak at 2376 cm−1 (at 40 GPa), which can be attributed to the terminal N≡N vibration of molecular N8 solid—a new nitrogen allotrope previously predicted to be stable at ambient conditions. Thus, we conjecture that the observed HA-II to N8 transformation possibly occurs together with azanes (NnHn+2) as the following: 8+4n[N2H5+N3]5N8+20NnHn+2. Note that for a given value of n greater than one, the number of moles produced for the product side of the reaction is less than that of the reactant ions present. Thus, this reaction is driven by the densification and favors to occur at high pressure, in accordance with the Le Chatelier’s principle.36 Finally, the present study also underscores the significance of HA in the synthesis of new forms of high energy density nitrogen allotropes and polymeric nitrogen, especially considering its metastability at relatively low pressures below 25 GPa down to 3 GPa at ambient temperature.

This work has been done in support of NEEC-NSWC (Grant No. N00174-16-C-0038), NSF-DMR (Grant No. 1701360), and ARO (Grant No. W911NF-17-1-0468). This work was also supported by the Defense Acquisition Program Administration and the Agency for Defense Development of Republic of Korea and by the National Research Foundation grant funded by the Korea government (Grant Nos. 2016K1A4A3914691 and U14358RF). We also thank Professor I. Olyenik at University of South Florida for scientific discussion. The present X-ray diffraction work was performed at HPCAT (Sector 16), Advanced Photon Source (APS), Argonne National Laboratory. We thank Dr. Ross Hrubiak for his assistance at the HPCAT beamline. HPCAT operations are supported by DOE-NNSA under Award No. DE-NA0001974 with partial instrumentation funding by NSF.

1.
P.
Yin
,
Q.
Zhang
, and
J. M.
Shreeve
,
Acc. Chem. Res.
49
,
4
(
2016
).
2.
M. I.
Eremts
,
A. G.
Gavriliuk
,
I. A.
Trojan
,
D. A.
Dzivenko
, and
R.
Boehler
,
Nat. Mater.
3
,
558
(
2004
).
3.
M. J.
Lipp
,
J. P.
Klepeis
,
B. J.
Baer
,
H.
Cynn
,
W. J.
Evans
,
V.
Iota
, and
C. S.
Yoo
,
Phys. Rev. B
76
,
014113
(
2007
).
4.
C.
Mailhiot
,
L. H.
Yang
, and
A. K.
McMahan
,
Phys. Rev. B
46
,
14419
(
1992
).
5.
U.
Teipal
.
Energetic Materials: Particle Processing and Characterization
(
John Wiley & Sons
,
2006
), p.
238
.
6.
V. E.
Zarko
,
Combust., Explos. Shock Waves
46
,
121
(
2010
).
7.
J. C.
Crowhurst
,
J. M.
Zaug
,
H. B.
Radousky
,
B. A.
Steele
,
A. C.
Landrville
, and
I. A.
Oleynik
,
J. Phys. Chem. A
118
,
8695
(
2014
).
8.
D.
Tomasino
,
M.
Kim
,
J.
Smith
, and
C. S.
Yoo
,
Phys. Rev. Lett.
113
,
205502
(
2014
).
9.
M. I.
Eremets
,
M. Y.
Popov
,
I. A.
Trojan
,
V. N.
Denisov
,
R.
Boehler
, and
R. J.
Hemley
,
J. Chem. Phys.
120
,
10618
(
2004
).
10.
M. I.
Eremets
,
R. J.
Hemley
,
H. K.
Mao
, and
E.
Gregoryanz
,
Nature
411
,
170
(
2001
).
11.
W.
Li
,
X.
Huang
,
K.
Bao
,
Z.
Zhao
,
Y.
Huang
,
L.
Wang
,
G.
Wu
,
B.
Zhou
,
D.
Duan
,
F.
Li
,
Q.
Zhou
,
B.
Liu
, and
T.
Cui
,
Sci. Rep.
7
,
39249
(
2017
).
12.
R. P.
Singh
,
R. D.
Verma
,
D. T.
Meshri
, and
J. M.
Shreeve
,
Angew. Chem. Int. Ed.
45
,
3584
(
2006
).
13.
E.
Horvath-Bordon
,
R.
Riedel
,
A.
Zerr
,
P. F.
McMillan
,
G.
Auffermann
,
Y.
Prots
,
W.
Bronger
,
R.
Kniep
, and
P.
Kroll
,
Chem. Soc. Rev.
35
,
987
(
2006
).
14.
M. I.
Eremets
,
M. Y.
Popov
, and
R. J.
Hemley
,
J. Chem. Phys.
120
,
10618
(
2004
).
15.
S. A.
Medvedev
,
M. I.
Eremets
,
J.
Evers
,
T. M.
Klapotke
,
T.
Palasyuk
, and
I. A.
Trojan
,
Chem. Phys.
386
,
41
(
2011
).
16.
B. A.
Steele
,
E.
Stavrou
,
J. C.
Crowhurst
,
J. M.
Zaug
,
V. B.
Prakapenka
, and
I. A.
Oleynik
,
Chem. Mater.
29
,
735
(
2017
).
17.
A.
Vij
,
W. W.
Wilson
,
V.
Vij
,
F. S.
Tham
,
J. A.
Sheehy
, and
K. O.
Christe
,
J. Am. Chem. Soc.
123
,
6308
(
2001
).
18.
M. J.
Greschner
,
M.
Zhang
,
A.
Majumdar
,
H.
Liu
,
F.
Peng
,
J. S.
Tse
, and
Y.
Yao
,
J. Phys. Chem. A
120
,
2920
(
2016
).
19.
X.
Wang
,
J.
Li
,
N.
Xu
,
H.
Zhu
,
Z.
Hu
, and
L.
Chen
,
Sci. Rep.
5
,
16677
(
2015
).
20.
Y.
Ma
,
A. R.
Oganov
,
Z.
Li
,
Y.
Xie
, and
J.
Kotakoski
,
Phys. Rev. Lett.
102
,
065501
(
2009
).
21.
B.
Hirshberg
,
R. B.
Gerber
, and
A. I.
Krylov
,
Nat. Chem.
6
,
52
(
2014
).
22.
Z.
Wu
,
E. M.
Benchafia
,
Z.
Iqbal
, and
X.
Wang
,
Angew. Chem. Int. Ed.
53
,
12555
(
2014
).
23.
P. G.
Chiglien
,
J.
Etienne
,
S.
Jaulmes
, and
P.
Laruelle
,
Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem.
30
,
2229
(
1974
).
24.
N. W.
Ashcroft
,
Phys. Rev. Lett.
92
,
187002
(
2004
).
25.
M. I.
Eremets
,
I. A.
Trojan
,
S. A.
Medvedev
,
J. S.
Tse
, and
Y.
Yao
,
Science
319
,
1506
(
2008
).
26.
H. K.
Mao
,
J.
Xu
, and
P. M.
Bell
,
J. Geophys. Res.
91
,
4676
, https://doi.org/10.1029/jb091ib05p04673 (
1986
).
27.
A. P.
Hammersley
,
S. O.
Svensson
,
M.
Hanfland
,
A. N.
Fitch
, and
D.
Häusermann
,
High Pressure Res.
14
,
235
(
1996
).
28.
C.
Prescher
and
V. B.
Prakapenka
,
High Pressure Res.
35
,
223
(
2015
).
29.
A. C.
Larson
and
R. B.
Von Dreele
, “
General structure analysis system (GSAS)
,” Report No. LAUR 86,
Los Alamos National Laboratory
,
2000
.
30.
B. H.
Toby
,
J. Appl. Cryst.
34
,
210
(
2001
).
31.
K.
Momma
and
F.
Izumi
,
J. Appl. Crystallogr.
44
,
1272
(
2011
).
32.
G.
Vaitheeswaran
,
N.
Yedukondalu
, and
B. M.
Abraham
,
J. Chem. Sci.
128
,
1631
(
2016
).
33.
F.
Zimmerman
,
T.
Lippert
,
C.
Beyer
,
J.
Stebani
,
O.
Nyuken
, and
A.
Wokaun
,
Appl. Spectrosc.
47
(
7
),
986
(
1993
).
34.
R. L.
Mills
,
B.
Olinger
, and
D. T.
Cromer
,
J. Chem. Phys.
84
,
2837
(
1986
).
35.
H.
Schneider
,
W.
Hafner
,
A.
Wokaun
, and
H.
Olijnyk
,
J. Chem. Phys.
96
(
11
),
8046
(
1992
).
36.
L.
Pauling
,
College Chemistry
, 3rd ed. (
Freeman
,
San Francisco
,
1964
).