We present the pressure-temperature (PT) induced physical and chemical transformations in ammonium perchlorates (APs) up to 50 GPa and 450 °C, using diamond anvil cells and confocal micro-Raman spectroscopy, which provide new constraints for the phase diagram of AP. The results show spectral evidences for three new polymorphs (III, IV, and VI) of AP, in addition to two previously known phases (I and II), at various PT conditions with varying degrees of hydrogen bonding and lack of strong spectral evidence for previously known high-temperature cubic phase (phase V). Upon further heating, AP chemically decomposes to N2, N2O, and H2O. The present phase diagram is, therefore, in sharp contrast to the previous one, underscoring a rich polymorphism, a large stability field for solids, and a replacement of the melt with a decomposition line.

Ammonium perchlorate (AP or NH4ClO4) is the salt product of perchloric acid (HClO4) and ammonia (NH3). It is also a powerful oxidizer that can be combined with fuels to manufacture non-conventional explosives.1–3 Mixing AP with fuels, polymeric binders, and metal (Al) powders generates a self-sustained combustion and makes it a powerful propellant used in rocket launching.3 Uncontrolled explosions involving AP, on the other hand, have caused significant damages, demonstrating its disastrous effects.4 Therefore, the research on AP has long been of interest to the propellant and energetic materials communities. Yet, AP is still one of the least understood energetic materials.2 This is in part due to the presence of chlorine in AP, which results in the decomposition mechanisms5,6 be more complex and different from other propellants and explosive ingredients.

Ammonium perchlorate has been subjected to an extensive level of thermal investigations at ambient conditions with a variety of techniques in order to understand its thermal decomposition mechanism5,6 and orientational dynamics in this ionic compound.7–9 At ambient conditions, AP crystallizes into an orthorhombic solid (phase I), which transforms to a cubic phase (phase V in Fig. 1(b)) at high temperatures above 238 °C–240 °C. A previous low-temperature neutron diffraction study8 attributed the gradual changes observed over the temperature range of 25 to −263 °C, to the rotational ordering of NH 4 + ions at low temperatures, within the same phase I. This result is, however, in contrast to a previous Raman study on AP single crystals,10 which has suggested structural phase transitions at −93 °C and −233 °C to phases with different orientational configurations of NH 4 + and ClO 4 ions.

FIG. 1.

(a) The phase diagram of AP reproduced from the previous studies.1,2,11 The previously suggested low pressure phase I-II11 and I-III2 boundaries are also marked as the red and blue bars in the inset, respectively. (b) The proposed phase diagram of AP based on the present Raman studies. For clarity, only data points at the vicinity of the proposed phase boundaries are shown in various symbols: △-phase I, □-phase II, ○-phase III, ✩-phase IV, •-phase V, and ♢-decomposition. The dashed lines are used to present the plausible phase boundaries and extensions of the proposed lines due to the scarcity of the available spectral details at high temperature above 250 °C below 30 GPa. The likelihood of the presence of phase V at low pressures is also presented. The crystal structure of the ambient P-T phase of AP is shown on top left (reproduced from crystal data in Ref. 8), where the N, Cl, and O atoms are shown in blue, green, and red balls, respectively. Since H atoms are disordered in this phase, their relative positions are not shown.

FIG. 1.

(a) The phase diagram of AP reproduced from the previous studies.1,2,11 The previously suggested low pressure phase I-II11 and I-III2 boundaries are also marked as the red and blue bars in the inset, respectively. (b) The proposed phase diagram of AP based on the present Raman studies. For clarity, only data points at the vicinity of the proposed phase boundaries are shown in various symbols: △-phase I, □-phase II, ○-phase III, ✩-phase IV, •-phase V, and ♢-decomposition. The dashed lines are used to present the plausible phase boundaries and extensions of the proposed lines due to the scarcity of the available spectral details at high temperature above 250 °C below 30 GPa. The likelihood of the presence of phase V at low pressures is also presented. The crystal structure of the ambient P-T phase of AP is shown on top left (reproduced from crystal data in Ref. 8), where the N, Cl, and O atoms are shown in blue, green, and red balls, respectively. Since H atoms are disordered in this phase, their relative positions are not shown.

Close modal

There have been several high-pressure studies on AP, reporting diverse results.2,11–15 For example, the early studies of Bridgman12 suggested a phase transition at 3.1 GPa at elevated temperatures. Similarly, a previous shock compression and a static powder x-ray diffraction (XRD) experiment on AP13 confirmed the presence of a phase transition at a slightly higher pressure, 4.7 GPa, to an asymmetric, non-cubic phase (either monoclinic or triclinic). The infrared study of Brill and Goetz,14 on the other hand, showed the disappearance of the ν 1 ClO 4 mode at 1–2.4 GPa and ambient temperature, which was attributed to the orthorhombic → cubic transition. Apparently, this is in contrast to a later result1 suggesting no phase transition to 26 GPa at ambient temperatures. Nevertheless, that same study1 suggested several phase transitions at high temperatures between 25 °C and 380 °C, and at high pressures to 26 GPa. Those include solid (orthorhombic) → solid (cubic), solid (orthorhombic or cubic) → liquid, and liquid → gas (decomposition) transitions, as reproduced in Fig. 1(a). Note that the pressure-induced orthorhombic to cubic transition in Fig. 1(a) shows a substantially weaker pressure dependence about −6.3 K/GPa to −7.5 K/GPa, than that of 216 K/GPa, reported by Richter and Pistorius.15 Interestingly, the study of Peiris et al.2 has shown clear spectral evidences for a few phase transitions in the pressure range below 5 GPa, including the discontinuous vibrational mode shifts above 0.9 GPa and 3 GPa and the emergence of new modes above 3 GPa. These transitions are marked in Fig. 1(a) as the blue vertical bars with pII and pIII labels, respectively. However, the XRD patterns obtained in the same study2 could not be indexed to an orthorhombic structure, underscoring low symmetry crystal structures for pII and pIII.2 The most recent study11 also found evidence for a first-order isostructural phase transition of phase I, but at a slightly higher pressure 4.6 GPa, in hydrostatic conditions. This transition is also marked in Fig. 1(a) as the vertical red bar and pII. Table I summarizes the above-discussed crystal structures of AP polymorphs.

TABLE I.

Crystal structures and stability fields of AP polymorphs. The proposed phase diagram in Fig. 1(b) is based on the phase identification presented herein. The phases I and II follow Hunter et al. identification,11 whereas the high temperature cubic phase is named phase V.

Phase Crystal structure Stable region Crystal ordering
Pnma, Z = 4  0–3.5 GPa at 25 °C  Freely rotating NH 4 + with ClO 4 locked in particular orientation within lattice sites 
a = 9.20 Å, b = 5.82 Å, c = 7.45 Å 
II  Pnma, Z = 4  above 4 GPa at 25 °C  Freezing of NH 4 + components and ordered ClO 4  
a ∼ 7.5 Å, b ∼ 6.25 Å, c ∼ 7.2 Å 
F4-3m, Z = 4  240 °C at ambient P  Freely rotating NH 4 + and almost freely rotating ClO 4  
a = 7.63 Å 
  Unknown  −93 to −233 °C at ambient P  Restricted rotation of NH 4 + and ordered ClO 4  
  Unknown  Below −93 °C at ambient P  Ordered NH 4 + and ClO 4 (both ions get reordered) 
Phase Crystal structure Stable region Crystal ordering
Pnma, Z = 4  0–3.5 GPa at 25 °C  Freely rotating NH 4 + with ClO 4 locked in particular orientation within lattice sites 
a = 9.20 Å, b = 5.82 Å, c = 7.45 Å 
II  Pnma, Z = 4  above 4 GPa at 25 °C  Freezing of NH 4 + components and ordered ClO 4  
a ∼ 7.5 Å, b ∼ 6.25 Å, c ∼ 7.2 Å 
F4-3m, Z = 4  240 °C at ambient P  Freely rotating NH 4 + and almost freely rotating ClO 4  
a = 7.63 Å 
  Unknown  −93 to −233 °C at ambient P  Restricted rotation of NH 4 + and ordered ClO 4  
  Unknown  Below −93 °C at ambient P  Ordered NH 4 + and ClO 4 (both ions get reordered) 

Clearly, the phase diagram of AP (Fig. 1(a))1,2,11 is poorly understood with some controversial results, underscoring several fundamental questions regarding the presence of phase transitions and the stability and structure of polymorphs. The present study is, therefore, focused on obtaining the phase diagram of AP in an extended region of pressures to 50 GPa and temperatures to 420 °C; the results are as summarized in Fig. 1(b). Note that the present phase diagram is very different from the previous one (Fig. 1(a)). In this paper, we will present the Raman spectral data, used to construct the present phase diagram, and explain this apparent discrepancy.

We used a highly polycrystalline form of AP (reagent grade from Alfa Aesar) without further purification. A small amount of AP samples is loaded in a small hole (0.1-0.2 mm in diameter and 0.03-0.1 mm in thickness), drilled on a pre-indented gasket, and placed on a membrane diamond anvil cell (m-DAC), together with a few small chips of ruby crystals. Depending on the required pressure, either 0.3 mm or 0.8 mm culet diamonds (∼0.3 carat of type Ia) were used. The internal pressures were determined from the temperature corrected R1 shift of ruby luminescence.16 The samples were externally heated to 420 °C using a band heater wrapped around the m-DAC, and the temperatures were measured using a K-type thermocouple in contact with one of the diamond anvils. Spatially resolved Raman spectra were collected using a custom-designed confocal micro-Raman system in a backscattering geometry with 514.5 nm excitation line of CW Ar+ ion laser. The laser power was maintained at a minimum level to avoid photo-induced transitions and/or decomposition of the sample.

High pressure-temperature (PT) experiments were performed along several thermal paths: (i) isothermal compressions at 20, 100, and 200 °C, (ii) isobaric heating at ∼3, 10, and 20 GPa, and (iii) quasi-isobaric heating below 2 GPa. High-pressure experiments were carried out using both SS and Re gaskets. However, only SS gaskets were used for high-temperature experiments to avoid chemical reactions of AP with Re observed above ∼230 °C.

Figure 2 shows typical appearances of AP samples at various PT conditions: along (a) isothermal compression at 200 °C and (b) isobaric heating at 20 GPa. Note that all AP samples remain solids at these PT conditions, which is evident from the presence of grain boundaries. It is also evident that the sample develops a reddish color just before the decomposition at 398 °C at 20 GPa, which in turn precludes discerning the melting of AP at high pressures. The extrapolation of this decomposition to ambient pressure is reasonably well compared with the ambient-pressure melting temperature, 258 °C.5 Therefore, the previously suggested low melting temperature (e.g., ∼20 °C at 24 GPa as shown in Fig. 1(a))1 may be due to a photo-induced chemical decomposition, which can also produce liquid or gaseous bubbles at relatively low temperatures upon illumination of intense laser pulses.

FIG. 2.

Micro-photographic images of AP samples, showing the visual appearance changes upon (a) isothermal compression at 200 °C and (b) isobaric heating at ∼20 GPa. The presence of lattice strains in all images confirms that AP remains as a solid in these pressure-temperature ranges in contrast to the previously suggested phase diagram.1 Note in (b) that AP discolorates at ∼398 °C at 20 GPa, just prior to decomposition.

FIG. 2.

Micro-photographic images of AP samples, showing the visual appearance changes upon (a) isothermal compression at 200 °C and (b) isobaric heating at ∼20 GPa. The presence of lattice strains in all images confirms that AP remains as a solid in these pressure-temperature ranges in contrast to the previously suggested phase diagram.1 Note in (b) that AP discolorates at ∼398 °C at 20 GPa, just prior to decomposition.

Close modal

Raman spectroscopy is useful to probe the phase transitions and map out the stability regions of phases at high PT conditions. Phase I of AP in an orthorhombic Pnma space group belongs to the D2h16 symmetry group. The factor group analysis of this phase gives a total of 60 Raman active modes (18Ag, 12B1g, 18B2g, and 12B3g). Table II summarizes the frequencies of Raman active modes for the phase I. A good agreement is found in the observed Raman modes between this and the previous studies.9,10,17–19 However, because of the relatively high cutoff frequency of the Raman notch filter used in the present setup, we were unable to measure the low frequency modes below 200 cm−1 that include lattice modes and weak internal modes of NH 4 + .

TABLE II.

Raman active vibrational modes of AP observed in previous studies at ambient conditions. The primed and unprimed ν represent NH 4 + and ClO 4 modes, respectively.

Ref. 10 (cm−1) Reference 17 (cm−1) References 9 and 18* (cm−1) Reference 19 (cm−1) Assignment
42    35  20  Lattice 
    51  60  Lattice 
72      81  Lattice 
150    150  140 158  Lattice 
      199 220  Lattice 
463  461  461*  463  ν2 ( ClO 4 ), (Ag, B2g
627  627  625*  627  ν4 ( ClO 4 ), (Ag, B3g
632    630*  632  ν4 ( ClO 4 ), (Ag, B2g
637    635*  637  ν4 ( 37 ClO 4 ), B2g 
922    921*  921.5  2 ( ClO 4 ), (Ag, B2g
937  933  935*  936.5  ν1 ( ClO 4 ), Ag 
1078  1060  1064*  106 3  ν3 ( ClO 4 ), B3g 
1109  1104  1108*  110 7  ν3 ( ClO 4 ), (Ag, B2g
1133  1129    113 3  ν3 ( ClO 4 ), Ag 
1410  1420      ν 4 ( NH 4 + ), (Ag, B1g
  1680      ν 4 ( NH 4 + ), B2g 
3212.5  3209      ν 1 ( NH 4 + ), Ag 
3267  3350      ν 3 ( NH 4 + ), (Ag, B2g
Ref. 10 (cm−1) Reference 17 (cm−1) References 9 and 18* (cm−1) Reference 19 (cm−1) Assignment
42    35  20  Lattice 
    51  60  Lattice 
72      81  Lattice 
150    150  140 158  Lattice 
      199 220  Lattice 
463  461  461*  463  ν2 ( ClO 4 ), (Ag, B2g
627  627  625*  627  ν4 ( ClO 4 ), (Ag, B3g
632    630*  632  ν4 ( ClO 4 ), (Ag, B2g
637    635*  637  ν4 ( 37 ClO 4 ), B2g 
922    921*  921.5  2 ( ClO 4 ), (Ag, B2g
937  933  935*  936.5  ν1 ( ClO 4 ), Ag 
1078  1060  1064*  106 3  ν3 ( ClO 4 ), B3g 
1109  1104  1108*  110 7  ν3 ( ClO 4 ), (Ag, B2g
1133  1129    113 3  ν3 ( ClO 4 ), Ag 
1410  1420      ν 4 ( NH 4 + ), (Ag, B1g
  1680      ν 4 ( NH 4 + ), B2g 
3212.5  3209      ν 1 ( NH 4 + ), Ag 
3267  3350      ν 3 ( NH 4 + ), (Ag, B2g

Figure 3 shows the Raman spectra of four AP polymorphs observed at high pressures and ambient temperature. Phase I exhibits a strong singlet for symmetric ClO 4 bending mode (ν2) at 463 cm−1, a strong singlet for symmetric ClO 4 stretching mode (ν1) at 937 cm−1, a medium doublet with different intensities for asymmetric ClO 4 deformation mode (ν4) at 627 and 632 cm−1, and weak and very features for asymmetric ClO 4 stretching modes (ν3) between 1070 and 1135 cm−1, as summarized in Table II. The overtone of the bending mode (2ν2) is clearly discernible as a low wavenumber shoulder of ν1 arising from the Fermi-resonance enhancement.17 Upon compression to 4.5 GPa and above, new modes start to appear on the low wavenumber side of both ν2 (marked with an asterisk in Fig. 3) and ν4 modes (a bent arrow), as shoulders, whereas the overtone (2ν2) moves further towards the ν1 mode. The intensity of the high wavenumber mode of the ν4 doublet starts increasing, which eventually becomes comparable. The ν3 modes, on the other hand, become noticeable and well separated. Importantly, the observed Raman frequencies of all these ClO 4 modes as well as NH 4 + modes above 4.5 GPa are in agreement with the phase identified above 3 GPa in the previous study of Peiris et al.2 (noted as pIII in Fig. 1(a)). However, we were unable to distinguish additional new peaks appearing on the high wavenumber side of the ν1 mode.2 Having recognized only one transition in this pressure range below 5 GPa, we considered the pIII in Ref. 2 to be the same as the pII in Ref. 11. Thus, we followed the nomenclature in Ref. 11 and denoted this phase as phase II in Fig. 1(b) (also in Table I).

FIG. 3.

Raman spectra of pure AP corresponding to different phases observed upon isothermal compression at room temperature (21 °C). The asterisks (*) and arrows (→) show the features that were the basis for the proposed transitions. Since NH 4 + modes were relatively weaker than the ClO 4 modes, only the changes observed in the internal modes of ClO 4 were shown here.

FIG. 3.

Raman spectra of pure AP corresponding to different phases observed upon isothermal compression at room temperature (21 °C). The asterisks (*) and arrows (→) show the features that were the basis for the proposed transitions. Since NH 4 + modes were relatively weaker than the ClO 4 modes, only the changes observed in the internal modes of ClO 4 were shown here.

Close modal

The pressure-induced Raman peak shifts (Fig. 4(a)) show several discontinuous spectral changes, signifying the onsets of phase transitions from phase I to II at ∼4.5 GPa, III at 10 GPa, and IV at 27 GPa. Subtle spectral changes in FWHM of the Raman peaks (Fig. 4(b)) further support the proposed transitions. The phase II → III transition at ∼10 GPa is, for example, characterized by appearance of a new peak on the ν2 doublet as a shoulder on the high wavenumber side. At the onset of the transition, the shoulder observed on the low wavenumber side in phase II becomes more discernible, such that it appears as a doublet. In addition, the intensity of the shoulder peak of the ν4 in phase II increases and becomes comparable to the other two peaks. The overtone (2ν2) seems hidden underneath the strong ν1 mode. As a result, the ν1 mode becomes sharper as shown in the FWHM of peaks in Fig. 4(b). The ν3 modes, on the other hand, become diffused while still maintaining the three-peak characteristic. However, these modes start merging to form a two-peak feature with pressure.

FIG. 4.

(a) The pressure-induced Raman peak shifts of the ν2 (in circles) and ν4 (in squares) modes of AP at room temperature, indicating the onset of phase transitions by vertical lines. The linear fits for phase IV remain to 55 GPa—the maximum pressure studied. (b) The normalized spectral FWHM of the ν2 (black circles) and ν1 (red pentagons) modes at three different temperatures 20, 100, and 200 °C. The onset of phase transition is indicated by the vertical lines, showing temperature-dependent transition pressures for the proposed phase transitions. The numbers on the vertical axes (left and right) are, respectively, the scales for the ν2 (black) and ν1 (red) modes.

FIG. 4.

(a) The pressure-induced Raman peak shifts of the ν2 (in circles) and ν4 (in squares) modes of AP at room temperature, indicating the onset of phase transitions by vertical lines. The linear fits for phase IV remain to 55 GPa—the maximum pressure studied. (b) The normalized spectral FWHM of the ν2 (black circles) and ν1 (red pentagons) modes at three different temperatures 20, 100, and 200 °C. The onset of phase transition is indicated by the vertical lines, showing temperature-dependent transition pressures for the proposed phase transitions. The numbers on the vertical axes (left and right) are, respectively, the scales for the ν2 (black) and ν1 (red) modes.

Close modal

Spectral evidences for the phase III → IV transition at 27 GPa, on the other hand, include (i) the intensity of triplet ν2 increasing, upon further compression, featuring satellite peaks on either side of the strong peak, (ii) the appearance of a new shoulder on the low wavenumber side of the asymmetric ν4 triplet, (iii) a slight broadening of the ν1 mode, and (iv) a two diffused-peak feature for ν3 mode. Although, it may seem that both ν2 and ν4 modes have characteristics of phase III, the discontinuities observed for both wavenumber shifts (Fig. 4(a)) and the FWHM analysis of peaks (Fig. 4(b)) support this proposed transition.

All observed Raman peaks of AP phases shift to higher wavenumber (blue shift) as pressure increases (see Fig. 4(a)). This is first hand expected as compression leads to a smaller unit cell (or a larger density) and enhances both intra- and inter-molecular bond strengths. Yet, the detailed spectral changes in peak positions, intensities, and widths across the phase transition can provide further information regarding molecular interactions in governing phases. Table III, for example, compares the observed and expected Raman frequencies of governing phases at the onset pressures of three solid-solid transitions. The expected values are the extrapolated peak positions of the parent phases. At the onset of phase I → II transition, the observed peak position for ν 1 decreases, whereas that for ν3 increases with no significant change for ν4. Such spectral changes are indicative of the strengthening of intermolecular hydrogen bonding upon the transition.10 The formation and/or strengthening of hydrogen bonding between the NH 4 + and ClO 4 ions in phase II leads to a restricted rotational freedom of NH 4 + ions. The ordered state of NH 4 + ions results in a reduction of their local site symmetry, which gives rise to additional peaks in the Raman spectrum, as observed across the phase I to II transition in Fig. 3. Thus, the present Raman changes support an ordered state of NH 4 + ions in phase II. This observation is then in agreement with the previous studies,2,11 suggesting restricted (either freezing or slowing down) rotations of NH 4 + ions upon the corresponding structural phase transition (i.e., from pI to pII in Ref. 11 or pII to pIII in Ref. 2).

TABLE III.

Comparison of Raman frequencies of selected vibrational modes between the initial and final phases associated with the phase transitions of I → II, II → III, and III → IV at the nearest onset pressures measured experimentally.

ν3 (cm−1) ν4 (cm−1) ν 1 (cm−1)
P (GPa) Parent New Parent New Parent New
5.4  1087.6  1089.2  634.1  634.0  3252.0  3238.2 
12.0  1101.8  1112.2  638.5  639.6  3265.0  3265.8 
27.4  1136.3  1138.9  671.1  669.4  3331.1  3328.3 
ν3 (cm−1) ν4 (cm−1) ν 1 (cm−1)
P (GPa) Parent New Parent New Parent New
5.4  1087.6  1089.2  634.1  634.0  3252.0  3238.2 
12.0  1101.8  1112.2  638.5  639.6  3265.0  3265.8 
27.4  1136.3  1138.9  671.1  669.4  3331.1  3328.3 

Upon phase II → III transformation, the observed peak positions for all ν3, ν4, and ν 1 modes increase (Table III). This indicates that the intra-ionic N–H bonding between partially ordered NH 4 + ions become strengthened, giving rise to a complete ordering of NH 4 + ions in phase III. As a result, we observed a shoulder peak for the ν 1 ( NH 4 + ) mode (see blue spectra in Fig. 5). This ordering in turn can induce subtle distortions in the tetrahedral (Td) geometry of ClO 4 ions in the lattice, giving rise to new Raman peaks near the ν2 and ν4 (marked in asterisks and arrows in Fig. 3).2 In contrast, at the onset of phase III → IV transition, we observed a decrease in Raman shift for ν 1 and an increase for ν3 and ν4 modes. Considering a similar spectral analysis applied for the I-II and II-III transformations, we conjecture that III → IV is a structural phase transition to an ordered state of AP.

FIG. 5.

Raman spectra of pure AP phases, phase I at 3 GPa (black), II at 10 GPa (pink) and III at 20 GPa (blue), showing in comparison with that of high-temperature phase V. Note that the intensity of the high-temperature spectra decreases by approximately three folds, yet all peaks are still present.

FIG. 5.

Raman spectra of pure AP phases, phase I at 3 GPa (black), II at 10 GPa (pink) and III at 20 GPa (blue), showing in comparison with that of high-temperature phase V. Note that the intensity of the high-temperature spectra decreases by approximately three folds, yet all peaks are still present.

Close modal

Raman spectra obtained during isothermal compressions at 100 °C and 200 °C also exhibit pressure-induced blue shifts analogous to those at ambient temperature. The similarities observed for the spectral changes at high temperatures in peak positions (not shown) and FWHM (Fig. 4(b)) indicate the presence of phases I, II, and III below 23 GPa. A majority of Raman modes of the same phase at comparable pressures show a subtle red shift (towards lower wavenumbers) or no changes, as temperature increases. This softening of vibrational modes is, again, expected as the lattice expands at elevated temperatures.

Figure 5 illustrates how the Raman spectra of phases I, II, and III change as the temperature is further elevated beyond 200 °C below 23 GPa. As each phase is further heated, the weak spectral features in all three phases become less distinguishable with all peaks in each phase becoming broadened. The new ClO 4 modes observed in phases II and III are still maintained above 200 °C as shown in Fig. 5. However, in comparison to the high temperature Raman spectra from these two phases (II and III), a noticeable red shift (towards lower wavenumbers) is observed in the high temperature Raman spectrum of phase I, especially for the comparatively stronger vibrational modes. Recall that our present instrumental setup does not allow the observation of lattice modes. Hence these observations on vibrational modes at elevated temperatures alone do not explain the existence of the previously reported phase V, as both the linewidth and frequency can be affected due to lattice anharmonicity at high temperature. However, we noted the possibility of the existence of high temperature phase V at low pressures (see Fig. 1(b)) based on the spectral similarities observed for vibrons in the high temperature cubic AN phase, according to our previous AN study.20 

Above 23 GPa, the phase IV is stable at least to 50 GPa at ambient temperature. However, upon elevating temperatures above ∼200 °C, we observed several notable spectral differences from phase IV above 23 GPa and 200 °C. These differences are shown in Fig. 6 and summarized as follows: (i) instead of a triplet, featuring satellite peaks in which the intensities increase with pressure as in phase IV, only a doublet is observed for the ν2 mode at 200 °C, which becomes diffused with pressure. (ii) New modes do not appear at the low wavenumber side of the ν4 mode (which is an asymmetric triplet), at the onset of the transition at 200 °C, unlike in phase IV. Instead, the lowest wavenumber peak of that triplet starts red shifting and becomes weaker with pressure. (iii) The highest wavenumber peak of ν2 modes at 100 °C and 200 °C has pressure dependences of 3.67 cm−1/GPa and 2.02 cm−1/GPa, respectively. Thus, unlike phase IV at lower temperatures, this mode does not engulf the lowest wavenumber ν4 mode, even at 46 GPa, (iv) the pressure dependence of ν1 modes varies only slightly, with relative intensities at 200 °C being considerably lower than that at low temperatures. (v) The ν3 modes appear as a two peak feature, which are considerably blue shifted to higher wavenumber at 200 °C, with the higher wavenumber mode having a higher intensity. (vi) The ν 1 mode shows a significantly weaker pressure dependent shift at 200 °C than that in phase IV. These observations suggest that the high-pressure phase observed at 200 °C and above 23 GPa is different from phase IV. Hence, we signify this new phase as phase VI in Fig. 1(b).

FIG. 6.

Raman spectral changes of pure AP at 100 (pink) and 200 °C (blue) above 23 GPa, showing the characteristics of phases IV and VI, respectively.

FIG. 6.

Raman spectral changes of pure AP at 100 (pink) and 200 °C (blue) above 23 GPa, showing the characteristics of phases IV and VI, respectively.

Close modal

The Raman spectra of phase VI (Fig. 6) show a remarkable red shift (to the lower wavenumber) for ν 1 mode and a significant blue shift (to the higher wavenumber) for ν3 mode in comparison to phase IV, at similar pressures. Following the same arguments stated before, we can say that further strengthening of the hydrogen bonding is observed in phase VI compared to that in phase IV. Note that the rate of change in the ν 1 mode is significantly smaller in phase VI (2.86 GPa/cm−1) than that in phase IV (4.64 GPa/cm−1). This means that the phase VI is less stiff than phase IV, arising from the strong hydrogen bonding network. In general, for molecular crystals, a softening of vibrational modes occurs due to lattice expansion, as the temperature is increased. Yet, the anomalous blue shift observed for the ν3 mode in phase VI with temperature may be due to a larger anharmonicity. The origin of this anharmonicity is not known due to lack of crystal structure information. However, based on the above characteristics, phase VI is likely an ordered phase with a different structure from phase IV.

Upon further heating, AP undergoes chemical decomposition, as observed previously.1 The decomposition accompanies a sudden color change of the sample to orange (see Fig. 2) that subsequently becomes opaque. In addition to the depletion of strong AP signals, new Raman peaks from N2 (2353 cm−1),21,22 N2O (527 cm−1),23 and H2O (652 cm−1)24 appear as shown in Fig. 7. However, “wet spots” which were attributed to melting in previous studies1 were not observed in these images, during heating and/or upon compression to higher pressures. The decomposition of phase VI was not measured above 20 GPa because of chemical reactions of AP with Re (and also Inconel) gaskets1 or mechanical failures of SS gaskets above 20 GPa and 300 °C.

FIG. 7.

The Raman spectra of AP at 365 °C (black) and 380 °C (red), at ∼9.4 GPa, showing the evidence for decomposition to N2O and N2.

FIG. 7.

The Raman spectra of AP at 365 °C (black) and 380 °C (red), at ∼9.4 GPa, showing the evidence for decomposition to N2O and N2.

Close modal

The present results provide new constraints for the phase diagram of AP as presented in Fig. 1(b). This phase diagram is different from the previous one in Fig. 1(a) in several aspects: (i) the presence of rich polymorphism, consisting of five solid polymorphs, (ii) the stability of solid phases at least to 50 GPa at ambient temperature, and (iii) the decomposition at high temperatures prior to melting, with detailed comparison as follows.

The present phase diagram consists of five polymorphs of AP. All solid-solid transitions occur reversibly with a little pressure hysteresis (within ±1 GPa). At ambient temperature, phase I (orthorhombic) transforms to II at 4.5 GPa, III at 9.9 GPa, and IV at 27 GPa—all are orientationally ordered. At high temperatures above 200 °C, phase IV transforms to phase VI. Upon further heating, decomposition to N2, N2O, and H2O is observed, hampering the observation of AP melting.

Both the previous (Fig. 1(a)) and present (Fig. 1(b)) phase diagrams show that the decomposition temperatures continue to increase with pressure. However, according to the present data, the previous study seems to overestimate the decomposition temperatures at all pressures to about 7 GPa (the maximum pressure that shows decomposition in Fig. 1(a)). Note that the decomposition temperature at ambient pressure in Fig. 1(b) is from the previous measurement,5 which is about 20 °C lower than the decomposition temperature reported in Fig. 1(a). On the other hand, the phase I (orthorhombic) → V (cubic) transition temperature shows a subtle decrease with pressure in Fig. 1(a), whereas our study lacks sufficient evidence to report its pressure dependence. The likelihood of its existence (phase V) at low pressure (to 3 GPa) is noted in a dashed line using the previously reported transition temperature at ambient conditions.25 The pressure dependent transition pressures corresponding to the melt line in the previous phase diagram in Fig. 1(a) predict melt AP at 25 GPa and 100 °C, where we observed the solid (phase III)–solid (IV) transition.26 

Present phase I → II transition is consistent with the previously suggested phase II → III transition by Peiris et al.2 and phase I → II transition by Hunter et al.11 albeit small differences in the transition pressures. Under non-hydrostatic conditions, the phase I → II transition (i.e., pII → pIII in blue in Fig. 1(a)) occurs at 3.3 GPa2 compared to the same transition at 4.5 GPa in present study. Under hydrostatic conditions, on the other hand, the same transition was reported to occur at different pressures in two different studies (above 3.0 GPa2 and over a pressure range of 3.98 and 4.6 GPa11) when two different pressure media (Fluorinert FC–752 and 4:1 methanol: ethanol,11 respectively) were used. Therefore, the apparent differences in the phase I → II transition pressure among the different experiments seem to reflect the differences in hydrostaticity of the pressure media in samples, time scale of pressure loadings, and also the sluggish nature of the transition. On the other hand, the present phase diagram shows no solid-solid phase transitions below 4 GPa. This is consistent with the previous observation by Hunter et al.11 but is in contrast to the solid-solid transition suggested at 1.5 GPa by Peiris et al.2 We attribute this apparent difference due to the sparseness of experimental data at smaller pressure intervals below 3 GPa and/or the sluggish nature of the transition.

The present Raman data provide some insights into the nature of hydrogen bonding in AP phases. For example, the pressure dependent Raman peak shifts seem to suggest that the strength of hydrogen bonding increases across the phase transitions at room temperature, whereas it weakens within the phase as a function of pressure. Interestingly the high-pressure high-temperature phase VI maintains strong hydrogen bonding in comparison to its low temperature counterpart, phase IV.

A limited amount of spectral evidences for the decomposition products challenges to understanding the decomposition mechanism of AP at high pressures. Previous studies suggest that it is possible to have relatively mobile NH 4 + units diffusing through a matrix of slowly rotating ClO 4 ions24 in AP’s cubic phase, which facilitates proton transfer that initiates its decomposition.17 The NH3 and HClO4 thus formed can be further oxidized to form N2 and N2O similar to AN,20 which explains our present observations. In addition, the increase in the degree of order in the high temperature phases of I, II, and III with pressure, thus explains the increase in decomposition temperatures as a function of pressure according to our studies.

In conclusion, the present study presents spectral evidences for the presence of reversible pressure-induced transitions resulting in new phases III, IV, and VI at varying P-T conditions in addition to confirming the stability of phases I and II in a wide P-T regime. Our revised phase diagram mapped out to 450 °C and 50 GPa consists of well-defined phase boundaries for phases I, II, III, and IV, as well as the decomposition line. Lack of ample spectral evidences has limited our ability to predicting the boundaries for phases V and VI. The nature and degree of hydrogen bonding in these different phases may play a role in its energetic behavior and thus needs to be further investigated.

This material is based upon work supported by the U.S. Department of Homeland Security, Science and Technology Directorate, Office of University Programs under Grant Award No. 2013-ST-061-ED0001; Defense Threat Reduction Agency under Grant Award No. HDTRA1-12-01-0020); and National Science Foundation, Division of Materials Research under Grant No. 1203834. The views and conclusions contained in this document are those of the authors and should not be interpreted as necessarily representing the official policies, either expressed or implied, of the U.S. Department of Homeland Security, Defense Threat Reduction Agency, and/or National Science Foundation.

1.
M. F.
Foltz
and
J. L.
Maienschein
,
Mater. Lett.
24
,
407
(
1995
).
2.
S. M.
Peiris
,
G. I.
Pangilinan
, and
T. P.
Russell
,
J. Phys. Chem. A
104
,
11188
(
2000
).
3.
C. R.
Pulham
,
D. I. A.
Millar
,
I. D. H.
Oswald
, and
W. G.
Marshall
, “
High pressure studies of energetic Materials
,” in
High-Pressure Crystallography: From Fundamental Phenomena to Technological Applications NATO Science for Peace Security Series B: Physics Biophysics
, edited by
E.
Boldyreva
and
P.
Dera
(
Springer-Science Business Media
,
Netherlands
,
2010
), pp.
447
457
.
4.
J. C.
Oxley
,
Terrorism Political Violence
5
,
30
(
1993
).
5.
J. C.
Oxley
,
J. L.
Smith
, and
B. R.
Valenzuela
,
J. Energ. Mater.
13
,
57
(
1995
).
6.
V. V.
Boldyrev
,
Thermochim. Acta
443
,
1
(
2006
).
7.
H. G.
Smith
and
H. A.
Levy
,
Acta Crystallogr.
15
,
1201
(
1962
).
8.
C. S.
Choi
,
H. J.
Prask
, and
E.
Prince
,
J. Chem. Phys.
61
,
3523
(
1974
).
9.
G. J.
Rosasco
and
H. J.
Prask
,
Solid State Commun.
16
,
135
(
1975
).
10.
T.
Chakraborty
,
S. S.
Khatri
, and
A. L.
Verma
,
J. Chem. Phys.
84
,
7018
(
1986
).
11.
S.
Hunter
,
A. J.
Davidson
,
C. A.
Morrison
,
C. R.
Pulham
,
P.
Richardson
,
M. J.
Farrow
,
W. G.
Marshall
,
A. R.
Lennie
, and
P. J.
Gould
,
J. Phys. Chem. C
115
,
18782
(
2011
).
12.
P. W.
Bridgman
,
Proc. Am. Acad. Arts Sci.
72
,
45
(
1937
).
13.
F. W.
Sandstorm
,
P. A.
Persson
, and
B.
Olinger
,
AIP Conf. Proc.
309
,
1409
(
1994
).
14.
T. B.
Brill
and
F.
Goetz
, in
Papers in Astronautics and Aeronautics
, edited by
T. L.
Boggs
and
B. T.
Zinn
(
AIAA
,
1978
), Vol.
63
, pp.
3
19
.
15.
P. W.
Richter
and
C. F. W. T.
Pistorius
,
J. Solid State Chem.
3
,
434
(
1971
).
16.
Fluorescence wavelength pressure Calculator, Java program written by james badro where the pressure dependence of ruby scale was obtained from,
H. K.
Mao
,
J.
Xu
, and
P. M.
Bell
,
J. Geophys. Res.
91
,
4673
, doi:10.1029/JB091iB05p04673 (
1986
);
The temperature dependence was from,
D. E.
McCumber
and
M. D.
Sturge
,
J. Appl. Phys.
34
,
1682
1684
(
1963
);
The script is based on FORTRAN program by Florent Occelli.
17.
Y. A.
Gruzdkov
,
J. M.
Winey
, and
Y. M.
Gupta
,
J. Phys. Chem. A
112
,
3949
(
2008
).
18.
D. J. J.
Van Rensburg
and
C. J. H.
Schutte
,
J. Mol. Struct.
11
,
229
239
(
1972
).
19.
W.
Zhu
,
T.
Wei
,
W.
Zhu
, and
H.
Xiao
,
J. Phys. Chem. A
112
,
4688
(
2008
).
20.
M.
Dunuwille
and
C. S.
Yoo
,
J. Chem. Phys.
139
,
214503
(
2013
).
21.
R.
Bini
,
L.
Ulivi
,
J.
Kreutz
, and
H. J.
Jodl
,
J. Chem. Phys.
112
,
8522
(
2000
).
22.
H.
Olijnyk
,
H.
Däufer
,
M.
Rubly
,
H. J.
Jodl
, and
H. D.
Hochheimer
,
J. Chem. Phys.
93
,
45
54
(
1990
).
23.
V.
Iota
,
J.-H.
Park
, and
C. S.
Yoo
,
Phys. Rev. B
69
,
064106
(
2004
).
24.
D. M.
Carey
and
G. M.
Korenowski
,
J. Chem. Phys.
108
,
2669
2675
(
1998
).
25.
L.
Rosso
and
M. E.
Tuckerman
,
Solid State Ionics
161
,
219
(
2003
).
26.
M.
Dunuwille
, Ph.D. dissertation,
Washington State University
,
2015
.