Ammonium Nitrate Crystal Structure, Polymorphism And Thermal Behavior.pdf

Chapter

AMMONIUM NITRATE: CRYSTAL STRUCTURE, POLYMORPHISM AND THERMAL BEHAVIOR Santiago García-Granda* and Jose M. Montejo-Bernardo Department of Physical and Analytical Chemistry. Faculty of Chemistry. University of Oviedo. Asturias, Spain.

ABSTRACT This chapter is divided into two parts. In the first part, the different polymorphs of ammonium nitrate are presented, their crystal structures are detailed, and the structural changes occurring in the successive phase transformations are described. At atmospheric pressure, at least five stable polymorphs have been reported (V, IV, III, II and I, from the lowest to highest temperature), with generally accepted transitions points at 255, 305, 357, 398 and 443 K (melting point), respectively. A sixth phase (VI) has been reported to exist at high temperature and high pressure, and a seventh phase (VII), existing at temperatures below 73 K, has been also put forward. These phase transitions are sometimes accompanied by significant volume changes in the unit cell, causing dimensional instability in the crystal and leading to disagreeable effects such as caking or breakage of the product. *

Corresponding author: Email: [email protected].

2

Santiago García-Granda and Jose M. Montejo-Bernardo In the second part, thermal aspects of ammonium nitrate are analyzed, focusing on the stability of each phase and the thermal behavior of phase transitions. Phase transitions are neither simple nor fully understood. Transition temperatures are known to actually depend of several variables, such as water content, impurity content, and the thermal history of the sample. Moreover, several of these transitions show considerable thermal hysteresis when the compound is analyzed via heating-cooling cycles. However, under certain conditions, phases II and IV are metastable with respect to phase III and reversible transitions between them are possible without passing through intermediate phase III.

INTRODUCTION Ammonium nitrate (AN) is a hygroscopic compound which crystallizes in ionic solids consisting of tetrahedral NH4+cations and planar triangular NO3anions. Due to their chemical nature, these ions interact with each other not only via electrostatic forces, but also through hydrogen bonds. There is at present no clear consensus among researchers as to the number of acknowledged crystalline phases of this salt and there is likewise disagreement regarding the existence or not of new crystalline phases when working under certain conditions of pressure and/or temperature. Even in phases that are already considered properly characterized, the emergence of new studies (generally based on Raman, IR and THz techniques, among others) puts some commonly accepted ideas into question. At the present time, the existence of five crystalline phases [1-3] is accepted under atmospheric pressure conditions, up to a temperature of 170ºC, at which the solid melts. These have been acknowledged since at least the beginning of the last century [4]. The phases are labeled with Roman numerals in the widely used system [5], in which I represents the phase which is stable at high temperatures. Figure 1 shows the generally accepted transition temperatures between these five phases, although these values may vary slightly depending on the source consulted [2, 6-9]. The determination of the internal structure of the five existing crystalline phases of ammonium nitrate at atmospheric pressure is a process whose origins date back more than eighty years and led to the emergence of a major body of work on the subject for more five decades, sometimes with conflicting results. In all cases, the final structures (unit-cell parameters, space group and atomic arrangement) were obtained by diffraction techniques (X-ray and neutron, employing single crystal or polycrystalline samples). In parallel to

Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 3 these studies, many others appeared based on other spectroscopic techniques (NMR, IR, Raman), which either supported or put into question the results published in diffraction studies. The first part of the chapter details the process of identification and structural determination of the existing phases of ammonium nitrate, primarily using X-ray and neutron diffraction techniques, and shows the importance of using these other spectroscopic techniques to put research on the right track. The second part of the text deals with the structural changes occurring in the successive phase transformation and analyses thermal aspects of the salt, focusing on the stability of each phase and its thermal behavior.

1. PHASES AND CRYSTAL STRUCTURE 1.1. Phases at Atmospheric Pressure Using X-ray powder diffraction data (taken at 195K and 240K), a paper published by Hendricks et al. in 1932 analyzing the crystal structure of the five known phases of ammonium nitrate under atmospheric pressure [1] proposed a hexagonal-like structure for phase V, with six molecules of salt in the unit cell. However, a subsequent study [10] based on single crystal X-ray diffraction data (at 120 K) describes it as tetragonal, with a P42 space group, eight AN molecules in the unit cell and a disordered arrangement in the case of the nitrate groups. Nevertheless, subsequent work showed that this proposal could not be correct since it was inconsistent with the results obtained by studying the crystal by infrared [11] and Raman [12] spectroscopy. Finally, two studies appeared almost simultaneously based on a Rietveld refinement of the neutron diffraction profile, at 78K [3] and 233K [13], respectively, which determined that phase V crystallizes in the orthorhombic system, in the Pccn space group, with eight molecules of the salt in the unit cell (Table 1). (As a curiosity, note that subsequent to this work, a Raman study even appeared in which it was stated that the experimental spectra obtained were consistent with the structure of P42 [14]). The crystal presents no evidence of orientational disorder, although the ammonium ion is slightly deformed with respect to ideal tetrahedral geometry. The ammonium ions as well as the nitrate ions are positioned forming CsCl-like lattices; i.e., each type of ion is surrounded by eight ions of the opposite type. This means that each ammonium is attached to four of its neighboring nitrate groups by strong hydrogen bonds in a

4

Santiago García-Granda and Jose M. Montejo-Bernardo

tetrahedral arrangement, resulting in a three-dimensional hydrogen bond network, with the shortest hydrogen bonds among the AN phases. The first models of the crystal structure of phase IV appeared in two papers published almost simultaneously and independently, the first of which used X-ray powder diffraction data [15] and the other, X-ray single crystal diffraction data [1]. In both papers, the unit-cell parameters were correctly determined (at room temperature), the space group was assigned and the internal crystal structure solved, specifying the relative positions of the ammonium cations and the nitrate anions within the cell, with no structural disorder. Ammonium nitrate phase IV crystallizes in the orthorhombic system, with two units of the salt in the unit cell (Table 1). Neutron single crystal diffraction techniques [16] were used to determine the role of the hydrogen atoms of the ammonium groups in the final crystal. Subsequent studies using X-ray single crystal diffraction [2] and neutron powder diffraction techniques [17] obtained more precise values of the unit-cell parameters and atomic positions. In this crystalline phase, each ammonium group is surrounded by twelve oxygen atoms from eight different nitrate groups, forming hydrogen bonds with four of these oxygen atoms [2], giving infinite hydrogen-bonded chains parallel to the a axis, taking the form N--H...O...H--N. These chains are in turn bound to one another by weak hydrogen bonds to form layers parallel to the (001) face [16]. A recent study [18] based on Terahertz spectroscopy (THz) and solid-state Density Functional Theory (DFT) concludes that the space group Pmmn is not a complete representation of the crystal structure of this phase, and that the correct space group is Pmn21. The authors state that the Pmmn space group represents only a vibrational averaging of the nitrate rotations, and that the Pmmn structure is actually based on a lower-symmetry (and lower-energy) Pmn21 crystal structure. On the other hand, the calculated theoretical models only can explain the experimental THz spectrum considering the Pmn21 space group and not the Pmmn space group. As can be seen in Table 1, the calculated crystal structure associated with the new space group implies small changes in the unit-cell parameters. Interestingly, an article published nearly 40 years ago [19] already pointed out the incompatibility of the Pmmn group (centrosymmetric) with experimental Raman results and proposed P21mn (non-centrosymmetric) as the space group. However, this paper received little attention and the centrosymmetric option remained as the correct space group in all following papers (in fact, it is not even referenced in the aforementioned THz-DFT study [18]).

Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 5

Figure 1.Temperature phase transitions for NH4NO3.The term ordered/disordered refers to the existence of several orientations in the crystal of the ions of the salt (see text for details).

The first reported model of the crystal structure of phase III was obtained from X-ray single crystal diffraction measurements [1] at a temperature of around 35ºC. Although the paper did not present particularly accurate values of the unit cell parameters, it correctly specified the crystal system and the space group and even proposed the arrangement of the ions in the crystal, indicating that it was very similar to that exhibited by aragonite or potassium nitrate. No mention was made of the existence of any kind of disorder in the crystal. The study showed some discrepancies between the observed and calculated intensities of some reflections (especially in those of “moderate” intensity). Several years later, a new analysis of the structure (at a temperature of about 42ºC) was published, also based on X-ray single crystal diffraction data [20]. This paper corroborated the unit-cell and space group data of the previous study, though significantly modifying the positions of the ions within the unit cell, especially the nitrate groups, thus achieving a better agreement between observed and calculated data. The paper makes no reference to the existence of disorder in the crystal, either. Phase III crystallizes in the orthorhombic system with four units of salt contained in the unit cell (Table 1). Note that, due to the way of choosing the axes, the space group was assigned in both studies as Pbnm instead of Pnma, which is the standard according to the International Tables for X-Ray Crystallography [21] and which is the one that appears in subsequent papers. Starting from the model proposed by Goodwin & Whetstone [20], a later study using neutron powder diffraction data [22] re-refined the structure, determining that the ammonium ions actually showed orientational disorder about their N atom positions, between two different orientations that were considered of equal weight. In the unit cell, the coordination sphere for every ammonium group consists of eleven oxygen atoms from seven nitrate anions [2], while studies based on neutron powder diffraction data report the existence of hydrogen-bonded chains parallel to the b axis [22, 23].

6

Santiago García-Granda and Jose M. Montejo-Bernardo

In a paper published in 1932 [1], Hendricks et al. accurately determined the unit-cell parameters of phase II (for a temperature of 100 ± 10 ºC) from Xray powder diffraction data. Making certain assumptions about the distances that must exist between the ions in the crystal consistent with obtained experimental reflections, these authors put forward a model of the crystal structure, but were unable to determine the space group. Phase II crystallizes in the tetragonal system, with two units of ammonium nitrate in the unit cell (Table 1). In a subsequent study analyzing the metastable transition IV↔II (see Section 2), Professor Shinnaka [24] postulated that the ammonium and nitrate ions are arranged forming CsCl lattices (each nitrate ion has eight of the nearest ammonium ions around it), with nitrate groups rotating in two orientations in opposite directions. This phenomenon could explain the experimental reflections as those corresponding to the P-421m space group. These possible rotations were later corroborated in an X-ray diffuse scattering study [25]. Two papers based on powder neutron diffraction finally clarified the issue of the space group and the disorder of the structure. The first [17] upheld the aforementioned assignation of the space group and determined that phase IV is in fact a disordered structure, not only for the nitrate ions, but also for the ammonium ions, with two alternative equivalent positions for each of the two types of ion. The second paper [23] went a step further, maintaining that if the occupancy factors of these two alternative sites are equal, this will imply the presence of certain planes of symmetry and the P-421m space group becomes P4/mbm. Seeing as the latter case involves very specific occupation factors (50%-50%), the space group is usually labeled as P-421m in papers, as this assignation is less restrictive. The latter also presents the existing hydrogen bonds in the crystal. The crystal structure of phase I was also analyzed by Hendricks et al.[1], using of X-ray powder diffraction measurements at a temperature of 155ºC. These authors determined that this phase has a CsCl-type cubic cell (although the lattice parameter is not too accurate) with a single molecule of the salt in the unit cell (Table 1). They were unable to establish the space group, however. The ammonium groups are located at the (0,0,0) position, while the nitrate groups, located in the (½, ½, ½) position, rotate with constant angular velocity in three orthogonal planes (though the possibility that they do so with approximately spherical symmetry is also considered). Studies based on X-ray diffuse scattering subsequently aimed to model the disorder of the nitrate ions by means of an 8-orientation [26] and a 12-orientation [9, 26] model (for a crystal measured at 150ºC), which provided a better fit to

Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 7 experimental data. One of these papers [9] also reports the existence of disorder in the ammonium groups, based on data from calorimetric studies [7]. Soon after, a study based on Raman analysis also proposed that the NO3groups are randomly oriented among 12 equivalent sites [27]. A more recent study based on neutron powder diffraction data [28] provided support for the 12-orientation model for the nitrate ions (preserving its planar equilateral triangle geometry) and corroborated the rotational disorder of the ammonium groups, stating that this occurred in two symmetry-related tetrahedral orientations. To reach these conclusions, the authors had to resort to a direct multipole analysis of the nuclear smearing functions, as a simple Rietveld refinement did not allow them to distinguish between the two proposed orientation models for the nitrate groups. Although the study was based on neutron diffraction data, the authors did not study the hydrogen bonding network due to the prevailing high disorder in the crystal. The unit cell and X-ray powder diffraction profiles of AN phases are shown in Figure 2. Different images representing the crystal structure of phases V [3, 13, 18, 29, 30], IV [16, 18, 29-32], III [22, 23, 30], II [17, 24, 29, 30, 32] and I [9, 29, 32] can be found in the literature. Table 1. Experimental crystallographic parameters of the five AN phases existing at atmospheric pressure

Space Group a (Å) b (Å) c (Å) V (Å3) T (ºC) Z

I [28]

II [23]&

III [22]

IV [17]

IV [18]*

V [3]

Pm3m

P4/mbn

Pnma

Pmmn

Pmn21

Pccn

4.3655(2) 4.3655(2) 4.3655(2) 83.2(1) >125% 1

5.6956(3) 5.6956(3) 4.9197(5) 159.6(1) 65# 2

7.7184(3) 5.8447(1) 7.1624(2) 323.1(1) 45 4

5.7574(1) 5.4394(1) 4.9298(1) 154.4(1) 25 2

5.2673 4.9246 5.8822 152.6 21 2

7.8850(2) 7.9202(2) 9.7953(2) 611.7(1) -195 8

* From a study using THz and DFT (see details in the text). # The crystal was obtained by direct transition of phases II to IV (see Figure 1), at that temperature, in the absence of humidity conditions (see details in Section 2). & Authors usually assign the P-421m space group, due to being less restrictive (see text for details). % The exact temperature is not reported in the paper.

8

Santiago García-Granda and Jose M. Montejo-Bernardo

Figure 2. Unit cell and X-ray powder diffraction profiles of AN phases II [17], III [22], IV [17] and V [3]. Pictures were taken using the Mercury program (a graphical software application from the Cambridge Crystallographic Data Centre).

In parallel to the work employing X-ray and neutron diffraction methods, the crystalline phases of the AN were studied also using other spectroscopic techniques mainly through Raman [12, 19, 33-36] and IR [11, 37-39] spectroscopy. Most of these studies (some of which have already been cited

Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 9 earlier in the text) focused mainly on determining the degree of disorder within the crystal, and on the different orientations of both types of ions of the salt in the unit cell in each of the crystalline phases [40-45]. Throughout the rest of the chapter, we shall see how these techniques have also been used on a regular basis to identify new AN phases at low temperatures and atmospheric pressure, at high pressure, or in the study of phase changes with the temperature at atmospheric pressure. Studies using NMR techniques [46-50] have also been conducted, though to a lesser extent. In addition to these clearly characterized five phases, several papers can be found in the literature providing evidence indicating the possible existence of another low-temperature phase at atmospheric pressure, phase VII, which would arise at very low temperatures. However, the evidence is not definitive and, in some cases, even conflicting. Each of the papers puts forward a different transition temperature from phase V--VII. Early studies based on calorimetric techniques [51-52] propose a phase change at a temperature of around 210 K [52], a result that was consistent with a subsequent infrared spectroscopy study [40]. Another study, however, based on DTA and X-ray diffraction measurements, places the appearance of the new phase at 100 K [53], while a paper based on a small anomaly in the heat capacity proposed that the phase change occurs at 156 K [7]. Later, a Raman study proposed a V-VII phase transition at around 100 K [54], although a subsequent study also based on Raman spectroscopy [55] refutes this result, reporting that no evidence was found of the existence of any low temperature phase at a temperature of 11K. Nevertheless, its existence is still reported in even later publications, without reference to any study, giving transition temperatures of 103 K [19, 56] or around 173 K [57]. To date, no possible structure has been reported for this new phase. In two studies based on powder neutron diffraction data at very low temperatures and using a Rietveld-type refinement (diffraction patterns were obtained at various temperatures from 5 K to 250 K) [3, 58], the results and conclusions were the same: the structures at all temperatures were successfully solved with the same space group, Pccn, with essentially identical crystal structures, corresponding to phase V. Interestingly, one of these studies on neutrons [58] puts forward an explanation from a previous Raman spectroscopy study on phase V at low temperatures [59] for the “phase change” found at 156 K [7]. The authors point out that it may actually be due to the existence of two effects: “the excitation of higher modes and the smearing action of increased thermal motion”, which can give rise to this anomalous behavior.

10

Santiago García-Granda and Jose M. Montejo-Bernardo

Figure 3. Simplified P-T phase diagram for AN from referenced data [32]. Dotted lines (in grey) show the original P-T phase diagram published by Rapoport and Pistorius [62], including phase VI proposed by Bridgman [61].

1.2. Studies of Phases at High Pressures Surprisingly, the possible phases of AN at pressures above atmospheric pressure (and room temperature) have received much less attention, despite the importance of knowing how this compound might behave at high pressures due to its explosive nature [60]. Even today, the existence and characteristics of new phases resulting from subjecting phase IV to high pressures remains a complex issue. Recent studies provide mixed results and seem to indicate that the possible occurrence of these phases depends primarily on the experimental conditions under which the study is conducted (an aspect that is directly associated with the metastable character of these phases). The first P-T phase diagram for AN was reported by Brigman almost a century ago, up to 1.2 GPa and 473 K [61]. This paper already proposed the emergence of a new phase, VI, at 0.9 GPa and 442 K. The area of study was later extended to 3.5 GPa and 623 K [62], employing differential thermal

Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 11 analysis results and proposing a triple point for the I↔VI, IV↔VI and I↔IV phase transition lines, around 1.9 GPa and 529 K (see Figure 3). In a paper employing IR spectroscopy to study shear-stress induced transitions [63], the variations in the IR spectra indicated the existence of a new phase (denoted as VIII) at 3.0 GPa. Furthermore, in a study based on changes in vibrational bands in Raman spectra data [64], shear-induced phase transformations at 0.45 and 2.7 GPa were also reported for phase IV. However, due to the similar results for the three phases, the authors report that the structures must all be closely related to phase IV. In an isothermal X-ray diffraction study at room temperature up to 7.7 GPa [65], no phase changes were observed in the AN sample. However, in this same study, shock loading experiments up to 20 GPa once again suggested the existence of a new high-pressure, high-density metastable phase at pressures below 3.5 GPa, with a major change in unit-cell volume. However, no evidence was found of such phase transition in a study published in 2009 [66], based on isotherm hydrostatic angle dispersive X-ray diffraction experiments up to 25 GPa. The authors also report in this study that, based on vibrational data obtained in their own laboratory and the X-ray structure analysis carried out in their study, any possible new phases must be closely related to phase IV, supporting the findings of an earlier study [64]. No phase transition was observed either in another paper published in 2011 [31], based on Raman spectroscopy data (up to 35 GPa) and X-ray powder diffraction (up to 20 GPa). The authors indexed the cell with the lattice parameters and space group (Pmmn) corresponding to phase IV (Table 2), and state that the Rietveld refinement shows the shortening of certain distances in the hydrogen bond network, mainly along the b-axis, which could explain why this axis shows the greatest shortening. The same behavior was also reported for this axis in previous studies [66]. In the same study [31], working under non-hydrostatic condition, there is no indication of any phase change below 3.0-3.5 GPa, while at pressures below approximately 10 GPa, behavior is similar to that of the hydrostatic case (Table 2). For higher values, however, the results begin to differ, the powder diffraction data showing a substantial change on reaching 21 GPa. This new powder profile is also indexed with the Pmmn space group (isostructural to phase IV), but with different unit-cell parameters (Table 2). This new phase was labeled as IV'. This possible phase transition was also observed in the Raman spectra, with the appearance of a new vibrational band between 15 and 19 GPa. The authors explain the differences observed in the X-ray and Raman experiments as being due to the fact that the phase transition (X-ray data)

12

Santiago García-Granda and Jose M. Montejo-Bernardo

occurs at higher pressure that the change in the local symmetry (Raman data) of the crystal. The authors propose that the transformation from IV--IV' is primarily favored by the strengthening of certain hydrogen bonds due to the shortening of the distance between an oxygen atom from the nitrate groups and the nitrogen atom from the amino groups. The new phase is stable, at least up to 35 GPa, and reverts back to phase IV when the pressure drops below 0.1GPa, indicating that phase IV' is a metastable phase. Note that the authors themselves acknowledge that they were unable to obtain phase IV' a second time on repeating the experiments. They consider this fact yet another argument in favor of the metastable nature of phase IV', indicating the presence of moisture in some of the samples, or the different solid form of the powder used in each experiment as possible causes. A paper [32] has recently appeared (two of whose authors also participated in the study in which the metastable phase IV' was detected [31]) that includes a new P-T phase diagram for AN based on the aforementioned studies and on new experimental data presented in this paper (Figure 3) aimed at resolving the discrepancies of previous studies. Using synchrotron X-ray powder diffraction and Raman spectroscopy measurements, the authors first show that phase IV is stable at room temperature up to 45 GPa and presents no phase transition from IV-IV'(with X-ray experimental data, the Raman data only go up to 27.4 GPa, but show the same results). The plot of the variation in unit-cell volume versus pressure is almost identical to that reported for experiments under non-hydrostatic conditions in the paper published in 2011 in which no phase IV' was observed [31]. Something similar occurs with the variation in the b and c unit-cell parameters. At pressures above 20 GPa, both axes converge in length and remain very close to the maximum applied pressure, which is precisely what occurred with the unit-cell parameters of phase IV' in the previous study [31] (Table 2). In addition, the collapse of the b-axis, with the b/c ratio approaching unity at high pressures, had also been previously reported [66]. In this study, however, this behavior is not associated with a phase transition (appearance of phase IV'), but simply with an “anisotropic compression of the IV phase”. For the authors, this behavior may be justified by considering it analogous to the anisotropic thermal expansion observed in an earlier paper for phase IV [67] (this feature is detailed in Section 2). Apart from that, they also suggest an explanation for the observed phase transitions at 0.45 and 2.7 GPa reported in a previous paper (discontinuities in Raman shift) [64], based on their own Raman results. Essentially, the explanation is that in molecular crystal analysis, nonlinear

Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 13 Raman shifts as a function of pressure are not unusual and can be interpreted as a discontinuity, particularly if multiple linear fits are used. In the new P-T phase diagram put forward (Figure 3), the existence of phase IV is confirmed up to 40 GPa and 467 K, new phase boundaries for phases IV, II and I are proposed, a larger stability region for phase II is detailed, though no evidence for the presence of phase VI [61] is included. Finally, in a quantum mechanical study of AN crystalline phases based on density functional theory (DFT) and the pseudo potential method [30], theoretical results were included on the hydrostatic compression of phase IV up to 600 GPa. According to this study, phase IV continues to exist up to 75 GPa at room temperature, while the anisotropic compressibility of the unit-cell axes indicated in previous papers occurs only up to 60 GPa. Above this pressure, compressibility along the three axes becomes almost equal. Calculations indicate that a slightly distorted monoclinic crystal phase is possible at pressures above 75 GPa (Table 2), with a P21/m space group and β ≈ 88º, and with an unit-cell volume very close to that calculated for the orthorhombic Pmmn symmetry. Based on these results, the authors propose that it may be possible for the orthorhombic phase IV to change to a monoclinic crystal system at higher temperatures and pressures. Table 2. Unit-cell parameters for phase-IV AN at high pressures reported in the literature

Atmospheric [17] Hydrostatic [66] Hydrostatic [31] Non-hydrostatic[31] Non-hydrostatic [31] Non-hydrostatic [31]CIF Non-hydrostatic [32] Non-hydrostatic [32] Hydrostatic [30]

Pressure

a (Å)

b (Å)

c (Å)

0 GPa 0.4 GPa 5 GPa 7 GPa 21 GPa 28 GPa 20 GPa 30 GPa 100 GPa

5.7574(1) 5.748 5.553(2) 5.507(2) 5.145(2) 4.94(5) 5.27 5.17 4.72

5.4394(1) 5.428 4.932(2) 4.885(2) 4.508(5) 4.32(7) 4.60 4.51 3.92

4.9298(1) 4.937 4.730(1) 4.724(3) 4.502(2) 4.31(8) 4.58 4.51 4.11

Space group Pmmn Pmmn Pmmn Pmmn Pmmn Pmmn Pmmn Pmmn P21/m

Phase IV IV IV IV IV' IV' IV IV ¿?

The values at 20 GPa and 30 GPa were determined from Figure 5 included in the paper [32]. The values at 100 GPa are from theoretical calculations and were determined from Figure 7 included in the paper [30]. All values are at 298 K.

2. PHASE TRANSITIONS AND THERMAL BEHAVIOR The phase transition temperatures indicated in Figure 1 could be defined as the figures for “ideal conditions”, since, as we shall see below, their values

14

Santiago García-Granda and Jose M. Montejo-Bernardo

are not always reproducible, not only for different samples, but even for the same sample. They depend on several factors, such as the water content of the sample, the presence of impurities, and even the thermal history of the solid [57, 68, 69]. Moreover, considering the transition temperatures that appear in various papers, a (rarely commented) factor that must be taken into account is whether deuterated salt has been used to carry out the studies, since deuteration produced a slight shift in the phase change temperatures as follows [13]: V--IV (=), IV--III (3.2 K), III--II (3.6 K), II--IV (-2.4 K), II--I (2.7 K). Under certain conditions, however, the order of the phase changes shown in Figure 1 also remains, and direct transitions between phases without passing through the intermediate stages are given. Even the process of decomposition of the molten phase also depends on several aspects related to the experimental conditions under which it occurs. As regards the successive phase changes, we shall look at three aspects: 1) Cell volume. On the one hand, its variation with temperature is different for each of the phases, while on the other hand, sudden changes in volume are given in each of phase changes; i.e., dimensional instability, giving rise to caking and breakage of prill effects yet to be generated. The specific volume of AN data as a function of temperature were measured in several studies employing powder diffraction techniques (X-ray and neutron) [1, 3, 67, 70-72], affording thermal expansion parameters whose graphical representations may be found in the literature (Figure 4) [3]. 2) Ion distributions. In all published studies, the successive phase changes are primarily related to changes in the ordered-disordered rotational motion of the nitrate ions [1, 29, 32, 36], and with the consequent variation of the hydrogen bond network inside the crystal. 3) Thermal behavior. Transitions between ammonium nitrate phases have also been also analyzed using thermal techniques such as Differential Scanning Calorimetry (DSC) [57, 73,74, 84, 85], DTA [56], and heat capacities [6, 7] in order to know accurately phase transition temperatures (Tt), the temperature range stability of each phase, and the enthalpy of each phase change (Figure 5). However, a problem arises on account of the sensitivity of the Tt with respect to the previously mentioned variables (moisture, presence of impurities, etc.), and especially to the thermal history of the sample. Thermal cycling (heating/cooling cycles) results in the presence of thermal

Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 15 hysteresis, which could be increased with the number of heating/cooling cycles (Figure 6).

Figure 4. Molecular volume variation with temperature. (Data from references [3] and [70]).

Figure 5. DSC for an AN sample with a 2% water content. Phase transition temperatures and enthalpies are detailed.

2.1. Phase V

16

Santiago García-Granda and Jose M. Montejo-Bernardo

This is the stable phase below 255-257 K (the value of Tt depends on the source consulted), and evolves to phase IV at this temperature (Figure 1). The variation in volume with temperature was measured in two papers published at same time [3, 58], giving very similar results, though with some small differences in the values of the unit-cell parameters. The authors of both studies argue that the differences are due to errors in the neutron wavelength and the different measured temperatures. The variation in volume is shown in Figure 4. The phase transition at Tt implies a volumetric change of -3.05%, because, although the molecular packing of phase V is similar to that of phase IV, its specific cell volume is significantly higher. This variation is related to the different orientation of the NO3- ions; in phase IV they occupy less space because they are stacked with their planes parallel [3] (see Figure 2), also giving rise to a different hydrogen bond network. A second phase change, the direct V--II phase transition has also been reported from heat capacities studies [7] and microscopic observations [8]. This transformation was first suggested a few years earlier [10] and, according to the researchers, this phenomenon only occurs if the sample is treated with a minute amount of surface active agent (octadecylamine acetate in this case). Really This transition actually occurs in two steps via a metastable phase V*via the scheme shown below, in a second-order transition [8], and that may be the result of a phenomenon of orientational order-disorder of the nitrate and ammonium ions [7]: V ─── 315 K ─── V* ─── 318 K ─── II

The structure of phase V* was not determined. However, on the basis of the measured X-ray powder diffraction pattern, the authors claim that its structure is only slightly different from phases V and II.

2.2. Phase IV The thermal stability range and phase change temperatures of this AN phase differ depending essentially on two factors: the presence or not of water in the sample, and its thermal history. Under dry conditions (absence of water), phase IV evolves directly to disordered phase II, whereas in the presence of a small amount of moisture, phase IV changes to disordered phase III (Figures 5 and 6). On the other hand, its thermal history gives rise to the

Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 17 presence of thermal hysteresis, and even to the coexistence of both phase transitions (Figure 6). The variation in volume with temperature for phase IV has been widely studied, (even doped with a small amount of NiO [23]), and experimental data show that the a-axis decreases slightly with temperature [3, 67], even though the unit-cell volume increases. The variation in volume is shown in Figure 4 (dashed line corresponds to extrapolated data for the IV--II phase change). The IV--II transition implies a volumetric change of only 1.69%, while the variation in volume in the IV--III transition is 3.84%. It is known that this phase change causes breakage and caking of the AN prills during storage, and details regarding the mechanism of caking have been put forward [75]. To avoid caking, several phase stabilizing agents have been used to obtain phase stabilized ammonium nitrate (PSAN) [68, 69, 76-80], eliminating the IV--III transition and directly leading to the IV--II transition.

Figure 6. DSC from AN sample recorded using heating-cooling cycles. a) Transitions for phase IV. b) Effect of several heating-cooling cycles on thermal hysteresis. (Data from reference [57]).

18

Santiago García-Granda and Jose M. Montejo-Bernardo

The change of phase from IV--II already appears in earlier papers [1, 4], which assigned one Tt at approximately 50 ºC, while recognizing that the nature and exact conditions under which the changes occur are unclear. The results of this research, along with that of subsequent studies analyzing the influence of the maximum temperature reached, time of heat treatment, and water content [24, 25, 57, 81, 82], has provided different information about this phase change: 



 

When the single crystal of phase IV was formed directly from phase II, or when it was obtained from an aqueous solution at room temperature by heating the sample, the IV--II transition occurs between 53-56ºC. The phase IV--II change occurs directly in dry AN crystals (without occluded water) when heating the sample to 55 ºC, and also in deuterated AN, with the small shift in the Tt previously indicated for the deuterated salt [17]. If a sample is heated up to phase I, the direct II--IV transition also occurs around 50 ºC on cooling. The FTIR study [45] proposed a temperature of 52 °C for the IV--II transition and 48 °C for II--IV.

This transition is of the ordered-disordered type and can be understood on account of the similar structure of both phases (CsCl-like structure, see Figure 2). Increasing the temperature causes an increase in the oscillation of the nitrate ions and a rotation in alternative planes of ±45º respectively relative to the orientation in phase IV. This leads to an expansion of the b-axis and a contraction of the a-axis, until the tetragonal unit cell of phase II is obtained. The new hydrogen bonding network, with weak hydrogen bonds, also allows the dynamic disorder of the ammonium ions [8, 29, 32]. The IV--III phase change, catalyzed by the presence of moisture, can occur in the 32-55 ºC temperature range depending on the moisture content and thermal history of the crystal [57, 68, 70]. The effect of the water content on the phase transition temperature was demonstrated long ago [70, 75, 76], and a recent study has shown how the temperature of this transition increases as the relative humidity (RH) decreases (while the temperature of the III--IV transition decreases when the RH decreases) [45]. On the other hand, the

Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 19 presence of this transition (and not the IV—II phase change) also depends on how phase IV was obtained. For example, if the sample is prepared by rapid cooling of melted AN, the IV--III transitions occurs between 35-51 ºC [82]. Currently, this phase change is not considered as a transition of the ordered-disordered type, but rather as a transition involving a complex process of dissolution and recrystallization [30, 45, 46, 83] (directly related to the presence of water in the sample). This assertion could be supported by the major difference in structure of both phases. A detailed explanation of the process (reorientation of the ions, geometrical relation between the unit-cell axes of both crystalline phases, and the role of water) can be found in the paper by Davey et al., [83]. The hydrogen bond network is significantly affected by the phase change [2]. The strong hydrogen bond network in layers parallel to the (001) face of phase IV is broken along one direction, converting the sheets into double chains [16, 23]. Moreover, the hydrogen bonds of phase III are weaker than in phase IV, allowing major thermal vibration in the ammonium ions and giving rise to the disordered character of this phase [30].

2.3. Phase III A priori, as shown in Figure 1 (and, as previously mentioned, in the presence of small amounts of moisture), phase III is observed from 305K, and does not change to phase II at 357K. However, some papers state that the III-II transition can occur at temperatures up to 362 K, while the reverse transition, II–III, can occur in an even wider range, between 357 and 321 K [1, 57]. Later we shall see that these results are related to the thermal hysteresis these two phases present (together with phase IV) under certain experimental conditions. The variation in volume with temperature [3] is shown in Figure 4. Due to the unusually large unit-cell volume of this phase, volumetric change at the transition temperature (considered as 357 K) implies a decrease of 1.59%. On the other hand, the phase III--II transition has a significant feature worth noting: it implies to change to a higher symmetry crystal system. However, this change is also accompanied by higher disordering, because the nitrate ions now show orientational disorder in phase II. Hence, the hydrogen bond network is modified.

20

Santiago García-Granda and Jose M. Montejo-Bernardo

2.4. Thermal Hysteresis “Thermal hysteresis” refers to the fact that the experimental temperature of the Tt transition between two phases is different in the heating process (A— B transition) than in the cooling process (B—A transition) when a sample is subjected to heating-cooling cycles; i.e., this temperature is affected by the thermal history of the sample. It is clear that this phenomenon affects the transitions between ammonium nitrate phases IV, III and II, since, as we have seen in the preceding paragraphs, many of the cited papers reported different Tt values for a phase change and its inverse. In addition, in the case of ammonium nitrate, heating-cooling cycles can cause phase changes to appear in the heating cycle which are different to those that occur in the cooling cycle. Furthermore, the intensity of the thermal hysteresis effect could increase with the number of heating-cooling cycles. An extensive study of these issues was published by Ingman et al., [57]. The main results of this study are shown in Figure 6 and will be discussed below. As the authors point out in the paper, the variation in Tt may be determined by phase metastability, which is a widely recognized phenomenon for phases IV, III and II of ammonium nitrate, as became clear when discussing the evolution of phase IV. Figure 6a shows the experimental result of one heating-cooling cycle. On heating, only the endotherm to the IV--II transition is observed (at a temperature of 53 ºC). On cooling, however, two exotherms appear. The first, at 57 ºC, corresponds to the II--III transition, and the second, at only 10 ºC, is due to the III--IV transition. Figure 6b shows the effect of several heating-cooling cycles (over 20100-0ºC) for a sample of AN. Several interesting effects may be highlighted: 1) The presence of three endotherms in the heating cycle, including both possible phase changes for phase IV. The first endotherm (≈ 40 ºC) corresponds to the IV--III transition, the second (≈ 55 ºC) to the IV-II transitions, and the third (≈ 87 ºC) to the III--II transition. 2) The same three transitions (three exotherms), appear in the cooling cycle, though at significantly lower temperatures. The phase II--III transition occurs at ≈ 60 ºC, the II--IV transition at ≈ 45 ºC, and the III--IV transition at ≈ 10 ºC. 3) As the number of heating-cooling cycles increases, the temperature of some of the transitions changes, following a clear trend. In the heating cycles, the phase IV--III and IV--II transitions are closer and shift to lower temperatures, whilst the phase III--II transition remains

Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 21 unchanged. In the cooling cycles, however, the phase II--IV transition remains unchanged, whilst the III--IV and II--III transitions shift to lower temperatures.

2.5. Phase II We have already seen that by submitting phase II to a cooling cycle, the temperatures of the Tt transition to stable lower temperature phases, are highly variable. However, we have not found any paper reporting that the same occurs in heating cycles, and the phase II—I transition can take place at different temperatures, taking 125 °C as a reference temperature (Figure 1). In fact, the precursor work on the study of the phases crystal of the AN [1] refers to heating and cooling around 125 °C, and contains no mention of any variation in Tt in the phase II--I and I--II transitions. Experimentally, however, the phase II--I transition is commonly observed to occur at slightly higher temperatures, up to 135 °C (see Figure 5). This fact is probably related to the presence of impurities in the sample or due to the precision of the calorimetric measurements. The variation in volume with temperature [3] is shown in Figure 4. The phase II--I transition is mainly driven by the temperature-induced distortion on the NO3- ions [32], and because the 12orientational model for phase I and tetragonal phase II are closely related [9]. The phase II--I transition implies a change to a higher symmetry crystal system, although this change also corresponds with greater disorder, given that phase I is characterized by free reorientations of the NO3- and NH4+ ions [32]. Details regarding the mechanism via which the phase change occurs have been recently published in a study based on molecular dynamics simulation [29].

2.6. Phase I Models of the variation of the volume of this phase with temperature are collected in the paper by Hendricks et al., [1], based on data from previous studies by other researchers. However, more recent studies with more precise data have not been found. This phase melts at 170 ºC (or 169 ºC in some papers), and the melt is a transparent colorless liquid. In a recent theoretical study [29], a melting temperature of 172±10 ºC was calculated, in good agreement with experimental values.

22

Santiago García-Granda and Jose M. Montejo-Bernardo

ACKNOWLEDGMENTS Financial support was provided by the Ministerio de Economía y Competitividad de España (MAT2010-15094, Factoría de Cristalización – Consolider Ingenio 2010) and ERDF.

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22]

Hendricks, SB; Posnjak, E; Kracek, FC. J. Am. Chem. Soc., 1932, 54, 2766. Holden, JR; Dickinson, Ch. W. J. Phys. Chem., 1975, 79, 249. Choi, CS; Prask, H. J. Acta Cryst. 1983, B39, 414. Bowen, NL. J. Phys. Chem., 1925, 30, 721. Parsonage, NG; Staveley, LAK. Disorder in Crystals. Oxford University Press, Oxford, 1978. Stephenson, CC; Bentz, DR; Stevenson, DA. J. Am. Chem. Soc., 1955, 77, 2161. Nagatani, M; Seiyama, T; Sakiyama, Suga, H; Seki, S. Bull. Chem. Soc. Jpn., 1967, 40, 1833. Seiyama, T; Yamazoe, N. J. Crystal Growth., 1968, 2, 255. Yamamoto, S; Shinnaka, Y. J. Phys. Soc. Japan., 1974, 37, 724. Amorós, JL; Arrese, F; Canut, M. Z. Kristallogr., 1962, 117, 92. Oxton, IA; Knop, V; Falk, M. J. Phys. Chem., 1976, 80, 1212. Kearley, GJ; Kettle, SFA; Oxton, IA. Spectrochim. Acta A., 1980, 36, 419. Ahtee, M; Smolander, KJ; Lucas, BW; Hewat, AW. Acta Cryst., 1983, C39, 651. Carrick, MT; James, DW. Austr. J. Chem., 1986, 39, 771. West, CD. J. Am. Chem. Soc. 1932, 54, 2256. Choi, CS; Mapes, JE; Prince, E. Acta Cryst., 1972, B28, 1357. Lucas, BW; Ahtee, M; Hewat, AW. Acta Cryst., 1979, B35, 1038. Witko, EM; Buchanan, WD; Korter, TM. J. Phys. Chem. A., 2011, 115, 12410. James, DW; Carrick, MT. Chem. Phys. Lett., 1974, 28, 117 Goodwin, TH; Whetstone, J. J. Chem. Soc., 1947, 1455. Hahn, Th; Aroyo, MI. (Eds.) International Tables for Crystallography. Volume A: Space-group symmetry, Wiley. UK, 2006. Lucas, BW; Ahtee, M; Hewat, AW. Acta Cryst, 1980, B36, 2005.

Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 23 [23] [24] [25] [26] [27] [28] [29] [30] [31]

[32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53]

Choi, CS; Prask, HJ; Prince, E. J. Appl. Cryst., 1980, 13, 403. Shinnaka, Y. J. Phys. Soc. Japan., 1956, 11, 393. Shinnaka, Y. J. Phys. Soc. Japan., 1959, 14, 1707. Shinnaka, Y. J. Phys. Soc. Japan., 1959, 14, 1073. Tang, HC; Torrie, BH. J. Phys. Chem. Sol., 1977, 38, 125. Ahtee, M; Kurki-Suonio, K; Lucas, BW; Hewat, AW. Acta Cryst., 1979, A35, 591. Velardez, GF; Alavi, S; Thompson, DL. J. Chem. Phys., 2004, 120, 9151. Sorescu, DC; Thompson, DL. J. Phys. Chem. A., 2001, 105, 720. Dunuwille, M; Davidson, AJ; Chellappa, RS; Dattelbaum, DM; Yoo, Ch-Sh. J. Phys. Chem. A., 2011, 115, 11889; J. Phys. Chem. A., 2012, 116, 7600 (corrigenda). Chellappa RS; Dattelbaum, DM.; Velisavljevic, N.; Sheffield, S. J. Chem. Phys., 2012, 137, 064504. Couture, L; Mathieu, JP. J. Phys. Radium., 1949, 10, 145. Penot, D; Poulet, H; Mathieu, JP. Compt. Rend., 1956, 243, 1303. Kearley, GJ; Kettle, SFA; Ingman, JS. J. Chem. Phys., 1980, 73, 2129. Akiyama, K; Morioka, Y; Nakagawa, I. Bull. Chem. Soc. Jpn., 1981, 54, 1662. Keller, WE; Halford, RS. J. Chem. Phys., 1949, 17, 26. Newman, R; Halford, RS. J. Chem. Phys., 1950, 18, 1276. Newman, R; Halford, R. S. J. Chem. Phys., 1950, 18, 1291. Theoret, A; Sandorfy, C. Can. J. Chem., 1964, 42, 57. Osterlund, K; Rosen, HJ. Solid State Commun., 1974, 15, 1355. Tang, HC; Torrie, BH. Proc. Int. Conf. Light Scattering Solids, 1976, 912. Kearley, GJ; Kettle, SFA. J. Mol. Struct., 1982, 79, 319. Dalterio, RA; Owens, FJ. J. Phys. Chem. Solids, 1988, 49, 1303. Wu, HB; Chan, MN; Chan, ChK. Aerosol Sci. Technol., 2007, 41, 581. Brown, RN; Mclaren, AC. Proc. R. Soc., 1962, 266, 329. Hovi, V; Järvinen, U; Pyykkö, P. J. Phys. Soc. Jpn., 1966, 21, 2742. Niemelä, L; Lohikainen, T. Eur. Phys. J., 1967, B6, 376. Riggin, MT; Knispel, RR; Pintar, MM. J. Chem. Phys., 1972, 56, 2911. Anderson-Altmann, KL; Grant, DM. J. Phys. Chem., 1993, 97, 11096. Jaffrey, J. Compt, Rend. 1947, 224, 1346. Crenshaw, JL; Ritter, I. Z. Physik. Chem. 1932, B16, 143. Volfkovich, SI; Rubinchik, SM; Kozhin, VM. Bull. Acad. Sci. USSR, Div. Chem. Sci (Engl. Transl.)., 1954, 167.

24

Santiago García-Granda and Jose M. Montejo-Bernardo

[54] Iqbal, Z. Chem. Phys. Lett., 1976, 40, 41. [55] Tang, HC; Torrie, BH. J. Phys. Chem. Solids., 1978, 39, 845. [56] Sowell, RR; Karnowsky, MM; Walters, LC. J. Therm. Anal., 1971, 3, 119. [57] Ingman, JS; Kearley, GJ; Kettle, SFA. J. Chem. Soc., Faraday Trans., 1. 1982, 78, 1817. [58] Ahtee, M; Smolander, KJ; Lucas, BW; Hewat, AW. Acta Cryst., 1983, B39, 685. [59] Jayasooriya, UA; Kearley, GJ; Kettle, SFA; Lauter, H. J. Phys. Status Solidi A, 1981, 63, 169. [60] Paterson, S; Davidson, JM. Nature, 1962, 195, 277. [61] Bridgman, PW. Proc. Am. Acad. Arts. Sci., 1916, 51, 581. [62] Rapoport, E; Pistorius, CWFT. J. Chem. Phys., 1966, 44, 1514. [63] Adams, DM; Sharma, SK. J. Chem. Soc., Faraday Trans., 2. 1976, 72, 2069. [64] Adams, DM; Sharma, SK. J. Chem. Soc. Faraday Trans., 2. 1981, 77, 1263. [65] Sandstrom, FW; Persson, PA; Olinger, B. AIP Conf. Proc., 1994, 309, 1409. [66] Robbins, DL; Sheffield, SA; Dattelbaum, DM; Velisavljevic, N; Stahl, DB. AIP Conf. Proc., 2009, 1195, 552. [67] Herrmann, M; Engel, W. Mat. Sci. Forum, 1996, 228-231, 359. [68] Kim, JH. J. Chem. Eng. Jpn., 1997, 30, 336. [69] Simoes, PN; Pedroso, LM; Portugal, AA; Campos, JL. Thermochim. Acta, 1998, 319, 55. [70] Griffith, EJ. J. Chem. Eng. Data, 1963, 8, 22. [71] Herrmann, M; Engel, W; Schneider, J; Goebel, H. Mat. Sci. Forum, 1994, 166, 489. [72] Herrmann, MJ; Engel, W. Prop. Expl. Pyrotech., 1997, 22, 143. [73] McDie, HG. Anal. Chem., 1963, 35, 1840. [74] Dellien, I. Thermochim. Acta, 1982, 55, 181. [75] Sjölin, Ch. J. Agr. Food Chem., 1972, 20, 895. [76] Sjölin, Ch. J. Agr. Food Chem., 1971, 19, 83. [77] Engle, W. Explosivstoffe, 1973, 1, 9. [78] Cady, HH. Prop. Expl. Pyrotech., 1981, 6, 49. [79] Owens, FJ. J. Appl. Phys., 1982, 53, 368. [80] Mathew, S; Krishnan, K; Ninan, KN. Prop. Expl. Pyrotech., 1998, 23, 150. [81] Tanaka, K; Fukuyama, I. J. Phys. Soc. Japan, 1953, 8, 428.

Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 25 [82] Kestila, E; Valkonen, J. Thermochim. Acta, 1993, 214, 305. [83] Davey, RJ; Ruddick, AJ; Mitchell, B; Maginn, SJ; Polywka, LA. J. Phys., D, 1991, 24, 176. [84] Oxley, JC; Kaushik, SM; Gilson, NS. Thermochim. Acta, 1989, 153, 269. [85] Oxley, JC; Smith, JL; Rogers, E; Yu, M. Thermochim. Acta, 2002, 384, 23.