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 NO3-
anions. 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 X-
ray 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 NO3-
groups 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

I [28] II [23]& III [22] IV [17] IV [18]* V [3]

Space
Pm3m P4/mbn Pnma Pmmn Pmn21 Pccn
Group
a (Å) 4.3655(2) 5.6956(3) 7.7184(3) 5.7574(1) 5.2673 7.8850(2)
b (Å) 4.3655(2) 5.6956(3) 5.8447(1) 5.4394(1) 4.9246 7.9202(2)
c (Å) 4.3655(2) 4.9197(5) 7.1624(2) 4.9298(1) 5.8822 9.7953(2)
V (Å3) 83.2(1) 159.6(1) 323.1(1) 154.4(1) 152.6 611.7(1)
T (ºC) >125% 65# 45 25 21 -195
Z 1 2 4 2 2 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

Space
Pressure a (Å) b (Å) c (Å) Phase
group
Atmospheric [17] 0 GPa 5.7574(1) 5.4394(1) 4.9298(1) Pmmn IV
Hydrostatic [66] 0.4 GPa 5.748 5.428 4.937 Pmmn IV
Hydrostatic [31] 5 GPa 5.553(2) 4.932(2) 4.730(1) Pmmn IV
Non-hydrostatic[31] 7 GPa 5.507(2) 4.885(2) 4.724(3) Pmmn IV
Non-hydrostatic [31] 21 GPa 5.145(2) 4.508(5) 4.502(2) Pmmn IV'
Non-hydrostatic [31]CIF 28 GPa 4.94(5) 4.32(7) 4.31(8) Pmmn IV'
Non-hydrostatic [32] 20 GPa 5.27 4.60 4.58 Pmmn IV
Non-hydrostatic [32] 30 GPa 5.17 4.51 4.51 Pmmn IV
Hydrostatic [30] 100 GPa 4.72 3.92 4.11 P21/m ¿?
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 20-
100-0ºC) 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 12-
orientational 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] Hendricks, SB; Posnjak, E; Kracek, FC. J. Am. Chem. Soc., 1932, 54,
2766.
[2] Holden, JR; Dickinson, Ch. W. J. Phys. Chem., 1975, 79, 249.
[3] Choi, CS; Prask, H. J. Acta Cryst. 1983, B39, 414.
[4] Bowen, NL. J. Phys. Chem., 1925, 30, 721.
[5] Parsonage, NG; Staveley, LAK. Disorder in Crystals. Oxford University
Press, Oxford, 1978.
[6] Stephenson, CC; Bentz, DR; Stevenson, DA. J. Am. Chem. Soc., 1955,
77, 2161.
[7] Nagatani, M; Seiyama, T; Sakiyama, Suga, H; Seki, S. Bull. Chem. Soc.
Jpn., 1967, 40, 1833.
[8] Seiyama, T; Yamazoe, N. J. Crystal Growth., 1968, 2, 255.
[9] Yamamoto, S; Shinnaka, Y. J. Phys. Soc. Japan., 1974, 37, 724.
[10] Amorós, JL; Arrese, F; Canut, M. Z. Kristallogr., 1962, 117, 92.
[11] Oxton, IA; Knop, V; Falk, M. J. Phys. Chem., 1976, 80, 1212.
[12] Kearley, GJ; Kettle, SFA; Oxton, IA. Spectrochim. Acta A., 1980, 36,
419.
[13] Ahtee, M; Smolander, KJ; Lucas, BW; Hewat, AW. Acta Cryst., 1983,
C39, 651.
[14] Carrick, MT; James, DW. Austr. J. Chem., 1986, 39, 771.
[15] West, CD. J. Am. Chem. Soc. 1932, 54, 2256.
[16] Choi, CS; Mapes, JE; Prince, E. Acta Cryst., 1972, B28, 1357.
[17] Lucas, BW; Ahtee, M; Hewat, AW. Acta Cryst., 1979, B35, 1038.
[18] Witko, EM; Buchanan, WD; Korter, TM. J. Phys. Chem. A., 2011, 115,
12410.
[19] James, DW; Carrick, MT. Chem. Phys. Lett., 1974, 28, 117
[20] Goodwin, TH; Whetstone, J. J. Chem. Soc., 1947, 1455.
[21] Hahn, Th; Aroyo, MI. (Eds.) International Tables for Crystallography.
Volume A: Space-group symmetry, Wiley. UK, 2006.
[22] Lucas, BW; Ahtee, M; Hewat, AW. Acta Cryst, 1980, B36, 2005.
 Ammonium Nitrate: Crystal Structure, Polymorphism and Thermal … 23

[23] Choi, CS; Prask, HJ; Prince, E. J. Appl. Cryst., 1980, 13, 403.
[24] Shinnaka, Y. J. Phys. Soc. Japan., 1956, 11, 393.
[25] Shinnaka, Y. J. Phys. Soc. Japan., 1959, 14, 1707.
[26] Shinnaka, Y. J. Phys. Soc. Japan., 1959, 14, 1073.
[27] Tang, HC; Torrie, BH. J. Phys. Chem. Sol., 1977, 38, 125.
[28] Ahtee, M; Kurki-Suonio, K; Lucas, BW; Hewat, AW. Acta Cryst., 1979,
A35, 591.
[29] Velardez, GF; Alavi, S; Thompson, DL. J. Chem. Phys., 2004, 120,
9151.
[30] Sorescu, DC; Thompson, DL. J. Phys. Chem. A., 2001, 105, 720.
[31] 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).
[32] Chellappa RS; Dattelbaum, DM.; Velisavljevic, N.; Sheffield, S. J.
Chem. Phys., 2012, 137, 064504.
[33] Couture, L; Mathieu, JP. J. Phys. Radium., 1949, 10, 145.
[34] Penot, D; Poulet, H; Mathieu, JP. Compt. Rend., 1956, 243, 1303.
[35] Kearley, GJ; Kettle, SFA; Ingman, JS. J. Chem. Phys., 1980, 73, 2129.
[36] Akiyama, K; Morioka, Y; Nakagawa, I. Bull. Chem. Soc. Jpn., 1981, 54,
1662.
[37] Keller, WE; Halford, RS. J. Chem. Phys., 1949, 17, 26.
[38] Newman, R; Halford, RS. J. Chem. Phys., 1950, 18, 1276.
[39] Newman, R; Halford, R. S. J. Chem. Phys., 1950, 18, 1291.
[40] Theoret, A; Sandorfy, C. Can. J. Chem., 1964, 42, 57.
[41] Osterlund, K; Rosen, HJ. Solid State Commun., 1974, 15, 1355.
[42] Tang, HC; Torrie, BH. Proc. Int. Conf. Light Scattering Solids, 1976,
912.
[43] Kearley, GJ; Kettle, SFA. J. Mol. Struct., 1982, 79, 319.
[44] Dalterio, RA; Owens, FJ. J. Phys. Chem. Solids, 1988, 49, 1303.
[45] Wu, HB; Chan, MN; Chan, ChK. Aerosol Sci. Technol., 2007, 41, 581.
[46] Brown, RN; Mclaren, AC. Proc. R. Soc., 1962, 266, 329.
[47] Hovi, V; Järvinen, U; Pyykkö, P. J. Phys. Soc. Jpn., 1966, 21, 2742.
[48] Niemelä, L; Lohikainen, T. Eur. Phys. J., 1967, B6, 376.
[49] Riggin, MT; Knispel, RR; Pintar, MM. J. Chem. Phys., 1972, 56, 2911.
[50] Anderson-Altmann, KL; Grant, DM. J. Phys. Chem., 1993, 97, 11096.
[51] Jaffrey, J. Compt, Rend. 1947, 224, 1346.
[52] Crenshaw, JL; Ritter, I. Z. Physik. Chem. 1932, B16, 143.
[53] 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.