Monotropic polymorphism in ester-based phase change materials from fatty acids and 1,4-butanediol

Article abstract of DOI:10.1021/cg400339z

This article reports on the synthesis, thermal behavior, and morphology of phase change materials based on diesters from 1,4-butanediol with palmitic (BD16), stearic (BD18), or behenic (BD22) acid. The crystalline contents and melting enthalpies of all substances exceed 90% and 180 J/g, respectively. The molecules display monotropic polymorphism. The metastable β′ form is built from outstretched molecules, whereas the denser, stable β form is composed of molecules with a kinked conformation. BD18 and BD22 crystallize into the β′ form during cooling. Only BD18 transforms into the β form during subsequent heating via a solid-solid transition. The resistance of BD22 to convert into the β form is believed to originate from the lack of mobility associated with longer aliphatic chains. In contrast, the polymorphic conversion is thought to be very efficient and to take place during cooling for the shorter BD16 molecules. The hypothesis is put forward that the original BD16 β′ nuclei are rapidly overgrown by the β phase by which crystal growth is postponed to larger supercooling where the β phase can grow from the melt. Consequently, BD16 only occurs in the β form. The melt crystallization of BD18 and BD22 into the β′ form hardly requires any supercooling.

Full text of DOI:10.1021/cg400339z

Article
pubs.acs.org/crystal
Monotropic Polymorphism in Ester-Based Phase Change Materials
from Fatty Acids and 1,4-Butanediol
†
†
‡
†
,†
S. Cabus, K. Bogaerts, J. Van Mechelen, M. Smet, and B. Goderis*
†Chemistry Department, Polymer Chemistry and Materials, Catholic University of Leuven, Celestijnenlaan 200F, 3001 Heverlee,
Belgium
‡
Faculty of Sciences (FNWI), Laboratory for Crystallography, University of Amsterdam, Science Park 904, 1098 XH Amsterdam, The
Netherlands
*
S Supporting Information
ABSTRACT: This article reports on the synthesis, thermal behavior, and
morphology of phase change materials based on diesters from 1,4-
butanediol with palmitic (BD16), stearic (BD18), or behenic (BD22)
acid. The crystalline contents and melting enthalpies of all substances
exceed 90% and 180 J/g, respectively. The molecules display monotropic
polymorphism. The metastable β′ form is built from outstretched
molecules, whereas the denser, stable β form is composed of molecules
with a kinked conformation. BD18 and BD22 crystallize into the β′ form
during cooling. Only BD18 transforms into the β form during subsequent
heating via a solid−solid transition. The resistance of BD22 to convert
into the β form is believed to originate from the lack of mobility
associated with longer aliphatic chains. In contrast, the polymorphic
conversion is thought to be very efficient and to take place during cooling
for the shorter BD16 molecules. The hypothesis is put forward that the original BD16 β′ nuclei are rapidly overgrown by the β
phase by which crystal growth is postponed to larger supercooling where the β phase can grow from the melt. Consequently,
BD16 only occurs in the β form. The melt crystallization of BD18 and BD22 into the β′ form hardly requires any supercooling.
1
. INTRODUCTION
Esters of fatty acids with different types of (poly)alcohols have
been studied to get rid of the fatty acid typical corrosivity, bad
odor, and sublimation during heating, while maintaining
reasonably high latent heats, narrow melting ranges, and little
Phase change materials (PCMs) are characterized by a large
latent heat during solid−liquid, liquid−gas, or solid−solid phase
transitions and can therefore be used for the storage and release
of thermal energy. Nowadays, PCMs are used to increase the
comfort and to optimize the energy management of buildings
or in textiles for clothing, for the smoothening of exothermic
temperature peaks in chemical reactions or the recovery of
waste heat, in medical hot−cold therapies, in solar heating
systems and green houses, as protective shielding in heat
7
−12
supercooling.
In addition, such esters are thermally very
reliable, withstanding numerous thermal cycles without loss of
9
−12
performance.
The downside of current lipid-based PCMs is
that they are limited to low temperature applications and that
they display low thermal conductivities, typically in the range
13
between 0.15 and 0.2 W/mK. For comparison, conductivity
1
,2
13
releasing, high-performance electronic equipment, etc.
values readily exceed 0.15 W/mK for inorganic substances.
Exploiting solid−liquid transitions in PCMs seems econom-
ically most attractive and since different applications require
storage capacity at different temperatures, a wide variety of
Currently, attempts are being made to increase the thermal
conductivity of lipid-based PCMs by adding highly conductive
fillers. Altering the length of the fatty acids or the alcohol
14
2
PCMs exists. Materials to be used for phase change thermal
residues, as well as blending different species, allows for a fine-
energy storage must have a large latent heat and high thermal
conductivity. They should have a melting temperature lying in
the practical range of operation, melt congruently, and
crystallize with minimum supercooling, be chemically stable,
7−10,12
tuning of transition temperatures and melting enthalpies.
To further widen the application range for lipid PCMs also
1
5,16
17
amide-
and carbonate-based derivatives have been
3
reported. The review by Sarier and Onder on organic PCMs
in textile applications contains a comprehensive overview of
low in cost, nontoxic, and noncorrosive. Given the current
climate changes and the growing awareness that the
exploitation of fossil resources should be limited, it is worth
investigating PCM materials based on renewable resources such
14
contemporary lipid-based PCMs.
4
as, for example, fatty acids.
Received: March 5, 2013
Revised: June 13, 2013
Published: June 14, 2013
5
,6
In this context, besides pure fatty acids and their blends,
also fatty acid derivatives have received considerable attention.
©
2013 American Chemical Society
3438
dx.doi.org/10.1021/cg400339z | Cryst. Growth Des. 2013, 13, 3438−3446

Crystal Growth & Design
Article
When browsing the literature, it is striking that the
optimization of novel lipid-based PCMs does not make use
of crystallographic information, although it is well-known that
dry dichloromethane (DCM) (30 mL) at room temperature.
The mixture was stirred for another 4 h prior to washing with
diluted HCl and NaHCO . After drying over MgSO , DCM
3
4
18
fatty acids and their derivatives display rich polymorphism.
was removed using a rotavapor. All compounds were purified
using a silica column with 100% DCM and appeared white.
They are denoted as BDz, with z being the number of carbon
atoms in the saturated fatty acid residue (i.e., BD16, BD18, and
BD22 for the compounds based on palmitic, stearic, and
The present paper wants to illustrate that gathering knowledge
on polymorphism is indeed relevant. The design of the studied
19
PCMs was inspired by the work of Li and Ding who
considered 1,4-butanediol distearate (BD18) as PCM because
1−
1
of its rather high melting enthalpy (182 J g ). In later work,
behenic acids, respectively.) H-NMR (BD18, 300 MHz,
8
this group also studied distearates with other linear diols,
CDCl , 298K): ∂ 0.88 (t, J = 6 Hz, 6H), ∂ 1.25 (s, 55H), ∂
3
whereas Alkan and co-workers explored the impact of replacing
1.60 (t, J = 2 Hz, 5H), ∂ 1.70 (q, J = 3 Hz, 4H), ∂ 2.29 (t, J = 7
Hz, 4H), ∂ 4.09 (t, J = 6 Hz, 4H).
1
5,16
the ester moyeties by amide groups.
This chemical
versatility allows for systematic studies on how molecular
structure, crystal packing, and polymorphism relate to the PCM
thermal behavior.
2.2. Differential Scanning Calorimetry (DSC). DSC
measurements were performed on a TA Instruments Q2000 in
combination with a RCS90 cooling device. The device was
purged with nitrogen and calibrated with sapphire and indium
for the temperature and the enthalpy. Samples of approximately
5 mg, enclosed in Tzero Aluminum hermetic pans, were
characterized during heating (first heating, referred to as ramp
A), cooling (ramp B), and heating (second heating, ramp C) at
In a first effort, we explored the impact of altering the
saturated fatty acid length (z in Figure 1) on the crystallization
and melting behavior of esters with 1,4-butanediol. Note that
fatty acids and 1,4 butanediol can be obtained from renewable
2
0
resources.
1
0 °C/min. Data were also collected during cooling at 10 °C/
min (ramp B) followed by heating at 1 °C/min (ramp D). The
experimental heat flow data were converted into specific heat,
cp(T), values by subtracting an empty pan measurement and
subsequent normalization to the sample mass and ramp rate.
The melting enthalpy, ΔH , was calculated by integrating the
m
area between the c (T) heating curve and a linear extrapolation
p
21
from the melt c (T) data. Crystallization temperatures, Tc,
p
Figure 1. Chemical structure of the PCM materials (abbreviated as
BDz), with z being the number of carbon atoms in the saturated fatty
acid residue.
were determined at the onset of crystallization. The
determination of melting peak onsets was not unambiguous.
Therefore, melting peak maxima are reported as a measure for
the melting temperature, Tm.
Interestingly, the melting enthalpy values of products with
shorter (palmitic acid) and longer (behenic acid) fatty acids
were found to be higher than for BD18. Moreover, the PCM
based on palmitic acid (BD16) displayed considerable super-
cooling in crystallization, whereas supercooling effects were
absent in PCMs based on behenic (BD22) and stearic (BD18)
acid. Explanations were found in polymorphism, crystal
perfection, and the pathway followed by the different species
to reach a given crystal form. The thermal behavior was
explored by means of differential scanning calorimetry (DSC),
the PCM crystal structures, and polymorphism by means of
temperature resolved in house and synchrotron X-ray powder
diffraction. Crystal morphology and nucleation efficiency were
addressed through optical microscopy.
2.3. Polarized Optical Microscopy (POM). Microscopic
analyses using crossed polarizers were conducted with an
Olympus BHS optical microscope (Olympus Belgium N.V.,
Aartselaar Belgium). The temperature was controlled using a
Linkam TMS 600 hot stage and a Linkam TMS 91 controller
(Linkam, Surrey, United Kingdom). A droplet of liquid material
was applied to the carrier glass at high temperature and covered
with a glass coverslip. The samples were imaged during cooling
at 10 °C/min (ramp B) every 6 s with a JVC TK-C138 color
video camera, using a script within the Leica Qwin software
(Leika, Germany). This protocol results in one image per 1 °C.
Data were also collected every °C for BD18 in a temperature
profile composed of ramp B followed by ramp D.
2.4. X-ray Diffraction. 2.4.1. In-House X-ray Powder
Diffraction and Crystal Structure Determination. X-ray
powder patterns were collected on an X’Pert MPD
diffractometer (PANalytical, Almelo, The Netherlands) equip-
ped with a sealed Cu tube, a focusing mirror, 0.01 rad soller
slits, and an X’Celerator strip detector. The samples were
loaded in a 0.7 mm glass capillary and were spinning
continuously during data collection. Data were collected at
room temperature right after synthesis and also during a ramp
D protocol [see Differential Scanning Calorimetry (DSC)] (i.e.,
during heating at 1 °C/min after cooling at approximately 10
°C/min from the melt). The temperature was controlled by a
Compact Cryostream (Oxford Instruments, Apdingdon, U.K.).
Determining crystal structures from powders of long-chain
organic compounds is quite laborious and not straightfor-
2. EXPERIMENTAL SECTION
2.1. Material Synthesis. Stearic acid was bought from
Fluka; palmitic and behenic acid were bought from Acros
Organics. They were used without further purification. Solvents
were of analytic grade and dried over calcium hydride.
Thionylchloride, 1,4-butanediol, and triethylamine were
purchased from Acros Organics and also used without further
purification. In a first step, 1 equivalent of a fatty acid [palmitic
acid (z = 16), stearic acid (z = 18) or behenic acid (z = 22),
respectively, with z being the number of carbon atoms in the
fatty acid] was added to 2 equivalents of thionylchloride
together with a drop of dimethylformamide (DMF). The
solution was refluxed and stirred for three hours. Excess
thionylchloride was removed using a rotary evaporator
2
2,23
ward
and was therefore limited to one representative
(
0
rotavapor). The resulting liquid was dripped to a solution of
.5 equivalents 1,4-butanediol and 1 equivalent triethylamine in
sample at room temperature (i.e., BD18). BD18 exists in two 2-
polymorphic forms, which were solved by using direct space
3
439
dx.doi.org/10.1021/cg400339z | Cryst. Growth Des. 2013, 13, 3438−3446

Crystal Growth & Design
Article
24
search techniques as described by David and Shankland. For
indexing the program McMaille was used, which generates a list
of potential unit cells, taking into account restrictions to the cell
2
5
dimensions. With the program Chekcell, a further selection
26
and cell refinement was carried out. For the ensuing structure
solution with FOX, a starting model with a flat molecular
27
conformation was built in a Z-matrix description. At the start-
up of the structure solution process (parallel tempering), only
translational and rotational freedom was given. At a later stage,
more flexibility was allowed while keeping some restraints to all
bond distances and bond angles. The two long saturated acyl
chains were restrained to be flat by dihedral restraints of 180°.
No dihedral angle restraints were applied to the central C4
moiety. Structure refinement was carried out with the program
2
8
GSAS. The background was modeled by a Chebyshev
polynomial. By using profile function 4, (hkl)-dependent peak
broadening as well as low-angle peak asymmetry was taken into
account. Soft restraints were applied to all distances and angles.
Soft planar restraints were applied to the saturated acyl chains
at either side of the central C4 moiety. Atomic displacement
parameters were not refined. Figures showing the agreement
between calculated and experimental powder diffraction data
are available in the Supporting Information. Figures, represent-
ing crystal structure features, were made, using the program
Mercury 3.0. Scattering patterns are represented as a function
of the scattering vector modulus, q = (4π/λ) sin(θ), with θ
being half the scattering angle and λ, the X-ray wavelength.
2.4.2. Time-Resolved Synchrotron X-ray Powder Diffrac-
tion. Temperature-resolved synchrotron X-ray powder dif-
fraction was performed at DUBBLE, the Dutch-Belgian
beamline (BM26) at the European Synchrotron Radiation
Facility (ESRF; Grenoble, France), using a wavelength, λ, of
1
.24 Å. The PCM samples were presented in a rubber ring,
Figure 2. DSC c (T) data: (A) first heating at 10 °C/min, (B)
sandwiched between thin mica sheets. The sandwich
constructions were kept together in crimped, perforated
aluminum Perkin-Elmer DSC pans. The small-angle X-ray
scattering (SAXS) parts of the powder patterns were measured
on a two-dimensional (2D) multiwire gas-filled detector placed
at 1.3 m from the sample after an evacuated tube. The wide-
angle X-ray diffraction (WAXD) parts were collected on a
Pilatus 300K−W detector from Dectris put closely to the
sample. The scattering angles were calibrated using the in-
house X-ray powder diffraction data of the materials. The SAXS
and WAXD patterns were normalized to the intensity of the
incoming beam, and measured by a photodiode at the beam
stop close to the SAXS detector. The 2D data were azimuthally
averaged using ConeX, corrected for the empty holder
scattering, and represented as a function of the scattering vector
modulus q. X-ray patterns were collected in consecutive frames
of 6 s during heating, cooling, and (second) heating at 10 °C/
min (i.e., a B ramp followed by a C ramp), which results in one
frame per 1 °C. Temperature was imposed through a Linkam
HFS 191 heating/freezing stage.
p
subsequent cooling at 10 °C/min, (C) subsequent heating at 10 °C/
min, and (D) heating at 1 °C/min after cooling at 10 °C/min for
BD16 (top), BD18 (middle), and BD22 (bottom). Heating runs are
plotted upward, and cooling runs are plotted downward. The Greek
letters refer to the melting or crystallization of a given polymorphic
form. The arrow in the middle panel points to some exothermic heat
developing during heating at 1 °C/min. The curves C and D are
shifted upward with 20 and 40 c (T) units, respectively, for clarity.
p
Table 1. DSC-Based Thermal Properties of All Lipid-Based
PCM, Obtained during Heating after (A) Synthesis, (B)
a
Subsequent Cooling, and (C) Heating at 10 °C/min.
29
BD16
BD18
BD22
Tm (°C)
62(A)
58(A)
74(A)
6
2(C)
54(C)
52(B)
70(C)
67(B)
Tc (°C)
51(B)
△
Hm (J g−1)
188(A)
186(A)
181(C)
209(A)
199(C)
1
88(C)
crystalline fraction
0.92(A)
0.93(A)
0.97(A)
0
.92(C)
0.91(C)
0.94(C)
△
△
Hm0 (J g−1
)
204(A)
199(A)
215(A)
3
. RESULTS AND DISCUSSION
1
98(C)
0.60(A)
.60(C)
212(C)
Figure 2 represents the DSC c (T) A, B, C, and D ramps for all
PCM materials collected between 20 and 80 °C. The
crystallization onset, T , and melting peak, T , temperatures
are listed in Table 1 together with the melting enthalpy, ΔH ,
values for the A and C ramps.
p
Sm0 (J g−1 K−1
)
0.61(A)
0.62(A)
0.62(C)
0
c
m
a
Crystalline fractions are extracted from the synchrotron WAXD data
m
at 20 °C.
All heating and cooling runs of BD16 are unimodal. The first
ramp A) and second (ramp C) heating runs are virtually
identical (see also data in Table 1). The endothermic melting
(
signal is narrower and the melting peak, therefore, occurs at
slightly lower temperatures in ramp D compared to in ramps A
3
440
dx.doi.org/10.1021/cg400339z | Cryst. Growth Des. 2013, 13, 3438−3446

Crystal Growth & Design
Article
and C, which is expected when lower heating rates are used.
The melting onset temperatures are not affected by the heating
rate. The cooling ramp B may seem odd as the exothermic peak
describes a loop as a function of the temperature. The heat
developed during the crystallization of BD16 makes the sample
temperature rise temporarily. This effect, which can be captured
thanks to the Tzero technology, is only possible for samples
that crystallize at high supercooling and the natural tendency to
30
return to the thermodynamic melting point. For BD16, the
difference between T and T is 11 °C. For the other samples,
c
m
there is hardly any supercooling and therefore also no looping
of the crystallization peak. The ramp A synchrotron WAXD
data in Figure 3 illustrate that only one crystal type, referred to
Figure 3. Selection of synchrotron X-ray WAXD data during ramp A
for BD16 (data are presented using a linear intensity scale). The right-
hand side numbers refer to the temperature in °C. At the left, the
crystal type or the melt state is indicated.
Figure 4. Synchrotron X-ray SAXS and WAXD data, taken at 25 °C
during the A ramps (left column panels) and C ramps (right column
panels) for the different BD samples, as indicated. The strongest
reflections are highlighted with vertical lines and are labeled with their
Miller indices. Note that a logarithmic intensity scale is used. The
patterns in the left and right columns are typical for β and β′ crystals,
respectively.
as the β form, exists for BD16 and that it converts into a melt
where DSC produces an endothermic peak. Similar WAXD
patterns are observed in the synchrotron B and C ramps and in
the in-house X-ray D ramp.
Figure 4 contains both the synchrotron SAXS and WAXD
data for BD16 taken in ramp A at 25 °C. The dominant
reflections are labeled with the β crystal Miller indices, of which
the corresponding crystal structure will be discussed further
below.
The melting point and enthalpy of BD18 during the first
heating run (ramp A) are both higher compared to in the
second heating run (ramp C) (see Table 1). The synchrotron
X-ray pattern of freshly synthesized BD18 strongly resembles
that of BD16, as illustrated in Figure 4, by means of the 25 °C
SAXS and WAXD patterns of ramp A. During heating in ramp
A, this BD18 β crystal form melts just like BD16 does, albeit at
a slightly lower temperature. However, upon cooling in ramp B,
another crystal form, the β′ form, is generated with a scattering
pattern, as illustrated in Figure 4, via the 25 °C patterns of ramp
C. The synchrotron X-ray data further reveal that this β′ form
simply melts during subsequent heating at 10 °C/min. The
ramp C endothermic peak in Figure 2 is labeled accordingly.
Clearly, the melting point difference between ramps A and C
for BD18 is related to polymorphism.
When heated at 1 °C/min after a ramp B cooling, the BD18
melting peak splits into two contributions (ramp D). Moreover,
a small exothermic signal appears shortly before the melting
onset, as indicated by the arrow in Figure 2. The corresponding
in-house X-ray data in Figure 5 reveal that the cooling-induced
β′ form converts into the more stable β structure exactly in the
temperature range of the exothermic DSC signal. This
3
441
dx.doi.org/10.1021/cg400339z | Cryst. Growth Des. 2013, 13, 3438−3446

Crystal Growth & Design
Article
Figure 5. Selection of in-house X-ray data during ramp D for BD18
(data are presented using a linear intensity scale). The right-hand side
numbers refer to the temperature in degrees Celcius. At the left, the
crystal type is indicated.
conversion seems to be fairly complete as judged from the X-
ray data. In DSC, however, the small endothermic peak at 54
°
C points to some remaining β′ phase that melts prior to the
Figure 6. Optical microscopy images, using crossed polarizers, for
BD18 at selected temperatures during heating at 1 °C/min (ramp D).
Prior to heating, the material was cooled at 10 °C/min (ramp B). Note
that the black patches in the images at 20.5 and 40 °C are not due to
the presence of liquidlike material but to crystals that are rotated so
that their optical axis runs parallel the polarizer or analyzer direction.
Black in the images at 57 and 58 °C truly corresponds to molten
material.
converted β fraction. Just like for BD16, the melting peak of
this β fraction occurs at a slightly lower temperature when
heated at 1 °C/min compared to when heated at 10 °C/min
(
compare with the BD18 β peak in ramp A). The absence of a
melting event prior to the conversion exotherm as well as the
absence of meltlike features in the X-ray patterns in this
temperature range demonstrate that the β′ to β transition is a
solid−solid transition. This notion is further supported by the
optical microscopy images in Figure 6. Except for small changes
in birefringence, no modifications to the crystals are observed
after the transition between 40 and 50 °C. In a melt-mediated
process, crystals with a totally different orientation would have
been created.
Clearly, in BD18, the β form is less easily created compared
to in BD16. This resistance to forming the β form is further
enhanced in BD22. The BD22 synchrotron X-ray patterns in
Figure 4 at 25 °C in ramp A, which represents the situation
right after the synthesis, clearly display β characteristics. The
DSC melting trace in ramp A, however, is bimodal (see Figure
which is also highlighted in Figure 7. Although certainly
present, no efforts were made to highlight the β′ features at
temperatures below 50 °C due to overlap with the dominating
β peaks. Since the β′ fraction is vanishingly small, only the β
melting peak temperature is included in Table 1 for ramp A.
When cooled down at 10 °C/min, BD22 converts into the β′
form. During subsequent heating and irrespective of whether it
happens at 10 (ramp C) or 1 °C/min (ramp D), no conversion
into the β form takes place prior to melting. Indeed, the
conversion from β′ to β seems more difficult for BD22
compared to that for BD18, where crystallization from a solvent
(which is what happens during the synthesis) or the application
of long-term high-temperature protocols suffice for a complete
conversion into the β form. In BD16, the β form dominates
completely. Noteworthy is that the optical microscopy
morphology of BD16 is more fine compared with that of
BD18 or BD12, as illustrated in Figure 8 for the room
temperature situation after cooling at 10 °C/min.
2
), revealing the melting of some β′ material in the low-
temperature peak in front of that of the β fraction.
The high-temperature synchrotron WAXD data in Figure 7
support this explanation. From above approximately 50 °C, the
β′ form 040 reflection (see Figure 4 for the assignment of the
reflections) starts to drift to smaller q values, away from below
the β 110 reflection. This small 040 feature is shaded black in
Figure 7 and disappears completely by melting before reaching
The fact that the β′ phase converts into the β phase via an
6
9 °C, in agreement with the DSC data. The β′ phase also
exothermic solid−solid transition makes the β′ and β phases
−1
31
contributes some less conspicuous intensity at q = 1.5 Å ,
monotropically related. In accordance with the enthalpy of
3
442
dx.doi.org/10.1021/cg400339z | Cryst. Growth Des. 2013, 13, 3438−3446

Crystal Growth & Design
Article
value for BD18 in Table 1, corresponding to the A ramp (β
form) is higher than that of the C ramp (β′ form).
However, in comparing melting enthalpies, one needs to
ensure that all samples are fully crystalline, or alternatively, that
deviating crystallinities are accounted for. Therefore, crystal-
linity values were estimated from the synchrotron WAXD data
at 25 °C. At first, a linear background connecting the X-ray
values at q = 1 and q = 2 Å− was subtracted. Next, a Gaussian
was used to approximate the liquid scattering. The ratio of the
crystalline contribution to the total background corrected
1
2
WAXD intensity in a q weighted intensity representation
yielded a crystallinity estimate, which was further multiplied
with 1.032 and 1.025 for the β and β′ phases, respectively, to
account for the crystalline contributions at q values below q = 1
−1
Å . These correction factors originate from an analysis of the
theoretical BD18 β′ and β scattering patterns that in turn result
from the crystal structures of the BD18 polymorphs, discussed
further below. Liquid and crystalline contributions above q = 2
−
1
Å
were neglected, assuming that their shares are proportional
−1
to those in the range 0 < q < 2 Å . Normalizing the ΔH
m
0
values to the corresponding crystallinities, yield the ΔH
m
values of fully crystalline material, represented in Table 1.
The ramp A values, representative for the β phase
(
predominantly the β phase for BD22) are still higher
compared to the ramp C values that correspond to the β′
phase melting. The differences are, however, strongly reduced
compared to when no crystallinity corrections were applied.
Figure 7. Selection of synchrotron X-ray WAXD data during ramp A
for BD22 (data are presented using a linear intensity scale). The right-
hand side numbers refer to the temperature in degrees Celcius. The
black-shaded areas highlight the β′ crystal form contribution. Except
for the melt pattern at 78 °C, all patterns are dominated by reflections
typical for the β crystal form.
0
The ratio of the ΔHm values to the corresponding melting
0
m
temperatures (in K) results in melting entropy values, ΔS ,
that are very similar for the different BD samples and
0
polymorphic forms (Table 1). These ΔSm values agree with
those calculated by means of the group contribution method
3
2,33
−1
−1
described by Dannenfelser et al.,
yielding 0.61 J g K ,
irrespective of the molecule at hand. The similarity between the
0
calculated and experimental ΔSm values implies that the PCM
chains gain full flexibility upon melting. That a disordered melt
is created can be deduced from the absence of a broad band in
18
the SAXS region (data not shown).
The crystallographic study served at finding molecular
explanations for some of the thermal characteristics listed
above. The crystallographic data, collected in Table 2, for BD18
were obtained via a full crystallographic analysis as explained in
Experimental Section. Details are available in the Supporting
Information (.cif file format). Data on the other samples are
limited to the crystalline unit cell, assuming space groups
identical to their BD18 β′ or β analogues. To obtain the unit
cell dimensions in these cases, the labeled reflections in Figure
4
were used. The cell optimization procedure started from a
first guess, based on the altered molecular length.
Figure 9 illustrates the crystallographically determined
molecular conformations in the BD18 β′ and β phases at 20
°
C. In the BD18 β′ phase, the two aliphatic chains of the
molecules are configured coaxially and the central butanediol
moiety is bent. In the β form, the aliphatic axes are displaced by
a kink at the central moiety.
Figure 10 contains projections of the BD18 crystal structures.
There are 4 and 2 symmetry-related molecules within the β′
and β unit cells, respectively. Both crystal forms are organized
as stacks of layers with a thickness given by the 001 spacing,
which for the β′ and β forms equals 51.88 Å and 46.78 Å,
respectively. Clearly, the chain tilt in the β form is larger than in
the β′ form. In both crystal forms, the aliphatic chains do not
adopt a perfect all-trans configuration. Although their packing is
Figure 8. Optical microscopy images taken at 20 °C after cooling at 10
°
C/min for (a) BD16, (b) BD18, and (c) BD22.
fusion rule, the higher melting polymorph of a monotropic pair
should also have the higher melting enthalpy. Indeed, the ΔHm
3
443
dx.doi.org/10.1021/cg400339z | Cryst. Growth Des. 2013, 13, 3438−3446

Crystal Growth & Design
Article
Table 2. Summary of the Crystallographic Unit Cell Parameters
BD16 (20 °C)
BD18 (20 °C)
BD22 (20 °C)
β
β′
β
β′
β
system
space group
a (Å)
triclinic
monoclinic
triclinic
monoclinic
triclinic
P1
P1
̅
P2 /a
1
P1
̅
4.52
P2 /a
1
̅
4.49
5.05
14.79
51.93
90
5.02
14.71
62.17
90
4.60
b (Å)
8.62
9.05
8.60
c (Å)
45.16
104.63
84.90
99.42
1665.51
2
50.24
110.97
82.27
100.75
1878.489
2
60.45
105.02
75.12
101.08
2210.87
2
α (°)
β (°)
92.46
90
91.62
90
γ (°)
volume (Å )
3
3876.853
4
4592.348
4
molecules in cell
density (g/dm3)
1.130
1.067
1.101
1.062
1.103
detect them. This happens in a temperature range where the
more stable β phase does not nucleate as a result of the kinetic
barriers associated with the formation and stacking of kinked
molecules. In accordance with classical nucleation theory, for a
crystal nucleus to be stable, it needs to reach a minimum size
before it can overcome the free-energy penalty associated with
the formation of crystal surfaces. On this account, a rather high
number of molecules would have to adopt a kinked
conformation in a concerted way, which seems highly
improbable. This number of molecules is particularly large at
low supercooling, where the driving force for crystallization is
low. Assuming that growth occurs via a process of secondary
nucleation, the growth of β phases at low supercooling is also
prohibited. In contrast, the β′ nucleation is easier as it happens
from the collection of chains in their straight configuration.
However, at low supercooling, the rate of β′−β conversion in
BD16 is so fast that it happens before the β′ nuclei were able to
substantially grow. Moreover, as soon as the β′−β conversion
reaches the β′ crystal growth front, its growth is promptly
arrested because only β phases are facing the melt. This
scenario may happen very frequently during the cooling
process, by which BD16 becomes loaded with a high number
of β nuclei. These β nuclei only grow at a sufficiently high
degree of supercooling where, as a result of the higher
crystallization driving force, smaller secondary nuclei can be
formed. Indeed, the nucleation density for BD16 is very high
Figure 9. Crystallographically determined molecular conformations in
the BD18 β′ and β phases at 20 °C (red represents oxygen, gray is
carbon, and white is hydrogen).
not perfectly parallel, it is fair to state that the chain packing
subcell of the β′ phase resembles an orthorhombic O⊥ packing
1
8
and that of the β phase the more dense triclinic T ∥ packing.
In accordance with the lipid crystal nomenclature, this justifies
18
labeling the two structures as β′ and β, respectively.
It is conceivable that the stacking of coaxial units is easier
than that of kinked objects. That a conversion into the
metastable, less dense β′ form is kinetically favored above the
34
stable β phase is in agreement with Ostwald’s rule of stages, at
least for BD18 and BD22. For monotropically related pairs, the
equilibrium interconversion temperature lies above the melting
31
point of both polymorphs and is thus virtual. Therefore, in
most cases, interconversions are melt-mediated. Solid−solid
conversions are only possible when the molecules within the
metastable solid phase display a minimum mobility and when
the interconversion is not too involving. The change of the
central butanediol configuration, together with a shearing-
rotation movement of the aliphatic chains seems to belong to
this category. Because of frictional forces, such a movement will
be more difficult the longer the molecules are and explains why
BD22 is more resistant to the β′−β conversion compared to
BD18. Furthermore, this solid−solid process involves nuclea-
tion, which takes time. When cooled and heated at 10 °C/min,
insufficient time is spent in the temperature range where the
mobility within the β′ phase is high enough for β nucleation in
BD18. This does not hold for heating at 1 °C/min, where
ample time is provided for nucleation. This time seems to be
insufficient for the less mobile BD22. Along this line of thought,
the β′−β conversion for shorter molecules, such as BD16,
should be quite efficient and could even be happening during
cooling. Ultimately, this conversion could be so fast that it
escapes the observation. The following hypothesis therefore
seems plausible.
(
Figure 8a) and crystallization at a supercooling of 11 °C
occurs very massively, giving rise to the looping phenomenon
observed in DSC (Figure 2).
When the β form is grown from the β′ form via a solid−solid
transition (such as in the BD18 and BD22 cases), one may
imagine a number of stacking defects where the conformational
change was only partially successful. The presence of such
defects will lower the density. If, on the other hand, the β form
is grown directly from the melt, such as in the BD16 case, the
amount of defects is expected to be much lower and the density
higher. The higher crystal density for BD16 (see Table 2) is
0
mirrored in an unexpectedly high ΔH
and T
(see Table 1).
m
m
0
Unexpected indeed, since in even normal paraffins ΔH
m
3
5
increases with increasing chain length. Such a trend is visible
for BD18 and BD22, for the β′ as well as for the β form. The
0
ΔH value and melting point of the BD16 β phase do not
m
follow this trend. Given the crystallographic density values,
0
however, it seems that the BD16 β phase ΔHm and T values
m
are not exceptionally high but that the values for the other two
BD samples are rather low, as a result of crystal defects. Note
that the presence of defects is also reflected in larger peak
When cooling BD16 from the melt, small β′ nuclei are
formed at such a low concentration that DSC or X-rays cannot
3
444
dx.doi.org/10.1021/cg400339z | Cryst. Growth Des. 2013, 13, 3438−3446

Crystal Growth & Design
Article
Figure 10. Projections of the BD18 crystal structures at 20 °C. The vertical direction in each image runs parallel to the 001 direction. The images A
and D are projections along the unit cell a axis (red axis), whereas the projections in B and C are along the unit cell b axis (green axis). The
crystallographic c axis is represented blue. The images A and B correspond to the BD18 β′ form, the images C and D to the β form.
widths for BD18 and BD22 compared to in BD16 (see Figure
cooling in crystallization, whereas supercooling effects are
absent in PCMs based on behenic (BD22) and stearic acid.
BD18 and BD22 display monotropic polymorphism with the β
phase being stable and the β′ phase metastable. Both BD18 and
BD22 occur completely or to a large extent in the β form when
crystallized from solution during purification after synthesis but
crystallize into the β′ form during cooling from the melt. A
detailed crystallographic study revealed that the β′ form is built
from the layer-wise stacking of outstretched molecules. During
subsequent heating, only BD18 transforms into the more stable
β form via a solid−solid transition, provided a slow heating rate
is used. In this conversion, the molecules adopt a kinked
configuration at the central butanediol moiety and stack at an
increased tilt angle with respect to the layer normal. The β form
is the stable form due to its higher crystal density and melting
enthalpy. The resistance of BD22 to convert into the more
stable β form during heating is believed to originate from a lack
of mobility as a result of the longer aliphatic chain length. BD16
only occurs in the β form, even when cooled from the melt.
0
4
). The data in Table 2 also reveal that the higher ΔHm values
for the β form (compared to the β′ form) correlate with the
higher densities for the β phase. This observation complies with
the density rule, stating that the polymorph with the highest
density is the stable one. Note that the densities of the different
polymorphic forms remain fairly constant over the temperature
range studied (data not shown).
4. CONCLUSIONS
This article reported on the synthesis, thermal behavior, and
temperature-dependent morphology of three phase change
materials (PCM) based on diesters from 1,4-butanediol with
saturated fatty acids of different lengths. All substances
crystallize rather well with crystalline contents exceeding 90%.
The melting enthalpy values of products with shorter (palmitic
acid) and longer (behenic acid) fatty acids are higher than for
products based on stearic acid (BD18). Moreover, the PCM
based on palmitic acid (BD16) display considerable super-
3
445
dx.doi.org/10.1021/cg400339z | Cryst. Growth Des. 2013, 13, 3438−3446

Crystal Growth & Design
Article
The hypothesis is put forward that when BD16 is cooled from
the melt, small β′ nuclei are nevertheless formed, but that the
rate of β′−β conversion is very rapid. As a result, the original β′
crystal growth front is rapidly overgrown by the β phase by
which the growth of the nucleus is postponed to larger
supercooling, where the driving force for growth in the β form
is sufficiently large.
(21) Mathot, V. B. F. In Calorimetry and Thermal Analysis of Polymers,
Mathot, V. B. F., Ed.; Hanser Publishers: New York, 1994; Chapter 5,
pp 105−167.
(
22) Van Mechelen, J. B.; Peschar, R.; Schenk, H. Acta Crystallogr.
006, B62, 1121−1130.
23) Van Mechelen, J. B.; Peschar, R.; Schenk, H. Acta Crystallogr.
006, B62, 1131−1138.
24) David, W. I. F.; Shankland, K. Acta Crystallogr. 2008, A64, 52−
4.
(25) Le Bail, A. Powder Diffr. 2004, 19, 249−254.
26) Laugier, J.; Bochu, B. Chekcell (http://www.ccp14.ac.uk/about.
htm), 2004.
27) Favre-Nicolin, V.; Cerny,
43.
(28) Larson, A. C.; Von Dreele, R. GSAS; Report No. LA-UR-86-
2
(
2
(
6
ASSOCIATED CONTENT
■
(
*
S
Supporting Information
X-ray crystallographic information files (CIF) are available for
the β and β′ polymorph of compound BD18, as well as images
showing the agreement between the calculated and exper-
imental powder diffraction data. This material is available free
of charge via the Internet at http://pubs.acs.org.
(
́
R. J. Appl. Crystallogr. 2002, 35, 734−
7
748; Los Alamos National Laboratory: New Mexico, USA, 1987.
(29) Gommes, C. J.; Goderis, B. J. Appl. Crystallogr. 2010, 43, 352−
3
55.
(
(
30) Sanders, B.; Rekstad, J. Sol. Energy 2006, 80, 616−625.
AUTHOR INFORMATION
■
31) Henck, J.-O.; Kuhnert-Brandstatter, M. J. Pharm. Sci. 1999, 88,
Corresponding Author
E-mail: bart.goderis@chem.kuleuven.be. Tel: +32 16327806.
Fax: +32 16327990.
103−108.
*
(32) Dannenfelser, R. M.; Yalkowsky, S. H. Ind. Eng. Chem. Res. 1996,
3
5, 1483−1486.
33) Dannenfelser, R. M.; Yalkowsky, S. H. J. Pharm. Sci. 1999, 88,
22−724.
34) Ostwald, W. F. Z. Zeitschrift fu
89−330.
(35) Domanska, U.; Wyrzkykowska-Stankiewicz, D. Thermochim.
Acta 1991, 179, 265−271.
(
7
(
2
Notes
The authors declare no competing financial interest.
̈
r Physikalische Chemie 1897, 22,
ACKNOWLEDGMENTS
■
The authors thank FWO-Vlaanderen for supporting DUBBLE
at the ESRF through the Big Science Project G.0707.08. S.C.
thanks the Flemish Agency for Innovation by Science and
technology (IWT) for a Ph.D. scholarship.
REFERENCES
■
(
1) Sharma, A.; Tyahi, V. V.; Chen, C. R.; Buddhi, D. Renewable
Sustainable Energy Rev. 2009, 13, 318−345.
2) Zalba, B.; Marin, J. M.; Cabeza, L. F.; Mehling, H. Appl. Therm.
Eng. 2003, 23, 251−283.
3) Farid, M. M.; Khudhair, A. A.; Razack, S. A. K.; Al-Hallaj, S.
Energy Convers. Manage. 2004, 45, 1597−1615.
4) Menoufi, K.; Castell, A.; Farid, M. M.; Boer, D.; Cabeza, L. F.
(
(
(
Renewable Energy 2013, 51, 398−403.
(
(
5) Kaygusuz, K.; Sari, A. Energy Sources, Part A 2006, 28, 105−116.
6) Kenisarin, M. M.; Kenisarina, K. M. Renewable Sustainable Energy
Rev. 2012, 16, 1999−2040.
7) Suppes, G. J.; Goff, M. J.; Lopes, S. Chem. Eng. Sci. 2003, 58,
751−176.
8) Li, W.-D.; Ding, E.-Y. Mater. Lett. 2007, 61, 4325−4328.
9) Sari, A.; Bicer, A.; Karaipekli, A.; Alkan, C.; Karadag, A. Sol.
(
1
(
(
̧
Energy Mater. Sol. Cells 2010, 94, 1711−1715.
(
(
(
(
10) Sari, A. Energy Convers. Manage. 2012, 64, 68−78.
11) Aydin, A. A. Sol. Energy Mater. Sol. Cells 2003, 108, 98−104.
12) Aydin, A. A. Sol. Energy Mater. Sol. Cells 2013, 113, 44−51.
̀
13) Tasidjodoung, P.; Le Pierres, N.; Luo, L. Renewable Sustainable
Energy Rev. 2013, 18, 327−349.
(
(
(
14) Sarier, N.; Onder, E. Thermochim. Acta 2012, 540, 7−60.
15) Canik, G.; Alkan, C. Sol. Energy 2010, 84, 666−672.
̈
16) Alkan, C.; Canik, G.; Dunya, H.; Sari, A. Sol. Energy Mater. Sol.
Cells 2011, 95, 1203−1207.
(
(
17) Kenar, J. A. Sol. Energy Mater. Sol. Cells 2010, 94, 1697−1703.
18) Larsson, K. In The Lipid Handbook; Gunstone, F. D.,
Hardwood, J. L.; Padley, F. B., Eds.; Chapman and Hall: London,
1
(
(
986, Chapter 8, pp 321−384.
19) Li, W.-D.; Ding, E.-Y. Mater. Lett. 2007, 61, 1526−1528.
20) Yim, H.; Haselbeck, R.; Niu, W.; Pujol-Baxley, C.; Burgard, A.;
Boldt, J.; Khandurina, J.; Trawick, J. D.; Osterhout, R. E.; Stephen, R.;
Estadilla, J.; Teisan, S.; Schreyer, H. B.; Andrae, S.; Yang, T. H.; Lee, S.
Y.; Burk, M. J.; Van Dien, S. Nat. Chem. Biol. 2011, 7, 445−452.
3
446
dx.doi.org/10.1021/cg400339z | Cryst. Growth Des. 2013, 13, 3438−3446