Fuel 222 (2018) 243–249

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Cloud point and crystallization in fatty acid ethyl ester biodiesel mixtures T
with and without additives
⁎
Patrick A. Leggieri, Michael Senra, Lindsay Soh
Department of Chemical and Biomolecular Engineering, Lafayette College, 740 High Street, Easton, PA 18042, United States

A R T I C L E I N F O A B S T R A C T

Keywords: Biodiesels, composed of saturated fatty acid alkyl esters (FAAE), have relatively high cloud points (CP), which
Biodiesel limit their commercial application. However, fatty acid ethyl esters (FAEE) have potentially advantageous
Fatty acid ethyl ester properties, including those related to cold flow, compared to their methyl analogs. This work systematically
Cloud point evaluates FAEE cloud point and crystallization data to aid in the development of FAEE biodiesel blends with
Crystallization
practical cold flow properties. Neat FAEE CP increases with alkyl-chain length and decreases with greater degree
Cold-flow properties
of unsaturation. Mixtures of saturated/saturated FAEE show partial cocrystallization, which impact the mixture’s
CP. The effects of cold-flow improving additives dimethyl azelate (DMAz) and triacetin on fatty acid methyl
esters (FAME) and FAEE are also analyzed. Additives lowered the overall CP, however the degree of depression
for each additive varied indicating different crystallization mechanisms. Thermodynamic ideal solution mod-
eling predicts binary FAEE CP with and without additives within 2% of experimental values.

1. Introduction esters to aid in the development of better fuel properties.

Biodiesel is a promising alternative to fossil-derived liquid fuels [1].
Some advantages of biodiesel include its renewability, biodegradability,
lubricity, and low emission of sulfurous and aromatic species [2,3].
Other favorable fuel properties include its high flash point, complete
miscibility with petrodiesel at all blend levels, and ability to directly
replace petrodiesel in diesel engines [2–5]. Other uses include home
heating oil, lubricants, and numerous other applications [6,7].
Biodiesel is synthesized via the transesterification of triglycerides
sourced from biofeedstocks to produce fatty acid alkyl esters (FAAE)
[3]. Typically, this reaction uses methanol as the reagent, yielding fatty
acid methyl esters (FAME) [8]. Methanol is favored due to its low
production cost [9]; however, it is toxic and commonly derived from
natural gas, meaning that it is nonrenewable [10]. An alternative to
FAME are fatty acid ethyl esters (FAEE), produced via transesterifica-
tion with ethanol (ethanolysis). In the United States, bioethanol is most
commonly derived from corn, [11] meaning that FAEE biodiesel can be
completely produced from renewable feedstocks. FAEE biodiesel also
has enhanced heat content and cetane number, and emits less carbon
monoxide than FAME biodiesel [12]. However, FAEE production may
demand increased feedstock costs as well as energy and separations
requirements compared to FAME [13]. These processing challenges
may be improved with technology and feedstock advancements [14,15]
as well as further characterization of these relatively under-studied
Despite its advantages, biodiesel may have higher NOx emissions,
lower oxidative stability, and poorer cold-flow properties (operability
at low temperatures) [16] compared to petrodiesel. The use of FAEE
instead of FAME has been shown to reduce NOx emissions [16] and can
potentially improve cold-flow behavior.
Solidification of biodiesel typically occurs between 263 and 298 K,
in comparison to 246 to 258 K for petrodiesel [7,17,18]. These higher
phase transition temperatures can lead to poor cold weather perfor-
mance; as such the use of biodiesel is commonly limited to petrodiesel
blends, typically at 20 wt% biodiesel (B20) or less [1,19].
Cloud point (CP) is often considered the most important and diffi-
cult cold-flow metric to improve [6,7,20]. It is defined as the tem-
perature at which solid wax crystals at least 0.5 µm in diameter are
formed [21], causing the solution to appear opaque and “cloudy.” As
such, the CP is the temperature at which engine problems begin to
occur due to formation of solids within the biodiesel fuel. Although no
numerical standards exist for biodiesel CP, ASTM D6751 specifies that
it must be reported [22]. Several approaches can be taken to improve
biodiesel CP: utilizing/manipulating biofeedstocks with advantageous
lipid profiles, blending with petrodiesel, modifying the FAAE transes-
terification product via fractionation or winterization, or adding cold-
flow improving chemical additives [20,23,24]. The appropriate selec-
tion of lipid profiles and additives for FAEE based biodiesel requires a
general understanding of the role of FAEE composition and

⁎
Corresponding author.
E-mail address: sohl@lafayette.edu (L. Soh).

https://doi.org/10.1016/j.fuel.2018.02.100
Received 18 September 2017; Received in revised form 11 February 2018; Accepted 14 February 2018
Available online 23 March 2018
0016-2361/ © 2018 Elsevier Ltd. All rights reserved.
P.A. Leggieri et al. Fuel 222 (2018) 243–249

interactions. However, most cold-flow research, especially the studies 3. Theory and calculations
assessing pure FAAE properties, has focused on FAME biodiesel. Some
studies [25–28] have analyzed the eutectic/metatectic behavior in 3.1. Thermodynamic cloud point model
binary and ternary FAEE mixtures using melting point, yet there is a
lack of extensive CP data in literature for both pure FAEE and FAEE A thermodynamic model based on solid-liquid equilibrium was used
blended with chemical additives. for modeling of the phase transitions based on the work of Imahara
Biofuel additives may be chosen based on their physicochemical et al. [39].
properties and their ability to be renewably-derived. Two potential L
⎛f ⎞ γ L xi ΔHfus,i TCP,i ΔC TCP,i ⎞ ΔCi ⎛ TCP,i ⎞
additives, triacetin and dimethyl azelate, show particular promise in ln ⎜ iS ⎟ = ln ⎛⎜ iS ⎞⎟ = ⎛ −1⎞− i ⎛ −1 + ln
these metrics. Triacetin, which can be produced via interesterification f
⎝ i ⎠ ⎝ γi z i ⎠ RTm ,i ⎝ T ⎠ R ⎝ T ⎠ R ⎝ T ⎠
of triglycerides with alkyl acetates [29,30] or through esterification of (1)
glycerol [31], has been shown to improve the CP of FAME biodiesel
where:
blends [4]. Dimethyl azelate can be derived from glycerol; it has a low
CP, is miscible with biodiesel, and has been shown to be a pour point
fiL and fiS are the liquid and solid phase fugacities of species i, re-
depressant for sunflower oil biodiesel [32].
spectively
In this work, a systematic analysis of the CP and crystallization
γiL and γiS are the liquid and solid phase activity coefficients of
behavior of neat and binary FAEE mixtures was conducted to deepen
species i, respectively
the understanding of FAEE biodiesel cold-flow properties and aid in the
xi and zi are the liquid and solid phase mole fractions of species i,
development of biodiesel blends with favorable cold-flow properties.
respectively
The effects and mechanisms of triacetin and dimethyl azelate as ad-
ΔHfus,i is the molar enthalpy of fusion for species i (J mol−1)
ditives were also evaluated for both pure FAEE and FAEE mixtures.
R is the gas constant (8.314 J mol−1 K−1)
Furthermore, the results were compared against a thermodynamic
TCP,i is the cloud point of pure species i (K)
model using solid-liquid equilibrium (SLE) with assumptions based on
T is the cloud point of species i in solution (K)
ideal solution theory to allow for the prediction of CPs of multi-com-
ΔCi is the difference of heat capacity of species i in the solid and
ponent FAEE and FAME blends.
liquid phases (J mol−1 K−1).

Since all FAAE mixture components analyzed are structurally si-

2. Materials and methods
All FAAE blends were prepared and analyzed via differential scan-
ning calorimetry (DSC, TA Instruments DSC 2920), using the procedure
reported previously [4]. The DSC was calibrated with deionized water;
six calibration trials yielded a melting onset temperature of 272.8 K,
with a standard deviation of 0.2 K.
The FAME and FAEE studied were chosen for the prevalence of their
corresponding triglycerides in common biofeedstocks [9,20,24,33].
Methyl palmitate (methyl-C16:0, > 99%), methyl stearate (methyl-
C18:0 ∼99%), and methyl oleate (methyl-C18:1, > 99%) were sourced
from Sigma-Aldrich. Methyl myristate (methyl-C14:0, > 99%) and
methyl linoleate (methyl-C18:2, > 99%) as well as all ethyl esters
–ethyl myristate (C14:0, > 99%), ethyl palmitate (C16:0, > 99%),
ethyl stearate (C18:0, > 99%), ethyl oleate (C18:1, > 99%), and ethyl
linoleate (C18:2, > 99%) – were sourced from Nu-Check Prep. All
binary FAAE mixtures studied were composed of 50:50 mass ratios.
Ternary mixtures were comprised of 50:50 FAAE ratios and additives
dimethyl azelate (> 98.5%, Sigma-Aldrich) and triacetin (99%, Sigma-
Aldrich), at 15 wt% of the total mixture. Additional experimentation
was performed to examine the effects of predictive modeling using
soybean oil derived biodiesel, SoyGold® (provided by Ag Processing
Inc).
In this work, the cloud point is defined as the location on the DSC
thermogram where the heat flow initially decreases due to phase
transition. For cocrystallizing mixtures or samples containing poly-
morphic species with multiple freezing peaks, the cloud point is defined
as the onset of the first (highest temperature) peak. Melting point is
defined as the peak melting temperature, corresponding to the point of
greatest endothermic heat consumption during heating; many other
works have used this definition [25,26,34–38]. Enthalpy of fusion data
corresponds to the heat released during crystallization of FAAE sam-
ples.
Experimental duplicates were conducted for all mixtures analyzed.
Random analytical duplicates showed an average CP standard deviation
of 0.2 K, and an average melting point standard deviation of 0.2 K.
milar, as in real biodiesel blends, ideal solution behavior and activity
coefficients of unity are assumed. This implies that:
1. All interactions between species in solution are identical
2. The change in heat capacity for species i between the liquid and
solid phase is negligible (ΔCi = 0 )
3. The first solid crystal at the mixture’s cloud point is composed of one
pure species, i, (zi = 1)
4. The fugacity of species i is equal to its liquid phase mole fraction
(since γiL = 1).
These assumptions greatly simplify the SLE equation, allowing the
mixture’s CP to be solved for explicitly by Eq. (2):
TCP,i
CP = RTCP ,i
1− ΔH ln(x i )
fus,i (2)
All species-dependent physicochemical parameters in Eq. (2) were
determined experimentally in this research, and were consistent with
literature ranges (see Supplementary Information).
4. Results and discussion
4.1. Pure FAEE cloud point, melting point, and enthalpy of fusion
To supplement the limited data on neat FAEE cold-flow properties
available in literature, data on FAEE cloud point, melting point, en-
thalpy of fusion, and qualitative co-crystallization behavior was col-
lected for pure and binary FAEE blends. For the sake of comparison,
cloud point data for both pure FAEE and FAME are presented in
Table 1.
The FAEE CPs in Table 1 are all statistically (p = 0.000) lower than
the FAME CPs with the same fatty acid moiety. This conclusion is
consistent with the literature consensus that FAEE inherently show
slightly favorable cold-flow properties in comparison to FAME for both
neat compounds [25,26,34–38] and biodiesel mixtures [24,40,41]. It
has been proposed that the additional –CH2 unit on the carbonyl end of
FAEE molecules decreases the amphiphilic nature of the molecule as a
whole, causing lower cloud and melting points [41]. Furthermore, the

244
P.A. Leggieri et al. Fuel 222 (2018) 243–249

Table 1
Cloud points and enthalpy of fusion (ΔHfus) for fatty acid alkyl esters with and without additives (triacetin, TA, or dimethyl azelate, DMAz).

Cloud Point (K) FAEE ΔHfus (kJ/mol)

Carbon-chain FAMEa FAEE FAEE + 15 wt% TA FAEE + 15 wt% DMAz This work Literatureb

C14:0 286.3 ± 0.2 276.8 ± 0.2 275.9 ± 0.1 273.6 ± 0.2 41.7 ± 0.2 39.8–45.9
C16:0 297.0 ± 0.1 289.6 ± 0.2 287.5 ± 0.2 286 ± 0.3 52.1 ± 1.2 53.8–55.8
C18:0 307.2 ± 0.7 301.4 ± 0.3 299.2 ± 0.5 298.2 ± 0.3 60.74 ± 1.0 59.7–61.6
C18:1 238.2 ± 0.0 231.6 ± 0.7 bdl 228.4 ± 0.5 44.8 ± 1.5 25.4–47.7
C18:2 bdl bdl bdl bdl bdl 24.39

bdl: below detection limit of DSC.

a
Ref. [4].
b
Refs. [25,28,35–38].

difference between FAEE and FAME CP and MP increases as molecular
length decreases; the added FAEE carbon has a more pronounced effect
on the CP and MP [40] of shorter-chain molecules since it constitutes a
greater fraction of the molecule. FAEE CP increases with carbon-chain
length, and decreases with unsaturation, with unsaturation having a
more pronounced effect (Table 1), which aligns with the trends seen for
FAME [39].
The enthalpy of fusion data (ΔHfus) collected in this work is gen-
erally consistent with literature values (Table 1). When collected via
DSC, ΔHfus is dependent on cooling rate due, in part, to polymorphism
in certain species [42]. This effect may account for the relatively broad
range of literature values (including this work) for each species. Neat
FAEE melting point (Tm) values collected in this work also generally
agreed with literature ranges (see Supplementary Information).
Polymorphism, evidenced by multiple DSC freezing peaks, was ob-
served in pure C16:0 and C18:0 (ΔHfus represent the sum of both peaks).
Other works have also noted polymorphism in these two species [37],
and that the transition from the α-form (observed first) to the β-form is
faster in C16:0 than in C18:0 [37]. It is potentially easier to observe this
behavior in smaller samples and at slower cooling rates, as this tran-
sition might be overwhelmed by larger liquid-to-α crystallization peaks
in large samples. As such, polymorphism was only observed during DSC
cooling of C16:0 when < 10 mg of sample used to determine pure
species melting point. Polymorphism in C18:0 was still observed when
additives were present, as seen in Fig. 1.
4.2. Binary FAEE mixture cloud point
Binary mixtures of FAEE were analyzed to study the effects of in-
teractions between saturated/saturated and saturated/unsaturated
species on CP (Table 2).
All binary mixtures of FAEE have CPs that are statistically lower
Fig. 1. DSC cooling traces of C18:0 with and without 15 wt% triacetin (TA) or dimethyl
azelate (DMAz). Polymorphism can be seen in all traces.
than those of the corresponding FAME mixtures (p = 0.000). As ex-
pected, when one mixture component is fixed, the CP of the mixture
increases as the alkyl-chain length of the second component increases,
but decreases if the second component is unsaturated. The latter effect
is more dramatic, and likely attributable to solubility of saturated FAEE
in unsaturated FAEE. Mixtures with C18:2 do not have markedly dif-
ferent CPs than those with C18:1, which suggests that the solubility of
saturated FAEE is comparable in mono- and diunsaturated FAEE. In
addition to differences in CP seen in Table 2, solubility effects of un-
saturated FAEE components cause differences in the crystallization
behaviors of FAEE.
4.2.1. Saturated/saturated binary FAEE mixtures
In saturated/saturated binary FAEE mixtures, all samples show a
single, irregularly shaped freezing peak or overlap of one large peak
and several smaller peaks (Fig. 2A). As such, all saturated-saturated
FAEE mixtures showed some degree of cocrystallization, or partial solid
phase miscibility, in which the longer molecule integrates the shorter
molecule or molecules into its crystal structure. Generally, cocrystalli-
zation is more likely to occur between species that are of similar
structure, shape, and pure crystal symmetry [43], and is common in
fatty ester mixtures [42]. The smaller peaks on the DSC freezing lines in
Fig. 2A likely represent either crystallization of the lower-freezing
component as a pure species after all of the higher-freezing component
has solidified in the cocrystal, or solid-phase transitions between mul-
tiple crystal structures composed of both species in the mixture.
The C14:0/C16:0 and C16:0/C18:0 peaks are considerably sharper
than that of the C14:0/C18:0 system, which is attributable to more
efficient packing of molecules of similar carbon-chain length into a
miscible solid structure. Molecules of similar length require less
bending of the longer molecule to accommodate the shorter molecule in
the cocrystal. This reduction in bending is thermodynamically favor-
able and related to faster crystallization.
To confirm that the cocrystallization peaks were not two in-
dependent but overlapping freezing peaks, DSC analysis was conducted
on all saturated/saturated mixtures at a slower cooling rate of 1 K/min
(Fig. 2A, bottom trace). At the slower cooling rate, single peaks with
distinctly unequal area were observed, suggesting that the mixtures do
exhibit cocrystallization. The enhanced definition of the smaller peak at
the slower cooling rate may be a result of either solid-phase transition
or heterogeneous crystallization being more pronounced with greater
temporal resolution on the DSC trace. It should be noted that the CP of
the slower cooling rate trial is about 3 K greater than that of the faster
rate trial due to crystal nucleation kinetics.
As seen in Table 2, all saturated/saturated FAEE mixtures had
smaller ΔHfus values than the weighted averages of their pure species
ΔHfus values. This is consistent with the tendency of cocrystallization to
decrease the amount of heat released in n-alkane petroleum systems
[44]. In addition, the systems in which components are separated by
two carbons (compared to four) showed greater reduction in mixture
ΔHfus values, further indicating greater thermodynamic favorability of

245
P.A. Leggieri et al. Fuel 222 (2018) 243–249

Table 2
Cloud points binary and enthalpy of fusion (ΔHfus) for binary mixtures of fatty acid alkyl esters with and without additives (triacetin, TA, or dimethyl azelate, DMAz).

Cloud Point (K) FAEE ΔHfus (J/g)

Carbon-chain FAME FAEE FAEE + 15 wt% TA FAEE + 15 wt% DMAz Experimental Pure Species weighted average

C14:0/C16:0* 285.6 ± 0.3a 282.1 ± 0.1 280.0 ± 0.1 278.9 ± 0.2 159.3 ± 0.2 181.0 ± 3.4
C14:0/C18:0* 292.6 ± 0.4a 289.3 ± 0.3 287.5 ± 0.2 286.4 ± 0.4 177.5 ± 0.3 183.8 ± 4.0
C14:0/C18:1 276 ± 0.3 268.9 ± 0.5 – – 152.1 ± 4.4 152.1 ± 5.6
C14:0/C18:2 276.9 ± 0.4 266.5 ± 0.3 – – bdl bdl
C16:0/C18:0* 298.9 ± 0.1a 295.5 ± 0.2 293.3 ± 0.2 291.8 ± 0.2 167.5 ± 0.1 187.3 ± 2.4
C16:0/C18:1 285.8 ± 0.3 278.6 ± 0.3 – – 151.97 ± 2.7 155.6 ± 4.0
C16:0/C18:2 287.6 ± 0.3 278.5 ± 0.3 – – bdl bdl
C18:0/C18:1 298.7 ± 0.2a 289.1 ± 0.2 289.6 ± 0.5 286.4 ± 0.4 172.9 ± 4.1 170.0 ± 6.4
C18:0/C18:2 297.6 ± 0.0a 289.7 ± 0.2 – – bdl bdl

“–”: No trials run with specified composition due to low expected CP.
bdl: Below detection limit of DSC.
a
Ref. [4].
* Samples displaying cocrystallization in FAME, FAEE, and FAEE plus additives.

cocrystallization between species of similar molecular length.

4.2.2. Saturated/unsaturated binary FAEE mixtures

As seen in Fig. 2B, cocrystallization was not observed in un-
saturated/unsaturated or saturated/unsaturated mixtures, even in the
C18:0/C18:1 mixture in which the two species have identical carbon-
chain lengths. Unsaturated FAAE act mostly as solvents during crys-
tallization and melting [35] of saturated species in multicomponent
mixtures due to the large differences in CP and melting point between
saturated and unsaturated components. The presence of such a solvent
can alter both crystallization mechanisms and the resulting crystal
structures. Thus, similar to cloud point, the cocrystallization behavior
of a FAEE mixture is more heavily influenced by degree of unsaturation
than carbon-chain length.
An example of unsaturated FAEE manipulating the crystallization of
saturated FAEE is seen in Fig. 2B. The presence of liquid C18:1 during
the crystallization of C18:0 prevents polymorphism at the specified
cooling rate. The shape of the DSC cooling traces of binary saturated/ Fig. 3. DSC traces of 50:50 C14:0/C16:0 FAEE and FAME mixtures with and without
unsaturated mixtures containing C18:1 and those containing C18:2 15 wt% triacetin (TA) or dimethyl azelate (DMAz). Dashed lines represent FAME systems;
solid lines represent FAEE systems.
were virtually identical (not shown).

4.3. Additive effects on binary FAEE and FAME cloud point and
crystallization
Two renewably-derived cold-flow improving additives, dimethyl
azelate (DMAz) and triacetin (TA) were analyzed for their effects on
FAEE cloud point (Table 2), crystallization behavior, and deviation
from ideal solution behavior.
Fig. 3 shows that the freezing peaks for binary methyl-C14:0/C16:0
and ethyl-C14:0/C16:0 are slightly different in shape, but they both
show signs of cocrystallization. Because FAME and FAEE show similar
CP and crystallization trends with respect to molecular structure, it is
reasonable to expect FAME and FAEE to respond similarly to cold-flow
improving additives.
4.3.1. Triacetin
Triacetin at 15 wt% decreased the CP of FAME and FAEE mixtures

Fig. 2. DSC cooling traces of binary FAEE saturated/saturated (A) and saturated/unsaturated (B) mixtures. Mixtures are 50:50 wt% with a cooling rate of 5 K/min (unless specified).

246
P.A. Leggieri et al.
by an average of ∼ 2 K. These results align with those obtained by Elias
et al. [4] that triacetin decreases the CP of pure and binary FAME
systems by 1–3 K at 10–20 wt%.
As seen in Fig. 3, the freezing peaks of FAEE and FAME blends with
triacetin are almost identical in shape to those of the binary mixtures
without triacetin. The melting point of triacetin is reported as 276 K
[45], which is within the freezing temperature range of the C14:0/
C16:0 systems. No distinct peak can be seen for triacetin crystallization,
so it is likely that triacetin crystallizes in the same temperature range as
the C14:0/C16:0 mixture. However, it is unlikely that triacetin co-
crystallizes with this mixture, because the molecular structure of tria-
cetin (C9H14O6) is highly disparate from that of FAEE. Because there is
no observable change in freezing peak shape between binary C14:0/
C16:0 mixtures with and without triacetin, it is unlikely that triacetin
alters the mechanism by which the mixture crystallizes, meaning that it
acts as a diluent in FAEE systems, as it has been shown to do in FAME
systems [4]. Fig. 1 shows that the C18:0 with 15 wt% triacetin trial still
showed polymorphism as expected, given the diluent behavior of tria-
cetin in these systems.
4.3.2. Dimethyl azelate (DMAz)
DMAz added at 15 wt% consistently reduced the CP of neat and
binary FAEE samples by about 3 K, shown in Table 2. In trials of both
FAME and FAEE mixtures of DMAz/C14:0/C16:0, CP was reduced on
average by about 4 K, suggesting that DMAz affects FAEE and FAME
similarly. DMAz appears to change the shape of the binary mixtures’
freezing peaks (Fig. 3) and thus is likely altering crystal formation. The
mixtures of FAEE and DMAz at 15 wt% show a small peak well below
the freezing temperature of the corresponding neat FAEE, which is
likely attributable to the phase transition of DMAz. Because DMAz re-
mains in the liquid phase throughout the entire crystallization process
of the C14:0/C16:0 mixture, it is likely that DMAz manipulates crys-
tallization mechanisms by solubilizing the FAEE.
Crystal modification from DMAz can also be seen in Fig. 1, where
the freezing peak of C18:0 is altered by DMAz but still shows poly-
morphism. DMAz and unsaturated FAEE both potentially affect satu-
rated FAEE crystallization via crystal modification, but unsaturated
FAEE prevent polymorphism in C18:0, as seen in Fig. 2B. Since DMAz is
more polar than FAAE due its additional carbonyl group, it is feasible
that crystal-modifying solvent effects of DMAz are different from those
of unsaturated FAEE.
4.4. Predictive thermodynamic cloud point modeling
The thermodynamic model outlined in Section 3 was used to predict
the CP of each mixture analyzed in this research, as well as those re-
ported in literature of biodiesel blends synthesized from various bio-
oils. Parity plots comparing predicted and experimental CPs (Fig. 4)
indicate that, generally, CP predictions based on the saturated com-
ponent with the longest carbon-chain most closely align with experi-
mental data, unless this component is present in miniscule concentra-
tion compared to another saturated species. For example, ethyl stearate
was used as species i in Eq. (2) to predict the CP of all experimental
mixtures containing it.
4.4.1. Thermodynamic modeling of pure/binary FAEE mixtures with and
without additives
All experimental CP values were predicted by the model within 2%
error, and a paired t-test indicated that the CPs of binary FAEE mixtures
with and without additives are accurately modeled by ideal solution
theory (p = 0.255). Mixtures that included C18:0 showed the greatest
deviation from the model. While all deviations from ideal solution
theory were minor, mixtures containing C18:0 would be expected to
exhibit the greatest deviations due to polymorphism and cocrystalli-
zation with other saturated FAEE. These polymorphisms represents
non-ideal solution behavior, as solid-phase transitions are unaccounted
247
Fuel 222 (2018) 243–249
for in the governing SLE fugacity equation.
Cocrystallization introduces deviation from ideal solution as it in-
validates assumption (3) outlined in Section 3, since the first crystal
formed in binary saturated/saturated mixtures is likely composed of
both species. Interestingly, cocrystallization in FAEE mixtures whose
constituent species are separated by 2 carbons causes the model to
slightly underpredict the CP. The increased thermodynamic favorability
of interactions between structurally similar species may explain the
onset of SLE at a higher temperature in these systems. Triacetin does
not introduce significant deviation of cocrystallizing mixtures from
ideal solution theory, as is expected of a diluent. On the other hand,
DMAz depresses the CP of these mixtures via crystal modification and
introduces no additional model error, perhaps due to a cancelling out of
two non-ideal behaviors. Underprediction of FAME CP in binary mix-
tures was also only observed in C14:0/C16:0 and C16:0/C18:0 mixtures
by Elias et al. [4], however it was observed in 75:25, not 50:50 mix-
tures.
The CPs of the three C14:0/C18:0 systems (binary, with triacetin,
with DMAz) are slightly overpredicted by the thermodynamic model.
The less thermodynamically favorable cocrystallization between species
of dissimilar molecular length may explain the onset of SLE at tem-
peratures slightly lower than ideal solution theory predicts. Elias et al.
[4] saw comparable overprediction of C14:0/C18:0 CP in FAME.
Despite small mixture-dependent deviations, the thermodynamic
SLE model with assumptions based on ideal solution theory can be used
to reasonably predict the CP of a two or three-component FAEE mixture
given its composition (mole fraction), and physicochemical data of its
constituent species obtainable via DSC.
4.4.2. Thermodynamic modeling of multicomponent biodiesel cloud point
The accuracy of the thermodynamic SLE CP model was tested on
methyl and ethyl esters derived from plant-based oils whose CPs are
reported in literature, and experiments conducted in this work using
SoyGold® FAME biodiesel. Predicted CPs for literature biodiesel fuels
were calculated using the reported fatty acid compositions of the bio-oil
used to synthesize the fuel.
FAME biodiesel blend CPs were more accurately predicted by the
model than FAEE CPs (Fig. 4D). The greatest deviation from ideal so-
lution in FAME CP was 2.6% in canola oil methyl ester, the reported CP
of which varies significantly in literature [9,20,24]. FAEE CPs deviate
from predicted CPs by up to 6.5%, and the greatest deviations were
those whose model CP predictions were based on C18:0. Given the
polymorphic/cocrystallizing behaviors of C18:0, these deviations from
solution ideality are expected. The most accurately predicted literature
FAEE biodiesel CP was that of palm oil, which was based on C16:0. The
variations seen for other biodiesel CP predictions may be due to small
(< 1%) amounts of 20:0 and 22:0, which may affect how materials
cocrystallize. Detailed information about model predictions of biodiesel
CP found in literature is available in Supplemental Information.
Large deviations of literature FAEE CP, but not FAME CP, values
from ideal solution theory suggest that multicomponent FAEE mixtures
featuring moderate concentrations of ethyl stearate may be quite
complex and exhibit non-ideal behaviors including cocrystallization or
heterogeneous crystallization as well as solid-phase transitions.
Analysis of SoyGold® showed that the model error was larger than a
number of the other biodiesels found in literature. This finding is likely
due to the large amount of cocrystallization that occurs between C16:0
and C18:0, which would deviate from the assumptions of ideal solution
theory. The addition of both dimethyl azelate and triacetin did not
greatly influence the percent error, indicating that these additives did
not strongly influence the solution thermodynamics of the biofuel.
The thermodynamic model presented here has potential applica-
tions in designing FAEE or FAME based biodiesels with desirable cloud
points. Lipid profiles may be improved upstream of thermochemical or
biochemical conversion steps in genetic modification of biofeedstocks,
particularly microalgae, to yield optimal cold flow properties of the
P.A. Leggieri et al.
Fig. 4. Parity plots of CP predicted via Eq. (2) for binary FAEE mixtures with no additives (A); 15 wt% triacetin, TA (B); 15 wt% dimethyl azelate, DMAz (C); FAME and FAEE of vegetable
oil derived biodiesels (D, [9,20,24,33]). Solid line: y = x. Shaded points in (A-C) represent mixtures containing C18:0.
resulting biodiesel [46].
5. Conclusion
The commercial feasibility of biodiesel production and adoption is
contingent on improving biodiesel’s technical shortfalls, including its
poor cold-flow properties. FAEE exhibit favorable cold-flow properties
compared to FAME and display a similar response to renewably-derived
cold-flow improving additives. Furthermore, the CP of FAAE blends can
be reasonably predicted via thermodynamic model. The data and ana-
lysis provided in this work on FAEE constituent species CPs, both in
isolation and in multi-component mixtures, assists in the engineering of
biodiesel blends with desirable FAEE profiles. The cold-flow analysis
may be used to informed techniques such as fractionation and feedstock
modification that can allow for the development of improved and/or
tailored biodiesel products.
Acknowledgements
The authors thank Professor Kenneth Haug and the Department of
Chemistry at Lafayette College for helping to provide access to the DSC.
Stephanie McCartney and Andrew Morris contributed to some sample
testing. The authors gratefully acknowledge the William T. Morris
Foundation and the Lafayette College EXCEL Scholar Foundation for
helping to fund this research.
Funding
This work was supported by the William T. Morris Foundation and
the Lafayette College EXCEL Scholar Program.
248
Fuel 222 (2018) 243–249
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.fuel.2018.02.100.
References
[1] Mohanan A, Bouzidi L, Li SJ, Narine SS. Mitigating crystallization of saturated
fames in biodiesel: 1. Lowering crystallization temperatures via addition of me-
tathesized soybean oil. Energy 2016;96:335–45.
[2] Basha SA, Gopal KR. A review of the effects of catalyst and additive on biodiesel
production, performance, combustion and emission characteristics. Renew Sustain
Energy Rev 2012;16(1):711–7.
[3] Knothe G. Production and properties of biodiesel from algal oils. In: Borowitzka MA,
Moheimani NR, editors. Algae for biofuels and energy. Netherlands. Dordrecht:
Springer; 2013. p. 207–21.
[4] Elias RC, Senra M, Soh L. Cold flow properties of fatty acid methyl ester blends with
and without triacetin. Energy Fuels 2016;30(9):7400–9.
[5] Meher LC, Vidya Sagar D, Naik SN. Technical aspects of biodiesel production by
transesterification—a review. Renew Sustain Energy Rev 2006;10(3):248–68.
[6] Knothe G, Razon LF. Biodiesel fuels. Prog Energy Combust Sci 2017;58:36–59.
[7] Sierra-Cantor JF, Guerrero-Fajardo CA. Methods for improving the cold flow
properties of biodiesel with high saturated fatty acids content: A review. Renew
Sustain Energy Rev 2017;72:774–90.
[8] Knothe G. Dependence of biodiesel fuel properties on the structure of fatty acid
alkyl esters. Fuel Process Technol 2005;86(10):1059–70.
[9] Lang X, Dalai AK, Bakhshi NN, Reaney MJ, Hertz PB. Preparation and character-
ization of bio-diesels from various bio-oils. Bioresour Technol 2001;80(1):53–62.
[10] Jones JC. On the use of ethanol in the processing of biodiesel fuels. Fuel
2010;89(5):1183.
[11] AFDC; 2017. https://afdc.energy.gov/data/.
[12] Clark SJ, Wagner L, Schrock MD, Piennaar PG. Methyl and ethyl soybean esters as
renewable fuels for diesel engines. JAOCS, J Am Oil Chem Soc
1983;61(10):1632–8.
[13] Marjanović AV, Stamenković OS, Todorović ZB, Lazić ML, Veljković VB. Kinetics of
the base-catalyzed sunflower oil ethanolysis. Fuel 2010;89(3):665–71.
[14] Lim S, Lee KT. Parallel production of biodiesel and bioethanol in palm-oil-based
biorefineries: Life cycle assessment on the energy and greenhouse gases emissions.
Biofuels, Bioprod Biorefin 2011;5(2):132–50.
P.A. Leggieri et al.
[15] Nhien LC, Long NVD, Lee M. Novel heatintegrated and intensified biorefinery
process for cellulosic ethanol production from lignocellulosic biomass. Energy
Convers Manage 2017;141:367–77.
[16] Bhale PV, Deshpande NV, Thombre SB. Improving the low temperature properties
of biodiesel fuel. Renewable Energy 2009;34(3):794–800.
[17] Sajjadi B, Raman AAA, Arandiyan H. A comprehensive review on properties of
edible and non-edible vegetable oil-based biodiesel: composition, specifications and
prediction models. Renew Sustain Energy Rev 2016;63:62–92.
[18] Sorate KA, Bhale PV. Biodiesel properties and automotive system compatibility is-
sues. Renew Sustain Energy Rev 2015;41:777–98.
[19] Harrow G. E85 and biodiesel deployment. National Renewable Energy Laboratory;
2007.
[20] Abe M, Hirata S, Komatsu H, Yamagiwa K, Tajima H. Thermodynamic selection of
effective additives to improve the cloud point of biodiesel fuels. Fuel
2016;171:94–100.
[21] Chiu C-W, Schumacher LG, Suppes GJ. Impact of cold flow improvers on soybean
biodiesel blend. Biomass Bioenergy 2004;27(5):485–91.
[22] Alleman TL, McCormick RL, Christensen ED, Fioroni G, Moriarty K, Yanowitz J.
Biodiesel Handling and Use Guide (Fifth Edition). National Renewable Energy
Laboratory (NREL), Golden, CO (United States); 2016.
[23] Lv P, Cheng Y, Yang L, Yuan Z, Li H, Luo W. Improving the low temperature flow
properties of palm oil biodiesel: Addition of cold flow improver. Fuel Process
Technol 2013;110:61–4.
[24] Smith PC, Ngothai Y, Dzuy Nguyen Q, O'Neill BK. Improving the low-temperature
properties of biodiesel: Methods and consequences. Renewable Energy
2010;35(6):1145–51.
[25] Boros L, Batista MLS, Vaz RV, Figueiredo BR, Fernandes VFS, Costa MC, et al.
Crystallization behavior of mixtures of fatty acid ethyl esters with ethyl stearate.
Energy Fuels 2009;23(9):4625–9.
[26] Boros LAD, Batista MLS, Coutinho JAP, Krahenbuhl MA, Meirelles AJA, Costa MC.
Binary mixtures of fatty acid ethyl esters: Solid-liquid equilibrium. Fluid Phase
Equilib 2016;427:1–8.
[27] Costa MC, Boros LAD, Batista MLS, Coutinho JAP, Krahenbuhl MA, Meirelles AJA.
Phase diagrams of mixtures of ethyl palmitate with fatty acid ethyl esters. Fuel
2012;91(1):177–81.
[28] Lopes JCA, Boros L, Krahenbuhl MA, Meirelles AJA, Daridon JL, Pauly J, et al.
Prediction of cloud points of biodiesel. Energy Fuels 2008;22(2):747–52.
[29] Casas A, Ramos MJ, Perez A. Kinetics of chemical interesterification of sunflower oil
with methyl acetate for biodiesel and triacetin production. Chem Eng J
2011;171(3):1324–32.
[30] Komintarachat C, Sawangkeaw R, Ngamprasertsith S. Continuous production of
palm biofuel under supercritical ethyl acetate. Energy Convers Manage
2015;93:332–8.
249
Fuel 222 (2018) 243–249
[31] Shafiei A, Rastegari H, Ghaziaskar HS, Yalpani M. Glycerol transesterification with
ethyl acetate to synthesize acetins using ethyl acetate as reactant and entrainer.
Biofuel Res J-Brj 2017;4(1):565–70.
[32] Sinsakulroj N, Pengprecha S. Synthesis of pour point depressant from sunflower oil.
In: International conference on chemical processes and environmental issues
(ICCEEI 2012); 2012.
[33] Joshi H, Moser BR, Toler J, Smith WF, Walker T. Ethyl levulinate: a potential bio-
based diluent for biodiesel which improves cold flow properties. Biomass Bioenergy
2011;35(7):3262–6.
[34] Knothe G, Dunn RO. A comprehensive evaluation of the melting points of fatty acids
and esters determined by differential scanning calorimetry. JAOCS, J Am Oil Chem
Soc 2009;86(9):843–56.
[35] Robustillo MD, Barbosa DF, Meirelles AJdA, Filho PdAP. Solid-liquid equilibrium in
ternary mixtures of ethyl oleate, ethyl laurate and ethyl palmitate. Fluid Phase
Equilib 2013;339:58–66.
[36] Robustillo MD, Barbosa DF, Meirelles AJdA, PessOa Filho PdA. Solid-liquid equi-
librium in ternary mixtures of ethyl laurate, ethyl palmitate and ethyl myristate.
Fluid Phase Equilib 2014;361:188–99.
[37] Robustillo MD, Barbosa DF, Meirelles AJdA, PessOa Filho PdA. Solid-liquid equi-
librium in ternary mixtures of ethyl laurate, ethyl palmitate and ethyl stearate.
Fluid Phase Equilib 2013;358:272–81.
[38] Robustillo MD, Barbosa DF, Meirelles AJdA, Pessoa Filho PdA. Solid-liquid equili-
brium of binary and ternary mixtures containing ethyl oleate, ethyl myristate and
ethyl stearate. Fluid Phase Equilib 2014;370:85–94.
[39] Imahara H, Minami E, Saka S. Thermodynamic study on cloud point of biodiesel
with its fatty acid composition. Fuel 2006;85(12–13):1666–70.
[40] Knothe G. “Designer” biodiesel: optimizing fatty ester (composition to improve fuel
properties. Energy Fuels 2008;22(2):1358–64.
[41] Knothe G, Van Gerpen J. The biodiesel handbook. Taylor & Francis; 2005.
[42] Maximo GJ, Costa MC, Coutinho JAP, Meirelles AJA. Trends and demands in the
solid-liquid equilibrium of lipidic mixtures. RSC Adv 2014;4(60):31840–50.
[43] Senra M, Scholand T, Maxey C, Fogler HS. Role of polydispersity and cocrystalli-
zation on the gelation of long-chained n-alkanes in solution. Energy Fuels
2009;23(12):5947–57.
[44] Senra M, Panacharoensawad E, Kraiwattanawong K, Singh P, Fogler HS. Role of n-
alkane polydispersity on the crystallization of n-alkanes from solution. Energy Fuels
2008;22(1):545–55.
[45] Haynes WM, Bruno TJ, Lide DR. CRC handbook of chemistry and physics. Boca
Raton, FL: CRC Press; 2015. p. 3–514.
[46] Tabatabaei M, Tohidfar M, Jouzani GS, Safarnejad M, Pazouki M. Biodiesel pro-
duction from genetically engineered microalgae: future of bioenergy in Iran. Renew
Sustain Energy Rev 2011;15(4):1918–27.