lipid thermodynamics: new perspectives on phase …
TRANSCRIPT
i
GUILHERME JOSÉ MAXIMO
LIPID THERMODYNAMICS: NEW PERSPECTIVES ON PHASE STUDIES FOR
APPLICATIONS IN ENGINEERING
TERMODINÂMICA DE LIPÍDIOS: NOVAS PERSPECTIVAS EM ESTUDOS DE
FASES PARA APLICAÇÕES EM ENGENHARIA
CAMPINAS
2014
ii
iii
UNIVERSIDADE ESTADUAL DE CAMPINAS
Faculdade de Engenharia de Alimentos
GUILHERME JOSÉ MAXIMO
LIPID THERMODYNAMICS: NEW PERSPECTIVES ON PHASE STUDIES FOR
APPLICATIONS IN ENGINEERING
TERMODINÂMICA DE LIPÍDIOS: NOVAS PERSPECTIVAS EM ESTUDOS DE
FASES PARA APLICAÇÕES EM ENGENHARIA
Thesis presented to the Faculty of Food
Engineering of the University of Campinas in
partial fulfillment of the requirements for the
degree of Doctor in Food Engineering.
Tese apresentada à Faculdade de Engenharia
de Alimentos da Universidade Estadual de
Campinas como parte dos requisitos exigidos
para a obtenção do título de Doutor em
Engenharia de Alimentos.
Supervisor/Orientador: ANTONIO JOSÉ DE ALMEIDA MEIRELLES
Co-supervisor/Co-orientador: MARIANA CONCEIÇÃO DA COSTA
Co-supervisor/Co-orientador: JOÃO ARAÚJO PEREIRA COUTINHO
ESTE EXEMPLAR CORRESPONDE À VERSÃO
FINAL DA TESE DEFENDIDA PELO ALUNO
GUILHERME JOSÉ MAXIMO E ORIENTADA
PELO PROF. DR. ANTONIO JOSÉ DE ALMEIDA
MEIRELLES
____________________________________________
CAMPINAS
2014
!"#$%"$&$'()*+,!"$
-.!/0*1!2$20%31&$24$'%20%5$67!.$1
8!9'!(&0"$%2$% $"4'2$20%20%3.)0.#$*!$%20%:'!60.&(1
;+*"!$%<0)!.$%=$*90'!.!%>0/!''$.(%?%5<8%@ABCDE
%% %
;$F!6(G%=4!'#0*60%H(1IG%JK@L? %
% ;DM! ;$FN!7!2%�*6(2O.$6!"1%P%.0Q%70*170"&!/01%(.%7#$10%1&42!01%,(*%$77'!"$&!(.1%!.
0.)!.00*!.)%A%=4!'#0*60%H(1I%;$F!6(R%S%5$67!.$1G%>T%P%U1R.RVG%LWJDR
%
% % %
% % ;$FX*!0.&$2(*P%:.&(.!(%H(1I%20%:'60!2$%;0!*0''01R
% % ;$F5((*!0.&$2(*01P%;$*!$.$%5(."0!YZ !"#!5(1&$ ! "#$#%:*$[\(%"#$#%$& '()*%+,(-
%%%%%%%%%%%%%%%%% F]010%^2(4&(*$2(_%S%-.!/0*1!2$20%31&$24$'%20%5$67!.$1G% $"4'2$20%20
3.)0.#$*!$%20%:'!60.&(1R
%
% % % %
% % ;$FJR%3`4!'a9*!(%1b'!2(?'a`4!2(R%LR%:')(*!&6(R%BR%>('4Yc01%1b'!2$1R%DR%Na`4!2(1
!d.!"(1R%MR%5*!1&$'%'a`4!2(R%eR%;0!*0''01G%:.&(.!(%H(1I%20%:'60!2$R%eeR%5(1&$G%;$*!$.$
5(."0!YZ(!"#R%eeeR% !"#$%&!'(H(Z(%:*$[\(%T0*0!*$R%efR%-.!/0*1!2$20%./*&0)&1 0#
5$67!.$1R% $"4'2$20%20%3.)0.#$*!$%20%:'!60.&(1R%fR%]a&4'(R
%
e.,(*6$Yc01%7$*$%8!9'!(&0"$%g!)!&$'
Título em outro idioma:%]0*6(2!.h6!"$%20%'!7a2!(1%P%.(/$1%70*170"&!/$1%06%01&42(1%20%,$1017$*$%$7'!"$Yc01%06%0.)0.#$*!$
Palavras-chave em inglês:>('!2?'!`4!2%0`4!'!9*!46
:')(*!
>('4&!(.1G%>('!2
e(.!"%'!`4!21
N!`4!2%"*O1&$'
Área de concentração:%3.)0.#$*!$%20%:'!60.&(1Titulação:%g(4&(*%06%3.)0.#$*!$%20%:'!60.&(1Banca examinadora::.&(.!(%H(1I%20%:'60!2$%;0!*0''01%UX*!0.&$2(*V
H0$.?N4"%g$*!2(.
>!6Z(%T02*(%20%:'60!2$%T!.#(
HI*d60%T$4'O
!"#$%&'()*($+,--,.$/,)0(-!
Data de defesa:%JE?WB?LWJDPrograma de Pós-Graduação:%3.)0.#$*!$%20%:'!60.&(1
!/
v
EXAMINATION COMITTEE / BANCA EXAMINADORA
________________________________________
Prof. Dr. Antonio José de Almeida Meirelles
Supervisor / Orientador
University of Campinas, Brazil
________________________________________
Prof. Dr. Jean-Luc Daridon
Titular member / Membro Titular
University of Pau and the Adour region, France
________________________________________
Prof. Dr. Simão Pedro de Almeida Pinho
Titular member / Membro Titular
University Salvador, Brazil
________________________________________
Prof. Dr. Jérôme Pauly
Titular member / Membro Titular
University of Pau and the Adour region, France
________________________________________
Prof. Dr. José Vicente Hallak Dangelo
Titular member / Membro titular
University of Campinas, Brazil
________________________________________
Prof. Dr. Rosiane Lopes da Cunha
Substitute member / Membro suplente
University of Campinas, Brazil
________________________________________
Prof. Dr. Martin Aznar
Substitute member / Membro suplente
University of Campinas, Brazil
________________________________________
Prof. Dr. Jorge Andrey Wilhelms Gut
Substitute member / Membro suplente
University of São Paulo, Brazil
vi
vii
ABSTRACT
The phase transition phenomena of lipidic systems have long since been evaluated by
several works in literature for developing chocolate, butters, dressings, spreads, cosmetic
creams, medicines or biofuels as well as for optimizing processes such as extraction,
refining, fractionation, crystallization or energy production. The aim of such works has
been to answer how the products’ composition can affect their physicochemical
characteristics especially that related to the melting processes. In fact, changes in
temperature highly impact the crystalline structure of fatty systems’ solid phase and,
consequently, in the sensorial and rheological properties of the products. These changes led
to so many thermodynamic behaviors that the determination of the solid-liquid equilibrium
of these systems can configure a particular challenge. However, the greater the complexity
of the system the lower the understanding of its behavior. In other words, despite the
number of works in literature involved in the investigation of the melting phenomena of
lipidic systems, there is still a lack of experimental data and modeling approaches for their
understanding. In this context, this work was conducted with two main goals. The first was
focused on the measurement and comprehension of the solid-liquid equilibrium phase
diagrams of lipidic binary systems of industrial interest. The second was aimed at the
development of theoretical alternatives to improve the phase diagram description based on
classical thermodynamic approaches. Thus, eleven binary systems composed by
triacylglycerols, fatty alcohols and fatty acids were evaluated. Such mixtures are potential
structuring, organogelating and energy storing agents for food, pharmaceutical and
materials industry. The solid-liquid phase diagrams of these mixtures presented distinct
behaviors depending on the formation of immiscible or miscible solid phases and the non-
viii
ideality of the system. Also, four systems built through a Brønsted acid-base reaction
between fatty acids and ethanolamines were also evaluated. In this case, the formation of
protic ionic liquid crystals with high self-assembling ability and marked non-Newtonian
behavior are promising for pharmaceutical and chemical applications. The problem
imposed by the partial miscibility of the solid phase in the construction of the phase
diagrams was overtaken by the implementation of an algorithm based on the resolution of a
non-linear equations system built by the solid-liquid equilibrium fundamental equations.
The “Crystal-T” algorithm was aimed at the determination of the temperature at which the
first and last crystal melts during the heating process. For this, the non-ideality of both
liquid and solid phases was evaluated using excess Gibbs energy equations, including the
group-contribution UNIFAC model, for the calculation of the activity coefficients. Taking
into account the growing increase of the world production and consumption of fat and oils,
this work, from industrial and academic emerging demands, contributed to overtake some
barriers on the understanding of the solid-liquid phase equilibrium of lipidic mixtures for
products and process engineering.
KEYWORDS: Solid-liquid equilibrium, Algorithm, Solid solution, Ionic liquids, Liquid
crystal.
ix
RESUMO
Para o desenvolvimento de chocolates, manteigas, molhos para salada, cremes cosméticos,
medicamentos ou biocombustíveis, assim como na otimização de processos de extração,
refino, fracionamento, cristalização e produção de energia, os fenômenos de transição de
fases dos sistemas lipídicos são temas, há muito tempo, de diversos trabalhos na literatura.
O objetivo desses trabalhos tem sido avaliar como a composição dos produtos altera as suas
propriedades físico-químicas e em especial aquelas relacionadas aos processos de fusão. De
fato, alterações na temperatura exercem um grande impacto na estrutura cristalina da fase
sólida dos sistemas graxos e, consequentemente, nas propriedades sensoriais e reológicas
dos produtos. Essas alterações produzem comportamentos termodinâmicos tão variados que
a determinação do equilíbrio de fases sólido-líquido desses sistemas representa um grande
desafio. Não obstante, quanto maior a complexidade do sistema, menor é a compreensão do
seu comportamento. Ou seja, apesar do grande número de trabalhos presentes na literatura
envolvidos na investigação dos fenômenos de fusão de sistemas lipídicos, novos dados
experimentais e abordagens teóricas para a modelagem dos diagramas são necessários para
sua compreensão. Neste contexto, este trabalho teve dois objetivos principais. O primeiro
relacionado à determinação e análise de diagramas de fases sólido-líquido de sistemas
lipídicos binários de interesse para a indústria. O segundo foi desenvolver alternativas
teóricas para aprimorar a representação dos diagramas de fases baseado em abordagens
termodinâmicas clássicas. Portanto, onze sistemas binários compostos por triacilgliceróis,
ácidos graxos e álcoois graxos foram avaliados. Esses sistemas são potenciais agentes de
estruturação, formação de organogéis, produção e armazenamento de energia na indústria
de alimentos, farmacêutica e de materiais. Os diagramas de equilíbrio sólido-líquido dessas
x
misturas apresentaram comportamentos distintos, dependentes da formação de fases sólidas
miscíveis ou imiscíveis e da não-idealidade do sistema. Além disso, foram estudados quatro
sistemas formados a partir da reação ácido-base de Brønsted entre ácidos graxos e
etanolaminas. Neste caso, a formação de líquidos iônicos próticos cristalinos com grande
habilidade para auto-organização e comportamento não-Newtoniano singular podem atuar
como auxiliares em diversas aplicações químicas e farmacêuticas. O problema imposto pela
miscibilidade da fase sólida na construção dos diagramas de fases foi superado pela
implementação de um algoritmo para a resolução de um sistema de equações não-lineares
baseado nas equações fundamentais do equilíbrio sólido-líquido. O objetivo do algoritmo
“Crystal-T” foi determinar a temperatura em que o primeiro e o último cristal se fundem
durante o aquecimento do sistema. Para isso, a não-idealidade de ambas as fases líquida e
sólida foi avaliada utilizando equações baseadas na energia de Gibbs em excesso, incluindo
o método de contribuição de grupos UNIFAC, para o cálculo dos coeficientes de atividade.
Considerando o aumento da produção mundial e do consumo de óleos e gorduras, este
trabalho, a partir de demandas emergentes da indústria e da pesquisa científica, contribuiu
na superação de alguns obstáculos relacionados à compreensão do equilíbrio de fases
sólido-líquido de sistemas lipídicos para a engenharia de produtos e processos.
PALAVRAS-CHAVE: Equilíbrio sólido-líquido, Algoritmo, Solução sólida, Líquidos
iônicos, Cristal líquido.
xi
SUMMARY
INTRODUCTION 1
OBJECTIVES 3
INTRODUÇÃO 7
OBJETIVOS 9
Chapter 1. TRENDS AND DEMANDS IN SOLID-LIQUID EQUILIBRIUM OF
LIPIDIC MIXTURES
13
1. Introduction 14
2. What do fat and oils provide? 16
3. The solid-liquid equilibrium of fatty systems 18
3.1. Evolution 19
4. Brief overview on the techniques for measuring SLE data 26
5. But, in fact, what is literature looking for? 28
6. Demands on modeling 30
7. Remarks 35
Chapter 2. SOLID-LIQUID EQUILIBRIUM OF TRIOLEIN WITH FATTY
ALCOHOLS
43
1. INTRODUCTION 44
1.1. Thermodynamic framework 46
2. MATERIALS AND METHODS 49
2.1. Differential Scanning Calorimetry (DSC) 50
2.2. Optical Microscopy 51
2.3. Modeling Procedure 51
3. RESULTS AND DISCUSSION 53
3.1. Performance of the Algorithm 53
3.2. Experimental and Predicted Solid-Liquid Equilibrium Data 53
4. CONCLUSIONS 64
Chapter 3. ON THE SOLID-LIQUID EQUILIBRIUM OF BINARY MIXTURES OF
FATTY ALCOHOLS AND FATTY ACIDS
69
1. INTRODUCTION 71
2. MATERIALS AND METHODS 72
2.1. Materials 72
2.2. Apparatus and procedure 73
2.2.1. Differential scanning calorimetry 73
2.2.2. Optical microscopy 75
2.2.3. X-ray diffraction 76
2.3. Thermodynamic modeling 76
xii
3. RESULTS AND DISCUSSION 79
3.1. Experimental SLE data 79
3.2. Eutectic point and immiscibility assumption 86
3.3. Evaluation of the solid-phase 89
3.4. Thermodynamic modeling of SLE data 91
4. CONCLUSIONS 97
SUPPLEMENTARY DATA 101
Chapter 4. LIPIDIC PROTIC IONIC LIQUID CRYSTALS 105
1. INTRODUCTION 107
2. EXPERIMENTAL 109
2.1. Materials 109
2.2. Differential Scanning Calorimetry (DSC) 110
2.3. Light-Polarized Optical Microscopy (POM) 111
2.4. Density, Viscosity and Critical Micellar Concentration (CMC) Measurements 111
2.5. Rheological Measurements. 112
3. RESULTS AND DISCUSSION 112
3.1. Phase Behavior 112
3.2. Self-assembling of the PILs 119
3.3. Rheological Characterization of the PILs and Their Binary Mixtures 121
4. CONCLUSIONS 130
SUPPORTING INFORMATION 137
Chapter 5. THE CRYSTAL - T ALGORITHM 143
1. INTRODUCTION 145
2. THEORY 147
2.1. Fundamentals 147
2.2. SLE algorithms 151
2.3. The Crystal-T algorithm 153
2.3.1 Adjustment of the i equations parameters 156
2.3.2. Experimental considerations 159
3. RESULTS AND DISCUSSION 161
3.1. Experimental data analysis 161
3.2. SLE modeling using the Crystal-T algorithm 168
4. CONCLUSIONS 177
SUPPORTING INFORMATION 183
CONCLUSIONS 189
CONCLUSÃO 193
xiii
I dedicate this thesis to my family and friends
Dedico esta tese à minha família e aos meus amigos
xiv
xv
ACKOWLEDGMENTS / AGRADECIMENTOS
Agradeço ao Prof. Dr. Antonio José de Almeida Meirelles, pela orientação deste
trabalho de tese, resultado de um processo crescente de compreensão do conteúdo e de
amadurecimento acadêmico. À Prof.a Dr.
a Mariana Conceição Costa, da Faculdade de
Ciências Aplicadas/UNICAMP, pela coorientação e apoio durante todo o processo de
desenvolvimento desse trabalho. Ao Prof. Dr. João Araújo Pereira Coutinho, da
Universidade de Aveiro, pela coorientação, durante minha estadia em Portugal, e
significativa contribuição a esse trabalho de tese. À Universidade Estadual de Campinas
(UNICAMP), à Faculdade de Engenharia de Alimentos (FEA) e ao Departamento de
Engenharia de Alimentos (DEA), na qualidade de instituições-sede dessa pesquisa. Ao
CNPq pela concessão da bolsa de doutorado e à CAPES pela concessão da bolsa de
doutorado sanduíche no exterior. À Prof.a Dr.
a Rosiane Lopes da Cunha, ao Prof. Dr. Celso
Costa Lopes e ao Prof. Dr. Eduardo Augusto Caldas Batista do DEA/FEA/UNICAMP, pelo
constante apoio acadêmico. À Prof.a Dr.
a Sandra Augusta Santos do Instituto de
Matemática, Estatística e Computação Científica (IMECC)/UNICAMP, pelas sugestões na
elaboração do algoritmo. A todos os companheiros do laboratório de Extração,
Termodinâmica Aplicada e Equilíbrio (EXTRAE) do DEA/FEA/UNICAMP, pela partilha
diária de conteúdo acadêmico, sugestões e experiências. A todos do laboratório PATh, da
Universidade de Aveiro, Portugal, pela recepção, apoio e companheirismo. Agradeço à
minha família e aos meus amigos por contribuírem com o intangível.
xvi
xvii
LIST OF FIGURES
INTRODUCTION
Figure 1. Action fields of the solid-liquid equilibrium studies on the vegetable oils
processing industries.
2
INTRODUÇÃO
Figura 1. Campos de ação dos estudos do equilíbrio sólido-líquido nas indústrias de
processamento de óleos vegetais.
8
Chapter 1. TRENDS AND DEMANDS IN SOLID-LIQUID EQUILIBRIUM OF
LIPIDIC MIXTURES
Figure 1. Most common phase diagrams of binary fatty mixtures with the following
nomenclature: mole fraction of the component a in the liquid (xa) and solid (za) phases,
respectively, liquid phase (l), solid phase with pure component a (Ca) or/and b (C
b), solid
phase comprising both compounds (s), solid phase with new structure formed by
peritectic transition (Cn).
20
Figure 2. Tammann plot of the eutectic reaction: enthalpy of the eutectic reaction as a
function of the concentration of the mixture (full line); mole fraction of the component a
at the eutectic point in the liquid phase xeut (dashed line), and in the solid phases (zeut) rich
in the component a (Ca) or b (C
b); represented in gray, solid solution regions rich in the
component a or b.
23
Figure 3. Sketch of the SLE phase diagram observed in mixtures of fatty acids (as
presented by Costa, et al. 37
) for some fatty acid + fatty acid binary systems. Ca
and
Cb are the solid solutions rich in one of the components of the mixture a or b; Cn is the
solid phase formed by the peritectic reaction; Cha and Ch
b are the solid phases formed
after the metatectic reaction; l is the liquid phase.
25
Chapter 2. SOLID-LIQUID EQUILIBRIUM OF TRIOLEIN WITH FATTY
ALCOHOLS
Figure 1. Deviation (this work) versus deviation from the literature (data from Table 2):
(○) Carareto et al., (2011), (×) Boros (2005), (+) Huang and Chen (2000) and (Δ) Jakob
et al. (1995). Dashed line represents the straight diagonal line and the continuous line is
the curve fitted with the data (y = 1.0123x – 0.017, r2 = 0.9966). Number of observations
= 31.
55
Figure 2. Solid-liquid phase diagram (temperature in Kelvin and mole fraction) and
thermograms for the triolein (1) + 1-hexadecanol (2) system. Symbols for phase
57
xviii
diagrams: (○) experimental melting point, (■) transition near triolein melting point, (×)
solid-solid triolein beta-prime transition and data modeled by (- - -) the three-suffix
Margules equation, (—) the UNIFAC-Dortmund model and (- -) the ideal assumption.
In detail, magnification of the thermograms close to the melting temperature for x1 > 0.7
and close to the triolein alpha polymorphic form temperature transition for x1 = 1.0.
Figure 3. Solid-liquid phase diagram (temperature in Kelvin and mole fraction) and
thermograms for the triolein (1) + 1-octadecanol (2) system. Symbols for phase diagrams:
(○) experimental melting point, (■) transition near triolein melting point, (×) solid-solid
triolein beta-prime transition and data modeled by (- - -) the three-suffix Margules
equation (—), the UNIFAC-Dortmund model and (- -) the ideal assumption. In detail,
magnification of the thermograms close to the melting temperature for x1 > 0.7.
58
Figure 4. Tammann plots of (a) triolein (1) + 1-hexadecanol (2) and (b) triolein (1) + 1-
octadecanol (2). (—) Linear regression for data on (■) transition close to the triolein
melting point, (○) pure triolein transition and (- - -) hypothetical plot with a eutectic
point.
60
Figure 5. Optical micrographs of samples of triolein (1) + 1-octadecanol (2) with
magnifications of 10 ((a) to (f)) and 40 (g) at 298.15 K.
62
Chapter 3. ON THE SOLID-LIQUID EQUILIBRIUM OF BINARY MIXTURES OF
FATTY ALCOHOLS AND FATTY ACIDS
Figure 1. Solid-liquid phase diagram and thermograms for the system 1-tetradecanol (1)
+ dodecanoic acid (2): experimental melting temperatures (○); transition of the pure
component (); eutectic transition (■); other transitions (+), and modeled data by three-
suffix Margules equation (); UNIFAC-Dortmund model ( ) and ideal assumption
(—). Thermogram in detail shows transition at x1 = 0.0000.
80
Figure 2. Solid-liquid phase diagram and thermograms for the system 1-hexadecanol (1)
+ tetradecanoic acid (2): experimental melting temperatures (○); transition of the pure
component (); eutectic transition (■); other transitions (+), and modeled data by three-
suffix Margules equation (); UNIFAC-Dortmund model ( ) and ideal assumption
(—). Thermograms in detail show transitions at x1 = (0.0000 and 0.6004).
81
Figure 3. Solid-liquid phase diagram and thermograms for the system 1-octadecanol (1)
+ hexadecanoic acid (2): experimental melting temperatures (○); transition of the pure
component (); eutectic transition (■); other transitions (+), and modeled data by three-
suffix Margules equation (); UNIFAC-Dortmund model ( ) and ideal assumption
82
xix
(—). Thermogram in detail shows transition at x1 = 0.5978.
Figure 4. Effect of the thermal annealing based treatment on thermograms of 1-
hexadecanol (1) + tetradecanoic acid (2) at (A) x1 = 0.8994 and at (B) x1 = 0.3042: I,
heating curve without the thermal treatment and; II, after thermal treatment. The
metastable transition is shown by the black arrow.
85
Figure 5. Tammann plots of the eutectic transition (■) for: (A) 1-tetradecanol +
dodecanoic acid; (B) 1-hexadecanol + tetradecanoic acid and (C) 1-octadecanol +
hexadecanoic acid as well as linear regression (- - -) of the data.
87
Figure 6. Optical micrographs of samples of 1-hexadecanol (1) + tetradecanoic acid (2)
at x1 = 0.3981 and (A) 311.75 K; (B) 313.65 K; (C) 313.75 K; (D) 314.55 K and 1-
octadecanol (1) + hexadecanoic acid (2) at x1 = 0.7993 and (E) 319.15 K; (F) 321.15 K;
(G) 322.25 K; (D) 325.65 K. Some details are shown by white arrows. Magnification of
5.
88
Figure 7. Difractograms at T = 298.15 K for (A) 1-hexadecanol (1) + tetradecanoic acid
(2), from up to down: x1 = (0.0000, 0.0982, 0.1993, 0.3042, 0.5020, 0.6004, 0.7152,
0.7951, 0.8994, 1.0000) mole fraction and for (B) 1-octadecanol (1) + hexadecanoic acid
(2), from up to down: x1 = (0.0000, 0.3010, 0.4999, 0.7010, 0.8990, 0.9508, 1.0000)
mole fraction.
90
Figure 8. (A) Liquid-phase activity coefficients and (B) Mixtures’ excess free Gibbs
energy calculate by three-suffix Margules equation (in detail, magnification for x1 < 0.1)
for 1-hexadecanol (1) + tetradecanoic acid (2): 1L ( ), 2
L () and i
L (○) calculated
from experimental data by SLE equation.
93
Figure 9. Effect of the pure compounds’ standard deviations data in the modeling of the
liquidus line for 1-hexadecanol + tetradecanoic acid: experimental data (○); error bars =
0.30 K (temperature’s experimental uncertainty); ideal assumption with mean values of
the pure compounds’ melting properties () and modeling considering the standard
deviations of the pure compounds’ properties (- - -).
96
SUPPLEMENTARY DATA
Figure S1. Fractional melting curves obtained by DSC. Data were evaluated using the
analysis software from TA Instruments (New Castle).
101
Figure S2. Thermogram and fractional melting curves for 1-hexadecanol. Data were
evaluated using the analysis software from TA Instruments (New Castle).
102
Figure S3. Replicates of the DSC thermograms for 1-octadecanol (1) + hexadecanoic 104
xx
acid (2) x1 = 0.7010. In detail, magnification highlighting a solid-solid transition at T =
292.74 K.
Chapter 4. LIPIDIC PROTIC IONIC LIQUID CRYSTALS
Figure 1. Brønsted acid-base reaction between a carboxylic acid and monoethanolamine
or diethanolamine for the formation of a protic ionic liquid. R represents the alkylic
moiety of the carboxylic acid.
109
Figure 2. Solid-liquid equilibrium phase diagram of the (A) oleic acid +
monoethanolamine, (B) oleic acid + diethanolamine, (C) stearic acid +
monoethanolamine and (D) stearic acid + diethanolamine systems by () differential
scanning calorimetry and () light-polarized optical microscopy presenting solid
monoethanolamine (SMEA), solid diethanolamine (SDEA), solid fatty acid (SFA), solid protic
ionic liquid (SPIL), liquid phase (L), normal hexagonal (HI), lamellar (L), and inverted
hexagonal (HII) mesophases. Lines are guides for the eyes. Solid lines were
experimentally characterized, and dashed lines describe hypothetical phase boundaries
based on the Gibbs phase rule.
117
Figure 3. Light-polarized optical micrographs showing the mesophases textures: (A) oily
texture in the stearic acid + monoethanolamine at xfatty acid = 0.2, T = 343.15 K; (B) oily
streaks in the oleic acid + diethanolamine at xfatty acid = 0.6, T = 293.15 K; (C) maltese
crosses in an isotropic backgroup oleic acid + diethanolamine at xfatty acid = 0.5, T =
357.15 K; (D) oily streaks with inserted maltese crosses oleic acid + diethanolamine at
xfatty acid = 0.4, T = 340.15 K, (E) mosaic texture in the stearic acid + monoethanolamine
at xfatty acid = 0.4, T = 363.15 K; and fan-shaped textures in the (F) oleic acid +
diethanolamine at xfatty acid = 0.3, T = 298.15 K; (G) oleic acid + monoethanolamine at xfatty
acid = 0.7, 293.15 K and (H) oleic acid + diethanolamine at xfatty acid = 0.7, T = 298.15 K.
118
Figure 4. Conductivity ( as a function of concentration of protic ionic liquids (PIL)
aqueous solutions. Solid lines are linear fitting of the data (R2 > 0.98) for critical micellar
concentration determination (CMC).
120
Figure 5. Density data ( / g cm-3
) as a function of temperature (T / K) for the (A) oleic
acid + monoethalonamine and (B) oleic acid + diethanolamine systems at xoleic acid = 0.0
(), 0.1 (), 0.2 (), 0.3 (), 0.4 (), 0.5 (), 0.6 (), 0.7 (), 0.8 (), 0.9 () and 1.0
(). At the bottom of each plot, density data as a function of oleic acid mole fraction at T
= 298.15, 323.15, and 353.15 K. Solid and dashed lines are guide for the eyes. Gray
regions represent the liquid crystal domain.
123
xxi
Figure 6. Experimental excess mole volume data of the () oleic acid +
monoethalonamine and () oleic acid + diethanolamine systems at 298.15 K. Dashed
lines are guides for the eyes. Vertical gray line at x = 0.5 (pure PIL) depicts the “true”
system indicated at the top of the plot.
124
Figure 7. Dynamic viscosity data of (A) oleic acid + monoethanolamine and (B) oleic
acid + diethanolamine at xoleic acid = 0.0 (); 0.1 (); 0.2 (); 0.3 (); 0.4 (); 0.5 ();
0.6 (); 0.7 (); 0.8 (); 0.9 (); 1.0 (). Details at the top right of the plots show
magnifications of the data at low values. At the bottom of each plot are viscosity data
as a function of fatty acid mole fraction at 353.15 K. Vertical gray lines at x = 0.5 (pure
PIL) depict the “true” systems indicated above.
125
Figure 8. Linearization procedure of the dynamic viscosity data using the Arrhenius
like equation 1. (A) Oleic acid + monoethanolanmine and (B) oleic acid + diethanolamine
at xoleic acid = 0.0 (); 0.1 (); 0.2 (); 0.3 (); 0.4 (); 0.5 (); 0.6 (); 0.7 (); 0.8
(); 0.9 (); 1.0 (). At the right of each plot are linear fittings of the data (R2 > 0.99)
at some concentrations. Gray regions represent the liquid crystal domain.
126
Figure 9. Shear stress () versus shear rate ( ) curves of (A) monoethanolammonium
oleate and (B) diethanolammonium oleate at 293.15 (), 298.15 (), 308.15 (), 323.15
(), 328.15 (), 348.15 (), 353.15 (+), and 363.15 () K. Solid lines are fitted curves.
Labels N, ST, V, BP, and HB stand for Newtonian, shear-thinning, Vocadlo, Bingham
plastic, and Herschel-Bulkley models, respectively.
129
SUPPORTING INFORMATION
Figure S1. Tammann plots of the eutectic transitions Heut in the ethanolamine-rich
region (x = 0.0 to 0.5 fatty acid molar fraction) and in the fatty acid-rich region (x = 0.5 to
1.0 fatty acid molar fraction) of the (+) oleic acid + monoethanolamine, () oleic acid +
diethanolamine, () stearic acid + monoethanolamine and () stearic acid +
diethanolamine systems. Solid lines are linear fitting of the data (R2 > 0.95 considering
that y = 0.0 at the pure compounds concentration). Dashed lines are supposed plots close
to pure component concentration. () Diethanolamine and () monoethanolamine
melting enthalpy for analysis purpose.
137
Chapter 5. THE CRYSTAL - T ALGORITHM
Figure 1. Thermodynamic cycle comprising the heating, melting and cooling processes
of the system, from T to Tfus, from a solid to a subcooled liquid state taking into account
148
xxii
the polimorphysm phenomena at Ttr.
Figure 2. Solid-liquid equilibrium phase diagram of a discontinuous binary solid solution
case.
151
Figure 3. Block diagram for the Crystal-T algorithm 156
Figure 4. Nested structure for the SLE modeling using the Crystal-T algorithm 159
Figure 5. Experimental ( by DSC and by microscopy) and modeled (solid lines) phase
diagrams using the 3-suffix Margules equation for the triacylglycerol + fatty alcohols
mixtures. AijS, Bij
S are the parameters for the i
S equation and Aij
L, Bij
L for the i
L equation
(kJmol-1
). A) trilaurin (1) + 1-hexadecanol (2), B) trilaurin (1) + 1-octadecanol (2), C)
trimyristin (1) + 1-hexadecanol (2), D) trimyristin (1) + 1-octadecanol (2), E) tripalmitin (1) +
1-hexadecanol (2), F) tripalmitin (1) + 1-octadecanol (2).
162
Figure 6. Thermograms for the mixture trimyristin (1) + 1-hexadecanol (2).
Magnification of the thermogram for x1 = 0.778 mole fraction, in detail, showing the
eutectic transition.
163
Figure 7. Tammann plot of the enthalpy of the eutectic transition for the trimyristin (1) +
1-hexadecanol (2) mixture. Dashed lines are linear regression.
163
Figure 8. Melting transition at the eutectic temperature (Teut = 319.46 K) for the
trimyristin (1) +1-hexadecanol (2) system at x1 = 0.694 molar fraction. A) 319.15 K, B)
319.65, C) 320.15 K. White arrows highlight details during the melting process.
164
Figure 9. Heating and melting processes of trimyristin (1) + 1-hexadecanol (2) system at
x1 = 0.8937 molar fraction at T = A) 313.15 K, B) 322.15 K, C) 323.35 K, D) 323.55 K,
E) 323.95 K, F) 324.35 K, G) 327.65 K and H) 329.05 K. White arrows highlight details
during the heating process.
166
Figure 10. Micrographies and thermogram of the system tripalmitin (1) + 1-octadecanol
(2) at x1 = 0.697 molar fraction. Pictures are at a temperature before (A. 293.15 K) and
after (B. 320.15 K) the solid-solid transition at 318.75 K. Black arrows in the thermogram
highlight the transitions. Eutectic transition in detail (T = 329.20 K).
168
Figure 11. SLE phase diagram of the tripalmitin (1) + 1-octadecanol (2) mixture using 2-
suffix Margules as iS equation and considering that i
L = 1.0 (simple solid line) (Aij
S =
6.93 kJmol-1
) or using the original UNIFAC model (bold line) (AijS = 6.23 kJmol
-1) and
the UNIFAC-Dortmund model (dashed line) (AijS = 7.63 kJmol
-1) as i
L equation. DSC
() and microscopy () data.
171
Figure 12. Deviations (Equation 14) from experimental data for the system tripalmitin 173
xxiii
+ octadecanol using the 2-suffix Margules equation as iL and i
S models. Aij
S and Aij
L are
the iS and i
L parameters (kJmol
-1), respectively. In detail, phase diagrams using
different parameters sets.
Figure 13. Deviations (Equation 14) from experimental data for the system tripalmitin
+ 1-octadecanol using original UNIFAC as iL model and 3-suffix Margules equation as
iS model with Aij
S and Bij
S as parameters. In detail, phase diagrams using different
parameters sets.
174
Figure 14. Deviations (Equation 14) from experimental data for the system tripalmitin
+ 1-octadecanol using 3-suffix Margules equation as both iL and i
S models. Aij
S and Bij
S
are parameters for iS
and AijL and Bij
L are parameters for i
L. In detail, phase diagrams
using different parameter sets.
176
SUPPORTING INFORMATION
Figure S1. Tammann plots of the experimental eutectic transition enthalpy eutH () and
linear regression (dashed lines) for the systems A) trilaurin (1) + 1-hexadecanol (2), B)
trilaurin (1) + 1-octadecanol (2), C) trimyristin (1) + 1-hexadecanol (2), D) trimyristin (1)
+ 1-octadecanol (2), E) tripalmitin (1) + 1-hexadecanol (2), F) tripalmitin (1) + 1-
octadecanol (2).
186
xxiv
xxv
LIST OF TABLES
Chapter 1. TRENDS AND DEMANDS IN SOLID-LIQUID EQUILIBRIUM OF
LIPIDIC MIXTURES
Table 1. Fatty acid mass composition of vegetable oils. 18
Table 2. Solid-liquid phase diagrams of binary mixtures of fatty compounds reported in
literature.
21
Chapter 2. SOLID-LIQUID EQUILIBRIUM OF TRIOLEIN WITH FATTY
ALCOHOLS
Table 1: Deviation data from the literature (lit.) and from the algorithm proposed in this
work
54
Table 2: Experimental solid-liquid equilibrium data for triolein (1) + 1-hexadecanol (2). 57
Table 3: Experimental solid-liquid equilibrium data for triolein (1) + 1-octadecanol (2). 58
Table 4: Estimated parameters (A and B) and mean absolute deviations of the
predicted data from the experimental data (in Kelvin).
60
Table 5: Pure fatty components properties from the literature (lit.) and in this work 63
Chapter 3. ON THE SOLID-LIQUID EQUILIBRIUM OF BINARY MIXTURES OF
FATTY ALCOHOLS AND FATTY ACIDS
Table 1. Fatty compounds used in this work, supplier, purification method, molar
fraction purity and method used to confirm the purity.
73
Table 2. Experimental melting temperatures Tfus, solid-solid transitions Ttr and total
melting enthalpies fusH of the pure components obtained in this work and from literature.
75
Table 3. Experimental solid-liquid equilibrium data for 1-tetradecanol (1) + dodecanoic
acid (2) for molar fraction x, temperature T and pressure p = 94.6 kPa.
80
Table 4. Experimental solid-liquid equilibrium data for 1-hexadecanol (1) +
tetradecanoic acid (2) for molar fraction x, temperature T and pressure p = 94.6 kPa.
81
Table 5. Experimental solid-liquid equilibrium data for 1-octadecanol (1) + hexadecanoic
acid (2) for molar fraction x, temperature T and pressure p = 94.6 kPa.
82
Table 6. Estimated parameters, A and B, and mean absolute deviations a between
predicted and experimental data.
92
SUPPLEMMENTARY DATA
Table S1. Solid-liquid and solid-solid transitions temperatures obtained in replicates for 8
binary mixtures.
103
Chapter 4. LIPIDIC PROTIC IONIC LIQUID CRYSTALS
xxvi
SUPPORTING INFORMATION
Table S1. Experimental solid-liquid equilibrium data for oleic acid + ethanolamine
systems for mole fraction x, temperature T †
and pressure p = 102.0 kPa.
138
Table S2. Experimental solid-liquid equilibrium data for stearic acid + ethanolamine
systems for mole fraction x, temperature T †
and pressure p = 102.0 kPa.
139
Table S3. Density data ( / g cm-3
) of the oleic acid-based binary mixtures. 140
Table S4. Dinamic viscosity ( / mPas) data of the oleic acid-based binary mixtures. 141
Table S5. Rheological models’ parameters related to shear stress / shear rate curves fitted
to flow experimental data for the PILs.
142
Chapter 5. THE CRYSTAL - T ALGORITHM
SUPPORTING INFORMATION
Table S1. Melting temperatures (Tfus / K) and enthalpies (fusH / kJmol-1
) and solid-solid
temperatures (Ttr / K) and enthalpies (trH / kJmol-1
) of the pure compounds used in the
modeling procedure and obtained in this work or from literature.
184
Table S2. Experimental solid-liquid equilibrium data for the triacylglycerols + fatty
alcohols systems for mole fraction x and temperature T for solidus and liquidus line.
185
xxvii
NOMENCLATURE
SLE solid-liquid equilibrium
xi mole fraction of compound i in the liquid phase
zi mole fraction of compound i in the solid phase
iL activity coefficient of the compound i in the liquid phase
iS activity coefficient of the compound i in the liquid phase
Tfus melting temperature of pure compound
fusH melting enthalpy of pure compound
R ideal gas constant
T melting temperature of the system
fusCp melting heat capacity of pure compound
Ttr temperature of the solid-solid transition of pure compound
trH enthalpy of the solid-solid transition of pure compound
xeut mole fraction of the compound in the liquid phase and at the eutectic point
zeut mole fraction of the compound in the solid phase and at the eutectic point
Teut temperature of the eutectic transition of the system
Heut enthalpy of the eutectic transition of the system
Aij , Bij binary interaction parameters of 2- and 3-suffix Margules equations
Rk, Qk UNIFAC structural parameters for groups
amn UNIFAC group-interaction parameters
UNIFAC UNIQUAC Functional-group Activity Coefficient
Fobj objective function
niL number of moles of compound i in the liquid phase
niS number of moles of compound i in the solid phase
G variation of the Gibbs energy
TAG triacylglycerol
VLE vapor-liquid equilibrium
ARD average relative deviations
RMSD root mean square deviation
xxviii
1
INTRODUCTION
The world production and consumption of vegetable oils overtook 150 millions of
tons in 2012 (F.A.O., 2013) and this reveals the relevance of the fats and oils industry in the
world economy. This scenario has been stimulating the construction of biorefinaries,
especially in emerging economies that attend for more than 70% of this amount, and
indicates the important role that the oilchemistry may play in the future bioeconomy.
Through the biorefinary concept, new demands for fats and oils rise as well as favorable
prospects for the development of processes and products based on derivatives of fatty
components (H. Clark et al., 2009; Octave e Thomas, 2009). In this context, an accurate
understanding of the most relevant thermodynamic mechanisms that rules the
physicochemical, sensorial and rheological phenomena involving such substances and their
mixtures is fundamental. One aspect of such an understanding is achieved by the
knowledge of the phase equilibrium behavior of fatty systems, and in particular their solid-
liquid transitions (Figure 1).
The melting process of lipidic compounds is highly complex when compared with
other phase changes. This complexity is especially due to the inherent ability of the fatty
crystals to self-assembling into stable and metastable structures in their multiple solid
phases. Also, in a mixture, fatty components can also present mutual miscibility or even
total immiscibility in the solid phase. Consequently, apart from the fact that such
phenomena largely affect microstructural changes and, therefore, the set of industrial
quality requirements of products, numerous operational and economical reasons emphasize
the importance of a critical thermodynamic evaluation of the fatty systems for process
design and optimization.
2
Figure 1. Action fields of the solid-liquid equilibrium studies on the vegetable oils processing
industries.
Experimental and modeled solid-liquid equilibrium (SLE) phase diagrams have
been long since evaluated by several authors (Wesdorp, 1990; Coutinho et al., 1995;
Domanska e Gonzalez, 1996; Inoue et al., 2004; Costa et al., 2012). Despite the great
number of experimental data on fatty systems, works in literature very frequently fails in
the complete description of their melting behavior. The reason is that, because the
complexity of the solid phase, the well established thermodynamic theory is usually
simplified, either by the lack of pure compounds experimental properties or by neglecting
3
the presence of other phenomena. Concerning to the optimization of extraction, refining,
crystallization or fractionation processes as well as the development of chocolate, butters,
dressings, spreads, cosmetic creams, medicines or biofuels, the determination of the SLE of
fatty systems as well as the improvement of modeling tools is more than an demand, but an
imperative.
OBJECTIVES
This work is aimed at deepening the understanding of solid-liquid equilibrium of
fatty systems by the evaluation of mixtures of interest in the industry, not yet available in
the literature, as well as improving the approaches for modeling the SLE of fatty systems,
by implementation of an alternative algorithm that includes both liquid and solid phases
non-ideality for the description of the phase diagram. This thesis is divided in five works
whose complexity increases as the comprehension of the systems increases. The specific
objectives, and a short summary of each work are related below:
The first work comprises a review on the current understanding of the SLE
behavior of fatty systems and on the approaches used in literature for their
modeling, taking into account that, despite the amount of available
experimental data, a deepest knowledge on SLE of fatty compounds is still
required.
The second work reports experimental data and the corresponding modeling
results for the solid-liquid equilibrium phase diagrams of binary systems
composed by triolein + 1-hexadecanol and triolein + 1-octadecanol. Triolein is
4
a representative triacylglycerol of a set of vegetable oils and fatty alcohols are
well known surfactant agents. The phase diagrams showed a eutectic profile
with immiscibility in the solid phase and eutectic point very close to pure
triolein. Modeling could accurately describe the non-ideality of the liquidus line
by using adjustable excess Gibbs equations and group contribution UNIFAC
models.
The third work determines the experimental data and reports the corresponding
equilibrium results for systems composed of 1-tetradecanol + dodecanoic acid,
1-hexadecanol + tetradecanoic acid and 1-octadecanol + hexadecanoic acid.
Mixtures of fatty acids and alcohols are structuring agents for the production of
organogels as well as suitable energy storing materials. Such mixtures
presented a small solid solution region and the liquid phase showed a particular
non-ideal profile, varying from positive to negative deviations due to the
particular interactions between acids and alcohols.
The forth work evaluates the SLE behavior of systems composed of fatty acids
and ethanolamines. Through a Brønsted acid-base reaction, such mixtures led to
the formation lipidic protic ionic liquids. The phase diagrams of these systems
depicted a congruent melting behavior. The high self-assembling ability of such
ionic liquids led to the formation of liquid crystals, highly promising for
numerous pharmaceutical applications.
The fifth work presents an algorithm for the calculation of SLE phase diagrams.
The Crystal-T algorithm considers the formation of solid solutions and
5
calculates both liquidus and solidus lines. The procedure was based in the
classical Bubble-T algorithm for calculation of vapor-liquid equilibrium. The
activity coefficients were determined by excess Gibbs energy based equations,
including the UNIFAC model. The effectiveness of the routine was evaluated
by the description of 6 experimentally evaluated discontinuous solid solution
systems of triacylglycerols + fatty alcohols.
6
7
INTRODUÇÃO
A produção e o consumo mundiais de óleos vegetais ultrapassaram 150 milhões de
toneladas em 2012 (F.A.O., 2013), relevando a importância da indústria de óleos e gorduras
na economia mundial. Este cenário tem estimulado a construção de biorefinarias,
especialmente nas economias emergentes, responsáveis por 70% deste total, e indica o
papel fundamental que a óleo-química pode exercer, em um futuro próximo, na
bioeconomia. O conceito da biorefinaria incentiva o surgimento de novas demandas no
processamento de óleos e gorduras, além de perspectivas favoráveis para o
desenvolvimento de novos processos e produtos baseados em derivados de compostos
graxos (H. Clark et al., 2009; Octave e Thomas, 2009). Neste contexto, uma compreensão
cuidadosa dos mecanismos termodinâmicos mais relevantes que determinam os fenômenos
físico-químicos, sensoriais e reológicos que envolvem essas substâncias, e suas misturas, é
fundamental. Um aspecto de tal compreensão pode ser obtido através do estudo do
equilíbrio de fases, em particular as transições sólido-líquido (Figura 1).
O processo de fusão dos compostos lipídicos é particularmente complexo quando
comparado a outras mudanças de fase. Esta complexidade é devida a inerente habilidade
dos cristais de compostos graxos de se auto-organizarem em estruturas estáveis e meta-
estáveis em suas múltiplas fases sólidas. Além disso, em uma mistura, os compostos graxos
também podem apresentar miscibilidade parcial ou até mesmo imiscibilidade total na fase
sólida. Esses fenômenos produzem importantes mudanças microestruturais nos sistemas,
que influenciam uma série de requisitos de qualidade industriais. Deste modo, inúmeras
razões operacionais e econômicas enfatizam a importância de uma avaliação
8
termodinâmica crítica dos sistemas graxos para a otimização e projeto de processos e
produtos.
Figura 1. Campos de ação dos estudos do equilíbrio sólido-líquido nas indústrias de processamento
de óleos vegetais.
Os diagramas de fases do equilíbrio sólido-líquido (ESL) têm sido avaliados por
diversos autores na literatura (Wesdorp, 1990; Coutinho et al., 1995; Domanska e
Gonzalez, 1996; Inoue et al., 2004; Costa et al., 2012). Porém, apesar do grande número de
dados experimentais de sistemas graxos, os trabalhos na literatura frequentemente falham
na completa descrição do seu comportamento de fusão. Isso porque, devido à complexidade
9
da fase sólida, a teoria termodinâmica do equilíbrio sólido-líquido é comumente
simplificada, seja pela falta de dados experimentais relativos às propriedades dos
compostos puros ou por negligenciar a presença de outros fenômenos. Considerando a
otimização de processos de extração, refino, cristalização ou processos de fracionamento
assim como o desenvolvimento de chocolates, manteigas, molhos para salada, cremes
cosméticos, medicamentos ou biocombustíveis, a determinação do equilíbrio sólido-líquido
de sistemas graxos, assim como o melhoramento de ferramentas para sua modelagem, além
de uma demanda, é um imperativo.
OBJETIVOS
Este trabalho tem o objetivo de aprofundar a compreensão do equilíbrio sólido-
líquido de sistemas graxos pela avaliação de misturas de interesse na indústria, ainda não
avaliados pela literatura, e aprimorar as abordagens para a modelagem do ESL de sistemas
graxos através da implementação de um algoritmo alternativo que inclui a não-idealidade
de ambas as fases sólidas e líquidas para a descrição dos diagramas de fase. Esta tese é
dividida em cinco trabalhos. Os objetivos específicos e um breve resumo dos trabalhos
estão descritos abaixo.
O primeiro trabalho é uma revisão bibliográfica sobre a compreensão atual do
ESL de sistemas lipídicos e as abordagens utilizadas pelos trabalhos na
literatura para a sua modelagem, levando em consideração que, apesar da
quantidade de trabalhos sobre o tema na literatura, um conhecimento mais
profundo sobre o ESL de sistemas graxos é ainda necessário.
10
O segundo trabalho reporta dados experimentais e modelados para o diagrama
do ESL de sistemas binários compostos por trioleína + 1-hexadecanol e
trioleína + 1-octadecanol. A trioleína é um triacilglicerol representativo de uma
classe de óleos vegetais e os alcoóis graxos são agentes surfactantes muito
conhecidos. Os diagramas de fases descreveram um comportamento eutético
com imiscibilidade na fase sólida e ponto eutético muito próximo à
concentração da trioleína pura. A modelagem pôde descrever com precisão a
não-idealidade da linha liquidus através da utilização de equações ajustáveis de
energia de Gibbs em excesso e do modelo de contribuição de grupos UNIFAC.
O terceiro trabalho determina dados experimentais e modelados do ESL de
sistemas compostos por 1-tetradecanol + ácido dodecanóico, 1-hexadecanol +
ácido tetradecanóico e 1-octadecanol + ácido hexadecanóico. Misturas de
ácidos e álcoois graxos são agentes estruturantes para a produção de organogels
e materiais para armazenamento de energia. Tais misturas apresentaram uma
pequena região de solução sólida e fase líquida com um perfil de não idealidade
variando entre desvios positivos e negativos devido às interações ácido/álcool.
O quarto trabalho avalia o ESL de sistemas formados por ácido graxos e
etanolaminas. Através de uma reação ácido-base de Brønsted, essas misturas
levaram à formação de líquidos iônicos próticos. Os diagramas de fases desses
sistemas apresentaram um perfil de fusão congruente. A habilidade de auto-
organização dos líquidos iônicos foi determinante para a formação de cristais
líquidos, promissores em inúmeras aplicações farmacêuticas.
11
O quinto trabalho apresenta um algoritmo para o cálculo do diagrama de fases
do ESL de sistemas graxos. O algoritmo Crystal-T considera a formação de
soluções sólidas e calcula a não-idealidade de ambas as fases líquida e sólida. O
procedimento foi baseado no algoritmo clássico Bubble-T utilizado no cálculo
do equilíbrio líquido-vapor. Os coeficientes de atividade foram determinados
através de equações de energia de Gibbs em excesso, incluindo o modelo
UNIFAC. A efetividade da rotina pôde ser avaliada através do ajuste de 6
sistemas com solução sólida descontínua compostos por triacilgliceróis +
alcoóis graxos, determinados experimentalmente.
12
REFERENCES / REFERÊNCIAS
COSTA, M. C. et al. Phase diagrams of mixtures of ethyl palmitate with fatty acid ethyl esters.
Fuel, v. 91, n. 1, p. 177-181, 2012.
COUTINHO, J. A. P.; ANDERSEN, S. I.; STENBY, E. H. Evaluation of activity coefficient
models in prediction of alkane solid-liquid equilibria. Fluid Phase Equilibria, v. 103, n. 1, p. 23-
39, 1995.
DOMANSKA, U.; GONZALEZ, J. A. Solid-liquid equilibria for systems containing long-chain 1-
alkanols .2. Experimental data for 1-dodecanol, 1-tetradecanol, 1-hexadecanol, 1-octadecanol or 1-
eicosanol plus CCl4 or plus cyclohexane mixtures. Characterization in terms of DISQUAC. Fluid
Phase Equilibria, v. 123, n. 1-2, p. 167-187, 1996.
F.A.O. OECD-FAO Agricultural Outlook 2013 (2013). Available in www.fao.org. Acessed in
09/01/2013.
H. CLARK, J.; E. I. DESWARTE, F.; J. FARMER, T. The integration of green chemistry into
future biorefineries. Biofuels, Bioproducts and Biorefining, v. 3, n. 1, p. 72-90, 2009.
INOUE, T. et al. Solid-liquid phase behavior of binary fatty acid mixtures 2. Mixtures of oleic acid
with lauric acid, myristic acid, and palmitic acid. Chemistry and Physics of Lipids, v. 127, n. 2, p.
161-173, 2004.
OCTAVE, S.; THOMAS, D. Biorefinery: Toward an industrial metabolism. Biochimie, v. 91, n. 6,
p. 659-664, 2009.
WESDORP, L. H. Liquid-multiple solid phase equilibria in fats, theory and experiments. 1990.
Delft University of Technology, Delft.
13
TRENDS AND DEMANDS IN SOLID-LIQUID EQUILIBRIUM OF
LIPIDIC MIXTURES
(submitted to RSC Advances)
Guilherme J. Maximo, a Mariana C. Costa,
b João A. P. Coutinho,
c and Antonio J. A.
Meirelles a
a Food Engineering Department, University of Campinas, R. Monteiro Lobato 80, 13083-
862, Campinas, São Paulo, Brazil. b
School of Applied Sciences, University of Campinas,
R. Pedro Zaccaria, 1300, 13484-350, Limeira, São Paulo, Brazil. c CICECO, Chemistry
Department, University of Aveiro, 3810-193, Aveiro, Portugal.
Chapter 1
14
ABSTRACT
The production of fats and oils presents a remarkable impact in the economy, in particular
in developing countries. Academic research aims at forecast and prepare to attend
forthcoming demands. In this context, the study solid-liquid equilibrium of fats and oils is
highly relevant as it may support the development of new processes and products as well as
improve those already existent. However, the task is daunting because lipidic compounds
present multiple solid phases leading to behaviors that deserve special attention. The high
complexity of these systems requires a particular effort for the experimental study and
theoretical description of their phase behavior. This work is addressed at perspectives to
better understand and overcome some of these issues.
KEYWORDS: Solid-liquid phase diagrams, Modeling, Triacylglycerols, Fatty alcohols,
Fatty acids, Fatty esters.
1. Introduction
The study of the solid-liquid equilibrium (SLE) of lipidic mixtures is a fundamental
tool for the development and optimization of the processing of fatty based products. Fats
and oils are remarkable suppliers of food, materials, chemicals and also energy. In this
context, vegetable oils represented more than 150 milions of tons 1 of produced and
consumed material in 2012. This information reveals the importance that fats and oils
should have in the RD&I policy agenda especially in the context of developing countries,
that attend for more than 70% of this amount. Finding alternatives for adding value to a
15
long list of products and byproducts extracted from crops, fruits or seeds, as proposed in the
biorefinary concept, is a task that is gaining a growing attention. A fats and oils biorefinary
involves all the manufacturing steps concerning the extraction of the oil from the feedstock
and production of any derived fatty based biochemical 2, 3
. It means that numerous
separation and purification processes such as extraction, refining, crystallization or
fractionation as well as products such as chocolate, butters, dressings, spreads, cosmetic
creams, medicines or biofuels with specific physicochemical properties shall be then
correctly designed and this can only be attended by a reliable evaluation of the phase
transition phenomena involving the compounds presents in fats and oils and their
derivatives.
This review is aimed at providing an overview on the current understanding of the
solid-liquid equilibrium behavior of fatty systems and on the approaches used in the
literature for their modeling, taking into account that a deepest knowledge on solid-liquid
equilibrium of fatty compounds is still required. The complex nature of fats and oils
suggests that the discussion on their phase equilibrium is best tackled starting with the
simplest systems. As the complexity of the system increases, the efforts for obtaining a
better understanding of the systems behavior also increases. Knowing that fatty systems
comprises a large class of compounds, this review addresses mixtures comprising these
components directly found in vegetable oils, namely acylglycerols (mono-, di- and
triacylglycerols) and fatty acids, as well as those directly produced from these sources,
namely fatty alcohols and esters.
16
2. What do fats and oils provide?
Fats and oils are composed of at least 90% of triacylglycerols plus a set of
minoritary components but a lot of further information can be added to this very simplified
view. First of all, triacylglycerols are esters built by a glycerol moiety and three long-chain
organic acids. Such fatty acids carbon-chains mostly range from 8 to 20 carbons comprising
both saturated and unsaturated compounds. Table 1 shows the mostly frequently found fatty
acids in the triacylglycerols present in vegetable oils. Thus, taking also into account that
such set of organic chains varies into 3 positions in the glycerol moiety structure, those
90% of triacylglycerols constitute, in fact, a very large variety of molecules. Therefore,
assuming that the carbon-chain length and the number of unsaturations in the molecule
have a significant effect on the feedstock’s physicochemical properties, the knowledge of
the composition of the oil is crucial.
Considering a predominant fraction composed by triacylglycerols, but neglecting
the other minority set of compounds is a mistake. Crude oils comprise a group of fatty
acids originated from the hydrolysis of the triacylglycerol molecule and consequently,
mono-, diacylglycerols and glycerols molecules are also present. Additionally, a small
fraction of phospholipids, sterols, carotenes, triterpene alcohols, chlorophylls, tocopherols
and tocotrienols are commonly found, as well as several other non-lipidic compounds,
minerals, metal ions and proteins 4-9
. Most of them are usually removed or reduced during
the refining processes, whilst a set of them still remain on the refined final product. From
the industrial point of view, the presence of some of them highly affects the
physicochemical, sensorial and rheological profile of the final product, despite their low
concentrations. Moreover, they may produce undesirable effects for the quality
17
requirements, such as changes in the crystallization kinetics, stimulating or retardating the
nucleation or growth of the crystals, solidification at low temperatures, changing the strain,
stress and the yield force during the flow or emulsification of the system, preventing the
effectiveness of separation processes, etc. 8, 10-12
Although such compounds are usually removed, they may be also deliberately
added to the fat and oils or other derived products in order to play significant roles in the
final profile of the product. In fact, even some triacylglycerols not present in the original
crude oil are sometimes added to the product in order to obtain the desired effects.
Additionally, they are also used for the production of other additives in the industry, such
as fatty alcohols and fatty esters. Fatty alcohols are produced from fatty acids or even from
triacylglycerols by different routes of esterification and hydrogenation 13
, but are also found
in small amounts in vegetable waxes such as those from sugar cane, sunflower or peanuts
14. Fatty esters are produced by transesterification of triacylglycerols using alcohols. Both
are widely used for the production of several other additives, especially surfactants, for
food and cosmetic products 15-17
. Furthermore, several works have long since been
evaluating the use of fatty acid esters mixtures as an alternative fuel for the replacement of
fossil fuels. The transesterification process for the biodiesel production is mainly based on
the utilization of methanol, but also ethanol, in convergence to the larger production of
sugar-cane, especially provided by several emerging countries 18-20
.
18
Table 1. Fatty acid mass composition of vegetable oils
Source <
C12
C12:0
Lauric
C14:0
Myristic
C16:0
Palmitic
C16:1
Palmitoleic
C18:0
Stearic
C18:1
Oleic
C18:2
Linoleic
C18:3
Linolenic
>
C18
Rice brain 21
- - 0.96 18.17 0.61 1.54 38.50 35.61 2.67 1.92
Palm oil 22
0.17 1.15 1.24 40.68 0.15 4.72 41.78 8.84 0.18 0.68
Macadamia nut oil 23
- - 0.98 9.38 19.28 3.40 59.76 2.03 0.14 5.02
Brazil nut oil 23
- - - 15.86 - 11.34 30.68 42.12 - -
Cottonseed oil 24
- - 0.32 15.53 0.12 3.40 21.23 54.07 4.59 0.72
Jatropha curcas oil 25
- - 0.06 13.34 0.89 6.30 41.68 37.00 0.21 0.34
Canola oil 22
- - 0.07 4.72 0.25 2.55 62.56 20.13 7.08 9.62
Soybean oil 22
- - 0.09 11.18 0.09 4.13 25.62 50.88 4.97 1.49
Coconut oil 26
13.10 51.00 23.0 6.10 6.10 - - - - -
Crambe oil 27
- - - 2.07 - 1.03 19.38 8.33 4.53 64.66
Fooder radish oil 27
- - - 5.11 - 2.36 39.47 16.69 12.19 24.18
Macauba oil 27
- - - 21.8 4.08 2.76 58.97 11.64 0.75 -
Cuphea oil 9 - 76.40 7.80 2.40 - 0.70 5.90 6.90 0.10 -
Olive oil 9 - - - 11.00 0.80 2.20 72.50 7.90 0.60 -
Grapeseed oil 28
- 0.03 0.11 8.14 0.16 4.05 15.12 71.20 0.57 0.52
Cocoa butter 9 - 0.10 25.40 0.20 33.20 32.6 2.80 0.10 - -
Palm kernel oil 9 54.20 16.40 8.10 - 2.80 11.4 1.60 - - -
Buriti fruit oil 5 - 0.03 0.08 16.78 0.32 1.77 74.06 4.94 1.04 0.83
Tucum oil 29
3.60 50.60 23.70 5.30 - 2.50 9.30 3.60 0.1 0.1
Sunflower oil 30
- 0.04 0.10 6.80 0.08 4.16 22.75 63.83 0.82 1.42
3. The solid-liquid equilibrium of fatty systems
The nature of the solid-liquid equilibrium of fatty mixtures is complex. Comprising
multiple solid phases at the same temperature and pressure condition, literature very
frequently fails to correctly describing the phase behavior of these mixtures and this
becomes worst as more complex behaviors are observed. What makes the SLE of fatty
mixtures so particular is their inherent ability to polymorphism phenomena 31-34
. It means
that the long-hydrocarbon chains of the fatty compounds and consequently their crystal
lattice can be configured into numerous micro- and macromolecular arrangements, directly
related to the thermal treatment to which the mixture is submitted. Triacylglycerols,
particularly, present three classical and well known polymorphic forms: α, β’ and β, not
19
accounting for additional sub modifications that also appear during thermal processes. The
first structure is formed into fast cooling. When submitted to a slow heating, such
arrangement melts and is recrystallized into a more stable structure, the ’ form, and
sequentially into the β form. Since different crystal lattices are observed in these cases,
namely hexagonal, orthorhombic and triclinic, different transition temperatures and
enthalpies are also observed. Crystalline structure of fatty acids, alcohols and esters can
also pack into different polymorphic forms, usually in temperatures very close to the most
stable form 31, 35-37
. Knowing that the polymorphism phenomena can be managed by
particular thermal treatments, the parameters of the processes to which mixtures are
submitted, time and temperature, can define the profile of the final product.
3.1. Evolution
The first studies on SLE of fatty mixtures published in the literature presented only
the melting temperature behavior, also known as the liquidus line 38-41
and most of them
only in a restricted concentration range, such that phase diagrams could not be completely
represented. On the other hand, the interest on the solid-liquid transitions of these systems
as well as the capacity of a better characterization of their solid phases has grown
significantly in the last decades. Table 2 shows the set of binary fatty mixtures that have
already been reported in the open literature. Efforts to better characterize these systems
have been growing, but increasing also the complexity of their description. In this context,
four principal types of phase diagrams could be identified for binary mixtures, as sketched
in Figure 1.
20
Figure 1. Most common phase diagrams of binary fatty mixtures with the following
nomenclature: mole fraction of the component a in the liquid (xa) and solid (za) phases,
respectively, liquid phase (l), solid phase with pure or rich in component a (Ca) or b (C
b),
solid phase comprising both compounds (s), solid phase with new structure formed by
peritectic transition (Cn).
326
0 1
Tem
per
atu
re
xa, z a
312
0 1
Tem
per
atu
re
xa, z a
Teut
Ca
l
Cb
Cb + lCa + l
l
Ca + Cb
327
0 1
Tem
per
atu
re
xa, z a
Teut300
335
0 1
Tem
per
atu
re
xa, z a
Teut
l
SS
SS
Cb + Cn
Ca + Cn
Ca + l
Cn + l
Cb + l
s + ls + l
Ca + Cb
l
s + l
s
T
x z
l
(B)(A)
(C) (D)
T
x z
21
Table 2. Solid-liquid phase diagrams of binary mixtures of fatty compounds reported in literature.
System Phase
Diagram
Ref. System Phase
Diagram
Ref.
Fatty ester + fatty ester
Fatty acids + fatty acids Ethyl caprylate + ethyl palmitate E 42
Caprilic acid + capric acid ESS/P/M 36
Ethyl caprylate + ethyl stearate E 43
Caprilic acid + lauric acid ESS/P/M 37
Ethyl caprate + ethyl palmitate E 42
Caprilic acid + myristic acid ESS/P/M 44
Ethyl caprate + ethyl stearate E 43
Caprilic acid + oleic acid ESS 45
Ethyl laurate + ethyl palmitate E 42
Capric acid + lauric acid ESS/P/M 36, 46
Ethyl laurate + ethyl stearate E 43
Capric acid + myristic acid ESS/P/M 37, 46
Ethyl myristate + ethyl palmitate E 42
Capric acid + palmitic acid ESS/P/M 44, 46
Ethyl myristate + ethyl stearate E 43
Capric acid + stearic acid E 46
Ethyl palmitate + ethyl stearate E 43, 47, 48
Capric acid + oleic acid ESS 45, 49
Ethyl palmitate + ethyl oleate E 42
Lauric acid + myristic acid ESS/P/M 36, 46
Ethyl palmitate + ethyl linoleate E 42
Lauric acid + palmitic acid ESS/P/M 37, 46
Ethyl stearate + ethyl oleate E 43
Lauric acid + stearic acid ESS/P/M 44, 46
Ethyl stearate + ethyl linoleate E 43
Lauric acid + oleic acid E 50
Methyl laurate + methyl palmitate E 51
Myristic acid + palmitic acid ESS/P/M 36
Methyl myristate + methyl palmitate ESS/P/M 35, 51, 52
Myristic acid + stearic acid ESS/P/M 37
Methyl myristate + methyl stearate ESS/P/M 35
Myristic acid + oleic acid E 50
Methyl palmitate + methyl stearate ESS/P/M 35, 47, 51
Palmitic acid + stearic acid ESS/P/M 36
Methyl palmitate + methyl oleate E 51
Palmitic acid + oleic acid E 53
Mehtyl palmitate + methyl linoleate E 51
Palmitic acid + linoleic acid E 53
Methyl stearate + methyl oleate E 51
Oleic acid + stearic acid E 49, 54
Methyl stearate + methyl linoleate E 51
Oleic acid + linoleic acid E 49
Methyl oleate + methyl linoleate E 51
Triacylglycerol + triacylglycerol
Fatty alcohols + fatty alcohols Tricaprylin + tristearin ESS 34
1-octanol + 1-decanol ESS/P/M 55
Trilaurin + tripalmitin ESS 34
1-decanol + 1-dodecanol ESS/P/M 55
Trilaurin + tristearin ESS 34
1-dodecanol + 1-tetradecanol Ɏ 55
Trimyristin + tripalmitin ESS 34
1-dodecanol + 1-hexadecanol ESS/P/M 55
Tripalmitin + tristearin ESS 56
1-dodecanol + 1-octadecanol ESS 57
Tripalmitin + triolein ESS 34, 53, 56
1-tetradecanol + 1-hexadecanol Ɏ
55 Tripalmitin + trilinolein ESS
53
1-tetradecanol + 1-octadecanol ESS/P/M 55
Tristearin + triolein E 34
1-hexadecanol + 1-octadecanol SS 48, 58
Triacylglycerol + fatty acid
Tricaprylin + myristic acid E 59
Fatty acid + fatty alcohol Tripalmitin + oleic acid ESS 49, 53
Capric acid + 1-dodecanol ESS/P 60
Tripalmitic + linoleic acid ESS 53
Lauric acid + 1-dodecanol P 60
Tristearin + palmitic acid ESS 56
Lauric acid + 1-tetradecanol ESS 61
Tristearin + linoleic acid ESS 56
Myristic acid + 1-dodecanol P 60
Triolein + palmitic acid E 53, 59
Myristic acid + 1-tetradecanol P 60
Trilinolein + palmitic acid E 53
Myristic acid + 1-hexadecanol E 61
Trilinolenin + stearic acid E 59
Palmitic acid + 1-dodecanol P 60
Palmitic acid + 1-tetradecanol E 60, 62
Triacylglycerol + fatty alcohol
Palmitic acid + 1-hexadecanol P 60
Trilaurin + 1-hexadecanol ESS 63
Palmitic acid + 1-octadecanol ESS 61
Trilaurin + 1-octadecanol E 63
Stearic acid + 1-dodecanol E 60
Trimyristin + 1-hexadecanol ESS 63
Stearic acid + 1-tetradecanol E 60
Trimyristin + 1-octadecanol ESS 63
Stearic acid + 1-hexadecanol P 60
Tripalmitin + 1-hexadecanol ESS 63
Stearic acid + 1-octadecanol P 60, 64
Tripalmitin + 1-octadecanol ESS 63
Triolein + 1-hexadecanol E 65
Triolein + 1-octadecanol E 65
E = Simple eutectic, P = Peritect, M = Metatectic, ESS = Eutectic with partial miscibility in the solid phase,
SS = Continuous solid solution, Ɏ
= not defined.
22
The first and simplest phase diagram is the so-called eutectic (Figure 1 A). Some
binary mixtures of fatty acids, mixtures of fatty esters or systems composed by fatty acids
and fatty alcohols present this behavior. Simple-eutectic mixtures are such that during the
cooling process, their compounds crystallize independently. The eutectic point establishes
the concentration at which the system melts at a single, well defined, and minimum
temperature. In a simple eutectic system there are 4 domains: one homogeneous liquid
phase (l); two biphasic solid-liquid domains (regions Ca + l and C
b + l); and one biphasic
solid-solid domain (region Ca + C
b). In such a system when a solid mixture is heated it
starts to melt always at the same temperature (eutectic temperature) and the composition of
the liquid phase changes along one of the branches of the liquidus line according to the
initial concentration of the solid mixture.
The second case (Figure 1 B) shows the phase diagram of a mixture with solid
solution in the extremes of the phase diagram. Depending on concentration of the mixture,
the molecules of a compound are incorporated into the lattice of another crystal 66
during
crystallization. It means that within this concentration range, in general close to one of the
pure components, the compounds do not crystallize independently. In case of lipidic
systems, solid solutions are particularly interesting because they can be intentionally
developed such that physicochemical properties could be adjusted to specific requirements
67-69. Several fatty binary mixtures present solid solution formation at some concentrations,
especially those composed of triacylglycerols or fatty alcohols 34, 56, 57, 61
. The phase
diagrams of such mixtures comprise 6 domains as shown in Figure 1 B. The domains are
explicitly circumscribed by two boundaries: the liquidus line, in which the last crystal of
the mixture melts, and the solidus line indicates the limits of the melt of the first crystal.
The eutectic point in this case establishes a triphasic equilibrium in which the liquid phase
23
is in equilibrium with two solid solutions with distinct compositions. The evaluation of the
eutectic point and the solubility region of the solid phase is usually carried out using the
Tammann plot (Figure 2) 45, 56, 65, 70
. This diagram relates the enthalpy of the eutectic
reaction as a function of the mixture concentration in which the enthalpy rises up to the
eutectic point when it begins to decrease. The concentration at which the solid solution is
formed is obtained when enthalpy of such reaction Heut = 0.
Figure 2. Tammann plot of the eutectic reaction: enthalpy of the eutectic reaction as a function of
the concentration of the mixture (full line); mole fraction of the component a at the eutectic point in
the liquid phase xeut (dashed line), and in the solid phases (zeut) rich in the component a (Ca) or b
(Cb); represented in gray, solid solution regions rich in the component a or b.
A behavior not so common occurs when the components of the liquid phase
crystallize as a solid solution throughout the entire concentration range. It means that only 3
domains are observed (Figure 1 C): the liquid region, the solid domain and the region of
coexistence of both liquid and solid phases. Many mixtures of triacylglycerols with similar
molecular sizes, such as PPS (Glycerol 1,2-palmitate,3-stearate) + PPP (tripalmitin) or SOS
24
(Glycerol 1,3-stearate,2-oleate) + POS (Glycerol 1-palmitate,2-oleate,3-stearate) 34
are able
to form solid solutions. In this case, systems are continuous solid solutions.
Frequently, in the crystallization processes of lipidic compounds, associative
phenomena can occur during the building of the crystal lattice. Such associative behavior
may lead to the appearance of a new crystalline structure that acts as a new compound. This
is the so-called peritectic reaction. This reaction is defined as an isothermal and reversible
process between a liquid and a solid phase resulting, during the cooling of a system, in new
solid phases 66
. It is possible to understand such phenomena taking into account the
inherent polymorphic ability of fatty compounds. Some works suggest to interpret such
phenomenon as a chemical reaction so that the original components form a product
according to a specified stoichiometry. If the new compound crystallizes independently of
the mixture’s compounds the phase diagram behavior is that presented in Figure 1 D. This
means that there are two biphasic solid regions, each with two immiscible crystalline
structures, the new compound Cn and one of the pure components of the original mixture Cb
or Ca , Furthermore, there are 3 solid-liquid domains involving one of the pure components
in the crystalline form and the liquid phase or the new compound and the liquid phase. Note
that the new compound exists only in the solid phase, and disappears during the system’s
melting. Works in literature report the appearance of peritectic profiles in binary mixtures
of fatty acids and some systems composed by methyl esters 35-37, 44
.
Despite the classification of many fatty mixtures into the typical four phase
diagrams shown in Figure 1, the understanding of the phase behavior of the fatty systems is
still far from complete. It is not uncommon that the solid phase presents further transitions
that remain to be clarified 71
. Metatectic transitions have also been identified on these
systems. In this case, the mixture can recrystallize during the melting process, i.e on heating
25
above the solidus line 36
. It means that the crystalline profile of the solid phase clearly
changes and probably a corresponding set of other particular properties. Recent studies 36,
37, 44 describe fatty systems with up to 15 different domains (Figure 3). In these mixtures all
the phenomena previously discussed are present in a single phase diagram, depending on
concentration and temperature: eutectic, peritectic and metatectic transitions as well as
partial miscibility of the solid phases.
Figure 3. Sketch of the SLE phase diagram observed in mixtures of fatty acids (as presented by
Costa, et al. 37
) for some fatty acid + fatty acid binary systems. Ca and
C
b are the solid solutions rich
in one of the components of the mixture a or b; Cn is the solid phase formed by the peritectic
reaction; Cha and Ch
b are the solid phases formed after the metatectic reaction; l is the liquid phase.
In fact, a large set of fatty binary mixtures were not evaluated yet: mixtures of
triacylglycerols; systems composed of triacylglycerols and fatty acids, alcohols or esters;
systems comprising methylic and ethylic fatty esters; mixtures of fatty esters and fatty acids
or alcohols; not to mention a long list of mono- and diacilglycerols mixtures. Moreover,
26
beyond these binary fatty systems, many works in literature, aiming at the design of
products and processes, have been evaluating the SLE of binary systems of these fatty
compounds with other organic solvents such as cyclohexane and ethanol 72
, alkanes 73
,
aromatic compounds 74
and drugs 75
. Additionally, the SLE of many lipid-derivative based
mixtures composed of surfactants 76
, phospholipids 70
, glycolipids 77
or lipidic-based ionic
liquids 78
have also been reported. These systems display a set of liquid crystalline states or
mesophases. Such mesophases are called as the “forth state of matter” since their
physicochemical properties are similar than those of a liquid state but the oriented nature of
the molecules confers particular properties of a solid state 79
. The presence of these
mesophases highly increases the complexity of the system equilibrium.
In summary, all these works reveals the diversity of the solid-liquid equilibrium
behavior of lipidic mixtures but also that several experimental studies are still required for a
detailed characterization and understanding of these systems.
4. Brief overview on the techniques for measuring SLE data
The temperatures and enthalpies of the transitions of both solid-liquid and solid-
solid phenomena, including those due to the polymorphic transitions, can be obtained by
differential scanning calorimetry (DSC). Such a technique is classically used for the
description of solid-liquid phase diagrams of fatty systems. However, owing to the
polymorphism of the solid phase, multiple crystalline structures can be present in a mixture.
Thus, some authors in literature 46
propose that samples should be submitted to a thermal
treatment before the measurement of the melting properties. First, with complete melting of
the mixture, the previously organized crystalline structures, that literature calls thermal
27
history, are deleted. Knowing that the polymorphic forms are highly dependent on the
temperature and the heating/cooling rates, specific procedures are required in order to
assess the temperatures and enthalpies of such different crystal structures. This can be
reached by melting the sample above the highest melting point of the components, cooling
at low rates, followed by an isothermal treatment at a very low temperature and a slow
heating run up to the melting of the component or mixture of compounds. In fact, the DSC
method is not a classical static method for the determination of equilibrium data. However,
slow rates allow to assess a quasi-equilibrium state in which crystal structures can be
rearranged such that temperatures and enthalpies of the transitions are measured. Moreover,
this technique is also able to detect endo- and exothermic events; it means that it detects not
only melting transitions but recrystallization phenomena of the solid phase, such as
polymorphic or metatectic transitions.
Besides DSC, a set of other analytical tecniques, such as X-ray diffraction and
microscopy, are frequently required in order to characterize in more detail the phase
behavior of the system and are often used in the literature for the evaluation of fatty
mixtures 10, 12, 37, 68, 80
. Single-crystal and powder X-ray diffraction tecniques allow assess
information on the crystal structure of the pure compounds and their mixtures. On the
molecular point of view, through the knowledge of the molecular structure of the
compounds, enthalpic and entropic molecular interactions can be inferred and the
ideal/non-ideal thermodynamic behavior of the mixture explained. Controlled-temperature
microscopy allows the visual observation of the beggining and the end of the melting,
crystallization and recrystallization phenomena, and the evalution of the growth and
disapearance of the crystals. Information obtained by these techniques can clarify the
28
transitions observed by DSC, such as polymorphic transitions, formation of solid solutions,
solid phase immiscibility, peritectic and metatectic reactions.
5. But, in fact, what is literature looking for?
Although several transitions or even complete behaviors have been clarified in the
literature, the solid-liquid phase diagrams of various fatty mixtures are still unknown or
were published with limited information concerning various aspects of their real behavior.
The knowledge of the SLE behavior of fatty systems open borders for new applications or
to the understanding of unclear industrial or biochemical phenomena involving such
substances. In fact, the phase equilibrium study really creates an additive design concept. It
means that if one aims to change the nutritional, sensorial or physicochemical profile of a
fatty product by the introduction or removal of components, increasing or decreasing the
melting point, the evaluation of the phase change profile is essential. In this context, a SLE
phase diagram is a powerful tool.
The solid-liquid equilibrium of fatty systems had been long evaluated to obtain an
assessment on the description of numerous processes or development of products 10, 11, 32, 69,
81. Some examples can be easily mentioned. a) Optimization of fractionation processes in
which the material are separated into fractions depending on the melting point by a set of
crystallization procedures 82
; b) Optimization of interesterification processes in which fatty
acids moieties are added or removed from the triacylglycerols in order to obtain lower or
higher melting points materials. This process is a classical alternative for the replacement of
the hydrogenation procedure which promotes the formation of saturated or trans-
unsaturated fatty compounds related as precursors of cardiovascular diseases 8; c)
29
Development of structured lipids for the improvement of the nutritional or physicochemical
characteristics of the product. This process involves chemical or enzymatic pathways and
is aimed at the modification of the triacylglycerols’ fatty acid profile 33
; d) Control of
crystallization phenomena in lubricants or oils for frying processes, in which solid particles
or even compounds that could promote crystallization are undesirable or in butters, creams
and analogous products, in which the crystalline structure is responsible for the desired
texture 10, 11, 69, 81
Other special cases take into account the production of new materials for the
applications in food, pharmaceutical, materials or energy industries. Mixtures of fatty acids
and fatty alcohols, for example, had been evaluated for the formulation of organogels 83, 84
and phase change materials 85-87
. Moreover, mixtures of fatty esters have long been used as
biofuels where the crystallization of saturated esters is a major challenge limiting their
operation at low temperatures 19, 20, 35, 88, 89
.
Organogels are structured materials formed by self-assembling of the components
through noncovalent interactions. They can become an alternative for structuring fat-based
systems making possible to replace traditional industrial processes as hydrogenation,
fractionation and interesterification. Fatty acids and alcohols are compounds 15, 16
that can
replace saturated fats by acting as gelator agents and promoting the formation of
organogels. Thus, the rheological properties of these products can be controlled by
formation of structured networks with different packing arrangements and particular shear
strain and stress profiles 90
.
Phase change materials are materials for energy storage. Concerning a phase-change
heat storage system, energy could be stored and released by reversible solid-solid or solid-
liquid phase changes. Fatty acids, alcohols and esters have relative low melting point, low
30
vapor pressure, low toxicity and high latent heats, being suitable for thermal storage
processes. 85, 91
Such materials play an important role as an environmentally friend way of
accumulating and saving energy, replacing traditional storage fluids such as those used in
refrigeration systems and contributing for the reduction of the emission of greenhouse-
gases.
Biodiesel is a mixture of the mono-alkyl esters obtained as derivatives of fat and oils
and thus, it is a renewable energy source. Consequently, they are environmentally
attractive, promoting the reduction of the emission of pollutant gases in replacement of
fossil fuels. Because of this, nowadays, biodiesel is mandatorily added to petrodiesel in
several countries. However, the biofuel industry is still looking for alternatives to overtake
several technical problems associated to biodiesel production, storage and use. The
solidification at low temperatures is one of them. The presence of fatty esters with high
melting temperature, such as methyl stearate and methyl palmitate is a factor that makes
difficult the use of this biofuel at low temperatures. Moreover, other minor components
such as monoacylglycerols of saturated fatty acids can also be present and can precipitate in
the storage tanks at low temperatures leading to the clogging of tubes and filters of the
vehicles. In all these cases a deeper knowledge of the solid-liquid equilibrium of fatty
mixtures can help optimizing industrial processes and improving the formulation of
products with desirable properties.
6. Demands on modeling
This growing number of new experimental data has not been followed by
proportional efforts to better model these complex phase diagrams. The greater the interest
31
in new systems or applications, the higher the demand to improve reliable modeling
approaches. The modeling of the melting profile of lipidic systems is best tackled by
depicting the melting temperature T as a function of the composition of the mixture or, in
other words, by depicting the phase diagram. This problem is solved by evaluating the well
known thermodynamic equilibrium condition. The thermodynamic equilibrium is
established when at the same temperature and pressure, the chemical potential of a
compound in every phase is also the same, in the present case, a liquid and a solid phase.
Theoretically, this problem is usually described by Equation 1 that relates the mole fraction
of the component i in the liquid phase xi and the melting temperature Tfus and enthalpy
fusH of the pure compound, so that the melting temperature of the mixture T can be
calculated 92
.
fusi
fus
1 1ln
Hx
R T T
(1)
where R is the gas constant. However, this equation is far from being adequate to accurately
depict the real state of the system. The SLE can be fully considered, and so better
described, by Equation 2 92, 93
.
Lfus tri i
fu
fus fus fus
s tri i 1
1 1 1 1ln 1ln
tr
pn
S
H Hx
R T T R
C T
T T R
T
T Tz
(2)
where zi is the mole fraction of component i in the solid phase, iL and i
S are the activity
coefficients of component i in the liquid and solid phases, respectively, Ttr and trH are
thermal transitions temperatures and enthalpies of the n solid-solid transitions (polymorphic
forms) of the component i and fusCp is the difference between the heat capacity of the pure
component i of the liquid and solid phases. If both equations are compared, it is observed
that Equation 1 was built under three main considerations. Firstly, it does not consider the
32
non-ideality of the phases by means of the calculation of the activity coefficients of
component i in both phases liquid iL and, if it is the case, solid i
S. Moreover, it assumes
that the solid phase is composed of immiscible pure components such that zi i
S = 1.0. Also,
it neglects the presence of polymorphic transitions and so the effects of the thermal
transitions temperatures Ttr and enthalpies trH that are present in the heating profile of the
components of the mixture.
Through Equation 2, the non-ideality of both phases is fully considered. Also, it
presents additionally two terms: The first term takes into account the effects of the
polymorphic forms of the mixture components on the melting profile. The second one is
related to the specific heat capacity fusCp that is usually neglected due to a clear lack of
reliable experimental data and because the magnitude of this property supposedly does not
affect the solid-liquid equilibrium calculation. Moreover, Equation 2 reveals that, the
modeling of fatty mixtures’ phase diagrams is based on a fundamental problem: the
miscibility of the solid phase or the presence of solid solutions, so that zi i
S 1.0. This is
related to the calculation of the solidus line. The solidus line depicts the behavior of the
melting temperature T as a function of the solid phase composition zi.
Despite the fact that solid solutions are more common than usually admitted, most
authors have adopted the modeling approach based on the solid phase immiscibility. In the
works that take into account a more comprehensive approach and consider the formation of
such solid solutions 33, 34, 89
two main focuses can be identified. From the academic focus,
the solid-liquid equilibrium theory is formulated for complete description of the solid-liquid
phase diagram of a system of interest, this means for calculating the liquidus and the
solidus lines. From the industrial focus, the knowledge of the solid content behavior of a
33
vegetable oil or a mixture of triacylglycerols has a particular interest for the formulation of
products. This behavior is best assessed by the construction of a diagram that represents the
solidified fraction of the system as a function of temperature. The aim is the manipulation
of the melting point of the mixture by the right choice of its composition. It is known that,
for instance, many sensorial aspects, such as crispness, spreadability, aroma and oiliness are
largely affected by the solid/liquid ratio in lipidic products, such as cocoa and peanut
butters, but also in case of cosmetic creams 10, 81
. Analogously, the melting behavior of a
biodiesel mixture is largely dependent on the original fat or oil composition as well as the
alcohol used in the transesterification reaction 19
. It means that blends with high
concentration of long carbon saturated fatty esters are more susceptible to high solid
content at low temperatures, worsening their flow properties 20, 51
. Other example for the
construction of solid fat content diagrams is the development of the so-called structured
lipids. By changing the fatty acid composition of triacylglycerols it is possible to improve
specific properties 33
. High amounts of saturated or trans-unsaturated fatty acids can, for
instance, bring negative effects concerning the nutricional claim of the product and,
additionally, can increase its melting point. Thus, by the adequate right choice of fatty acids
the desired solid content profile of a mixture can then be designed avoiding, for instance,
the presence of trans-unsaturated fatty acids.
In order to assess the SLE described by Equation 2, works in literature apply mainly
two well known algorithms: the minimization of the Gibbs energy and the isoenthalpic
flash calculation 34
. The first is based on the stability criteria in which a mixture at
equilibrium condition present the minimum Gibbs energy. The second one, in analogy with
the vapor-liquid equilibrium, considers the feeding of a mixture in an isothermic and
isobaric vessel and the equilibrium condition is then calculated through the analysis of the
34
mass balance and the fundamental SLE equations. Both approaches are robust in obtaining
the equilibrium state of the mixture but sensitive to initial concentrations estimates.
For the description of the non-ideality of the system, these optimization procedures
calculate the activity coefficients of the liquid and solid phases. The activity coefficients of
the components in the liquid phase iL are usually calculated using equations for the excess
Gibbs energy 92
such as Margules or UNIQUAC equations 94
or the group-contribution
UNIFAC model 95, 96
. Equation 3 shows the 2-suffix Margules equation, commonly used in
the SLE modeling of binary systems, comprising one adjustable parameter A12, related to
the interaction between the compounds of the binary.
L 2
1 12 2 ln( )RT A x (3)
where x is the composition of the component in the liquid phase, T is the melting
temperature of the mixture and R is the ideal gas constant. Despite the slight non-linearity
of this model, what is favorable in case of optimization procedures, its disadvantage is the
calculation of symmetric non-ideal behaviors. It means that the compounds should present
positive or negative deviations but not both behaviors depending on the concentration. The
UNIFAC equation is a predictive method commonly used in case of liquid-liquid and
vapor-liquid equilibrium and that can be also applied for the calculation of the iL in case of
SLE studies 96
. This model is based on the group-contribution concept so that the activity
coefficient of the component is calculated by a sum of terms related to enthalpic and
entropic interactions. It is well applied for the description of the liquid phase in the SLE of
fatty systems presenting slight deviations from ideality. However, the model can fail in the
prediction of strong non-ideal behaviors such those observed in cases where the compounds
are very different in size.
35
Two approaches were up to now used for the calculation of activity coefficients of
the components in the solid phase of fatty systems: the two-suffix Margules equations 34
for
the description of the solid phase of mixtures of triacylglycerols and a predictive version of
the UNIQUAC model 89, 97
for the description of mixtures of fatty esters. Applying
Margules equation, Equation 3 is used but evaluating the composition of the compound in
the solid phase zi. In the predictive version of the UNIQUAC model, the binary interactions
parameters are calculated by using the physicochemical properties of the pure compound,
as described by the authors.
Apart from discussions on the effectiveness these approaches, taking into account
the description of the lipidic systems listed in Table 2 that present different kind of
behaviors as well as numerous mixtures that were not evaluated yet, the SLE modeling of
fatty systems is a field yet to be explored.
7. Remarks
In this large and opened context, several demands are clearly evident. The first and
most obvious is the enlargement of the number of reliable experimental data of binary,
ternaries and even more complex mixtures of fatty systems. Additionally, pure compound
experimental data, melting temperatures and enthalpies as well as solid-solid transitions
properties and specific heat capacity of the phases are highly demanded, since all of these
parameters are required in the modeling procedure. The second demand is the improvement
of the description of the non-ideality of the solid phase by using reliable equations or
predictive models as a tool for the design of mixtures with specific behaviors. The third is
the evaluation of the modeling of systems comprising the peritectic reaction whose efforts
36
are too incipient in the literature. The forth is deepening the understanding of the solid
phases, with experimental and modeling tools in order to describe its behavior according to
the well known thermodynamic theories. The better the description of the SLE behavior of
fatty systems the easiest the development or the optimization of fatty-based products and
processes will be. New alternatives can be thus found to overcome industrial problems,
opening pathways for new academic studies.
ACKNOWLEDGEMENTS
The authors are grateful to the national funding agencies FAPESP (2012/05027-1,
2008/56258-8), CNPq (304495/2010-7, 483340/2012-0, 140718/2010-9, 479533/2013-0)
and CAPES (BEX-13716/12-3) for the financial support.
REFERENCES
1. F.A.O., OECD-FAO Agricultural Outlook 2013, available in http://www.fao.org, accessed in
09/01/2013, 2013.
2. S. Octave and D. Thomas, Biochimie, 2009, 91, 659.
3. J. H. Clark, V. Budarin, F. E. I. Deswarte, J. J. E. Hardy, F. M. Kerton, A. J. Hunt, R. Luque, D.
J. Macquarrie, K. Milkowski, A. Rodriguez, O. Samuel, S. J. Tavener, R. J. White and A. J. Wilson,
Green Chem., 2006, 8, 853.
4. S. M. Silva, K. A. Sampaio, R. Ceriani, R. Verhé, C. Stevens, W. De Greyt and A. J. A.
Meirelles, J. Food Eng., 2013, 118, 341.
5. S. Silva, K. Sampaio, T. Taham, S. Rocco, R. Ceriani and A. A. Meirelles, J. Am. Oil Chem.
Soc., 2009, 86, 611.
6. R. Ceriani and A. A. Meirelles, J. Am. Oil Chem. Soc., 2004, 81, 305.
37
7. K. M. Phillips, D. M. Ruggio, J. I. Toivo, M. A. Swank and A. H. Simpkins, J. Food Compos.
Anal., 2002, 15, 123.
8. F. Shahidi, Bailey's Industrial Oil and Fat Products, Wiley Interscience, 2005.
9. J. A. Kent, Kent and Riegel's Handbook of Industrial Chemistry and Biotechnology, Springer,
New York, 2007.
10. C. Himawan, V. M. Starov and A. G. F. Stapley, Adv. Colloid Interfac., 2006, 122, 3.
11. K. Smith, K. Bhaggan, G. Talbot and K. Malssen, J. Am. Oil Chem. Soc., 2011, 88, 1085.
12. A. G. Marangoni, N. Acevedo, F. Maleky, E. Co, F. Peyronel, G. Mazzanti, B. Quinn and D.
Pink, Soft Matter, 2012, 8, 1275.
13. U. Kreutzer, J. Am. Oil Chem. Soc., 1984, 61, 343.
14. M. H. Gordon, in Oils and fat authentication, ed. M. Jee, Oxford, 2002.
15. R. W. Johnson and E. Fritz, Fatty Acids in Industry: Processes, Properties, Derivatives,
Applications, Marcel Dekker, Nova York, 1989.
16. H. Maag, J. Am. Oil Chem. Soc., 1984, 61, 259.
17. R. D. Farris, J. Am. Oil Chem. Soc., 1979, 56, 770A.
18. M. B. Dantas, M. Conceição, V. J. Fernandes Jr, N. Santos, R. Rosenhaim, A. B. Marques, I. G.
Santos and A. G. Souza, J. Therm. Anal. Calorim., 2007, 87, 835.
19. G. Knothe, Fuel Process. Technol., 2005, 86, 1059.
20. G. Knothe, Energ. Fuels, 2008, 22, 1358.
21. C. E. C. Rodrigues, R. Antoniassi and A. J. A. Meirelles, J. Chem. Eng. Data, 2003, 48, 367.
22. M. Lanza, G. Sanaiotti, E. A. C. Batista, R. J. Poppi and A. J. A. Meirelles, J. Chem. Eng. Data,
2009, 54, 1850.
23. C. E. C. Rodrigues, F. A. Silva, A. Marsaioli and A. J. A. Meirelles, J. Chem. Eng. Data, 2005,
50, 517.
24. C. E. C. Rodrigues, É. C. D. Reipert, A. F. de Souza, P. A. P. Filho and A. J. A. Meirelles, Fluid
Phase Equilibr., 2005, 238, 193.
25. C. A. S. da Silva, G. Sanaiotti, M. Lanza, A. J. A. Meirelles and E. A. C. Batista, J. Chem. Eng.
Data, 2010, 55, 2416.
26. J. Bezard, M. Bugaut and G. Clement, J. Am. Oil Chem. Soc., 1971, 48, 134.
27. R. C. Basso, F. H. Miyake, A. J. A. Meirelles and E. A. C. Batista, Fuel, 117, 590.
38
28. I. Dalmolin, M. A. Mazutti, E. A. C. Batista, M. A. A. Meireles and J. V. Oliveira, J. Chem.
Thermodyn., 2010, 42, 797.
29. J. R. d. O. Lima, R. B. da Silva, E. Miranda de Moura and C. V. Rodarte de Moura, Fuel, 2008,
87, 1718.
30. M. S. Cuevas, C. E. C. Rodrigues and A. J. A. Meirelles, J. Food Eng., 2009, 95, 291.
31. L. Ventolà, M. Ramírez, T. Calvet, X. Solans, M. A. Cuevas-Diarte, P. Negrier, D. Mondieig, J.
C. van Miltenburg and H. A. J. Oonk, Chem. Mater., 2002, 14, 508.
32. R. E. Timms, Prog. Lipid Res., 1984, 23, 1.
33. M. Santos, G. C. Roux and V. Gerbaud, J. Am. Oil Chem. Soc., 2011, 88, 223.
34. L. H. Wesdorp, Liquid-multiple solid phase equilibria in fats, theory and experiments, Doctoral
Thesis, Delft University of Technology, Delft, 1990.
35. M. C. Costa, L. A. D. Boros, J. A. P. Coutinho, M. A. Krähenbühl and A. J. A. Meirelles,
Energ. Fuels, 2011, 25, 3244.
36. M. C. Costa, M. Sardo, M. P. Rolemberg, J. A. P. Coutinho, A. J. A. Meirelles, P. Ribeiro-Claro
and M. A. Krähenbühl, Chem. Phys. Lipids, 2009, 160, 85.
37. M. C. Costa, M. Sardo, M. P. Rolemberg, P. Ribeiro-Claro, A. J. A. Meirelles, J. A. P. Coutinho
and M. A. Krähenbühl, Chem. Phys. Lipids, 2009, 157, 40.
38. E. Müller and H. Stage, Experimentelle vermessung von dampf-flüssigkeits-
phasengleichgewichten, Springer, Berlin, 1961.
39. B. J. Grondal and D. A. Rogers, Oil Soap, 1944, 21, 303.
40. F. Francis, F. J. E. Collins and S. H. Piper, Proc. Roy. Soc. A., 1937, 158, 691.
41. A. E. Bailey, Melting and solidification of fats, Interscience publishers, New York, 1950.
42. M. C. Costa, L. A. D. Boros, M. L. S. Batista, J. A. P. Coutinho, M. A. Krähenbühl and A. J. A.
Meirelles, Fuel, 2012, 91, 177.
43. L. Boros, M. L. S. Batista, R. V. Vaz, B. R. Figueiredo, V. F. S. Fernandes, M. . osta, . .
r henb hl, A. J. A. Meirelles and J. A. P. Coutinho, Energ. Fuels, 2009, 23, 4625.
44. M. C. Costa, M. P. Rolemberg, A. J. A. Meirelles, J. A. P. Coutinho and M. A. Krähenbühl,
Thermochim. Acta, 2009, 496, 30.
45. T. Inoue, Y. Hisatsugu, M. Suzuki, Z. Wang and L. Zheng, Chem. Phys. Lipids, 2004, 132, 225.
46. M. C. Costa, M. P. Rolemberg, L. A. D. Boros, M. A. Krähenbühl, M. G. de Oliveira and A. J.
A. Meirelles, J. Chem. Eng. Data, 2007, 52, 30.
39
47. E. S. Lutton and F. R. Hugenberg, J. Chem. Eng. Data, 1962, 7, 208.
48. J. C. Smith, J. Chem. Soc., 1931, 0, 802.
49. M. P. Rolemberg, Equilíbrio Sólido-Líquido de Ácidos Graxos e Triglicerídeos: Determinação
Experimental e Modelagem, Doctoral Thesis, University of Campinas, Campinas, 2002.
50. T. Inoue, Y. Hisatsugu, R. Ishikawa and M. Suzuki, Chem. Phys. Lipids, 2004, 127, 161.
51. H. Imahara, E. Minami and S. Saka, Fuel, 2006, 85, 1666.
52. C. A. Lockemann and E. U. Schluender, J. Chem. Eng. Data, 1993, 38, 432.
53. K. Nishimura, K. Maeda, H. Kuramochi, K. Nakagawa, Y. Asakuma, K. Fukui, M. Osako and
S.-i. Sakai, J. Chem. Eng. Data, 2011, 56, 1613.
54. T. Inoue, Y. Hisatsugu, R. Yamamoto and M. Suzuki, Chem. Phys. Lipids, 2004, 127, 143.
55. N. D. D. Carareto, Estudo do equilibrio solido-liquido em sistemas binarios de alcoois graxos
atraves da calorimetria exploratoria diferencia, Master Thesis, University of Campinas, Campinas,
2010.
56. M. C. Costa, L. A. D. Boros, J. A. Souza, M. P. Rolemberg, M. A. Krähenbühl and A. J. A.
Meirelles, J. Chem. Eng. Data, 2011, 56, 3277.
57. N. D. D. Carareto, M. C. Costa, M. P. Rolemberg, M. A. Krähenbühl and A. J. A. Meirelles,
Fluid Phase Equilibr., 2011, 303, 191.
58. D. G. Kolp and E. S. Lutton, J. Chem. Eng. Data, 1962, 7, 207.
59. M. C. Costa, L. A. D. Boros, M. P. Rolemberg, M. A. Krähenbühl and A. J. A. Meirelles, J.
Chem. Eng. Data, 2010, 55, 974.
60. M. C. Costa, Estudo do equilíbrio sólido-líquido de misturas de alcoóis graxos com ácidos
graxos, Post-Doctoral Report, University of Campinas, Campinas, 2011.
61. G. J. Maximo, N. D. D. Carareto, M. C. Costa, A. O. dos Santos, L. P. Cardoso, M. A.
Krähenbühl and A. J. A. Meirelles, Fluid Phase Equilibr., 2014, 366, 88.
62. J. Zeng, Z. Cao, D. Yang, F. Xu, L. Sun, L. Zhang and X. Zhang, J. Therm. Anal. Calorim.,
2009, 95, 501.
63. G. J. Maximo, Lipid Thermodinamics: New Perspectives on Phase Studies for Aplications in
Engineering, Doctoral Thesis, University of Campinas, Campinas, 2014.
64. F. G. Gandolfo, A. Bot and E. Flöter, Thermochim. Acta, 2003, 404, 9.
65. G. J. Maximo, M. C. Costa and A. J. A. Meirelles, Braz. J. Chem. Eng., 2013, 30, 33.
40
66. IUPAC, Compendium of Chemical Terminology, Blackwell Scientific Publications, Oxford,
1997.
67. E. Deffense, J. Am. Oil Chem. Soc., 1993, 70, 1193.
68. A. G. Marangoni and S. S. Narine, Food Res. Int., 2002, 35, 957.
69. S. S. Narine and A. G. Marangoni, Food Res. Int., 1999, 32, 227.
70. G. G. Chernik, Adv. Colloid Interfac., 1995, 61, 65.
71. M. C. Costa, M. P. Rolemberg, L. A. D. Boros, M. A. Krähenbühl, M. G. Oliveira and A. J. A.
Meirelles, J. Chem. Eng. Data, 2007, 52, 30.
72. U. Domańska, J Solution Chem., 1989, 18, 173.
73. M. Benziane, K. Khimeche, A. Dahmani, S. Nezar and D. Trache, J. Therm. Anal. Calorim.,
2013, 112, 229.
74. U. Domańska, P. Morawski and M. Piekarska, J. Chem. Eng. Data, 2009, 54, 1271.
75. S. Benessam, K. Khimeche, F. Djellouli, M. Benziane and A. Dahmani, J. Therm. Anal.
Calorim., 2013, 112, 317.
76. G. G. Chernik, Curr. Opin. Colloid In., 1999, 4, 381.
77. R. Koynova and M. Caffrey, Chem. Phys. Lipids, 2002, 115, 107.
78. G. J. Maximo, R. J. B. N. Santos, J. A. Lopes da Silva, M. C. Costa, A. J. A. Meirelles and J. A.
P. Coutinho, ACS Sustainable Chem. Eng., 2014 (in press)
79. I. Dierking, Textures of liquid crystals, Wiley-VCH Weinheim, 2003.
80. M. C. Costa, M. P. Rolemberg, A. O. dos Santos, L. P. Cardoso, M. A. Krähenbühl and A. J. A.
Meirelles, J. Chem. Eng. Data, 2010, 55, 5078.
81. K. W. Won, Fluid Phase Equilibr., 1993, 82, 261.
82. R. E. Timms, Eur. J. Lipid Sci. Tech., 2005, 107, 48.
83. H. M. Schaink, K. F. van Malssen, S. Morgado-Alves, D. Kalnin and E. van der Linden, Food
Res. Int., 2007, 40, 1185.
84. F. G. Gandolfo, A. Bot and E. Floter, Journal of the American Oil Chemists Society, 2004, 81,
1.
85. G. J. Suppes, M. J. Goff and S. Lopes, Chem. Eng. Sci., 2003, 58, 1751.
86. L. Shilei, Z. Neng and G. H. Feng, Energ Buildings, 2006, 38, 708.
87. J. J. Zhang, J. L. Zhang, S. M. He, K. Z. Wu and X. D. Liu, Thermochim. Acta, 2001, 369, 157.
41
88. J. V. Gerpen, Fuel Process. Technol., 2005, 86, 1097.
89. . . . outinho, . on alves, M. J. Pratas, M. L. S. Batista, V. F. S. Fernandes, J. Pauly and
J. L. Daridon, Energ. Fuels, 2010, 24, 2667.
90. J. Daniel and R. Rajasekharan, J. Am. Oil Chem. Soc., 2003, 80, 417.
91. D. Feldman, M. M. Shapiro, D. Banu and C. J. Fuks, Sol. Energ. Mater., 1989, 18, 201.
92. R. C. Reid, J. M. Prausnitz and B. E. Pouling, The properties of gases and liquids, McGraw-
Hill, New York, 1987.
93. J. Gmehling, B. Kolbe, M. Kleiber and J. Rarey, Chemical Thermodynamics for Process
Simulation, Wiley-VHC, Weinheim, 2012.
94. D. S. Abrams and J. M. Prausnitz, AIChE J., 1975, 21, 116.
95. A. Fredenslund, R. L. Jones and J. M. Prausnitz, AIChE J., 1975, 21, 1086.
96. J. G. Gmehling, T. F. Anderson and J. M. Prausnitz, Ind. Eng. Chem. Fund., 1978, 17, 269.
97. J. A. P. Coutinho, Ind. Eng. Chem. Res., 1998, 37, 4870.
42
43
SOLID-LIQUID EQUILIBRIUM OF TRIOLEIN WITH FATTY
ALCOHOLS
(published in Brazilian Journal of Chemical Engineering v. 30, n. 1, p. 33-43, 2013)
Guilherme J. Maximo1, Mariana C. Costa
2 and Antonio J. A. Meirelles
1*
1 Laboratory of Extraction, Applied Thermodynamics and Equilibrium, Department of Food
Engineering, School of Food Engineering, Phone: + (55) (19) 3521 4037, Fax: + (55) (19)
3521 4027, University of Campinas, Av. Monteiro Lobato 80, 13083-862, Campinas - SP,
Brazil. E-mail: [email protected] 2
School of Applied Sciences, University of
Campinas, R. Pedro Zaccaria 1300, 13484-350, Limeira - SP, Brazil.
Chapter 2
44
ABSTRACT
Triacylglycerols and fatty alcohols are used in the formulation of cosmetic, pharmaceutical
and food products. Although information about the phase transitions of these compounds,
and their mixtures, is frequently required for design and optimization of processes, and
product formulation, these data are still scarce in the literature. In the present study, the
solid-liquid phase diagrams of two binary systems composed of triolein + 1-hexadecanol
and triolein + 1-octadecanol were evaluated by differential scanning calorimetry (DSC) and
optical microscopy. The experimental data were compared with predicted data by solving
the phase equilibrium equations using an algorithm implemented in MATLAB. The liquid-
phase activity coefficients were calculated using the Margules equation (two- and three-
suffix) and the UNIFAC model (original and modified Dortmund model). The approaches
used for calculating the equilibrium condition allowed gave accurate predictions of the
liquidus line, with deviations from the experimental data lower than 0.57 K.
KEYWORDS: Triolein; Fatty alcohol; Solid-liquid equilibrium; Differential scanning
calorimetry; Modeling; UNIFAC.
1. INTRODUCTION
Fats and oils are multicomponent systems composed of a mixture of triacylglycerols
(TAGs) and minor components, such as free fatty acids and nutraceuticals. Under thermal
treatment, triacylglycerols are responsible for the formation of a crystalline network and,
consequently, particular microstructures that influence the sensorial, rheological and
45
thermodynamic profiles of food and chemical products (Bruin, 1999; O’Brien, 2004).
Within the food and chemical industry, there is a clear trend towards the development of
new ways of structuring these systems, such as the utilization of surfactants and co-
surfactants. In particular, fatty alcohols have been evaluated as gelling and surfactant agents
in food, chemical and pharmaceutical oil-based products (Egan et al., 1984; Kreutzer, 1984;
Daniel and Rajasekharan, 2003). Therefore, the study of the phase equilibrium of these
compounds is a powerful tool for process and product design.
The differential scanning calorimetry technique (DSC) is often used for the
experimental determination of solid-liquid equilibrium (SLE). Some data on solid-liquid
transitions of fatty mixtures, phase diagrams and pure component properties using this
technique, including TAGs, fatty acids, esters and alcohols, can be found in the literature
(Wesdorp, 1990; Costa et al., 2007, 2009, 2010a, 2010b, 2011; Carareto et al., 2011). The
phase transition behavior can be modeled by solving a system of equations based on the
conditions of thermodynamic equilibrium (Wesdorp, 1990; Dos Santos et al., 2011).
Several approaches for this procedure can be found in the literature; the majority is based
on evaluation of the activity coefficients of the compounds in the equilibrium phases
(Wesdorp, 1990; Slaugther and Doherty, 1995; Coutinho et al., 1998). The activity
coefficients in the solid-liquid equilibrium can be estimated with adjustable models such as
the Margules equation, or with predictive methods such as the group-contribution UNIFAC
model (Fredenslund et al., 1975; Gmehling et al., 1978; Chiavone Filho and Rasmussen,
2000; Coutinho et al., 2001), both based on the calculation of excess Gibbs free energy.
Binary solid-liquid equilibrium data on systems composed of mixtures of
triacylglycerols, mixtures of fatty alcohols or mixtures of these components with other fatty
compounds are available in the literature, but data on systems composed of triacylglycerols
46
+ fatty alcohols are still scarce. In the present work, solid-liquid equilibrium data on two
fatty mixtures, triolein + 1-octadecanol and triolein + 1-hexadecanol, were measured by
differential scanning calorimetry (DSC), evaluated with optical microscopy and modeled
by resolution of the thermodynamic equilibrium equations using the two- and three-suffix
Margules equations, the original UNIFAC (Fredenslund et al., 1975) and the modified
UNIFAC-Dortmund model (Gmehling et al., 1993) for calculation of the liquid-phase
activity coefficient. An optimization algorithm was implemented in MATLAB for
prediction of the melting point.
1.1. Thermodynamic framework
The solid-liquid equilibrium of component i in a binary or two-phase system is
represented by Equation (1), as discussed by Prausnitz et al. (1986).
L Lfusi i
S Sfusi i
1 1exp
Hx
R T Tx
(1)
where Lix and S
ix are the mole fractions of component i in the liquid (L) and solid (S)
phases, Li and S
i are the activity coefficients of component i in the same phases, fusH is
the melting enthalpy, and fusT is the melting temperature of pure component i, T is the
temperature of the system, and R is the ideal-gas constant (8.314 Jmol-1K
-1). The Margules
equation and the group-contribution UNIFAC model are often used for calculation of i
Using the two-suffix Margules equation for a binary mixture, the of component 1 is
given by Equations (2) and (3). Using the three-suffix Margules equation, the of the
47
components in a binary mixture are calculated using Equations (4) and (5) (Reid et al.,
1987).
21 2R lnT Ax (2)
22 1R lnT Ax (3)
2 31 2 2R ln ( 3 ) 4T A B x Ax (4)
2 32 1 1R ln ( 3 ) 4T A B x Ax (5)
where A and B are adjustable parameters. According to the UNIFAC model (Equations (6)
to (17)) (Fredenslund et al., 1975), the activity coefficient can be written as the sum of a
combinatorial contribution Ci , which accounts for the effects of size and shape, and a
residual contribution Ri , which accounts for interactions among groups.
C Ri i iln ln ln (6)
i i i
i i i
Ci i i j j jln ln 5 ln
x xq l x l
(7)
(i) (i)Ri k kk kln (ln ln ) (8)
i i i i5( ) ( 1)l r q r (9)
i ii
j j j
r x
x r
(10)
i ii
j j j
q x
x q
(11)
(i)i k kkr R (12)
(i)i k kkq Q (13)
48
m km
n nmk k m m mk m n
ln 1 ln( )Q
(14)
m mm
n n n
Q X
Q X
(15)
( j)j m j
m ( j)j n n j
xX
x
(16)
nmnm exp
a
T
(17)
where i is the volume fraction of molecule i, i is the area fraction of molecule i, ri is the
molecular volume parameter, qi is the surface area parameter, ik is the number of k-type
groups in molecule i, k is the activity coefficient of group k in the mixture composition,
(i)k is the activity coefficient of group k for a group composition corresponding to pure
component i, i is the surface area fraction of component i in the mixture, mX is the group
mole fraction of group m, nm is the Boltzmann factor, Rk and Qk are the structural
parameters for the groups and anm is the temperature-dependent group-interaction
parameter (Reid et al., 1987). In this work, in addition to the original UNIFAC model, the
modified UNIFAC-Dortmund model (Gmehling et al., 1993) was also used. In this model,
the combinatorial term Ci (Equation (18)) was empirically changed to make it possible to
deal with compounds that are very different in size. Also, the Boltzmann factor nm
(Equation (20)) of the residual term is temperature-dependent, with three new group-
interaction parameters, anm, bnm and cnm.
C i i i ii i
i i i i
ln 1 ln 5 1 lnqx x
(18)
49
3/4i i
i 3/4j j j
x r
x r
(19)
2nm nm nm
nm expa b T c T
T
(20)
For systems that show only a eutectic transition and whose components are
immiscible in the solid phase, it is possible to assume that the solid phase is composed of a
pure component, so Equation (21) becomes valid.
S Si i 1x (21)
With these considerations, a system of equations can be constructed for prediction
of the melting point using: i) the equation expressing the equilibrium condition for each
component (Equation (1)), ii) a model for calculation of the activity coefficients of the
liquid phase (Equations (2) to (20)) and iii) the global molar balance of the system.
2. MATERIALS AND METHODS
Triolein (C57H104O6), 1-octadecanol (C18H38O) and 1-hexadecanol (C16H34O) with
purity higher than 99% by mass were supplied by Sigma-Aldrich (USA). Samples (1 g) of
triolein + 1-octadecanol and triolein + 1-hexadecanol were prepared gravimetrically in
amber flasks using an analytical balance (Precisa Gravimetrics AG, Dietkon) with a
precision of 2×10-4
g and a mole fraction of component 1 from 0.0 to 1.0. The samples were
pretreated on Thermoprep (Metrohm, Herisau) by heating in a nitrogen atmosphere to 15 K
above the highest melting point of the components. The mixtures were sealed under inert
50
atmosphere, allowed to cool to room temperature, and kept in a freezer at 273.15 K until
DSC analysis.
2.1. Differential Scanning Calorimetry (DSC)
SLE of pure triolein, fatty alcohols and their mixtures were characterized by DSC
using a DSC8500 calorimeter (PerkinElmer, Waltham) that was equipped with a
refrigerated cooling system, which in this work operated between 223 and 343 K using
nitrogen as a purge gas. Indium (certificated by PerkinElmer, Waltham) and cyclohexane
(C6H12) (Merk, USA), both with 0.99 mass fraction purity, had previously been used as
primary standards to calibrate the melting temperature and heat flow at a heating rate of 1
Kmin−1
. Samples (2-5 mg) of each mixture, triolein + 1-hexadecanol and triolein+1-
octadecanol, were weighed in a microanalytical balance AD6 (PerkinElmer, Waltham) with
a precision of 2×10−6
g and put in sealed aluminum pans. In order to erase previous thermal
histories related to the effects of polymorphism, a pretreatment was carried out based on the
methodology described by Costa et al. (2007). For this purpose, each sample was kept at a
temperature of 15 K above the highest melting point of the components. After 20 min at
this temperature, the samples were cooled to 40 K below the lowest melting point of the
components at a cooling rate of 1 Kmin−1
and allowed to remain at this temperature for 20
min. After this pretreatment, each sample was analyzed in a heating run at a heating rate of
1 Kmin−1
. Measurements were evaluated using PerkinElmer (Waltham) analysis software.
The temperatures of the thermal transitions were the peak top temperatures and the
enthalpies were calculated from the area of the corresponding peaks.
51
2.2. Optical Microscopy
In order to evaluate the behavior of the solid:liquid fraction of the systems as a
function of the concentration of the components, as well as the assumption of solid phase
immiscibility, both systems were analyzed at 298.15 K by optical microscopy using a Carl
Zeiss model Scope A1 microscope (Zeiss, Germany). Samples of each mixture of both
systems were heated from storage temperature (above 273.15 K) to 298.15 K on a
Thermoprep (Metrohm, Herisau). The images obtained were taken with magnifications of
10 and 40.
2.3. Modeling Procedure
SLE modeling was carried out according thermodynamic approach previously
discussed. The binary parameters A and B of the Margules equations were numerically
adjusted by an algorithm implemented in MATLAB 7.0, and based on the calculation of the
melting temperatures by Equation (1), rewritten as a function of T, and by Equations (2) to
(5) as the Li model. The one-variable optimization problem (using the two-suffix Margules
equation) was implemented using a parabolic interpolation method (MATLAB function
fminbnd). The two-variable optimization problem (using the three-suffix Margules
equation) was carried out using a sequential quadratic programming method (MATLAB
function fmincon). In both cases, the objective function was the root mean square deviation
(RMSD) (Equation (22)) between the experimental (Texp) and calculated temperatures
(Tcalc).
52
2i 1 exp calc i2
( )RMSD
n T T
n
(22)
where n is the number of observations. In the case of the UNIFAC or the UNIFAC-
Dortmund model for calculation of Li , equilibrium Equation (1) cannot be rewritten as a
function of T, and thus an optimization algorithm was implemented in MATLAB 7.0 using
sequential quadratic programming (MATLAB function fmincon) for the calculation of the
melting temperature of the mixtures. In this case, the objective function was the square of
the absolute deviation between the experimental temperature (Texp) and the calculated one
(Tcalc) (Equations (23) and (24)) for a composition Lix . The routine was repeated for 0 < L
ix
< 1 and the first iteration started with a melting temperature sufficiently close to that of the
optimal solution, taken as the melting point of pure component 1 or 2.
2obj exp calc iF ( )T T (23)
fus fuscalc L L
fus fus i iR ln
H TT
H T x
(24)
For modeling, the pure component properties (temperature and enthalpy of fusion)
were taken from the experimental data measured in this work. Also, data from the literature
(Wesdorp, 1990; Ventolà et al., 2004; Nichols et al., 2006) was used for evaluation of the
significance of these properties for modeling accuracy. Accounting for the immiscibility in
the solid phase, the procedure described above was applied along the entire composition
range for calculating the mixtures’ melting points, assuming that either the light or the
heavy component represents the pure solid phase. Therefore, two temperatures were
obtained, with the equilibrium temperature (liquidus line) being the higher. This is the usual
procedure for estimating equilibrium temperatures in a simple eutectic mixture, so the left-
53
hand curve corresponds to a solid phase composed of the pure heavy component and the
right-hand curve to a solid phase of the pure light component and the predicted eutectic
mixture occurs exactly at their intersection.
3. RESULTS AND DISCUSSION
3.1. Performance of the Algorithm
To evaluate the performance of the algorithm, solid-liquid equilibrium experimental
data on fatty mixtures, and other organic binaries, from the literature were calculated by the
aforementioned modeling procedure. The deviations obtained were compared with
deviations obtained by the literature works’ modeling procedure (Table 1) and plotted for
evaluation (Figure 1). The agreement between the data from the literature and the data in
this study was good since the fitted curve (continuous line in Figure 1) determined by linear
regression shows only a slight deviation from the straight diagonal line. Apart from
differences in numerical procedures, the observed deviations are probably due to
differences in pure component properties, which have an important effect on the results of
the equilibrium estimation, as will be discussed below.
3.2. Experimental and Predicted Solid-Liquid Equilibrium Data
The reproducibility of the data was checked by performing three repeat runs for the
calibration standard (indium) and for the pure components (triolein, 1-hexadecanol and 1-
octadecanol). The standard deviation of the melting point data obtained was 0.13 K. This
54
deviation was assumed to be the uncertainty of all of the melting points measured in this
work. Also, these experimental data were compared with experimental data obtained from
the literature (Ferguson and Lutton, 1947; Kolp and Lutton, 1951; Wesdorp, 1990;
Rolemberg, 2002; Ventolà et al., 2004; Carareto et al., 2011; National Bureau of Standards,
2011). The average relative deviations (ARDs) (%) of the melting data measured in this
work (Twork
) in relation to data found in the literature (Tlit
) were calculated according to
Equation (25), where n is the number of observations. The agreement between both sets of
data was good, since the ARD was 0.14 %.
lit worki in
i liti
100 1ARD (%)
n
T T
T
(25)
Table 1. Deviationa data from the literature (lit.) and from the algorithm proposed in this work
System
Margules
two-suffix
Margules
three-suffix UNIFAC
UNIFAC -
Dortmund Ideal
lit. this
work lit.
this
work lit.
this
work lit.
this
work lit.
this
work
palmitic acid + tristearinb 0.36 0.34 0.16 0.14 - - 0.46 0.46 - -
trilinolein + stearic acidb 1.07 1.06 1.02 0.94 - - 5.20 5.92 - -
capric acid + palmitic acidb 0.86 0.84 0.56 0.57 - - 2.89 2.89 - -
capric acid + stearic acidb 0.78 0.76 0.67 0.67 - - 1.68 1.68 - -
triolein + tripalmitinb 0.40 0.39 0.10 0.10 - - 0.57 0.57 - -
1-octanol + 1-dodecanolc 2.00 1.90 1.80 1.85 - - 2.60 2.58 2.60 2.46
1-dodecanol + 1-octadecanolc 1.00 1.04 0.40 0.40 - - 4.30 3.94 4.30 4.05
1-decanol + 1-hexadecanolc 1.10 1.15 0.80 0.70 - - 7.30 7.31 7.30 7.53
2-aminobenzoic acid + benzoic acidd - - - - 6.95 7.10 - - - -
tetrachloromethane + 2-dodecanonee - - - - 2.75 2.94 - - - -
tetrachloromethane + 2-undecanonee - - - - 1.04 1.09 - - - -
indane + dodecanee - - - - 0.50 0.74 - - - -
a RMSD (Equation (22)) data from
bBoros (2005),
cCarareto et al. (2011),
dHuang and Chen (2000)
and average of the absolute deviations ni calc exp i
nT T from eJakob et al. (1995).
55
0
1
2
3
4
5
6
7
8
0 1 2 3 4 5 6 7 8
Dev
iati
on
fro
m t
his
wo
rk
Deviation from literature
Figure 1: Deviation (this work) versus deviation from the literature (data from Table 2): (○)
Carareto et al., (2011), (×) Boros (2005), (+) Huang and Chen (2000) and (Δ) Jakob et al. (1995).
Dashed line represents the straight diagonal line and the continuous line is the curve fitted with the
data (y = 1.0123x – 0.017, r2 = 0.9966). Number of observations = 31.
The thermograms obtained with DSC runs and the phase diagrams of the two binary
systems evaluated, (1) triolein + (2) 1-hexadecanol and (1) triolein + (2) 1-octadecanol, are
presented in Figures 2 and 3. Figures 2 and 3 also show modeling results with the ideal
assumption, using the adjustable three-suffix Margules equation and the predictive
UNIFAC-Dortmund model for Li calculation. In addition, the experimental data related to
the thermogram analysis are presented in Tables 2 and 3.
For pure 1-hexadecanol and 1-octadecanol two overlapping peaks could be seen in
the region very close to the melting temperature, as observed by other authors (Ventolà et
al., 2004; Carareto et al., 2011). In fact, fatty alcohols have rotator forms during
crystallization processes (TFA rotator form, Tables 2 and 3) that are structurally different but
with very similar thermodynamic properties (Ventolà et al., 2004). Despite the fact that the
melting temperatures of these rotator forms could easily be found on the thermograms, the
56
proximity of their values makes determination of the enthalpy of each of these pure
component forms difficult.
For pure triolein two well-defined peaks and a third mild peak at a lower
temperature were observed. In fact, TAGs usually have three polymorphic forms, known as
alpha, beta-prime and beta, with differences in the crystal lattice packing, temperatures and
heats of fusion (Wesdorp, 1990; Rolemberg, 2002; Costa, 2008). According to Fergusson
and Lutton (1947), triolein also has these three polymorphic forms, melting at 241.15 K,
260.15 K and 278.65 K, respectively, and these values are very similar to the temperatures
found in this work (Tables 2 and 3).
The thermograms of the mixtures of triolein and 1-hexadecanol or 1-octadecanol
clearly showed three well-defined peaks, which were also observed by other authors for
fatty mixtures with triacylglycerols (Costa et al., 2010a, 2010b, 2011). The thermal
transition at the highest temperature is related to the solid-liquid equilibrium temperature of
the mixture. The transition at the lowest temperature is close to the solid-solid transition of
the triolein beta-prime polymorphic form, and thus is probably related to this transition.
The last one (Ttr, Tables 2 and 3) is a transition temperature close to the triolein melting
point, which will be discussed below. The transition related to the alpha polymorphic form
of triolein, the polymorphic form with the lowest transition temperature, was not well
identified in the thermograms of the mixtures and is not shown in Figures 2 and 3.
57
Table 2. Experimental solid-liquid equilibrium data (K) for triolein (1) + 1-hexadecanol (2).
xtriolein Tfus TFA rotator form Ttr T’melting T melting
0.0000 322.34 321.62
0.1018 320.40 277.18 259.74
0.1975 319.07 277.41 260.12
0.3000 317.92 277.80 260.44
0.4017 316.48 278.10 260.56
0.5014 314.51 278.15 260.62
0.5963 312.42 278.28 260.78
0.7006 307.58 277.78 260.20
0.7840 304.03 278.29 260.47
0.8833 296.43 278.00 260.47
0.9463 287.33 277.82 259.74
1.0000 278.16 259.31 241.34
240
260
280
300
320
340
0.0 0.2 0.4 0.6 0.8 1.0
Tem
per
atu
re /
K
x1
x1 = 0.0
x1 = 1.0
Heat Flow / u.a.
x1 = 0.80
x1 = 0.90
x1 = 0.95
Temperature / K280 295
286 300
290 315
Hea
t F
low
/ u
.a.
Temperature / K
Hea
t F
low
x1 = 1.0
248233
Figure 2: Solid-liquid phase diagram (temperature in Kelvin and mole fraction) and thermograms
for the triolein (1) + 1-hexadecanol (2) system. Symbols for phase diagrams: (○) experimental
melting point, (■) transition near triolein melting point, solid-solid triolein () alpha and (×) beta-
prime transitions and data modeled by (- - -) the three-suffix Margules equation, (—) the UNIFAC-
Dortmund model and (- -) the ideal assumption. In detail, magnifications of the thermograms close
to the melting temperature for x1 > 0.7 and close to the triolein alpha polymorphic form temperature
transition for x1 = 1.0.
58
Table 3. Experimental solid-liquid equilibrium data (K) for triolein (1) + 1-octadecanol (2).
xtriolein Tfus TFA rotator form Ttr T’melting T melting
0.0000 331.04 330.35
0.1048 329.93 277.65 258.87
0.2030 328.66 277.87 259.80
0.3097 326.50 277.51 260.53
0.4046 325.23 277.52 260.69
0.5004 323.49 277.61 259.73
0.5910 321.13 278.19 261.29
0.6993 317.70 278.24 260.77
0.7996 313.58 278.03 260.72
0.8975 305.38 278.00 260.13
0.9467 296.60 278.72 259.74
1.0000 278.16 259.31 241.34
240
260
280
300
320
340
0.0 0.2 0.4 0.6 0.8 1.0
Tem
per
atu
re /
K
x1
x1 = 0.0
x1 = 1.0
Heat Flow / u.a.
x1 = 0.80
Temperature / K
295.5 298
290 314
300 325
Hea
t fl
ow
/ u
.a.
x1 = 0.90
x1 = 0.95
Figure 3: Solid-liquid phase diagram (temperature in Kelvin and mole fraction) and thermograms
for the triolein (1) + 1-octadecanol (2) system. Symbols for phase diagrams: (○) experimental
melting point, (■) transition near triolein melting point solid-solid triolein () alpha and (×) beta-
prime transitions and data modeled by (- - -) the three-suffix Margules equation (—), the UNIFAC-
Dortmund model and (- -) the ideal assumption. In detail, magnification of the thermograms close
to the melting temperature for x1 > 0.7.
59
Figure 4 contains the Tammann plots of the mixtures, showing the enthalpy
variation of the observed invariant transitions as a function of triolein molar concentration.
In general, the polymorphic transitions and the eutectic reaction, which separates the solid-
liquid region from the solid-solid, are the most common invariant thermal events of
mixtures of triacylglycerols and fatty compounds. The Tammann plot allows a better
determination of the eutectic point, since enthalpy increases as the mixture concentration
approximates the eutectic composition, where the enthalpy should begin to decrease. In
several cases, the increase in the difference between melting temperatures of pure
components increases the proximity of the eutectic point to the melting temperature of the
pure lighter component (Levine, 2002). The linear regression (with r2 > 0.99) of the
Tammann curves for the transition close to the triolein melting point (Figure 4) shows that,
for x = 1, the enthalpy variation is very close to the corresponding value for pure triolein.
This means that this invariant thermal event may correspond to a eutectic reaction with a
composition very close to that of pure triolein, making it virtually impossible to
differentiate the eutectic composition from pure triolein. In agreement with this
observation, triolein + 1-octadecanol and triolein + 1-hexadecanol modeled phase diagrams
had eutectic points higher than 0.99 and 0.98 triolein mole fraction, respectively, for all
Li models.
60
y = 117.69x 0.76
r² = 0.99990.0
20.0
40.0
60.0
80.0
100.0
120.0
0.0 0.2 0.4 0.6 0.8 1.0
H
( k
Jm
ol-1
)
x1
y = 115.05x + 0.14r² = 0.9998
0.0
20.0
40.0
60.0
80.0
100.0
120.0
0.0 0.2 0.4 0.6 0.8 1.0
H
( k
Jm
ol-1
)
x1
(a) (b)
Figure 4: Tammann plots of (a) triolein (1) + 1-hexadecanol (2) and (b) triolein (1) + 1-octadecanol
(2). (—) Linear regression for data on (■) transition close to the triolein melting point, (○) pure
triolein transition and (- - -) hypothetical plot with a eutectic point.
Table 4. Estimated parameters (A and B) and mean absolute deviationsa of the predicted data
from the experimental data (in Kelvin).
Model triolein +
1-hexadecanol
triolein +
1-octadecanol Average
Two-suffix Margules A 1065.77 855.15
0.53 0.60 0.56
Three-suffix Margules A 1186.79 1026.36
B 396.02 507.89
0.32 0.34 0.33
UNIFAC 0.45 0.42 0.43
UNIFACDortmund 0.68 0.47 0.57
Ideal 2.31 1.78 2.04
a
exp predni i i
n
T T
The binary parameters (Jmol-1
) for both Margules equations and the deviations
between calculated and experimental data for each modeling approach are shown in Table
4. It can easily be observed that the liquidus line was well described by all models, since
61
the average absolute deviation was always less than 0.57 K. The Margules parameter values
obtained were very similar to those found in the literature for binary fatty mixtures with
TAGs and for systems with fatty alcohols (Costa et al., 2010a, 2011; Carareto et al., 2011).
The ideal assumption shows the highest deviations, indicating that the systems of triolein +
1-hexadecanol and triolein + 1-octadecanol are slightly non-ideal. In fact, the activity
coefficients calculated by all the models were higher than 1 for both systems and the
highest values were obtained for the most accurate model, the three-suffix Margules
equation. In this case the Li values were close to 2 when L
ix was close to 0.
Solid phase immiscibility is usually assumed for the purpose of modeling fatty
systems (Wesdorp, 1990; Rolemberg, 2002; Costa et al., 2007). Nevertheless, several fatty
mixtures have regions of solid solution close to the pure component composition on the
phase diagram (Costa et al., 2009, 2011; Carareto et al., 2011). In order to check the
immiscibility assumption and thus comprehend the behavior of the solid:liquid fraction in
the solid-liquid biphasic region (under the liquidus line), optical micrographs of both
systems were taken at 298.15 K. Some of the images of the binary mixture of triolein (1) +
1-octadecanol (2) are shown in Figure 5. Firstly, it was clearly observed that increasing the
concentration of triolein decreased the solid fraction, in agreement with the lever rule
(Figures 5(a) to 5(f)). Since the melting points of the pure saturated fatty alcohols 1-
hexadecanol and 1-octadecanol are higher than that of triolein, in this region of the phase
diagram the solid fraction of the system is probably composed only of pure fatty alcohol.
Secondly, the micrograph of the system with a triolein mole fraction of 0.1 at a
magnification of 40 (Figure 5(g)) clearly shows that there is a liquid fraction even at this
relatively low concentration. Thus, there is probably no solid solution at the concentration
62
close to the pure heavy component (x1 = 0.0). Tammann plots of both systems also
corroborate this probability, since enthalpy values tend to zero at x1 = 0.0. Both
observations validate our modeling approach.
(a)
(b)
(c)
x1 = 0.3
10
x1 = 0.1
10
x1 = 0.0
10
x1 = 0.1
40
x1 = 0.7
10
x1 = 0.9
10
x1 = 1.0
10
(d)
(f)
(e)
(g)
Figure 5. Optical micrographs of samples of triolein (1) + 1-octadecanol (2) with magnifications of
10 ((a) to (f)) and 40 (g) at 298.15 K.
Finally, the influence of pure component properties on the accuracy of the modeling
approach was investigated. Component properties taken from the literature (Wesdorp,
1990; Ventolà et al., 2004; Nichols et al., 2006) as well as the values obtained in this work
are presented in Table 5. The values from the literature were also used with the modeling
63
procedure presented above. In the case of the pure component properties taken from the
literature, the global average absolute deviation between the calculated and the
experimental results, including the ideal solution assumption, was 0.90 K, while for the
pure component properties obtained in this work the deviation was 0.79 K. Firstly, the
differences in the pure compounds’ experimental values observed in Table 5 can probably
be attributed to differences between the experimental method used in the literature and in
this work (different heating and cooling rates and isothermal conditions). In addition, the
effects of polymorphism could produce peaks that overlap or are too close, which disrupts the
determination of enthalpy, especially for fatty alcohols, as discussed previously. In case of the
interactions parameters and the corresponding liquid phase activity coefficients obtained by using
literature’s pure compounds properties, an increasing of the positive deviations was observed, when
Li values were compared with that obtained in this work, as previously discussed. Literature data
presented Li values slightly higher than 2 when L
ix was close to 0 for three-suffix Margules
equation, also the most accurate model in this case. This highlights the fact that pure component
properties are important for a reliable modeling approach, being significant for the solid-liquid
equilibrium calculation.
Table 5. Pure fatty components properties from the literature (lit.) and in this work.
Component Temperature (K) Enthalpy (kJmol
-1)
lit. this work lit. this work
Triolein 277.95 a 278.15 100.000
a 115.560
1hexadecanol 321.60 b 322.34 58.400
c 56.591
1octadecanol 330.30 b
331.04 69.600 c
65.642
a Wesdorp (1990),
b Ventolà et al. (2004),
c Nichols et al. (2006).
64
4. CONCLUSIONS
In the present work the solid-liquid phase diagrams for the binary mixtures of
triolein + 1-hexadecanol and triolein + 1-octadecanol were measured using differential
scanning calorimetry. Both diagrams exhibited three well-defined transitions: the mixture
melting temperature, the solid-liquid transition related to the melting of triolein or of a
triolein-rich eutectic mixture and a solid-solid transition related to the triolein beta-prime
polymorph. The two- and three-suffix Margules equations and the group-contribution
UNIFAC and UNIFAC-Dortmund models were accurate for estimating the activity
coefficients of the liquid phase and provided a melting point of the fatty mixtures evaluated
with a global mean absolute deviation of 0.47 K. Micrographs showed that the assumption
of solid phase immiscibility was valid for the systems investigated. The present work also
showed that pure component properties are very important for developing a reliable
modeling approach.
ACKNOWLEDGMENTS
The authors are grateful to FAPESP (2008/56258-8 and 2009/54137-1) and CNPq
(304495/2010-7, 480992/20009-6 and 140718/2010-9) for the financial support.
REFERENCES
Boros, L. A. D., Modelagem matemática e termodinâmica do equilíbrio sólido-líquido de sistemas
graxos. M.Sc. Thesis, University of Campinas (2005). (In Portuguese).
65
Bruin, S., Phase equilibria for food products and process design. Fluid Phase Equilibria, 158, 657
(1999).
Carareto, N. D. D., Costa, M. C., Rolemberg, M. P., Krähenbühl, M. A. and Meirelles, A. J. A., The
solid-liquid phase diagrams of binary mixtures of even saturated fatty alcohols. Fluid Phase
Equilibria, 303, No. 2, 191 (2011).
Chiavone Filho, O. and Rasmussen, P., Modeling of salt solubilities in mixed solvents. Brazilian
Journal of Chemical Engineering, 17, No. 2, 117 (2000). (In Portuguese).
Costa, M. C., Determinação experimental do equilíbrio-sólido líquido de sistemas binários de
ácidos graxos saturados: Estudo detalhado da fase sólida. Ph.D. Thesis, University of Campinas
(2008). (In Portuguese).
Costa, M. C., Rolemberg, M. P., Boros, L. A. D., Krähenbühl, M. A., Oliveira M.G. and Meirelles
A. J. A., Solid-liquid equilibrium of binary fatty acid mixtures. Journal of Chemical & Engineering
Data, 52, No. 1, 30 (2007).
Costa, M. C., Sardo, M., Rolemberg, M. P., Coutinho, J. A. P., Meirelles, A. J. A., Ribeiro-Claro, P.
and Krähenbühl, M. A., The solid–liquid phase diagrams of binary mixtures of consecutive, even
saturated fatty acids. Chemistry and Physics of Lipids, 160, No. 2, 85 (2009).
Costa, M. C., Boros, L. A. D, Rolemberg, M. P., Krähenbühl, M. A. and Meirelles, A. J. A., Solid-
liquid equilibrium of saturated fatty acids + triacylglycerols. Journal of Chemical & Engineering
Data, 55, No. 2, 974 (2010a).
Costa, M. C., Rolemberg, M. P., dos Santos, A. O., Cardoso, L. P., Krähenbühl, M. A. and
Meirelles, A. J. A., Solid-liquid equilibrium of tristearin with refined rice bran and palm oils.
Journal of Chemical & Engineering Data, 55, No. 11, 5078 (2010b).
Costa, M. C., Boros, L. A. D, Souza, J. A., Rolemberg, M. P., Krähenbühl, M. A. and Meirelles, A.
J. A., Solid-liquid equilibrium of binary mixtures containing fatty acids and triacylglycerols. Journal
of Chemical & Engineering Data, 56, No. 8, 3277 (2011).
Coutinho, J. A. P., Predictive UNIQUAC: A new model for the description of multiphase solid-
liquid equilibria in complex hydrocarbon mixtures. Industrial & Engineering Chemistry Research,
37, No. 12, 4870 (1998).
Coutinho, J. A. P., Pauly, J. and Daridon, J. L., A thermodynamic model to predict wax formation
in petroleum fluids. Brazilian Journal of Chemical Engineering, 18, No.4, 411 (2001).
Daniel, J. and Rajasekharan, R., Organogelation of plant oils and hydrocarbons by long-chain
saturated FA, fatty alcohols, wax esters, and dicarboxylic acids. Journal of the American Oil
Chemists’ Society, 80, No. 5, 417 (2003).
Dos Santos, M. T., Le Roux, G. A. C. and Gerbaud, V., Phase equilibrium and optimization tools:
application for enhanced structured lipids for foods. Journal of the American Oil Chemists’ Society,
88, No. 2, 223 (2011).
66
Egan, R. R., Earl, G. W. and Ackerman, J., Properties and uses of some unsaturated fatty alcohols
and their derivatives. Journal of the American Oil Chemists’ Society, 61, No. 2, 324 (1984).
Ferguson, R. H. and Lutton, E. S., The polymorphism of triolein. Journal of the American Chemical
Society, 69, No. 6, 1445 (1947).
Fredenslund, A., Jones, R. L. and Prausnitz, J. M., Group–contribution estimation of activity
coefficients in nonideal liquid mixtures. AIChE Journal, 21, No. 6, 1086 (1975).
Gmehling, J., Anderson, T. F. and Prausnitz, J. M., Solid-liquid equilibria using UNIFAC.
Industrial & Engineering Chemistry Fundamentals, 17, No. 4, 269 (1978).
Gmehling, J., Li, J. and Schiller, M., A Modified UNIFAC Model. 2. Present parameter matrix and
results for different thermodynamic properties. Industrial & Engineering Chemistry Research, 32,
No. 1, 178 (1993).
Huang, C. C. and Chen, Y. P., Measurements and model prediction of the solid-liquid equilibria of
organic binary mixtures. Chemical Engineering Science 55, No. 16, 3175 (2000)
Jakob, A., Joh, R., Rose, C. and Gmehling, J., Solid-liquid equilibria in binary mixtures of organic
compounds. Fluid Phase Equilibria, 113, No.1-2, 117 (1995).
Kolp, D. G. and Lutton, E. S., The polymorphism of n-hexadecanol and n-octadecanol. Journal of
the American Chemical Society, 73, No. 12, 5593 (1951).
Kreutzer, U. R., Manufacture of fatty alcohols based on natural fats and oils. Journal of the
American Oil Chemists’ Society, 61, No. 2, 343 (1984).
Levine, I. N., Physical Chemistry. McGrawHill, New York (2002).
National Bureau of Standards. NIST Chemistry WebBook <webbook.nist.gov> (Accessed in April
10, 2011).
Nichols, G., Kweskin, S., Frericks, M., Reiter, S., Wang, G., Orf, J., Carvallo, B., Hillesheim, D.
and Chickos, J., Evaluation of the vaporization, fusion, and sublimation enthalpies of the 1-
alkanols: the vaporization enthalpy of 1-, 6-, 7-, and 9-heptadecanol, 1-octadecanol, 1-eicosanol, 1-
docosanol, 1-hexacosanol and cholesterol at T = 298.15 K by correlation gas chromatography.
Journal of Chemical & Engineering Data, 51, No. 2, 475 (2006).
O’Brien, R. D., Formulating and Processing for Applications. CRC Press, Boca Raton (2004).
Prausnitz, J. M., Lichtenthaler R. N. and Azevedo, E. G., Molecular Thermodynamics of Fluid-
Phase Equilibria. Prentice-Hall, New Jersey (1986).
Reid, R. C., Praunitz, J. M. and Poling, B. E., The Properties of Gases and Liquids. MacGrawHill,
New York (1987).
Rolemberg, M. P., Equilíbrio sólido-líquido de ácidos graxos e triglicerídeos: Determinação
experimental e modelagem. Ph.D. Thesis, University of Campinas (2002). (In Portuguese).
67
Slaughter, D. W. and Doherty, M. F., Calculation of solid-liquid equilibrium and crystallization
paths for melt crystallization processes. Chemical Engineering Science, 50, No. 11, 1679 (1995).
Ventolà, L., Calvet, T., CuevasDiarte, M. A., Ramírez, M., Oonk, H. A. J., Mondieig, D. and
Negrier, Ph., Melting behaviour in the n-alkanol family: Enthalpy–entropy compensation. Physical
Chemistry Chemical Physics, 6, No. 8, 1786 (2004).
Wesdorp, L. H., Liquid–multiple solid phase equilibria in fats, theory and experiments. Ph.D.
Thesis, Delft University of Technology (1990).
68
69
ON THE SOLID-LIQUID EQUILIBRIUM OF BINARY MIXTURES
OF FATTY ALCOHOLS AND FATTY ACIDS
(published in Fluid Phase Equilibria, v. 366, p. 88-98, 2014)
Guilherme J. Maximo a, Natália D. D. Carareto
a, Mariana C. Costa
b, Adenilson O. dos
Santos c, Lisandro P. Cardoso
d, Maria A. Krähenbühl
e, Antonio J. A. Meirelles
a*
a Laboratory of Extraction, Applied Thermodynamics and Equilibrium, School of Food
Engineering, University of Campinas (UNICAMP), R. Monteiro Lobato 80, 13083-862, Campinas,
São Paulo, Brazil. b School of Applied Sciences, University of Campinas, R. Pedro Zaccaria, 1300,
13484-350, Limeira, São Paulo, Brazil. c Social Sciences, Health and Technology Center, University
of Maranhão, 65900-410, Imperatriz, Maranhão, Brazil. d Institute of Physics Gleb Wataghin
(IFGW), University of Campinas, 13083-859, Campinas, São Paulo, Brazil. e DDPP - School of
Chemical Engineering, University of Campinas, 13083-970, Campinas, São Paulo, Brazil.
GRAPHICAL ABSTRACT
SLE phase diagrams of three fatty alcohol + fatty acid systems. Experimental measurements by
DSC and thermodynamic modeling for iL calculation. Liquid phase activity coefficients’
evaluation. Micrographs of the solid-liquid transition and X-ray diffraction of the solid-phase.
Chapter 3
70
ABSTRACT
Fatty alcohols and fatty acids are used in the cosmetic, pharmaceutical and food industries
as surfactants. They are also considered phase change materials for thermal storage
processes. Information on their thermal properties is required for optimizing production
processes as well as for improving their industrial and home use. In the present study, the
solid-liquid phase diagrams of three binary systems of 1-tetradecanol + dodecanoic acid, 1-
hexadecanol + tetradecanoic acid and 1-octadecanol + hexadecanoic acid were determined
by differential scanning calorimetry. The phase-transition phenomena were further
investigated by optical micrographs and X-ray diffraction patterns. The experimental data
showed that the systems present eutectic transitions and some exhibit also partial solid
phase miscibility. The liquid phase activity coefficients were calculated by Margules 2 and
3-suffix and by UNIFAC and UNIFAC-Dortmund methods. The modeling approach
resulted in an accurate prediction, with average absolute deviations from experimental data
lower than 1.16 K. The values of excess Gibbs energy present an unusual behavior, with
positive deviations at very low alcohol concentrations and negative ones at high
concentrations of this component. This occurs due to changes in the H-bonding interactions
along the concentration range of the mixture.
KEYWORDS: Solid-Liquid Equilibrium; Differential Scanning Calorimetry; Fatty
alcohols; Fatty acids; X-ray diffraction; Thermodynamic modeling.
71
1. INTRODUCTION
Fatty alcohols and fatty acids are biocompounds found in fats and oils, extracted
from them by separation processes or produced by specific routes from triacylglycerols.
They are widely used in the cosmetic, pharmaceutical and food industries as surfactant and
co-surfactant structuring agents for emulsification, gellation and coating [1-3]. Moreover,
different authors evaluated them as suitable phase change materials for thermal storage
processes, as required in solar energy application, cooling or heating systems and industrial
heat recovery [4, 5]. Under specific thermal conditions, fatty acids and alcohols exhibit
particular crystalline microstructures that influence the rheological and physical chemical
patterns of products involving such substances [6, 7]. Thus, taking into account design and
optimization of such applications, information about phase transitions of these compounds
and their mixtures is required. Nevertheless, these data are yet scarce in the literature.
Differential scanning calorimetry (DSC) and other techniques, such as optical
microscopy or X-ray diffraction, were frequently used for determination and analysis of
solid-liquid transitions of several pure substances and their mixtures, including fatty
compounds [8-19]. These techniques allow the evaluation of the phase diagrams,
particularly solid-liquid transitions or crystal formation behavior. In fact, recent studies
have shown that phase diagrams of binary systems of fatty acids are more complex than
they were usually presented by literature [11-13], suggesting a similar complexity for
systems of fatty acid + fatty alcohols.
In the present study, solid-liquid equilibrium data of three fatty mixtures, 1-
tetradecanol + dodecanoic acid, 1-hexadecanol + tetradecanoic acid and 1-octadecanol +
hexadecanoic acid, were measured by differential scanning calorimetry. Also, some thermal
72
transitions were evaluated with the aid of optical microscopy at controlled temperatures and
Xray diffraction analysis. The systems were modeled by solving the thermodynamic
equilibrium equations with an algorithm implemented in MATLAB. The liquid phase
activity coefficients were calculated using the two and three-suffix Margules equations
[20], the original UNIFAC [21] or the modified UNIFAC-Dortmund methods [22].
2. MATERIALS AND METHODS
2.1. Materials
Fatty compounds used in this work, supplier, mass fraction purity and method used
to confirm the purity are indicated in Table 1. Samples (1 g) of 1-tetradecanol + dodecanoic
acid, 1-hexadecanol + tetradecanoic acid and 1-octadecanol + hexadecanoic acid systems
were prepared gravimetrically on amber flasks in the whole composition range using an
analytical balance (Precisa Gravimetrics AG, Dietkon) with precision of 1×10-4
g. The
samples were pre-treated on Thermoprep (Metrohm, Herisau) by heating under a nitrogen
atmosphere until 15 K above the higher melting point of the components. The mixtures
were sealed at inert atmosphere, allowed to cool down to room temperature, and kept in a
freezer (below 273.15 K) until DSC analysis. The uncertainty of the compositions x1,
obtained by error propagation from the values of the weighed masses, was estimated as not
higher than 5×10−4
(in mole fractions).
73
Table 1. Fatty compounds used in this work, supplier, purification method, mole fraction purity and
method used to confirm the purity.
Chemical name Source
Mass
fraction purity
as received a
Purification
method b
Purity
analysis
method
Mole
fraction
purity
1-tetradecanol
Sigma-Aldrich
> 0.990 -
DSC c
0.995
1-hexadecanol > 0.990 - 0.994
1-octadecanol > 0.990 - 0.992
Dodecanoic acid > 0.990 - 0.991
Tetradecanoic acid > 0.990 - 0.994
Hexadecanoic acid > 0.990 - 0.995
a as reported by the supplier.
b no further purification method was required.
c using
the analysis
software from TA Instruments (New Castle). Fractional melting curves in Supplementary data.
2.2. Apparatus and procedure
2.2.1. Differential scanning calorimetry
The SLE data of the mixtures were characterized by DSC at local ambient pressure
p = 94.6 0.1 kPa, using a MDSC 2920 calorimeter (TA Instruments, New Castle)
equipped with a refrigerated cooling system which, in this work, operated between 250 K
and 350 K, using nitrogen as a purge gas. The melting temperature and heat flow were
previously calibrated with primary calibration standards, indium (certificated by
PerkinElmer, Waltham), naphthalene and cyclohexane (Merck, Whitehouse Station), with
purity greater than 0.99 mass fraction, at a heating rate of 1 K min−1
(such as that used in
the methodology described below). Samples of each mixture (2–5 mg) were weighed in a
microanalytical balance AD6 (PerkinElmer, Waltham) with precision of 1×10−6
g and put
in sealed aluminum pans. The methodology, as described by Costa et al. [10], is based on a
pre-treatment in order to erase previous thermal histories due to the fatty compounds
74
polymorphism effects. Thus, each sample was isothermally kept at 15 K above the highest
melting point of the components for 30 min. After this the samples were cooled to 40 K
below the lowest melting point of the components at a cooling rate of 1 Kmin−1
, and
allowed to stay at this temperature for 30 min. Finally, the samples were analyzed in a 1
Kmin−1
heating run. The temperatures of the thermal transitions were taken as the peak top
temperatures and the enthalpies were calculated from the area of the corresponding peaks.
The data were evaluated using the analysis software from TA Instruments (New Castle).
The uncertainties were calculated by performing up to five repeated DSC runs for
the pure compounds studied, and 2 – 4 repeated runs for eight binary mixtures selected
among the samples of the systems evaluated (replicates for the binary mixtures are reported
in Supplementary data). Table 2 shows the experimental pure components’ properties and
their standard deviations. The mean standard deviations of temperature values, including
the results obtained for the binary mixtures with 3 – 4 replicates, were not higher than 0.30 K
and this value was taken as the temperature experimental uncertainty. Moreover, the
experimental pure components’ properties were compared with data obtained from
literature [12, 18, 23-36]. The root-mean-square deviations were 0.63 K and 2.02 kJmol-1
for melting temperatures and enthalpy values, respectively, and the absolute deviations
obtained were close to the experimental errors.
75
Table 2. Experimental melting temperatures Tfus, solid-solid transitions Ttr and total melting
enthalpies fusH of the pure components obtained in this work and from literature.
Tfus / K Ttr / K fusH / ( kJ mol
-1 )
Component this work a literature this work
a literature this work
a literature
1-tetradecanol 311.39 0.08 311.2 [24] 310.45 0.10 310.8 [24] 47.6 1.0 48.4 [28]
311.2 [29] 310.8 [29]
47.0 [24]
311.2 [18] 310.1 [18]
49.5 [29]
312.0 [30]
47.3 [30]
1-hexadecanol 323.17 0.10 322.3 [29] 322.27 0.09 322.2 [24] 56.4 1.5 58.4 [29]
322.3 [24] 322.2 [29]
58.4 [24]
322.9 [23]
57.7 [27]
323.3 [18]
58.4 [28]
322.2 [27]
1-octadecanol 331.82 0.17 331.8 [23] 330.97 0.08 330.6 [32] 65.4 1.8 69.6 [28]
331.6 [18] 330.0 [36]
66.6 [33]
331.2 [32] 329.5 [33]
66.7 [32]
331.2 [25]
Dodecanoic acid 318.29 0.19 317.5 [26] 317.90 0.24 317.6 [12] 35.6 0.6 36.3 [26]
318.1 [12]
38.7 [12]
318.0 [34]
36.7 [24]
37.2 [35]
Tetradecanoic acid 328.13 0.16 327.5 [26] 327.65 0.16 328.2 [12] 43.8 0.8 45.2 [26]
328.9 [12]
44.0 [12]
327.3 [24]
45.2 [24]
328.0 [34]
49.5 [29]
Hexadecanoic acid 336.34 0.26 335.8 [26] - 53.3 1.1 52.3 [26]
335.4 [12]
55.9 [12]
335.7 [24]
54.9 [24]
337.5 [31]
52.0 [35]
336.4 [35]
53.4 [31]
337.7 [30]
51.4 [30]
a Experimental data obtained in this work and standard deviations of the values. Uncertainty for
melting temperature and solid-solid transitions temperature is 0.30 K and for melting enthalpy,
evaluated from the standard deviation.
2.2.2. Optical microscopy
The behavior of the solid-liquid transition of some mixtures among the evaluated
systems was additionally analyzed by temperature controlled optical microscopy using a
76
microscope Leica DM LM (Leica Microsystems, Wetzlar). The images were obtained by
putting a sample on a cover-slip in a hot stage FP82H, connected to a Mettler Toledo FP 90
(Mettler-Toledo, Columbus) central processor unit programmed to heat the sample at a
heating rate of 0.1 Kmin−1
until the melting of the sample. The images were acquired at
each 0.5 K with a magnification of 5.
2.2.3. X-ray diffraction
The solid phase of the systems was also evaluated by X-ray diffraction technique.
Diffractograms of the samples (0.2 - 0.4 g) were obtained using a Philips X’Pert
diffractometer (Philips Analytical, Almelo) operating in the reflection mode with an anode
of Cu–K (= 1.5406 Å) and diffracted-beam graphite monochromator. Diffraction data
were collected from 4° to 50° with a Bragg-Brentano geometry (:2) with steps of 0.02° at
each 2s. The procedure used anti–scatter slits for incident and diffracted beams of 1° and
receiving slit of 0.1 mm.
2.3. Thermodynamic modeling
Phase diagram modeling was carried out using the solid-liquid equilibrium equation
for each component of the mixture (Equation 1) as presented elsewhere [37, 38]
L Lfus tr fus fusi i
Sfu
fus
s tri i 1
1 1 1 1e ln 1xp
n
S
p
tr
H Hx
R T T R T T
C
R
T
T Tx
T
(1)
77
where xiL and xi
S are the mole fractions of the component i in the liquid (L) and solid (S)
phases, iL and i
S are the activity coefficients of the component i in the same phases, fusH
and Tfus are the melting enthalpy and temperature of the pure component i , T is the melting
temperature of the mixture, trH and Ttr are the solid-solid transitions enthalpies and
temperatures of the pure component i and fusCp is the heat capacity change of the pure
component i at melting temperature.
The modeling procedure was applied considering only the occurrence of a eutectic
reaction, based on the methodology described by Maximo et al. [19]. As a consequence, the
immiscibility in the solid phase was assumed, so that this phase is composed of pure fatty
acid or pure fatty alcohol (xiSi
S = 1). Moreover, the term related to fusCp can be neglected
as explained elsewhere [20, 37-39]. Also, as will be discussed later, fatty acids and alcohols
show polymorphism phenomena during the melting process with transition temperature
very close to melting temperature, i.e Ttr Tfus. In this case, the solid-solid transition
enthalpy trH can be embodied in the total melting enthalpy fusH. Thus, the SLE equation
1 can be simplified as presented in equation 2.
L L fusi i
fus
1 1exp
Hx
R T T
(2)
The iL values were calculated using the two and three-suffix Margules equations or
the group-contribution UNIFAC and modified UNIFAC-Dortmund models. Furthermore,
the systems were also modeled assuming the liquid phase as an ideal solution (iL
= 1).
Since the liquidus line was calculated for a simple eutectic mixture, the left-hand curve
corresponded to the equilibrium with a solid phase composed of pure fatty acid and the
78
right-hand curve corresponded to the equilibrium with a solid phase composed of pure fatty
alcohol, being the eutectic point their intersection.
In this procedure, the binaries parameters of Margules equations were numerically
estimated by an algorithm implemented in MATLAB 7.0. The algorithm was based on the
calculation of the melting temperatures by equation 2, rewritten as a function of T. The
adjustment of the two-suffix Margules equation’s parameter was a one-variable
minimization problem and implemented using a parabolic interpolation method (MATLAB
function fminbnd). Using the three-suffix Margules equation, a two-variable minimization
problem was carried out using a sequential quadratic programming method (MATLAB
function fmincon). In this case, the objective function was the summation of the square of
the absolute deviation ibetween experimental temperature Texp and the calculated one Tcalc
(Equations 3 and 4) for each composition xiL. When UNIFAC or UNIFACDortmund were
used as iL model, the algorithm was based on the sequential quadratic programming
(MATLAB function fmincon) for the calculation of the melting temperature of the mixtures
by minimization of the objective function (Equation 3). In this case, the routine was
repeated for 0 < xiL < 1 and the first iteration started with a melting temperature value
sufficiently close to the optimal solution. Moreover, the pure components’ melting
temperatures and enthalpy values, Tfus and fusH, used in equation 2, were taken from
experimental data measured in this work (Table 2).
2i exp calc i( ) T T (3)
fus fuscalc L L
fus fus i iln
H TT
H R T x
(4)
79
3. RESULTS AND DISCUSSION
3.1. Experimental SLE data
Figures 1 to 3 show the solid-liquid phase diagrams and the thermograms of the
pure components and mixtures evaluated in the temperature range close to the biphasic
region: 1-tetradecanol + dodecanoic acid (Figure 1), 1-hexadecanol + tetradecanoic acid
(Figure 2) and 1-octadecanol + hexadecanoic acid (Figure 3). The experimental melting
temperatures obtained from DSC thermograms are presented in Tables 3 to 5. Since some
transitions presented in Tables 3 to 5 are not apparent in the phase diagrams of Figures 1 to
3 due to magnification effects, some details are presented in such figures in order to clearly
show some of these transitions (see inserts in Figures 2 and 3). For each one of the pure
alcohols, 1-tetradecanol, 1-hexadecanol and 1-octadecanol, two overlapped peaks were
observed in the region very close to melting temperature, probably related to rotator forms
(Ttrans,pure, Tables 3 to 5) that appear during melting and crystallization processes [18, 19,
40]. These rotator forms are structurally different but present very similar thermodynamic
properties what explains the overlapping of the peaks. The melting profiles of pure
dodecanoic and tetradecanoic acids also showed more than one and overlapped transitions
(see inserts in Figures 1 and 2, highlighted by black arrows) in agreement with literature
[12], indicating that probably there are also polymorphic transitions during the melting
processes of these fatty acids. With the occurrence of a marked overlap phenomenon in all
these cases, the determination of the enthalpy of the transition htr by means of peak area
measurement does not lead to accurate and reliable values.
80
Table 3. Experimental solid-liquid equilibrium data for 1-tetradecanol (1) + dodecanoic acid (2) for
mole fraction x, temperature T and pressure p = 94.6 kPa
a.
a Uncertainties for mole fraction, temperature and pressure are 0.0005, 0.30 K and 0.1 kPa,
respectively. b pure dodecanoic acid (2).
c pure 1-tetradecanol (1).
Figure 1. Solid-liquid phase diagram and thermograms for the system 1-tetradecanol (1) +
dodecanoic acid (2): experimental melting temperatures (○); transition of the pure component ();
eutectic transition (■); other transitions (+), and modeled data by three-suffix Margules equation
(); UNIFAC-Dortmund model ( ) and ideal assumption (—). Thermogram in detail shows
transition at x1 = 0.0000.
295.0
300.0
305.0
310.0
315.0
320.0
0.0 0.2 0.4 0.6 0.8 1.0
T/ K
x1-tetradecanol
Heat Flow
-15
-10
-505
Heat Flow (W/g)
-15
-10-505
Heat Flow (W/g)
27
52
80
28
52
90
29
53
00
30
53
10
31
53
20
Te
mpe
ratu
re (
K)
Exo
Up
Univ
ersa
l V4.5
A T
A I
nst
rum
ents
315
320
0.0
00
0
0.1
03
1
0.2
00
5
0.2
99
0
0.4
00
7
0.4
99
6
0.5
48
4
0.6
01
8
0.6
27
3
0.6
49
90
.69
96
0.7
99
0
0.9
00
2
x1 = 0.0000
x1
1.0
000
x1 Tfus / K T trans,pure / K Teutectic / K Ttrans1 / K Ttrans2 / K Ttrans3 / K Solid phase
0.0000 318.29 317.90 2b
0.1031 314.98 299.48 279.19 2
0.2005 312.70 300.20 278.40 2
0.2990 309.26 299.91 276.07 2
0.4007 306.95 299.73 278.92 2
0.4996 303.79 299.71 275.70 2
0.5484 301.61 299.03 2
0.6018 299.52 298.56 278.30 1c
0.6273 301.12 299.53 298.54 279.95 1
0.6499 301.41 299.51 298.21 281.48 1
0.6996 303.77 300.01 298.85 284.77 1
0.7990 306.44 299.92 298.40 289.24 285.62 1
0.9002 308.99 299.16 288.03 1
1.0000 311.39 310.45 1
81
Table 4. Experimental solid-liquid equilibrium data for 1-hexadecanol (1) + tetradecanoic acid (2)
for mole fraction x, temperature T and pressure p = 94.6 kPa
a.
x1 Tfus / K Ttrans,pure / K Teutectic / K Ttrans1 / K Ttrans2 / K Ttrans3 / K Ttrans4 / K Ttrans5 / K Solid phase
0.0000 328.13 327.65 2b
0.0982 325.88 312.44 2
0.1993 323.76 312.39 2
0.3042 321.80 312.57 306.71 2
0.3981 318.78 312.63 306.88 288.13 2
0.5020 315.20 312.66 311.82 307.43 287.39 2
0.5497 312.06 311.04 307.48 284.47 1c
0.6004 313.56 312.52 311.39 306.73 288.21 1
0.6506 314.21 312.06 310.85 307.53 291.21 284.10 1
0.7152 316.32 312.81 311.65 306.52 293.61 290.80 1
0.7951 319.19 312.81 311.53 305.97 299.30 294.47 1
0.8994 320.91 312.16 298.98 1
1.0000 323.17 322.27 1
a Uncertainties for mole fraction, temperature and pressure are 0.0005, 0.30 K and 0.1 kPa,
respectively; b pure tetradecanoic acid (2);
c pure 1-hexadecanol (1).
Figure 2. Solid-liquid phase diagram and thermograms for the system 1-hexadecanol (1) +
tetradecanoic acid (2): experimental melting temperatures (○); transition of the pure component ();
eutectic transition (■); other transitions (+), and modeled data by three-suffix Margules equation
(); UNIFAC-Dortmund model ( ) and ideal assumption (—). Thermograms in detail show
transitions at x1 = (0.0000 and 0.6004).
305.0
310.0
315.0
320.0
325.0
330.0
335.0
0.0 0.2 0.4 0.6 0.8 1.0
T/ K
x1-hexadecanol
Heat Flow
-50510
Heat Flow (W/g)
-5-3-113579
Heat Flow (W/g)
28
52
90
29
53
00
30
53
10
31
53
20
32
53
30
33
5
Tem
pera
ture
(K
)E
xo U
pU
nive
rsal
V4.
5A T
A I
nstr
umen
ts
325
330
305
307
0.0
000
0.0
98
2
0.1
993
0.3
042
0.3
98
1
0.5
02
0
0.5
497
0.6
004
0.6
50
6
0.7
15
2
0.7
951
0.8
994
1.0
00
0
x1
x1 = 0.6004
x1 = 0.0000
82
Table 5. Experimental solid-liquid equilibrium data for 1-octadecanol (1) + hexadecanoic acid (2)
for mole fraction x, temperature T and pressure p = 94.6 kPa
a.
x1 Tfus / K T trans,pure / K Teutectic / K Ttrans1 / K Ttrans2 / K Ttrans3 / K Ttrans4 / K Solid phase
0.0000 336.34 2b
0.0965 334.09 321.95 293.93 2
0.1981 332.36 322.05 294.23 2
0.3010 330.39 322.42 293.52 2
0.3992 327.63 322.34 290.86 2
0.4506 325.29 321.60 294.05 2
0.4999 322.80 312.48 292.58 2
0.5534 322.15 316.01 293.62 1c
0.5978 323.64 322.13 314.59 293.10 1
0.7010 326.00 321.82 315.67 303.69 292.74 1
0.7993 327.82 322.69 314.86 307.70 304.76 1
0.8990 329.93 321.67 319.88 307.58 1
0.9508 330.65 1 + 2 d
1.0000 331.82 330.97 1
a Uncertainties for mole fraction, temperature and pressure are 0.0005, 0.30 K and 0.1 kPa,
respectively; b pure hexadecanoic acid (2);
c pure 1-octadecanol (1);
d solid solution composed by 1-
octadecanol (1) and hexadecanoic acid (2).
Figure 3. Solid-liquid phase diagram and thermograms for the system 1-octadecanol (1) +
hexadecanoic acid (2): experimental melting temperatures (○); transition of the pure component ();
eutectic transition (■); other transitions (+), and modeled data by three-suffix Margules equation
(); UNIFAC-Dortmund model ( ) and ideal assumption (—). Thermogram in detail shows
transition at x1 = 0.5978.
310.0
315.0
320.0
325.0
330.0
335.0
340.0
0.0 0.2 0.4 0.6 0.8 1.0
T/ K
x1-octadecanol
Heat Flow
-12
-10
-8-6-4-202
Heat Flow (W/g)
-15
-10-505
Heat Flow (W/g)
29
03
10
33
0
Te
mpe
ratu
re (
K)
Exo
Up
Univ
ersa
l V4.5
A T
A I
nst
rum
ents
311
316
0.0
00
0
0.0
965
0.1
98
1
0.3
01
0
0.3
99
2
0.4
506
0.4
999
0.5
53
4
0.5
97
8
0.7
010
0.7
993
0.8
99
0
1.0
000
x1
0.9
508
x1 = 0.5978
83
The thermal analysis of the mixtures showed at least two well-defined endothermic
peaks. The thermal transition at the higher temperature is related to the melting temperature
of the mixture (Tfus, Tables 3 to 5). The second transition is the eutectic reaction
(Teutectic,Tables 3 to 5). For the systems with 1-tetradecanol and 1-octadecanol the intensity
of the eutectic thermal transition decreased when the concentration was close to pure fatty
alcohol and at x1-octadecanol 0.95 mole fraction no thermal transition at the eutectic
temperature was observed, showing that the compounds involved in these mixtures may
probably be partially miscible, close to this concentration range. However, since the
observation of no eutectic peaks could be also related to weak transitions due to the
proximity to the pure compound concentration, this aspect will be further investigated by
Tammann plots evaluation, powder X-Ray diffractometry and later discussed.
The thermograms of the mixtures also presented endothermic peaks below the
eutectic reaction temperature. In some cases, these peaks, which are related to solid-solid
thermal transitions, are overlapped, indicating the existence of a complex phase diagram
probably due to the fatty compounds’ polymorphism. The solid-solid transitions
temperatures (Ttrans1 to Ttrans5) are also presented in Tables 3 to 5. Moreover, it was observed
that some thermograms of the mixtures presented thermal transitions inside the biphasic
region (above eutectic temperature and below liquidus line). These transitions could be
probably related to metastable molecular conformations of the fatty compounds but
eventually to peritectic or metatectic reactions [11, 13]. In order to check this supposition,
some of these samples were submitted to a heating-cooling cycle treatment during which
the sample was cooled down from 15 K above the highest melting point of the components
to 40 K below the lowest melting point of the components at a cooling rate of 5 Kmin−1
84
and allowed to stay at this temperature for 30 min. Then, the sample was heated to the onset
temperature at which it was observed the supposed metastable transition and kept there for
60 min. The final cycle followed the aforementioned methodology by which the sample
was cooled down to the lower temperature, kept isothermally for 30 min and heated again
at the same rate of 1 Kmin-1
until sample melting. Figure 4 shows the results of the thermal
annealing for mixtures of 1-hexadecanol (1) + tetradecanoic acid (2) at x1 = (0.3042 and
0.8994) mole fractions whose related transitions are indicated by the black arrows. The
thermograms obtained before the thermal treatment (Figure 4, curves I) show two peaks in
the region of solid-liquid transition related to melting and eutectic reaction, and a very
small one comprised between the melting and eutectic peaks. As can be seen, after thermal
treatment (Figure 4, curves II) the very small peak was clearly not observed, indicating that,
in fact, it is related to a metastable transition. According to literature [26, 41] the crystals
network structure of the fatty compounds presents, during thermal processes, transitions
through hexagonal and metastable chain packings to orthorhombic and more stable states.
When submitted to a thermal annealing based treatment, which consisted in an isothermal
condition in a specific temperature followed by a cooling process at a low rate, the structure
of the material is supposedly converted. Considering that after the thermal treatment such
transitions were no longer observed, one can suppose that they were related to a residual
solid-phase in hexagonal chain packing that was converted to the orthorhombic and more
stable form. In fact, as previously observed, the thermograms of the fatty acids and fatty
alcohols presented polymorphic forms with different chain packing and stability which
probably induced the appearance of these metastable transitions observed in the
calorimetric measurements. Therefore, since it was confirmed that these transitions are
metastable phenomena that can be eliminated by a proper thermal treatment of the sample,
85
they were not presented in Tables 3 – 5. Note that solid-solid transitions below eutectic
reaction are reported in data Tables 3 – 5 but further analyses are still required.
Figure 4. Effect of the thermal annealing based treatment on thermograms of 1-hexadecanol (1) +
tetradecanoic acid (2) at (A) x1 = 0.8994 and at (B) x1 = 0.3042: I, heating curve without the thermal
treatment and; II, after thermal treatment. The metastable transition is shown by the black arrow.
The phase diagrams of the systems were very similar to each other. In all cases, the
liquidus lines showed just one inflexion point in the temperature of the invariant thermal
transition temperature observed in the thermograms, namely the eutectic transition. The
phase diagrams presented at least four well-defined regions, with the first one related to a
liquid phase above liquidus line, two biphasic regions between the melting temperature and
the eutectic transition and a solid phase below eutectic temperature. At the left side of the
diagram, the biphasic region corresponds to a liquid mixture in equilibrium with pure fatty
acid in the solid phase. At the right side the biphasic region corresponds to a liquid mixture
in equilibrium with pure fatty alcohol or, as it will be discussed later, a fatty alcohol-rich
solid phase. In this last case a fifth region occurs in the right side of the diagram. This
behavior is very similar to the so-called eutectic system that is commonly found in the
295 305 315 325 335
Hea
t fl
ow
T / K
295 305 315 325 335
T / K-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
He
at
Flo
w (
W/g
)
295 305 315 325 335
Temperature (K)Exo Up Universal V4.5A TA Instruments
-2
0
He
at
Flo
w (
W/g
)
295 305 315 325 335
Temperature (K)Exo Up Universal V4.5A TA Instruments
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
He
at
Flo
w (
W/g
)
295 305 315 325 335
Temperature (K)Exo Up Universal V4.5A TA Instruments
II
I
(B)(A)
II
I
86
literature in case of fatty mixtures [10, 17], for which small regions of miscibility may
occur in the solid phase, usually close to the extremes of the phase diagram. Tables 3 to 5
present the compound in the solid phase that is in equilibrium with the liquid phase, being a
pure compound or a solid solution, in case of miscibility of the solid phase, according the
evaluation of the Tammann plots, discussed as follows.
3.2. Eutectic point and immiscibility assumption
Figure 5 exhibits the behavior of the enthalpy variation of the systems’ invariant
transitions as a function of the concentration. This diagram is the so-called Tammann plot
and allowed the characterization of the eutectic point, since the corresponding enthalpy
increases as the mixture concentration comes close to the eutectic composition and should
begin to decrease after this transition [17, 42]. Experimental data were submitted to linear
regression and equations were presented in Figure 5. In the hypothesis of immiscibility of
solid phase, it was assumed that heutetic = 0 for x1 = 1 or 0. Since R2 > 0.98 for all cases,
the assumption is, most probably, valid. Fitted data allowed the determination of the
experimental eutectic points, xalcohol, eutetic = (0.5750, 0.5490 and 0.5100) mole fraction for the
systems 1-tetradecanol + dodecanoic acid, 1-hexadecanol + tetradecanoic acid and 1-
octadecanol + hexadecanoic acid, respectively.
87
Figure 5. Tammann plots of the eutectic transition (■) for: (A) 1-tetradecanol + dodecanoic acid;
(B) 1-hexadecanol + tetradecanoic acid and (C) 1-octadecanol + hexadecanoic acid as well as linear
regression (- - -) of the data.
The experimental behavior around the eutectic temperature could also be observed
by optical micrographs. Figure 6 shows results for the 1-hexadecanol + tetradecanoic acid
system at x1-hexadecanol = 0.3981 mole fraction (micrographs A to D) and for the 1-
octadecanol + hexadecanoic acid system at x1-octadecanol = 0.7993 mole fraction (micrographs
E to H). For the first system, at T = 311.75 K (Figure 6 A) both components were in the
solid phase. As the temperature was raised to T = 313.65 K (Figure 6 B) a fraction of the
solid crystals started to melt and this phenomenon is indicated in the Figure 6 B by white
arrows. A slight increase of the temperature from T = 313.65 K to T = 313.75 K (Figure 6
C) was sufficient to induce the increase of the amount of melted mixture in the sample, as
could be clearly observed in the border of the crystals. With the continuous increase in
temperature inside biphasic region the fraction of the liquid phase also increased (Figure 6
D), as expected and in agreement with the lever rule. The optical micrographs of the system
1-octadecanol + hexadecanoic acid (Figures 6 E to H) presented a similar behavior, since
the liquid phase fraction increased as temperature increased with the beginning of melting
y = 63.632xR² = 0.9898
y = -103.45x + 96.03R² = 0.9817
0.0
10.0
20.0
30.0
40.0
50.0
h
eu
teti
c/
kJm
ol-1
x1-tetradecanol
y = 54.04xR² = 0.9836
y = -65.73x + 65.73R² = 0.9822
x1-hexadecanol
y = 81.769xR² = 0.9909
y = -105.36x + 95.222R² = 0.9552
x1-octadecanol
(A) (B) (C)
0.0 1.0 0.0 0.01.0 1.00.50.50.5
88
transition at the eutectic temperature. In this case, some of the melting phenomena observed
are indicated by white arrows. Moreover, some images of these mixtures at other
compositions were taken at temperatures below the eutectic point and, as expected, it was
observed no liquid fraction, in agreement with experimental DSC results.
Figure 6. Optical micrographs of samples of 1-hexadecanol (1) + tetradecanoic acid (2) at x1 =
0.3981 and (A) 311.75 K; (B) 313.65 K; (C) 313.75 K; (D) 314.55 K and 1-octadecanol (1) +
hexadecanoic acid (2) at x1 = 0.7993 and (E) 319.15 K; (F) 321.15 K; (G) 322.25 K; (D) 325.65 K.
Some details are shown by white arrows. Magnification of 5.
The Tammann curves also allowed to determinate the regions of miscibility in the
phase diagrams. If complete immiscibility of the solid phase occurs, the enthalpy values of
the eutectic reaction tend to zero when mixture concentration tends to the pure components.
However, the Tammann plots of the mixtures of 1-tetradecanol + dodecanoic acid and 1-
octadecanol + hexadecanoic acid showed that the solid components of these binaries in the
right side of the phase diagram are partially miscible because the enthalpy variation value
200 mm
200 mm
200 mm 200 mm 200 mm
200 mm 200 mm 200 mm
A B C D
HGFE
89
tended to zero at x1-tetradecanol = 0.9283 mole fraction and at x1-octadecanol = 0.9038 mole
fraction.
3.3. Evaluation of the solid-phase
In order to evaluate some aspects of the behavior of the solid phase of the systems,
diffraction patterns at T = 298.15 K and at different compositions of the 1-hexadecanol (1)
+ tetradecanoic acid (2) and 1-octadecanol (1) + hexadecanoic acid (2) systems were
carried out at 15.0° < 2 < 27.5°, which is the diffraction angle range in which it was
observed the most meaningful diffraction profiles (see Figure 7). According to the
system’s phase diagram (Figures 1 to 3) at T = 298.15 K both components in all systems
are in the solid state. Figure 7 A, that presents the 1-hexadecanol (1) + tetradecanoic acid
(2) system, shows that pure tetradecanoic acid (x1 = 0.0000 mole fraction) presents some
overlapped diffraction peaks at the regions close to 2θ (19°, 21.5° and 24°) and pure 1-
hexadecanol (x1 = 1.0000 mole fraction) exhibits two intense diffraction peaks at 2θ 21.5°
and 2θ 24.5° and a mild peak at 2θ 20.5°. For this system, the diffraction patterns of the
mixtures shows that pure components’ diffraction peaks could be observed throughout the
concentration range, except for x1 = 0.8994 mole fraction. At this concentration, the
behavior differed from the pure component pattern suggesting that the crystalline structure
of the mixture was changed. In fact, according to Table 4, at x1 = 0.8994 mole fraction the
DSC measurements showed that at T = 298.98 K a solid transition was observed (Ttrans 4,
Table 4).
90
Figure 7. Difractograms at T = 298.15 K for (A) 1-hexadecanol (1) + tetradecanoic acid (2), from
up to down: x1 = (0.0000, 0.0982, 0.1993, 0.3042, 0.5020, 0.6004, 0.7152, 0.7951, 0.8994, 1.0000)
mole fraction and for (B) 1-octadecanol (1) + hexadecanoic acid (2), from up to down: x1 =
(0.0000, 0.3010, 0.4999, 0.7010, 0.8990, 0.9508, 1.0000) mole fraction.
Furthermore, the overall evaluation of the diffractograms suggested that both pure
components coexist in the solid phase and they were independently crystallized, in
agreement with the previous evaluation of the immiscibility assumption that was took into
account in the modeling approach and observed by the DSC data and the Tammann plots
(Figure 5 B). Figure 7 B, that presents the 1-octadecanol (1) + hexadecanoic acid (2)
system, shows that pure hexadecanoic acid (x1 = 0.0000 mole fraction) presents two intense
diffraction peaks at 2θ 21.0° and 2θ 24.5° and a mild peak at 2θ 20.5° and pure 1-
octadecanol (x1 = 1.0000 mole fraction) exhibits also two intense diffraction peaks at the
same diffraction angles and two other mild peaks at 2θ 20.5° and 2θ 22.5°. In this case,
the diffraction patterns of the mixtures shows that both pure 1-octadecanol and
hexadecanoic acid coexist in the solid phase in a concentration range up to x1 = 0.8990
mole fraction, because the pure components’ diffraction peaks could be clearly observed up
15.0 17.5 20.0 22.5 25.0 27.5
Inte
nsi
ty (
u.a
.)
2
(B)
15.0 17.5 20.0 22.5 25.0 27.5
Inte
nsi
ty (
u.a
.)
2
(A)
91
to this concentration. However, the mixture at x1 = 0.9508 mole fraction presented a
diffraction pattern equal to that of pure 1-octadecanol. In fact, according to the evaluation
of the Tammann plots this system presented partial miscibility at x1 > 0.9038 mole fraction.
Thus, at this concentration, both components were not independently crystallized and
because this solid solution is a 1-octadecanol-rich mixture, its diffraction pattern is similar
to that one observed for pure 1-octadecanol.
3.4. Thermodynamic modeling of SLE data
Table 6 shows the deviations between calculated and experimental data for each
modeling approach as well as the binary parameters (Jmol-1
) for both Margules equations.
First of all, since it was observed a very small solubility of the solid phase for two systems
and a complete insolubility for one of them, the modeling assumed that the mixtures
behaves as a simple eutectic system. It was observed that liquidus line was well described
by all models since the average absolute deviations were less than 1.2 K for all, including
the ideal assumption. The modeling approach using the three-suffix Margules equation as
iL model allowed a better description of the systems’ behavior.
The values of the liquid phase activity coefficients indicated that these systems are
slightly non-ideal. This could be observed by the modeled data in the SLE phase diagrams
(Figures 1 to 3) by which for the 1-tetradecanol + dodecanoic acid and 1-hexadecanol +
tetradecanoic acid systems, the deviations of the ideal liquidus lines were positive at low
concentrations of fatty alcohols and negative at high concentrations of this compound.
Moreover, for the 1-octadecanol + hexadecanoic acid system, the deviation of the ideal
92
liquidus line was negative throughout the phase diagram. These observations are in
agreement with literature’s liquid phase activity coefficients at infinite dilution for
analogous fatty mixtures or other acid-alcohol mixtures [43-45].
Table 6. Estimated parameters, A and B, and mean absolute deviations a between
predicted and experimental data.
Model 1-tetradecanol +
dodecanoic acid
1-hexadecanol +
tetradecanoic acid
1-octadecanol +
hexadecanoic acid
Two-suffix Margules A / J mol-1
-496.86 -501.24 -938.79
/ K 0.92 0.71 0.36
Three-suffix Margules A / J mol-1
-912.33 -1027.24 -1168.16
B / J mol-1
-1351.78 -1199.45 -440.745
/ K 0.50 0.28 0.28
UNIFAC / K 1.07 0.88 0.44
UNIFAC-Dortmund / K 1.16 0.92 0.68
Ideal / K 1.11 0.93 0.63
aexp predn
i i i
n
T T
Figure 8 shows the liquid phase activity coefficients for the system 1-hexadecanol +
tetradecanoic acid (Figure 8 A) as well as the behavior of the corresponding excess Gibbs
free energy (GE) (Figure 8 B), calculated using the three-suffix Margules equation, the most
accurate model. It is worth noting that this system exhibits a simple eutectic behavior, with
total immiscibility in the solid phase. In this case, the modeling approach based on equation
2 requires absolutely no approximation, increasing the reliability of the liquid phase activity
coefficients estimated using the three-suffix Margules model.
93
Figure 8. (A) Liquid-phase activity coefficients and (B) Mixtures’ excess free Gibbs energy
calculate by three-suffix Margules equation (in detail, magnification for x1 < 0.1) for 1-hexadecanol
(1) + tetradecanoic acid (2): 1L ( ), 2
L () and i
L (○) calculated from experimental data by
SLE equation.
It could be observed that when alcohol is added to pure fatty acid, the acid activity
coefficient increases until reaching a maximum value, when it begins to decrease with the
increasing of alcohol concentration. On the other hand, when acid is added to pure fatty
alcohol, the alcohol activity coefficient tends to decrease until reaching a minimum value,
when it begins to increase. In addition, GE values presented negative deviations throughout
the concentration range, except for slightly positive deviations at the region of low
concentration of fatty alcohol.
This behavior can be probably explained on the basis of two main factors, namely
intermolecular H-bonding interactions and changes observed in these interactions along the
entire concentration range of the mixture. Mixtures containing fatty acids and/or alcohols
exhibit van der Waals interactions between their carbon chains and H-bonding interactions
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
1.2
0.0 0.2 0.4 0.6 0.8 1.0
i
x1-hexadecanol
-350
-250
-150
-50
50
0.0 0.2 0.4 0.6 0.8 1.0
GE
n-1
/ Jm
ol-1
x1-hexadecanol
-5
0
5
0.0 0.1
(A) (B)
94
between their polar groups. It is well known that the energy of H-bonding interactions
between carboxyl moieties is greater than that observed between the hydroxyl groups
belonging to alcohol molecules, because the adjacent hydroxyl and carbonyl groups of the
carboxyl moiety makes this moiety even more strongly polarized [46, 47]. Furthermore, H-
bonding interactions between carboxylic acid molecules are responsible for their trend to
self associate and form dimers.
Considering the addition of alcohol molecules to the pure organic acid, some of the
H-bonding interactions between two carboxylic acids are broken and replaced by the
relatively weaker H-bonding interactions between hydroxyl and carboxyl groups, which
explains the positive deviations of the fatty acid activity coefficients as well as the positive
deviations of the GE
values and so the lower stability of the mixture than the corresponding
ideal one at low alcohol concentrations. On the other hand, the H-bonding interactions
between hydroxyl and carboxyl groups may engage a large number of molecules, so that
this kind of H-bonding can build a network of interacting molecules, compensating its
relative weakness in comparison to H-bonding involving only carboxyl moieties by the
pervasiveness of its effect along the mixture. Such behavior probably predominates as the
alcohol concentration raises, since this increases the likelihood that carboxyl moieties
interact by H-bonding with hydroxyl groups of the alcohol molecules. It also favors the
stabilization of the system (GE < 0) and explains the negative deviations of the fatty acid
activity coefficients.
Considering the addition of acid molecules to pure fatty alcohol, some of the H-
bonding interactions between two alcohol molecules are broken and replaced, in the present
case, by the relatively stronger H-bonding interactions between hydroxyl and carboxyl
groups. This justifies the observed negative deviations of the alcohol activity coefficients.
95
On the other hand, the continuous increase of the concentration of fatty acid raises the
likelihood of the energetically more favorable H-bonding interactions between two
carboxyl moieties. This means that, after some concentration value of fatty acid, the H-
bonding interactions between acid molecules tend to predominate, causing an inversion in
the trend of decreasing alcohol activity coefficients. Furthermore, the alcohol activity
coefficient assumes values above 1.0 in the diluted alcohol solutions, probably because the
alcohol molecules, in this range of concentrations, are surrounded by organic acids
interacting strongly with one another, i.e. they are in fact surrounded mainly by organic
acid dimers.
Moreover, Figure 8 A shows the experimental activity coefficients of the liquid
phase calculated from the SLE model (Equation 2). This estimation took into account the
standard deviations of the experimental pure compounds properties fusH and Tfus from
Table 1, so that the iL-errors are also shown in Figure 8 A. As it can be seen, the
experimental values are in good agreement with those calculated using the three-suffix
Margules equation. It is important to note that, as immiscibility assumption was taken into
account, the experimental iL were calculated only in the concentration range that the
species were in both phases, i.e., compound 2 before eutectic concentration and compound
1 after this.
Additional information on the non-ideal behavior of the systems could be obtained
if the experimental errors of the pure compounds’ properties (Table 2) are taken into
account in the estimation of the ideal melting temperatures. Figure 9 shows the description
of the SLE diagram of the system 1-tetradecanol + dodecanoic acid with such assumptions.
It could be observed that there is no significant difference between most of experimental
96
melting point data and the calculated liquidus line below the eutectic concentration. Similar
results were obtained for the other two systems. As mentioned before, at this concentration
range, the most accurate model, three-suffix Margules equation, provided slightly positive
deviations and this observation is in reasonable agreement to literature review.
Nevertheless, modeled melting temperatures are within the uncertainty range of values
around the ideal behavior, so that these mixtures may be approximated by ideal systems in
case of fatty alcohol concentrations lower than the eutectic point. Moreover, after this
concentration, the system presents a significant non-ideal behavior with negative deviations
from ideality, confirming the above discussion.
Figure 9. Effect of the pure compounds’ standard deviations data in the modeling of the liquidus
line for 1-hexadecanol + tetradecanoic acid: experimental data (○); error bars = 0.30 K
(temperature’s experimental uncertainty); ideal assumption with mean values of the pure
compounds’ melting properties () and modeling considering the standard deviations of the pure
compounds’ properties (- - -).
310.0
314.0
318.0
322.0
326.0
330.0
0.0 0.2 0.4 0.6 0.8 1.0
T/
K
x1-hexadecanol
97
4. CONCLUSIONS
Solid-liquid equilibrium for the binary mixtures of 1tetradecanol + dodecanoic
acid, 1hexadecanol + tetradecanoic acid and 1-octadecanol + hexadecanoic acid were
measured using differential scanning calorimetry. Phase diagrams exhibited two well-
defined transitions that were the mixture melting temperature and the eutectic reaction.
Systems also presented many thermal transitions below eutectic temperature and other
metastable transitions above eutectic temperature. Micrographs allowed the observation of
the melting behavior around the eutectic temperature and the results were in agreement
with the lever rule. According to the DSC data, Tammann plots and X-ray diffraction
analysis, the mixtures of 1tetradecanol + dodecanoic acid and 1-octadecanol +
hexadecanoic presented partial miscibility in the solid phase within the concentration range
close to the pure fatty alcohol, unlike the 1-hexadecanol + tetradecanoic acid system for
which the pure compounds were independently crystallized along the entire concentration
range. The two-suffix Margules equations, the group-contribution UNIFAC and UNIFAC-
Dortmund methods were accurate for describing the melting point of the evaluated fatty
mixtures, since the mean absolute deviations were less than 1.16 K. The approach based on
the three-suffix Margules equation gave the best results with mean absolute deviation less
than 0.36 K.
98
ACKNOWLEDGEMENTS
The authors are grateful to FAPESP (2008/56258-8, 2009/54137-1 and 2012/05027-
1) and CNPq (304495/2010-7, 483340/2012-0 and 140718/2010-9) for the financial
support.
REFERENCES
[1] R.R. Egan, G.W. Earl, J. Ackerman, J. Am. Oil Chem. Soc., 61 (1984) 324-329.
[2] J. Daniel, R. Rajasekharan, J. Am. Oil Chem. Soc., 80 (2003) 417-421.
[3] H.M. Schaink, K.F. van Malssen, S. Morgado-Alves, D. Kalnin, E. van der Linden, Food Res.
Int., 40 (2007) 1185-1193.
[4] D. Feldman, M.M. Shapiro, D. Banu, C.J. Fuks, Sol. Energ. Mater., 18 (1989) 201-216.
[5] G.J. Suppes, M.J. Goff, S. Lopes, Chem. Eng. Sci., 58 (2003) 1751-1763.
[6] S. Bruin, Fluid Phase Equilibr., 158 (1999) 657-671.
[7] C. Himawan, V.M. Starov, A.G.F. Stapley, Adv. Colloid Interfac., 122 (2006) 3-33.
[8] L.H. Wesdorp, Liquid-Multiple Solid Phase Equilibria in Fats, Theory and Experiments, Delft
University of Technology, Delft, 1990.
[9] A.G. de Souza, J.C.O. Santos, M.M. Conceição, M.C.D. Silva, S. Prasad, Braz. J. Chem. Eng.,
21 (2004) 265-273.
[10] M.C. Costa, M.P. Rolemberg, L.A.D. Boros, M.A. Krähenbühl, M.G. de Oliveira, A.J.A.
Meirelles, J. Chem. Eng. Data, 52 (2007) 30-36.
[11] M.C. Costa, M.P. Rolemberg, A.J.A. Meirelles, J.A.P. Coutinho, M.A. Krähenbühl,
Thermochim. Acta, 496 (2009) 30-37.
[12] M.C. Costa, M. Sardo, M.P. Rolemberg, J.A.P. Coutinho, A.J.A. Meirelles, P. Ribeiro-Claro,
M.A. Krähenbühl, Chem. Phys. Lipids, 160 (2009) 85-97.
[13] M.C. Costa, M. Sardo, M.P. Rolemberg, P. Ribeiro-Claro, A.J.A. Meirelles, J.A.P. Coutinho,
M.A. Krähenbühl, Chem. Phys. Lipids, 157 (2009) 40-50.
99
[14] M.C. Costa, L.A.D. Boros, M.P. Rolemberg, M.A. Krähenbühl, A.J.A. Meirelles, J. Chem.
Eng. Data, 55 (2010) 974-977.
[15] M.C. Costa, M.P. Rolemberg, A.O. dos Santos, L.P. Cardoso, M.A. Krähenbühl, A.J.A.
Meirelles, J. Chem. Eng. Data, 55 (2010) 5078-5082.
[16] M.C. Costa, L.A.D. Boros, J.A.P. Coutinho, M.A. Krähenbühl, A.J.A. Meirelles, Energ. Fuels,
25 (2011) 3244-3250.
[17] M.C. Costa, L.A.D. Boros, J.A. Souza, M.P. Rolemberg, M.A. Krähenbühl, A.J.A. Meirelles,
J. Chem. Eng. Data, 56 (2011) 3277-3284.
[18] N.D.D. Carareto, M.C. Costa, M.P. Rolemberg, M.A. Krähenbühl, A.J.A. Meirelles, Fluid
Phase Equilibr., 303 (2011) 191-192.
[19] G.J. Maximo, M.C. Costa, A.J.A. Meirelles, Braz. J. Chem. Eng., 30 (2013) 33-43.
[20] R.C. Reid, J.M. Prausnitz, B.E. Poling, The Properties of Gases and Liquids, McGraw-Hill,
New York, 1987.
[21] A. Fredenslund, R.L. Jones, J.M. Prausnitz, Aiche J., 21 (1975) 1086-1099.
[22] J. Gmehling, J.D. Li, M. Schiller, Ind. Eng. Chem. Res., 32 (1993) 178-193.
[23] D.G. Kolp, E.S. Lutton, J. Am. Chem. Soc., 73 (1951) 5593-5595.
[24] E.S. Domalski, E.D. Hearing, J. Phys. Chem. Ref. Data, 25 (1996) 1-525.
[25] U. Domanska, J.A. Gonzalez, Fluid Phase Equilibr., 123 (1996) 167-187.
[26] T. Inoue, Y. Hisatsugu, R. Ishikawa, M. Suzuki, Chem. Phys. Lipids, 127 (2004) 161-173.
[27] J. Xing, Z.C. Tan, Q. Shi, B. Tong, S.X. Wang, Y.S. Li, J. Therm. Anal. Calorim., 92 (2008)
375-380.
[28] G. Nichols, S. Kweskin, M. Frericks, S. Reiter, G. Wang, J. Orf, B. Carvallo, D. Hillesheim, J.
Chickos, J. Chem. Eng. Data, 51 (2006) 475-482.
[29] C. Mosselman, J. Mourik, H. Dekker, J. Chem. Thermodyn., 6 (1974) 477-487.
[30] J. Zeng, Z. Cao, D. Yang, F. Xu, L. Sun, L. Zhang, X. Zhang, J. Therm. Anal. Calorim., 95
(2009) 501-505.
[31] F.O. Cedeño, M. M. Prieto, A. spina, . . Garc a, Thermochim. Acta, 369 (2001) 39-50.
[32] J.C. van Miltenburg, H.A.J. Oonk, L. Ventola, J. Chem. Eng. Data, 46 (2000) 90-97.
[33] L. Ventolà, M. Ramírez, T. Calvet, X. Solans, M.A. Cuevas-Diarte, P. Negrier, D. Mondieig,
J.C. van Miltenburg, H.A.J. Oonk, Chem. Mater., 14 (2002) 508-517.
[34] A.D. Bond, New J. Chem., 28 (2004) 104-114.
100
[35] J.-J. Zhang, J.-L. Zhang, S.-M. He, K.-Z. Wu, X.-D. Liu, Thermochim. Acta, 369 (2001) 157-
160.
[36] T. Yamamoto, K. Nozaki, T. Hara, The Journal of Chemical Physics, 92 (1990) 631-641.
[37] J.M. Prausnitz, R.N. Lichtenthaler, E.G. Azevedo, Molecular Thermodynamics of Fluid-Phase
Equilibria, 2nd ed., Prentice-Hall, New Jersey, 1986.
[38] J. Gmehling, B. Kolbe, M. Kleiber, J. Rarey, Chemical Thermodynamics for Process
Simulation, Wiley-VHC, Weinheim, 2012.
[39] J.A.P. Coutinho, S.I. Andersen, E.H. Stenby, Fluid Phase Equilibr., 103 (1995) 23-39.
[40] L. Ventola, T. Calvet, M.A. Cuevas-Diarte, M. Ramirez, H.A.J. Oonk, D. Mondieig, P.
Negrier, Phys. Chem. Chem. Phys., 6 (2004) 1786-1791.
[41] L. Carreto, A.R. Almeida, A.C. Fernandes, W.L.C. Vaz, Biophys. J., 82 (2002) 530-540.
[42] G.G. Chernik, Adv. Colloid Interfac., 61 (1995) 65-129.
[43] G. Akerlof, J. Am. Chem. Soc., 52 (1930) 2353-2368.
[44] R. Carta, S. Dernini, J. Chem. Eng. Data, 28 (1983) 328-330.
[45] P.C. Belting, J. Rarey, J. Gmehling, R. Ceriani, O. Chiavone-Filho, A.J.A. Meirelles, J. Chem.
Thermodyn., 55 (2012) 42-49.
[46] S.N. Vinogradov, R.H. Linnell, Hydrogen Bonding., Van Nostrand Reinhold Company, New
York, 1971.
[47] O. Ferreira, T. Fornari, E.A. Brignole, S.B. Bottini, Latin Am. Appl. Res., 33 (2003) 307-312.
101
SUPPLEMENTARY DATA
Purity analysis of the pure compounds
Figure S1 depicts the fractional melting curves for the pure compounds used in this
work and the fitted curves by linear regression. Note that, except for hexadecanoic acid,
polymorphic phenomena were observed in all pure compounds. However, temperature of
the polymorphic form is very close to the melting temperature of the compound, being the
difference between both temperatures of up to 1.0 K approximately. Thus, even with the
changes in the fractional melting curve profile generated by this phenomenon, purity mole
fractions are higher than 0.990 for all cases. Figure S2 show the thermogram for pure 1-
hexadecanol that present a polymorphic form, as well as fractional melting curves,
depicting this fact. Data were evaluated using the analysis software from TA Instruments
(New Castle).
Figure S1. Fractional melting curves obtained by DSC. Data were evaluated using
the analysis
software from TA Instruments (New Castle).
y = - 0.128 x + 317.540R² = 0.994
y = - 0.102 x + 327.640R² = 0.997
y = - 0.149 x + 335.940R² = 0.994
y = - 0.357 x + 311.000R² = 0.985
y = - 0.265 x + 322.920R² = 0.983
y = - 0.174 x + 331.210R² = 0.984
305
310
315
320
325
330
335
340
0 4 8 12
Tem
pera
ture / K
Total area / partial area
DtotalH /DpartialH
(with area correction)
Dodecanoic acid
Tetradecanoic acid
Hexadecanoic acid
1-Tetradecanol
1-Hexadecanol
1-Octadecanol
102
Figure S2. Thermogram and fractional melting curves for 1-hexadecanol. Data were evaluated
using the analysis software from TA Instruments (New Castle).
103
Table S1. Solid-liquid and solid-solid transitions temperatures obtained in replicates for 8 binary
mixtures a.
System x1 replic. Tfus / K Teut / K Ttrans / K
1-tetradecanol (1) + dodecanoic acid (2)
0.6018 1 299.52 298.56 278.30
2 300.73 298.53 277.72
1-hexadecanol (1) + tetradecanoic acid (2) 0.3042 1 321.80 312.57 306.71
2 321.40 312.31 305.86
3 321.58 312.46 305.82
1-hexadecanol (1) + tetradecanoic acid (2)
0.6004 1 313.56 312.39 311.39 306.73 288.21
2 313.98 312.83 311.75 306.37 288.30
1-hexadecanol (1) + tetradecanoic acid (2) 0.8994 1 320.91 312.16 298.98
2 320.69 311.87 299.04
3 320.69 311.60 298.91
4 320.85 312.02 298.98
1-octadecanol (1) + hexadecanoic acid (2)
0.4999 1 322.80 312.48 292.58
2 322.42 313.10 293.33
1-octadecanol (1) + hexadecanoic acid (2) 0.7010 b 1 326.00 321.82 315.67 303.69 292.74
2 325.99 322.27 315.56 303.69 292.70
3 326.02 322.46 315.93 302.82 293.20
1-octadecanol (1) + hexadecanoic acid (2) 0.7993 1 327.82 322.69 314.86 307.70 304.76
2 327.80 321.91 314.51 307.83 304.31
3 327.71 321.94 314.93 308.03 304.84
a the mean standard deviation taking into account Tfus, Teut and Ttrans and related to the results
obtained for the systems with 3-4 replicates is 0.24 K. b
Results for this composition were depicted
in Figure S3.
104
Figure S3. Replicates of the DSC thermograms for 1-octadecanol (1) + hexadecanoic acid (2) x1 =
0.7010. In detail, magnification highlighting a solid-solid transition at T = 292.74 K.
290 300 310 320 330
Hea
t fl
ow
Temperature / K
3rd replicate
2nd replicate
1st replicate
290 300 310
105
LIPIDIC PROTIC IONIC LIQUID CRYSTALS
(published in ACS Sustainable Chemistry & Engineering, 2014, in press)
Guilherme J. Maximo, †‡
Ricardo J. B. N. Santos, ‡ José A. Lopes-da-Silva,
§ Mariana C.
Costa, ║
Antonio J. A. Meirelles † and João A. P. Coutinho*
‡
† Laboratory of Extraction, Applied Thermodynamics and Equilibrium, School of Food
Engineering, University of Campinas, Campinas, São Paulo, Brazil. ‡ CICECO, Chemistry
Department, University of Aveiro, 3810-193, Aveiro, Portugal. § QOPNA, Chemistry Department,
University of Aveiro, 3810-193, Aveiro, Portugal. ║
School of Applied Sciences, University of
Campinas, 13484-350, Limeira, São Paulo, Brazil.
GRAPHICAL ABSTRACT
Mixtures of fatty acids and ethanolamines show a complex multiphase diagram resulting in
the formation of protic ionic liquid crystals with non-Newtonian behavior.
Chapter 4
106
ABSTRACT
Protic ionic liquids (PILs) based on lipidic compounds have a range of industrial
applications, revealing the potential of oil chemistry as a sustainable basis for the synthesis
of ionic liquids. PILs of fatty acids with ethanolamines are here disclosed to form ionic
liquid crystals, and their mixtures with the parent fatty acids and ethanolamines display a
lyotropic behavior. Aiming at characterizing their rheologic and phase behavior, four
ethanolamine carboxylates and the mixtures used for their synthesis through a Brønsted
acid-base reaction are investigated. Their phase diagrams present a complex multiphase
profile, exhibiting lyotropic mesophases as well as solid-liquid biphasic domains with a
congruent melting behavior. These PILs present a high self-assembling ability and a non-
Newtonian behavior with yield stress in the liquid crystal mesophase. The appearance of
lamellar and hexagonal structures, with probably normal and inverted configurations in the
mixtures, due to the formation of the PIL is responsible for the high viscoelasticity and
notable nonideality that is mainly ruled by hydrophobic / hydrophilic interactions.
Considering their renewable origin, the formation of liquid crystalline structures, in
addition to the non-Newtonian behavior and ionic liquids properties, and the mixtures here
evaluated possess a great potential, and numerous applications may be foreseen.
KEYWORDS: Ionic liquids, Solid-liquid Equilibrium, Fatty acid, Ethanolamine, Liquid
crystal, Rheology.
107
1. INTRODUCTION
More than a remarkable presence in the open literature during the past decade, ionic
liquids and mesotherm (i.e. low melting point) salts have long since been embodied in a
large range of detergents, personal care cosmetic, pharmaceuticals and chemical marketed
products. The interest for ionic liquids, salts with melting temperatures below 100°C, was
driven by their peculiar properties: negligible vapor-pressure, wide liquid phase range, no
or limited flammability and a tunability that offers freedom for the design and optimization
of chemical processes and product formulation. Such flexibility makes them also
environmentally friendly by reducing the amount of byproducts in the industry 1-3
. A wide
range of mesotherm salts based on fatty acids and their derivatives have been used long
since as surfactants, and for many other purposes, without actually being recognized as
ionic liquids. Protic ionic liquids (PILs) are produced through a Brønsted acid-base reaction
(Figure 1) and therefore, are capable of particular chemical interactions, including proton
acceptance and proton donation. Because they are easily prepared, often from low-cost or
even renewable reactants, PILs may be used in several industrial applications besides their
use as surfactants, such as the design of electrolyte fuel cells, biocatalytic reactions,
separation, and self-assembly processes 4-6
.
Addressing the claims of sustainable chemistry, the search for natural sources for
the synthesis of ionic liquids with low environmental impacts seems to be more than a
forthcoming trend; it seems to be an imperative demand 7-9
. In this way, taking into account
the renewable aspects and relevance of several industrial sectors in the actual worldwide
economic scenario, oil chemistry shows great potential to serve as a sustainable supplier of
green-chemicals. In fact, nowadays, the world production and consumption of vegetable
108
oils is larger than 150 millions of tons 10
. Considering not only the refining process, but
also several other steps of the oil production chain, the fatty acids emerge as a widely
relevant byproduct. Being edible, chemically attractive, due to the particular properties of
their long carbon chain, and having low toxicity, they are widely used by the industry as the
anion for the production of salts, and potentially of PILs, with properties tailored by choice
of an appropriate proton acceptor 11-15
. Alkanolamines, such as mono-, di- or tri-
ethanolamine, are compounds naturally found in phospholipid biological membranes. They
are often taken as the cation of the acid-base reaction with fatty acids to produce soaps,
which are actually PILs, used in the formulation of industrial and hand-cleaners, cosmetic
creams, aerosol shave foams, and industrial lubricants due to their excellent emulsifying,
thickening and detergent ability in oil-in-water emulsions 16-19
. In particular, the
monoethanolamine oleate is long been embodied in the pharmaceutical industry. Approved
by U.S. Food and Drug Administration (FDA), it acts as sclerosant agent when a diluted
solution is administrated intravenously in the treatment of leg veins, bleeding esophageal
varices, mucosal varicosities, and other similar surgical procedures 20-23
. This is quite
curious taking into account that ILs’ research is looking for effective pharmaceutical
applications, up to now with limited success 24
.
These ionic liquids with long alkyl chains and their mixtures, as happens to other
surfactant compounds, present thermotropic and lyotropic liquid crystalline stable
mesophases 25-31
respectively, associated to complex rheologic behaviors, often
characterized by non-Newtonian flows and specific viscoelasticities 32-34
that are
characteristics that are so far poorly exploited. Because of their amphiphilic and
asymmetric nature such PILs can organize themselves in oriented crystalline matrices
above the melting temperature, despite presenting liquid flow properties, and are thus able
109
to be used as supporting media for chemical reactions, development of self-assembling
drug delivery systems, materials for membrane separation processes, or electrical
conducting media 35-36
.
This work aims at the characterization of the rheological and thermotropic behavior
of four protic ionic liquid and the solid-liquid equilibrium of the binary mixtures of their
precursors: oleic acid + monoethanolamine, oleic acid + diethanolamine, stearic acid +
monoethanolamine, and stearic acid + diethanolamine. These compounds were chosen
because of the sustainable character of fatty acids, and despite the use of ethanolamine
soaps in commercial products, these systems have not been previously investigated in the
literature. This choice of systems allows the characterization of the effects of molecular
structure, alkyl chain length, and presence of unsaturation in the thermodynamic and
rheological behavior of these systems and the protic ionic liquids that they originate.
Figure 1. Brønsted acid-base reaction between a carboxylic acid and monoethanolamine or
diethanolamine for the formation of a protic ionic liquid. R represents the alkylic moiety of the
carboxylic acid.
2. EXPERIMENTAL SECTION
2.1. Materials
R
O
OH
+OH
H2NOH
+NH3
carboxylic acid monoethanolamine monoethanolammonium carboxylate
diethanolammonium carboxylatediethanolamine
R
O
OH
carboxylic acid
HO
HN
OH
R
O
O
R
O
O HO
H2N
OH
+
110
Oleic acid, stearic acid, monoethanolamine, and diethanolamine were supplied by
Sigma-Aldrich (St. Louis, MO) with mass purities higher than 99%. Samples of
approximately 1.0 g of oleic acid + monoethanolamine, oleic acid + diethanolamine, stearic
acid + monoethanolamine, and stearic acid + diethanolamine systems were prepared
gravimetrically on flasks using an analytical balance (Mettler Toledo, Inc., Columbus, OH)
with a precision of ± 1×10−4
g and concentration of component 1 varying from x1 = (0.0 to
1.0) mole fractions for the complete description of the phase diagrams. The mixtures were
stirred at 10 K above the mesophases domain of the samples in an oil bath and under
injection of nitrogen to avoid oxidation of the compounds. The uncertainty x of the
compositions x1, obtained by error propagation from the values of the weighed masses, was
estimated as x = 5×10−4
(mole fraction).
2.2. Differential Scanning Calorimetry (DSC)
The solid-liquid equilibrium (SLE) data of the mixtures were characterized using a
DSC8500 calorimeter (PerkinElmer, Waltham, MA) previously calibrated with indium
(PerkinElmer, Waltham), naphthalene and cyclohexane (Merck, Whitehouse Station, NJ) at
ambient pressure p = (102.0 ± 0.5) kPa . The samples were submitted to a cooling-heating
cycle at 1 K min-1
based on a methodology established for measuring of the SLE of fatty
systems 37-38
. The temperatures of the thermal transitions were then analyzed in the last
heating run and taken as the peak top temperatures. The uncertainty of the temperature
wasT 0.40 K and this value was obtained by the average of the standard deviation
111
obtained by triplicates of melting temperature values of the pure compounds and mixtures
of them.
2.3. Light-Polarized Optical Microscopy (POM)
The melting profile of the liquid crystalline mesophases of the mixtures were
evaluated using a BX51 Olympus polarized optical microscope (Olympus Co., Tokyo,
Japan) equipped with a LTS120 Linkam temperature controlled stage (Linkam Scientific
Instruments, Ltd., Tadworth, U.K.) that in this study operated between 243.15 and 393.15
K. Samples (2 mg, approximately) were put in concave slides with coverslips, cooled to
243.15 K, and allowed to stay at this temperature for 30 min. After this, the samples were
observed in a 0.1 K min-1
heating run. The uncertainty of the temperature obtained from
POM measurements was taken as not higher thanPOM 1.0 K. The value was estimated
by the mean standard deviation obtained by triplicate evaluation of some of the binary
mixtures.
2.4. Density, Viscosity and Critical Micellar Concentration (CMC)
Measurements
Density and dynamic viscosity measurements of the mixtures were performed
between 288.15 and 373.15 K using an automated SVM 3000 Anton Paar rotational
Stabinger viscometer-densimeter (Anton Paar, Graz, Austria) as described elsewhere 39
.
The relative uncertainty for the dynamic viscosity was 0.35%, and absolute uncertainty for
density was 5×10−4
g cm−3
. CMC measurements of the protic ionic liquids were performed
112
by conductivity using a Mettler-Toledo SevenMulti pHmeter/Conductivimeter (Mettler
Toledo Inc., Columbus, OH) with a precision of ± 0.01 S cm-1
. Aqueous mixtures of
approximately 103
M of each protic ionic liquid were prepared with Milli-Q water and kept
at 298.15 0.1 K with a thermostatic water bath. Other concentrations were obtained by
successive dilution. The values of the CMC were taken as the interceptions of two
successive linear behaviors in the conductivity versus concentration plot.
2.5. Rheological Measurements
The stress-strain flow behavior of the protic ionic liquids was determined using a
Kinexus-pro stress controlled rotational rheometer (Malvern Instruments Ltd.,
Worcestershire, U.K.). The curves were obtained with a shear rate ranging from 0 to 100
s−1
at 298.15 – 388.15 K using a stainless steel cone and plate geometry with a diameter of
40 mm, angle of 4º, and gap set to 50 m. The temperature was controlled ( 0.1 K) by a
Peltier system built in the lower fixed flat plate.
3. RESULTS AND DISCUSSION
3.1. Phase behavior
The solid-liquid equilibrium (SLE) phase diagrams of the fatty acid + ethanolamine
systems exhibit a complex multiphasic profile sketched in Figure 2. Experimental SLE data
for these systems are reported in Supporting Information (SI). In all the binary mixtures
113
under study, there is the formation of a protic ionic liquid, i.e., the ethanolammonium salt
of a fatty acid, according to the proton transfer reaction sketched in Figure 1, as shown by
Alvarez et al. 12
. In fact, at the equimolar composition (x1 = 0.5 mole fraction), the
thermogram shows no thermal transitions below the melting temperature, unlike what was
observed for the other mixtures, as reported in the experimental data tables in SI. At this
concentration, the system comprises only the pure PIL. Considering the formation of such
an intermediate compound, the phase diagrams should be read taking into account two
well-defined regions. The left-hand side of the diagram, or the region at x = 0.0 to 0.5 fatty
acid mole fraction, describes the equilibrium between the ethanolamine and the protic ionic
liquid. The right-hand side of the diagram, or the region at x = 0.5 to 1.0 fatty acid mole
fraction, describes the equilibrium between the protic ionic liquid and the fatty acid. It
means that the four solid-liquid phase diagrams here reported display an intermediate
compound with a congruent melting profile and thus represent, indeed, eight equilibrium
profiles, namely, monoethanolammonium oleate + monoethanolamine, oleic acid +
monoethanolammonium oleate, diethanolammonium oleate + diethanolamine, oleic acid +
diethanolammonium oleate, monoethanolammonium stearate + monoethanolamine, stearic
acid + monoethanolammonium stearate, diethanolammonium stearate + diethanolamine,
and stearic acid + diethanolammonium stearate.
These systems present several thermal invariant transitions. In each phase diagram,
two of them are related to the eutectic reaction that bounds the solid phase and the other
biphasic domains. The first eutectic is observed in the ethanolamine-rich composition
region of the diagram and the second in the fatty acid-rich composition region. They are
thus related to the minimum point of the liquidus line, i.e., the curve that describes the
behavior of the melting temperature of the system. The Tammann plots of these invariant
114
transitions are reported in the SI. Their behavior is typically observed for eutectic
transitions, with the corresponding enthalpy increasing to the eutectic composition and then
decreasing after this transition 40
. These plots suggest that the systems present a full
immiscibility in the solid phase because enthalpy is close to zero at pure compound
concentration. For stearic acid-based mixtures, the extrapolation of the trend is very close
to the pure ethanolamine melting enthalpy value meaning that the eutectic point in such a
region is probably very close to pure ethanolamine concentration. The large difference
between the pure stearic ionic liquid and pure ethanolamine melting temperatures supports
such an evaluation.
Figure 3 presents the principal textures of the lyotropic and thermotropic
mesophases (in the case of mixtures or pure PIL at x1=0.5, respectively) above the liquidus
line. Five types of profiles were characterized, those described as oily textures (Figure 3A),
oily streaks (Figure 3B), oily streaks with inserted maltese crosses (Figure 3C), maltese
crosses in an isotropic background (Figure 3D), and mosaic textures (Figure 3E), related to
the appearance of a lamellar mesophase and those with well-defined fan-shaped textures
(Figures 3F-H) related to the presence of hexagonal mesophases. These structures are
typical of the liquid crystalline state 26-28,41-42
. Well-defined lamellar mesophases were
observed in all systems at a particular composition range. Hexagonal mesophases could be
clearly observed in the oleic acid-based systems at the extremes of the liquid crystal
domain, and this is in agreement with the self-assembly behavior of surfactant molecules.
In fact, in aqueous surfactant solutions, hexagonal ensembles are structured at high water
concentration, while lamellar matrices appear at intermediate concentrations26-28
. Thus,
considering that the solute, in this case, is the ethanolamine or the fatty acid, the appearance
of hexagonal mesophases before lamellar ones is expected. Moreover, when the ionic liquid
115
is in solution with the ethanolamine, the solvent is hydrophilic. Otherwise, with the fatty
acid, the solvent is hydrophobic. Thus, in the ethanolamine-rich region, the hexagonal
mesophase observed probably assumes a normal structure, unlike in the fatty acid-rich
region, where it probably assumes inverted ones. Normal and inverted hexagonal
mesophases are known in the literature as type I and II, respectively. It means that, in
normal type I hexagonal mesophases, the hydrophilic moiety of the surfactant interacts with
the medium, while the hydrophobic moiety is inside of the ensemble. On the other hand, in
inverted type II hexagonal mesophases, the opposite structure is observed. Moreover, as
previously mentioned, hexagonal mesophases were not observed in the stearic acid-based
mixtures. This probably happens because the linear structure of the fatty acid carbon-chain
prefers to self-assemble in layers instead of cylindrical shapes 43
. This will be further
discussed.
Because liquid crystals were not observed in the pure ethanolamines and fatty acids,
such structures could be related to the formation of the protic ionic liquids. In fact, oily
textures and maltese crosses were clearly observed throughout the liquid crystal phase
temperature range at x = 0.5 fatty acid mole fraction, i.e., the pure protic ionic liquid. The
same textures were reported in the literature for similar compounds 44-48
. They occur due to
the amphiphilic nature of the ethanolammonium soap and the complex lyotropic behavior
of some of these mixtures, resulting from the different packaging profiles that depend on
the polarity of the coexistent compound, either the polar ethanolamine in the ethanolamine-
rich composition region or the non-polar fatty acid in the fatty acid-rich composition
region.
Taking into account the existence of these mesophases, it is possible to characterize
the other thermal invariant transitions that are observed in the biphasic domain. Oleic acid-
116
based mixtures (Figure 2A and B) present, at the left-hand side of the diagram, two
invariant transitions. The first, at a higher temperature, is related to the formation of the
hexagonal mesophase and the second is related to the formation of the lamellar mesophase.
In the right-hand side of the diagrams, the ionic liquid-rich region shows one invariant
transition related to the lamellar mesophase formation and the fatty acid-rich region
presents one transition, probably related to the formation of the inverted hexagonal
mesophase. On the other hand, because stearic acid-based mixtures (Figure 2C and D)
showed only lamellar mesophases, the invariant thermal transitions in the biphasic region
are related to the formation of such structure.
The profile of the phase diagrams, including the solid-liquid equilibrium and the
mesophases, could be sketched in Figure 2A-D, considering the experimental observations
and the Gibbs phase rule imposing that at a constant pressure and temperature, at most three
phases can coexist in a binary mixture 40,49-51
. Taking into account the coexistent phases and
mesophases, namely, solid ethanolamine (SEA), solid fatty acid (SFA), solid protic ionic
liquid (SPIL), isotropic liquid (L), normal hexagonal (HI), lamellar (L), and inverted
hexagonal (HII), it is possible to summarize all biphasic domains that are present in the
phase diagrams. In the oleic acid + monoethanolamine (Figure 2A) and oleic acid +
diethanolamine (Figure 2B) systems, we have SEA + SPIL, SEA + L, SEA + HI, SEA + L, SPIL
+ SFA, SPIL + L, SPIL + HII, SFA + HII, and SFA + L. In the stearic acid + monoethanolamine
system (Figure 2C) we have SEA + SPIL, SEA + L, SEA + L, SPIL + SFA, SPIL + L, SPIL + L,
and SFA + L. In the stearic acid + diethanolamine system (Figure 2D), we have SEA + SPIL,
SEA + L, SPIL+ L, SPIL + SFA, SPIL + L, SFA + L, and SFA + L.
117
Figure 2. Solid-liquid equilibrium phase diagram of the (A) oleic acid + monoethanolamine, (B) oleic
acid + diethanolamine, (C) stearic acid + monoethanolamine and (D) stearic acid + diethanolamine
systems by () differential scanning calorimetry and () light-polarized optical microscopy presenting
solid monoethanolamine (SMEA), solid diethanolamine (SDEA), solid fatty acid (SFA), solid protic ionic
liquid (SPIL), liquid phase (L), normal hexagonal (HI), lamellar (L), and inverted hexagonal (HII)
mesophases. Lines are guides for the eyes. Solid lines were experimentally characterized, and dashed
lines describe hypothetical phase boundaries based on the Gibbs phase rule.
260
270
280
290
300
310
320
330
340
350
360
370
380
390T
/ K
SMEA + SPILSFA + SPIL
L
L
SPIL + L
SPIL + LHII
HI
SMEA + L
SFA + L
SFA+ HIISPIL+ HII
L
L
277
281
0.00 0.12
SMEA+ L
SMEA+ HI
HI
SMEA+ L
270
280
290
300
310
320
330
340
350
360
370
380
390
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
T/
K
xfatty acid
SFA + SPIL
SPIL + L
SFA + SPIL
SPIL + L
SMEA + L
SMEA + L
L
L
SPIL + L
SFA + L
LL
L
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
xfatty acid
L L
LL
SFA + L
SFA + L
SFA + SPIL
SDEA + SPIL
SPIL + L
SMEA + L
SPIL + L
SPIL + L
L
HII
HI
L
SDEA + SPIL
SFA + L
SFA + SPIL
SDEA + L
SDEA + HI
SFA + HII
SDEA + L
SPIL + HII
SPIL + LSPIL + L
L
L
(A) (B)
(D)(C)
118
Figure 3. Light-polarized optical micrographs showing the mesophases textures: (A) oily texture in
the stearic acid + monoethanolamine at xfatty acid = 0.2, T = 343.15 K; (B) oily streaks in the oleic
acid + diethanolamine at xfatty acid = 0.6, T = 293.15 K; (C) maltese crosses in an isotropic backgroup
oleic acid + diethanolamine at xfatty acid = 0.5, T = 357.15 K; (D) oily streaks with inserted maltese
crosses oleic acid + diethanolamine at xfatty acid = 0.4, T = 340.15 K, (E) mosaic texture in the stearic
acid + monoethanolamine at xfatty acid = 0.4, T = 363.15 K; and fan-shaped textures in the (F) oleic
acid + diethanolamine at xfatty acid = 0.3, T = 298.15 K; (G) oleic acid + monoethanolamine at xfatty acid
= 0.7, 293.15 K and (H) oleic acid + diethanolamine at xfatty acid = 0.7, T = 298.15 K.
119
3.2. Self-Assembling of the PILs
It was mentioned before that the appearance of the liquid crystal mesophases is due
to the formation of the protic ionic liquid. In fact, it presents an amphiphilic and
asymmetric molecular structure, containing a hydrophilic (ethanolammonium cation) and
hydrophobic moiety (fatty acid anion). Such profile characterizes a mesogen, i.e., a
compound that under suitable temperature and pressure conditions can build liquid crystal
matrices 36,42
. For the understanding of the self-organizing behavior of the protic ionic
liquids formed by reaction of the fatty acids with ethanolamines mixtures the critical
micellar concentration of these compounds in water was measured by conductivity.
Figure 4 shows the conductivities of the diluted aqueous solutions of the four protic
ionic liquids here studied. As the conductivity increases with the concentration of the ionic
liquids in solution, the slope changes defining essentially three different linear regions.
Thus, considering that the different dependence of conductivity upon PIL concentration is
due to differences in the self-assembly of the mesogens, two CMC values were obtained (as
shown in Figure 4). Literature shows that with an increase in the surfactant concentration in
aqueous solutions, spherical micelles, cylindrical micelles, or bilayers are formed from the
monomers52
. After the first CMC value, the protic ionic liquid is probably self-assembled in
spherical micelles. In fact, taking into account the molecular geometry (local) constraints,
the critical packing parameter53
in all cases assumed values lower than 1/3. This value is
related to the formation of spherical micelles with the polar headgroup (ethanolammonium
cation) interacting with the aqueous medium (as depicted in Figure 4). Moreover,
considering the global constraints imposed by an increase in the concentration of the
mixture and intermolecular interactions, the second CMC value could be probably related
120
to the formation of more complex structures, according to similar observations in the
literature54-56
. The formation of such structures is more evident in the oleate ionic liquids
because they present a more pronounced shift in the slope and lower CMC2 values.
Assuming, in this case, both global and local constraints, the presence of the unsaturation in
the carbon-chain of the oleic ionic liquids might probably favor the formation of cylindrical
self-assembling. On the other hand, the geometry of the saturated alkyl chain of the stearic
ionic liquids probably led to the formation of bilayer structures. This observation
corroborates with the fact that stearic acid based mixtures did not present hexagonal
crystals, unlike oleic acid ones, as previously discussed.
Figure 4. Conductivity ( as a function of concentration of protic ionic liquids (PIL) aqueous
solutions. Solid lines are linear fitting of the data (R2 > 0.98) for critical micellar concentration
determination (CMC).
0
5
10
15
20
25
30
35
0.0 0.2 0.4 0.6 0.8 1.0
/
S
cm-1
PIL / mM
PIL CMC1 CMC2
[MEA][Oleic] 0.159 0.714
[DEA][Oleic] 0.184 0.632
[MEA][Stearic] 0.159 0.721
[DEA][Stearic] 0.092 0.744
more complex
structures
121
3.3. Rheological Characterization of the PILs and Binary Mixtures
Taking into account the organized nature of a liquid crystal microstructure, its
properties should present a particular behavior, or even a different profile, when compared
with an isotropic media. In fact, visually, some mixtures close to x = 0.5 fatty acid mole
fraction, including the equimolar concentration, presented a very viscous gel-like behavior.
The phase diagrams of the mixtures here studied show that above the liquidus line, there is
a very large liquid crystalline domain in a broad concentration range and with maximum
temperature values close to 390 K. Thus, density and viscosity measurements were
determined in order to characterize these fluids, fostering the understanding of the
interactions between the compounds in the mixture because in this work the compounds
react to form a new substance (density and viscosity data are reported in SI). Unfortunately,
due to their high melting points, the mixtures of stearic acid could not be studied. Figure 5
presents the densities of the oleic acid + monoethanolamine and oleic acid +
diethanolamine binary systems from 288.15 to 373.15 K because such mixtures are liquid
above 288.15 K. As a first observation, the results seem to be quite trivial with density
values at fixed concentrations decreasing linearly (R2 0.99) with increasing temperature,
ranging from = 1.10 to 0.85 g cm-3
, i.e., from pure ethanolamine to pure oleic acid density
values. However, if the density results are seen on their composition dependency at a fixed
temperature, the values also decrease with an increase in oleic acid concentration, but a
more complex behavior is revealed. Three different regions are now apparent: (1) one in the
ethanolamine rich region before the liquid crystal formation, (2) one delimited by the liquid
crystal domain (dashed lines of the Figure 5), and (3) one in the fatty acid rich region where
122
no liquid crystal is observed. These differences in the density result from the peculiar
packing profiles that liquid crystals present, which induce changes in the densities of the
mixtures. The excess molar volume VEX
/ cm3 mol
-1 data calculated from the experimental
density data (Figure 6) are in agreement with the observed nonlinear density dependency
and provide further significant details on the molecular interactions. It is important to note
that the excess molar volume was calculated considering that at x = 0.0 to 0.5 oleic acid
mole fraction the mixture is the ethanolamine + ionic liquid binary system and at x = 0.5 to
1.0 oleic acid mole fraction the system is the fatty acid + ionic liquid mixture. The results
clearly show that, at a fixed temperature, the excess molar volumes of the ethanolamine +
protic ionic liquid mixtures present a significant negative deviation from ideal behavior. On
the other hand, the excess molar volumes of the oleic acid + protic ionic liquid mixtures
present significant positive deviations. Because negative deviations were observed in the
first part of the diagram, one can infer that from the point of view of excess molar volumes
the interactions between ethanolamine and the ionic liquid are more favorable than the
interactions between fatty acid and the ionic liquid. In fact, ethanolamines and fatty acid
have quite different size, shape, and polarities. In the ethanolamine + ionic liquid mixture,
due to the polar behavior of the ethanolamine, the interactions with the hydrophilic moiety
of the ionic liquid are preferred. The large size differences may also favor the packing of
the ensemble. On the opposite, in the fatty acid + ionic liquid systems, the interactions with
the hydrophobic moiety of the ionic liquid are preferred due to the apolar behavior of the
organic acid. Thus, the rigid structure of the liquid crystal formed and the large size of the
long carbon chain probably do not allow much possibilities of packing leading to positive
deviations from ideal behavior.
123
Figure 5. Density data ( / g cm-3
) as a function of temperature (T / K) for the (A) oleic acid +
monoethalonamine and (B) oleic acid + diethanolamine systems at xoleic acid = 0.0 (), 0.1 (), 0.2
(), 0.3 (), 0.4 (), 0.5 (), 0.6 (), 0.7 (), 0.8 (), 0.9 () and 1.0 (). At the bottom of
each plot, density data as a function of oleic acid mole fraction at T = 298.15, 323.15, and 353.15 K.
Solid and dashed lines are guide for the eyes. Gray regions represent the liquid crystal domain.
124
Figure 6. Experimental excess mole volume data of the () oleic acid + monoethalonamine and
() oleic acid + diethanolamine systems at 298.15 K. Dashed lines are guides for the eyes. Vertical
gray line at x = 0.5 (pure PIL) depicts the “true” system indicated at the top of the plot.
Figure 7 shows the measured dynamic viscosity data for the same systems from
288.15 to 358.15 K except at and close to the equimolar composition for the oleic acid +
monoethanolamine system due to its high viscosity with a gel-like behavior. In the case of
pure PILs, shear rate-dependent rheological measurements were conducted and will be later
discussed. An exponential decay of the viscosity with temperature was observed, being
more pronounced close to the pure ionic liquid. This is clearly visible considering the
isothermal behavior at 353.15 K of the viscosity with the fatty acid concentration at the
bottom of Figure 7. The viscosity values increased with the concentration from pure
ethanolamine composition to the protic ionic liquid, and then they start to decrease again to
the value of pure oleic acid. Another singular behavior from the viscosity measurements
can be highlighted if the data is fitted to an Arrhenius-like exponential equation (Eq. 1)
AEln( ) ln(A)
RT (1)
-6.0
-4.0
-2.0
0.0
2.0
4.0
0.0 0.2 0.4 0.6 0.8 1.0
VE
/ cm
3
mo
l-1
xoleic acid
PIL + ethanolamine oleic acid + PIL
125
where is the dynamic viscosity (Pa s), EA is the activation energy (J mol-1
), R is the ideal
gas constant (8.314 J mol-1
K-1
), T is the absolute temperature (K), and A is the pre-
exponential factor (Pa s). Thence, Figure 8 shows the logarithm of the viscosity data, ln
(), as a function of T-1
. The fitting procedure, which is presented at the right side of Figure
8, shows that the trend presents a shift with a significant change in the activation energy EA
at a temperature (Tinters in Figure 8) close to the melting of the mesophases. Considering the
meaning of the coefficients of the Arrhenius equation, the results reveal that the activation
energy of the liquid crystalline state is higher than that in the isotropic liquid phase. The
crystalline mesophase needs, thus, more energy to flow than the isotropic liquid 57
.
Figure 7. Dynamic viscosity data of (A) oleic acid + monoethanolamine and (B) oleic acid +
diethanolamine at xoleic acid = 0.0 (); 0.1 (); 0.2 (); 0.3 (); 0.4 (); 0.5 (); 0.6 (); 0.7 ();
0.8 (); 0.9 (); 1.0 (). Details at the top right of the plots show magnifications of the data at
low values. At the bottom of each plot are viscosity data as a function of fatty acid mole fraction
at 353.15 K. Vertical gray lines at x = 0.5 (pure PIL) depict the “true” systems indicated above.
0
500
1000
1500
2000
2500
3000
3500
285 300 315 330 345 360 375
/ m
Pa
s
T / K
(A)
285 300 315 330 345 360 375
T / K
(B)
0
70
140
0.0 0.2 0.4 0.6 0.8 1.0
/ m
Pa
s
xoleic acid
353.15 K
0.0 0.2 0.4 0.6 0.8 1.0xoleic acid
353.15 K
0
250
500
285 300
0
250
500
290 300 310
PIL + ethanolamine oleic acid + PIL PIL + ethanolamine oleic acid + PIL
126
Figure 8. Linearization procedure of the dynamic viscosity data using the Arrhenius like equation
1. (A) Oleic acid + monoethanolanmine and (B) oleic acid + diethanolamine at xoleic acid = 0.0 ();
0.1 (); 0.2 (); 0.3 (); 0.4 (); 0.5 (); 0.6 (); 0.7 (); 0.8 (); 0.9 (); 1.0 (). At the
right of each plot are linear fittings of the data (R2 > 0.99) at some concentrations. Gray regions
represent the liquid crystal domain.
Figure 9 shows some of the flow curves of the oleate-based PILs (x = 0.5 oleic acid
mole fraction) from 293.15 to 363.15 K, describing the shear rate-dependence of the
samples’ shear stress. For the monoethanolammonium oleate sample at T > 348.15 K, the
y = 41.55x - 2.57
y = 51.44x - 5.64
9
13
Tinters. = 324 K
7.50
8.75
10.00
11.25
12.50
13.75
15.00
ln (/
Pas)
(A)
y = 54.20x - 4.27
y = 64.73x - 7.54
11
14
Tinters. = 322 K
y = 47.02x - 2.89
y = 57.47x - 6.29
10
14
0.28 0.3510-2 T-1
Tinters. = 307 K
y = 55.49x - 4.22
y = 67.19x - 7.99
11
15
Tinters. = 310 K
y = 41.55x - 2.57
y = 53.84x - 5.16
10
14
Tinters. = 307 K
y = 60.93x - 6.15
y = 72.75x - 9.94
11
15
Tinters. = 312 K
7.50
8.75
10.00
11.25
12.50
13.75
15.00
0.27 0.29 0.31 0.33 0.35
ln (/
Pa
s)
10-2T-1
(B)
127
shear rate is directly proportional to the shear stress, with no evidence of yield stress (0),
characterizing a clear Newtonian behavior (Eq. 2). The same result is observed for
diethanolammonium oleate samples at T > 323.15 K. Below such temperature values, for
both PILs, the increase in shear rates leads to a nonlinear increase in shear stress. At 298.15
T 348.15 K, the monoethanolammonium oleate sample presented a significant yield
stress (outlined by the dashed circles in Figure 9), whose values increase with a decrease in
temperature from 0 = 10 to 70 Pa. At this range, the shear rate/shear stress dependence
changes gradually to a marked nonlinear behavior that could be adjusted by the Vocadlo
model 58-59
(Eq. 3) up to T > 293.15 K. At this low temperature, the yield stress increases to
0 > 200 Pa with a linear dependence between shear rate and shear stress such that the flow
curve is well-characterized by the Bingham plastic model (Eq. 4, R2 > 0.99).
Comparatively, the non-Newtonian temperature domain of the diethanolammonium oleate
samples did not present yield stress up to 308.15 K, where the flow curve is well adjusted
(R2 > 0.99) by a shear-thinning profile (Eq. 5). However, with the temperature decreasing
(T < 308.15 K) the flow curve presents a yield stress and the Herschel-Bulkley model (Eq.
6) could be accurately adjusted (R2 > 0.95). The model’s parameters fitted to flow
experimental data are in SI. The flow models adopted here 60
can be described as follows
(2)
Vv
n1 n
0V VK
/ (3)
0 ' (4)
n K (5)
Hn
0H H K (6)
128
where is the shear stress (Pa), is the viscosity (Pa s), is the shear rate (s-1
), 00V
0Hare the yield stresses (Pa) related to each model, K, KV and KH are the consistency
indexes related to each model, ’ is the Bingham plastic viscosity, n, nV and nH are flow
behavior indexes related to each model, and the subscripts v and H make reference to the
Vocadlo and Herschel-Bulkley models, respectively. The non-Newtonian behavior
displayed by these PILs is thus a unique characteristic of these compounds, and the flow
phenomena here observed play a key role in numerous industry applications 32-36,57
. The
self-assembly ability of the PILs and formation of thermotropic mesophases with bilayered
lamellar-oriented structures above the melting point are the main reason for the non-
Newtonian rheological behavior here described. Moreover, the chemical interactions
among the molecules appear to be strong enough for the requirement of a critical shear
stress. The increasing temperatures lead to a decrease in the apparent viscosity. This, for
practical purposes, says there is an attenuation of the non-Newtonian behavior followed by
the appearance of a Newtonian profile at sufficiently high temperatures. The temperature
increase induces the breaking of the interactions that support the liquid crystalline
microstructure. The concentration of the isotropic phase increases with depletion of the
mesophase, as observed in the phase diagrams (Figure 1) and in the polarized optical
micrographs (Figure 2A, typical lamellar texture at low temperatures; Figure 2C, textures at
high temperatures). It was also observed that the non-Newtonian behavior of
monoethanolammonium oleate is more evident than for the diethanolammonium oleate in
special at low temperatures where such a PIL follows a Bingham plastic profile, presenting
very high yield stresses. So, regarding the size of the cation in the surfactant molecule
structure, one might suppose that the bilayers constructed by the monoethanolammonium
129
oleate are more packed with stronger intermolecular hydrogen bonding networks than the
diethanolammonium oleate structure, whose larger cation decreases the self-assembly
ability. In fact, complex structures are more easily formed by monoethanolammonium
oleate due to its lower CMC2 value. The non-Newtonian behavior exhibited by the
ethanolammonium oleates is quite interesting concerning their use in the formulation of
products and considering the particular physicochemical properties of the protic ionic
liquids for specific applications 4-6,12,15-17,35-36
.
Figure 9. Shear stress () versus shear rate ( ) curves of (A) monoethanolammonium oleate and
(B) diethanolammonium oleate at 293.15 (), 298.15 (), 308.15 (), 323.15 (), 328.15 (),
348.15 (), 353.15 (+), and 363.15 () K. Solid lines are fitted curves. Labels N, ST, V, BP, and HB
stand for Newtonian, shear-thinning, Vocadlo, Bingham plastic, and Herschel-Bulkley models,
respectively.
0 20 40 60 80 100
/ s-1
(B)
.
0
100
200
300
400
500
600
700
0 20 40 60 80 100
/ P
a
/ s-1
(A)
.
0
15
30
0 50 100
0
5
10
15
20
25
30
0 50 100
N
V
BP
N
ST
HB
V
V
V
ST
N
HB
130
4. CONCLUSIONS
Four PILs based on oleic and stearic acids, with mono- and di-ethanolamines as
cations, and the binary mixtures used for their preparation were here investigated. The
protic ionic liquids are shown to display an interesting thermotropic behavior with the
formation of lamellar mesophases. The SLE phase diagrams of the mixtures of their
precursors display a congruent melting behavior with lamellar and hexagonal lyotropic
mesophases. Moreover, the oleic acid based systems also present a non-Newtonian
behavior clearly observed in the flow curves obtained by the stress-controlled rheological
assays. Although many ethanolammonium carboxylates have a range of industrial
applications known, considering their renewable origin, the novel characteristics here
reported, such as the formation of liquid crystalline structures in addition to the non-
Newtonian behavior and the unique properties resulting from their ionic liquid character,
confer to the compounds here investigated a great potential. On the basis of these results,
numerous applications may be foreseen for these compounds, besides those currently in
use, which should be the object of future studies.
SUPPORTING INFORMATION
Tables with experimental solid-liquid equilibrium data, Tammann plots of eutectic
transitions, conductivity data for aqueous solutions of PILs, density, and viscosity data for
the binary mixtures, shear stress, and shear rate data of PILs at different temperatures, and
rheological models’ parameters fitted to experimental flow curves. This material is
available free of charge via the Internet at http://pubs.acs.org.
131
ACKNOWLEDGEMENTS
The authors are grateful to the national funding agencies FAPESP (2012/05027-1,
2008/56258-8), CNPq (304495/2010-7, 483340/2012-0, 140718/2010-9, 479533/2013-0)
and CAPES (BEX-13716/12-3) from Brazil, and FCT (Pest-C/CTM/LA0011/2013) from
Portugal for the financial support. The authors declare no competing financial interest.
REFERENCES
(1) Brennecke, J. F.; Maginn, E. J. Ionic liquids: Innovative fluids for chemical processing.
Aiche J. 2001, 47 (11), 2384-2389.
(2) Plechkova, N. V.; Seddon, K. R. Applications of ionic liquids in the chemical industry.
Chem. Soc. Rev. 2008, 37 (1), 123-150.
(3) Rogers, R. D.; Seddon, K. R. Ionic liquids-solvents of the future? Science 2003, 302
(5646), 792-793.
(4) Greaves, T. L.; Weerawardena, A.; Krodkiewska, I.; Drummond, C. J. Protic ionic liquids:
Physicochemical properties and behavior as amphiphile self-assembly solvents. J. Phys. Chem. B
2008, 112 (3), 896-905.
(5) Greaves, T. L.; Weerawardena, A.; Fong, C.; Krodkiewska, I.; Drummond, C. J. Protic
ionic liquids: Solvents with tunable phase behavior and physicochemical properties. J. Phys. Chem.
B 2006, 110 (45), 22479-22487.
(6) Belieres, J. P.; Angell, C. A. Protic ionic liquids: Preparation, characterization, and proton
free energy level representation. J. Phys. Chem. B 2007, 111 (18), 4926-4937.
(7) Ventura, S. P. M.; Gurbisz, M.; Ghavre, M.; Ferreira, F. M. M.; Gonçalves, F.; Beadham,
I.; Quilty, B.; Coutinho, J. A. P.; Gathergood, N. Imidazolium and Pyridinium Ionic Liquids from
Mandelic Acid Derivatives: Synthesis and Bacteria and Algae Toxicity Evaluation. ACS
Sustainable Chem. Eng. 2013, 1 (4), 393-402.
(8) Handy, S. T. Greener solvents: Room temperature ionic liquids from biorenewable sources.
Chem-Eur. J. 2003, 9 (13), 2938-2944.
132
(9) Hou, X.-D.; Li, N.; Zong, M.-H. Facile and simple pretreatment of sugar cane bagasse
without size reduction using renewable ionic liquids–water mixtures. ACS Sustainable Chem. Eng.
2013, 1 (5), 519-526.
(10) FAO OECD-FAO Agricultural Outlook 2013. Available at http://www.fao.org. Accessed in
04/01/2013
(11) Chu, Z.; Feng, Y. Vegetable-derived long-chain surfactants synthesized via a “green” route.
ACS Sustainable Chem. Eng. 2012, 1 (1), 75-79.
(12) Alvarez, V. H.; Mattedi, S.; Martin-Pastor, M.; Aznar, M.; Iglesias, M. Synthesis and
thermophysical properties of two new protic long-chain ionic liquids with the oleate anion. Fluid
Phase Equilibr. 2010, 299 (1), 42-50.
(13) Reinish, M. D. Soaps from fatty acids. J. Am. Oil Chem. Soc. 1952, 29 (11), 506-510.
(14) Maag, H. Fatty-acid derivatives - Important surfactants for household, cosmetic and
industrial purposes. J. Am. Oil Chem. Soc. 1984, 61 (2), 259-267.
(15) de Souza, R. L.; de Faria, E. L. P.; Figueiredo, R. T.; Freitas, L. d. S.; Iglesias, M.; Mattedi,
S.; Zanin, G. M.; dos Santos, O. A. A.; Coutinho, J. A. P.; Lima, Á. S.; Soares, C. M. F. Protic ionic
liquid as additive on lipase immobilization using silica sol–gel. Enzyme Microb. Tech. 2013, 52 (3),
141-150.
(16) Trusler, R. B. Ethanolamine soaps. Ind. Eng. Chem. 1929, 21, 685-687.
(17) Zhu, S.; Heppenstall-Butler, M.; Butler, M. F.; Pudney, P. D. A.; Ferdinando, D.; Mutch, K.
J. Acid soap and phase behavior of stearic acid and triethanolamine stearate. J. Phys. Chem. B 2005,
109 (23), 11753-11761.
(18) Zhu, S.; Pudney, P. D. A.; Heppenstall-Butler, M.; Butler, M. F.; Ferdinando, D.; Kirkland,
M. Interaction of the acid soap of triethanolamine stearate and stearic acid with water. J. Phys.
Chem. B 2007, 111 (5), 1016-1024.
(19) Pudney, P. D. A.; Mutch, K. J.; Zhu, S. P. Characterising the phase behaviour of stearic
acid and its triethanolamine soap and acid-soap by infrared spectroscopy. Phys. Chem. Chem. Phys.
2009, 11 (25), 5010-5018.
(20) Meyer, N. E. Monoethanolamine oleate: A new chemical for the obliteration of varicose
veins. Am. J. Surg. 1938, 40 (3), 628-629.
(21) Yamamoto, K.; Sakaguchi, H.; Anai, H.; Tanaka, T.; Morimoto, K.; Kichikawa, K.;
Uchida, H. Sclerotherapy for simple cysts with use of ethanolamine oleate: Preliminary experience.
Cardiovasc. Inter. Rad. 2005, 28 (6), 751-755.
(22) Kiripolsky, M. G. Commentary: more on ethanolamine oleate as a vascular sclerosant.
Dermatol. Surg. 2010, 36 (7), 1153-1154.
133
(23) Gomes, C. C.; Gomez, R. S.; do Carmo, M. A.; Castro, W. H.; Gala-Garcia, A.; Mesquita,
R. A. Mucosal varicosities: case report treated with monoethanolamine oleate. Med. Oral Patol.
Oral Cir. Bucal 2006, 11 (1), E44-6.
(24) Hough, W. L.; Smiglak, M.; Rodriguez, H.; Swatloski, R. P.; Spear, S. K.; Daly, D. T.;
Pernak, J.; Grisel, J. E.; Carliss, R. D.; Soutullo, M. D.; Davis, J. J. H.; Rogers, R. D. The third
evolution of ionic liquids: active pharmaceutical ingredients. New J. Chem. 2007, 31 (8), 1429-
1436.
(25) Greaves, T. L.; Weerawardena, A.; Fong, C.; Drummond, C. J. Formation of amphiphile
self-assembly phases in protic ionic liquids. J. Phys. Chem. B 2007, 111 (16), 4082-4088.
(26) Lawrence, M. J. Surfactant systems - Their use in drug-delivery. Chem. Soc. Rev. 1994, 23
(6), 417-424.
(27) Sagar, G. H.; Arunagirinathan, M. A.; Bellare, J. R. Self-assembled surfactant nano-
structures important in drug delivery: A review. Indian J. Exp. Biol. 2007, 45 (2), 133-159.
(28) Muller-Goymann, C. C. Physicochernical characterization of colloidal drug delivery
systems such as reverse micelles, vesicles, liquid crystals and nanoparticles for topical
administration. Eur. J. Pharm. Biopharm. 2004, 58 (2), 343-356.
(29) Bowlas, C. J.; Bruce, D. W.; Seddon, K. R. Liquid-crystalline ionic liquids. Chem.
Commun. 1996, (14), 1625-1626.
(30) Holbrey, J. D.; Seddon, K. R. The phase behaviour of 1-alkyl-3-methylimidazolium
tetrafluoroborates; ionic liquids and ionic liquid crystals. J. Chem. Soc. Dalton 1999, (13), 2133-
2139.
(31) Li, X.-W.; Zhang, J.; Dong, B.; Zheng, L.-Q.; Tung, C.-H. Characterization of lyotropic
liquid crystals formed in the mixtures of 1-alkyl-3-methylimidazolium bromide/p-xylene/water.
Colloid. Surface. A 2009, 335 (1–3), 80-87.
(32) Burrell, G. L.; Dunlop, N. F.; Separovic, F. Non-Newtonian viscous shear thinning in ionic
liquids. Soft Matter 2010, 6 (9), 2080-2086.
(33) Amann, T.; Dold, C.; Kailer, A. Rheological characterization of ionic liquids and ionic
liquid crystals with promising tribological performance. Soft Matter 2012, 8 (38), 9840-9846.
(34) Mukherjee, I.; Manna, K.; Dinda, G.; Ghosh, S.; Moulik, S. P. Shear- and temperature-
dependent viscosity behavior of two phosphonium-based ionic liquids and surfactant triton x-100
and their biocidal activities. J. Chem. Eng. Data 2012, 57 (5), 1376-1386.
(35) Binnemans, K. Ionic liquid crystals. Chem. Rev. 2005, 105 (11), 4148-4204.
(36) Gin, D. L.; Pecinovsky, C. S.; Bara, J. E.; Kerr, R. L. Functional lyotropic liquid crystal
materials. Struct. Bond. 2008, 128, 181-222.
134
(37) Costa, M. C.; Rolemberg, M. P.; Boros, L. A. D.; Krahenbuhl, M. A.; de Oliveira, M. G.;
Meirelles, A. J. A. Solid-liquid equilibrium of binary fatty acid mixtures. J. Chem. Eng. Data 2007,
52 (1), 30-36.
(38) Maximo, G. J.; Costa, M. C.; Meirelles, A. J. A. Solid-liquid equilibrium of triolein with
fatty alcohols. Braz. J. Chem. Eng. 2013, 30 (1), 33-43.
(39) Carvalho, P. J.; Regueira, T.; Santos, L. M. N. B. F.; Fernandez, J.; Coutinho, J. A. P.
Effect of water on the viscosities and densities of 1-butyl-3-methylimidazolium dicyanamide and 1-
butyl-3-methylimidazolium tricyanomethane at atmospheric pressure. J. Chem. Eng. Data 2010, 55
(2), 645-652.
(40) Chernik, G. G. Phase-equilibria in phospholipid water-systems. Adv. Colloid Interfac.
1995, 61, 65-129.
(41) Greaves, T. L.; Weerawardena, A.; Fong, C.; Drummond, C. J. Many protic ionic liquids
mediate hydrocarbon-solvent interactions and promote amphiphile self-assembly. Langmuir 2007,
23 (2), 402-404.
(42) Dierking, I., Textures of Liquid Crystals. Wiley-VCH Weinheim, 2003.
(43) Novales, B.; Navailles, L.; Axelos, M.; Nallet, F.; Douliez, J.-P. Self-assembly of fatty
acids and hydroxyl derivative salts. Langmuir 2008, 24 (1), 62-68.
(44) Sagnella, S. M.; Conn, C. E.; Krodkiewska, I.; Moghaddam, M.; Drummond, C. J.
Endogenous nonionic saturated monoethanolamide lipids: solid state, lyotropic liquid crystalline,
and solid lipid nanoparticle dispersion behavior. J. Phys. Chem. B 2010, 114 (4), 1729-1737.
(45) Sagnella, S. M.; Conn, C. E.; Krodkiewska, I.; Drummond, C. J. Nonionic diethanolamide
amphiphiles with saturated hydrocarbon chains: Neat crystalline and lyotropic liquid crystalline
phase behavior. J. Colloid. Interf. Sci. 2012, 385 (1), 87-95.
(46) Sagnella, S. M.; Conn, C. E.; Krodkiewska, I.; Drummond, C. J. Nonionic diethanolamide
amphiphiles with unsaturated C18 hydrocarbon chains: thermotropic and lyotropic liquid crystalline
phase behavior. Phys. Chem. Chem. Phys. 2011, 13 (29), 13370-13381.
(47) Fay, H.; Meeker, S.; Cayer-Barrioz, J.; Mazuyer, D.; Ly, I.; Nallet, F.; Desbat, B.; Douliez,
J.-P.; Ponsinet, V.; Mondain-Monval, O. Polymorphism of natural fatty acid liquid crystalline
phases. Langmuir 2011, 28 (1), 272-282.
(48) Zheng, M.; Wang, Z.; Liu, F.; Mi, Q.; Wu, J. Study on the microstructure and rheological
property of fish oil lyotropic liquid crystal. Colloid. Surface. A 2011, 385 (1–3), 47-54.
(49) Reid, R. C.; Prausnitz, J. M.; Pouling, B. E., The Properties of Gases and Liquids. McGraw-
Hill: New York, 1987.
(50) Laughlin, R. G., The Aqueous Phase Behavior of Surfactants. Academic Press: London,
1994.
135
(51) Chernik, G. G. A differential scanning calorimetry study of a binary-system .1.
Interpretation of DSC curves for isothermal and nonisothermal phase-transitions in binary-systems.
J. Colloid. Interf. Sci. 1991, 141 (2), 400-408.
(52) Figueira-Gonzalez, M.; Francisco, V.; Garcia-Rio, L.; Marques, E. F.; Parajo, M.;
Rodriguez-Dafonte, P. Self-aggregation properties of ionic liquid 1,3-didecyl-2-methylimidazolium
chloride in aqueous solution: from spheres to cylinders to bilayers. J. Phys. Chem. B 2013, 117
(10), 2926-2937.
(53) Hyde, S.; Andersson, S.; Larsson, K.; Blum, Z.; Landh, T.; Lidin, S.; Ninham, B. W., The
Language of Shape - The Role of Curvature in Condensed Matter: Physics, Chemistry and Biology.
Elsevier Science: Amsterdam, 1997.
(54) Kucharski, S. Surface properties and micelle formation of n,n-di-(2-hydroxy-3-
alkoxypropyl)ethanolamine and n-(2-hydroxy-3-alkoxypropyl)diethanolamine surfactants. J.
Colloid. Interf. Sci. 1978, 64 (2), 300-310.
(55) Malik, W. U.; Chand, P. Critical micelle concentration of lauric acid-diethanolamine
condensate by polarographic and spectrophotometric methods - A comparative study. J. Am. Oil
Chem. Soc. 1966, 43 (7), 446-448.
(56) Mukherjee, P.; Dash, S.; Patel, S.; Mishra, B. K. Self-organized assemblies of surfactants
derived from ethanolamines. J. Disper. Sci. Technol. 2012, 33 (6), 881-886.
(57) Weiss, R. G. Thermotropic liquid crystals as reaction media for mechanistic
investigations1. Tetrahedron 1988, 44 (12), 3413-3475.
(58) Rao, M. A., Rheology of Fluid and Semisolid Foods. Springer: New York, 2007.
(59) Vocadlo, J. J.; MooYoung, M. Rheological properties os some commercially available fats.
Can. Inst. Food Technol. J. 1969, 2, 137-140.
(60) Rao, M. A. Rheology of liquid foods - A review. J. Texture Stud. 1977, 8 (2), 135-168.
136
137
SUPPORTING INFORMATION
Figure S1. Tammann plots of the eutectic transitions Heut in the ethanolamine-rich region (x = 0.0
to 0.5 fatty acid mole fraction) and in the fatty acid-rich region (x = 0.5 to 1.0 fatty acid mole
fraction) of the (+) oleic acid + monoethanolamine, () oleic acid + diethanolamine, () stearic
acid + monoethanolamine and () stearic acid + diethanolamine systems. Solid lines are linear
fitting of the data (R2 > 0.95 considering that y = 0.0 at the pure compounds concentration). Dashed
lines are supposed plots close to pure component concentration. () Diethanolamine and ()
monoethanolamine melting enthalpy for analysis purpose.
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
0.0 0.2 0.4 0.6 0.8 1.0
H
eu
t/ k
Jm
ol-1
xfatty acid
138
Table S1. Experimental solid-liquid equilibrium data for oleic acid + ethanolamine systems for
mole fraction x, temperature T †
and pressure p = 102.0 kPa
‡.
oleic acid (1) + monoethanolamine (2) oleic acid (1) + diethanolamine (2)
x1 Teut / K T1 / K Tm / K TLCm / K x1 Teut / K T1 / K T2 / K Tm / K TLCm / K
0.0000 280.01 0.0000
300.75
0.1008 328.15 0.1033 272.81 275.22 282.53 296.43
0.1025 271.80 277.97 279.24 0.1993 275.25 277.38 280.74 293.00
0.1964 373.15 0.2986
308.15
0.2058 272.65 276.31 0.3206 275.69 278.15
281.98
0.2504 274.58 275.90 0.4089
353.15
0.2953 272.44 277.40 0.4097 274.48
279.89
0.2999 382.15 0.4445 273.76
283.07
0.3381 273.65 278.59 0.5000
283.60 360.95
0.3972 271.67 281.54 0.5383 262.72 271.29
283.65
0.4001 376.45 0.6200
323.15
0.4259 272.35 284.22 0.6184 262.83 271.35
275.00
0.5000 288.08 358.15 0.6880
304.35
0.5626 276.89 281.96 284.79 0.6940
265.37
0.6020 324.35 0.7984 265.80
271.40 279.97
0.6042 277.15 280.92 284.23 0.9019 267.06
270.48 285.08
0.7000 298.05 1.0000
287.11
0.7200 278.82
0.7762 279.19 281.53
0.8991 278.48 282.26 285.43
1.0000 287.16
† Teut is the temperature of the eutectic transition, T1 and T2 are the temperatures of the other
invariant transitions, Tm is the melting temperature (liquidus line) and TLCm is the melting of the
liquid crystal. ‡ Uncertainties for mole fraction and pressure are 0.0005 and 0.5 kPa, respectively. Uncertainty
for temperatures Teut, T1, T2, Tm is 0.4 K and for temperature TLCm is 1.0 K.
139
Table S2. Experimental solid-liquid equilibrium data for stearic acid + ethanolamine systems for
mole fraction x, temperature T †
and pressure p = 102.0 kPa
‡.
stearic acid (1) + monoethanolamine (2) stearic acid (1) + diethanolamine (2)
x1 Teut / K Tm / K TLCm / K x1 Teut / K T1 / K Tm / K TLCm / K
0.0000
280.01
0.0000
300.80
0.0230
348.45 0.0987 298.68
324.24
0.0996 277.12 339.08
0.2009 298.05
327.38
0.1000
375.00 0.2983 297.99 328.88 331.78 353.15
0.1971 274.93 344.16
0.3957
373.15
0.2004
388.15 0.4103 298.08 328.97 336.36
0.2973 277.18 346.75
0.5000
339.48 380.15
0.3960
388.15 0.5521 330.43
333.28
0.3996 276.49 350.58
0.5940
355.15
0.5000
351.65 373.15 0.6088
330.55
0.5455 344.15 351.15
0.6698 331.16
332.98
0.5495
365.15 0.6700
334.00
0.6061 344.06 350.65
0.8053 332.18 334.55 337.84
0.6600
351.15
0.8944 331.90 334.65 341.53
0.7078 340.60 350.40
1.0000
343.57
0.7936 341.35 347.34
0.9003
342.06
1.0000
343.57
† Teut is the temperature of the eutectic transition, T1 is the temperature of the other invariant
transitions, Tm is the melting temperature (liquidus line) and TLCm is the melting of the liquid crystal. ‡ Uncertainties for mole fraction and pressure are 0.0005 and 0.5 kPa, respectively. Uncertainty
for temperatures Teut, T1, T2, Tm is 0.4 K and for temperature TLCm is 1.0 K.
140
Table S3. Density data ( / g cm-3
) of the oleic acid-based binary mixtures.
oleic acid (1) + monoethanolamine (2)
x1
T / K 0.0000 0.1008 0.1964 0.2999 0.4001 0.5000 0.6020 0.6700 0.7900 0.8991 1.0000
288.15 1.021 0.989 0.971 - - - - 0.927 0.914 0.905 0.896
293.15 1.017 0.985 0.968 - - 0.944 0.931 0.924 0.911 0.901 0.892
298.15 1.013 0.981 0.965 - - 0.940 0.928 0.920 0.907 0.898 0.889
303.15 1.009 0.978 0.961 - - 0.937 0.925 0.917 0.904 0.894 0.885
308.15 1.005 0.974 0.958 - - 0.934 0.921 0.913 0.901 0.891 0.882
313.15 1.001 0.971 0.955 - - 0.931 0.918 0.910 0.898 0.887 0.879
318.15 0.997 0.967 0.952 - - 0.928 0.915 0.906 0.894 0.884 0.875
323.15 0.993 0.964 0.949 - - 0.925 0.912 0.903 0.891 0.881 0.872
328.15 0.989 0.961 0.945 - - 0.922 0.908 0.900 0.888 0.877 0.868
333.15 0.985 0.958 0.942 - - 0.919 0.905 0.896 0.884 0.874 0.865
338.15 0.981 0.954 0.939 - - 0.916 0.902 0.893 0.881 0.871 0.861
343.15 0.977 0.951 0.935 - - 0.913 0.899 0.890 0.878 0.867 0.858
348.15 0.973 0.947 0.932 - - 0.910 0.895 0.886 0.875 0.864 0.855
353.15 0.969 0.943 0.929 - - 0.906 0.892 0.883 0.871 0.861 0.851
oleic acid (1) + diethanolamine (2)
x1
T / K 0.000 0.1033 0.1993 0.3010 0.4089 0.5000 0.6000 0.6940 0.8050 0.9019 1.000
293.15 1.098 1.054 1.025 1.000 - - - 0.931 0.916 0.902 0.892
298.15 1.094 1.051 1.022 0.996 - 0.960 0.944 0.928 0.913 0.899 0.889
303.15 1.091 1.048 1.018 0.993 - 0.958 0.941 0.925 0.909 0.896 0.885
308.15 1.088 1.044 1.015 0.990 - 0.955 0.937 0.921 0.906 0.892 0.882
313.15 1.084 1.041 1.012 0.987 - 0.953 0.934 0.918 0.903 0.889 0.879
318.15 1.081 1.038 1.009 0.984 - 0.950 0.931 0.915 0.899 0.885 0.875
323.15 1.078 1.035 1.006 0.981 - 0.947 0.928 0.912 0.896 0.882 0.872
328.15 1.075 1.032 1.003 0.978 - 0.944 0.925 0.909 0.893 0.879 0.868
333.15 1.071 1.029 0.999 0.975 - 0.941 0.922 0.905 0.889 0.875 0.865
338.15 1.068 1.025 0.997 0.972 - 0.938 0.919 0.902 0.886 0.872 0.861
343.15 1.065 1.022 0.993 0.969 - 0.935 0.915 0.899 0.883 0.869 0.858
348.15 1.061 1.019 0.990 0.966 - 0.932 0.912 0.896 0.880 0.865 0.855
353.15 1.058 1.016 0.987 0.963 - 0.929 0.909 0.893 0.876 0.862 0.851
141
Table S4. Dinamic viscosity ( / mPas) data of the oleic acid-based binary mixtures.
oleic acid (1) + monoethanolamine (2)
x1
T / K 0.0000 0.1008 0.1964 0.2999 0.4001 0.5000 0.6020 0.6700 0.7900 0.8991 1.0000
288.15 31.30 204.61 634.31 - - - - 744.92 210.56 81.52 42.47
293.15 24.14 148.95 441.61 - - - 3070.60 543.59 161.51 64.42 34.22
298.15 18.94 109.37 310.11 - - - 2069.50 398.04 124.75 51.81 28.30
303.15 15.11 82.03 224.88 - - - 1441.80 299.60 98.36 42.18 23.55
308.15 12.22 62.55 166.37 - - - 1028.80 229.27 78.66 34.76 19.81
313.15 10.02 48.52 126.06 - - - 749.08 179.05 63.86 28.94 16.76
318.15 8.31 38.08 96.30 - - - 556.01 140.33 52.18 24.39 14.42
323.15 6.97 30.31 75.12 - - - 419.19 111.94 43.19 20.72 12.46
328.15 5.90 24.45 59.38 - 369.41 - 320.92 90.35 36.09 17.75 10.85
333.15 5.05 20.00 47.74 - 301.39 - 249.20 73.88 30.53 15.34 9.49
338.15 4.35 16.48 38.66 - 236.87 - 195.50 60.70 25.90 13.33 8.39
343.15 3.78 13.75 31.77 - 182.26 - 155.19 50.43 22.20 11.67 7.45
348.15 3.31 11.59 26.36 - 142.32 - 124.15 42.23 19.16 10.27 6.65
353.15 2.91 9.87 22.14 - 112.36 - 100.17 35.62 16.67 9.10 5.94
358.15 - - - - 89.73 - - - - - -
363.15 - - - - 72.22 - - - - - -
368.15 - - - - 58.63 - - - - - -
358.15 - - - - 89.73 - - - - - -
oleic acid (1) + diethanolamine (2)
x1
T / K 0.000 0.1033 0.1993 0.3010 0.4089 0.5000 0.6000 0.6940 0.8050 0.9019 1.000
293.15 859.75 1483.70 2123.60 2890.50 - - - 604.10 160.54 60.43 34.22
298.15 565.34 973.89 1391.60 1893.20 - - - 434.85 123.88 48.73 28.30
303.15 383.69 659.35 944.52 1274.90 - - 1007.10 322.16 97.57 39.75 23.55
308.15 266.73 456.87 655.31 878.11 - - 697.98 242.99 77.81 32.81 19.81
313.15 190.57 325.42 465.64 619.97 - - 501.73 187.33 62.89 27.33 16.76
318.15 137.49 233.45 334.96 443.22 - - 367.53 144.99 51.13 23.06 14.42
323.15 101.53 171.54 245.99 323.76 - - 275.12 114.38 42.15 19.60 12.46
328.15 75.82 128.30 183.67 240.44 - - 209.10 91.41 35.11 16.80 10.85
333.15 58.00 97.69 139.44 181.85 - - 161.66 74.07 29.62 14.52 9.49
338.15 44.83 75.16 106.96 139.50 - - 126.43 60.26 25.09 12.64 8.39
343.15 35.25 58.76 83.50 108.56 - - 100.18 49.70 21.48 11.07 7.45
348.15 28.08 46.45 66.00 85.54 - - 80.26 41.36 18.51 9.75 6.65
353.15 22.68 37.24 52.77 68.18 84.94 88.12 64.94 34.74 16.10 8.64 5.94
358.15 - - - - - 70.20 - - - - -
363.15 - - - - 54.48 56.68 - - - - -
368.15 - - - - 44.39 46.10 - - - - -
373.15 - - - - 36.56 37.97 - - - - -
378.15 - - - - 30.32 31.35 - - - - -
142
Table S5. Rheological models’ parameters related to shear stress / shear rate curves fitted to flow
experimental data for the PILs.
monoethanolammonium oleate diethanolammonium oleate
T / K 0 / Pa K † n
‡ R
2 model 0 / Pa K
† n
‡ R
2 model
293.15 217.78 4.600 1.000 0.9908 Bingham plastic 20.79 30.438 0.589 0.9634 H-B
298.15 74.22 2.260 106 0.307 0.9706 Vocadlo 14.34 10.822 0.873 0.9520 H-B
303.15 62.58 5.803 105 0.316 0.9586 Vocadlo 2.09 1.740 0.870 0.9908 H-B
308.15 39.69 1.010 106 0.298 0.9556 Vocadlo 0.00 1.706 0.813 0.9908 Shear thinning
313.15 36.47 3.479 105 0.303 0.9795 Vocadlo 0.00 0.797 0.987 0.9999 Shear thinning
318.15 26.19 2.792 105 0.296 0.9296 Vocadlo 0.00 0.486 0.988 0.9999 Shear thinning
323.15 19.39 5.939 104 0.314 0.9314 Vocadlo 0.00 0.343 0.996 0.9999 Shear thinning
328.15 25.17 1.060 105 0.298 0.9585 Vocadlo 0.00 0.251 1.000 0.9999 Newtonian
333.15 15.97 2.249 104 0.318 0.9517 Vocadlo 0.00 0.188 1.000 0.9999 Newtonian
338.15 14.10 1907.900 0.364 0.9862 Vocadlo 0.00 0.144 1.000 0.9999 Newtonian
343.15 11.10 4191.800 0.330 0.9691 Vocadlo 0.00 0.112 1.000 0.9999 Newtonian
348.15 9.21 504.509 0.343 0.9679 Vocadlo 0.00 0.087 1.000 0.9999 Newtonian
353.15 0.00 0.072 1.000 0.9999 Newtonian 0.00 0.069 1.000 0.9999 Newtonian
358.15 0.00 0.052 1.000 0.9999 Newtonian 0.00 0.055 1.000 0.9999 Newtonian
363.15 0.00 0.041 1.000 0.9999 Newtonian 0.00 0.044 1.000 0.9999 Newtonian
† KV for Vocadlo model, ’ for Bingham plastic model, KH for H-B model and for Newtonian
model. ‡ nV for Vocadlo model, n for shear-thinning model and nH for H-B model.
143
THE CRYSTAL-T ALGORITHM
(submitted to Physical Chemistry Chemical Physics)
Guilherme J. Maximo a, Mariana C. Costa
b, and Antonio J. A. Meirelles
a*
a Food Engineering Department, University of Campinas, R. Monteiro Lobato 80, 13083-862,
Campinas, São Paulo, Brazil. b School of Applied Sciences, University of Campinas, R. Pedro
Zaccaria, 1300, 13484-350, Limeira, São Paulo, Brazil.
GRAPHICAL ABSTRACT
The Crystal-T algorithm accurately depicts the SLE phase diagrams of lipidic mixtures
presenting solid solution, measured in this work, by using liquid and solid phase models.
Chapter 5
144
ABSTRACT
The thermodynamic modeling procedures sometimes fail in the correct description of the
phase behavior and their inability increases with the complexity of the system. Lipidic
mixtures present a particular phase change profile highly affected by their unique
crystalline structure. Consequently, marked deviations from ideality do not allow
reasonable accuracy by using the classical solid-liquid equilibrium (SLE) modeling
approach according to which the solid phase is assumed to be a pure component.
Considering the non-ideality of both liquid and solid phases, this study describes the
Crystal-T algorithm for the calculation of the SLE of binary mixtures. The algorithm is
aimed at the determination of the temperature in which the first and last crystal of the
mixture melts. The evaluation is focused on experimental data measured and reported in
this work for systems composed by triacylglycerols and fatty alcohols. The effectiveness of
the description of the liquidus and solidus lines of discontinuous solid solutions were tested
by using excess Gibbs energy based equations, and the group contribution UNIFAC model
for the calculation of the activity coefficients. Very low deviations between theoretical and
experimental data evidenced the strength of the algorithm, contributing to the enlargement
of the scope of the SLE modeling.
KEYWORDS: Solid-liquid equilibrium, Modeling, Solid solution, Non-ideality,
Triacylglycerols, Fatty alcohols.
145
1. INTRODUCTION
Several works in literature have long since been engaged in the understanding of the
melting and crystallization phenomena of systems formulated by lipidic compounds as well
as the modeling of their solid-liquid equilibrium (SLE) behavior1, 2
. Notwithstanding the
success of this task is inversely proportional to the complexity of the system. Fats and oils
mixtures are remarkably known by presenting a particular crystalline structure highly
dependent on the processes conditions to which the mixtures are submitted. A special and
common phenomenon in this context is the formation of solid solutions, in which crystals
of a compound are fitted into the lattice of another crystal3. Lipid-based products
commonly present this phenomenon, highly affecting the rheological and physicochemical
profile of the system4, 5
. Mixtures of triacylglycerols and fatty alcohols, here evaluated, are
an example. Fatty alcohols are used as surfactant structuring agents in lipidic-based systems
in replacement of hydrogenated vegetable oils or saturated TAGs. Moreover, TAGs and
fatty alcohols are used in microbiostatic coating for foods, formulation of organogels or
controlled release medicines6-12
.
In fact, world production and consumption of vegetable oils actually overtake 150
millions of tons per year13
. Being one of the most important items of the human diet as well
as a distinguished flavor and texture builder, fats and oils becomes also an important
renewable energy source, increasing the worldwide economic impact of the oilchemistry in
industrial and academic fields. Paradoxically, in some cases, they can act as promoters of
serious cardiovascular diseases, drawing attention of international policy and academic
research towards finding effective solutions. Taking into account these demands, the phase
equilibrium modeling is the ideal first step for the optimization and design of industrial
146
operations as well as the formulation of products with the desired properties. However,
literature presents a set of approaches that fails in describing the SLE behavior of lipidic
mixtures either by the lack of pure compounds experimental properties or by describing
them as simple eutectic systems and consequently neglecting the presence of solid
solutions. In fact, few works in literature went beyond the limitation of the simple eutectic
behavior in order to evaluate the SLE of fatty systems2, 14, 15
. For this reason, this work was
aimed at describing the SLE phase diagrams of lipid-based mixtures presenting solid
solutions by using an algorithm, namely Crystal-T. The evaluation was focused on binary
systems of triacylglycerols (TAGs) and fatty alcohols experimentally determined by
differential scanning calorimetry and microscopy.
The “Crystal-T” algorithm was developed with the aim of improving the modeling
of the SLE of lipidic based mixtures presenting solid solutions. The procedure was based in
the classical “Bubble-T” algorithm16-19
used for the determination of the bubble point in the
classical vapor-liquid equilibrium (VLE) calculation. The routine considers the non-ideality
of both liquid and solid phases and uses the classical isofugacity equilibrium
thermodynamic criteria and activity coefficient models in order to calculate the liquidus and
solidus lines of the phase diagrams. The calculation was anchored to the experimental data
obtained in the evaluation of the Tammann plots of the eutectic transition. The set of non-
linear equations was solved by an optimization routine that identified the temperature and
the composition in which the first and last crystal melts.
147
2. THEORY
2.1. Fundamentals
Considering an isobaric system, the classical thermodynamic establishes a well
defined theory for the description of the solid-liquid equilibrium (SLE) relating 3
fundamental variables: the mole fraction of the compound i in the liquid phase xi, in the
solid phase, zi and temperature T. The equation that relates these variables is built through
the isofugacity criteria17-19
. Taking into account the relation between the fugacities of the
compounds in both phases, the SLE is described by the calculation of the Gibbs energy of a
defined heating-colling process, depicted in Figure 1. Further details are in Supporting
Information. The SLE is so given by Equation 1.
Lfus tri i
fu
fus fus fus
s tri i 1
1 1 1 1ln 1ln
tr
pn
S
H Hx
R T T R
C T
T T R
T
T Tz
(1)
where iL and i
S are the activity coefficients of the component i in the liquid and solid
phases, respectively, T is the melting temperature (K) of the mixture, R is the ideal gas
constant (8.314 J·mol-1
·K-1
), Tfus and fusH are the melting temperature (K) and enthalpy
(J·mol-1
) of the component i, Ttr and trH are thermal transitions temperatures (K) and
enthalpies (J·mol-1
) of the n solid-solid transition (polymorphic forms) of the component i
and fusCp is the difference between the heat capacity (Jmol-1
K-1
) of the pure component i
of the liquid and solid phases.
148
Figure 1. Thermodynamic cycle comprising the heating, melting and cooling processes of the
system, from T to Tfus, from a solid to a subcooled liquid state taking into account the polimorphysm
phenomena at Ttr.
In order to describe the SLE of a mixture, the calculation of T, xi and zi must be
considered by using Equation 1 and the adjustment of the i equations. This is classically
obtained by a numerical optimization procedure. However, considering the high non-
linearity of Equation 1, some theoretical approximations, well applied for simplest cases,
are frequently found in literature. The general suppositions are that i) the specific heat
capacity of the pure compounds can be neglected when compared with the magnitude of the
fusH; ii) the compounds in the solid phase are immiscible such that the term related to the
non-ideality of the component i in the solid phase are zi i
S = 1.0; and iii) the liquid phase
behaves as in an ideal system, being the liquid phase activity coefficient iL
= 1.0. Also,
pure compounds’ solid-solid (polymorphic) transitions are often neglected although, when
compared to the fusH values, the magnitude of the trH can significantly impact in the
modeling results. Despite the eventual non reliability of one or other of these hypotheses,
the intrinsic nature of the adjustment procedure can generate good results. Consequently, all
those assumptions must be carefully evaluated.
A marked feature of fatty mixtures’ SLE is the appearance of solid solutions. From
the SLE point of view, solid solutions are so that zi iS 1.0. Therefore, the activity
Tfus
T Solid Subcooled
Liquid
Melting
Ttr
149
coefficients of the component i in both liquid iL and solid phases i
S should
necessarily be
taken into account. In this work, the well-known Margules equation, was used for the
calculation of both iL
and iS
models as well as the group contribution UNIFAC equation
for the description of iL. The Margules equation is a function relating T, xi or zi that is able
to accurately model systems comprising chemicals with similar molar volumes17
. In case of
binary mixtures, this equation can be written using one or two adjustable parameters,
namely the 2- or 3-suffix Margules equations, respectively Equations 2 and 3, considering
the utilization of up to quadratic or cubic terms17
as follows:
2
ij j
iexp
R
A z
T
(2)
2 3 2 3
ij ij j ij j ij ij i ij i
i j
3 4 3 4exp exp
R R;
A B z B z A B z B z
T T
(3)
where i, j = components 1, 2 and Aij and Bij (kJmol-1
) are empirical parameters related to
the thermodynamic interactions between the two compounds in the mixture. The UNIFAC
(UNIQUAC Functional-group Activity Coefficient) model is a predictive model based on
the group-contribution concept20, 21
in which the activity coefficient can be found by means
of a sum of both enthalpic and entropic contributions of the chemical groups in the mixture.
Some modifications of this model consider the temperature-dependence of the enthalpic
contribution or changes in the entropic contribution22
. Details concerning the UNIFAC
model are described elsewhere20-22
. Thus, in this work, the following approaches were used
for calculating the activity coefficients: i) original UNIFAC or UNIFAC-Dortmund models
for iL and 2-suffix Margules equation for i
S; ii) original UNIFAC model for i
L and 3-
suffix Margules equation for iS; iii) 2-suffix Margules equation for both i
L and i
S; and iv)
150
3-suffix Margules equation for both iL and i
S. In case of UNIFAC models, the parameters
where taken from literature22, 23
.
Considering the non-ideality of the solid phase (ziiS 1.0), the system can present
either continuous or discontinuous solid solutions. In this work, the mixtures evaluated
presents discontinuous solid solutions, in which the compounds are miscible in the solid
phase in a restrained concentration range. This case is analogous to the VLE of azeotropy-
heterogeneous mixtures. The classical shape of this phase diagram is sketched in Figure 2.
The curves that circumscribe the biphasic region are called liquidus and solidus lines and
describe, respectively, the melting temperature T of the mixture as a function of xi and zi. At
T = Teut, the behavior of both curves presents a discontinuity. This is due to the formation of
an additional solid phase, establishing a solid-solid-liquid equilibrium state. According to
the isofugacity criteria, this triphasic point is such that
0 0 0L L S S S S
i i i i i i i iI II
ix f z f z f (4)
where 0L
if and
0S
if are the standard state fugacities of the compound i in the liquid phase
and in both solid phases I and II. This point is the so-called eutectic point, in which the
mixture at a composition xi, ziI, zi
II melts in a minimum and single temperature Teut. From
the industrial point of view, the correct determination of the eutectic point is of utmost
importance if one is interested in the formulation of mixtures with low melting point, for
instance in order to avoid the solidification by low temperatures, as in case of biodiesel
mixtures24-26
, lubricants or controlled melting mixtures27, 28
.
151
Figure 2. Solid-liquid equilibrium phase diagram of a discontinuous binary solid solution case.
2.2. SLE algorithms
The equilibrium thermodynamic theory is long since well established for the
resolution of the vapor-liquid and liquid-liquid equilibria. On the other hand, the SLE
problems have been developed by analogies, using for instance, the same models for the
description of the non-ideality of the phases. Two algorithms had been already presented in
the literature for the description of the SLE: the stability test based on the Gibbs energy
minimization and the isoenthalpic flash calculation. Both are detailed explained elsewhere2,
14, 15, 29. In the first case, the algorithm considers that for a given temperature or composition
a system in equilibrium has the minimum Gibbs energy G. The function that represents G is
written by relating the activity coefficients of the phases iL and i
S, as described in
Equation 5.
326
0 1
Tem
per
atu
re
x1, z 1
Tfus 2
Tfus 1
Teut
xeut zeut IIzeut I
Solid Solution
Solid + Liquid
S+L
Liquid phase
Solid phase
liquidus line
solidus line
Teut x z0 1
152
nc
L L S S L L L
i
f
i i
us fu
1
S fus tri
1 1fu
s fu
tr
s
s
R ln
1 1 1 1 R
ln
i
nc n
i tr
p
G n g n g g n T x x
H Hn T z
R T T R
C T T
T
T
T
TR
S
i iln1 z
(5)
where nL , n
S , g
L and g
S are, respectively, the number of moles n and the mole Gibbs
energy g of both liquid L and solid S phases. Mathematically, the minimum of Equation 4 is
obtained by an optimization procedure in which given a defined mixture at a fixed
temperature T the composition of the system in the equilibrium condition is obtained.
Considering a binary mixture this problem becomes a bidimensional problem and the
independent variables could be for instance, the number of moles of the liquid phase niL.
The number of moles of the solid phase niS is so calculated through the mass balance of the
system.
In case of the isoenthalpic flash calculation, the algorithm considers that a system
with a known composition feeds a vessel at a fixed temperature T and pressure P where the
equilibrium condition must be attended. The final composition is calculated by equations
relating the mass balances as well as the SLE equilibrium equation (Equation 1). The
solution is based on the minimization of an objective function written by this set of
equations, being that the most known is the Rachford and Rice equation (Equation 6), as
presented elsewhere16, 30
.
2 2 2i i
i i
i=1 i 1 i 1 i
10
1 1
K nx z
K
(6)
153
where Ki is the xi / zi ratio, from Equation 1, ni is the number of moles of the compound i in
the feeding and is the ratio between the liquid phase amount L and the total feed mixture
F.
In this work, in order to evaluate the effectiveness of the proposed Crystal-T
algorithm, the Gibbs and flash algorithms were implemented for the description of the SLE
phase diagrams of the binary fatty systems. The minimization of the free Gibbs energy was
written using a procedure based on a multidirectional optimization problem. In this case, a
Sequential Quadratic Problem (SQP) algorithm was used, given by the MATLAB function
fmincon. The isoenthalpic flash calculation algorithm was implemented using an
optimization routine written in LINGO (Lindo Systems) for the minimization of the
Rachford and Rice equation (Equation 5). For this, a Generalized Reduced Gradient (GRG)
method for non-linear problems was considered. Markedly, both algorithm are highly
dependent on the initial estimate, since it must be within the biphasic domain for the
complete description of the liquidus and solidus lines of the SLE diagram. Once the
biphasic domain is unknown, the introduction of additional algorithms to test the initial
estimate or the number of phases in the equilibrium state becomes necessary.
2.3. The Crystal-T algorithm
The Crystal-T algorithm is aimed at the construction of the liquidus and solidus
lines of the SLE phase diagram, as sketched in Figure 1, which means to represent the
temperature T as a function of xi and zi. It was applied for depicting the thermodynamic
behavior of binary mixtures. For this, a routine was formulated in which the mole fraction
of the compound i in the solid phase zi and the melting temperature T are calculated in order
154
to attend the equilibrium criteria (Equation 1) for a given mole fraction of the compound i
in the liquid phase xi. The procedure was based in finding the root of the function F,
defined as the mole balance of the solid phase (Equation 7).
2
i1
i
F z (7)
where,
nLfus tri i
if
fus fus
us t
u
ri r
f s
t 1
ln1
11 1 1
exp lnp
S
H Hxz
R T T R T T
C T T
T TR
(8)
iL = f ( xi, T, parameters of the equation ) (9)
iS = f ( zi, T, parameters of the equation) (10)
0.0 < xi, zi < 1.0 ( xi, zi ; i = 1, 2 ) (11)
T > 0.0 ( T ) (12)
Figure 3 sketches the block diagram of the routine. 1) The algorithm starts for a
given composition xi, a set of phase transition properties Tfus, fusH, Ttr, trH, fusCp and
parameters for the i equations. The parameters are Aij (Jmol-1
) in case of 2-suffix Margules
equation, Aij and Bij (Jmol-1
) in case of 3-suffix-Margules equation and, for the UNIFAC
model, the structural parameters for each group k, Rk and Qk, and the group-interaction
parameters amn between the groups m and n in the mixture. A melting temperature T is
firstly assumed and iS taken as 1.00. 2) In the first interaction, zi is calculated (Equation 8),
the solid phase mole fractions are normalized and the iS calculated considering the
estimated T and the new zi values. 3) In the following step, zi is newly calculated
(Equation 8) and compared with the previously one, with a tolerance level lower than 110-
4. Once such comparison is attended, 4) the mole balance of the solid phase is then tested
155
with the same tolerance level. The values for T and z are then obtained if such criterion is
established. If the criterion is not attended, 5) a new estimate for T (T*) is done through a
modified quasi-Newton method in which only the first derivative of the function F is
considered (Equation 13). The numerical derivative of the function F’ was calculated by an
infinitesimal range of = 510-3
. The convergence is established at a tolerance level lower
than 110-5
. The new T* value is then used in the next iteration.
FT T
F '* (13)
Since the eutectic transition establishes a minimum point in the liquidus and solidus
lines, in the domain restrained by Tfus < T < Teut, being Tfus the lower melting temperature
between both pure compounds, there are two composition pairs (xi, zi) attending the
equilibrium criteria for one single temperature T. Thus, a well approximated estimate for T
in the procedure is essential. For this reason, the algorithm is calculated for 0.0 < xi < 1.0
and sequentially for 1.0 < xi < 0.0. The first calculation starts at T = Tfus,2 and the second
one at T = Tfus,1. At each iteration, the initial estimate for T is given by the melting
temperature of the previous iteration and the composition step xi 0.01. This procedure
led to the appearance of two profiles T, xi, zi as depicted in Figure 2, being the interception
of the curves the solid-solid-liquid equilibrium state (Equation 4), corresponding to the
eutectic point. The supposed extension of the melting behavior of each profile is depicted
by dashed lines in Figure 2. They are metastable melting transitions possibly present in the
heating process of the mixture. It is remarkable that in case of no discontinuity in the
behavior of the functions, the algorithm naturally attend the continuous solid solution case
with no further procedures, comprising or not minimum/maximum points.
156
Figure 3. Block diagram for the Crystal-T algorithm.
2.3.1. Adjustment of the i equation parameters
Once the algorithm for the calculation of the SLE was established, the parameters of
the 2- and 3-suffix Margules equations were adjusted using an optimization routine that
embodies the Crystal-T algorithm. Considering the calculation of iL and i
S, one, two or
four parameters were fitted depending on the approach used. In this work, the adjustment of
one parameter, in case of using UNIFAC model and 2-suffix Margules equation for iL and
iS calculation, respectively, was implemented using an algorithm based on the Nelder-
.T.
.F..T.
.F.
.T.
.F.|[zi]I
[zi]II|< 10-4
START
|1[zi]n|< 10-4
xi
Aij, Bij, Rk, Qk, amn
T = Tfus
iS = [1.0, 1.0]
END
F = 1 zi
T’ = T + 0.0005
F’ = 1 z’i
T * = T ‒ F / F’T = T *
| F | < 10-5
[zi] Izi = zi / zi
iS
[zi] II
157
Mead Simplex direct search (MATLAB function fminsearch). The objective function used
in this procedure was built as proposed by literature31
, taking into account the ratio between
the square absolute deviations between calculated and experimental (exp.) data and the
experimental uncertainties 2, as follows:
exp exp exp exp expeut euti eut eutI II
2 22 2 2 exp expexp exp exp
eut eut eut euti i eut eut eut eut I II
2 2 2 2 2
n
i T xT z z
z z z zT T T T x x
(14)
where Ti are the mixture melting temperature and (T, xi, ziI, zi
II)eut represents the eutectic
point i.e. the solid-solid-liquid equilibrium SSLE point with two solid phases I and II. The
experimental eutectic point was calculated using the well known Tammann plot, through
the evaluation of the behavior of the eutectic enthalpies as a function of the concentration,
as explained elsewhere32-34
. The uncertainties were obtained experimentally, as will be
discussed later.
In the adjustment of two parameters, in case of using UNIFAC model and 3-suffix
Margules equation for iL and i
S calculation, respectively, or using 2-suffix Margules
equation for the calculation of both iL and i
S and in the adjustment of four parameters, by
using the 3-suffix Margules equation for both phases, the numerical optimization fails in
find a unique solution. It means that there is not a single set of adjustable parameters that
answer an optimization routine. This problem is well known in the literature17, 31
concerning the calculation of non-linear thermodynamic problems. By determining both
liquidus and solidus lines, the number of independent variables increases, when compared
with description of only the liquidus line, as well as the number of adjustable parameters.
Considering that the number of experimental data points is the same, the degrees of
freedom of the problem also increase. This prevents obtaining a unique solution. For this
158
reason, a heuristic method was adopted. The problem is finding the set of parameters Aij
and Bij of the 2- and/or 3-suffix Margules equation (Equations 2 and 3) that minimizes the
deviation between calculated and experimental data (Equation 13). Considering n number
of parameters to adjust, depending on the case, n vectors with m elements were built. The m
elements were possible solutions for the problem. Thus, n m sets of calculated data were
found and the deviation (Equation 13) was calculated for each set. A multi-dimensional
surface relating the n m sets of parameters and the deviation were then found. The surface
presented a minimum region that circumscribed the set of parameters and answered the
problem.
The modeling procedure can be summarized by Figure 4. The routine is represented
by a three level nested structured: 1) the first one is the Crystal-T algorithm, where the
equilibrium point (T, xi, zi) is calculated; 2) the second one comprises the execution of the
first level for 0.0 < xi < 1.0 and sequentially for 1.0 < xi < 0.0 in order to depict the SLE
phase diagram; 3) the third one is the adjustment of the i parameters by a numerical or a
heuristic optimization procedure in which the first and the second levels are executed until
convergence is established.
159
Figure 4. Nested structure for the SLE modeling using the Crystal-T algorithm.
2.3.2. Experimental considerations
The Crystal-T algorithm was used in the modeling of SLE experimental data of the
following triacylglycerols + fatty alcohols mixtures: trilaurin + 1-hexadecanol, trilaurin +
1-octadecanol, trimyristin + 1-hexadecanol, trimyristin + 1-octadecanol, tripalmitin + 1-
hexadecanol and tripalmitin + 1-octadecanol. As previously specified, the SLE problem
takes into account the knowledge of pure compounds transitions properties. In this work,
specific heat capacity fusCp, solid-solid transitions temperature Ttr and enthalpy trH were
taken from literature2, 35-38
. Melting temperatures Tfus and enthalpies fusH of the pure
triacylglycerols and fatty alcohols (Sigma-Aldrich, 99% mass fraction) were obtained by
six-replicates using differential scanning calorimetry with a DSC8500 calorimeter
(PerkinElmer, Waltham) at ambient pressure following a specific methodology developed
for fatty systems, as described elsewhere32, 34
. Melting temperatures and enthalpies of pure
compounds were compared with literature values2, 35, 36, 38-44
and low mean relative
Numerical or heuristicoptimization procedure for the adjustment of the i equations
Evaluation of the Crystal-T algorithm for 0.0 < xi < 1.0
and for 1.0 < xi < 0.0.
Crystal - T algorithm(T, xi,zi)
1st Level
2nd Level
3rd Level
160
deviations of 1.1% and 5.5%, respectively, were observed. The triacylglycerol + fatty
alcohol mixtures were prepared gravimetrically and their melting properties were obtained
following the same methodology.
Because the beginning of the melting process (solidus line), in case of solid solution
formation is difficult to identify by using the DSC data, being often overestimated, the
mixtures were evaluated by temperature controlled microscopy using a BX51 Olympus
optical microscope (Olympus Co., Tokyo) coupled to a LTS120 Linkam temperature-
controller apparatus (Linkam Scientific Instruments Ltd., Tadworth). Samples were
observed in a 0.1 K min-1
heating run. Melting temperatures of the pure compounds and
some of their mixtures were evaluated by microscopy and results compared with the DSC
data. Deviations between experimental data obtained by DSC and by microscopy were
estimated to be not higher than 1.0 K.
The uncertainty of the mixture’ melting temperature was obtained by the evaluation
of at least three-replicates of the pure compounds and some of the binary mixtures. The
uncertainty of the eutectic point was calculated by error propagation method, considering
the equations obtained in the fitting procedure of the Tammann plot. The xi experimental
errors were also calculated by error propagation from the values of the weighted masses.
Uncertainties for melting temperatures and mole fractions as well as for the eutectic
temperature and mole fraction were estimated as not higher than 0.38 K, 0.001, 0.58
K and 0.005, respectively.
Since the evaluation of the SLE experimental data is fundamental for the accuracy
of the modeling procedure, their results will be detailed explained.
161
3. RESULTS AND DISCUSSION
3.1. Experimental data analysis
Figure 5 depicts the solid-liquid transitions experimental data of the binary mixtures
as well as the modeled liquidus and solidus lines. Tables with experimental data are
reported in the Supporting Information. The DSC technique was able to clearly show the
melting temperature behavior of the mixtures, i.e the liquidus line and the eutectic reaction,
when it exists. Figure 6 sketches the melting behavior observed by the DSC thermograms
obtained for the trimyristin (1) + 1-hexadecanol (2) mixture. In all cases, the liquidus line
shows a minimum and an invariant transition is observed at the same temperature of this
inflexion point. This behavior is typically observed in case of systems presenting eutectic
transitions.
The eutectic transition and the formation of solid solution were investigated using
Tammann plots and temperature-controlled microscopy. Figure 7 presents the Tammann
plots of the trimyristin (1) + 1-hexadecanol (2) mixture for the invariant transition observed
and Figure 8 the optical micrographs for the melting process of the same mixture at x1 =
0.694 mole fraction. At this concentration, DSC data show the presence of an invariant
transition at T = 319.46 K. Micrographies taken at temperatures before and after T = 319.46
K show that at this temperature the mixture clearly starts to melt, delimiting the biphasic
region. In fact, the well-described triangle-shape of the Tammann plot of the invariant
transition, considering the linear regressions of the data (R2 > 0.98), was able to show the
typical profile of a eutectic transition. It means that the enthalpy of the transition increases
up to the eutectic point (x1 = 0.124 0.005 mole fraction) when it begins to decrease.
162
Figure 5. Experimental ( by DSC and by microscopy) and modeled (solid lines) phase diagrams
using the 3-suffix Margules equation for the triacylglycerol + fatty alcohols mixtures. AijS, Bij
S are the
parameters for the iS equation and Aij
L, Bij
L for the i
L equation (kJmol
-1). A) trilaurin (1) + 1-
hexadecanol (2), B) trilaurin (1) + 1-octadecanol (2), C) trimyristin (1) + 1-hexadecanol (2), D)
trimyristin (1) + 1-octadecanol (2), E) tripalmitin (1) + 1-hexadecanol (2), F) tripalmitin (1) + 1-
octadecanol (2).
312
316
320
324
328
332
0.0 0.2 0.4 0.6 0.8 1.0
T /
K
x, z trilaurin
312
316
320
324
328
332
0.0 0.2 0.4 0.6 0.8 1.0
T /
K
x, z trilaurin
(A) (B)
315
319
323
327
331
335
0.0 0.2 0.4 0.6 0.8 1.0
T /
K
x, z trimyristin
315
319
323
327
331
335
0.0 0.2 0.4 0.6 0.8 1.0
T /
K
x, z trimyristin
315
320
325
330
335
340
0.0 0.2 0.4 0.6 0.8 1.0
T /
K
x, z tripalmitin
320
325
330
335
340
0.0 0.2 0.4 0.6 0.8 1.0
T /
K
x, z tripalmitin
(C) (D)
(E) (F)
(Aij, Bij) L = (10.0, -2.0)
(Aij, Bij) S = (1.6, -0.5)
(Aij, Bij) L = (2.2, 0.5)
(Aij, Bij) L = (9.0, -5.0)
(Aij, Bij) S = (-0.1, 0.1)
(Aij, Bij) L = (12.0, -3.0)
(Aij, Bij) S = (0.5, -1.5)
(Aij, Bij) L = (5.9, -4.3)
(Aij, Bij) S = (-0.5, 0.1)
(Aij, Bij) L = (5.0, -5.0)
(Aij, Bij) S = (-0.5, 0.5)
163
Figure 6. Thermograms for the mixture trimyristin (1) + 1-hexadecanol (2). Magnification of the
thermogram for x1 = 0.778 mole fraction, in detail, showing the eutectic transition.
Figure 7. Tammann plot of the enthalpy of the eutectic transition for the trimyristin (1) + 1-
hexadecanol (2) mixture. Dashed lines are linear regressions.
315
319
323
327
331
335
0.0 0.2 0.4 0.6 0.8 1.0
T /
K
x, z trimyristin
H = 93.042x + 72.158
R² = 0.98
H = 489.09 x
R² = 0.99
0
10
20
30
40
50
60
70
0.0 0.2 0.4 0.6 0.8 1.0
eu
tH/ k
Jm
ol-1
xtrimyristin
(A) (B)
164
Figure 8. Melting transition at the eutectic temperature (Teut = 319.46 K) for the trimyristin (1) +1-
hexadecanol (2) system at x1 = 0.694 mole fraction. A) 319.15 K, B) 319.65, C) 320.15 K. White
arrows highlight details during the melting process.
Note that the enthalpy associated to the eutectic transition disappears at x1 = 0.0
mole fraction in the left-hand side of the Tammann plot and at x1 = (0.775 0.005) mole
fraction in the right-hand side. It means that, in case of the trimyristin (1) + 1-hexadecanol
(2) mixture, at the left-hand side of the phase diagram, the solid phase in the biphasic
region is composed by pure fatty alcohol or, considering the experimental uncertainty, a
solid solution with a composition very close to pure fatty alcohol. However, in the right-
hand side of the diagram, the system is a mixture with both compounds in the solid-phase
as a solid solution (x1 0.775 mole fraction). For the other mixtures, solid phase miscibility
was also observed at the triacylglycerol-side of the diagram. The Tammann plots for all
systems are in the Supporting Information. In summary, trilaurin (1) + 1-hexadecanol (2)
presents a eutectic at (x1, z1I, z1
II) (0.552, 0.000, 0.949) mole fraction, trimyristin (1) + 1-
octadecanol (2) at (x1, z1I, z1
II) (0.521, 0.000, 0.937) mole fraction, tripalmitin (1) + 1-
hexadecanol (2) at (x1, z1I, z1
II) (0.025, 0.000, 0.743) mole fraction and tripalmitin (1) + 1-
octadecanol (2) at (x1, z1I, z1
II) (0.112, 0.000, 0.682) mole fraction. Thus, at x1 > xeut, the
solid phase of the mixture is composed by both compounds at the biphasic domain. On the
other hand, at x1 < xeut experimentally is virtually impossible to know if the solid phase of
the system is really immiscible or if there is a small miscibility gap very close to pure fatty
165
alcohol, i.e. z2 1.00. Tammann plot for trilaurin (1) + 1-octadecanol (2) shows an almost
simple eutectic behavior with solid phase immiscibility through the entire concentration
range. In this case, the eutectic point is at (x1, z1I, z1
II) (0.802, 0.000, 1.000) mole fraction.
In summary, solid solutions are clearly evidenced in almost all cases and always at the
right-hand side of the diagram, indicating that they are triacylglycerol-rich solutions. Also,
the higher the difference between the pure compounds’ melting temperatures or the higher
the triacylglycerol carbon chain, the larger the solid solution region.
When the compounds in solid phase are miscible such that a solid solution is
formed, the transition for the description of the solidus line, i.e when the first crystal melts,
was difficult to be observed by the DSC measurements (see highlights in Figure 6). Thus,
in order to verify firstly the previous supposition on the existence of solid solution and to
indentify the temperature of the solidus line, the samples were also submitted to thermal-
controlled microscopy at this defined region. Figure 9 shows the melting process of the
trimyristin (1) + 1-hexadecanol (2) system at x1 = 0.894 mole fraction beginning from a
temperature T < Teut. White arrows were used to highlight the appearance of the liquid
phase at T = 323.55 K (Figure 9 D). A slight enlargement of the crystal is observed with the
rounding of the angulated boundaries of the solid particle. Consequently, the amount of the
liquid phase starts to increase with the entrance in the biphasic region. This means that the
melting starts at a temperature higher than that previously evaluated as being the eutectic
transition (T = 319.46 K). Thus, at a temperature lower than T = 323.55 K there is only a
solid solution.
166
Figure 9. Heating and melting processes of trimyristin (1) + 1-hexadecanol (2) system at x1 = 0.894
mole fraction at T = A) 313.15 K, B) 322.15 K, C) 323.35 K, D) 323.55 K, E) 323.95 K, F) 324.35
K, G) 327.65 K and H) 329.05 K. White arrows highlight details during the heating process.
As previously mentioned, fatty compounds generally present some thermal
transitions in the solid phase and this is the case of both fatty alcohols and triacylglycerols.
This phenomenon is very known by literature and the compounds evaluated in this work
have already been reported by presenting polymorphic forms during the heating process of
the solid phase2, 32, 35, 44, 45
. All of the triacylglycerols used in this work, trilaurin, trimyristin
and tripalmitin present three principal polymorphic forms, namely , ’ and , with
different crystal packing, respectively hexagonal, orthorhombic and triclinic and with
increasing thermal stability and transition properties, temperatures and enthalpies. Apart
from these principal polymorphic forms, some sub modifications in the crystal structure
167
could happen during the heating process, implying in additional rearrangements of the
long-carbon chains. Thus, a single thermogram may present several solid-solid transitions,
related to these numerous conformational rearrangements of the compound’s crystal lattice.
Fatty alcohols, specifically 1-hexadecanol and 1-octadecanol, investigated in this work,
also present one characteristic polymorphic form whose transition temperature is very close
to the melting temperature35, 36
. All these transitions are clearly evident in the thermograms
of the pure compounds but are also present in case of mixtures. The pure components
melting properties and the solid-solid transitions are presented in Supporting Information.
Micrographs before and after the solid-solid transition evidenced at T = 318.75 K for the
system tripalmitin (1) + 1-octadecanol (2) at x1 = 0.697 mole fraction are shown in Figure
10. Changes in the crystalline structure of the mixture and so in the refraction properties of
the solid particle is clearly observed. This transition is probably related to one of the
polymorphic forms of the tripalmitin since, at the same temperature, the pure compound
present a solid-solid transition that, according to literature2, is related to the form (Ttr =
317.85 K). This thermal event is also identified by the thermogram of the mixture, as
shown in Figure 10. The thermogram clearly shows an endothermic transition followed by
an exothermic one. This fact configures a resolidification phenomenon probably related to a
rearrangement into the ’ polymorphic form. The identification of the pure compounds’
solid-solid transitions is important because for transition temperatures higher than the
eutectic point, the solid-liquid equilibrium behavior of the mixture is influenced by these
transitions. This is described by the Equation 1. Details about the SLE equation are
presented in Supporting Information.
168
Figure 10. Micrographies and thermogram of the system tripalmitin (1) + 1-octadecanol (2) at x1 =
0.697 mole fraction. Pictures are at a temperature before (A. 293.15 K) and after (B. 320.15 K) the
solid-solid transition at 318.75 K. Black arrows in the thermogram highlight the transitions. Eutectic
transition in detail (T = 329.20 K).
3.2. SLE modeling using the Crystal-T algorithm
Taking into account the comprehension of the experimental melting behavior of the
systems, the modeling of the SLE phase diagrams was carried out. Firstly, some remarks
can be done considering the performance of the procedure. At the first level routine of the
nested structure of the modeling procedure (Figure 4) the Crystal-T algorithm was executed
and the equilibrium point (T, xi, zi) of a mixture for a given liquid phase composition was
calculated. Since the given liquid phase composition xi lay on the liquidus line the
equilibrium point could be directly determined and no additional routine to test the number
of phases was required. On the other hand, by using the Gibbs energy minimization, and the
isoenthalpic flash algorithms, the calculation of the SLE was very sensitive to the initial
100 mm 100 mm
BA
307.0 313.5 320.0 326.5 333.0 339.5 346.0
Hea
t fl
ow
T / K
327 331
169
estimate. If the composition was inside the biphasic region, the mixture is separated in two
phases and the answer is the phase change point (T, xi, zi). However, out of this domain, the
mixture has only one phase and the liquidus and solidus lines are not described, requiring a
routine to test the number of phases. Considering that the description of the entire SLE
phase diagram implies in the evaluation of a relatively large set of points that would be
additionally followed by an adjustment procedure for the i models’ parameters, the
numerical efforts increases exponentially. Apart from discussions on the performance of the
procedures, the results obtained by using the Crystal-T, the Gibbs or flash algorithms were
(and should be) exactly the same because they use the same thermodynamic theory
expressed by Equation 1.
At the second level of the modeling, the Crystal-T algorithm was executed
considering each one of the pure compounds as initial estimate and generating two profiles,
being the interception of both profiles the solid-solid-liquid equilibrium point (Figure 2).
This procedure avoided additional phase test routines and problems with the discontinuity
of the profile, as already mentioned.
At the third level of the procedure the parameters of the liquid and solid i equations
were estimated. The results obtained were here exemplified through the system tripalmitin
(1) + 1-octadecanol (2) because this case presented the larger solid solution region. For the
estimation procedure, this work considered that both liquid and solid phases are non-ideal
and thus equations for depicting iL and i
S were used. Two kinds of i equations were
evaluated. The UNIFAC model20, 22
is absolutely predictive and this is a great advantage for
the design of a phase change based unitary operation if one does not know any
experimental information about the mixture. On the other hand, the Margules equation is a
170
function with adjustable parameters. Although such model is not predictive, the function
presents a lower non-linearity, reducing the numerical efforts, but presenting a good
agreement with the experimental data as it has been already reported in the literature34, 46, 47
.
Figure 11 depicts the SLE diagram of the system tripalmitin (1) + 1-octadecanol (2)
using the original or modified UNIFAC models to calculate the iL
and the 2-suffix
Margules equation to calculate the iS. Additionally, the liquid phase were also taken as
ideal (iL
= 1.0). In these cases, the optimization procedure is unidirectional, taking into
account the adjustment of the Aij parameter of 2-suffix Margules equation. These
approaches fail in the prediction of the liquidus line as well as the SSLE (eutectic)
temperature and composition. The iL values obtained from both UNIFAC models did not
deviate significantly from the ideality. However, in case of original UNIFAC model, the
solid phase positive deviation calculated by 2-suffix Margules equation was slightly
decreased when compared with the other approaches. The lower the positive deviation from
ideality, the lower the tendency to phase separation. Consequently, when liquid phase was
calculated by original UNIFAC, the 2-suffix Margules could described a solidus line with a
solid solution domain larger than that described by the approach using the modified
UNIFAC version. These observations show that probably neither the ideal assumption nor
the UNIFAC models can accurately depict the liquid phase behavior or 2-suffix Margules
equation are not able to describe the solid phase non-ideality. Furthermore, original and
modified UNIFAC sketch different SLE profiles. If original and modified UNIFAC models
are compared, different combinatorial and residual terms are observed 20, 22
. This means
that, considering the same system, both models can predict different molecular interactions
and entropic effects. In fact, in some previous works 32, 48
original UNIFAC model were
171
slightly more accurate in depicting the liquid phase behavior in the SLE of fatty systems
than the modified version, despite modifications to improve the non-ideality calculation.
Also, some works show that fatty compounds molecules can be better represented using a
different way of counting groups, such as the ester group in TAG molecules, or using new
or readjusted UNIFAC interaction parameters 49, 50
. These considerations allied to the fact
that UNIFAC’s parameters were adjusted by VLE experimental data 20
, show that the
utilization of the UNIFAC model for accurate description of SLE are probably restrict to
simple cases, such as that considering solid phase immiscibility. However, in this work, the
use of UNIFAC model in more complex approaches is additionally investigated. For this,
and based on the aforementioned comments, the original version was chosen.
Figure 11. SLE phase diagram of the tripalmitin (1) + 1-octadecanol (2) mixture using 2-suffix
Margules as iS equation and considering that i
L = 1.0 (simple solid line) (Aij
S = 6.93 kJmol
-1) or
using the original UNIFAC model (bold line) (AijS = 6.23 kJmol
-1) and the UNIFAC-Dortmund
model (dashed line) (AijS = 7.63 kJmol
-1) as i
L equation. DSC () and microscopy () data.
326
328
330
332
334
336
338
340
0 0.2 0.4 0.6 0.8 1
T /
K
x, z tripalmitin
172
Using the 2-suffix Margules equation for the calculation of both phases’ activity
coefficients or the 3-suffix Margules equation and the original UNIFAC model for the
calculation of iS and i
L, respectively, the optimization procedure is bidirectional. Owing to
the set of experimental data points with an inherent experimental uncertainty as well as the
number of parameters to fit the data, the problem could not be answered by numerical
procedures and was solved by a heuristic method. For this, tridimensional surfaces were
built relating the parameters used for the iL and i
S calculation and the deviation (Equation
14) generated by them. Figures 12 and 13 show the contour plots of the surfaces generated
by these approaches and sketch the changes in the calculated SLE phase diagrams
according to the different sets of parameters used to represent the behaviors. The phase
diagram with the lowest deviation in each case is represented in the bottom right of the
figure.
Both approaches improved the accuracy of the theoretical liquidus line and,
consequently, the description of the solidus line when compared to the previous one-
variable problem. The minimum deviations observed (Equation 14) were described by
valleys and were not single points. However, the valleys were as small as the accuracy of
the model to represent the activity coefficients or the quality and quantity of experimental
data increased. In the first case, sketched in Figure 12, an adjustable equation was
introduced for the calculation of both phases’ activity coefficients. In the second case,
sketched in Figure 13, an equation with two parameters was used for calculating solid
phase activity coefficients but keeping the predictive character of the liquid phase activity
coefficients. In general, the second approach, i.e using 3-suffix Margules as iS equation
and original UNIFAC as iL equation, described more precisely the phase diagrams. The
173
mean deviation decreased from = (18.10 to 3.0), approximately, especially due to a better
characterization of the solidus line. In fact, the calculation of the non-ideality of the
compounds by using 2-suffix Margules equation is such that the behavior is strictly
symmetrical reflecting the specific feature of this model (Equation 2). Symmetrical activity
coefficient values say that the compounds have similar mutual interactions. However, such
assumption is not probably the best in present cases. In fact, by using the non-symmetrical
3-suffix Margules equation the activity coefficients values assume very distinguished
behaviors.
Figure 12. Deviations (Equation 14) from experimental data for the system tripalmitin +
octadecanol using the 2-suffix Margules equation as iL and i
S models. Aij
S and Aij
L are the i
S and
iL parameters (kJmol
-1), respectively. In detail, phase diagrams using different parameters sets.
328
332
336
340
0 0.5 1T
/ K
x1, z1
328
332
336
340
0 0.5 1
T/
K
x1, z1
7.2 7.6 8.0 8.4 8.8 9.2 9.6 10.00.2
0.4
0.6
0.8
1.0
1.2
1.4
B i
j (k
J.m
ol-1
)
A ij (kJ.mol
-1)
18.5
20.0
22.0
24.0
28.0
30.0
32.0
328
332
336
340
0 0.5 1
T/
K
x1, z1
328
332
336
340
0 0.5 1
T/
K
x1, z1
328
332
336
340
0 0.5 1
T/
K
x1, z1
AijS= 12.0
AijL = 4.0
AijS = 12.0
AijL = - 1.0
AijS = 7.9
AijL = 0.7
AijS = 4.0
AijL = 4.0
AijS = 4.0
AijL = - 1.0
AijS (kJ mol-1)
Aij
L (k
J m
ol-1
)
= 18.1= 100.0
= 503.6
= 78.0
= 27.62
174
Figure 13. Deviations (Equation 14) from experimental data for the system tripalmitin + 1-
octadecanol using original UNIFAC as iL model and 3-suffix Margules equation as i
S model with
AijS and Bij
S as parameters. In detail, phase diagrams using different parameters sets.
Figure 14 sketches the results obtained by applying the 3-suffix Margules equation
for the description of both liquid and solid phases non-ideality. The optimization method
considered so a four-directional search. In this case, similarly to the previous cases, a cross-
evaluation of two matrices AijS Bij
S (parameters for i
S equation) and Aij
L Bij
L
(parameters for iL equation) were performed and the region with the lowest deviations was
depicted. The first contour plot AijS Bij
S presents the results at a fixed set of i
L parameters
(AijL, Bij
L) and the second one Aij
L Bij
L at a fixed set of i
S parameters (Aij
S, Bij
S). The fixed
328
332
336
340
0 0.5 1
T/
K
x1, z1
328
332
336
340
0 0.5 1
T/
K
x1, z1
328
332
336
340
0 0.5 1
T/
K
x1, z1
328
332
336
340
0 0.5 1
T/
K
x1, z1
328
332
336
340
0 0.5 1
T/
K
x1, z1
AijS = 10.0
BijS = 1.0
AijS = 10.0
BijS = -7.0
AijS = 6.5
BijS = -4.0
AijS = 4.0
BijS = 1.0
AijS = 4.0
BijS = -7.0
4.5 5.0 5.5 6.0 6.5 7.0-5.0
-4.5
-4.0
-3.5
-3.0B
ij (
kJ.
mo
l-1)
A ij (kJ.mol
-1)
3.03.13.33.54.07.010.015.020.0
AijS (kJ mol-1)
Bij
S (k
J m
ol-
1)
= 3.0
= 9.6
= 42.3 = 171.1
= 65.7
175
parameters were inside the best fitting region found. The adjustment was considerably
improved with deviations (Equation 14) lower than = 1.6. With this improvement one
could suppose that the behavior of the activity coefficients was more precisely assessed. In
the last cases, the liquid phase activity coefficients were very close to the ideality
presenting positive or negative deviations not higher than 10%. However, by this approach,
the deviation from ideality of the liquid phase presented marked negative deviations of up
to 25% and in fact, Figure 11 showed that by considering the liquid phase as ideal the
liquidus line presented a slight negative deviation. Likewise, the positive non-ideality of the
solid phase was slightly decreased when compared with the last cases. It means that the
improvement of the description of the liquid phase non-ideality, and consequently the
liquidus line, improved the non-ideal profile of the solid phase. Thus, the solidus line was
also more precisely assessed.
Figures 12 to 14 also show, in detail, the evolution of the behavior of the phase
diagrams shape by using different parameters sets. This is interesting since the parameters
of the i equations were directly related to the non-ideality of the system. In general, when
parameters of both liquid and solid phases i equations were low, the higher the trend to an
ideal behavior (i 1) and consequently the trend to show a continuous solid solution
SLE phase diagram. Increasing the values of the solid phase i equations parameters led to
the increase of the positive deviation from ideality and consequently to the solid-solid
phase separation. The behavior of the liquidus and solidus lines thus showed a
discontinuity, corresponding to the eutectic point. The higher the positive deviation in iS
values the smaller the solid solution region until the compounds were completely
immiscible in the solid phase.
176
Figure 14. Deviations (Equation 14) from experimental data for the system tripalmitin + 1-
octadecanol using 3-suffix Margules equation as both iL and i
S models. Aij
S and Bij
S are parameters
for iS
and AijL and Bij
L are parameters for i
L. In detail, phase diagrams using different parameter
sets.
As mentioned before, by using the heuristic method the optimized answer was a
region and not a single set of parameters. It means that into the minimum deviation region,
the set of parameters float around a small range of values. This means that once liquidus
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2-0.1
0.0
0.1
0.2
0.3
0.4
D i
j (k
J.m
ol-1
)
C ij (kJ.mol
-1)
1.60
1.70
1.80
1.90
2.00
3.00
326
330
334
338
342
0.0 0.2 0.4 0.6 0.8 1.0
T/ K
x1, z1
AijS, Bij
S = 5.9, -4.3 AijL, Bij
L = - 0.5, 0.1
5.8 5.9 6.0 6.1 6.2 6.3 6.4-4.8
-4.7
-4.6
-4.5
-4.4
-4.3
-4.2
B i
j (k
J.m
ol-1
)
A ij (kJ.mol
-1)
1.60
1.65
1.70
1.80
1.90
2.00
AijS, Bij
S = 5.90, -4.40AijL, Bij
L = - 0.50, 0.10
328
332
336
340
0.0 0.5 1.0
T/
K
x1, z1
AijS, Bij
S = 1.0, -1.0
AijL, Bij
L = 0.5, 0.1
AijS (kJ mol-1)
328
332
336
340
0.0 0.5 1.0
T/
K
x1, z1
AijS, Bij
S = 5.9, -4.3
AijL, Bij
L = 5.0, 5.0
328
332
336
340
0.0 0.5 1.0
T/
K
x1, z1
AijS,Bij
S = 5.9, -4.3
AijL, Bij
L = -3.0, -5.0
Bij
S(k
J m
ol-1
)
= 1.60
= 391.80
328
332
336
340
0.0 0.5 1.0
T/
K
x1, z1
AijS, Bij
S = 15.0, -2.0
AijL, Bij
L = 0.5, 0.1
= 2.60
AijL (kJ mol-1)
Bij
L(k
J m
ol-1
)
= 614.2
= 173.5
177
line is accurately predicted by a defined set of parameters, the modeling can fail in the
description of the solidus line. Likewise, when a set of parameters are used for the accurate
description of the solidus line, the behavior of the liquidus line is slightly altered. Thus, the
minimum region assumes so a rough surface-type shape which prevented the utilization of
optimization procedure based on a classical numerical method.
The modeling results for all binary mixtures experimentally evaluated in this work
are depicted in Figure 5 and, as indicated, 3-suffix Margules equation was considered for
the calculation of both liquid and solid phases non-ideality. Quantitatively, this approach
provided the best description of the SLE behavior, with the lowest deviations between
calculated and experimental data. In fact, when compared with the other approaches, the
increase of the accuracy was in this case obtained by means of the increase of the number
of adjustable parameters. However, this implied in the decrease of the predictive ability of
the model. Qualitatively, the utilization of the UNIFAC model as iL equation and the 3-
suffix Margules equation for the solid phase also presented reasonable results. Even with
slightly higher deviations, this approach presented a higher predictive ability and was also
able to represent the overall thermodynamic behavior of the mixtures.
4. CONCLUSIONS
The non-ideality of the solid phase could be embodied in the SLE modeling by an
effective procedure, here called as Crystal-T algorithm. In contrast with prior works in
literature that usually consider lipidic mixtures as exhibiting a solid phase with compounds
independently crystallized, triacylglycerol + fatty alcohol systems presented a
discontinuous solid solution behavior. The solid solution region could be accurately defined
178
by the Tammann plots of the eutectic transition and the results obtained by such evaluation
anchored to the modeling procedure. The algorithm was not sensitive to initial estimate and
thus, very robust and effective. This because it calculates the SLE from a given liquid phase
xi, that is a composition of the the liquidus line, from x1 = 0 to 1 and from x1 = 1 to 0.
Consequently, the algorithm avoids the problem with the discontinuity of the function in
the eutectic point, and additional routines to test the number of phases, used in Gibbs and
flash SLE algorithms. The most accurate results were obtained by using the 3-suffix
Margules equation for both liquidus and solidus lines description, due to the increasing of
the number of adjustable parameters. However, considering the future evaluation of a large
set of binary mixtures, and consequently, the construction of a parameters databank, the
SLE of more complex systems could be effectively described. Otherwise, considering a
reasonable qualitative overall description, the utilization of original UNIFAC model as iL
associated with 3-suffix Margules equation as iS
should also be applied.
This work contributed for the expansion of the boundaries of the classical SLE
modeling comprising not trivial mixtures as in case of food, oilchemistry and
pharmaceutical fields.
SUPPORTING INFORMATION
Description of the SLE equations, Tables with pure components melting properties
and mixtures’ experimental equilibrium data and Tammann plots of the eutectic transitions.
179
ACKNOWLEDGEMENTS
The authors are grateful to the brazilian funding agencies FAPESP (2008/56258-8,
2012/05027-1), CNPq (304495/2010-7, 483340/2012-0, 140718/2010-9) and CAPES
(BEX-13716/12-3) for the financial support. The authors are also grateful to Professor Dr.
Sandra Augusta Santos from IMECC/UNICAMP for the suggestions.
REFERENCES
1. R. E. Timms, Prog. Lipid. Res., 1984, 23, 1.
2. L. H. Wesdorp, Liquid-multiple solid phase equilibria in fats, theory and experiments, Doctoral
Thesis, Delft University of Technology, Delft, 1990.
3. IUPAC, Compendium of Chemical Terminology, Blackwell Scientific Publications, Oxford,
1997.
4. A. G. Marangoni and S. S. Narine, Food Res. Int., 2002, 35, 957.
5. A. G. Marangoni, N. Acevedo, F. Maleky, E. Co, F. Peyronel, G. Mazzanti, B. Quinn and D.
Pink, Soft Matter, 2012, 8, 1275.
6. F. G. Gandolfo, A. Bot and E. Flöter, Thermochim. Acta, 2003, 404, 9.
7. M. Pernetti, K. F. van Malssen, E. Flöter and A. Bot, Curr. Opin. Colloid In., 2007, 12, 221.
8. A. Bot, Y. S. J. Veldhuizen, R. den Adel and E. C. Roijers, Food Hydrocolloids, 2009, 23,
1184.
9. J. Daniel and R. Rajasekharan, J. Am. Oil Chem. Soc., 2003, 80, 417.
10. L. Dassanayake, D. Kodali, S. Ueno and K. Sato, J. Am. Oil Chem. Soc., 2009, 86, 1163.
11. R. R. Egan, G. W. Earl and J. Ackerman, J. Am. Oil Chem. Soc., 1984, 61, 324.
12. H. M. Schaink, K. F. van Malssen, S. Morgado-Alves, D. Kalnin and E. van der Linden, Food
Res. Int., 2007, 40, 1185.
13. F.A.O., OECD-FAO Agricultural Outlook 2013, 2013, available at http://www.fao.org,
accessed in 09/01/2013.
180
14. K. W. Won, Fluid Phase Equilibr., 1993, 82, 261.
15. M. Santos, G. C. Roux and V. Gerbaud, J. Am. Oil Chem. Soc., 2011, 88, 223.
16. S. M. Walas, Phase Equilibria in Chemical Engineering, Butterworth, Boston, 1985.
17. R. C. Reid, J. M. Prausnitz and B. E. Pouling, The Properties of Gases and Liquids, McGraw-
Hill, New York, 1987.
18. J. Gmehling, B. Kolbe, M. Kleiber and J. Rarey, Chemical Thermodynamics for Process
Simulation, Wiley-VHC, Weinheim, 2012.
19. J. M. Prausnitz, R. N. Lichtenthaler and E. G. Azevedo, Molecular Thermodynamics of Fluid-
Phase Equilibria, Prentice-Hall, New Jersey, 1986.
20. A. Fredenslund, R. L. Jones and J. M. Prausnitz, AIChE J., 1975, 21, 1086.
21. J. Gmehling, T. F. Anderson and J. M. Prausnitz, Ind. Eng. Chem. Fund., 1978, 17, 269.
22. J. Gmehling, J. D. Li and M. Schiller, Ind. Eng. Chem. Res., 1993, 32, 178.
23. H. K. Hansen, P. Rasmussen, A. Fredenslund, M. Schiller and J. Gmehling, Ind. Eng. Chem.
Res., 1991, 30, 2352.
24. M. C. Costa, L. A. D. Boros, J. A. P. Coutinho, M. A. Krähenbühl and A. J. A. Meirelles,
Energ. Fuels, 2011, 25, 3244.
25. L. oros M. L. S. atista . V. Va . . Figueiredo V. F. S. Fernandes M. C. Costa M. A.
r henb hl, A. J. A. Meirelles and J. A. P. Coutinho, Energ. Fuels, 2009, 23, 4625.
26. . A. . Coutinho M. on alves, M. J. Pratas, M. L. S. Batista, V. F. S. Fernandes, J. Pauly and
J. L. Daridon, Energ. Fuels, 2010, 24, 2667.
27. A. A. Nyqvist-Mayer, A. F. Brodin and S. G. Frank, J. Pharm. Sci., 1986, 75, 365.
28. P. W. Stott, A. C. Williams and B. W. Barry, J. Control. Release, 1998, 50, 297.
29. R. Gautam and W. D. Seider, AIChE J., 1979, 25, 991.
30. H. H. Rachford and J. D. Rice, J. Petrol. Tech., 1952, 4, 327-328.
31. T. F. Anderson, D. S. Abrams and E. A. Grens, AIChE J., 1978, 24, 20.
32. G. J. Maximo, M. C. Costa and A. J. A. Meirelles, Braz. J. Chem. Eng., 2013, 30, 33.
33. G. G. Chernik, J. Colloid. Interf. Sci., 1991, 141, 400.
34. M. C. Costa, M. P. Rolemberg, L. A. D. Boros, M. A. Krähenbühl, M. G. de Oliveira and A. J.
A. Meirelles, J. Chem. Eng. Data, 2007, 52, 30.
181
35. L. Ventolà, M. Ramírez, T. Calvet, X. Solans, M. A. Cuevas-Diarte, P. Negrier, D. Mondieig, J.
C. van Miltenburg and H. A. J. Oonk, Chem. Mater., 2002, 14, 508.
36. C. Mosselman, J. Mourik and H. Dekker, J. Chem. Thermodyn., 1974, 6, 477.
37. N. D. D. Carareto, M. C. Costa and A. J. A. Meirelles, presented in VI Congresso Brasileiro de
Termodinâmica Aplicada, Salvador, 2011.
38. G. Charbonnet and W. S. Singleton, J. Am. Oil Chem. Soc., 1947, 24, 140.
39. D. G. Kolp and E. S. Lutton, J. Am. Chem. Soc., 1951, 73, 5593.
40. U. Domanska and J. A. Gonzalez, Fluid Phase Equilibr., 1996, 123, 167.
41. E. S. Domalski and E. D. Hearing, J. Phys. Chem. Ref. Data, 1996, 25, 1.
42. J. Xing, Z. C. Tan, Q. Shi, B. Tong, S. X. Wang and Y. S. Li, J. Therm. Anal. Calorim., 2008,
92, 375.
43. G. Nichols, S. Kweskin, M. Frericks, S. Reiter, G. Wang, J. Orf, B. Carvallo, D. Hillesheim and
J. Chickos, J. Chem. Eng. Data, 2006, 51, 475.
44. N. Garti, J. Schlichter and S. Sarig, Lipid Fett, 1988, 90, 295.
45. M. Kellens, W. Meeussen and H. Reynaers, Chem. Phys. Lipids, 1990, 55, 163.
46. M. C. Costa, L. A. D. Boros, J. A. Souza, M. P. Rolemberg, M. A. Krähenbühl and A. J. A.
Meirelles, J. Chem. Eng. Data, 2011, 56, 3277.
47. M. C. Costa, L. A. D. Boros, M. P. Rolemberg, M. A. Krähenbühl and A. J. A. Meirelles, J.
Chem. Eng. Data, 2010, 55, 974.
48. G. J. Maximo, N. D. D. Carareto, M. C. Costa, A. O. dos Santos, L. P. Cardoso, M. A.
Krähenbühl and A. J. A. Meirelles, Fluid Phase Equilibr., 2014, 366, 88.
49. G. F. Hirata, C. R. A. Abreu, L. C. B. A. Bessa, M. C. Ferreira, E. A. C. Batista and A. J. A.
Meirelles, Fluid Phase Equilibr., 2013, 360, 379.
50. P. C. Belting, J. Rarey, J. Gmehling, R. Ceriani, O. Chiavone-Filho and A. J. A. Meirelles,
Fluid Phase Equilibr., 2014, 361, 215.
182
183
SUPPORTING INFORMATION
Deduction of the SLE equations considering polymorphism phenomena
The isofugacity criteria establish that, in case of solid-liquid equilibrium (SLE), the
fugacity of the compound fi in both solid (S) and liquid (L) phases are equal.
S L
i if f S1
Equation S1 can be written in terms of the standard fugacity of the compounds fi0 in
each phase as follows:
0
0 0
0
L SL L S i i i
i i i i i i Li i i
S
S
x fx f z f
z f
S2
where xi and zi are mole fractions of the compound i in both liquid and solid phases,
respectively, and iL and i
S are the activity coefficients of both liquid and solid phases,
respectively. It means that the temperature T of the system must be the same in both solid
and liquid phases. This is attended considering a subcooled liquid state. The L S
i i i ix z
ratio can be calculated by the variation of the Gibbs energy G as follows,
0
0
S
i
L
i
R lnf
G Tf
S3
of the following process: heating the system from T to the melting temperature Tfus,
followed by the complete melting and then cooling to the temperature T. The G of such a
process is calculated by Equation S4, knowing the variation of the enthalpy H (Equation
S5) and entropy S (Equation S6) of the process, and considering the n solid-solid
transitions.
184
G H T S S4
tr fus
fus tr
n nS S L L
i tr i fus i i
tr=1 tr=1
d d + d d
T T T T
T T T T
H Cp T H Cp T H Cp T Cp T
S5
tr fus
fus tr
S S L Ln ntr fusi i i i
tr=1 tr=1tr fus
d d d dT
T T T T
T T T T
H HCp Cp Cp CpS T T T T
T T T T T
S6
where fusH and Tfus are the melting enthalpy and temperature of the compound i, and trH
and Ttr are the solid-solid transitions enthalpy and temperature of the compound i. Knowing
that the behavior of the heat capacity of the compound i in both liquid CpiL
and solid CpiS
phases as a function of the temperature is small1-3
(in the range evaluated), CpiL
and Cpi
S
can be considered as constants as well as the variation of the heat capacity fusCp = CpL
CpS. Thus, the SLE can be determined by Equation S7 from Ttr < T < Tfus, by Equation S8
for T < Ttr , and analogously, in case of more than one solid-solid transition.
Lfusi i
fusi i
fus fus fus1 1ln 1ln
S
pC T T
T
H
R T
x
T T Rz
S7
Lfus tri i
f
fus fus f
i
us
us tri
1 1 1 1ln 1ln
S
pH Hx
R T T R T
C T T
Tz TR T
S8
Table S1. Melting temperatures (Tfus / K) and enthalpies (fusH / kJmol-1
) and solid-solid
temperatures (Ttr / K) and enthalpies (trH / kJmol-1
) of the pure compounds used in the
modeling procedure and obtained in this work or from literature.
Compound Tfus ref. fusH ref. Ttr1 ref. tr1H lit. Ttr2 ref. tr2H ref.
Trilaurin 319.39
this work
116.45
this work
308.25 4 86.00
4 288.75
4 69.80
4
Trimyristin 330.90 141.97 319.05 4 106.00
4 305.75
4 81.90
4
Tripalmitin 338.98 165.88 328.85 4 126.50
4 317.85
4 95.80
4
1-hexadecanol 322.61 33.60 5 321.89
this work 23.70
5 - -
1-octadecanol 331.42 40.10 6 330.55 26.50
6 - -
185
Table S2. Experimental solid-liquid equilibrium data for the triacylglycerols + fatty alcohols
systems for mole fraction x and temperature T for solidus and liquidus line.†
trilaurin (1) +
1-hexadecanol (2)
trilaurin (1) +
1-octadecanol (2)
trimyristin (1) +
1-hexadecanol (2)
x1 Solidus
/ K
Liquidus
/ K
x1 Solidus
/ K
Liquidus
/ K
x1 Solidus
/ K
Liquidus
/ K
0.000 322.61 0.000 331.42 0.000 322.61
0.097 315.39 321.13 0.101 317.31 330.00 0.042 320.35 321.68
0.197 315.44 319.48 0.197 317.42 328.56 0.100 319.98
0.295 315.55 318.38 0.294 317.50 327.25 0.137 320.82
0.402 315.96 317.30 0.395 317.67 325.84 0.192 320.11 322.05
0.503 315.86 0.501 317.56 323.71 0.246 319.89 322.98
0.592 315.85 316.28 0.597 318.01 321.73 0.394 319.11 324.71
0.703 315.18 316.79 0.701 317.76 319.85 0.498 319.30 326.30
0.781 315.06 317.15 0.790 317.76 318.67 0.594 318.99 326.78
0.863 315.00 318.58 0.884 317.70 318.77 0.694 318.80 327.59
1.000 319.39 1.000 319.39 0.778 319.10 328.24
0.894 323.55 ‡ 329.24
1.000 330.90
trimyristin (1) +
1-hexadecanol (2)
tripalmitin (1) +
1-hexadecanol (2)
tripalmitin (1) +
1-octadecanol (2)
x1 Solidus
/ K
Liquidus
/ K
x1 Solidus
/ K
Liquidus
/ K
x1 Solidus
/ K
Liquidus
/ K
0.000 331.42 0.000 322.61 0.000 331.42
0.100 325.99 329.88 0.050 322.60 0.058 329.20 330.27
0.199 326.26 328.60 0.101 322.90 328.69 0.099 328.90 329.80
0.300 326.01 327.28 0.201 321.90 331.98 0.151 329.56
0.393 325.90 0.299 321.93 333.62 0.200 328.95 331.40
0.493 325.88 0.408 321.68 334.75 0.300 328.80 333.40
0.592 325.77 326.73 0.501 321.90 335.93 0.401 329.50 334.79
0.691 325.65 327.67 0.597 322.00 336.61 0.502 328.80 335.63
0.793 325.48 328.90 0.701 321.82 337.35 0.595 329.00 336.63
0.878 325.61 329.52 0.800 330.45 ‡ 338.32 0.697 329.20 337.35
1.000 330.90 0.894 336.05 ‡ 338.85 0.797 332.05 ‡ 338.16
1.000 338.98 0.906 335.85 ‡ 338.82
1.000 338.98
† Uncertainties for molar fraction and liquidus line temperature are 0.001 and 0.38 K.
Uncertainty for solidus line temperature measured by DSC and by microscopy ‡ is 0.58 K.
186
Figure S1. Tammann plots of the experimental eutectic transition enthalpy eutH () and linear
regression (dashed lines) for the systems A) trilaurin (1) + 1-hexadecanol (2), B) trilaurin (1) + 1-
octadecanol (2), C) trimyristin (1) + 1-hexadecanol (2), D) trimyristin (1) + 1-octadecanol (2), E)
tripalmitin (1) + 1-hexadecanol (2), F) tripalmitin (1) + 1-octadecanol (2).
y = -93.04 x + 72.16
R² = 0.981
y = 506.01 x - 1.40
R² = 0.993
0
10
20
30
40
50
60
70
0.0 0.2 0.4 0.6 0.8 1.0
eu
tH/ k
Jm
ol-1
xtrimyristin
y = 255.42 x - 0.45
R² = 0.9998
y = -319.09 x + 298.97R² = 0.994
0
20
40
60
80
100
120
140
0.0 0.2 0.4 0.6 0.8 1.0
eu
tH /
kJm
ol-1
xtrimyristin
y = 140.53 x - 0.11R² = 0.996
y = -566.99x + 566.99R² = 0.999
0
20
40
60
80
100
120
0.0 0.2 0.4 0.6 0.8 1.0
eu
tH/ k
Jm
ol-1
xtrilaurin
(B)
(D)
y = 194.20 x - 1.02
R² = 0.998
y = -267.63x + 253.93
R² = 0.987
0
20
40
60
80
100
120
0.0 0.2 0.4 0.6 0.8 1.0
eu
tH/ k
Jm
ol-1
xtrilaurin
y = -36.42 x + 27.17R² = 0.990
0
10
20
30
0.0 0.2 0.4 0.6 0.8 1.0
eu
tH/ k
Jm
ol-1
xtripalmitin
Supposed plot
y = -92.08 x + 62.807R² = 0.985
y = 510.34x - 3.749R² = 0.999
0
10
20
30
40
50
60
0.0 0.2 0.4 0.6 0.8 1.0
eu
tH/ k
Jm
ol-1
xtripalmitin
(A)
(E) (F)
(C)
187
REFERENCES
1. G. Charbonnet and W. S. Singleton, J. Am. Oil Chem. Soc., 1947, 24, 140.
2. R. Ceriani, R. Gani and A. J. A. Meirelles, Fluid Phase Equilibr., 2009, 283, 49.
3. J. Xing, Z. C. Tan, Q. Shi, B. Tong, S. X. Wang and Y. S. Li, J. Therm. Anal. Calorim., 2008,
92, 375.
4. L. H. Wesdorp, Liquid-multiple solid phase equilibria in fats, theory and experiments, Doctoral
Thesis, Delft University of Technology, Delft, 1990.
5. C. Mosselman, J. Mourik and H. Dekker, J. Chem. Thermodyn., 1974, 6, 477.
6. L. Ventolà, M. Ramírez, T. Calvet, X. Solans, M. A. Cuevas-Diarte, P. Negrier, D. Mondieig, J.
C. van Miltenburg and H. A. J. Oonk, Chem. Mater., 2002, 14, 508.
188
189
CONCLUSIONS
Concerning the use of natural renewable resources and the valuation of
biocompounds extracted from vegetable oils, this work was addressed at the study of the
phase transition behavior of several lipidic mixtures, especially those related to melting
phenomena as a tool for the design of products and processes.
Among the binary systems composed of triacylglycerols, fatty acids and fatty
alcohols, whose solid-liquid equilibrium (SLE) phase diagrams were experimentally
characterized in this work by differential scanning calorimetry, the formation of eutectic
mixtures with partial miscibility in the solid phase was observed. It means that solid
solutions can be formed during the crystallization process of these mixtures. The formation
of solid solutions could be evaluated through X-Ray diffraction and optical microscopy.
The concentration of the mixture at which solid solution is obtained could be calculated
through the Tammann plot of the eutectic reaction. Also, in some systems, compounds
presented both positive and negative deviations from ideality in the liquid phase depending
on the compound in excess. Due to this behavior, the Margules equation with two
adjustable parameters should be used for the calculation of the activity coefficient. In
addition to binary mixtures composed by two lipidic components, the SLE of systems
composed by fatty acids and ethanolamines was also evaluated. These mixtures led to the
formation of ethanolamine carboxylates. These compounds are protic ionic liquids formed
during the proton transfer reaction between the reagents of the mixture. Due to the
formation of this new compound, the SLE diagrams of these mixtures presented a
congruent melting profile. Moreover, the formation of lamellar and hexagonal mesophases
190
(liquid crystals) could be identified by using light polarized optical microscopy. Such ionic
liquid crystals showed a great structuring ability and a special non-Newtonian behavior.
Based on the thermodynamic complexity observed, one may conclude that the
theoretical description of the lipidic systems’ phase diagrams could not be carried out with
the simplification of the SLE theory, which is in fact a usual procedure in literature. The
formation of solid solutions suggested that the non-ideality of both liquid and solid phases
could play an important role on the description of the phase diagram. Thus, the calculation
of the SLE diagrams of lipidic mixtures with partial miscibility in the solid phase could be
effectively carried out by the algorithm “Crystal-T”, developed in this work. The
calculation of the activity coefficient of the liquid and solid phases using the Margules
equation with 2 adjustable parameters, the consideration of the polymorphic phenomena
and the solid phase solubility information obtained by the Tammann plots were
fundamental for the accuracy of the modeling. The solid phase of the mixtures could be
quantified and characterized by the description of the solidus line of the phase diagram.
Consequently, the description of the liquidus line was also improved.
From the experimental point of view, the results suggested that, for future works,
the determination of a larger number of fatty compounds’ physicochemical properties, such
as melting and solid phase transitions temperatures, enthalpies and heat capacities is
fundamental, since they can play a significant role for the description of the SLE phase
diagrams. Likewise, the results highlighted that, future works on the determination of the
SLE of binary and multicomponent fatty systems of industrial and academic interest are
still required in order to improve the understanding of the thermodynamic behavior of such
mixtures. From the theoretical point of view, this work converges to the improvement of
the approaches for the calculation the SLE diagrams of fatty systems presenting more
191
complex thermodynamic behaviors. Thus, it is necessary to develop works related to the
design of algorithms that takes into account mixtures presenting congruent melting profiles,
peritectic reactions, compounds with strong deviation from ideality, unusual non-ideal
behaviors, and the utilization of different activity coefficient models for both liquid and
solid phases.
The diversity of the lipidic mixtures’ solid-liquid equilibrium behaviors observed in
this work through the formation of eutectic mixtures with miscible or immiscible solid
phases, congruent melting profiles or liquid crystal mesophases is the key for the
development of several applications for the food, pharmaceutical and material industry. The
SLE phase diagram can be a useful tool for the formulation of lipidic-based products such
as cosmetic creams, pharmaceuticals and biofuels, for the manipulation of specific
physicochemical properties and for the development of industrial processes such as
crystallization, purification, flowing, and energy generation. In general, the use of fatty
compounds extracted or produced from vegetable oils, in the context of the “green
chemistry”, can promote the decrease of environmental and health negative impacts.
192
193
CONCLUSÕES
Considerando o reaproveitamento de recursos renováveis naturais e o aumento do
valor agregado dos biocompostos extraídos de óleos vegetais, este trabalho contribuiu para
o estudo experimental e teórico do comportamento das transições de fases de diversas
misturas lipídicas, em especial aquelas relacionadas aos fenômenos de fusão, como
ferramenta para o desenvolvimento de produtos e processos.
Entre os sistemas binários formados por triacilgliceróis, ácidos graxos e alcoóis
graxos, cujos digramas de fases do equilíbrio sólid-líquido (ESL) foram caracterizados
experimentalmente neste trabalho por calorimetria diferencial exploratória, foi observada a
formação de misturas eutéticas com miscibilidade parcial na fase sólida. Ou seja, soluções
sólidas podem ser formadas durante o processo de cristalização dessas misturas. A
formação de soluções sólidas pôde ser avaliada através de difração de raio-X e microscopia
ótica. A concentração da mistura em que a solução sólida foi formada pôde ser obtida a
partir do diagrama de Tammann da reação eutética. Além disso, em alguns sistemas, os
compostos apresentaram desvios da idealidade positivos e negativos na fase líquida,
dependendo do composto em excesso. Devido a este comportamento, a equação de
Margules com 2 parâmetros ajustáveis foi utilizada para o cálculo do coeficiente de
atividade. Além de misturas binárias formadas por dois compostos lipídicos, sistemas
formados por ácidos graxos e etanolaminas foram também avaliados. Essas misturas
levaram a formação de carboxilatos de etanolamina. Esses compostos são líquidos iônicos
próticos formados durante a reação de transferência de prótons entre os reagentes da
mistura. Pela formação desses novos compostos, os diagramas de ESL dessas misturas
apresentaram um perfil de fusão congruente. Além disso, a formação de mesofases
194
lamelares e hexagonais (cristais líquidos) pode ser identificada através de microscopia de
luz polarizada. Tais líquidos iônicos cristalinos mostraram grande habilidade estruturante e
comportamento não-Newtoniano singular.
Com base na complexidade termodinâmica observada, conclui-se que a descrição
teórica dos diagramas de fase dos sistemas lipídicos não pode ser realizada através da
simplificação da teoria do ESL, o que é um procedimento usual encontrado na literatura. A
formação de soluções sólidas sugeriu que a não-idealidade de ambas as fases líquidas e
sólidas teve um papel importante na descrição do diagrama de fases. Portanto, o cálculo dos
diagramas de ESL de misturas graxas com miscibilidade parcial na fase sólida pôde ser
efetivamente realizado através do algoritmo “Crystal-T” desenvolvido neste trabalho. O
cálculo do coeficiente de atividade das fases líquidas e sólidas utilizando a equação de
Margules com 2 parâmetros ajustáveis, a consideração dos fenômenos polimórficos e das
informações experimentais sobre a solubilidade da fase sólida obtidas pelo diagrama de
Tammann da reação eutética foram fundamentais para a precisão da modelagem. A fase
sólida das misturas puderam ser quantificadas e caracterizadas a partir da descrição da linha
solidus do diagrama de fases. Consequentemente, a descrição da linha liquidus foi, da
mesma forma, melhorada.
Do ponto de vista experimental, os resultados apresentados sugerem que, como
trabalho futuro, é fundamental a determinação das propriedades físico-químicas
(temperatura, entalpia e capacidade calorífica da fusão dos compostos puros e das
transições da fase sólida) de um maior número de compostos graxos, como os
triacilgliceróis, acilgliceróis parciais, alcoóis, ácidos e ésteres graxos. Essas propriedades
podem influenciar na descrição dos diagramas de fases sólido-líquido. Os resultados ainda
destacam que, trabalhos futuros envolvidos na determinação do ESL de sistemas lipídicos
195
binários e multicomponentes de interesse industrial ou acadêmico são necessários para
ampliar a compreensão do comportamento termodinâmico dessas misturas. Do ponto de
vista teórico, os resultados convergem para o aperfeiçoamento das abordagens para o
cálculo dos diagramas de ESL de sistemas graxos que possuam comportamentos
termodinâmicos mais complexos. Para isso, é necessária o desenvolvimento de trabalhos
relacionados à elaboração de algoritmos que considerem casos de misturas que apresentam
fusões congruentes, reações peritéticas, compostos com fortes desvios da idealidade,
comportamentos não-ideais incomuns, e a utilização outros modelos para o cálculo do
coeficiente de atividade das fases.
A diversidade do comportamento termodinâmico de sistemas binários de compostos
lipídicos observada neste trabalho através da formação de misturas eutéticas com fases
sólidas miscíveis ou imiscíveis, fusão congruente ou formação de líquidos cristalinos é a
chave para o desenvolvimento de novas aplicações para a indústria de alimentos,
farmacêutica e de materiais. O ESL pode ser ferramenta útil na formulação de produtos
com base lipídica, como cremes cosméticos, fármacos e biocombustíveis, na manipulação
de propriedades físico-químicas e no desenvolvimento ou otimização de processos
industriais como a cristalização, purificação de compostos, escoamento ou geração de
energia. De modo geral, a utilização de compostos graxos extraídos ou produzidos a partir
de óleos vegetais, no contexto da “química verde”, pode promover a diminuição de
impactos negativos no ambiente e na saúde humana.