Journal of Thermal Analysis and Calorimetry

, Volume 110, Issue 3, pp 1353–1365

Quality of Brazilian vegetable oils evaluated by (modulated) differential scanning calorimetry

Authors

    • Institute for Forest UtilizationAlbert-Ludwigs-University Freiburg
  • Gustaaf Schoukens
    • Department of TextilesGhent University
  • Leo Vonck
    • Topchim N.V
  • Dirk Stanssens
    • Topchim N.V
  • Henk Van den Abbeele
    • Topchim N.V
Article

DOI: 10.1007/s10973-011-2132-2

Cite this article as:
Samyn, P., Schoukens, G., Vonck, L. et al. J Therm Anal Calorim (2012) 110: 1353. doi:10.1007/s10973-011-2132-2

Abstract

Vegetable oils are increasingly replacing fossil-oil-based polymers, and therefore aimed at being used in polymerization reactions from −20 to 100 °C. Therefore, phase transitions and heat capacities in this temperature range should be well characterized to optimize processing conditions and energy inputs. By using the DSC analysis, only small primary correspondence or divergence between different oil types are seen as a function of their degree of unsaturation, but it does not clearly distinguish detailed features such as shoulder bands related to the separate melting processes of single fatty acid components. By using modulated DSC analysis, the combined analysis of reversing and non-reversing heat signals provides better results. The latter confirms that the melting is not a physical one-step process, but equilibrates between phase transitions and enthalpic reorganizations of the fatty acids that can be monitored separately. The specific heat capacities measured during modulated DSC are somewhat lower than traditional calorimetric measurements, but relate to the degree of unsaturation. The thermal behavior of palm-, soy-, sunflower-, corn-, castor-, and rapeseed-oil is discussed in relation to their composition, by applying a first or second heating scan.

Keywords

Thermal analysisHeat FlowHeat capacityModulated DSCVegetable oil

Introduction

Vegetable oils have become popular in several domains such as energy production [1], and materials research [2] as a sustainable replacement for depleting fossil oil resources. The thermal polymerization of vegetable oils is common for creation of thermoset-like polymers [3], or vehicles for printing inks and paints [4]. The thermal analysis of renewable resources has become a key issue in sustainable development [5]. In Brazil, huge efforts are undertaken to commercialize oil-based products from palm, soy, corn, palmkernel, sunflower, castor, coconut, and rapeseed or canola [6]. To name some examples, the thermal stability of sunflower oil from the northeast part of Brazil was studied by pyrolysis up to 900 °C [7], or in presence of antioxidants [8] for biodiesel applications. Also dynamic kinetics during thermal decomposition of Brazilian castor oil [9] or corn oil [10] diesel were studied. However, the use of Brazilian vegetable oils in material synthesis remains scarce. We are using specific vegetable oil types as additives for paper coatings in a temperature range of −20 to 100 °C [11]. Therefore, it is essential to better detail the thermal behavior of the vegetable oils in this temperature range for a rigorous control of further processing conditions in relation to physical properties.

Thermoanalytical methods such as thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) received many interest for characterization of oils and fats. Some early studies were made for rapid screening under sub-ambient conditions [12], or determination of melting points [13], crystallization and oxidation of edible oils [14] and level of saturation [15]. More recent studies by DSC were made for determination of heat capacities [16] compared to adiabatic equipment [17], determination of composition [18], solid fat contents [19], spontaneous ignition temperatures [20], melting properties before and after chemical interesterification [21], crystallization and melting behavior of specific oils [22, 23] or hydrogenated and interesterified oils [24], determination of total polar compounds [25] and blends [26, 27]. Kinetic studies demonstrated that thermal decomposition of oils occurred in three stages related to the decomposition of polyunsaturated, monounsaturated and saturated fatty acids [28]. The thermal oxidation processes were studied to find out quality deterioration in parallel with the percentage of polar compounds and viscosity during frying [29]. Indeed, the rheological parameters are indicative for the oil quality, e.g., the presence of different types of shortenings influences both the crystallization behavior and storage or shear moduli [30].

Besides conventional DSC measurements, ultra-fast scan rates were applied to study the melting of edible oils [31]: the sensitivity increased with higher scan rates, but also the peak temperatures increased and shoulder peaks were eliminated due to the polymorphic behavior of the triacylglycerols. A more general study on the effect of heating rates during comparative DSC scanning for vegetable oils illustrated that the number of endothermic transition peaks and the area under the melting curves depend on the heating rate [32]. The multiple melting behavior was explained by the melting of triacylglycerides (TAG) with different melting peaks and crystal reorganization effects. However, many of those enthalpy transitions are coupled with each other and/or strongly overlap. Better insight in these mechanisms may be obtained by using the so-called modulated differential scanning calorimetry (M-DSC). Although the number of oil species characterized by this method is limited, it was illustrated that this technique is useful in following oxidation of virgin olive oil [33]. We present here measurements for some Brazilian oil types, which can be used for a quality assessment of oils before further polymerization. The development of good qualification methods relies on a combination of different techniques. In a previous study, we performed a spectroscopic classification for these oils in combination with statistical analysis [34]. The thermal study presented here will also provide better insight in coupled and non-coupled transition phenomena of the selected vegetable oils.

Experiments

Materials

We received different refined vegetable oils from Cargill Agricola S/A (Mairinque, SP, Brazil), including palm-oil (PO), soy-oil (SO), sunflower-oil (ZO), corn-oil (MO), castor-oil (CO), rapeseed or canola oil (KO). The oils were stored in transparent plastic bottles under normal room conditions (23 °C, 50% RH, natural daylight exposure) until used for characterization and thermal analysis.

The oil composition given by manufacturer’s data is represented in Table 1. The fatty acid composition was quantified by the American Oil Chemists Society (AOCS) standard method Ce 1-62. The oils were transformed into methyl esters, then extracted and analyzed by gas chromatography (GC). The fatty acids were identified from the peak area percentage as a ratio to the total area of all methyl esters. Further characteristics of the oils in Table 2 are based on our previous work [34], by using Fourier Transform Infrared (FTIR) and Raman spectroscopy. The portions of saturated (%S), mono-unsaturated (%M) and poly-unsaturated (%P) fatty acids can be distinguished from FTIR spectra, with a proportional absorption band at 1,398 cm−1; vegetable oils generally contain considerable amounts of %M and %P. The iodine values (IV) were determined from Raman spectroscopy spectra, with a proportional absorption band at 1,660 cm−1; the values compare well against traditional titration based on the Association of Official Analytical Chemists (AOAC) Method 993.20. We preferred this method above the more accurate and sensitive NMR technique [35] because of ease, cost, and available automated procedures. Based on the spectroscopic analysis of three samples, the standard deviation on IV was ±1.2. The peroxide values were determined according to AOCS Method Cd 8b-90 and checked from analysis of the FTIR absorption band at 3,444 cm−1 [36], confirming its low values [34]. Based on the FTIR analysis of three samples, the peroxide value could be determined with a variation of 5%. The water content could also be determined according to FTIR analysis [37], and was <0.05% for all samples [34].
Table 1

Fatty acid composition of vegetable oil types, reported by manufacturers

Fatty acid

Saturated fatty acids

Mono-unsaturated fatty acids

Poly-unsaturated fatty acids

C12:0

C14:0

C16:0

C18:0

C20:0

C22:0

C16:1

C18:1

C18:1 OH

C20:1

C18:2

C18:3

 

La

M

P

S

A

B

Po

O

 

Ga

L

Ln

PO

0.3

0.7

43.0

4.5

0.3

0.3

41.0

0.1

9.5

0.4

MO

11.0

2.0

25.0

54.0

1.0

SO

10.5

4.0

0.5

0.5

22.0

54.5

7.5

KO

4.0

2.0

61.0

21.0

9.0

CO

1.0

1.0

6.0

85.0

5.0

1.0

ZO

5.0

4.0

15.0

1.0

62.0

La lauric acid, M myristic acid, P palmitic acid, S stearic acid, A arachidic acid, B behenic acid, Po palmitoleic acid, O oleic acid, Ga gadoleic acid, L linoleic acid, Ln linolenic acid

Table 2

Properties of vegetable oil types

 

Compositiona

Iodine valueb/g(I2) (100 g oil)−1

Peroxide value/meq kg−1

M/%

P/%

S/%

PO

>90

56

5.4

MO

28

58

14

112

3.0

SO

25

61

14

122

2.7

KO

61

25

14

101

1.0

CO

87

11

2

92

1.2

ZO

78

2

20

82

4.8

aCalculated from FTIR spectra

bCalculated from Raman spectra (see [34])

Thermal analysis

DSC and M-DSC were performed on a Q2000 equipment (TA Instruments V3.9A, Zellik, Belgium) under continuous nitrogen flow to exclude atmospheric oxidation and retain data only related to the intrinsic oil composition. We used hermetically sealed aluminum sample pans to avoid the evaporation of oils. Relatively small sample masses were used, as smaller sample mass gives a better peak resolution, while large sample sizes give peaks that are less resolved at fast scan rates [30]. For DSC, sample sizes of about 5.0 ± 0.1 mg were prepared and afterward, the scans were normalized for small mass variations. Two heating cycles were applied from −35 to 100 °C at a rate of 10 °C/min, with a cooling cycle in between at similar rate. An intermediate scanning rate was selected in order to optimize the sensitivity and resolution for the applied sample mass. For M-DSC, smaller sample sizes of 2.0 ± 0.1 mg were chosen to improve the sensitivity. The samples were heated over two cycles from −35 to 100 °C at a rate of 2 °C/min with a temperature modulation amplitude of ±2 °C every 60 s. The cooling rate for M-DSC was similar to the heating rate. At the outer temperatures of −35 and 100 °C, the samples were isothermally stabilized for about 5 min. The instantaneous increase in heating rate with more intense heat flow during a short time may increase the occurrence of thermal events. On the other hand, we did not apply ultra-high DSC scan rates. They usually exclude the effects of recrystallization and polymorphism in which we are interested. Instead, broad peaks often occur as two or more TAG melt simultaneously at too fast scanning speeds. Before testing, the DSC was calibrated with indium (melting point 156.6 °C, melting enthalpy ΔH = 28.45 J/g) and gallium (melting point 29.8 °C). The measurements were referred against a run with an empty hermetical aluminum pan for corrections of the baseline. The presented results are averaged from three separate scans. There was a relatively good repeatability with a variation on peak transition temperatures within ±0.5 °C. Automated functions of the TA Universal Analysis Software were applied for interpretation of the test results. Transitions are characterized by peak temperatures as defined by Deman et al. [38].

Results and discussion

Identification of vegetable oils by DSC

The heat flow curves for vegetable oils measured by DSC are presented in Fig. 1 for two subsequent heating cycles. The different melting ranges and curve shapes represent the combined effects of fatty acid composition, polymorphism and thermal history of the TAG. Therefore, oils do not show a specific melting point but they have a broad transition region at −35 to 50 °C with multiple endotherms related to separate melting events of the TAGs during heating. However, the exact analysis of thermal profiles for oils by DSC is rather complex due to the great variety and distribution in composition. During the first heating cycle, the applied heating rates largely influence the transition temperatures, peak shapes, and number of observed transition peaks. During the second heating cycle, the transitions are also influenced by the previous thermal history. The annealing during the first heating cycle and subsequent cooling may result in crystallization and reorganization of the fatty acid components, thereby influencing measurements during the second heating cycle. Melting and crystallization are two main concurring events in thermal treatment of vegetable oils, which require the intake or release of thermal enthalpy. The corresponding values for heat enthalpy at specific transition temperatures are given in Table 3. We compare the thermal behavior of oils and group them according to main characteristics, as below:
  • The PO presents endotherm transitions at 5 and 40 °C related to phase transitions of the fatty acid constituents under heating. The melting enthalpy at 5 °C is most intense with a broad peak, while the enthalpy at 40 °C is smaller and more narrow distributed. However, the physical appearance of PO suggests that the main melting process happens at 40 °C as it was the only solid oil under room conditions (23 °C, 50 RH) in agreement with the high portions long-chain saturated fatty acids. As such, parallel transitions between the different oil constituents are expected between 5 and 40 °C, leading to final melting at 40 °C. The main component at 5 °C corresponds to palm olein, while the higher melting point at 40 °C correspond to a stearin fraction [39], and/or overlap with α and β polymorphic forms [40]. Highly saturated TAG (tri-saturated SSS) obviously melt at higher temperatures than highly unsaturated ones (tri-unsaturated UUU), while the intermediate ones (mono-unsaturated SSU and mono-saturated SUU) melt in between. For PO, the fraction of tri-saturated TAGs (SSS) clearly shows up at the highest melting temperature. Moreover, the long hydrocarbon chains connected to the TAG have strong tendency for polymorphism and can be packed into different crystal structures [41], occurring by solid-state transformations before final melting. Comparing the first and second heating cycle, the shape of the melting peak at 40 °C changes more significantly than the melting endotherm at 5 °C, which indicates that it is more sensitive to the crystallization behavior during intermediate cooling. In particular, the better organized crystalline structure developing after a first heating cycle causes higher melting temperature (41.88 °C vs. 38.09 °C) and higher melting enthalpy (18.65 J/g vs. 12.05 J/g) during the second heating. Indeed, the crystalline structure of the highly saturated TAG strongly depend on the thermal history: during the first heating cycle, some of the less thermally stable TAG polymorphs melt while the remaining ones rearrange and recrystallize into more stable polymorphs that melt at higher temperatures. The crystallization during intermediate cooling is characterized by two exotherms at 17.73 and −0.87 °C; the peaks in the cooling traces are most pronounced for PO compared to the other oils (data not shown). This is obviously due to the high amount of saturated fatty acids in PO, which crystallize more effectively than unsaturated ones in other oils. The crystallization peaks represent two distinct crystallization processes of high melting TAG at 15 to 5 °C (stearin fraction) and low melting TAG at 0 to −20 °C (olein fraction) in agreement with Chiavarro et al. [42]. The crystallization temperatures are nearly similar to Chen et al. [43], which indicates almost no impurities that could stimulate the crystallization process at smaller degree of undercooling. Moreover, the intensity of the high temperature crystallization peak is highest and the low temperature peak is broader. The sum of both exotherm enthalpies for crystallization during cooling (−13.92 J/g at 17.73 °C and −34.81 J/g at −0.87 °C = −48.73 J/g) is somewhat smaller than the sum of endotherm melting enthalpies during the second heating cycle (36.54 J/g at 4.92 °C and 18.65 J/g at 38.09 °C = 55.19 J/g). It suggests that the crystals formed during cooling are not completely stable and recrystallize upon heating. Therefore, tempering of the oil during subsequently planned polymerization processes will be important. Based on the melting enthalpies, the crystallization in the second heating cycle has increased relatively to the first heating cycle (35.80 J/g at 3.99 °C and 12.05 J/g at 41.88 °C = 47.85 J/g).

  • The MO and SO have endotherm transitions at lower temperatures due to higher double bond densities and/or lower tendency for crystallization in the present temperature range. The weak crystallization was also confirmed by very small exotherm peaks at −12.85 °C (MO) or −12.82 °C (SO) during intermediate cooling. The MO and SO show very similar transitions during two subsequent heating cycles, as can be expected from their fatty acid composition containing mainly poly-unsaturated constituents (Tables 1, 2). For MO, two endotherm transitions happen at −25 and −9 °C (first heating) or −24 and −11 °C (second heating), and for SO, the endotherms appear at −25 and −8 °C (first heating) or −25 and −11 °C (second heating). Related to the fatty acid composition, the lowest endotherm represents melting of the tri-unsaturated moieties (UUU) and the upper one represents the di- or mono-unsaturated (SUU, SSU) ones. During subsequent heating, the lower melting peak remains relatively well in place and the upper one shifts to lower temperatures in the second scan, which is a characteristic for poly-unsaturated fatty acid mixtures. During intermediate cooling, the crystallization for MO happens partially at −12.85 °C with an enthalpy of −5.52 J/g, which is recovered during melting with an enthalpy of 1.16 J/g (−24.6 °C) and 3.22 J/g (−11.5 °C). The crystallization for SO is very similar at −12.82 °C with an enthalpy of −5.49 J/g, in parallel with recovery during melting with an enthalpy of 1.28 J/g (−25.1 °C) and 3.58 J/g (−11.5 °C). During both heating cycles, a very small exotherm reaction at around 20 °C for MO and around 25 °C for SO indicates reorganization or co-crystallization processes of mainly unsaturated TAG moieties (UUU). The related crystallization enthalpies are higher for MO (0.42 J/g) compared to SO (0.15 J/g). The present qualities of SO and MO suggest that they have comparable thermal stability and behavior, but SO seems to have a more stable crystalline structure that is less sensitive to recrystallization.

  • The KO and CO show also similar thermal transitions, due to the relatively high amount of monounsaturated fatty acids. For KO, two melting endotherm peaks occur at −23 and −12 °C (first heating), while the upper temperature endotherm disappears in a second heating due to very weak intermediate crystallization effects. For CO, the low temperature endotherm appears also at −23 °C (first and second heating), while the second endotherm around −7 °C is hardly present due to minor melting of crystal zones. For both oils, a strong exotherm transition appears in the first- and second heating cycle with comparable temperatures (15–20 °C) and enthalpies (Table 3), representing the significant recrystallization of the monounsaturated moieties in the liquid phase. The present qualities of KO and CO suggest that very weak crystalline structures form during intermediate cooling. Indeed, very weak crystallization effects were observed during intermediate cooling with an enthalpy of −0.40 J/g at −21 °C for KO and −0.12 J/g at −22 °C for CO.

  • The ZO has a very clear single endotherm at −4 °C with a small shoulder at around −10 °C, which remains stable during both a first and second heating step. The structure does almost show no intermediate crystallization during cooling and consequently no recrystallization during heating. The melting processes of ZO sees more like a one-step process that is finished at 5 °C in contrast to the multiple transitions of previous oils. The high content of monounsaturated fatty acids may reduce the oil reactivity. In general, the higher the degree of unsaturation in fatty acids, the lower the melting point of the triacylglycerol is in agreement with Table 3.

https://static-content.springer.com/image/art%3A10.1007%2Fs10973-011-2132-2/MediaObjects/10973_2011_2132_Fig1_HTML.gif
Fig. 1

DSC heat flow curves for different vegetable oils during (i) first heating cycle, and (ii) second heating cycle. Only for PO, the heat flow curve during cooling is represented (others did not show specific transitions)

Table 3

Reaction enthalpy for different vegetable oils according to DSC measurements in the first (i) and second (ii) heating cycle, near important transition points

Transition point

PO

CO

KO

MO

ZO

SO

Tpeak/°C

ΔH/J g−1

Tpeak/°C

ΔH/J g−1

Tpeak/°C

ΔH/J g−1

Tpeak/°C

ΔH/J g−1

Tpeak/°C

ΔH/J g−1

Tpeak/°C

ΔH/J g−1

1

 (i)

3.99

35.80

−22.80

0.66

−23.89

0.55

−25.95

1.35

−25.50

5.73

−24.83

1.36

 (ii)

4.92

36.54

−22.91

0.44

−24.05

0.73

−24.64

1.16

−26.30

6.52

−25.09

1.28

2

 (i)

27.15

2.02

−9.61

0.38

−12.36

0.34

−8.86

3.18

−3.99

70.10

−8.87

3.89

 (ii)

24.86

2.87

−8.48

0.43

−11.91

0.13

−11.51

3.22

−4.12

69.53

−11.52

3.58

3

 (i)

38.09

12.05

23.13

0.23

21.56

0.24

23.82

0.42

    

 (ii)

41.88

18.65

17.61

0.17

14.86

0.19

15.89

0.15

    

In conclusion, the DSC analysis shows small primary correspondence or divergence between different oils that can be roughly classified according to their IV (Table 2): PO with IV = 56, ZO with IV = 82, KO and CO with IV = 90–105, MO and SO with IV = 112–122. However, DSC does not clearly distinguish detailed features such as shoulder bands related to the separate melting processes of single fatty acid components. This is due to the complex overlap between different processes, including interactions and (re)organization of the different fatty acids. The combination of melting and crystallization properties in oils is affected by the TAG profiles and nature of bound fatty acid chains [19]. Therefore, the DSC measurements were only made as a first estimation of the oil quality and more sensitive analysis is needed, such as M-DSC presented below.

Identification of vegetable oils by modulated DSC

The thermal evaluation of vegetable oils by M-DSC in Fig. 2 includes (i) the total heat flow, (ii) non-reversing heat flow, and (iii) reversing heat flow curves during the first heating cycle. The maximum peak temperatures Tpeak and corresponding reaction enthalpies ΔH for transitions during first heating cycle are summarized in Table 4. The total heat flow for M-DSC during first and second heating cycles is compared in Fig. 3. Although Kanavouras et al. [33] reported small differences in subsequent heating cycles during M-DSC compared to conventional DSC, we observed some important variations that are further analyzed for each oil type below. In general, the compositional ingredients are better resolved in M-DSC than in our conventional DSC test (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs10973-011-2132-2/MediaObjects/10973_2011_2132_Fig2_HTML.gif
Fig. 2

Modulated DSC analysis of vegetable oils with (i) total heat flow curve, (ii) non-reversing heat flow curve, and (iii) reversing heat flow curve during the first heating cycle

Table 4

Reaction enthalpy for different vegetable oils according to modulated DSC measurements in first heating cycle, near important transition points

Transition point

PO

CO

KO

MO

ZO

SO

Tpeak/°C

ΔH/J g−1

Tpeak/°C

ΔH/J g−1

Tpeak/°C

ΔH/J g−1

Tpeak/°C

ΔH/J g−1

Tpeak/°C

ΔH/J g−1

Tpeak/°C

ΔH/J g−1

1

 (a)

−0.81

13.83

−25.77

4.11

−25.10

0.70

−27.02

0.59

−27.40

56.50

−25.71

7.83

 (b)

−0.93

7.40

−25.85

3.82

−25.18

0.77

−26.93

0.34

−27.26

60.19

−25.71

7.48

 (c)

−0.75

6.43

−27.82

0.20

−27.99

0.18

−26.61

0.24

−23.86

1.22

−27.78

0.39

2

 (a)

12.56

4.89

−21.54

0.53

−22.92

0.43

−23.05

0.96

−8.38

70.45

−21.71

1.49

 (b)

12.52

2.39

−21.54

0.56

−21.55

0.40

−23.69

0.87

−8.38

39.06

−21.76

1.55

 (c)

3.48

3.91

−24.15

0.18

−24.33

0.12

−22.92

0.10

−8.35

29.17

−24.26

0.45

3

 (a)

20.88

0.75

17.35

24.12

19.00

0.24

−11.20

4.21

  

−17.45

0.18

 (b)

21.87

0.68

17.35

24.28

−17.23

0.28

−10.67

3.37

  

−17.49

0.15

 (c)

21.94

0.08

−20.08

0.04

−13.48

0.87

  

−20.79

0.11

4

 (a)

37.30

21.06

  

−0.25

0.63

      

 (b)

37.30

18.78

  

0.36

0.50

      

 (c)

37.77

2.45

  

−0.89

0.18

      

Values are determined from (a) total heat flow, (b) non-reversing heat flow, and (c) reversing heat flow

https://static-content.springer.com/image/art%3A10.1007%2Fs10973-011-2132-2/MediaObjects/10973_2011_2132_Fig3_HTML.gif
Fig. 3

Modulated DSC total heat flow curves of vegetable oils during (i) first heating cycle, and (ii) second heating cycle for a PO, b MO, c SO, d KO, e CO and f ZO

The modulated technique may be more sensitive to the individual TAG transitions than conventional DSC, as the instantaneously superposed high heating rates allow to differentiate between phenomena related to changes in heat capacity (reversing heat flow: glass transition and melting) and interfering kinetic processes (non-reversing heat flow: crystallization, structural perfection, decomposition, and evaporation). Therefore, the total heat flow curves are split into reversible events that follow the temperature modulations and non-reversible components that are incapable to follow the temperature modulation. Most transitions in the total heat curve have components in the reversing and non-reversing heat curves, which indicate that the transitions show a combination of melting and crystallization of the respective fatty acids. The contribution of each process can be estimated from the enthalpy of both reversing and non-reversing events (Table 4). The oils with a high degree of saturation have higher melting enthalpies than those with low degree of saturation, as the former ones require more energy for melting. In most cases, the reaction enthalpy from the total heat flow equals the sum of enthalpies of reversing and non-reversing heat flows. In some instances, however, crystallization and melting can contribute to the reversing heat flow [44], as a result of rapid interfacial crystallization and melting within the modulation cycle [45]. The thermal characteristics are studied in detail for our vegetable oils, as M-DSC analysis is rarely found to precisely describe the oil quality.
  • For PO (Fig. 2a), the melting of fatty acids is best observed in the reversing heat flow, together with endothermic relaxations in the non-reversing heat flow. The reversing signal compares to a typical melting curve for PO, as measured during conventional DSC by Nassu et al. [13]; two melting endotherms of fractions at −10 and 40 °C are separated by an exothermic peak at 20 to 30 °C. These complex transitions represent different contributions of melting and crystallization phenomena. The multiple shoulder bands in the reversing heat flow show the stepwise melting of the oil structure: typically, the low temperature peaks during heating in the thermograph represent β2 and α polymorphs, while the high temperature peaks stand for \( \beta_{1}^{\prime } \) and β1 polymorphs [46]. The β2 and α polymorphs for PO occur at relatively high temperature and provide delayed melting compared to the other oils. After deconvolution, the melting endotherm in the reversing heat flow includes four overlapping peaks at −8, 0, 6, and 12 °C for the major TAGs. The melting endotherm at −8 °C (olein fraction) has a total enthalpy of ΔH = 13.83 J/g, composed of 7.40 J/g consumed by non-reversing processes (endothermic relaxations) and about 6.43 J/g by reversing processes (melting). The following melting endotherm at 13 °C (palmitic fraction) has a total enthalpy of 4.89 J/g with interferences between reversible melting (3.91 J/g) and enthalpic relaxations (2.39 J/g). However, the mean melting peak at 40 °C with a total enthalpy of 21.06 J/g has a contribution of only 2.45 J/g due to reversing melting and 18.78 J/g due to non-reversing rearrangements. These structural reorganizations in the semi-liquid phase are characteristic for PO in viscous state, and only seen in the non-reversing heat flow. As such, the final melting is not a physical one-step process but exists as equilibrium between melting and kinetic reordering of partially molten fatty acid components in combination with non-molten species. Siew et al. [47] concluded that the melting point is essentially determined by slip of the hard fraction of stearins, indicated by the major melting peak. This behavior is important to take into account for further processing of the oil. The oil is completely molten at temperatures above 46 °C when heated at 2 °C/min. The sum of melting enthalpies during modulated DSC (ΔH = 13.81 + 5.27 + 0.75 + 20.83 = 40.66 J/g) is somewhat lower than during normal DSC, which can be attributed to the different experimental conditions. In general, the total melting enthalpy for M-DSC was smaller than general values for PO (ΔH = 78 J/g). The second heating scan (Fig. 3a) has a more intense exotherm intermediate peak as recrystallization phenomena in thermally pre-treated oils become more pronounced after better organization of fatty acids along each other.

  • For MO (Fig. 2b), the principal melting process happens below −20 °C. The endotherm peaks in the non-reversing and reversing heat flow have a small shift of about 3 °C, indicating that some reordering happens before final melting. The broadening of endotherms is due to the overlap of several transitions, including solid–solid transitions between several crystalline forms at −20 to 0 °C and solid–liquid transitions of the TAG constituents and crystals at 0 to 40 °C. The latter is specified by a broad region due to the different crystal forms of unsaturated moieties. During a second heating scan (Fig. 3b), the crystal formation of thermally pretreated MO becomes more stable and shows less transitions. The primary melting point increases from −26 (first heating) to −22 °C (second heating), as expected for more stable crystals.

  • For SO (Fig. 2c), the melting happens at −27 to −20 °C and differences between reversing and non-reversing heat flow are more evident that for other oils. The melting region of SO has three endotherms that are only well-resolved in the reversing signal: the lowest peak (−27 °C) is due to highly polyunsaturated moieties (LLL), the middle one (−23 °C) is due to mono-unsaturated ones (OLL, PLL, POL/SSL) and the small endotherm at highest temperatures (−20 °C) represents a minor fraction of almost fully saturated moieties (PPP, PPO). The small exotherm peak around 0 °C in the reversing heat flow represents some constraints in TAGs. They are caused by small water uptake, hydrogenation or crystallization of the fatty acids with high density of double bonds, typically leading to broadening of the transition processes. The SO hydrolysis may results in trans-isomers and non-intersoluble TAGs that increase the melting range. These processes interfere with a kinetic reorganization represented by multiple endotherms in the non-reversing heat flow. On the other hand, also crystallization may contribute to better stability and higher melting temperatures. The differences in reversing and non-reversing heat flow are very clear for SO, likely due to high amounts of poly-unsaturated fatty acids that make the oil more reactive among others and cause both chemical constraints (released during melting) and physical constraints (released during enthalpic relaxation) around unsaturated moieties. The higher reactivity of SO among other oils was also seen from high iodine values (Table 2). During a second heating scan (Fig. 3c), the three melting sequences of SO become clearer and shift to a lower temperature. The thermal history may little degrade the crystalline quality of SO by oxidation.

  • For KO (Fig. 2d), some differences between reversing and non-reversing signals also represent non-coupled processes. The endotherms at −30 and −20 °C in the reversing heat flow correspond to the main melting reported by Kawamura et al. [23], and include several crystal forms α (−27 °C), β2 (−24 °C) and β1 (−22 °C). The intense exotherms at −22 and −18 °C in the non-reversing heat flow suggest a recrystallization process of β2 into β1 phase. Therefore, small melting peaks of residual β2 and β1 crystals are also observed at −5 and 5 °C in the reversing heat flow. The endotherm region is actually relatively broad and relates to fused crystals with lower and higher melting temperatures. Overall, the reversible and non-reversible heat signals for KO are better coupled to each other than for SO, as the KO has more saturated and mono-unsaturated fatty acids. Therefore, KO is somewhat less reactive and interactions between the fatty acid chains leading to enthalpic relaxations diminish in intensity, as they are theoretically mainly coupled to melting events rather than physical relaxations. The second heating scan (Fig. 3d) shows the same features as before, with a slight shift of melting endotherms toward a lower temperature.

  • For CO (Fig. 2e), the melting processes (reversible heat flow) and enthalpic relaxations (non-reversible heat flow) at −25 to 0 °C correspond relatively well. It is the only oil where additional relaxation is measured in the liquid state, characterized by a broad endotherm at 10 to 25 °C in non-reversible heat flow. As CO contains larger amounts of hydrogenated moieties compared to other oils (Table 1), the saturation of double bonds by hydrogenation artificially increases the melting temperature that is around −30 °C in case of non-hydrogenated CO. Therefore, the low intensity of melting endotherms at −25 °C represents a small fraction non-hydrogenated fatty acids. Otherwise, the main melting process happens in a sharp temperature zone around 0 °C because the concentration of C18:1-OH is dominant compared to the non-hydrogenated fraction. The hydrogenation can also explain different reversing and non-reversing heat flow above the melting temperature (CO is liquid at room temperature), as hydrogen bonds may imply additional molecular constraints that are gradually released at around 20 °C. During the second heating scan (Fig. 3e), the hydrogenation effects and transitions in the molten state disappear. Instead, a small exotherm at 8 °C suggests the formation of a metastable organized structure.

  • For ZO (Fig. 2f), the multiple melting peaks at around −25 °C manifest as a small endotherm in the reversing heat flow. The highly unsaturated fatty acids melt at −27 and −25 °C, comparable to SO. On the other hand, the melting process becomes complicated by significant recrystallization. This co-crystallization is typical for poly-unsaturated fatty acids including LLL, OLL, and OOL, which have highest concentrations for ZO (Table 1). The latter is evidenced by an extremely important exotherm reaction in the non-reversing heat flow around −25 °C, and manifests in a broader endotherm process corresponding to melting of the formed crystal phase at −10 to −5 °C. The exotherm process shifts to higher temperatures of −20 °C during a second heating scan (Fig. 3f), as the structure of thermally treated oils further improves by more perfect crystallisation.

In conclusion, the M-DSC provides more details on the oil quality, which may be due to the locally higher heating rates that sharpen the transition peaks and the separation in reversing and non-reversing heat flow relating to different transition processes. It provides better results than conventional DSC at high scanning rates [32], as higher scanning rates often provide clear transitions but lower resolution.

Further interpretation of modulated DSC measurements

Besides the heat flow, the specific heat capacities are more fundamental material properties and can be measured during a sinusoidal modulated M-DSC experiment in quasi-isothermal mode. The knowledge of specific heat capacities is useful to determine the energy input during heating of oils in future polymerization processes. However, it has to be taken into account that the values are only valid in a given temperature region, and it is often difficult to find information on the complete temperature range including different transitions. The values recorded during two heating cycles are given in Fig. 4. The transition phases are better illustrated in the derivative specific heat signals in Fig. 5. The values measured during the intermediate cooling cycle are plotted in Fig. 6.
https://static-content.springer.com/image/art%3A10.1007%2Fs10973-011-2132-2/MediaObjects/10973_2011_2132_Fig4_HTML.gif
Fig. 4

Specific heat capacity of vegetable oils as a function of the heating temperature determined from modulated DSC measurements during (i) the first heating cycle and (ii) the second heating cycle for a PO, b MO, c SO, d KO, e CO, and f ZO

https://static-content.springer.com/image/art%3A10.1007%2Fs10973-011-2132-2/MediaObjects/10973_2011_2132_Fig5_HTML.gif
Fig. 5

First derivative of the specific heat capacity as a function of the heating temperature determined from modulated DSC measurements during the first heating cycle

https://static-content.springer.com/image/art%3A10.1007%2Fs10973-011-2132-2/MediaObjects/10973_2011_2132_Fig6_HTML.gif
Fig. 6

Specific heat capacity of vegetable oils as a function of the heating temperature determined from modulated DSC measurements during the intermediate cooling cycle for a PO, b MO, c SO, d KO, e CO, and f ZO

By definition, the heat capacity measurements are derived from the reversible heat flow curves (Fig. 2, curve (iii)). The heat capacity values at a specific temperature of, e.g., 50 °C representative for the liquid oil state, are read from Fig. 4: in general, the specific heat increases for higher degree of unsaturation, as observed for CO (1.49 J/g °C), KO (1.58 J/g °C), MO (1.63 J/g °C), ZO (1.85 J/g), and SO (3.2 J/g °C). Indeed, the differences in heat capacity depend on the unsaturation degree [48], and can be explained by the different concentrations of monounsaturated fatty acids corresponding to conclusions from traditional calorimetric experiments [43]. Based on thermal analysis, we conclude that the SO is highly unsaturated and consequently very reactive which explains also some difference in specific heat during first and second heating cycle, which is smaller for other oils. The saturated PO presents high specific heat capacity, which is attributed to important transitions during heating (see Fig. 5) and formation of ordered polymorph structures as discussed before. The specific heat capacities determined by M-DSC are often lower than reported values from calculations, or determinations by calorimetric measurements [49, 50]. Santos et al. [16] reported values of 1.96 J/(gK) for KO, 1.98 J/(gK) for ZO, and 2.40 J/(gK) for SO. Tochitani et al. [17] reported values of 2.2 J/(gK) for PO, 1.9 J/(gK) for KO, and 2.0 J/(gK) for MO from traditional DSC at 50 °C. While the values from M-DSC and DSC agree relatively well for saturated oils such as PO, the heat-capacities for unsaturated oils measured by M-DSC are lower as the latter method is more sensitive to the oil reactivity. Moreover, the test results may depend on the used method, heating program and selected modulation parameters. However, heat capacity measurements from M-DSC are not often reported. The values give a relative impression between oils that all run under the same conditions, as such that values outside the range of transition temperatures may be considered as representative for the oil quality. In general, the specific heat capacities linearly rise with increasing temperature [51], which is theoretically explained by the heat consumption during expansion and density variations of the substance under heating [52]. The latter increase therefore mainly manifests in the liquid phase, and the less pronounced increase for ZO may be attributed to relatively high degree of crystallization and/or the formation of a liquid ordered phase intrinsic to the present oil quality that provides better stability. During the cooling cycle (Fig. 6), the changes in specific heat capacity are very clear and directly relate to the mobility of the fatty acid molecules in the different states. The transition in heat capacity is sensitive to the liquid–solid state transition upon cooling, and does not significantly relate to intermediate structural reorganization. The latter are obviously retarded during cooling from the liquid phase.

The presented enthalpy variations near endothermic transitions and the specific heat values are both important to consider during further processing of the vegetable oils. First, their values are specific for given oil compositions and provide, therefore, a good reference to evaluate the oil quality. Second, the polymerization of oils together with polymers requires a specific energy input for heating the reactor mixture to a given polymerization temperature between 20 and 100 °C [11]. Finally, the crystallization and variations in internal structure change the viscosity and rheological properties of the vegetable oils. During heating in a specific temperature range, it is clear that the structure of each vegetable oil type specifically changes by molecular reorganizations, as discussed in this article. In this respect, the M-DSC measurements especially provide more detailed information on structural evolutions of vegetable oils during heating with a distinctive “fingerprint” reference pattern at −30 to 100 °C. The highlighted differences between reversible and irreversible transitions measured in modulated DSC are new in respect to traditional DSC and allow to assessing the oil transformations and reactivity under heating more efficiently. With the presented measurements, we provided a unique and detailed thermal evaluation of vegetable oils connected to the intrinsic oil structure (not only fatty acid composition, but also higher order structures), which can be used for further optimization of polymerization processes. These reactions will be presented in a following report.

Conclusions

We characterized vegetable oils by DSC and M-DSC in order to evaluate their quality based on thermal transitions at −20 to 100 °C. Both methods offer complementary information and give fundamental understanding of the oil composition besides practical heating values for further processing.

From traditional DSC, differences in melting and crystallization are revealed during subsequent heating and intermediate cooling. Depending on the shape of endotherms and eventually following exotherms, the oils could be roughly grouped in parallel with their IV value. At low IV (PO), the long-chain saturated fatty acids expose different crystal structures that form by solid-state transformations before final melting. The crystalline structure organizes better during intermediate cooling and further improves by recrystallization in the second heating cycle. At intermediate IV (ZO), the high amount of mono-unsaturated fatty acids causes very low reactivity and almost no tendency for crystallization during intermediate cooling or heating. At some higher IV (CO, KO), only very weak crystallization happens during intermediate cooling that is compensated by very strong exotherm recrystallization effects of the mono-unsaturated moieties during heating. At highest IV (MO, SO), endotherms related to poly-unsaturated moieties are very sensitive to crystallisation at intermediate cooling and therefore slightly change the melting endotherm during a second heating scan. While a slight exotherm during heating corresponds to reorganization or co-crystallization effects, the crystalline organized structure for soy-oil seems most stable.

From M-DSC, the stepwise melting of different fatty acid components is best monitored by the reversing heat flow, while enthalpy relaxations occur in parallel in the non-reversing signal. As such, the final melting is not seen as a physical one-step process, but rather as equilibrium between melting and kinetic reordering of partially molten moieties together with non-molten species. The transitions strongly change after thermal pre-treatment during a first heating cycle. For saturated oils, the non-reversing heat flow illustrates reorganization after melting of the olein fraction and before melting of the palmitic fraction, while simultaneous structural reorganizations even become larger near the main melting endotherm. The reversing and non-reversing heat flows are relatively well coupled in presence of mono-unsaturated moieties, but deviate stronger in presence of poly-unsaturated fatty acids as an indication of structural rearrangements or co-crystallisation before melting. The presence of hydrogenated moieties is evidenced from non-reversing heat flow as reorganizations in the liquid state, and disappears after thermal pre-treatment. The heat capacities increase with heating temperatures and for oils with higher degree of unsaturation, but measurements by M-DSC are somewhat lower than traditional literature reports.

The combination of thermal characterization from DSC and MDSC provides a good reference that can be used in further quality assessment of vegetable oils, before they are submitted to, e.g., polymerization. Further processing parameters may be optimized based on these values, as crystalline structures can interfere with kinetics and the specific heat determines energy input.

Acknowledgements

P. Samyn acknowledges the Robert Bosch Foundation for support in the Junior professorship program. H. Van den Abbeele and G. Schoukens thank the Institute for the Promotion of Innovation by Science and Technology in Flanders (I.W.T.) for a funding program ‘SNAP’ (contract grant IWT-080213).

Copyright information

© Akadémiai Kiadó, Budapest, Hungary 2011