Physicochemical Exploration of Cocoa Butter During Spontaneous Fermentation: A Comparative Study Across Three Latin American Countries

Abstract

This study characterized the physicochemical properties of cocoa butter (CB) extracted from cocoa beans of the Criollo Nativo (Peru), Criollo (Mexico), and Forastero (Brazil) varieties subjected to spontaneous fermentation under traditional local conditions in each country. Cocoa samples were collected at 24-h intervals, and CB was extracted to evaluate its lipid composition through fatty acid profiling and spectroscopic techniques (FT-IR and NMR). Also, the thermal and structural properties via differential scanning calorimetry (DSC), including melting and crystallization profiles, crystallization kinetics, and polymorphism, were determined. The results revealed that stearic, oleic, and palmitic acids were predominant in all varieties, while trace levels of myristic and pentadecanoic acids contributed to molecular packing. FT-IR identified bands associated with glycerol chain formation in TAGs, which were confirmed by NMR through chemical shifts linked to the distribution of POS, SOS, and POP species. CB exhibited melting temperatures between 19.6 and 20.5 °C, favoring polymorphic transitions toward more stable forms. Form I (γ) predominated during early fermentation, while Forms II (α) and III (β′₂) were subsequently identified, particularly in Criollo varieties. These findings demonstrate that fermentation time significantly influences the chemical composition, oxidative stability, and crystalline structure of CB, providing valuable insights for optimizing cocoa processing and the development of high-quality chocolate products.

1. Introduction

Chocolate has become the world’s most desired food, with production reaching up to 7.7 million tons in recent years [1,2]. Cocoa butter (CB) constitutes the continuous phase responsible for forming a crystalline network that enables the dispersion of solid particles [3]. This network is essential for manufacturing high-quality chocolates, characterized by thermal stability, composition, rheological behavior, and organoleptic properties [4,5]. According to Golodnizky et al. [6], CB accounts for approximately 30% of chocolate and serves as a key ingredient in various confectionery products due to its distinctive features, including pale yellow color, neutral flavor, and narrow melting range [7,8,9]. For this reason, cocoa butter has become a focal point of research aimed at understanding its composition and properties, which are closely linked to quality in the food industry.

Studies report that cocoa butter (CB) contains saturated fatty acids (SFAs) ranging around 56% to 64% and unsaturated fatty acids (UFAs) ranging around 34% to 43% [10]. Among these, the three predominant FAs in CB are palmitic acid (C16:0), stearic acid (C18:0), and oleic acid (C18:1). Together with other acids such as linoleic (C18:2) and arachidic acid (C20:0), they comprise more than 98% of the total FA content in CB [11,12]. These fatty acids are mainly stored in CB as triglycerides (TAGs), which are composed of three FAs bound to a glycerol backbone via esterification [13,14]. Triglycerides are composed of 80% monounsaturated fatty acids, 3% saturated fatty acids, and 1% triple unsaturated and diunsaturated fatty acids [15]. More than 25 different TAG species have been identified in CB, with the majority being POP, SOS, and POS—primarily formed from stearic, palmitic, and oleic acids [16,17,18]—and reported as responsible for CB’s physical and chemical properties [19].

The melting point of cocoa butter (Theobroma cacao) ranges from 27 to 35 °C [12,20], and is attributed to its FA composition—primarily saturated fatty acids—and its triglyceride (TAG) content [8,21]. Studies indicate that the melting point is closely related to CB’s crystalline structure and thermal stability. Both the crystalline arrangement and thermal behavior are determined by the ability of the substance to adopt various unit cell configurations, resulting from different molecular packing modalities, which give rise to polymorphic forms [9,22]. Ghazani & Marangoni [23] note that in the late 1920s, three polymorphs were initially identified in CB—α, β′, and β—arranged in ascending order of melting point and stability. Later, in 1966, a more comprehensive classification was published, identifying six polymorphic forms (melting point): Form I or γα (~17.3 °C), Form II or α (~23.2 °C), Form III or β′1 (~25.5 °C), Form IV or β′2 (~27.5 °C), Form V or β1 (~33.8 °C), and Form VI or β2 (~36.3 °C) [23,24]. Among these, Form V (β1) is recognized as the most stable and desirable for chocolate [6,25], while the metastable Form β′ is where CB typically crystallizes [26].

The chemical composition of cocoa butter (CB) confers notable properties, including its melting profile, crystallization behavior, solid fat content, and polymorphic characteristics—all distinct to other fats [3]. These properties vary depending on factors such as geographic origin, cocoa bean maturity, fermentation conditions, and the bean’s microbiome [9,27]. Among these, two factors are particularly influential: origin (which also implies varietal differences) and type of fermentation. Regarding origin, each cocoa-producing country hosts more than one variety adapted to local growing conditions. Studies have reported variations in TAG and FA composition as well as crystallization rates in CB across different countries [1,23]. On the other hand, the effect of fermentation on CB is still under study, with conflicting findings. Some authors suggest that fermentation exerts a “minor” influence on lipid composition, with minimal variation in triglyceride and fatty acid content [5,28]. This assumption relies on two premises: (a) fermentation primarily affects hydrophilic compounds or those with hydrophilic bonds, and (b) CB contains few free unsaturated fatty acids, which reduces its susceptibility to oxidation, thus favoring its stability. However, this presumed stability has not been specifically tested under fermentation conditions [28]. Conversely, Chun & Meng [12] report that fermentation can markedly affect the physicochemical properties of CB. During this process, lipases hydrolyze TAGs, releasing FAs [29]. These FAs may undergo esterification reactions which can lead to ester formation [30], altering the lipid profile. Moreover, the acidic environment combined with elevated temperatures during fermentation promotes the hydrolysis of ester bonds within TAGs, further contributing to the release of FAs [28]. This cascade influences the CB crystallization behavior, forming metastable forms of CB (β′1 and β′2) [31].

Therefore, this study explored the differences in CB’s physicochemical properties during the spontaneous fermentation of “Criollo” cocoa grown in northeastern Peru and the Gulf Coast of Mexico [32], and “Forastero” cocoa cultivated in Brazil [33].

2. Materials and Methods

2.1. Biological Material

This study was conducted using cocoa varieties cultivated in Brazil (Forastero), Peru (Criollo Nativo), and Mexico (Criollo). The Forastero variety was sourced from Fazenda Michinori, located in the municipality of Tomé-Açu, Pará State (02°28′41.3″ S, 48°16′50.7″ W). The Criollo Nativo variety was obtained from the APROCAM Cooperative in the district of Copallín, Amazonas region (5°41′13.70″ S, 78°24′20.58″ W). The Mexican Criollo variety was collected from “Las Delias” farm, situated along the Comalcalco local road, postal code 86650, in the city of Comalcalco, Tabasco State (18°13′08″ N, 93°14′35.6″ W).

2.2. Spontaneous Fermentation of Cocoa Beans and Sampling

From the onset of spontaneous fermentation to the sampling of the beans, all procedures were conducted in each country of origin following the protocols used by local producers in each country [34]. Briefly, 150 kg of cocoa beans was placed into fermentation boxes and covered with polyethylene sheets. The fermentation of the “Forastero” variety lasted three days, with turnings every 24 h [35]. The “Criollo Nativo” variety underwent a seven-day fermentation with turnings every 48 h [36]. The Mexican Criollo variety was fermented over seven days, beginning with the first turning at 48 h, followed by turnings every 24 h until the final day [37].

Ferment sampling was conducted following the protocol described by Balcázar-Zumaeta et al. [38], with minor modifications. Aseptic extractions of 150 g of cocoa beans (separated from the pulp) were performed every 24 h for the duration of each fermentation, except for the Mexican cocoa, for which sampling began at 48 h. The cocoa bean samples were placed in sterilized bags and transported in liquid nitrogen to the designated laboratories in each country—the Laboratory of Biotechnological Processes (Universidade Federal do Pará, Brazil); the Laboratory for Research in Food Engineering and Postharvest Technology (Universidad Nacional Toribio Rodríguez de Mendoza de Amazonas, Peru); and the Laboratory of the Academic Division of Agricultural Sciences (Universidad Juárez Autónoma de Tabasco, Mexico)—where they were stored under ultra-freezing conditions (−86 °C).

2.3. Cocoa Bean Sampling

In the laboratory, cocoa samples were freeze-dried using a lyophilizer, as described by Balcázar-Zumaeta et al. [36]. In brief, the beans were placed in 50 mL Falcon tubes and subjected to 0.008 bar and −84 °C for 18 h. Subsequently, fat was extracted from the lyophilized beans using petroleum ether in a Soxhlet extractor, following the methodology outlined by Abbas et al. [39] and Balcázar-Zumaeta et al. [38]. For each extraction, approximately 10 g of lyophilized fermented cocoa was used, with 8 siphons regulated at 75 °C. The extracted fat was placed in a desiccation chamber for 12 h to eliminate residual solvent. Thereafter, samples were heated in an oven at 65 °C for 30 min and stored in hermetically sealed vials at 4 °C until analysis.

2.4. ATR FT-IR Spectra

Spectral data from CB fat from Criollo Nativo (Peru), Criollo (Mexico), and Forastero (Brazil) were obtained at the Laboratory of Biotechnological Processes (LAPROBIO) of the Federal University of Pará (UFPA). The procedure followed the methodology described by da Silva et al. [40], with minor modifications. Spectra were acquired using mid-infrared (MIR) spectroscopy in the range of 650 to 4000 cm⁻¹, employing an Agilent Cary 630 FT-MIR spectrometer (Santa Clara, CA, USA) coupled to an attenuated total reflectance (ATR) accessory equipped with a zinc selenide crystal. Spectral resolution was set at 16 cm⁻¹, with 32 scans per sample using approximately 0.1 g of cocoa butter. After each measurement, the ATR accessory was cleaned with isopropanol to prevent cross-contamination between CB samples [41].

2.5. Nuclear Magnetic Resonance (NMR) Spectra

The ¹H and ¹³C NMR spectra, and the homonuclear correlation maps (HOMO-COSY), were obtained using a Bruker Ascend™ spectrometer (Rheinstetten, Germany), operating at 400 MHz (¹H) and 100 MHz (¹³C). Samples consisted of 100 mg of cocoa butter (CB) dissolved in 600 µL of deuterated chloroform (CDCl₃), transferred to standard 5 mm NMR tubes. Spectral data were processed with TopSpin software version 3.6.0. Free induction decays (FIDs) underwent Fourier transformation (line broadening: 0.3 Hz), and manual analysis was conducted, including a baseline correction and calibration. The residual solvent signals were used as internal references: 7.25 ppm for ¹H and 77.0 ppm for ¹³C spectra. Peak identification relied on the chemical shift (δ) and coupling constant (J) values derived from the one-dimensional (1D) ¹H and ¹³C spectra. Two-dimensional (2D) ¹H–¹H HOMO-COSY correlation maps were also obtained. The detected signals were consistently observed on the three varieties of CB samples [42,43,44,45]. Finally, the iodine value (IV) and saponification value (SV) were estimated based on integrated areas of the ¹H spectra [46,47,48]. One measurement was performed per sample.

2.5.1. Estimation of Iodine Value ($IV$)

Iodine values ($IVs$) were estimated using proton nuclear magnetic resonance (¹H NMR) spectra by integrating signals corresponding to olefinic, allylic, methylenic, and methyl protons. These integrated areas served in calculating the total number of protons and the degree of unsaturation using the following Equation (4):

$$\mathrm{Proton}\mathrm{area}\text{ : }{A}_P=\left({\displaystyle \frac{i+h}{4}}\right)$$

(1)

$$\mathrm{Olefinic}\mathrm{protons}(V)\text{ : }V={\displaystyle \frac{\left[k+j\right]-A_P}{A_P}}$$

(2)

$$\mathrm{Total}\mathrm{protons}(T)\text{ : }T={\displaystyle \frac{\mathrm{Á}rea total integrada}{A_P}}$$

(3)

$$\mathrm{Iodine}\mathrm{value}(IV)\text{ : }IV=253.81\times {\displaystyle \frac{V}{MW}}$$

(4)

where $(k+j)$ represents the integrated area of vinyl protons obtained directly from the spectra; $(i+h)$ corresponds to the protons of the two methylene groups in glycerol; and $MW$ is the molecular weight of the triglyceride in cocoa butter (CB).

2.5.2. Saponification Value ($SV$)

Saponification values ($SVs$) were determined based on ¹H NMR spectral data, integrating signals corresponding to olefinic, allylic, methylenic, and methyl protons. These areas were used to calculate the total proton contribution and subsequently the $SV$, applying the following Equation (5):

$$SV=\left[MW\times \left(-0.2358\right)\right]+398.42$$

(5)

The average molecular weight ($MW$) of the triglycerides was previously determined based on integrated areas from the ¹H NMR spectra, using the same equations applied for iodine value ($IV$) estimation. Spectral data were compiled and organized in Microsoft Excel spreadsheets. Saponification values were subsequently analyzed to compare sample quality and oxidative status across cocoa butter (CB) samples.

2.6. Lipidomic Profiling (GC/MS)

To determine the lipid content in cocoa butter (CB), a transesterification procedure in alkaline medium was performed according to the methodology described by Nascimento et al. [49]. Briefly, 20 mg of CB sample was placed in a polypropylene conical tube (Tb1) and mixed with 100 μL of potassium methoxide solution (2 N). The mixture was sonicated for 15 min at 60 °C to facilitate esterification.

Following this, 200 μL of n-hexane was added, and the sample was centrifuged at 10,000 RPM for 2 min. The organic phase (sediment) was transferred to a second tube (Tb2). To the remaining aqueous phase in Tb₁, 110 μL of hydrochloric acid (2 N) was added and manually agitated for 10 s. A solvent mixture of dichloromethane (CH₂Cl₂) and n-hexane (1:1 v/v; 200 μL) was added, followed by an additional 10-s agitation and centrifugation under the same conditions. The resulting supernatant was transferred to tube Tb₂. The residue remaining in Tb₁ was subjected to automated evaporation (35 °C, 5 psi) to remove residual liquids prior to derivatization into fatty acid methyl esters (FAMEs). These were subsequently analyzed by gas chromatography–mass spectrometry (GC/MS) for lipid profile quantification.

For the derivatization of non-transesterified free molecules, 50 μL of N,O-bis(trimethylsilyl) trifluoroacetamide (BSTFA) containing 1% trimethylchlorosilane (TMCS) was added to tube Tb₁. The mixture was incubated in an ultrasonic bath at 60 °C for 15 min to achieve effective silylation. After cooling, the transesterified lipids were transferred to tube Tb2, which was subsequently sonicated for 5 min and centrifuged under the previously described conditions. The supernatant was then deposited into a 2 mL vial containing buffer for subsequent GC/MS analysis, as described by Barra et al. [50].

Analyses were performed using a gas chromatograph (GC) equipped with an autosampler and coupled to a single quadrupole mass spectrometer (MS). The system employed a ZB-5HT capillary column (5% phenyl, 95% dimethylpolysiloxane; 30 m × 0.25 mm, 0.10 μm film thickness). Helium served as the carrier gas at a flow rate of 1.0 mL/min, with an injection volume of 1.0 μL. The injector was maintained at 250 °C. The oven temperature was programmed to rise from 200 °C to 300 °C at 20 °C/min, followed by a 10-min hold at 300 °C. MS detection was conducted at 275 °C with the ion source held at 230 °C. Ionization was achieved via electron impact at 70 eV, with a solvent delay of 5 min and a scan range set between 500 and 600 Da. Compound identification was based on mass spectral comparisons with reference libraries: NIST2011, WILEY2009, and FAMES2011. Retention times were validated using Kovats retention indices calculated from a homologous hydrocarbon series. Quantification of lipid species (FAMEs) was performed by normalizing the area of each chromatographic peak to the total peak area, expressed as percentage values [50].

2.7. Crystallization and Melting Behavior

To evaluate the crystallization and melting behavior of cocoa butter (CB) during spontaneous fermentation, differential scanning calorimetry (DSC) was performed using a Discovery DSC 2500 instrument (TA Instruments, New Castle, DE, USA), following the protocol described by Castro-Alayo et al. [51] with minor modifications. A 10 mg portion of CB was placed in a low-mass aluminum pan (Tzero, New Castle, DE, USA) and hermetically sealed using a Tzero press (TA Instruments). An empty pan was used as a reference for baseline adjustment. Samples were first heated from ambient temperature (~16 °C) to 60 °C for 10 min to eliminate crystalline memory and ensure sample homogeneity. Subsequently, the samples were cooled at a rate of 10 °C/min to −35 °C and held for 10 min, followed by heating at the same rate up to 60 °C. Using TRIOS software version 5.3, crystallization temperature (T_c) and melting temperature (T_m) were determined from the exothermic and endothermic peaks, respectively, within the heat flow vs. temperature thermograms. Onset (T_onset) and endset (T_endset) temperatures were identified for each peak. Baseline interpolation was used to calculate the enthalpies of crystallization and melting from peak areas. All measurements were performed in triplicate.

2.8. Isothermal Crystallization and Kinetic Modeling

Isothermal crystallization of cocoa butter (CB) was conducted following the procedures described by Castro-Alayo et al. [51] and Fernandes et al. [52]. Based on the recommendations of Rashid et al. [53], three crystallization temperatures (T_c: 16, 17, and 18 °C) were selected, representing values 0–4 °C below the melting point (T_m) previously determined in Section 2.7. These temperatures were calculated according to the established criteria for supercooling suitable for CB crystallization analysis.

Approximately 9–10 mg of CB was sealed in low-mass aluminum pans (Tzero) using a Tzero press, with an empty pan as reference. Thermal history was erased by heating each sample to 50 °C for 3 min, followed by rapid cooling at 60 °C/min to the selected Tc. The samples were then held isothermally at these temperatures for 90 min, during which exothermic events corresponding to crystal formation were recorded. Subsequently, samples were reheated at a rate of 20 °C/min up to 50 °C to capture endothermic transitions. The temperature and enthalpy of melting for crystals formed at each Tc were recorded, enabling identification of the polymorphic forms resulting from isothermal crystallization (Figures S1–S3).

The crystallization kinetics of the cocoa butter (CB) samples were calculated based on isothermal crystallization data using the Avrami equation. Data fitting was performed with the Crystallization Fit package available in Origin Pro 2024 (version 10.1.0.178), as described by Castro-Alayo et al. [31]. This approach enabled the extraction of the Avrami parameters, the rate constant ($k$) and Avrami exponent ($n$), which provide insights into nucleation mechanisms and crystal growth behavior (6).

$$1-V_c\left(t\right)=\exp \left(-k{t}^n\right)$$

(6)

where $V_c$ is the fraction of crystals formed at time t during isothermal crystallization (7).

$$\ln \left[-\ln \left(1-V_c\left(t\right)\right)\right]=lnk+n·lnt$$

(7)

From the equation $\ln \left[-\ln \left(1-V_c\left(t\right)\right)\right]$ vs. $lnt$, the Avrami exponent ($n$) was determined from the slope of the linear fit. The time required to reach 50% crystallization ($t_{1/2}$) was calculated based on the extracted parameters $k$ and $n$, using the following Equation (8):

$$t_{1/2}={\left[{\displaystyle \frac{0.69315}{k}}\right]}^{{\displaystyle \frac{1}{n}}}$$

(8)

Finally, the induction time of crystallization (t₀)—defined as the time required for the onset of the crystallization peak—was determined following the approach described by Castro-Alayo et al. [51]. This parameter was calculated from the isothermal thermograms, where t₀ corresponds to the elapsed time between the start of the isothermal hold and the initial deviation in heat flow from the baseline, indicating the onset of crystal nucleation.

2.9. Data Analysis

Thermal behavior variations observed in DSC analyses were subjected to one-way analysis of variance (ANOVA) to evaluate the effect of spontaneous fermentation time. For broader analytical comparison, multivariate techniques were applied, including hierarchical cluster analysis (HCA) and principal component analysis (PCA), to explore patterns across thermal and lipidomic variables. These multivariate evaluations were performed using RMarkdown within RStudio (version 2024.12.0+467; Boston, MA, USA), enabling reproducible workflows and comprehensive visualization of data structure and sample relationships.

3. Results and Discussion

3.1. Cocoa Butter Lipidomics

According to Herrera-Rocha et al. [5], the lipid fraction accounts for over 50% of the cocoa bean, consisting mainly of fatty acids (FAs) [15]. As shown in Figure 1, the predominant FAs in cocoa butter (CB) were stearic, oleic, and palmitic acids, consistent with findings reported by Alvarez et al. [54], Bergenholm et al. [21], da Silva Santos et al. [55], and Yucel et al. [15]. These components, linked by a glycerol backbone, are associated with the formation of triglycerides such as POS, SOS, and POP—lipid structures characteristic of CB [56,57]. In this study, stearic acid content ranged between 30.2% (Peru, Figure 1a) and 36.7% (Brazil, Figure 1b) during spontaneous fermentation, in line with previously reported values of 32.9–36.1% for this FA [12,58,59,60]. Overall, stearic acid levels in CB from all three countries remained stable, though an increase was noted between 72 and 120 h of fermentation, similarly described by Herrera-Rocha et al. [5]. In contrast, during short-term fermentation (CB from Brazil), stearic acid content remained around 36% (Figure 1b). According to Fibrianto et al. [61], this concentration leads to CB with greater hardness and a higher melting point, attributed to the long-chain nature and high melting temperature of stearic acid [62].

Oleic acid content ranged from 32.4% (Brazil) to 35.9% (Mexico), as illustrated in Figure 1. In short-duration fermentations (72–96 h), this fatty acid exceeded 34% in the Criollo variety (Figure 1a,c), except for Brazil (~32%, Figure 1b), which may be attributed to the use of the Forastero variety. Despite these differences, the oleic acid concentration in CB was consistent with values reported by Barišić et al. [58] and Lapčík et al. [60], who found levels between 32.7% and 37%. These results align with descriptions of Latin American-origin CB, recognized for its high oleic acid content [23]. Furthermore, the presence of oleic acid in cocoa butter (CB) is a distinctive feature of this edible fat, contributing to reduced susceptibility to oxidation [55]. Regarding palmitic acid concentration, levels ranged from 24.4% (Mexico, Figure 1c) to 29.5% (Peru, Figure 1a), similar to those reported by Chun & Meng [12] and Lapčík et al. [60]. This fatty acid is one of the three predominant saturated FAs in CB, corroborating findings from Alvarez et al. [54] and Herrera-Rocha et al. [5]. Also, palmitic acid content decreased by approximately 2–3%; however, from the third day onward, its concentration exceeded 25% across all studied varieties. These results indicate that this FA is present regardless of variety, as reported by Silveira et al. [63], who noted that FA availability is associated with an increase in TAGs containing saturated fatty acids, which in turn contribute to weak co-crystallization and the formation of β′-type polymorphs in CB [21,64].

Trace amounts of specific fatty acids (FAs) were detected in cocoa butter (CB) during this study. Myristic acid, a saturated FA essential to cocoa [10,65], was found in concentrations ranging from 0.01% (Brazil, Figure 1b) to 0.08% (Mexico, Figure 1c). This FA contributes to enhanced molecular packing, promoting the formation of more stable crystalline structures in CB [64]. Throughout fermentation, its concentration remained relatively stable, with the exception of the Brazilian samples, in which it disappeared after 48 h. These low levels are consistent with prior findings by Iswari et al. [65] and Sabahannur & Alimuddin [66], who reported concentrations of 0.01–0.05% in CB. Such reduced quantities are considered favorable, as elevated levels of this FA have been associated with increased total cholesterol and plasma LDL concentrations [67]. Pentadecanoic acid, on the other hand, remained stable at approximately 0.02% during fermentation across all three cocoa varieties. Although present at low concentrations, this FA contributes to the development of the final flavor in products such as chocolate and has been linked to a reduced risk of comorbidities [68,69].

Margaroleic acid was detected in the Criollo variety (Peru) at up to 72 h of fermentation, with concentrations ranging from 0.03% to 0.04%, consistent with reference ranges reported by Chun & Meng [12], Cooney et al. [70], and Ramel et al. [17], citing values between 0.02% and 0.05% in CB. Palmitoleic acid, a monounsaturated FA characteristic of CB [10,21], reached concentrations of up to 0.24% at 72 h (Figure 1a) during fermentations shorter than 120 h. Additionally, linoleic and eicosenoic acids peaked at up to 0.07% in CB samples from all three countries. Linoleic acid—an unsaturated FA—together with stearic, oleic, and palmitic acids accounted for over 95% of total FA content in CB [12]. Eicosenoic acid, also unsaturated, was reported in CB from Criollo Nativo (Peru) and Criollo (Mexico) varieties and was not detected in Forastero (Brazil), suggesting its potential as a varietal marker. Furthermore, palmitoleic acid reached 0.24% and margaric acid 0.3% across all three countries—values aligned with prior studies and associated with reduced cardiovascular disease risk [12,17,54,60,70,71,72,73,74].

Behenic and lignoceric acids were reported at concentrations of 0.23% and 0.1%, respectively (Figure 1), during short-term fermentation (72–96 h) in CB from all three countries. These values are consistent with those found by Alvarez et al. [54], Chun & Meng [12], Figueira & Luccas [71], Joshi et al. [72], Liu et al. [75], Norazlina et al. [76], and Ramel et al. [17]. Both are classified as long-chain fatty acids (C22 and C24) with high melting points. Additionally, other FAs such as linolelaidic acid, tricosanoic acid, and azelaic acid were detected. These are also considered long-chain FAs (with more than 18 carbon atoms), and their formation has been associated with hydrolytic processes occurring in CB [77,78].

Based on the fatty acid (FA) profiles identified in cocoa butter (CB) samples during spontaneous fermentation in Peru, Brazil, and Mexico, a multivariate analysis was conducted using principal component analysis (PCA) and hierarchical cluster analysis (HCA). PCA results revealed that the first two principal components (Figure 2a) accounted for 98.68% of the total variance, supported by high eigenvalues of 6.67 and 2.06, respectively. The biplot projection along PC1 and PC2 demonstrated a clear separation pattern among the fermentation processes of the three countries, indicating distinct lipidomic signatures. This differentiation was further corroborated by the HCA dendrogram (Figure 2b), which presented a top-down clustering pattern with three main branches, each corresponding to CB samples from one country. The node rankings, indicated by numerical annotations as per Balcázar-Zumaeta et al. [36], validated the geographical clustering. The graphical representation confirmed the presence of three predominant FAs—oleic, stearic, and palmitic—in CB samples from all three countries. Variable correlations, assessed via factor map distances, highlighted FA contributions to sample quality: the farther a variable is from the origin, the better its representation in the factorial space. Accordingly, CB from the Criollo variety (Peru) was distinguished by elevated palmitic acid content, consistent with other local cultivars [79]. The Forastero variety (Brazil) exhibited higher stearic acid levels, as previously noted by Ghazani & Marangoni [23], while Criollo CB from Mexico was marked by oleic acid prevalence. Importantly, these dominant FAs were consistently observed in significant proportions within fermentations lasting less than 120 h, further supporting the notion that short-term fermentation not only promotes the formation of stearic, oleic, and palmitic acids but also enhances levels of linoleic and arachidic acids [11].

Cluster dissimilarity among fermentation times was calculated using the complete linkage method with Euclidean distance. Each node in the dendrogram displays two p-values derived from distinct bootstrapping algorithms: the value on the left represents the approximately unbiased (AU) p-value, calculated via multiscale bootstrap resampling; the value on the right reflects the bootstrap probability (BP) p-value, obtained through conventional bootstrap resampling. The AU p-value offers a more reliable measure of how strongly each cluster is supported by the data structure.

3.2. Thermal Profile of Cocoa Butter (DSC)

Table 1. Crystallization and melting profile of cocoa butter from three varieties.

Variety	Fermentation Time	Crystallization				Melting
		Tonset (°C)	Tc (°C)	Tendset (°C)	Enthalpy (J/g)	Tonset (°C)	Tm (°C)	Tendset (°C)	Enthalpy (J/g)
Criollo Nativo (Peru)	0 h	14.48 ± 0.07 a	9.65 ± 0.09 ab	14.30 ± 0.15 a	53.86 ± 1.19 ab	14.69 ± 0.08	20.58 ± 0.02 a	26.27 ± 0.12 a	70.70 ± 1.02 ab
	24 h	14.29 ± 0.14 ab	9.94 ± 0.16 a	14.17 ± 0.17 a	54.84 ± 1.53 a	14.39 ± 0.06	19.74 ± 0.13 e	25.60 ± 0.10 c	72.95 ± 1.79 a
	48 h	14.28 ± 0.08 ab	9.49 ± 0.18 b	14.01 ± 0.07 a	55.99 ± 0.62 a	14.37 ± 0.03	20.51 ± 0.09 ab	25.98 ± 0.12 b	72.73 ± 1.71 a
	72 h	13.75 ± 0.08 c	9.72 ± 0.12 ab	13.55 ± 0.13 b	55.19 ± 0.23 a	13.90 ± 0.13	19.15 ± 0.06 f	26.07 ± 0.08 ab	69.50 ± 0.97 abc
	96 h	13.15 ± 0.10 d	8.63 ± 0.17 c	13.02 ± 0.12 c	50.92 ± 0.59 bc	14.52 ± 0.14	20.23 ± 0.08 bc	25.99 ± 0.14 ab	66.09 ± 1.49 c
	120 h	14.10 ± 0.18 b	9.87 ± 0.14 ab	14.00 ± 0.19 a	54.23 ± 1.27 ab	13.55 ± 1.16	19.77 ± 0.07 e	25.39 ± 0.02 c	73.49 ± 0.65 a
	144 h	13.51 ± 0.04 c	8.68 ± 0.02 c	13.27 ± 0.03 bc	52.49 ± 1.45 abc	14.25 ± 0.25	20.09 ± 0.07 cd	26.07 ± 0.10 ab	66.79 ± 0.91 bc
	168 h	13.12 ± 0.13 d	8.99 ± 0.26 c	13.00 ± 0.17 c	49.89 ± 2.34 c	14.38 ± 0.23	19.89 ± 0.21 de	25.51 ± 0.06 c	65.71 ± 2.25 c
Forastero (Brazil)	0 h	14.69 ± 0.08 c	10.96 ± 0.04 b	13.03 ± 2.29	64.18 ± 0.84	12.40 ± 0.09 b	19.65 ± 0.05 c	25.71 ± 0.49 b	79.84 ± 0.53
	24 h	15.60 ± 0.04 b	11.07 ± 0.02 b	15.29 ± 0.03	63.79 ± 0.84	12.31 ± 0.15 b	20.17 ± 0.01 b	26.38 ± 0.04 ab	78.93 ± 0.49
	48 h	14.81 ± 0.21 c	11.07 ± 0.18 b	14.64 ± 0.17	64.48 ± 1.53	13.55 ± 0.37 a	20.08 ± 0.09 b	25.54 ± 0.30 b	79.80 ± 1.47
	72 h	16.03 ± 0.04 a	11.45 ± 0.17 a	15.61 ± 0.12	63.94 ± 0.15	12.18 ± 0.13 b	20.49 ± 0.19 a	27.03 ± 0.38 a	79.67 ± 0.96
Criollo (Mexico)	0 h	14.46 ± 0.08	10.61 ± 0.10	14.26 ± 0.11	52.42 ± 0.46	14.04 ± 0.06 b	20.06 ± 0.09 ab	24.74 ± 0.09	70.69 ± 0.92
	48 h	14.58 ± 0.03	11.11 ± 0.08	14.46 ± 0.13	53.29 ± 1.66	15.02 ± 0.02 a	19.97 ± 0.06 ab	24.98 ± 0.29	72.61 ± 0.14
	72 h	14.24 ± 0.05	11.01 ± 0.12	14.26 ± 0.06	54.68 ± 0.20	14.94 ± 0.10 a	19.73 ± 0.10 ab	24.56 ± 0.63	73.54 ± 0.74
	96 h	14.53 ± 0.06	11.02 ± 0.05	14.45 ± 0.03	54.19 ± 1.70	14.99 ± 0.09 a	19.76 ± 0.10 ab	24.68 ± 0.14	74.74 ± 0.32
	120 h	14.40 ± 0.04	11.05 ± 0.23	14.32 ± 0.03	53.62 ± 0.95	14.91 ± 0.11 a	19.77 ± 0.20 ab	24.76 ± 0.28	72.38 ± 0.18
	144 h	14.40 ± 0.19	10.79 ± 0.38	14.30 ± 0.21	53.57 ± 2.57	15.28 ± 0.16 a	20.31 ± 0.46 a	25.17 ± 0.45	71.03 ± 4.64
	168 h	14.23 ± 0.39	10.63 ± 0.46	14.14 ± 0.39	55.69 ± 0.61	14.24 ± 0.36 b	19.60 ± 0.19 b	25.18 ± 0.15	70.58 ± 2.75

Values within the same column with different letters are significantly different (p < 0.05). Each value in the table represents the mean ± SD of three replicates.

summarizes the crystallization and melting behavior of cocoa butter (CB) under spontaneous fermentation across Peru, Brazil, and Mexico. CB from Peru (Criollo Nativo) began crystallizing at between 13.12 °C and 14.48 °C, with a maximum crystallization temperature (T_c) of 9.94 °C. In contrast, Brazilian CB exhibited crystallization onset between 14.69 °C and 15.60 °C, reaching a T_c of 11.45 °C. The Criollo variety from Mexico displayed an onset of crystallization near 14.5 °C, with a T_c peak of 11.11 °C. Moreover, the period between 72 and 96 h was characterized by the highest T_c and T_onset, consistent with previous reports [31,80].

The T_c reflects an exothermic transition from amorphous to crystalline solid. As shown in Table 1, T_c values remained elevated during the initial 72 h of fermentation across all three regions. During this period, CB from Brazil and Mexico presented higher crystallization peaks (11.45 °C and 11.11 °C, respectively) compared to Peru (9.94 °C). These outcomes align with previous findings: those of Aumpai et al. [81] (10.6 °C), Bayés-García et al. [80] (13.5 °C), and Castro-Alayo et al. [31] (12.9 °C). The observed variation may be attributed to the crystallization rate [3] of cocoa butter extracted from raw cocoa. The thermal behavior of CB during fermentation suggests the presence of α and β′ polymorphic forms in the cocoa beans [31,82]. These forms may evolve into more stable polymorphs during chocolate manufacturing [83]. Crystallization enthalpies measured in CB were 64.48 J/g (Forastero), 55.9 J/g (Criollo Nativo), and 55.6 J/g (Criollo), consistent with Castro-Alayo et al. [31], who noted enthalpy increases alongside Tc. Notably, the lowest enthalpy values occurred at the early stages of fermentation.

The melting behavior of cocoa butter (CB) in the bean at the onset of fermentation exhibited a melting range between approximately 17.5 °C (T_onset) and 28.6 °C (T_endset), in agreement with values reported by Bayés-García et al. [84] and Castro-Alayo et al. [31]. These findings closely match those obtained in the present study, as shown in Table 1, where the fermentation time had a statistically significant effect (p < 0.05). Regarding the melting peak temperature (T_m), fermentation duration was also a significant factor (p < 0.05). Although higher T_m values were observed at the beginning of fermentation (e.g., 20.58 °C), extended fermentation periods (e.g., 7 days in varieties from Peru and Mexico) did not yield a consistent trend in T_m variability. In contrast, shorter fermentation time (Forastero) led to an increase in T_m, reaching 20.49 °C on day 4. Previous studies have reported that fermentation periods shorter than four days promote stable melting properties, which are associated with the formation of desirable polymorphs in CB [34]. An increase in T_m is linked to improved thermal stability, attributed to changes in molecular packing density [55].

According to Table 1, the melting peak temperature (T_m) ranged between ~19.6 °C and 20.58 °C, which is consistent with the presence of α polymorphs in CB [31,51]. These temperatures remain below the melting points of metastable forms (α and β’), suggesting a potential facilitation of transition toward the more stable β polymorphs [6,85]. Concurrently, short fermentation periods yielded high enthalpy values (~79.8 J/g), indicative of increased energy release—critical for achieving the threshold nucleus size and promoting tightly packed, thermally stable crystalline structures [86,87,88].

The Avrami index (n) provides insight into crystallization kinetics by describing nucleation type and crystal growth dimensionality [3,89]. As shown in

Table 2. Kinetics of crystallization parameters of cocoa butter during fermentation from three varieties.

Crystallization Temperature (°C)	Fermentation Time	Criollo Nativo					Forastero					Criollo
		n	k (min−n)	t1/2 theo (min)	t1/2 exp (min)	t0 (min)	n	k (min−n)	t1/2 theo (min)	t1/2 exp (min)	t0 (min)	n	k (min−n)	t1/2 theo (min)	t1/2 exp (min)	t0 (min)
16	0 h	3.48 ± 0.10 abcdef	5.69 × 10−5 ± 1.65 × 10−5 bcd	15.01 ± 0.18 ghi	14.97 ± 0.18 ef	12.68 ± 0.52 fghi	2.93 ± 0.24 d	9.22 × 10−4 ± 6.22 × 10−4 a	10.06 ± 0.39 f	9.98 ± 0.45 e	10.40 ± 1.38 d	3.07 ± 0.08 ab	1.05 × 10−4 ± 2.05 × 10−5 c	17.63 ± 0.69 abcde	17.61 ± 0.52 abcd	21.70 ± 0.52 bcdef
	24 h	3.69 ± 0.23 abcd	5.93 × 10−5 ± 3.59 × 10−5 bcd	13.23 ± 0.56 i	13.35 ± 0.54 f	10.40 ± 0.73 i	3.21 ± 0.22 bcd	1.75 × 10−4 ± 1.16 × 10−4 b	14.09 ± 1.38 de	14.19 ± 1.47 cd	11.00 ± 0.83 bcd	NIS	NIS	NIS	NIS	NIS
	48 h	3.027 ± 0.33 bcdef	1.28 × 10−4 ± 1.47 × 10−4 bcd	18.29 ± 0.16 defghi	18.99 ± 0.55 bcdef	14.43 ± 0.65 efghi	3.52 ± 0.37 abc	1.49 × 10−4 ± 9.21 × 10−5 b	14.28 ± 0.69 cde	14.31 ± 0.73 cd	13.11 ± 0.73 abcd	2.04 ± 0.14 c	3.56 × 10−2 ± 1.29 × 10−2 a	4.37 ± 0.42 h	4.56 ± 0.42 e	10.16 ± 1.18 l
	72 h	3.083 ± 0.27 bcdef	1.89 × 10−4 ± 1.08 × 10−4 bcd	15.09 ± 0.06 ghi	15.64 ± 0.10 def	12.44 ± 1.01 ghi	3.06 ± 0.06 cd	1.88 × 10−4 ± 3.97 × 10−5 b	14.71 ± 0.32 bcde	14.73 ± 0.38 bcd	14.12 ± 0.38 abc	2.86 ± 0.09 abc	3.21 × 10−4 ± 7.90 × 10−5 c	14.72 ± 0.25 efg	15.03 ± 0.45 cd	12.81 ± 0.31 jkl
	96 h	3.01 ± 0.35 bcdef	1.19 × 10−4 ± 9.99 × 10−5 bcd	19.52 ± 1.08 defghi	20.19 ± 1.43 bcdef	16.35 ± 1.03 defghi	NIS	NIS	NIS	NIS	NIS	2.95 ± 0.09 ab	3.31 × 10−4 ± 8.13 × 10−5 c	13.47 ± 0.09 g	13.77 ± 0.10 d	12.75 ± 0.52 jkl
	120 h	3.35 ± 0.10 abcdef	9.43 × 10−5 ± 2.99 × 10−5 bcd	14.43 ± 0.39 ghi	14.79 ± 0.36 f	11.42 ± 0.42 hi	NIS	NIS	NIS	NIS	NIS	2.87 ± 0.07 abc	2.05 × 10−4 ± 6.26 × 10−5 c	17.11 ± 0.59 cdefg	17.07 ± 0.63 abcd	12.38 ± 0.28 kl
	144 h	2.96 ± 0.25 bcdef	1.26 × 10−4 ± 2.29 × 10−5 bcd	14.71 ± 1.95 ghi	15.33 ± 2.05 def	11.54 ± 1.30 hi	NIS	NIS	NIS	NIS	NIS	2.80 ± 0.06 abc	3.68 × 10−4 ± 1.38 × 10−4 c	15.07 ± 1.35 efg	15.03 ± 0.93 cd	12.20 ± 0.63 kl
	168 h	2.43 ± 0.34 f	2.27 × 10−4 ± 4.39 × 10−5 bc	24.85 ± 6.75 abcd	25.61 ± 7.33 ab	20.68 ± 5.77 cdefg	NIS	NIS	NIS	NIS	NIS	2.75 ± 0.06 abc	5.92 × 10−4 ± 2.79 × 10−4 c	13.65 ± 2.65 fg	14.31 ± 3.41 cd	12.87 ± 0.75 jkl
17	0 h	4.33 ± 0.22 a	5.24 × 10−6 ± 3.01 × 10−6 d	15.67 ± 0.21 fghi	15.99 ± 0.21 def	13.65 ± 0.52 fghi	3.16 ± 0.11 bcd	2.06 × 10−4 ± 5.76 × 10−5 b	13.24 ± 0.08 ef	13.35 ± 0.01 d	11.54 ± 1.30 abcd	3.58 ± 0.12 a	2.19 × 10−5 ± 1.06 × 10−5 c	18.42 ± 0.54 abcde	18.39 ± 0.48 abc	23.99 ± 0.36 bc
	24 h	3.61 ± 0.04 abcde	4.71 × 10−5 ± 6.41 × 10−6 bcd	14.33 ± 0.19 ghi	14.31 ± 0.21 f	13.41 ± 0.58 fghi	3.49 ± 0.18 abcd	4.01 × 10−5 ± 2.38 × 10−5 b	17.34 ± 1.66 abc	17.68 ± 1.65 ab	10.82 ± 1.43 cd	NIS	NIS	NIS	NIS	NIS
	48 h	2.97 ± 0.39 bcdef	1.39 × 10−4 ± 9.87 × 10−5 bcd	19.41 ± 1.44 defghi	20.44 ± 2.14 bcdef	19.89 ± 0.99 cdefgh	3.37 ± 0.36 abcd	7.19 × 10−5 ± 8.73 × 10−5 b	18.11 ± 1.85 a	18.39 ± 1.95 a	11.82 ± 1.72 abcd	2.35 ± 0.21 bc	1.58 × 10−2 ± 8.96 × 10−3 b	5.25 ± 0.65 h	5.41 ± 0.65 e	14.85 ± 0.99 ijk
	72 h	3.64 ± 0.89 abcd	5.36 × 10−5 ± 4.97 × 10−5 bcd	18.96 ± 1.09 defghi	19.84 ± 1.18 bcdef	15.27 ± 2.46 defghi	3.18 ± 0.08 bcd	1.46 × 10−4 ± 1.41 × 10−4 b	17.11 ± 1.58 abcd	17.08 ± 1.57 abc	14.55 ± 1.98 a	2.95 ± 0.14 ab	2.08 × 10−4 ± 6.46 × 10−5 c	15.85 ± 0.82 defg	16.17 ± 1.09 bcd	18.64 ± 0.63 efghi
	96 h	2.79 ± 0.25 cdef	1.05 × 10−4 ± 8.06 × 10−5 bcd	22.47 ± 3.36 bcdef	23.19 ± 3.75 abcd	23.99 ± 2.98 bcd	NIS	NIS	NIS	NIS	NIS	3.09 ± 0.09 ab	1.74 × 10−4 ± 5.06 × 10−5 c	14.82 ± 0.21 efg	14.97 ± 0.18 cd	17.37 ± 0.10 fghi
	120 h	3.87 ± 0.26 abc	1.64 × 10−5 ± 1.34 × 10−5 cd	16.71 ± 0.49 efghi	17.07 ± 0.38 cdef	13.83 ± 0.55 efghi	NIS	NIS	NIS	NIS	NIS	2.97 ± 0.12 ab	1.46 × 10−4 ± 5.14 × 10−5 c	17.48 ± 0.35 abcdef	17.31 ± 0.18 abcd	19.54 ± 0.55 cdefgh
	144 h	2.90 ± 0.21 bcdef	1.68 × 10−4 ± 5.84 × 10−5 bcd	18.01 ± 1.78 defghi	18.52 ± 1.46 bcdef	15.57 ± 2.85 defghi	NIS	NIS	NIS	NIS	NIS	2.93 ± 0.15 ab	2.11 × 10−4 ± 1.19 × 10−4 c	16.32 ± 0.69 defg	16.77 ± 0.83 bcd	17.07 ± 0.68 ghij
	168 h	2.51 ± 0.23 ef	4.99 × 10−4 ± 6.63 × 10−5 a	20.92 ± 3.99 cdefgh	22.72 ± 4.51 abcde	16.66 ± 3.72 defghi	NIS	NIS	NIS	NIS	NIS	3.38 ± 0.43 a	8.82 × 10−5 ± 9.07 × 10−5 c	16.131 ± 0.52 defg	16.65 ± 0.99 bcd	15.69 ± 2.73 hijk
18	0 h	3.72 ± 0.18 abcd	2.60 × 10−5 ± 1.28 × 10−5 bcd	15.92 ± 0.33 fghi	16.17 ± 0.28 def	22.78 ± 0.52 bcde	3.29 ± 0.08 abcd	6.07 × 10−5 ± 2.63 × 10−5 b	17.37 ± 1.43 abc	17.74 ± 1.30 ab	10.46 ± 0.48 d	3.04 ± 0.21 ab	7.55 × 10−5 ± 5.49 × 10−5 c	21.32 ± 0.43 a	21.34 ± 0.52 a	32.10 ± 0.31 a
	24 h	3.91 ± 0.07 ab	2.33 × 10−5 ± 5.29 × 10−6 cd	13.97 ± 0.31 hi	14.25 ± 0.31 f	17.73 ± 0.73 defghi	3.72 ± 0.06 ab	1.65 × 10−5 ± 1.65 × 10−6 b	17.52 ± 0.54 ab	18.04 ± 0.48 a	14.79 ± 0.90 a	NIS	NIS	NIS	NIS	NIS
	48 h	3.19 ± 0.29 bcdef	2.45 × 10−5 ± 2.64 × 10−5 cd	29.14 ± 3.98 ab	29.76 ± 4.02 a	30.18 ± 0.99 ab	3.53 ± 0.09 abc	2.19 × 10−5 ± 9.36 × 10−6 b	19.11 ± 0.69 a	19.72 ± 0.68 a	14.61 ± 0.79 a	2.49 ± 0.45 bc	1.35 × 10−2 ± 1.28 × 10−2 bc	5.54 ± 1.13 h	5.74 ± 1.26 e	19.12 ± 1.30 defghi
	72 h	2.72 ± 0.07 def	6.19 × 10−5 ± 1.29 × 10−5 bcd	30.97 ± 0.21 a	29.88 ± 0.52 a	27.29 ± 2.98 abc	3.78 ± 0.06 a	8.68 × 10−6 ± 1.56 × 10−6 b	19.79 ± 0.33 a	20.14 ± 0.38 a	14.43 ± 1.01 ab	2.93 ± 0.13 ab	1.51 × 10−4 ± 3.87 × 10−5 c	17.96 ± 0.72 abcde	18.34 ± 0.99 abc	24.23 ± 1.15 b
	96 h	2.92 ± 0.39 bcdef	6.84 × 10−5 ± 8.21 × 10−5 bcd	28.16 ± 0.77 abc	28.38 ± 0.37 a	34.93 ± 5.54 a	NIS	NIS	NIS	NIS	NIS	3.07 ± 0.15 ab	1.22 × 10−4 ± 5.09 × 10−5 c	17.14 ± 0.56 bcdefg	17.25 ± 0.55 abcd	22.12 ± 0.83 bcde
	120 h	3.43 ± 0.87 abcdef	8.34 × 10−5 ± 1.22 × 10−4 bcd	21.49 ± 0.43 cdefg	22.85 ± 0.58 abcd	21.46 ± 4.55 bcdef	NIS	NIS	NIS	NIS	NIS	3.17 ± 0.26 ab	5.52 × 10−5 ± 4.43 × 10−5 c	21.02 ± 0.81 ab	21.58 ± 0.85 a	23.15 ± 1.03 bcd
	144 h	3.11 ± 0.18 bcdef	7.30 × 10−5 ± 1.57 × 10−5 bcd	23.39 ± 3.15 bcde	24.07 ± 2.52 abc	20.02 ± 5.81 cdefgh	NIS	NIS	NIS	NIS	NIS	2.97 ± 0.19 ab	1.05 × 10−4 ± 6.52 × 10−5 c	20.35 ± 1.09 abc	20.44 ± 1.06 ab	21.16 ± 1.16 bcdefg
	168 h	2.93 ± 0.42 bcdef	2.36 × 10−4 ± 1.04 × 10−4 b	20.65 ± 3.94 defgh	22.66 ± 4.80 abcde	16.23 ± 4.64 defghi	NIS	NIS	NIS	NIS	NIS	3.38 ± 0.95 a	1.53 × 10−4 ± 1.66 × 10−4 c	19.47 ± 4.07 abcd	20.20 ± 4.73 ab	21.16 ± 5.00 bcdefg

Values within the same column with different letters are significantly different (p < 0.05). Each value in the table represents the mean ± SD of three replicates. Abbreviations: crystallization rate constant (k), Avrami index (n), not in study (NIS).

, fermentation time significantly affected n (p < 0.05), with mean values ranging from 2.04 to 3.78. Values near 2 suggest needle-like growth from sporadic nuclei (Forastero; 16–17 °C), while values around 3 (Criollo Nativo and Criollo; 17–18 °C) indicate disk- or spherulitic-like growth from sporadic or instantaneous nucleation. Values approaching 4 reflect spherulitic growth from instantaneous nuclei [89,90]. The highest n values were observed between the third and fifth days of fermentation across all varieties, revealing a mixed nucleation and growth mechanism (rod and spherulitic) [70]. Crystallization rate (k), known as the Avrami constant, is strongly dependent on crystallization temperature [19]. Both k and theoretical/experimental half-time (t₁/₂), reported in Table 2, decreased throughout fermentation, likely due to changes in CB’s chemical composition and the emergence of compounds influencing crystallization behavior [31]. Furthermore, kinetic parameters yielded an average R² of ~0.999, indicating an excellent model fit to CB crystallization data [31,91].

Ostrowska-Ligęza et al. [8] and Zhao & James [92] report that cocoa butter (CB) is composed of TAGs capable of crystallizing into six polymorphic forms, designated I through VI, in order of increasing stability. This study is the first to characterize the polymorphic forms present in CB during the fermentation of cocoa beans from three genetic varieties (

Table 3. Principal polymorphic forms of cocoa butter from three varieties characterized by differential scanning calorimetry (DSC).

Variety	Fermentation Time	Form I (γ)	Form II (α)	Form III (β’2)
Criollo Nativo	0 h	61.61 ± 1.15 abc	0.00 ± 0.00	38.39 ± 1.15 a
	24 h	70.12 ± 2.55 a	0.00 ± 0.00	29.88 ± 2.55 b
	48 h	64.12 ± 7.15 ab	0.00 ± 0.00	35.81 ± 7.07 ab
	72 h	61.64 ± 0.16 abc	0.00 ± 0.00	38.36 ± 0.16 a
	96 h	53.73 ± 3.03 c	46.27 ± 3.03	0.00 ± 0.00 c
	120 h	68.20 ± 2.51 ab	0.00 ± 0.00	31.80 ± 2.51 ab
	144 h	60.88 ± 1.85 bc	0.00 ± 0.00	39.12 ± 1.85 a
	168 h	70.42 ± 1.75 a	0.00 ± 0.00	29.58 ± 1.75 b
Forastero	0 h	84.15 ± 3.40 a	15.85 ± 3.40 b	0.00 ± 0.00 c
	24 h	71.43 ± 3.65 b	0.00 ± 0.00 c	28.57 ± 3.65 a
	48 h	0.00 ± 0.00 c	83.18 ± 3.21 a	16.82 ± 3.21 b
	72 h	0.00 ± 0.00 c	83.44 ± 1.74 a	0.00 ± 0.00 c
Criollo	0 h	76.99 ± 0.20 a	0.00 ± 0.00	22.62 ± 0.42 b
	48 h	74.74 ± 0.27 a	0.00 ± 0.00	25.26 ± 0.27 b
	72 h	73.35 ± 0.43 a	0.00 ± 0.00	26.91 ± 0.07 b
	96 h	71.93 ± 0.97 ab	0.00 ± 0.00	28.10 ± 1.01 ab
	120 h	78.99 ± 0.45 a	0.00 ± 0.00	21.01 ± 0.45 b
	144 h	74.10 ± 0.46 a	0.00 ± 0.00	25.90 ± 0.46 b
	168 h	65.69 ± 6.79 b	0.00 ± 0.00	34.31 ± 6.79 a

Values within the same column with different letters are significantly different (p < 0.05). Each value in the table represents the mean ± SD of three replicates.

). The results demonstrate that fermentation time significantly influences the presence of specific polymorphs (p < 0.05). Form I (γ) was the most predominant (53.74–84.15%) and is considered the initial crystalline structure formed upon cooling of the lipid fraction extracted from the bean [15,84]. In the Forastero variety (Brazil), CB initially exhibited Form I, which subsequently transitioned into more stable forms (α and β′₂), a phenomenon known as polymorphic transformation [84,93]. Notably, the presence of Form II in CB under short fermentation conditions (Brazil) corresponds to a kinetically favorable but thermodynamically unstable polymorph [86].

Form III (β’₂) was more predominant in the Criollo variety (Peru and Mexico). During fermentation, Form I decreased while Form III increased, especially between the third and fifth days of fermentation. This shift was associated with a rise in melting point (>20 °C) [87,93,94,95]. According to Table 3, Form III ranged from 21.0 to 39.1% in cocoa butter (CB) from the Criollo variety (Peru and Mexico), and reached 28.5% in the Forastero variety (Brazil). This was possibly due to static CB conditions (untempered samples), evidencing the presence of a considerable quantity of metastable polymorphs [16,87].

Furthermore, fermentation times shorter than seven days were associated with an increased presence of thermally stable polymorphs [31]. The polymorphic structures found in CB extracted from raw, unprocessed beans indicate that Form V—which is considered desirable in the chocolate industry—had not yet been reached. Achieving this form requires the tempering phase, which promotes the formation of over 90% of this stable polymorph [16,88,96].

3.3. Spectral Characterization and Multivariate Assessment Using ATR FT-IR

Based on the FT-IR spectra, four distinct spectral regions were identified according to bond types (Figures S4–S6). Stretching vibrations of single bonds (O–H, C–H, and N–H) were detected between 2500 and 4000 cm⁻¹, double bonds within the 1500–2000 cm⁻¹ region, and triple bonds between 2000 and 2500 cm⁻¹. In the region spanning 650–1500 cm⁻¹, complex vibration patterns appeared, useful for molecular identification; this region is commonly referred to as the fingerprint region [97]. The fingerprint region is associated with the position and intensity of absorption bands that are strictly representative of individual organic compounds [98]. Within this region, an absorption band at 715.6 cm⁻¹ (Figure 3a) was identified; this peak intensity allowed the differentiation of CB samples from various origins, as observed in Criollo Nativo cocoa (Peru) compared to other varieties [99]. Additionally, an absorption band at 723.1 cm⁻¹ was noted for CB samples from Brazil and Mexico (Figure 3b,c); this band is attributed to the overlapping of CH₂ rocking vibrations and out-of-plane bending of cis-disubstituted olefins [100,101].

Further peaks were observed at 872.2, 887.1, and 894.6 cm⁻¹ with low intensity (A.U.), corresponding to in-plane oscillatory vibrations of =CH groups. Antony et al. [102] and Deus et al. [103] identified an intense band region between 1260 and 1000 cm⁻¹ attributed to C–O stretching vibrations in the O–C–C fragment of esters. In this study, during the spontaneous fermentation of the three CB varieties, peaks at 1115 and 1170 cm⁻¹ were detected. Moreover, a unique band at 1245 cm⁻¹ was found exclusively in Criollo Nativo CB (Peru), associated with v(C–H) stretching in carbohydrates and v(C–O) stretching in acids [104], indicating the presence of free fatty acids (FFAs). Moreover, bands at 1372 and 1379 cm⁻¹, related to CH₃; bending and lipid presence in CB, were also identified [101,105]. Additionally, bands at 1461 and 1469 cm⁻¹ within the fingerprint region corresponded to methylene (CH₂) functional groups and C–O stretching [99,106]. Finally, various peaks in the fingerprint region provided evidence of simple lipids (UFA esters), compound lipids (simple lipids conjugated with non-lipid molecules), and fatty acids [106].

In the double-bond region, absorption bands between 1800 and 1700 cm⁻¹ were identified, corresponding to carbonyl ester stretching vibrations $v$(C=O) [107,108]. During the fermentation of Criollo Nativo cocoa butter (CB), an absorption band at 1737 cm⁻¹ was observed (Figure 3a), indicating a stretching vibration associated with the presence of linoleic acid [109,110]. In contrast, CB samples from Brazil and Mexico showed an absorption band at 1744 cm⁻¹ (Figure 3b,c), which corresponds to the C=O stretching of esters and free fatty acids (FFAs). This band is associated with the conformational state of the glycerol backbone in TAGs found in foods such as CB [100,102,111,112]. According to Bresson et al. [107], Goodacre & Anklam [113], and Lucarini et al. [114], both bands (1737 and 1744 cm⁻¹) are correlated with the stretching vibration of carboxyl groups in TAG structures.

In the triple-bond region (Figure 3), peak intensities were mainly low. For instance, absorption bands at 2102 and 2110 cm⁻¹ were attributed to terminal alkyne C≡C groups (monosubstituted), while the band at 2117 cm⁻¹ corresponded to medial alkyne C≡C groups (disubstituted) [115]. Within the single-bond region, low-intensity bands between 3300 and 2500 cm⁻¹ were identified, associated with O–H stretching vibrations typical of carboxylic acids [109]. In addition, the spectral zone between 3200 and 2700 cm⁻¹ was linked to $v$(C–H) stretching modes [107,108], which reflect lipid content [116]. Notable peaks at 2855, 2915, and 2922 cm⁻¹ were consistently observed during the fermentation of the three CB varieties, corresponding to symmetric C–H stretching of CH₂ and CH₃ groups [102,110,114]. These peaks illustrate intra- and interchain interactions among TAG molecular structures [112]. Finally, low-intensity signals were reported within the 3550–3200 cm⁻¹ range. This signal is associated with O–H stretching vibrations due to the presence of intermolecular hydrogen bonding [103,105].

An exploratory analysis was performed using all spectral data after baseline correction (Figure 4). The hierarchical cluster analysis (HCA) dendrogram revealed three distinct groups, corresponding to the spontaneous fermentation processes of CB from the three studied varieties, indicating chemical differentiation among them (Figure 4a). The right branch grouped CB samples from the Criollo variety cultivated in Peru and Mexico. In contrast, samples from Brazil (Forastero) formed a separate cluster. These HCA results aligned with the principal component analysis (PCA), in which PC1 and PC2 jointly explained 94% of the total data variance (Figure 4b,c). Moreover, when fermentation time was considered by country, PCA revealed clearer group formation, not only based on genetic variety—as suggested by HCA—but also in relation to geographic origin. This confirms that CB chemical composition is influenced by variety, country of origin, and postharvest handling. Additional factors include environmental conditions, plantation age, and pod maturity [28], which—consistent with observations in Figure 1 and Figure 2—affect the profile of fatty acids and triglycerides [117].

3.4. Characterization of Cocoa Butter by Nuclear Magnetic Resonance (NMR)

3.4.1. ¹H NMR Spectra Acquired from Cocoa Butter of Three Varieties

Cocoa butter (CB) is primarily composed of a mixture of unsaturated and/or polyunsaturated fatty acids at the sn-2 position and saturated fatty acids at the sn-1 and sn-3 positions of the glycerol backbone. This structural arrangement results in the formation of three dominant TAG species in CB: POP, POS, and SOS [42,118,119]. As illustrated in Figure 5, the ¹H NMR protonic profiles of CB from the three varieties remained consistent throughout the spontaneous fermentation (Figures S7–S9), aligning with findings reported by Colella et al. [42], where signal assignments for CB samples (Figure 5) exhibited similar patterns to those observed in other vegetable oils [120,121]. The ¹H spectral data were validated using the 2D ¹H COSY spectrum (Figure 6), which revealed all expected correlations for TAGs containing saturated (SFAs) and unsaturated fatty acids (UFAs). According to the homonuclear COSY map, overlapping signals with minor chemical shift variations were observed. Accordingly,

Table 4. ¹H NMR chemical shifts (δH, ppm) and ¹H/¹H proton correlations of fatty acids in TAGs in CDCl₃ for cocoa butter.

Position	δH, Multiplicity [J in Hz]	COSY
A	2.29 td (7.5)	B
B	1.56–1.62 m	A,C
C	1.23–1.32 m	B,D,E
D	0.86 t (6.78)	C
E	1.96–2.04 m	C,F
F	5.31–5.34 m	E,G
G	2.73–2.77 t (6.5)	F
H, L (Gly)	4.13 dd (11.9, 5.9)	H’,L’,I
I (Gly)	5.22–5.27 m	H,L’,L,L’
H’, L’ (Gly)	4.28 dd (11.8)	H,L,I

Abbreviations: dd—doublet of doublets; t—triplet; m—multiplet; Gly—glycerol. Signal multiplicity was indicated when identifiable, and experimental coupling constants (J, Hz) were extracted from the spectrum. The letters shown below, representing protons from various functional groups of fatty acids (FAs) in triglycerides (TAGs), refer to the labeling scheme described by Colella et al. [42,122].

reports the corresponding multiplets (m) along with their chemical shift ranges.

Table 4 summarizes all proton signals related to the spectra presented in Figure 5 (¹H NMR) and Figure 6 (¹H–¹H COSY) for cocoa butter (CB). The signal at 0.86 ppm is attributed to the overlap of the methyl proton triplet (t) from acyl groups (LA) [42,121,122]. The doublet of doublets (dd) signal at 4.13 ppm corresponds to the 1′a, 3′a–CH₂–OCO– functional group, indicating the presence of the sn-1,3 DAG compound [123]. Additionally, the chemical shift ranging from 1.56 to 1.62 ppm (multiplet, m) denotes acyl group protons [120], while shifts between 5.22 and 5.27 ppm (m) are associated with the glycerol backbone [120]. Finally, the signals at 4.13 and 4.28 ppm reflect glycerol protons in TAGs of the lipid fraction in CB [42,123].

3.4.2. Analysis of ¹³C NMR Spectra in Cocoa Butter

Figure 7 presents the carbon signals corresponding to various positions within triglycerides (TAGs) (

Table 5. ¹³C NMR chemical shifts (δ¹³C, ppm) in CDCl₃ for cocoa butter (CB).

Triglycerides			δ 13C ppm
		Palmitic Acid	Stearic Acid	Oleic Acid	Linoleic Acid
C-1	sn 1–3
	sn 2			172.8	172.8
C-2	sn 1–3	34.0	34.0	-	-
	sn 2	-	-	34.2	34.2
C-3	sn 1–3	24.9	24.9	-	-
	sn 2	-	-	24.9	24.9
C-4	sn 1–3	29.1	29.1	-	-
	sn 2	-	-	-	-
C-5	sn 1–3	29.3	29.3	-	-
	sn 2	-	-	29.2	-
C-6	sn 1–3	29.5	29.5	-	-
	sn 2	-	-	29.1	29.1
C-7	sn 1–3	29.7	-	-	-
	sn 2	-	-	29.8	-
C-8	sn 1–3	29.6–29.8		-	-
	sn 2			27.2	27.21
C-9	sn 1–3	29.6–29.8		-	-
	sn 2			129.7	-
C-10	sn 1–3	29.6–29.8		-	-
	sn 2			130.0	126.7
C-11	sn 1–3	29.6–29.8		-	-
	sn 2			27.2	-
C-12	sn 1–3	29.6–29.8		-	127.9
	sn 2			29.8	-
C-13	sn 1–3	29.4	-	-	-
	sn 2	-	-	29.4	130
C-14	sn 1–3	31.9	-	-	-
	sn 2	-	-	29.5	27.2
C-15	sn 1–3	22.7	29.4	-	-
	sn 2	-	-	29.3	29.4
C-16	sn 1–3	14.1	31.9	-	-
	sn 2	-	-	31.9	31.6
C-17	sn 1–3	-	22.7	-	-
	sn 2	-	-	22.7	-
C-18	sn 1–3	-	14.1	-	-
	sn 2	-	-	14.1	-

The ¹³C NMR results reflect chemical shift values detected in all cocoa butter (CB) samples from the three countries. These values clearly demonstrate that the chemical shifts (δ¹³C, ppm) of fatty acids in CB depend significantly on the positional arrangement of the acyl chain within the glycerol backbone (denoted as sn-1,2,3). Carbon atom numbering was assigned according to the most precise structural representation and labeling system (IUPAC nomenclature) for the major and most abundant fatty acids in natural CB, as reported in the publication by Colella et al. [42].

), revealing identical spectral profiles among the three cocoa butter (CB) varieties during fermentation (Figures S10–S12), in line with findings reported by Colella et al. [42]. Distinct regions of the spectrum exhibited two chemical shift values for the same carbon (C) of the fatty acids (FAs), depending on the positional arrangement of the acyl chain within the TAG molecule. This variation is attributed to the distribution of the three major TAG species prevalent in CB—POS, SOS, and POP [119,124,125]. Furthermore, our results corroborate previous observations by Colella et al. [42], who identified chemical shifts between 34.0 and 34.2 ppm for C-2 carbons, 126.7 and 130.0 ppm for olefinic carbons in oleic and linoleic acids, and 172.8 and 173.3 ppm for the carbonyl C-1 positions.

3.4.3. Iodine Value (IV) and Saponification Value (SV)

Table 6. Iodine and saponification values for cocoa butter (CB) from ¹H−NMR.

Variety	Fermentation Time	Iodine Value (g I2/100 g CB)	Saponification Value (mg KOH/g CB)
Criollo Nativo	0 h	31.0762	190.5300
	24 h	31.2386	160.9245
	48 h	31.4988	188.6276
	72 h	29.5908	146.3459
	96 h	32.8891	182.6926
	120 h	30.4913	183.4519
	144 h	32.1449	180.9272
	168 h	31.6446	185.6525
Forastero	0 h	18.2358	66.3548
	24 h	19.3730	64.5709
	48 h	28.9005	184.8556
	72 h	24.4219	136.4746
Criollo	0 h	31.5657	188.5245
	48 h	31.8905	184.4995
	72 h	31.5402	184.1428
	96 h	31.9401	186.9054
	120 h	30.8025	179.8106
	144 h	31.8358	189.0276
	168 h	16.4199	186.1338

reports the iodine value (IV) determined from the ¹H NMR spectra, which, according to Darmawan et al. [126] and Duodu et al. [127], reflects the relative degree of unsaturation in fats and oils—an indicator of susceptibility to rancidity. For cocoa butter (CB) samples from Peru and Mexico (Criollo), the IV ranged between 29.5 and 32.6 g I₂/100 g of CB, consistent with previously reported values for cocoa butter (33–42 g I₂/100 g) found in studies by Darmawan et al. [126], Jahurul et al. [128], Mounjouenpou et al. [129], and Norazlina et al. [130]. The IV for Brazilian CB (Forastero) was the lowest, ranging from 18.2 to 28.9 g I₂/100 g, which is associated with lower content of unsaturated fatty acids (UFAs) and greater oxidative resistance—possibly linked to antioxidant pigments responsible for the yellowish hue of CB [39,129,130]. Although the IV fluctuates during fermentation processes, all values remain within reported thresholds for this food type, indicating resistance to rancidity, photooxidation, metal-catalyzed oxidation, and thermal/atmospheric degradation [127].

The saponification value (SV) is a quality indicator for fats and oils, expressed as the amount of alkali (mg KOH/g of sample) required to saponify a given quantity of material [131,132]. Similar to the iodine value (IV), the SV for cocoa butter (CB) of the Criollo variety (Peru and Mexico) ranged from 179.8 to 190.5 mg KOH/g during fermentation. This result is associated with the presence of fatty acids with higher molecular weight [127]. In contrast, the SV for CB of the Forastero variety (Brazil) started at 64 mg KOH/g at the beginning of fermentation, peaking at 184.8 mg KOH/g on the second day. All SVs fall within the previously reported threshold for cocoa butter, which is less than 198 mg KOH/g [126,129]. These values are primarily attributed to the presence of long-chain fatty acids (e.g., C18 and C16) [132].

4. Conclusions

The predominant fatty acids (FAs) in cocoa butter (CB) in the three varieties were stearic (Forastero), oleic (Criollo), and palmitic (Criollo Nativo) acids. Additionally, trace amounts of myristic acid—characteristic of cocoa—were detected, contributing to the stable molecular packing of CB. Together with pentadecanoic acid, these compounds are involved in the final flavor profile of chocolate. Multivariate analysis revealed that clear differences in the spontaneous fermentation processes among countries and short fermentation periods were found to promote the formation not only of these three major FAs but also of linoleic and arachidic acids.

FT-IR spectroscopy in the CB during spontaneous fermentation revealed the chemical nature of simple and complex lipids, free fatty acids, and carboxyl groups in TAGs, and symmetric C–H stretching of CH₂ and CH₃ groups. Nuclear magnetic resonance (NMR) spectroscopy proved to be a powerful technique for elucidating chemical modifications in cocoa butter (CB), complementing the FT-IR spectroscopy results. The use of tools for spectroscopy in this study (IR and NMR) demonstrated that CB is notable for its characteristic composition of glycerolipids (mainly TAGs), di- and monoglycerides, and free fatty acids, among others.

In the crystallization kinetics of cocoa butter (CB), short fermentation periods (3–5 days) promoted the development of needle-like, disc-like, and spherulitic crystals, indicating a mixed nucleation and growth mechanism. Likewise, the fermentation time significantly influenced the polymorphic forms identified, showing that Form I (γ), considered the initial crystalline structure of CB, was predominant during fermentation, though it declined between days 3 and 5. Additionally, Form III (β′₂) was detected, with greater prevalence in the Criollo varieties.

This study highlighted the importance of the fermentation process of cocoa cultivated in Latin American countries in relation to the composition and thermal properties of cocoa butter (CB). It confirmed that fermentation time influences CB stability, where metastable forms (α and β′) are initially generated and subsequently transformed, through chocolate processing, into the stable Form V, which is responsible for high-quality chocolate. This transformation is also associated with the composition of CB, particularly the presence of fatty acids such as stearic, palmitic, and oleic acids, as well as triglycerides (POP, POS, and SOS).