Optimization of Mono- and Disaccharide Extraction from Cocoa pod Husk

Abstract

Cocoa pod husk (CPH) is a potential material to produce value-added products. The objective of this study was to optimize the microwave-assisted hydrothermal pretreatment (MA-HTP) of CPH and CPH hemicellulose (HMC-CPH) using only water as the extraction medium, in combination with response surface analysis (RSA), Box–Behnken design (BBD), and proton nuclear magnetic resonance identification and quantification (¹H NMR Qu) to provide an efficient protocol for the extraction of mono- and disaccharides, as a novel method for which no precedent was found. The methodology consisted of 15 CPH MA-HTPs and 15 HMC-CPH MA-HTPs (triplicate) designed by RSA-BBD; the experimental variables were time, temperature, and power, and the response was the concentration of extraction products. Glucose, sucrose, and fructose were identified as products of the extractions by ¹H NMR. With 95% confidence, higher sucrose content was determined for CPH (45.62%) compared to HMC-CPH (17.34%), high fructose content for both CPH and HMC-CPH (37.88% and 35.37%, respectively), and minimal glucose concentrations were obtained in both CPH and HMC-CPH (4.57% and 0.93%, respectively). Using RSA-BBD, optimal temperature, power, and time points were predicted for glucose CPH: 135.4 °C, 180.6 W, and 5.8 min; sucrose: 154.3 °C, 256.3 W, and 20. 2 min; fructose 129.5 °C, 173.8 W, and 5.27 min. For HMC-CPH, the optimal conditions were as follows: glucose: 142.2 °C, 204.4 W, and 10.5 min; sucrose: 148.8 °C, 215.6 W, and 14.3 min; fructose: 151.6 °C, 231.6 W, and 13 min.

1. Introduction

The cocoa industry generates large amounts of waste; in 2024, cocoa bean production was around 4.3 million tons [1]. Cocoa beans make up about 10% of the total weight of the fruit and are used to make chocolate [2]. The cocoa pod husk (CPH) is the main by-product of the cocoa industry, accounting for approximately 75% of the total weight of the fruit [3,4], with annual estimates of 48 million tons of CPH worldwide [5]. Management of these residues is costly and complex, and they generally remain on the land, causing foul odors, soil contamination, greenhouse gas emissions, and excessive growth of pathogenic fungi that cause diseases that affect crop production [6,7]. However, it is also an important and challenging renewable source for biorefining and has been the focus of research interest for several decades [8]. CPH is rich in biologically active molecules with nutraceutical properties, consisting of carbohydrates, lignin, proteins, lipids, pectin, minerals, theobromine, phenolic compounds, and tannins [4]. Sugars in CPH pectin include xylose, arabinose, rhamnose, galactose, mannose, glucose, and galacturonic acid, making it a good source for the food industry [9]. Hemicellulose is rich in xylose with neutral sugar substitutions and phenolic acid esters used in the food and pharmaceutical industries [10]. In this sense, CPH can be pretreated to separate its components such as lignin from cellulose and hemicellulose, and to obtain molecules from xylo-oligosaccharides [11]. There are several types of pre-treatment, including thermal, chemical, physical, biological, and combinations of these [12,13]. Microwave-assisted hydrothermal pretreatment (MA-HTP) is a pretreatment that combines microwave methodology with chemical reagents and is known to be more efficient than conventional heating [14]. This study proposes combining MA-HTP and ¹H NMR without the use of chemical reagents as a novel method for extracting and quantifying mono- and disaccharides from CPH and HMC-CPH. Several studies have been reported on the extraction of pectin from CPH with MA-HTP in combination with strong acids, organic acids, enzymatic methods, etc. [2,6,15,16]; the results highlight the impact on time savings, which is why it is considered a green technology [10,17]. MAE has also been used in combination with strong acids to delignify lignocellulosic materials with maximum lignin removal [14]. The efficacy of MA-HTP can be assessed using a mathematical model that predicts the statistical significance of the dependent variables and their interactions, providing optimal conditions using tools that reduce the number of experiments [18], these are response surface analyses (RSAs), which allow the variables of an experiment to be optimized to optimize a response [2]. An example is the Box–Behnken design (BBD), a mathematical model with first- and second-order coefficients, which is a three-level incomplete factorial design for three factors [19]. The BBD is slightly more efficient than the central composite design and much more efficient than the three-level full factorial designs [18,20]. Therefore, MA-HTP in combination with RSA is a technique used to extract active compounds from plant materials, where the relationship between solvent, extraction time, and irradiation power is studied [21]. On the other hand, NMR has been successfully used as a quantitative method for natural products, as all components resonate at very low concentrations, just above the detection threshold (5–10 µM) [22]; it is also over 98% accurate and is therefore considered a reliable technique for quantitative estimation. This has been established through validation procedures for precision, accuracy, linearity, reproducibility, robustness, selectivity, and specificity. However, this is only true if the sampling and processing parameters are well-known [23]. ¹H NMR spectra of carbohydrates have constant shifts concerning a reference because they are not affected by pH or ionic strength due to the absence of ionizable groups. Therefore, the acquisition parameters remain constant for each sample and a list of frequencies can be constructed [24]. To date, no publication proposes the extraction of mono- and disaccharides from CPH by MA-HTP using only water as an extraction medium, so it is feasible to seek the optimization of the process parameters, reducing time and cost using a statistical model. Therefore, the present study aims to optimize the conditions for the extraction of carbohydrates from CPH and HMC-CPH by MA-HTP, using DBB response surface analysis and ¹H NMR quantification, to make a green pretreatment for this lignocellulosic material more efficient.

2. Materials and Methods

2.1. General Methodology

In this study, CPH and HMC-CPH were used as raw materials. The species studied was Theobroma cacao L. variety Carmelo, collected at the Rancheria Rio Seco, municipality of Cunduacan [latitude: 18°7′55.90″ N, longitude: 93°18′4.49″ W] and at the farm Jesús María, municipality of Comalcalco [latitude: 18°11′0.22″ N, longitude: 93°14′28.02″ W], state of Tabasco, Mexico, at altitudes of 10 and 13 m above sea level (MASL). The CPH was dried under ambient conditions with indirect sun exposure (average maximum temperature 21 °C, average minimum temperature 6 °C, and relative humidity less than 10%). The CPH was then mechanically ground in a Thomas Wiley Model 4 laboratory mill (TP4274E70520A, Thomas Scientific, Swedesboro, NJ, USA) and the particle size was graded using a US STD 100 laboratory sieve (W. S. Tyler, Mentor, OH, USA). The CPH material used in this study had an average particle size of less than 150 µm. Finally, this material was oven-dried at 102 °C ± 3 °C to constant weight. From a fraction of this material, hemicellulose was obtained according to the methodology reported by Peng and She [25]. Finally, CPH and HMC-CPH were reserved for further processing.

2.2. Reagents

The standards and solvents used for the analysis of the pretreatments were D-fructose ≥ 99% (CAS no. 57-48-7), sucrose ≥ 99.5% (CAS no. 57-50-1), and D-glucose ≥ 99.5% (CAS no. 50-99-7), deuterium oxide 99.9% (CAS no. 7789-20-0) and 3-(trimethylsilyl) propionic acid-2,2,2,3,3-d4-acid sodium salt 98% (CAS no. 24493-21-8), all from Sigma Aldrich (Merck KGaA, Saint Louis, MO, USA).

2.3. Microwave-Assisted Hydrothermal Pretreatment of CPH and HMC-CPH

A MARS 6TM microwave digestion system (CEM Corporation, Mecklenburg, NC, USA) was used for microwave-assisted extraction of CPH and HMC-CPH. A measurement of 1 g of CPH fines less than 150 µm and 1 g of HMC-CPH were weighed and placed separately in lidded silicone cups (Xpress Plus, CEM Corporation) to which 10 mL of distilled water was added to obtain a 1:10 ratio (CPH: distilled H₂O); the solid–liquid ratio was established based on a series of preliminary tests, which showed that a reduction in water volume was insufficient to achieve efficient extraction, while an increase in water volume did not generate significant improvements in the performance of the extraction process. The CPH and HMC-CPH cups with their respective triplicates were placed in the carousel of the microwave oven programmed with a temperature, power, and time previously defined by the BBD. As a result of the pretreatment, a heterogeneous mixture of two phases with an acidic pH of 3.5 to 4.5 was obtained, consisting of a solid phase and an aqueous phase. The solid phase corresponded to fine particles of the sample that showed no visible alterations, while the aqueous phase contained the compounds extracted by the hydrothermal process. The mixture was transferred to a 50 mL conical tube (Eppendorf) and centrifuged at 3700 rpm for 40 min in a Centra CL2 centrifuge (cat. no. 426, Thermo Scientific, Needham, MA, USA). The aqueous fraction was collected and freeze-dried at −50 °C under reduced pressure of 0.045 bar in a 2.5 L Freezone Legacy freeze-dryer (Labconco Corporation, Kansas City, MO, USA), and the freeze-dried CPH and HMC extract were then analyzed by nuclear magnetic resonance (NMR).

2.4. Acquisition of ¹H NMR Spectra

NMR spectroscopy was performed on a VARIAN 600 MHz (14.1 T) Premium COMPAC spectrometer (Agilent Technologies Inc., Santa Clara, CA, USA). For the analysis of CPH and HMC-CPH extracts, analytes were prepared by dissolving 30 mg of each lyophilized extract in 600 µL of a 5 mM solution of TSP (3-(trimethylsilyl) propionic acid-2,2,2,3,3-d4-acid sodium salt) and D₂O (deuterium oxide). For this purpose, 5 mm NMR tubes were used and sonicated for 20 min to completely dilute the sample. The ¹H NMR spectra were obtained using experimental parameters designed to achieve higher resolution and quantification of carbohydrates, which were as follows: 64 scans of 32 K complex points at 25 °C, a spectral width of 16 ppm, an acquisition time of 4 s, a relaxation time of 2 s, an angle of 90°, and an acquisition time of 4 min. The water suppression scheme used was PRESAT, the presaturation during the relaxation delay was performed at the minimum power for complete water suppression, and the receiver gain was set to 30 and held constants for all spectra. Spectrum acquisition and data processing were performed according to the methodology of Del Campo et al. [22] and Hernández Bolio et al. [26]. The phase correction, baseline correction, and integration of the spectra obtained on the signals of interest were performed in MNova software, version 14.2.0-26256 (Mestrelab Research SL, Santiago de Compostela, Spain). The integral values were entered into Microsoft Excel spreadsheets for the quantification process. Quantification was carried out using the following formula (Equation (1)):

$$P\_x=I\_x/I\_std\times (N\_std)/N\_x\times M\_x/M\_std\times W\_std/W\_x\times P\_std$$

(1)

where I, N, M, W, and P are the area of the integral, the number of nuclei in the molecule, the molar mass, the gravimetric weight, and the purity of the analyte (x) and standard (std), respectively [23,27].

2.5. Response Surface Analysis, Box–Behnken Design

The RSA methodology was used to optimize the microwave-assisted hydrothermal pretreatment for carbohydrate extraction from CPH and HMC-CPH, using the Box–Behnken design to model and optimize the experimental conditions using the second-order polynomial equation (Equation (2)):

$$y=\beta o+{\sum }_{i=1}^k\beta iXi+{\sum }_{i=1}^k\beta ii{X}^{2}i+{\sum }_{i=1}^k{\sum }_{j=1}^k\beta ijXiXj+\epsilon$$

(2)

where $y$ is the dependent variable; $k$ is the sample number; $ij$ are the index numbers of the samples; $\beta o$ is the lag term; $\beta i$ is the first-order linear effect of the input factor ($Xi$); $\beta ii$ is the quadratic effect of the input factor (squared) ($Xi$); and $\beta ij$ is the linear–linear interaction effect between the input factors $Xi-Xj$ [2,14,19,21]. In the Box–Behnken design, there is a relationship between the uncoded and coded independent variables, which is shown in Equation (3).

$$X_i=(x_i-x_0)/{ ∆}_{xi}$$

(3)

where $X\_i$, $x\_i$, and $x\_0$ are coded value, natural value, and natural value at the midpoint (of the i-th independent variable) and ${ ∆}_{xi}$ is the change value of an independent variable [19].

The derivation of the equation in this case will be obtained by encoding the factors −1, 0, and 1, which allows the coefficients to be comparable and the model to be more numerically stable. The response depends on linear effects (how it changes and when a single factor is modified), quadratic effects (if there is curvature in the relationship with a maximum or minimum), and interaction effects (how the combination of two factors influences y). Finally, regression methods (least squares) are used to adjust the coefficients without collinearity based on experimental data. Three independent variables with three levels each (−1, 0, and 1) were used. The independent variables were temperature (T = 100, 150, and 200 °C), power (P = 100, 200, and 300 W), and time (t = 5, 10, and 15 min), resulting in a total of 15 experiments performed in triplicate. The series was performed in order according to the BBD, and the response variable was the quantification of carbohydrates determined by NMR (

Table 1. Box–Behnken design for three factors.

Essay	Temperature (°C)	Power (W)	Time (min)
1	−1 [100]	−1 [100]	0 [10]
2	1 [200]	−1 [100]	0 [10]
3	−1 [100]	1 [300]	0 [10]
4	1 [200]	1 [300]	0 [10]
5	−1 [100]	0 [200]	−1 [5]
6	1 [200]	0 [200]	−1 [5]
7	−1 [100]	0 [200]	1 [15]
8	1 [200]	0 [200]	1 [15]
9	0 [150]	−1 [100]	−1 [5]
10	0 [150]	1 [300]	−1 [5]
11	0 [150]	−1 [100]	1 [15]
12	0 [150]	1 [300]	1 [15]
13	0 [150]	0 [200]	0 [10]
14	0 [150]	0 [200]	0 [10]
15	0 [150]	0 [200]	0 [10]

The 15 experimental combinations performed in triplicate (45 runs). The response variable was the quantification of carbohydrates using NMR.

). To ensure the reliability and accuracy of the model using the Box–Behnken design, an analysis of variance was performed, the coefficient of determination R² was analyzed to evaluate the explanatory power of the model, and response surface graphs were generated to visualize the behavior of the model as a function of two factors at a time. Analyses were performed in R software (RStudio Inc. Version 4.2.3, Boston, MA, USA).

3. Results and Discussion

3.1. ¹H NMR Spectra Elucidation

As a result, two monosaccharides (glucose and fructose) and one disaccharide (sucrose) could be identified by proton nuclear magnetic resonance from the freeze-dried aqueous extracts obtained from microwave-assisted hydrothermal treatments of cocoa pod husk, as well as from the hemicellulose extracted from the same.

Table 2. Characteristics of the ¹H NMR signals observed in MA-HTP of CPH and HMC-CPH.

D-Fructose, Sucrose, and D-Glucose Standards			CPH and HMC-CPH MA-HTP Extracts
Assignment	Chemical Shifts Used for Identity Check		Chemical Shift
	δ (ppm)	Multiplicity	Identity Check δ (ppm)	Quantity δ (ppm)	Multiplicity
Fructose	4.11	d		4.11	d
Fructose	4.01	t		4.01	t
Fructose	3.89	dd	3.90		m
Sucrose	5.40	d		5.40	d
Sucrose	4.21	d		4.20	d
Sucrose	4.05	t	4.05		t
Sucrose	3.76	t	3.76		t
Sucrose	3.67	s	3.67		s
Sucrose	3.55	dd	n/d		n/d
Sucrose	3.47	t	3.47		t
Glucose	5.24	d		5.24	d
Glucose	4.65	d		4.65	d
Glucose	3.89	dd	n/d		n/d
Glucose	3.53	dd	3.53		dd
Glucose	3.25	t	3.25		t

s, singlet; d, doublet; t, triplet; m, multiplet; dd, doublet–doublet.

shows the chemical shifts, and the multiplicity of signals identified in the experimental ¹H NMR spectra of MA-HTP of CPH and HMC-CPH, as well as in the spectra of the D-fructose, sucrose, and D-glucose standards.

As can be seen in Table 2, the identification of the monosaccharides and disaccharides was supported by the spectra of the standards as well as by data from the literature. Three signals were identified for fructose elucidation and only two for quantification. In contrast, six signals were identified for sucrose elucidation, two of them for quantification, as well as two of the four signals identified for glucose quantification. In this respect, del Campo et al. [22] report, for fructose from honey samples, two signals identified as multiplets at the same chemical shifts reported in this study (4.00 ppm and 4.10 ppm), but with different multiplicity. On the other hand, for sucrose, they report one signal with a doublet multiplicity at the chemical shift of 5.42 ppm, which coincides with the same signal reported in this study. Finally, they reported five signals for glucose; for β-glucose, one doublet and two doublets of doublets were identified at chemical shifts of 4.65 ppm, 3.40 ppm, and 3.25 ppm, respectively. In contrast, for α-glucose two signals, one doublet and one doublet of doublets were identified at chemical shifts of 5.22 ppm and 3.42 ppm, respectively. Four of the five signals identified in this study, two for β-glucose and two for α-glucose, were coincident, with one doublet of doublets missing at the chemical shift of 3.40 ppm corresponding to β-glucose. This could be because the H-2 proton of β-glucose is usually coupled with H-1 and H-3. If H-1 and H-3 have similar coupling constants or if there is overlap with other signals from the ring, which is very likely, the double pattern of doubles may not be evident. Another recurring reason is that β-glucose may be in equilibrium with other forms, such as α-glucose or open forms, which can cause exchange between conformations, averaging the couplings of the signals until the double–double pattern is lost. On the other hand, Al-Mekhlafi et al. [28] reported six, five, and two signals, respectively, at chemical shifts 4.08 ppm, 4.00 ppm, 3.93 ppm, 3.79–3.85 ppm, 3.69 ppm, and 3.51–3.57 ppm for fructose, 5.39 ppm, 4.16 ppm, 4.07 ppm, 3.87–3.67 ppm, and 3.53–3.42 ppm for sucrose, and 5.19 ppm and 4.59 ppm for glucose, with differences of about 0.04–0.12 ppm for most chemical shifts. Finally, agreement was also found with the data of Spiteri et al. [29]. They identified a signal for fructose with a chemical shift of 4.1 ppm, a signal for sucrose at 4.22 ppm, and two signals for glucose at 5.23 ppm and 3.24 ppm, corresponding to α-glucose and β-glucose, respectively, but did not report signal multiplicity data. Identification of sugars by ¹H NMR is limited because oligosaccharides incorporated into an isolated spin system per monomer unit are separated from each other by glycosidic bonds and, considering that oligosaccharides can be formed from the same monomers and in the same sequences, they will overlap in the spectra so that only a few sugars can be identified [24].

According to the PubChem databases (CID: 5984), fructose is a monosaccharide found in fruits and honey, soluble in water, ether, and alcohol, and is mainly used as a preservative and as an intravenous infusion in parenteral nutrition. Sucrose (PubChem CID: 5988), a glycosyl-glycoside composed of glucose and fructose units, is used as an osmolyte, food sweetener, human metabolite, etc., and is mainly obtained from sugar cane and sugar beet. Finally, according to the PubChem databases (CID: 5793), D-glucose or dextrose, the most abundant isomer of glucose in nature, is produced by photosynthesis in plants and by hepatic gluconeogenesis in humans. On the other hand, a study by Perwitasari et al. [8] hydrolyzed CPH to monosaccharides, which were then metabolized by A. niger via the tricarboxylic acid cycle to citric acid, which could also be used in the food industry.

3.2. Optimization of CPH and HMC-CPH MA-HTP Conditions Through RSA-BBD

The ¹H NMR Qu results were used to optimize the MA-HTP of CPH and HMC-CPH using BBD response surface analysis. The BBD indicates the optimal points of the factor’s temperature, power, and time in the MA-HTP of CPH and HMC-CPH. The three factors were coded as −1, 0, and 1 for low, medium, and high levels, respectively. This study presents the optimal level for each dependent variable, as well as the quadratic polynomial equations and three-dimensional graphical representations.

Table 3. BBD model for MA-HTP optimization of CPH and HMC-CPH by ¹H NMR Qu.

Independent Variables			Dependent Variable Y
Coded −1,0,1 [Uncoded]			Yield CPH			Yield HMC-CPH
Temperature (°C)	Power (W)	Time (min)	Glucose (%)	Sucrose (%)	Fructose (%)	Glucose (%)	Sucrose (%)	Fructose (%)
−1 [100]	−1 [100]	0 [10]	1.0	51.10	29.90	1.2	7.70	16.91
1 [200]	−1 [100]	0 [10]	1.0	43.17	27.80	0.4	12.00	25.16
−1 [100]	1 [300]	0 [10]	9.4	68.02	70.26	2.2	26.23	44.78
1 [200]	1 [300]	0 [10]	1.0	39.20	27.07	0.6	8.67	23.87
−1 [100]	0 [200]	−1 [5]	1.2	50.39	29.56	0.5	8.15	20.20
1 [200]	0 [200]	−1 [5]	0.8	43.38	26.87	1.3	17.19	27.49
−1 [100]	0 [200]	1 [15]	0.7	39.41	23.51	1.1	20.90	35.43
1 [200]	0 [200]	1 [15]	1.1	37.91	28.08	1.7	26.92	49.61
0 [150]	−1 [100]	−1 [5]	0.5	34.63	20.94	1.4	13.98	21.58
0 [150]	1 [300]	−1 [5]	1.1	36.78	29.23	2.3	25.76	47.07
0 [150]	−1 [100]	1 [15]	7.5	52.79	57.48	2.8	16.42	41.23
0 [150]	1 [300]	1 [15]	14.0	61.78	88.57	1.8	22.96	44.31
0 [150]	0 [200]	0 [10]	1.0	42.15	27.91	7.8	35.80	75.56
0 [150]	0 [200]	0 [10]	0.8	35.13	23.10	2.3	28.57	51.86
0 [150]	0 [200]	0 [10]	1.4	43.38	31.83	2.1	17.45	34.10

shows the ¹H NMR Qu of carbohydrates; glucose, sucrose, and fructose were obtained from 15 MA-HTP of CPH and 15 MA-HTP of HMC-CPH.

3.3. RSA—BBD

The application of the Box–Behnken statistical design in this study allowed us to model the influence of experimental parameters on the extraction of mono- and disaccharides from cocoa pod husks. The coefficients of determination obtained were R² = 0.7426, 0.7648, and 0.7090 for glucose, sucrose, and fructose in CPH, respectively, and R² = 0.7749, 0.7882, and 0.7904 for the same sugars in HMC-CPH. Although it is recognized that a lack of adjustment could compromise the accuracy of the predictions, the model exhibits acceptable predictive power, explaining more than 70% of the observed variability. This level of performance is considered adequate, given the structural complexity of lignocellulosic biomass, the variability inherent in thermal extraction processes, and uncontrolled environmental conditions.

The concentration of glucose, sucrose, and fructose extracted from CPH and HMC-CPH by MA-HTP were determined in percent by ¹H NMR Qu for each treatment designed by BBD. Table 3 shows an apparent difference in mono- and disaccharide concentrations between the CPH and HMC-CPH whole matrix; however, by analysis of variance only a statistically significant difference (p ≤ 0.05) was found between the sucrose concentrations of CPH and HMC-CPH, with the highest concentration being the sucrose content extracted from cocoa pod shells.

On the other hand, a significant statistical difference (p ≤ 0.05) was found between the concentrations of sucrose, fructose, and glucose extracted from CPH, given the variables of time and power. With 95% confidence, two groups of means were identified for sugar concentrations, two groups for time, and two groups for power, finding sucrose and fructose from CPH in the same group with a mean of 45.62% and 37.88%, respectively, being the sugars with the highest concentration in cocoa pod husk, while power of 300 W and time of 15 min are identified with the highest extraction of sugars, 37.69% and 36.08%, respectively. Similarly, a statistically significant difference (p ≤ 0.05) was obtained between sucrose, fructose, and glucose concentrations of HMC-CPH; for sugars extracted from hemicellulose, temperature and power were significant. Using analysis (Tukey), two groups were identified for temperature and two groups for power, with the temperature of 150 °C giving the highest concentration of sugars with a mean of 23.37% and the power of 200 to 300 W with mean concentrations of 20.78% and 20.77% of sugars, while for the concentration of mono- and disaccharides three groups were obtained, the highest concentration being fructose, followed by sucrose and finally glucose with mean values of 35.37%, 17.34%, and 0.93%, respectively. Minimal glucose content was found in both CPH and HMC-CPH matrices, which could indicate the absence or minimal depolymerization of cellulose. The higher glucose content can be attributed to free sugars reported in the literature as an extractable fraction of the lignocellulosic matrix [4,5]. The hemicellulose fraction extracted from the cocoa pod shell is predominantly rich in fructose, while the whole cocoa shell matrix is rich in sucrose and fructose [30]. The results obtained are consistent with studies on microwave-assisted thermal hydrolysis in lignocellulosic matrices, which have shown that higher energy intensity favors the breakdown of glycosidic bonds and the release of reducing sugars. It has also been reported that sucrose is probably the predominant sugar in cocoa agro-industrial residues, especially in the parenchymatous fraction of the husk, while the low concentration of glucose detected in both matrices could be explained by limited or even zero depolymerization of cellulose, or by its low presence as free sugar [31,32]. According to Vegas Niño et al. [33], cellulose requires more intensive pretreatments to release glucose efficiently due to its highly recalcitrant crystalline structure. In contrast, hemicellulose, which is more amorphous and accessible in nature, allows for more efficient release of fructose and sucrose under moderate thermal conditions.

Finally, these findings confirm that the complete CPH matrix has an abundant profile of free sugars and accessibility to microwave pretreatment using water as a solvent, with relevant implications for its use in biotechnological valorization processes.

Regarding the identification of optimal conditions for sugar extraction from lignocellulosic cocoa matrices, we used the Box–Behnken design of experiments (BBD), considering critical variables such as temperature, power, and treatment time. The approach allowed modeling and predicting the behavior of glucose, sucrose, and fructose extraction based on the evaluated operating parameters.

Table 4. Optimum values provided by RSA-BBD for glucose, sucrose, and fructose from CPH and HMC-CPH.

Source	Carbohydrate	Threshold (0.01)
		Temperature (°C)	Power (Watt)	Time (Min)
CPH	Glucose	135.4	180.6	5.8
CPH	Sucrose	154.3	256.3	20.2
CPH	Fructose	129.5	173.8	5.2
HMC-CPH	Glucose	142.2	204.4	10.5
HMC-CPH	Sucrose	148.8	215.6	14.3
HMC-CPH	Fructose	151.6	231.6	13.0

presents the optimal results obtained for each sugar, both in the complete matrix of the cocoa pod husk (CPH) and in its hemicellulose fraction (HMC-CPH), under the experimental conditions defined by the BBD.

As can be seen, the predicted optimum temperatures, power, and time for the extraction of glucose, sucrose, and fructose from CPH range between 129.5 °C and 154.3 °C, between 173.8 W and 256.3 W, and between 5.27 min and 20.2 min, with higher conditions for sucrose and lower for fructose. These are contrary to the conditions proposed for the extraction of glucose, sucrose, and fructose extracted from HMC-CPH, which ranged between 142.2 °C and 151.6 °C, power between 204.4 and 231.6 W, and times between 10.5 min and 13 min, with central values prevailing in all cases. According to the results obtained in this study, glucose can be extracted at lower temperatures and in shorter times. This molecule has a more stable structure and slower degradation kinetics, making it more resistant in controlled thermal processes and minimizing the formation of unwanted by-products. On the other hand, more aggressive conditions were required for the efficient separation of fructose and sucrose. Sucrose, being a disaccharide, must be hydrolyzed into glucose and fructose before participating in thermal reactions, which requires more energy and time. However, these molecules undergo faster decomposition reactions at high temperatures, so controlled conditions must be maintained.

The coded factor equations can predict the response to determine the levels of each factor, with high levels coded as 1, low levels as −1, and intermediate levels coded as 0. Positive represents the synergistic relationship, while negative represents the antagonistic relationship between the variables [30]. Uncoded quadratic polynomial equations of the independent variables to explain the efficacy of MA-HTP, constructed by multiple regression analysis of the BBD matrix for all dependent variables (Equations (4)–(9)), are presented.

$$\begin{array}{cc}CPH glucose=\hfill & -3.8 +0.223 T- 0.0843 P - 1.24 t -0.000558 T\times T+0.000343 P\times P+0.0512 t\times t\hfill \\ & -0.000420 T\times P+0.00080 T\times t+0.00295 P\times t\hfill \end{array}$$

(4)

$$\begin{array}{cc}\hfill HMC-CPH glucose& \\ & =-25.2+0.241T+0.0577P+1.04 t-0.000778 T\times T 0.000102 P\times P-0.0388 t\times t\hfill \\ & -0.000040 T\times P-0.00020 T\times t-0.00095 P\times t\hfill \end{array}$$

(5)

$$\begin{array}{cc}CPH sucrose=\hfill & 134.8-0.765 T-0.265 P-3.11 t+0.00269 T\times T+0.001044 P\times P+0.113 t\times t-0.00104 T\times P\hfill \\ & +0.0055 T\times t+0.0034 P\times t \hfill \end{array}$$

(6)

$$\begin{array}{cc}\hfill HMC-CPH sucrose& \\ & =-127.9+1.160 T+0.475 P+2.67 t-0.00302 T\times T-0.000607 P\times P-0.057 t\times t\hfill \\ & -0.001093 T\times P-0.0030 T\times t-0.00262 P\times t \hfill \end{array}$$

(7)

$$\begin{array}{cc}CPH fructose=\hfill & 10+0.88 T-0.371 P-4.97 t-0.00218 T\times T+0.001660 P\times P+0.194 t\times t-0.00205 T\times P\hfill \\ & +0.0073 T\times t+0.0114 P\times t\hfill \end{array}$$

(8)

$$\begin{array}{cc}\hfill HMC-CPH fructose& \\ & =-236+2.14 T+0.816 P+6.48 t-0.00631 T\times T-0.001040 P\times P-0.196 t\times t\hfill \\ & -0.00146 T\times P+0.0069 T\times t-0.0112 P\times t\hfill \end{array}$$

(9)

The model is significant (p ≤ 0.05) for sucrose and fructose from CPH; the response surface curves are shown in Figure 1D–I. The model is significant (p ≤ 0.05) for fructose from HMC-CPH; the response surface curves are shown in Figure 2G,H. Regarding the calculated coefficient of determination, it was found that there was no lack of fit (p > 0.05) for the CPH models for glucose, sucrose, and fructose, and for HMC-CPH for sucrose.

Figure 1 shows the three-dimensional valley response plots of the effect of the variable’s temperature, power, and time of CPH MA-HTP; these follow a minimum stationary point pattern indicating that the response (carbohydrate yield) increases with central values of temperature and power and a low level of time for CPH glucose concentration. However, for CPH sucrose, the yield is maximized at high power and time and a central value of temperature. For CPH fructose, the yield is maximized with central values of temperature and power and a low level of time.

The optimal extraction temperatures of 129.5 °C and 154.3 °C correlate with nominal powers of 173.8 W and 256.3 W for the CPH of glucose, sucrose, and fructose, respectively. Extraction times range from 5.27 min to 20.2 min for average yields of 4.57% for glucose (Figure 1A–C), 45.62% for sucrose (Figure 1D–F), and 37.88% for fructose (Figure 1G–I). The extraction yield is indicated on the Y axis.

The three-dimensional response surface plots in Figure 2 follow a maximum stationary point type pattern, with optimum yields of mono- and disaccharides increasing with a central value of temperature, power, and time for HMC-CPH glucose, and central values of temperature and power with a high level of time for maximum yields in HMC-CPH sucrose, and a central value of temperature and high levels of power and time required for optimum values of HMC-CPH fructose.

The optimal extraction temperatures for HMC-CPH carbohydrates range from 142.2 °C to 151.6 °C, and the optimal power for these carbohydrates ranges from 204.4 W to 231.6 W for HMC-CPH glucose, sucrose, and fructose. Extraction times range from 10.5 to 14.3 min for average yields of 0.931% for glucose (Figure 2A–C), 17.34% for sucrose (Figure 2D–F), and 35.37% for fructose (Figure 2G–I).

As can be seen in Figure 1, for the optimization of glucose from CPH, an upward trend in temperature is observed, suggesting that the extraction of this sugar is favored at high temperatures. On the other hand, the effect of power is less linear, identifying an optimal intermediate zone to maximize extraction and, finally, with moderate increases in time, extraction is improved. This suggests that prolonged exposure to high temperatures would induce Maillard reactions and/or degradation [33,34].

Sucrose shows greater thermal sensitivity, reaching optimal yield in a medium temperature range, with a defined maximum power, indicating that a high value can induce the breakdown of glycosidic bonds. Finally, treatment time has a positive effect up to a certain threshold, after which a decrease in sucrose concentration is observed, possibly due to conversion to monosaccharides [35].

Finally, for fructose, the temperature shows more stable behavior, with a gradual increase in its extraction, although with a lower slope than glucose. A zone of maximum efficiency is identified at medium power, suggesting that fructose is mainly released by physical breakdown of the matrix. Time has a moderate positive effect, but excessive prolongation could induce thermal degradation reactions [36].

The Box–Behnken design allowed the identification of optimal operating zones for each sugar, revealing differences in their thermal and mechanical sensitivity. Glucose responds favorably to high temperatures; sucrose requires more controlled conditions to avoid hydrolysis; and fructose benefits from intermediate power and moderate times. These results are key to adjusting selective extraction protocols, depending on the desired sugar profile and utilization objective.

Figure 2 shows that, in relation to glucose extraction from HMC-CPH, even under high temperature conditions, the yield was considerably limited, indicating the unavailability of free glucose in the hemicellulose fraction. This trend remained constant across the entire power range evaluated, reinforcing the hypothesis that glucose is not found in free form, but rather integrated into more recalcitrant structures that require more intensive pretreatments for its release [33]. Likewise, increasing the exposure time did not produce significant improvements in extraction, indicating that the kinetics of glucose release are restricted by the structural resistance of the lignocellulosic matrix [34].

In the case of sucrose, an optimal extraction zone was identified under conditions of medium temperature, intermediate to high power, and moderate treatment times. These results suggest that excessive temperatures or prolonged exposures could induce hydrolysis or thermal degradation of the disaccharide [35]. This compromises its stability and reduces its yield during the extraction process.

As for fructose, a sustained increase in its extraction is observed around the increase in temperature and time at intermediate powers, suggesting that fructose does not require extreme conditions and that its release kinetics are more efficient than those of glucose [36,37].

4. Conclusions

The conditions for microwave-assisted hydrothermal pretreatment of cocoa pod husk and cocoa pod husk of hemicellulose were optimized for glucose, sucrose, and fructose and predicted by Box–Behnken response surface design. For the optimization of mono- and disaccharides from cocoa pod husk, the predicted values of optimum temperature, power, and time were as follows: (1) for glucose, 135.4 °C, 180.6 W, and 5.8 min, (2) for sucrose, 154.3 °C, 256.3 W, and 20.2 min, and (3) for fructose, 129.5 °C, 173.8 W, and 5.27 min. For the optimization of mono- and disaccharides from cocoa pod husk of hemicellulose, the predicted optimal temperature, power, and time values were as follows: (4) for glucose, 142.2 °C, 204.4 W, and 10.5 min, (5) for sucrose, 148.8 °C, 215.6 W, and 14.3 min, and (6) for fructose, 151.6 °C, 231.6 W, and 13 min. The proposed optimization of microwave-assisted hydrothermal pretreatment for the extraction of glucose, sucrose, and fructose can be a green protocol for the utilization of cocoa pod husk.

This study provides precise conditions for temperature, power, and time for the extraction of mono- and disaccharides from cocoa pod shells, which would facilitate the standardization of the process on a larger scale. Other favorable aspects worth noting are that there are large-capacity microwave devices that can be adapted to continuous processes under certain technical adjustments. There is also demand for bioproducts; the extracted glucose, sucrose, and fructose can be used in functional foods, bioplastics, and bioethanol, among others, which could justify an investment. However, the technical and investment challenges of scaling up to industrial scale would have to be considered, as well as the validation of the design under real production conditions. It is recommended to replicate this study for the extraction of other valuable compounds such as organic acids, phenolics, lignin, etc., found in waste or co-products from the chocolate industry. It is also recommended to explore synergies with other agro-industrial wastes and measure feasibility by integrating life cycle studies, energy impact, and economic viability.