Mamoncillo peels and seeds were collected from fruits and vegetables stores in the city of Manizales, Colombia. These wastes were reduced in size to dimensions of less than 1 cm. Then, they were dried at 45 °C until they reached constant weight in order to avoid the loss of substances with low molecular weight. Afterwards, they were reduced in size using a disc mill to obtain a powder of less than 250 pm (-60 mesh). Finally, the powdered samples were stored in plastic bags and placed in a desiccator to prevent moisture absorption.

In this work, the kinetic parameters and the reaction mechanism for the thermal degradation of mamoncillo peels and seeds were determined. The thermal decomposition of lignocellulosic biomass is a heterogeneous process because the biomass components undergo thermochemical reactions in the solid state, with very complex reaction mechanisms (Santos et al., 2020). Therefore, determining the pyrolysis kinetics of biomass requires the use of different reaction kinetic models.

The so-called model-free methods are based on determining Eα without knowing the reaction's kinetic model (Chen et al., 2017; Liu et al., 2020). They are also known as isoconversional methods because i) Eα is obtained as a function of conversion, and ii), as an isoconversional principle, at a particular conversion, the reaction rate for the thermal decomposition of biomass is function of temperature (non-isothermal analysis) (Singh et al., 2020a). Among the most used isoconversional methods to calculate kinetic parameters are the KAS, FWO, and Starink integral methods, in addition to the Friedman differential method (Singh et al., 2020a). The Equations of these methods are presented below:

FWO method (Santos et al., 2020):

Starink method (Singh et al., 2020a):

Friedman method (Gogoi et al., 2018):

where α is the fractional conversion; Eα , A _a , and T _a are the activation energy, pre-exponential factor, and absolute temperature for a value a, respectively; β is the heating rate; R is the universal gas constant (8,314 J/kmol.K); g(α) is a function that describes a reaction model in its integral form; f(α) is a function that describes a reaction model in its differential form; and C is a constant. Table 1 presents the algebraic expression of 18 different reaction models, which are a function of ^f(α) and ^g(α).

Table 1

Source: Emiola-Sadiq et al. (2021), Mishra and Bhaske (2014), Vyazovkin et al. (2011)

Here, f(0,5) and g(0,5) are functions that describe various reaction models in differential and integral form, respectively. They are evaluated at a fractional conversion of 0,5. T ₀₅ is the absolute temperature at a conversion value of 0,5. The Z(α) master-plot consists of theoretical curves from different reaction kinetic models for solid-state degradation reactions, which are independent of the kinetic parameters (Wang et al., 2019; Singh et al., 2021). In these curves, the integral and differential forms of the reaction models are combined. These curves are used to identify the most appropriate theoretical kinetic model that describes an experimental reaction process.

To predict the mechanisms or kinetic model of thermal degradation reactions, the recommendations given by the ICTAC (Vyazovkin et al., 2011 ) were followed, which involve using the master-plots method in association with the Criado method (Santos et al., 2020). To this effect, the following Equation was used (Gogoi et al., 2018; Santos et al., 2020; Aboulkas et al., 2010):

The theoretical master-plots were constructed by plotting for each kinetic model, while the experimental master-plot was obtained by plotting with experimental data (Santos et al., 2020). Both terms of Equation (5) were plotted as a function of fractional conversion. The coincidence of a theoretical master-plot with the experimental one indicates that the thermal degradation of biomass is carried out by means of the reaction kinetic model given by the theoretical master curve (Aboulkas et al., 2010).

In this research, the four isoconversional methods (KAS, FWO, Starink, and Friedman) were used to determine E _α . A _α was calculated by means of Equation (6), while the reaction mechanism or kinetic model (f(α)) of the thermal decomposition for mamoncillo waste was obtained using master-plots.

The results obtained for the proximate analysis on a dry basis, the ultimate analysis on a dry ash free basis, the atomic ratios, and the HHV of mamoncillo wastes are shown in Table 2. Note that the total moisture, ash, and fixed carbon content are higher in the peels than in the seeds. Mamoncillo peels have four times more ash content than the seeds. The peels can cause issues related to slags, corrosion, or clogging in the equipment (Chen et al., 2018). Additionally, a high ash content can cause poor heat generation and reduced energy conversion (Singh et al., 2020a). The high volatile matter contents in the seeds should facilitate ignition at low temperatures and benefit the formation of gaseous and liquid products during the pyrolysis process (Santos et al., 2020). These wastes have the same fixed carbon content, and they are in the range required for efficient burning, which is between 15 and 25% (Santos et al., 2020). The empirical formulas for the peels and seeds, as obtained from the ultimate analysis, are C₂₀H₂₉,₅O_13,7N₀₂₀S_0,04 and C₂₀H_{29 4}O_13,9N_0,14S₀,₀₃, respectively. Both wastes have the same empirical formula. These similar results lead to similar values for the the atomic ratios H/C and O/C and the HHV. However, the little difference in HHV is due to the fact that the mamoncillo peels have a higher ash content (Rojas and Flórez, 2019).

Table 2

*Dry basis. **Dry ash free basis.

Figure 1 shows the TG/DTG curves for the peels (Figure 1a) and the seeds (Figure 1b) at three heating rates. Note that thermal decomposition of mamoncillo waste can be described using three main zones (Santos et al., 2020). The first zone indicates the drying of wastes, and it is called the passive zone, where the total moisture is removed (Wang et al., 2019; Singh et al., 2021). This zone is between 30 and 170 °C for peels and 30 and 190 °C for seeds. In the second zone, the main active pyrolysis zone (Emiola-Sadiq et al., 2021), the elimination of volatile matter takes place, which happens between 170 and 650 °C for peels and 190 and 590 °C for seeds. In the last zone, there is char formation, with temperatures of >650 and >590 °C for peels and seeds, respectively. The peak temperatures for the three different heating rates are 303,22, 314,50, and 326,20 °C for peels, and 291,49, 303,17, and 315,56 °C for seeds. These results indicate that mamoncillo seeds are pyrolytically more reactive than mamoncillo peels.

Figure 1 Source: Authors

The kinetic parameters corresponding to the pyrolysis of mamoncillo waste were evaluated by calculating E _α and A _α via the KAS, FWO, Starink, and Friedman isoconversional methods. To obtain the kinetic properties, only the second zone in the DTG curve was considered, as it is characterized by the highest mass loss has the fraction that is considered to be pyrolysable (Mehmood et al., 2017).

Figures 2a and 2b show the variations of E _α with conversion (a=0,05-0,80) as estimated via the isoconversional methods for mamoncillo peels and seeds. Regarding mamoncillo peels, it was found that E _α varies with conversion ranges of 132,94-365,40, 134,20-357,21, 132,21-362,92 and 152,24-578,27 kJ/mol for the KAS, FWO, Starink, and Friedman methods, respectively. As for the seeds, Eα varies with conversion ranges of 120,07-389,62, 122,03-380,32, 119,44-386,94, and 135,13-491,09 kJ/mol. The average Eα for the peels are 240,15, 237,36, 238,63, and 296,84 kJ/mol for these methods. For the mamoncillo seeds, the average E _α are 198,22, 197,49, 197,10, and 228,25 kJ/mol. It was also found that, for the KAS, FWO, and Starink methods, the activation energy deviation from the average ^Eα was between 29 and 31% for the peels and from 33 to 35% for the seeds. For the Friedman method, these deviations were 47 and 42% for peels and seeds, respectively. These results indicate that the integral methods were more accurate than the differential method (Vyazovkin et al., 2011).

Figure 2 Source: Authors

During the pyrolysis of mamoncillo waste, E _α showed little variations with regard to the KAS, FWO, and Starink methods. For the peels, the average values obtained were 240,15, 237,36, and 238,63 kJ/mol, respectively. For the seeds, these values were 198,22, 197,49, and 197,10 kJ/ mol. Meanwhile, the Friedman method reported significant variations, with values of 296,84 and 228,25 kJ/mol for the peels and seeds, respectively. These results indicate that mamoncillo seeds are more reactive than mamoncillo peels during the pyrolysis process (Singh et al., 2020a). Moreover, Figure 2 shows that conversions of up to 0,20 and low Eα values (184 kJ/mol for peels and 161 kJ/mol for seeds) are needed to break the weaker bonds between molecules and remove light components. This indicates the start of the pyrolysis reaction (Pawar et al., 2021). For conversions between 0,20 and 0,70, the value of Eα increases to 270 kJ/mol for peels and 201 kJ/mol for seeds. This may be mainly due to cellulose as well as hemicellulose pyrolysis (Santos et al., 2020). For conversions greater than 0,70, a higher Eα was found, which is due to lignin pyrolysis (Pawar et al., 2021).

As mentioned earlier, model-free methods can generally be split into two categories: differential and integral. These are used for estimating kinetic parameters (pre-exponential factor, activation energy, and reaction order) (Vyazovkin et al., 2011). The activation energy may show differences depending on the method used for calculation. These differences are due to the inherent and unavoidable inaccuracy and imprecision of differential methods. Therefore, they can sometimes exhibit numerical instability when compared to integral methods; when the reaction heat varies noticeably with the temperature program and when they are applied to differential data, their accuracy can be limited due to the difficulty in determining the baseline (Vyazovkin et al., 2011; Cai et al., 2018). The Friedman method is the most general isoconversional differential technique, and it requires derivative conversion data, which leads it to be numerically unstable and sensitive to noise (Cai et al., 2018). In fact, in this work, the Friedman method showed a higher average Eα than KAS, FWO, and Starink for both mamoncillo wastes. This behavior can also be explained by the fact that the Friedman method normally depends on the instantaneous rate of biomass conversion without any pre-assumptions, which it is not possible in integral methods (Pawar et al., 2021). The authors used this method for the sake of comparison, but they do not recommend using it.

Table 3 presents the variations in A _a with conversion (α=0,05-0,80) at 10 °C/min as estimated via isoconversional methods for mamoncillo waste. This factor, also called the frequency factor, measures the frequency of active collision that occurs between the reactant molecules. It is used to explain the chemistry of the pyrolysis reaction, and it serves to optimize the experimental conditions of pyrolytic processes (Santos et al., 2020). It is known that, for A _α <10⁹ s^-1, only surface reactions are taking place, while A _a >10⁹ s^-1 indicates that i) there is a complex reaction, which does not depend on the contact surface area; ii) a high collision of molecules is required; and iii) a high Eα is needed for biomass pyrolysis (Kaur et al., 2017). In our study, the A _a values calculated via the KAS, FWO, and Starink methods were found to be between 10⁹ and 10³¹ s^-1 for peels and between 10⁹ and 10³⁴ s^-1 for seeds. It was found that A _α varies with conversion as ^Eα changes, i.e., A _a increases when the conversion is higher.

Table 3

Source: Authors

It was also found that A _α decreases with the increase in heating rate. This indicates that, at low conversions (<0,20), the reactivity of the thermal decomposition of mamoncillo wastes is low, with the highest values at high conversion (Singh et al., 2020a).

We predicted the reaction model during the pyrolysis of mamoncillo waste using 18 reaction kinetic models (solidstate mechanisms) with the corresponding experimental and theoretical master-plot curves. The experimental and theoretical curves at 10 °C/min are presented in Figure 3.

Figure 3 Source: Authors

Regarding the mamoncillo peels, it can be seen (Figure 3a) that, for conversions from 0,05 to 0,50, the experimental curve almost overlaps the R2 model or the contracting cylinder model, and, for conversion ranges between 0,50 and 0,80 the experimental curve is close to the F3 model or the third-order reaction model. It can be concluded that the decomposition of mamoncillo peels in each conversion range (0,05-0,50 and 0,50-0,80) can be described by a single reaction model. Meanwhile, regarding the mamoncillo seeds (Figure 3b), note that, for conversions between 0,05 and 0,50, the experimental curve is close to the F2 model or the second-order reaction model, and, for conversion ranges of 0,50-0,80, the experimental curve almost overlaps the R3 model or the contracting sphere model. This indicates that the pyrolysis of mamoncillo seeds can be described by a complex reaction mechanism.

The thermodynamic parameters ΔH, ΔG, and ΔS involved in the thermal decomposition of mamoncillo wastes were calculated using the activation energy obtained via the KAS, FWO, Starink, and Friedman methods. The values of ΔH, ΔG, and ΔS were calculated using Equations (7) to (9). The changes in ΔH, ΔG and ΔS with the conversion value at 10°C/min are reported in Tables 4 - 9, respectively. ΔH is the minimum energy required by the biomass to form products during pyrolysis (Singh et al., 2021; Kaur et al., 2017). Thereupon, AH evaluates the total energy consumed in the conversion of mamoncillo waste into products (Santos et al., 2020). AG represents the increase in the system's total energy for the formation of activated complexes (Maia and Morais, 2016), and it can be used to determine whether the reactions are spontaneous. A negative AG value indicates the occurrence of spontaneous reactions, while a positive value denotes a nonspontaneous reaction (Chen et al., 2023). Meanwhile, ΔS is associated with the measure of randomness or disorder of energy and matter in a system, as well as to the formation of new chemical compounds (Kumar et al., 2020).

Table 4

Source: Authors

In Tables 4 and 5, it can be seen that, for both wastes, i) ΔH increases with conversion for all methods except for Friedman; ii) the average values are similar for the KAS, FWO, and Starink methods; iii) the H values for the Friedman method are higher than those of KAS, FWO, and Starink; iv) the AH values for mamoncillo seeds are lower than those reported for mamoncillo peels; and v) all ΔH values are positive.

Table 5

Source: Authors

Positive ΔH values indicate that the thermal degradation of mamoncillo wastes took place via endothermic reactions. This means that an external source of heat is required to transform biomass into gas, oil, and charcoal during the pyrolysis process (Chen et al., 2017; Kumar et al., 2020). It was found that the difference between the values of Eα and AH at each conversion level is less than 5 kJ/mol. This small difference indicates the formation of low-energy activated complexes in the product-forming potential barrier (Santos et al., 2020). Therefore, the pyrolysis of mamoncillo wastes for energy generation is viable (Kumar et al., 2020; Santos et al., 2020). Moreover, for the mamoncillo peels, the average ΔH values at three heating rates (10, 20, and 40°C/ min) for the KAS, FWO, Starink, and Friedman methods were 235,27 ± 0,10, 232,47 ± 0,10, 233,74 ± 0,10, and 291,95 ± 0,10 kJ/mol, respectively. For mamoncillo seeds, these values were 193,42 ± 0,10, 192,70 ± 0,10, 192,30 ± 0,10, and 223,46 ± 0,10 kJ/mol, respectively. These results show that the effect of the heating rate on AH is negligible. Similar results were reported for banana leaves (Singh et al., 2020a) and acai seeds (Santos et al., 2020). It can be seen that mamoncillo seeds consume less total energy than mamoncillo peels during their conversion into products.

The variations in ΔG with the conversion level for mamoncillo peels and seeds as obtained via the KAS, FWO, Starink, and Friedman methods are shown in Tables 6 and 7. The average AG values for mamoncillo waste are in the range between 162 and 164 kJ/mol. The ΔG values are similar for both wastes. However, the average ΔG values the peels were slightly higher than those obtained for the seeds, which means that, for thermal decomposition, peels require more heat. Similar results have been reported for the pyrolysis of agricultural residues (Chen et al., 2017). In addition, ΔG was found to decrease with conversions of 0,05-0,80, implying that the energy of the reaction system decreased when the pyrolysis started (Yuan et al., 2017). It was also found that, for both wastes, ΔG is positive, which indicates that the thermal decomposition reaction is nonspontaneous. Therefore, this reaction requires energy consumption for the chemical bonds to break. This result suggests that the reactivity of the thermal degradation was low and that the pyrolysis process requires an external energy source to obtain the activated complex (Chen et al., 2017).

Table 6

Source: Authors

Table 7

Source: Authors

Tables 8 and 9 present the variations in AS with the conversion level for mamoncillo peels and seeds as obtained via the KAS, FWO, Starink, and Friedman methods. ΔS is negative at conversion of up to 0,10 for the peels, and the seeds exhibit negative values up to a fractional conversion of 0,30. Low ΔS values indicate that these wastes undergo small physical and chemical charges during thermal degradation, transitioning to a new condition close to thermodynamic equilibrium (Dhyani et al., 2017), increasing the pyrolysis reaction time (Singh et al., 2020b). For high ΔS values, the reactivity is high, the reaction time decreases, and the process is far from thermodynamic equilibrium (Singh et al., 2021; Mallick et al., 2018). Here, ΔS increased when the fractional conversion increased for the KAS, FWO, and Starink methods and both wastes, implying that the pyrolysis was far from reaching equilibrium. The ΔS values calculated via the Friedman method are all negative for mamoncillo waste. ΔS followed similar trend as ΔH and E _α , which increased with the increasing fractional conversion. This can be attributed to the fact that the reaction rate increased for the 0,05-0,80 conversions (Singh et al., 2021).

Table 8

Source: Authors

Table 9

Source: Authors

The knowledge of thermogravimetric, kinetic, and thermodynamic parameters is fundamental in determining the feasibility and efficiency of the pyrolysis process, designing pyrolyzers, determining the energy balance, and calculating the energy consumption or requirements (Kumar et al., 2020, Singh et al., 2020a). These parameters can be used to determine the optimal operating conditions to obtain a specific pyrolysis product (bio-oil, biochar, and syngas) (Li et al., 2023) from mamoncillo wastes. Therefore, this study can contribute to understanding mamoncillo waste pyrolysis and its future applications. For example, it was found that these wastes have an ignition temperature (Figure 1) between 210 °C (peels) and 250 °C (seeds), as well as high volatile matter contents (between 77 and 84% on a dry basis). This indicates that the pyrolysis of these wastes can produce more bio-oil than biochar and syngas (Nawaz et al., 2021). On the other hand, by controlling the operating conditions, it is possible to produce a higher proportion of biochar than bio-oil and syngas, or a higher proportion of syngas than bio-oil and biochar. These wastes have different applications. Bio-oil can be used as solid biofuel for cooking purposes and domestic heating, or as a solvent for extracting important chemical compounds. Biochar can be used as a soil and water conditioner (it is a good biosorbent), as a solid fuel for cooking purposes and domestic heating, or as a supercapacitor (Nawaz et al., 2021; Li et al., 2023). On the other hand, condensable gases or synthesis gas (syngas) can be used as a biofuel for combustion engines or boilers (Nawaz et al., 2021).