Aroma Compound Release from Starches of Different Origins: A Physicochemical Study

Abstract

Sensory properties, particularly aroma, play a crucial role in consumers’ acceptance and perceived quality of food. The release and the perception of aroma compounds is affected by their interaction with nonvolatile ingredients of foods, such as proteins, lipids, and polysaccharides. These interactions, whether reversible or irreversible, significantly influence aroma retention and release. Starch, a common food constituent, has been found to interact with aroma compounds, impacting flavor dynamics through processes like complexation and encapsulation. In this study, reversed-flow gas chromatography (RF-GC) is employed for the estimation of the release behavior of polar (diacetyl) and non-polar (dl-limonene) aroma compounds from starches of various origins (corn, wheat, rice, and potato). The results show that aroma compound release is influenced by the matrix composition, environmental conditions, and physicochemical properties of both starch and aroma compounds. The temperature-dependent mass transfer coefficients and activation energies reveal the strong influence of polar and non-polar characteristics on aroma compound behavior. Additionally, significant variations in retention and release are observed based on the starch type and the type of bonds involved in aroma compound interactions, underscoring the critical role of thermodynamic and kinetic parameters in flavor dynamics.

1. Introduction

Aroma is a key factor in food acceptance, playing a vital role in both the perceived quality and the overall sensory experience. Beyond flavor enhancement, aromas trigger emotional and memory responses during eating. The interactions between aroma compounds and food ingredients such as proteins, polysaccharides, lipids, etc., are of great importance in terms of the behavior of these compounds in food systems [1,2,3,4].

Aroma in foodstuffs is either natural or added by using small or large molecules of aroma compounds that are connected to the food’s ingredients. The strength of this connection depends on the aroma compound molecules and the food’s ingredients, as well as on the bonds that develop [5,6,7,8,9]. Aroma has drawn more attention as a crucial component of improving food quality and competitiveness in the global market, especially in light of its stability within food systems [10].

The release of aroma compounds into the gas phase, a complicated process regulated by several factors, is the basis for food flavor detection. This detection can take place in vivo during consumption or in vitro through the allocation of the aroma compounds between the headspace and food. Both the rheological characteristics of the food matrix [11,12] and the physicochemical characteristics of the aroma compounds (solubility, polarity, and hydrophobicity), as well as their interactions with food ingredients like proteins, lipids, polysaccharides, and salts, affect the distribution of aroma compounds [13,14].

Polysaccharides interact with aroma compounds through multiple mechanisms including adsorption, entrapment, complexation, encapsulation, and hydrogen bonding. These interactions can be either reversible (through weak physical forces) or irreversible (through strong chemical forces like covalent bonds with amino and sulfhydryl groups) [15,16,17,18]. Non-covalent bonds are particularly valuable as they enhance aroma retention while allowing controlled release through temperature changes [19,20].

Starch, commonly used as a thickener, stabilizer, and carrier in foods, can interact with volatile aroma compounds such as alcohols, aldehydes, lactones, terpenes, etc. [15,21]. It is well known that amylose and amylopectin form inclusion complexes with low-molecular weight aroma compounds, affecting their release at the gas/liquid interface [22,23]. The synthesis and the concentration of aroma compounds and carbohydrates are two examples of the many variables that affect the interactions between these two substances, which are typically low energy [24]. The diffusivity of aroma compounds within the food matrix strongly influences the release of these compounds [25], while the release of certain aroma compounds can be increased by the presence of mono- and di-saccharides [26,27,28].

The control of aroma compounds’ intensity and quality in the headspace is mainly determined by thermodynamic and kinetic factors, which influence their retention and release. The factors affecting the parameters include the physicochemical characteristics of the aroma compounds, the food matrix, the environmental conditions, and the interactions between aroma compounds and matrix components. Thermodynamic characteristics are associated with the air/matrix partition coefficient at equilibrium, while kinetic properties determine the speed at which aroma compounds are emitted from the matrix into the headspace [14].

In order to ascertain the type of protein–aroma chemical interactions, fluorescence spectroscopy is frequently employed to measure thermodynamic characteristics including changes in Gibbs energy (ΔG), enthalpy (ΔH), and entropy (ΔS) [19,29]. Static headspace measurement makes it possible to quantify the retention and release of aroma components from food products by determining the thermodynamic and kinetic parameters [30,31]. The kinetic value indicates the rate at which aroma chemicals are released from food, whereas the thermodynamic parameter is associated with the gas/matrix partition coefficient (volatility), as determined under equilibrium conditions [32,33]. According to experimental research, the physicochemical characteristics of the matrix, the affinity of aroma compounds for matrix components, ambient factors, and other factors all affect the magnitudes of thermodynamic and kinetic parameters [34].

In addition to the above techniques, inverse gas chromatography (IGC) has been effectively applied for the study of the interactions between starch and aroma compounds, as well as the investigation of volatile compound–material interactions [35,36,37,38,39,40,41]. In this technique, a solid r liquid coating on a solid support fills the chromatographic column, playing the role of the stationary phase, and the aroma compounds act as the probe that interacts with the stationary phase, and from the calculated physicochemical parameters the surface characteristics are extracted.

Although IGC is a versatile and reliable technique for the determination of physicochemical parameters, it has two limitations—the first one is that the probe needs to be at infinite dilution, and the second is that it is unable to determine kinetic parameters.

In this study, reversed-flow gas chromatography (RF-GC), an IGC sub-technique, is used for the first time to study the mechanisms of the release and evaporation of aroma compounds both from bulk liquid and from different starches. This technique was introduced in 1980 [42], and it has since been used in a wide range of scientific fields, including the calculation of adsorption isotherms [43], the kinetic study of alcoholic fermentation [44,45,46,47], the analysis of interactions between aroma compounds and starch [48], the determination of diffusion and mass transfer coefficients [49,50,51], and more.

The aim of the present study is to apply the RF-GC technique in order to determine the physicochemical parameters of the release of two aroma compounds—one polar (diacetyl) and one non-polar (dl-limonene)—from a bulk liquid phase as well as from different botanical starches (corn, wheat, rice, and potato), providing valuable insights into the physicochemical mechanisms governing these processes.

2. Materials and Methods

Diacetyl and dl-limonene of 99% purity were obtained from Merck A.G. (Darmstadt, Germany). In

Table 1. The physicochemical properties of the aroma compounds ^a: molecular weight (M_W), molar volume (M_V), hydrophobicity (Log P), water solubility (W_Sol ^b), boiling point (b_P, 760 mm Hg), density (d ^b), saturated vapor pressure (P_sat ^b), and odor descriptor.

	Chemical Formula	Mw (g mol−1)	Mv (cm3 mol−1)	Log P (25 °C)	WSol (g L−1) a	bp (°C) b	d (g mL−1) a	Psat (mmHg)	Odor Descriptor
dl-limonene	C10H16	136.24	163.3	4.57	0.0076	178	0.842	1.55	lemon
diacetyl	C4H6O2	86.09	88.8	−1.34	200.0	88	0.990	56.80	buttery

^a Data retrieved from the following database: https://pubchem.ncbi.nlm.nih.gov/ (accessed on 15 November 2024); ^b determined at 25 °C.

, a comprehensive overview of some of their physicochemical properties is given.

Starch granules derived from various botanical sources, including corn, wheat, potato, and rice, were also used in the present research. All these were supplied by Sigma-Aldrich (St. Luis, MO, USA) and have the following product numbers: S9676 for corn (Batch #020N0198), S4251 for potato (Lot#058K01572V), S5127 for wheat (Lot#BCBP4613V), and S7260 for rice (Lot#BCBN7121V).

To characterize these starches, several analytical techniques were employed.

The specific surface area (SSA) of each starch sample was determined in triplicate using the Micromeritics Tristar 3000 instrument. The N₂ adsorption–desorption isotherms at liquid nitrogen temperature (77 K) were recorded to study the textural properties of the samples using a piece of Micromeritics apparatus (Micromeritics Tristar 3000 porosimeter, Norcross, GA, USA). Using the corresponding data, each specific surface area was calculated by the Brunauer, Emmett, and Teller (BET) equation. The calculation of micropores and external surface area was performed by the t-plot method. An identical amount of each sample was firstly freeze-dried and then placed in a piece of degassing apparatus (Micromeritics, FlowPrep 060 sample degas system, Norcross, GA, USA) at 70 °C under a nitrogen stream for 5 h. Each dried sample was weighed accurately before being placed in the porosimeter. The glass transition temperature (T_g) of each starch sample was measured using the TG 209 F3 Tarsus instrument from Netzsch (Burlington, MA, USA). Ten (10) mg of each sample was added into the instrument cell and the temperature was regulated from 30 °C to 300 °C.

The morphology and the size distribution of all starches were illustrated by taking photos with a Zeiss Supra 35VP FEG Scanning Electron Microscope (Carl Zeiss; Oberkochen, Germany). A small amount of each sample was placed on an appropriate holder and covered with a thin layer of conducting material, (Au/Pd) alloy, using a Quorum Q150V Plus Automatic Coater Sputter (East Sussex, UK).

Finally, FTIR spectra were obtained for all raw materials used, as well as for the mixtures of different starches and aroma compounds, by using a Cary 630 FTIR Spectrometer (Agilent S.A., Santa Clara, CA, USA) with an ATR sampling module. The collection of FTIR spectra was performed across a wavenumber range of 400 to 4000 cm⁻¹. The resolution was 2 cm⁻¹, while the data collection interval was 1 cm⁻¹. The scanning speed was 1 cm/s, while the scanning was repeated 64 times for every sample. Background subtraction was applied for all spectra.

For the implementation of the RF-GC technique, helium of 99.999% purity, provided by Air Liquide (Athens, Greece), was used as the carrier gas and was dried by being passed through a gas purifier, No. 452 of Matheson Gas Products (East Rutherford, NJ, USA), with a constant flow rate (corrected each time at the column temperature) of 0.35 × 10⁻⁶ m³ s⁻¹. For the FID flame, the hydrogen used was produced by a hydrogen generator (Claind, Como, Italy), while the air was produced by a compressor. The pressure drop along l + l′ (cf. Figure 1) in all experiments was negligible.

2.1. Reversed-Flow Gas Chromatography Instrumentation

The basis of the RF-GC technique is the alteration of the flow direction of the carrier gas by time. The configuration of the system consisted of a gas chromatograph produced by Shimadzu (Kyoto, Japan) model 14A, a Flame Ionization Detector (F.I.D.), a sampling cell constructed from stainless steel, and a four-port electronic gas valve from VICI (Valvo Instrument Co., Houston TX, USA), as illustrated in Figure 1.

The sampling column’s two lengths were l = l′ = 60 cm. The length L₁ of the diffusion column was 54 cm, while that of the glass vessel, L₂, was 8 cm. The internal diameter of both the diffusion column and the vessel was 4 mm. The chromatographic oven’s temperature was set at the proper experimental temperature (303.15 K, 313.15 K, 323.15 K, and 333.15 K) and the detector’s temperature was set at 523.15 K in all experiments.

While a stagnant column of carrier gas filled the diffusion column and the glass vessel, the chromatographic peaks in RF-GC were created by changing the flow direction of the carrier gas in the sample cell by means of the four-port valve (cf. Figure 1). When the valve was in the location shown by the solid lines, the carrier gas entered the column at x = 0 and exited it from x = l + l′ in the direction of the detector. The carrier gas flowed in the other direction and entered the column at x= l + l′ when the valve was in the other position (dashed lines). In both portions l and l′, the time reversal was less than the gas hold-up time.

When the flow was restored to the initial position, narrow, rather symmetrical chromatographic peaks, known as sampling peaks, could be superimposed over the continuous concentration–time curve.

2.2. Experimental Procedure

In order to determine the mass transfer coefficient needed for the evaporation of aroma compounds, an amount of the aroma compound under study was introduced to the vessel, with L₂ at a height of 6 cm. Afterwards, the glass vessel with the aroma compound was connected at the free end of the diffusion column by a Swagelok union.

By the time a concentration–time curve—referred to the evaporated aroma compound—appeared, the flow of the carrier gas was inversed in order to begin the chromatographic procedure. The flow reversal lasted for 6 s, a period of time shorter than the gas hold-up time in both sections of the sampling column. By applying the above procedure, a sequence of sampling peaks was obtained (cf. Figure 2).

Each experiment stopped when four consecutive peaks appeared to be at the same height. There is a proportional relationship, as a function of time t when the flow reversal occurs, between the concentration of the substance under study at the x = l′ junction of the sampling cell at time t, c (l′, t) and the area under the curve or height H of the continuous signal of the sampling peaks [52].

H^1/M = 2 c (l′, t)

(1)

where M is equal to 1 for the linear FID detector.

In Figure 3 the variation of the height H of the sampling peaks versus time t is given.

For the determination of the mass transfer coefficient needed for the release of the aroma compounds from starches of different origins, a quantity of 0.5 g of each starch was weighed accurately and then 0.5 mL of the aroma compound was added. After homogenization, each mixture was placed in a glass container measuring up to 6 cm. The glass container with the mixture was then connected to the free end of the diffusion column with a Swagelok union and the procedure was carried out as described above.

3. Theory

As previously stated, each sampling peak’s height reaches a maximum value of H = H_∞ and then stays constant. Equation (2) provides the relationship between the maximum height H_∞ and the concentration of the evaporated aroma component, c(l) [53].

$$c\left(\mathrm{g}\right)={\displaystyle \frac{c\left(l\right)}{K}}={\displaystyle \frac{u{H}_{\mathrm{\infty }}}{2}}({\displaystyle \frac{L_1}{D_{\mathrm{g}}}}+{\displaystyle \frac{1}{k_{\mathrm{c}}}})$$

(2)

where L₁ is the length of the diffusion column (see Figure 1), D_g is the diffusion coefficient of the aroma compound in the carrier gas, k_c is the mass transfer coefficient needed for the evaporation of the aroma compound from the liquid, H_∞ is the maximum value of the sample peak’s height, u is the linear velocity of the carrier gas, and c(g) is the gaseous concentration of the aroma compound vapor in equilibrium with the bulk liquid phase with the concentration c(l).

The height of each sample peak for the evaporation of a substance can be determined by the following formula [54]:

$$H^{{\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$M$}\right.}}=2c\left(l’,t\right)=2{\displaystyle \frac{k_{\mathrm{c}}{D}_{\mathrm{g}}{c}_{\mathrm{o}}}{u(k_{\mathrm{c}}{L}_1+D_{\mathrm{g}})}}\left\{1-\exp \left[-{\displaystyle \frac{2\left(k_{\mathrm{c}}{L}_1+D_{\mathrm{g}}\right)t}{L_1^2}}\right]\right\}$$

(3)

where c₀ is the vapor’s concentration in equilibrium with the bulk liquid phase or when absorbed in the starch.

The value of H_∞, which corresponds to the height of the sample peak after a long period of time, is given by Equation (4).

$$H_{\infty }^{{\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$M$}\right.}}={\displaystyle \frac{{2k}_{\mathrm{c}}{D}_{\mathrm{g}}{c}_{\mathrm{o}}}{u(k_{\mathrm{c}}{L}_1+D_{\mathrm{g}})}}$$

(4)

From Equations (3) and (4), Equation (5) obtained [54]:

$$\ln \left(H_{\infty }^{{\displaystyle \frac{1}{M}}}-H^{{\displaystyle \frac{1}{M}}}\right)=\ln H_{\infty }^{{\displaystyle \frac{1}{M}}}-\left[{\displaystyle \frac{2\left(k_{\mathrm{c}}{L}_1+D_{\mathrm{g}}\right)}{L_1^2}}\right]t..$$

(5)

The slope of the plot of $\ln \left(H_{\infty }^{{\displaystyle \frac{1}{M}}}-H^{{\displaystyle \frac{1}{M}}}\right)$ against t is as follows:

$$\mathrm{slope}=-2(k_{\mathrm{c}}{L}_1+D_{\mathrm{g}})/L_1^2.$$

(6)

By replacing the known values of L₁ and D_g in Equation (6), the k_c value can be determined.

4. Results

4.1. Characterization of Starch Granules

The SSA results and the corresponding standard deviation, as well as the T_g results, are given in

Table 2. The physicochemical properties (specific surface area (SSA) with the corresponding standard deviation and glass transition temperature (T_g)) of starch granules of different origins.

Starch Origin	SSA (m2 g−1)	Tg (°C)
Potato	0.277 ± 0.013	71.2 ± 0.1
Wheat	0.239 ± 0.016	72.8 ± 0.3
Corn	0.450 ± 0.009	65.3 ± 0.1
Rice	1.075 ± 0.012	69.1 ± 0.1

providing a comparative view of the different starch types. The measurements for SSA were performed in triplicate.

Scanning electron microscope images of starches of different botanical origins are given below in Figure 4. From these photographs, the differences in morphology as well as in the size and size distribution of the starch particles of different origins can be observed.

The FTIR spectra of the different starches, the aroma compounds, and the mixtures of starches with these aroma compounds are depicted below in Figure 5, Figure 6, Figure 7 and Figure 8.

From these spectra, the binding of the aroma compounds with the different starches is obtained by the enhancement of the peaks in all spectra obtained from the starch and aroma compound mixtures compared to the same peaks in the pure starch spectra.

4.2. Mass Transfer Coefficients for Evaporation of Aroma Compound from Bulk Liquid

In order to determine the mass transfer coefficient of aroma compounds from the bulk liquid phase to the carrier gas, the experimental procedure described above was applied. Using the linear plot of $\ln \left(H_{\infty }^{{\displaystyle \frac{1}{M}}}-H^{{\displaystyle \frac{1}{M}}}\right)$ versus t, the slope was calculated, and by applying Equation (6), the values of k_c were extracted. Furthermore, from the plot of lnk_c against 1/T, the activation energy, E_a, needed for the transfer of the aroma compound to the carrier gas could be calculated. The experimental temperatures utilized were 303.15 K, 313.15 K, 323.15 K, and 323.15 K.

The given values for the mass transfer coefficients, k_c, are the mean values after three replications at each temperature. In order to apply the aforementioned method, its precision should be estimated. For this reason, the obtained mean values and the corresponding standard deviation were applied.

The precision for each quantity is computed from the following relation:

$$\mathrm{precision}(\%)=100-100\times {\displaystyle \frac{\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{r}\mathrm{d}\mathrm{d}\mathrm{e}\mathrm{v}\mathrm{i}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}{\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e}}}$$

(7)

The results of the mass transfer coefficients, k_c, as well as the activation energies needed for the evaporation are illustrated in

Table 3. The mean values of the mass transfer coefficients, k_c, required for the evaporation of diacetyl and dl-limonene from the bulk liquid to the gaseous phase at various temperatures, along with the standard deviation and the corresponding activation energy, E_a, needed for the procedure of evaporation.

T/K	Diacetyl		Dl-Limonene
	103 kc/cm s−1	Ea/kJ mol−1	103 kc/cm s−1	Ea/kJ mol−1
303.15	12.64 ± 0.34	15.16 ± 0.69	10.99 ± 0.56	5.30 ± 0.20
313.15	14.79 ± 0.45		11.88 ± 0.89
323.15	18.18 ± 0.93		12.56 ± 0.23
333.15	21.55 ± 0.73		13.31 ± 0.38

.

4.3. Mass Transfer Coefficients for Evaporation of Aroma Compounds from Starches of Different Origins

As described above, when the glass vessel contained starches of different botanical origins impregnated with an aromatic compound and connected to the free end of the diffusion column, the mass transfer coefficients needed for the evaporation of the aroma compounds from the starch granules to the carrier gas could be calculated.

By applying Equation (6) as above, the values of k_c were calculated. In addition, from the plot of lnk_c versus 1/T, the activation energy E_a required for the evaporation of aroma compounds from starch was determined. The experimental temperatures used were 303.15 K, 313.15 K, 323.15 K, and 323.15 K. All experiments were performed in triplicate in order to calculate the average values of the mass transfer coefficients with their standard deviation.

The results of the mass transfer coefficients, k_c, as well as the activation energies needed for the evaporation of aroma compounds from starch are illustrated in

Table 4. The mean values of the mass transfer coefficients, k_c, and the activation energy, E_a, needed for the evaporation of diacetyl and dl-limonene from starch granules of different origins to the gaseous phase at various temperatures, along with their standard deviation.

T/K	Starch Origin	Diacetyl		Dl-Limonene
		103 kc/cm s−1	Ea/kJ mol−1	103 kc/cm s−1	Ea/kJ mol−1
303.15	Rice	12.34 ± 1.26	19.14 ± 1.88	7.40 ± 0.54	49.66 ± 2.44
313.15		16.86 ± 1.65 a		14.62 ± 1.21 a
323.15		19.26 ± 0.86 b		23.45 ± 2.05 b
333.15		25.19 ± 2.32		45.40 ± 2.48
303.15	Wheat	0.49 ± 0.02	98.65 ± 7.18	15.95 ± 1.13	9.03 ± 0.12
313.15		2.24 ± 0.09		17.77 ± 1.28
323.15		7.16 ± 0.15		19.85 ± 1.12
333.15		16.46 ± 0.78		22.00 ± 1.23
303.15	Corn	9.04 ± 0.86 c	20.03 ± 0.80	7.76 ± 0.23 c	16.37 ± 1.00
313.15		11.24 ± 0.93 d		10.03 ± 0.68 d
323.15		14.81 ± 1.53 e		11.84 ± 0.42 e
333.15		18.27 ± 1.23		14.03 ± 0.53
303.15	Potato	10.84 ± 0.85	26.06 ± 2.82	7.85 ± 0.15	44.11 ± 2.68
313.15		17.09 ± 1.02		13.27 ± 0.18
323.15		22.46 ± 1.23 f		25.25 ± 0.56 f
333.15		27.72 ± 1.56		36.40 ± 1.23

The same letter indicates non-statistically significant values.

.

5. Discussion

From the results listed in Table 3 regarding the evaporation of aroma compounds from the bulk liquid phase, the following conclusions can be drawn:

(i)

The values of k_c increased with the temperature, which is in accordance with previous results [54], although the increment in the case of dl-limonene is smaller than in the case of diacetyl. The values of the mass transfer coefficient for each temperature are higher for diacetyl than for limonene. The higher the temperature value, the greater the difference in values. The latter could be attributed to the higher boiling point of dl-limonene compared to the boiling point of diacetyl.

(ii)

The above is supported by applying statistical analysis (one-way ANOVA, P ≤ 0.01) to the mean value of the mass transfer coefficient at each temperature. All values are statistically different, except for the mean values corresponding to the temperature of 303.15 K.

(iii)

The polar nature of diacetyl and the non-polar nature of dl-limonene may be the cause of the variations in the activation energies, E_a, needed for the evaporation of diacetyl and dl-limonene from the bulk liquid. The differences in the calculated activation energies, E_a, needed for the evaporation of diacetyl and dl-limonene from the bulk liquid could be attributed to the polar character of diacetyl and the non-polar character of dl-limonene. Van der Waals dipole–dipole interactions could be present at the polar diacetyl, while the non-polar dl-limonene could not develop such interactions.

From the results listed in Table 4, as well as from Figure 9 and Figure 10, the following conclusions can be drawn:

(i)

The values of k_c increased with the temperature in all cases. These values are in accordance with the results given in the literature [55] for the mass transfer coefficient of diacetyl from stirred water. Although the values of k_c increased with the temperature in each system, comparing the values of two aroma compounds with the same starch at the same temperature does not show a pattern. In the case of the systems aroma compounds and wheat, the k_c values are statistically different for the two aroma compounds at all temperatures, while in the case of the systems aroma compounds and corn, the k_c values are statistically different only at the highest temperature. This may also explain the small difference observed in the activation energy for the two systems (diacetyl + corn and dl-limonene + corn).

(ii)

The different variations in k_c values with temperature for the diacetyl + starch systems (from corn, rice, and potato) compared to those of the diacetyl + starch system from wheat suggest different mechanisms of diacetyl release for these different starches. On the other hand, the release of dl-limonene seems to follow the same mechanism for wheat- and corn-derived starches and a different mechanism for rice- and potato-derived starches. These conclusions are also supported by the activation energy values. The activation energies needed for the release of diacetyl from starches derived from corn, potato, and rice have the same values (more or less), whereas for release from starches derived from wheat, the activation energy requirement was calculated to be about four times higher.

Also, the activation energy requirements for the release of dl-limonene from starches derived from corn and rice are (more or less) the same, while somewhat smaller values for the dl-limonene + potato and dl-limonene + wheat systems have been calculated.

(iii)

The calculated activation energy, E_a, needed for the release of diacetyl from wheat was found to be equal to 98.65 kJ mol⁻¹, indicating chemical sorption of the polar compound diacetyl (ketone). This could be attributed to the fixation of the polar molecule of diacetyl by amino groups of proteins, since wheat starch has the highest protein content compared to starches from other botanical sources. This conclusion is further supported by the lower k_c values observed in comparison with the other systems being studied. In contrast, the activation energy required for the evaporation of dl-limonene from wheat was determined to be equal to 9.03 kJ mol⁻¹, which is a lower value in comparison to those observed in the other systems being studied.

(iv)

In all the other systems studied, the activation energies needed for the evaporation of diacetyl from rice-, corn-, and potato-based starch were calculated to be 19 kJ mol⁻¹, 20 kJ mol⁻¹, and 26 kJ mol⁻¹, respectively—values which are in accordance with those reported (with opposite signs) in the literature [56] for the enthalpy of adsorption of ethanol on starchy substrates and which indicate weak energy bonds. The highest value was observed in the case of starch from potato, which contains the bigger amount of proteins compared with the other two starches, while the other two values are almost the same, which could be attributed to their lower contents of protein. Also, the above results can be attributed to the number of chains per branch, which is zero for the potato starch, between 4.7 and 8.7 for the rice, 4.4 for the corn, and between 11.9 and 19.2 for the wheat [57].

(v)

The activation energies needed for the evaporation of dl-limonene from rice, corn, and potato were calculated to be equal to 49 kJ mol⁻¹, 16 kJ mol⁻¹, and 44 kJ mol⁻¹, respectively. Since dl-limonene molecules cannot develop any bonds with starch granules, the formation of inclusions between dl-limonene and the substrate is possible and could lead to these differences in E_a. Also, these values are on the same order of magnitude with those reported (with opposite signs) for the enthalpy of adsorption of heptane and octane on cellulose [58].

(vi)

According to the literature [48], the enthalpy of the physicochemical interaction between diacetyl and starch of different origins adheres to the following pattern: wheat > potato > rice > maize. The values of the activation energy needed for the release of diacetyl from the different starches follow the same pattern.

(vii)

On the other hand, for the interaction of dl-limonene with different starches, the following pattern has been proposed: corn > rice > potato > wheat, while for the activation energy required for the release of dl-limonene, the pattern has been found to be as follows: rice > potato > corn > wheat. This difference could be attributed to the size of the dl-limonene molecule and the fact that rice shows the biggest specific surface area (SSA), while all the other starches have a somewhat smaller SSA.

(viii)

The activation energy values were statistically evaluated by one-way ANOVA. Differences between different systems were considered at a significance level of P  ≤  0.01 using Tukey’s post hoc test for multiple comparisons. From the calculated results of the interactions of diacetyl with different starches, we determined that the activation energy requirements for the diacetyl + corn, diacetyl + potato, and diacetyl + rice systems are not statistically different, while according to the interactions between dl-limonene and the starches, the activation energy requirements for the dl-limonene + rice and dl-limonene + potato systems have no statistical differences. Finally, the amounts of activation energy needed for the release of diacetyl from different starches are statistically different from the corresponding values for the release of dl-limonene from different starches.

(ix)

Because complex algebraic equations were applied for the calculation of the physico-chemical parameters, accuracy and uncertainties could not be calculated. However, by using Equation (7), the precision of the mass transfer coefficients and activation energies for all the systems can be determined, while the obtained results show that the RF-GC method appears to have good precision (90−99%).

6. Conclusions

In this study, the mass transfer coefficients and activation energies needed for the evaporation of diacetyl and dl-limonene from both bulk liquids and starch granules of different botanical origins were investigated. It was found that the mass transfer coefficient k_c increased with temperature for both compounds. Diacetyl, a polar compound, exhibited a significantly higher activation energy requirement for evaporation from wheat starch (98.65 kJ/mol), indicating strong chemical sorption. This could be due to interactions with amino groups in proteins or hydrogen bonding with starch hydroxyl groups. In contrast, the non-polar compound dl-limonene showed a much lower activation energy requirement for evaporation from wheat starch (9.03 kJ/mol), suggesting weaker interactions with the starch matrix.

The results also showed that the activation energies required for diacetyl’s release from rice, corn, and potato starches ranged between 19 and 26 kJ/mol, which aligns with previously reported values for similar interactions between ethanol and starch. The activation energies needed for dl-limonene’s evaporation from these starches were between 16 and 49 kJ/mole, values which are comparable to the enthalpy of adsorption for hydrocarbons like heptane and octane on cellulose, further confirming the different sorption mechanisms of polar and non-polar aroma compounds on starchy substrates. These findings underscore the roles of chemical structure and starch composition in influencing the evaporation behavior of aroma compounds.

The usage of RF-GC with the appropriate mathematical approaches to the experimental results can lead to the determination of various physicochemical and kinetic parameters concerning the interaction of volatile or semi-volatile compounds with food ingredients.

The results from this and other similar studies have important implications for the food industry, particularly for new product development. Understanding the mechanism of the release of aroma compounds from the food matrix by calculating the mass transfer coefficients and activation energies needed for the release of aroma compounds from one of the main food components, such as starch, allows the food industry to optimize the sensory properties of foods. The reversed-flow gas chromatography (RF-GC) technique can also be used to study the release of aroma compounds from other substrates as well as from mixtures of these substrates in different proportions. This knowledge can lead to improved product quality, better taste stability during storage, and more controlled release of flavorings during food consumption.