Comparison of Vacuum and Atmospheric Deep-Fat Frying of Osmo-Dehydrated Goldenberries

Abstract

Colombian goldenberries that do not reach sufficient quality for export are exposed to waste, so the search for processes that provide added value while guaranteeing the conservation of this fruit is paramount. Thus, snacks by vacuum frying from goldenberries (Physalis peruviana L.) with low export quality were made. Goldenberry slices previously subjected to ultrasound-assisted osmotic dehydration were used for this purpose. Response surface methodology with different levels of temperature (110 °C and 130 °C), vacuum pressure (0.3 bar and 0.5 bar), and time (2 min and 6 min) was used to optimize the process. At optimal vacuum frying conditions (i.e., 108 °C, 0.5 bar, and 5.5 min), snacks with lower (p ≤ 0.05) oil content were produced, compared to atmospheric frying chips. The optimized snacks had 9% oil, 7% moisture, ΔE of 13 (with respect to fresh fruit), a_w of 0.3, and hardness of 14 N. The kinetics and modeling of moisture loss and oil uptake were performed under optimal conditions, obtaining the best fit with the Page (R² = 99%) and the first-order (R² = 96%) models, respectively. There was a clear correlation between oil uptake and moisture loss, as the highest oil retention in the product took place when the product had lost the greatest amount of water; therefore, the low initial moisture in the product due to pretreatment resulted in lower oil uptake in it. The obtained goldenberry snack showed adequate physicochemical properties, and the pretreatment yielded a product with much healthier characteristics (i.e., lower oil content, and therefore, a lower caloric intake); so, the proposed process could represent an alternative to the processing of low-export quality Colombian goldenberries.

1. Introduction

Snacks are one of the fastest-growing sectors of the food market and their consumption has increased substantially in recent years in all age groups around the world [1]. Even though the consumption of unhealthy and energy-dense snacks is an important risk factor for obesity and the development of cardiovascular diseases, consumers continue to prefer this type of snack. This is due to the organoleptic attributes such as crunchiness, golden color, palatability, flavor, and an aroma characteristic of fried snacks, which are caused by the oil and other compounds generated during the manufacturing process, which other alternative preparation methods fail to mimic [2]. Yet, in recent years, consumer eating patterns have also been drastically affected by changes in people’s lifestyles, lack of time to prepare a proper meal, and constant information on the influence of diet on health status [3]. Consumer trends toward healthier foods require alternative strategies to traditional fried snack foods, which are rich in simple sugars, saturated lipids, and salt. One option is fruit-based fried chips, which have good sensory and nutritional properties [4].

Vacuum frying is a promising alternative to deep-fat frying because, unlike the latter, the food is processed hermetically at pressures below atmospheric pressure, thus requiring lower processing temperatures to obtain products with similar organoleptic attributes [5]. Low process temperatures (i.e., 90 °C–140 °C) and the reduction of oxygen confers advantages such as flavor preservation, less nutrient degradation, as well as protection of oil quality and reduction of undesirable chemical reactions such as the generation of acrylamides [6]. Vacuum frying has been widely used to preserve other fruits and their quality [6,7,8,9,10]. Along with vacuum frying, osmotic dehydration as a pretreatment has been shown to improve product quality [11]. In addition, the osmotic dehydration pretreatment combined with ultrasound helps to remove some of the water from the raw material, improving its stability during storage and marketing [12].

The goldenberry or cape gooseberry (Physalis peruviana L.) belongs to the Solanaceae family and the genus Physalis, is native to the South American Andes, and has more than eighty varieties found in the wild that are characterized by their sugary fruit enclosed in a calyx [13]. Goldenberry has great nutritional value and is reported to have a considerable content of bioactive compounds such as polyphenols [14]. Colombia has been recognized worldwide as a leading exporter of goldenberry [15], followed in South America by Peru and Ecuador [16]. The fresh goldenberry is exported to international markets such as the Netherlands (71%), the United States (11%), and Germany (6%); however, there is a high percentage of non-exportable quality fruit that requires the development of high-value-added processing alternatives [17].

Vacuum frying could have growth potential in the healthy snacks market and represents an opportunity to diversify the exportable supply of exotic Andean fruits such as goldenberries, creating added value to these crops. Therefore, the objective of this study was to identify the best vacuum frying process conditions to produce healthy and quality goldenberry snacks, as well as to adjust the kinetics of moisture loss and oil uptake during vacuum frying to empirical mathematical models.

2. Materials and Methods

2.1. Raw Material

Colombia ecotype goldenberries with low export quality from a farm in the municipality of Gualmatán (Nariño, Colombia) were used at a commercial maturity stage between 4 and 5 [18]. The raw material was washed with distilled water and cut into (2.0 ± 0.1) mm thick slices to improve mass transfer and avoid difficulties in frying due to wax and impermeability of the fruit skin [19]. Finally, the oil used for frying was a commercial blend of refined vegetable oils (i.e., soybean oil and palm olein).

2.2. Pretreatment

Goldenberry slices were first subjected to ultrasound-assisted osmotic dehydration. For this, the slices were immersed 60 min at 37 kHz and (50 ± 2) °C in ultrasound equipment (Elmasonic S10, Elma Schmidbauer GmbH, Germany) containing a 70°Bx sucrose solution (Food Grade, Sigma-Aldrich, Colombia) with distilled water, using a sample: solution ratio of 1:5 w/w. Finally, the mixture was manually stirred every 5 min to avoid surface sugar crusting [20].

2.3. Frying

The experiments were carried out in Pasto (2527 MASL, 14 °C, 70% humidity, 0.7873 bar; Nariño, Colombia). Vacuum frying was carried out at 110 °C and 130 °C, 0.3 bar and 0.5 bar, and 2 min and 6 min in a homemade prototype manufactured by Centricol^® LTDA (Colombia). As can be seen in Figure 1, the vacuum fryer was a 6 L stainless-steel vacuum chamber connected to a vacuum and condensation system that collected the water vapor removed from the sample.

200 g of osmodehydrated goldenberry slices were used for each test, with a sample: oil ratio of 1:10 w/v. The samples were fed into a basket fixed to the lid handle and driven by a pulley system. The fryer and vacuum valve were then closed. Once the treatment conditions were achieved, the sample was completely submerged in hot oil and the vacuum pump was left on to extract the steam generated, while an automatic system kept the vacuum pressure in the system. Once the frying process was completed, the basket was lifted, the vacuum valve was opened to equalize the pressure, and the frying chamber was opened. Then samples were taken out of the fryer, placed on absorbent paper, and allowed to cool. Finally, samples were packed in aluminum bags and stored at room temperature for subsequent analysis.

For comparative purposes, traditional deep-fat frying was performed using the same equipment and oil: sample ratio, but without vacuum generation.

2.4. Analysis

Effects of process conditions on moisture and oil content, color, water activity (a_w), color change (∆E), and hardness were studied (n = 3 × 2).

2.4.1. Moisture and Oil Contents

Both oil uptake and moisture loss were monitored according to the official methods of analysis approved by the AOAC [21]. The moisture (method 925.10) was measured in an oven (UN 110, Memmert, Germany) at 105 °C. Meanwhile, oil was measured by Soxhlet extraction (method 999.13) with ethyl ether (99.8%, Sigma-Aldrich, Colombia) and 8 h reflux.

2.4.2. Color

Color difference (ΔE) between before (0) and after (f) frying was calculated with Equation (1). CIELab units (L*, a*, and b*) were measured using the D65 illuminant and an observation angle of 10° with a CM-5 spectrophotometer (Konica Minolta Sensing Americas^® Inc., USA).

$$∆E=\sqrt{{{(L_0^{\ast }-L_t^{\ast })}^2+(b_0^{\ast }-b_t^{\ast })}^2+{(a_0^{\ast }-a_t^{\ast })}^2}$$

(1)

2.4.3. Water Activity (a_w)

a_w was measured with a HygroLab C1^® dew point hygrometer (Rotronic AG, Bassersdorf, Switzerland).

2.4.4. Hardness

Snack hardness was measured using the method proposed by Da Silva and Moreira [7]. A Lloyd LS1^® texturometer (Test and Calibration Instruments Ametek, Lloyd Materials Testing, USA) was used. The sample was placed between two supports and the load was applied centrally. Each sample was analyzed using 16 mm between supports, a 3 mm diameter circular tipped rod, a preload of 0.1 N, and 20 mm/min. Hardness (N) was the parameter used to compare the crispiness of goldenberry snacks, where a low breaking force was considered as a product with high crispiness [22]. To reduce the variability associated with the shape heterogeneity of the samples, 10 measurements were made for each treatment, and the arithmetic mean was calculated.

2.5. Experimental Design

A rotating composite central design (CCD) and response surface methodology (RSM) were used to optimize the frying process. The experimental factors were temperature (110 °C and 130 °C), vacuum pressure (0.3 bar and 0.5 bar), and time (2 min and 6 min). Each factor had a central point and two axial points. The overall design included eight factorial points, six axial points, and six replicates at the central point. The experiment was carried out in triplicate for a total of 60 trials. This design was used to determine the levels of the experimental factors that contributed to obtaining the desired healthy and quality attributes for goldenberry snacks; that was, those that minimized the values of a_w, moisture content, oil content, color change (ΔE), and hardness.

2.6. Kinetics Modeling

To model the moisture loss and oil uptake, moisture and oil contents were measured in samples every 30 s at optimal conditions of the vacuum frying process. The modeling assumed that: (1) during the process the oil temperature was constant; (2) the initial water content (t = 0) in the sample was uniform, and (3) the initial oil content (t = 0) in the chip was 0.

Experimental oil content data were fitted to the empirical models of Monod (Equation (2)), Moyano-Pedreschi (Equation (3)), and first-order (Equation (4)):

$$A={\displaystyle \frac{Y\ast t}{b+t}}$$

(2)

$$A={\displaystyle \frac{Y\ast k_1\ast t}{1+{(k}_1\ast t)}}$$

(3)

$$A=Y\ast (1-e^{(-k_2\ast t)})$$

(4)

where A was oil at time t; Y was the equilibrium oil content or the maximum oil content at infinite time; k₁ and k₂ were the oil absorption constants (s⁻¹); t was the oil content during frying at time t; and b was the frying time (s) when half of the peak oil capacity was achieved.

Moisture loss data were expressed in terms of dimensionless moisture ratio (MR), as shown in Equation (5) and used to express the experimental data as a ratio with values between 0 and 1, relating the moisture content at time t with respect to the initial moisture.

$$MR={\displaystyle \frac{M_t-M_{\infty }}{M_i-M_{\infty }}}$$

(5)

where M_t was the moisture at t; M_i was the initial moisture at t = 0; M_∞ was the equilibrium moisture, and t was the frying time. Data were then fitted to the empirical models of Newton (Equation (6)), Henderson-Pabis (Equation (7)), and Page (Equation (8)):

$$MR=e^{(-k_1\ast t)}$$

(6)

$$MR=a e^{(-k_2\ast t)}$$

(7)

$$MR=e^{(-k_3\ast t^n)}$$

(8)

where k₁, k₂, and k₃ were the moisture loss constants (s⁻¹); t was the time of moisture loss (s) during frying; a and n were model constants.

The kinetic parameters were estimated using a nonlinear regression analysis method. The coefficient of determination (R², Equation (9)) and the root mean square error of prediction (RMSEP, Equation (10)) were calculated:

$$R^2=1-{\displaystyle \frac{{\sum }_{i=1}^N{(D_{ex}-D_p)}^2}{{\sum }_{i=1}^N{(D_{ex}-{\overline{D}}_e)}^2}}$$

(9)

$$RMSEP=\sqrt{{\displaystyle \frac{1}{N}}\ast \sum _{i=1}^N{(D_{ex}-D_P)}^2}$$

(10)

where R² was the coefficient of determination; D_ex was the experimental value; D_P was the value predicted by the model, D_e was the mean experimental value, and N was the number of observations.

2.7. Statistical Analysis

Analysis of experimental data and model fitting were performed using Statgraphics Centurion XIX^® (Statgraphics Technologies, Inc., USA). Measurements were expressed as the mean ± standard deviation. Analysis of variance (ANOVA) was done considering the fitting error in the results obtained. Mean values were considered significantly different when p ≤ 0.05.

3. Results and Discussions

As a result of the ultrasound-assisted osmotic dehydration pretreatment, the moisture content of the goldenberry slices was reduced from (81.4 ± 1.0)% to (25.9 ± 1.1)%. Response variables for each of the vacuum frying experimental conditions are presented in

Table 1. Experimental design and response variables of goldenberry snacks produced by vacuum frying and ultrasound-assisted osmotic dehydration as a pretreatment.

Run	T (°C)	t (min)	P (bar)	Oil (%)	Moisture (%)	L*	a*	b*	ΔE	aw	Hardness (N)
1	110	6.0	0.5	8.4 ± 0.3	8.0 ± 0.0	50.1 ± 0.1	11.3 ± 0.2	35.5 ± 0.2	9.0 ± 0.8	0.28 ± 0.00	24.8 ± 0.6
2	120	4.0	0.4	13.6 ± 0.1	7.4 ± 0.1	49.0 ± 0.7	18.5 ± 0.3	35.5 ± 0.1	8.0 ± 0.1	0.32 ± 0.00	13.6 ± 2.3
3	110	6.0	0.3	13.3 ± 0.2	8.6 ± 0.3	45.9 ± 0.3	14.8 ± 0.2	31.0 ± 1.1	8.5 ± 0.3	0.37 ± 0.01	16.5 ± 1.3
4	110	2.0	0.5	13.1 ± 0.1	8.5 ± 0.0	46.3 ± 0.4	12.8 ± 0.2	40.4 ± 1.1	4.6 ± 1.1	0.31 ± 0.01	8.7 ± 2.2
5	130	6.0	0.5	13.0 ± 0.4	5.8 ± 0.2	33.6 ± 0.4	16.5 ± 0.5	27.9 ± 2.0	14.2 ± 1.5	0.27 ± 0.00	8.3 ± 1.3
6	110	2.0	0.3	9.6 ± 0.2	11.7 ± 0.1	43.2 ± 0.6	11.0 ± 0.3	37.5 ± 1.1	4.3 ± 1.6	0.38 ± 0.00	19.6 ± 0.3
7	120	4.0	0.4	15.4 ± 0.1	7.3 ± 0.1	47.0 ± 1.3	17.2 ± 0.3	31.9 ± 1.5	8.5 ± 0.2	0.30 ± 0.00	13.2 ± 2.1
8	120	0.6	0.4	10.9 ± 0.5	15.8 ± 0.1	40.3 ± 2.3	17.8 ± 0.2	31.6 ± 0.9	8.1 ± 0.2	0.43 ± 0.00	30.0 ± 2.5
9	120	4.0	0.4	14.3 ± 0.1	6.9 ± 0.0	46.9 ± 1.1	16.6 ± 0.3	31.6 ± 1.4	8.6 ± 0.2	0.33 ± 0.00	14.6 ± 1.9
10	120	4.0	0.4	13.5 ± 0.0	6.6 ± 0.1	47.2 ± 1.1	18.5 ± 0.3	32.9 ± 1.5	8.3 ± 0.1	0.33 ± 0.01	14.0 ± 2.5
11	120	4.0	0.2	12.8 ± 0.2	7.7 ± 0.4	57.9 ± 0.6	18.0 ± 0.3	52.7 ± 1.1	20.9 ± 1.6	0.35 ± 0.00	24.9 ± 2.3
12	120	4.0	0.4	13.2 ± 0.2	6.4 ± 0.0	49.7 ± 0.3	18.5 ± 0.4	35.0 ± 0.3	8.8 ± 0.3	0.33 ± 0.00	14.3 ± 2.1
13	120	4.0	0.4	13.6 ± 0.0	6.5 ± 0.1	45.9 ± 1.2	17.3 ± 0.4	31.9 ± 0.4	8.0 ± 0.3	0.32 ± 0.00	14.0 ± 2.4
14	130	2.0	0.3	9.2 ± 0.1	7.8 ± 0.3	50.6 ± 0.2	15.0 ± 0.1	37.7 ± 1.0	8.1 ± 0.1	0.32 ± 0.00	17.8 ± 1.3
15	137	4.0	0.4	11.1 ± 0.0	5.4 ± 0.3	42.1 ± 0.5	22.5 ± 0.4	28.6 ± 2.6	12.7 ± 1.1	0.32 ± 0.00	28.8 ± 2.0
16	120	7.4	0.4	14.7 ± 0.1	6.3 ± 0.3	46.6 ± 0.3	16.6 ± 0.3	32.5 ± 0.3	7.6 ± 1.2	0.31 ± 0.01	20.0 ± 2.7
17	130	6.0	0.3	11.9 ± 0.1	6.8 ± 0.3	33.1 ± 2.1	14.8 ± 0.2	22.2 ± 2.5	19.1 ± 0.6	0.31 ± 0.00	25.2 ± 1.6
18	130	2.0	0.5	10.1 ± 0.0	7.1 ± 0.2	49.5 ± 0.5	22.4 ± 0.2	48.8 ± 1.5	14.2 ± 0.2	0.28 ± 0.00	14.4 ± 1.7
19	120	4.0	0.6	13.4 ± 0.1	4.7 ± 0.1	58.6 ± 3.1	18.5 ± 0.1	36.6 ± 0.2	16.5 ± 0.9	0.29 ± 0.00	11.4 ± 1.5
20	103.182	4.0	0.4	8.7 ± 0.3	9.4 ± 0.2	46.7 ± 1.1	21.0 ± 0.2	40.6 ± 2.1	7.5 ± 0.5	0.40 ± 0.00	19.0 ± 1.3

. Meanwhile,

Table 2. Effect of factors (p value) on the response variables of goldenberry snacks.

Response	Temperature (T)	Pressure (P)	Time (t)	T × P	P × t	T × t
Oil	0.2499	0.5711	0.0129	0.1768	0.0139	0.0325
Moisture	0.0002	0.0012	0.0000	0.1434	0.1149	0.3356
aw	0.0005	0.0004	0.0010	0.0357	0.5517	0.2930
ΔE	0.0000	0.0007	0.0000	0.6760	0.0001	0.0457
Hardness	0.0009	0.0000	0.2254	0.0000	0.0086	0.0003

contains the ANOVA of the effect of the factors on the quality variables analyzed in the goldenberry snacks.

3.1. Effect of Process Conditions on Oil Content

According to Table 2, temperature influenced oil content, a finding that is in agreement with the results of Soto et al. [6]. It was also observed that oil content increased significantly with frying time, with no influence of vacuum pressure; probably due to a higher oil transfer until reaching the equilibrium or saturation stage [23]. The interaction between temperature and time showed that a greater reduction in oil content was achieved at short frying times and low frying temperatures, which could be easily achieved under vacuum conditions.

Furthermore, according to Table 2, vacuum pressure favored oil uptake, possibly due to the sponge effect in the product during the recovery of atmospheric pressure in the vacuum system and the cooling stage [24]. The interaction between vacuum pressure and time showed that the use of long frying times, along with a high vacuum, resulted in less oil entering the snack, suggesting the benefit of using vacuum, a finding that is in agreement with the results of Garayo and Moreira [23].

3.2. Effect of Process Conditions on Water Activity (a_w) and Moisture Content

During the vacuum frying process, goldenberry chips’ moisture decreased, following the same behavior reported by Soto et al. [6]. Water loss started drastically when the sample came in contact with the hot oil due to superficial moisture loss, and decelerated over time as the sample came to equilibrium [25].

As evidenced in Table 2, all the factors studied had a significant effect on the final moisture content of the snacks. Longer times, higher frying temperatures, and higher vacuum resulted in greater moisture loss in the sample due to the inverse relationship between vacuum pressure and water boiling point. Consequently, the water in the product vaporized quicker from the inside of the food; a phenomenon that increased with temperature and time, as previously reported [5,8].

Low a_w ensures the microbiological stability of food products, since a_w < 0.6 greatly inhibits microbial growth [26]. Table 2 shows that temperature, vacuum pressure, and frying time had a significant effect on a_w, ensuring safe products with the proposed vacuum frying process (see Table 1). As expected, higher levels of time, temperature, and vacuum favored a decrease in a_w. In contrast, Villamizar et al. (2012) [8] observed that vacuum had little effect on a_w variation.

3.3. Effect of Process Conditions on Color

According to Table 2, significant variations of ∆E value were observed due to the increase in temperature, with a decrease in b* (see Table 1), which indicated a darkening of the snack. The above could be attributed to browning, caramelization, and pigment degradation reactions [6]. Additionally, Table 1 shows a slight decrease in a*, which was attributed to a slight degradation of carotenoids due to the use of lower temperatures due to the use of higher vacuum conditions, coinciding with what was observed by Dueik et al. [9] with carrots. Color variation increased at higher temperature levels, which agreed with the findings by Soto et al. [6], who described that products processed at high temperatures and short times showed more pronounced color changes compared to the use of low temperatures and long processing times.

3.4. Effect of Process Conditions on Hardness

During the first 90 s of frying, the snacks did not reach the minimum hardness of 8.5 N, a value at which a crunchy texture was evident. At first, the texture was softer due to the loss of cell integrity, reduced cell adhesion, and free diffusion of cell contents throughout the tissue [24]. After a certain time, the texture increased due to the dehydration of the outer cells of the fruit and the formation of a crust [22]. Thus, as can be seen in Table 2, vacuum and temperature affected the hardness of the snacks, since it decreased as temperature increased under conditions of higher vacuum, which conferred greater brittleness and crunchiness to the snack.

The above coincided with that reported by Esan et al. [22] who found that, during frying, temperature yielded faster hardness changes and crust formation, affecting breaking strength. This could be explained by the progressive development of a dehydrated crust that increased the breaking strength, coupled with sugar caramelization and cellulose crystallization.

3.5. Optimization

Oil temperature, vacuum pressure, and frying time influenced the color, texture, moisture, a_w, and oil content of goldenberry chips. It was observed that an increase in frying temperature greatly decreased the moisture content and a_w of the chips, as well as the increase in frying time. As observed in Figure 2, long frying times affected product desirability due to negative effects on variables, such as higher ∆E and oil content. The former could be because a less moist surface may darken or burn more easily [6]. As for the latter, a higher oil content could be due to a higher mass transfer [25].

Figure 2A suggested that short processing times decreased desirability, which was attributed to the fact that the desired hardness characteristics were not achieved in short times, as reported by Soto et al. [6], who observed that adequate hardness in papaya chips was found with a_w from 0.1 to 0.3, due to longer processing. Crunchiness loss in the storage of fried chips was linked to higher water content [27].

Vacuum generation proved to be beneficial in terms of desirability (see Figure 2B,C), which could be due to the ease with which this factor allowed the development of faster and more effective water elimination, as well as color preservation. In addition, vacuum generation allowed the use of lower temperatures than those used in atmospheric frying, which made it possible to obtain a product with more attractive physicochemical and sensory qualities, as well as greater preservation of the oil quality [24].

The process conditions that achieved the highest desirability in terms of lowest moisture content, lowest a_w, lowest oil uptake, lowest color change and adequate hardness were 108.24 °C, 0.543 bar, and 5.53 min. Similar optimum processing conditions were described by Dueik et al. [9] with potatoes, carrots, and apples; Villamizar et al. [8] with mango, Esan et al. [22] with sweet potatoes; and Wexler et al. [27].

Table 3. Physicochemical characterization of goldenberry.

Parameter	Optimized Snack	Osmodehydrated	Fresh
Sample mass (g)	0.4 ± 0.0	0.3 ± 0.1	0.9 ± 0.2
L*	49.6 ± 0.8	48.6 ± 0.8	42.6 ± 1.8
a*	11.8 ± 0.7	13.2 ± 1.0	15.0 ± 1.0
b*	36.6 ± 1.6	38.2 ± 1.8	38.8 ± 2.0
∆E	8.0 ± 1.4	6.6 ± 0.7	-
aw	0.29 ± 0.00	0.76 ± 0.01	0.95 ± 0.01
Moisture (%)	7.6 ± 0.4	26.3 ± 0.8	81.5 ± 0.9
Hardness (Newtons)	13.8 ± 2.4		-
Oil (%)	8.6 ± 0.1		-

shows the physicochemical characteristics of the goldenberry snacks obtained under the optimized conditions of the vacuum frying process in this work, compared to those of the osmodehydrated and fresh fruits. Table 3 shows the variation of some physicochemical characteristics from their initial state, through the intermediate pretreatment and final characteristics after vacuum frying. As for the sensory qualities of the snacks (not evaluated by the panel), a golden color and a crunchy structure stood out.

The optimized goldenberry chips in this study had a lower oil content and higher moisture content than those reported by Garayo and Moreira [23] on French fries, who used more drastic conditions, and therefore higher energy consumption (i.e., 0.03 bar and 144 °C). The same happened when compared to Da Silva and Moreira [7], who worked at 0.01 bar and >120 °C. Finally, the hardness value obtained under optimal vacuum frying conditions was similar to that reported by Yamsaengsung et al. [10] in plantain chips.

3.6. Moisture Loss and Oil Uptake Kinetics

Three trials were carried out under optimum process conditions. Samples were taken every 30 s for 5.53 min.

3.6.1. Moisture Loss Kinetics

As observed in Figure 3, the rate of moisture loss expressed in terms of MR was marked during the initial 90 s of the process; and a tendency to equilibrium was observed after 200 s. This behavior could be explained by the progressive product shrinkage because of microstructural stresses due to the migration of free water from the porous to the snack surface, which made the pores smaller, hindering the exit of residual moisture from the food from the inside [10]. Additionally, the formation of a crust with low thermal conductivity at the surface reduced heat transfer at the crust-food interface, which in turn reduced moisture loss from the interior of the food [25].

As shown in

Table 4. Goodness-of-fit kinetic models used to predict moisture loss during snack processing by vacuum frying.

Newton		Henderson-Pabis		Page
${\mathit{R}}^{\mathbf{2}}$	RMSEP (%)	${\mathit{R}}^{\mathbf{2}}$	RMSEP (%)	${\mathit{R}}^{\mathbf{2}}$	RMSEP (%)
97.66	0.04162	97.85	0.03989	98.95	0.02784

, the three models analyzed showed an adequate fit to the data; however, Page’s model stood out for its higher R² and lower RMSEP conditions suitable for the representation of experimental data [28].

The kinetic parameters obtained by nonlinear regression for the resolution of the Page model were a = 0.78476 and k = 0.0356833. Therefore, the Page equation describing the behavior of cape goldenberry moisture loss kinetics during vacuum frying under optimal conditions was described by Equation (11):

$$MR=e^{(-k_3\ast t^n)}=e^{(-0.0356833\ast t^{0.78476})}$$

(11)

3.6.2. Oil Uptake Kinetics

Oil uptake begins simultaneously with the formation of the crust and the increase in product temperature [5]. Previous authors [6,9] indicated that the general pattern should show a rapid initial increase in oil uptake, followed by a gradually decreasing rate, with a final increase for a longer frying time. However, as shown in Figure 4, the characteristic curve was not reached in this study, probably due to the low initial moisture of the goldenberry (i.e., 26%) caused by ultrasound-assisted osmotic dehydration. The pretreatment yielded a process with less evaporation of surface water, thus reducing the number of capillary spaces left by the water and allowing a rapid formation of surface crust that limited the amount of oil that could be absorbed while controlling the final oil content and absorption during the frying process. All this ensured a product with much healthier characteristics [2].

In parallel, the vacuum process generated the force that made the oil adhere to the product surface and forced it to penetrate the food (i.e., sponge effect), which located the oil mainly near the food surface [24,27]. Once the free water was reduced to a critical level (i.e., 240 s in Figure 4), the oil uptake decreased during the pressurization process, which was due to a negligible difference between the internal pore pressure and the medium before depressurization. Thus, during the depressurization process, the diffusivity of air was much higher than that of oil, so air diffused faster than oil in the pores, thus blocking the passage of oil [5,23].

For the above reason, Mir-Bel et al. [29] suggested that the volume of oil absorbed is inversely proportional to the depressurization rate, so the cooling stage is where the highest oil absorption occurrs, regardless of the use of vacuum during frying. Therefore, it is necessary to drain the oil from the surface to control and reduce oil permeation into the food, which can be achieved by centrifugation before restoring the system pressure [27,30].

As for the kinetics of oil content, as shown in

Table 5. Goodness-of-fit kinetic models used to predict oil uptake under optimal conditions.

Monod		Moyano-Pedreschi		First Order
R2	RMSEP (%)	R2	RMSEP (%)	R2	RMSEP (%)
91.09	0.7042	93.02	0.6824	96.12	0.5087

, the First-Order model showed the highest R² and the lowest RMSEP.

To use the first-order model, it was necessary to establish the oil content at equilibrium. For this purpose, the frying process was executed for 20 min at optimum temperature and vacuum pressure, obtaining an oil content of 9.7% at equilibrium. The kinetic constant for the first-order model obtained by nonlinear regression was k = 0.00576705. Therefore, the equation for this model that described the behavior of the oil uptake kinetics under optimum conditions was the one described in Equation (12):

$$A=Y\ast \left(1-e^{-k\ast t}\right)=9.7065\ast \left(1-e^{-0.00576705\ast t}\right)$$

(12)

3.7. Correlation Between Oil Uptake and Moisture Loss

The comparison between the kinetics of moisture loss and oil uptake shown in Figure 5 shows a correlation between the two, indicating that the phase of highest oil uptake was reached at 180 s of the process, where the moisture tended to equilibrium, a behavior that agreed with that reported by Garayo and Moreira [23] in French fries. Likewise, Ziaiifar et al. [31] indicated that oil absorption increased as moisture content decreased.

The correlation between oil uptake and moisture loss was a consequence of structural changes, which reflected the formation of the crust and the reduction of its permeability [9,30]. As frying progressed, the external surface dried out, losing its lipophobicity, facilitating the adherence of oil to the product, so that the condensed vapor generated a pressure difference between the surrounding medium and the pore during the cooling stage. This pressure gradient facilitated the infiltration of the oil adhering to the surface of the pore space. Consequently, oil absorption stemmed from moisture loss as it determined the volume available for infiltration [9,22]. These observations agreed with those of Mir-Bel et al. [29] and Mariscal and Bouchon [30], who found that the highest oil retention in the product took place when the product had lost the greatest amount of water during the frying process, which was why low initial moisture in the product resulted in lower oil absorption.

3.8. Vacuum vs. Atmospheric Frying

Vacuum frying was performed at optimum vacuum pressure and temperature (i.e., 0.543 bar and 108.24 °C) and 1.5 min. Meanwhile, atmospheric frying was done at 170 °C, without a vacuum, at the same time. Processes were carried out with fresh goldenberry slices, as well as with samples subjected to ultrasound-assisted osmotic dehydration. The oil content obtained by both frying methods and with both samples is shown in

Table 6. Oil content in goldenberry snacks produced by atmospheric vs. vacuum frying.

Sample	Oil (%)
	Atmospheric	Vacuum
Fresh	27.1 ± 1.5	17.2 ± 0.9
Osmo-dehydrated	14.2 ± 1.9	8.3 ± 0.3

.

Table 6 confirmed the relationship between initial moisture content and oil absorption, with a higher percentage of oil absorbed in fresh goldenberries obtained by both frying mechanisms compared to osmodehydrated samples, which agreed with the observations made by Montero-Castillo et al. [5]. Regarding the effect of pressure, a significant reduction (p ≤ 0.05) of 37% in oil content was observed in vacuum frying of fresh samples compared to atmospheric pressure frying, as well as a significant reduction (p ≤ 0.05) of 42% in oil content in vacuum frying of goldenberries treated with ultrasound-assisted osmotic dehydration regarding atmospheric frying. Similarly, Dueik et al. [9] reported a 50% reduction in oil content in carrot chips prepared by vacuum frying compared to conventional frying. However, Wexler et al. [27] observed that the oil content in papaya chips (92–156 g/kg) was in all cases higher than that obtained by vacuum frying (<93 g/kg).

4. Conclusions

Vacuum frying proved to be a viable process for obtaining goldenberry chips with low oil content and water activity, preserving the natural color of the product. Pretreatment with ultrasound-assisted osmotic dehydration improved the final characteristics of the product, yielding a presumably healthier product compared to the conventional deep-fat frying process. The use of lower temperatures prevented decomposition and loss of sensitive components, providing acceptable physicochemical characteristics. The process proposed in this work could be an alternative for processing low-quality fruits for export, being an option for adding value and reducing agro-industrial waste. In addition, the pretreatment can represent a reduction in frying operating costs, since the initial material has lower moisture content.

Mathematical models were satisfactorily adjusted to experimental data of moisture loss and oil uptake kinetics of the vacuum frying process of goldenberry snacks, being suitable for purposes of description, prediction, and control of the process on a commercial scale.

For future research, it would be interesting to carry out, under a holistic approach, an economic study comparing the process proposed in this work with the conventional immersion frying process. Such an analysis should then consider, along with investment and operating costs, the nutritional quality of the final product, its shelf life, and an analysis of the environmental impact of the process. For this analysis, the added value given to the product obtained in this work should be considered; value reflected in the fact that the snack obtained would be healthier as it would have a lower oil content (and therefore a lower caloric intake); in addition to the fact that it could have a longer shelf life due to its lower water activity.