OPTIMIZATION of VACUUM FRYING CONDITIONS of EGGPLANT (Solanum Melongena L.) SLICES by RESPONSE SURFACE METHODOLOGY

OPTIMIZATION OF VACUUM FRYING CONDITIONS OF EGGPLANT (Solanum melongena L.) SLICES BY RESPONSE SURFACE METHODOLOGY

José D. Torres, Armando Alvis, Diofanor Acevedo, Piedad M. Montero and Diego F. Tirado

SUMMARY

Due to the global trend to consume low-fat products, efforts minance (L*) decreased with increasing temperature and frying are being made to reduce oil uptake in fried products. In this time in all processed samples. Oil uptake, color change (∆E) paper, the effect of pretreatments and parameters of the vacu- and breaking force increased significantly with the increase of um frying process on quality attributes of purple creole egg- these same factors. The highest desirability (0.73) was obtained plant slices is presented. A lab-scale vacuum fryer was used in blanched samples treated at 130°C during 210s. The opti- to perform vacuum frying at a maximum pressure of 30kPa. mum values in responses were: moisture 64.77%, oil 3.89%, Numerical optimization was carried out using the response ∆E 16.67, luminosity 82.59, breaking force 0.96N, color 4.65, surface methodology (RSM) through a randomized Box-Behnk- odor 4.06, taste 4.25 and greasiness 4.35. The correlation coef- en experimental design, with three factors: frying temperature ficients R2 for the response variables indicated a good fit of the (120, 130 and 140°C), frying time (120, 210, and 300s) and data to second-order regression models. Vacuum frying is an pretreatment (control, blanching and drying), which were es- alternative to process eggplant slices with low-oil content and tablished through preliminary tests. Moisture content and lu- high acceptability.

Introduction combined with other culinary oil is submitted to high tem- of water, allowing the remov- presentations (Lo Scalzo peratures in the presence of al of water from the product Eggplant (Solanum melon- et al., 2016). water and air, which leads to at a higher rate (Mariscal and gena L.) is a horticultural Frying is one of the most the formation of a wide vari- Bouchon, 2008). species with high nutritional popular and important meth- ety of compounds through Air absence during vacuum value, since it is rich in vita- ods of food processing around hydrolytic, thermic and oxida- frying can inhibit lipid oxida- mins and phenolic compounds the globe (Tirado et al., 2013; tive reactions (Dueik and tion and darkening of fruits in the flesh, and anthocyanins 2015). It can be defined as a Bouchon, 2011). The use of and vegetables (Shyu and in the peel, all with antioxi- special type of cooking by oil vacuum is an alternative to Hwang, 2001; Shyu et al., dant properties (Aramendiz- immersion at a temperature improve quality of products 2005). Therefore, dehydrated Tatis et al., 2010). Eggplant is above the boiling point of during deep fat frying food produced by vacuum one of the most abundantly water (Tirado et al., 2012). (Garayo and Moreira, 2002). frying can exhibit a crunchy grown vegetables in the During the process, foods ex- In the vacuum frying process- texture, suitable color and Colombian Caribbean Region, perience starch gelatinization, es, foods are submerged into flavor, and appropriate nutri- especially in Cordoba, Sucre protein denaturation and other oil in a hermetic system ent retention (Da Silva and and Bolívar departments changes in microstructure, where pressure is decreased Moreira, 2008; Diamante (Correa et al., 2010). It can be physical and organoleptic below atmospheric levels. et al., 2012; Šumić et al., consumed dried, cooked, as properties (Pedreschi, 2012). This makes it possible to re- 2016). Some authors suggest paste, pickled, and fried or During atmospheric frying, duce the boiling temperature that evaporated water from

KEYWORDS / Moisture Loss / Oil Uptake / Optimization / Quality Attributes / Solanum melongena L. / Vacuum Frying / Received: 01/26/2017. Modified: 09/27/2017. Accepted: 10/05/2017.

José D. Torres. Food Engineer, Uni- (Univalle), Colombia. Resear- Salle, Colombia. Specialist in Engineering, Universidad Com- versidad de Cartagena (Unicar- cher, Unicórdoba, Colombia. Food Science and Technology, plutense de Madrid, Spain. tagena), Colombia. M.Sc. Agro- Diofanor Acevedo. Food Engi- UNAL, Colombia. M.Sc. In Researcher, Unicartagena, Food Sciences, Universidad de neer and Pharmaceutical Che- Food Science and Technology, Colombia. Address: Grupo de Córdoba (Unicórdoba), Colom- mist, Unicartagena, Colombia. Universidad de Zulia, Vene- Investigación Nutrición, Salud bia. Ph.D. student in Engi- Specialist in Food Science and zuela. Ph.D. in Sciences, y Calidad (NUSCA), Uni- neering Sciences, Ponti-ficia Technology, Universidad Nacio- Universidad Rafael Belloso cartagena. Ave. del Consulado, Universidad Católica de Chile. nal de Colombia (UNAL), Chacín, Venezuela. Researcher, calle 30 Nº 48-152. Campus Armando Alvis. Food Engineer, Colombia. Ph.D. In Food Engi- Unicartagena, Colombia. Piedra de Bolívar. Cartagena, Universidad INCCA de Colom- neering, Univalle, Colombia. Diego F. Tirado. (Corresponding Colombia. e-mail: dtiradoa@ bia. Specialist in Educational Researcher, Unicartagena, author). Food Engineer and unicartagena.edu.co Management, UNICÓRDOBA, Colombia. M.Sc. in Environmental En- Colombia. Ph.D. in Food Engi- Piedad M. Montero. Food En- gineering, Unicartagena, Colom- neering, Universidad del Valle gineer, Universidad de la bia. Ph.D. student in Chemical

OCTOBER 2017 • VOL. 42 Nº 10 0378-1844/14/07/468-08 $ 3.00/0 683 OPTIMIZACIÓN DE LAS CONDICIONES DE FRITURA AL VACÍO DE RODAJAS DE BERENJENA (Solanum melongena L.) UTILIZANDO LA METODOLOGÍA DE SUPERFICIE DE RESPUESTA José D. Torres, Armando Alvis, Diofanor Acevedo, Piedad M. Montero y Diego F. Tirado RESUMEN Dada la tendencia mundial de consumir productos bajos en nuyeron con el aumento de temperatura y tiempo de fritura grasa se hacen esfuerzos para reducir la absorción de grasa en todas las muestras procesadas. La absorción de aceite, en los productos sometidos a fritura. En este trabajo se pre- cambio de color (∆E) y fuerza de ruptura aumentaron signi- senta el efecto de pretratamientos y parámetros del proceso ficativamente con el incremento de estos factores. Se obtuvo de fritura al vacío sobre los atributos de calidad de rodajas la mayor deseabilidad (0,73) en muestras escaldadas tratadas de berenjenas de la variedad criolla morada. Se utilizó una a 130ºC durante 210s. Los valores óptimos en las respuestas freidora a escala de laboratorio para llevar a cabo la fritura fueron: humedad 64,77%; aceite 3,89%; ∆E 16,67; luminosi- al vacío a presión máxima de 30kPa. Se realizó optimización dad 82,59; fuerza de ruptura 0,96N; color 4,65; olor 4,06; sa- numérica utilizando la metodología de superficie de respuesta bor 4,25; y grasosidad 4,35. Los coeficientes de correlación (MSR) a través de un diseño experimental Box-Behnken alea- R2 para las variables respuestas indicaron buen ajuste de los torizado, con tres factores: temperatura de fritura (120, 130 y datos a los modelos de regresión de segundo orden. La fritu- 140ºC), tiempo de fritura (120, 210 y 300s) y pretratamiento ra al vacío es una alternativa para procesar y obtener roda- (control, escaldado y secado), establecidos en ensayos preli- jas de berenjena con bajo contenido de aceite y alta acepta- minares. El contenido de humedad y luminosidad (L*) dismi- bilidad sensorial.

OPTIMIZAÇÃO DAS CONDIÇÕES DE FRITURA A VÁCUO DE RODELAS DE BERINJELA (Solanum melongena L.) UTILIZANDO A METODOLOGIA DE SUPERFÍCIE DE RESPOSTA José D. Torres, Armando Alvis, Diofanor Acevedo, Piedad M. Montero e Diego F. Tirado RESUMO Devido à tendência mundial em consumir produtos baixos em luminosidade (L*) diminuíram com o aumento de temperatura e gordura, são realizados esforços para reduzir a absorção de tempo de fritura em todas as amostras processadas. A absorção gordura nos produtos submetidos a fritura. Neste trabalho se de óleo, mudança de cor (∆E) e força de ruptura aumentaram apresenta o efeito de pré-tratamentos e parâmetros do proces- significativamente com o incremento de estes fatores. Obteve-se so de fritura a vácuo sobre os atributos de qualidade de rode- a maior desejabilidade (0,73) em amostras escaldadas tratadas las de berinjelas roxa. Utilizou-se uma fritadeira em escala de a 130°C durante 210s. Os valores óptimos nas respostas foram: laboratório para realizar a fritura a vácuo na pressão máxima umidade 64,77%; óleo 3,89%; ∆E 16,67; luminosidade 82,59; de 30kPa. Realizou-se optimização numérica utilizando a meto- força de ruptura 0,96N; cor 4,65; olor 4,06; sabor 4,25; e gor- dologia de superfície de resposta (MSR) através de um desenho dura 4,35. Os coeficientes de correlação R2 para as variáveis experimental Box-Behnken aleatorizado, com três fatores: tem- respostas indicaram bom ajuste dos dados aos modelos de re- peratura de fritura (120, 130 e 140°C), tempo de fritura (120, gressão de segunda ordem. A fritura a vácuo é uma alternativa 210 e 300s) e pré-tratamento (controle, escaldado e secado), es- para processar e obter rodelas de berinjelas com baixo conteú- tabelecidos em ensaios preliminares. O conteúdo de umidade e do de óleo e alta aceitabilidade sensorial.

food during deep frying is to- ments are useful to improve several independent varia- Materials and methods tally replaced by oil that is and control oil uptake, obtain- bles (Esan et al., 2015; Abtahi accumulated in the crust, ing suitable textures, promoting et al., 2016; Yuksel and Sample preparation and which would be the way how sugar lixiviation and air re- Kayacier, 2016). storage conditions absorption takes place but, so moval from tissues. It is im- So far, no vacuum frying far, the mechanisms have not portant to have adequate infor- process for eggplant has been Fresh eggplant (S. melonge- been fully clarified (Zhang mation and methods for moni- reported in the literature, de- na L.), variety creole purple et al., 2016). An alternative to toring the process and to en- spite the importance of this with yellow seeds, character- reduce surface oil uptake in sure the quality of products type of research for the devel- ized in a previous work by vacuum-fried products is the under vacuum frying. Nu- opment of healthy low-oil con- Aramendiz-Tatis et al., (2010) use of pretreatments such as merical optimization by means tent foods. This becomes of was used in this study. The blanching, vacuum or micro- of the response surface meth- special interest since this vege- fruits were purchased in the wave drying, osmotic dehydra- odology (RSM) makes it possi- table is part of the gastronomic Central Market of Cartagena tion, freezing and edible coat- ble to find appropriate condi- culture of the Colombian de Indias, Colombia, selected ings, and post-treatments in- tions for food processing with- Caribbean Region. The main according to size uniformity cluding hot air drying and out increasing the cost of ex- objective of this study was to and verifying that there were centrifugation (Nunes and periments. It also improves optimize the vacuum frying no diseases or external defects Moreira, 2009). Pedreschi processes in which the desired conditions of pretreated egg- due to mishandling. The raw (2012) claims that pretreat- responses are influenced by plant slices by RSM. material was stored at 4ºC in a

684 OCTOBER 2017 • VOL. 42 Nº 10 refrigerator until the next day. process or ‘equivalent thermal Ltd., UK) coupled with Texture independent variables was im- The initial moisture content of driving forces’: ∆T1= 50ºC, Expert Exceed software version plemented. Each variable had samples (86.54%) was mea- ∆T2= 60ºC and ∆T3= 70ºC. 2.64 was used. This equipment three levels and three center sured by oven drying at 105ºC Therefore, the oil temperatures had a 50kg load cell and a ve- points per block. The factors according to AOAC 930.15 selected were 120, 130 and locity head of 5mm·s-2. These were set as frying temperature (AOAC, 2005). Afterwards, 140ºC. Frying times of 120, values were selected from pre- (X1), frying time (X2) and eggplants were washed and cut 210 and 300s were established liminary tests. Samples were pretreatment (X3). The com- in slices with diameters of from preliminary tests. The oil placed on two parallel supports mercial statistical software 3.8cm ±0.2cm and thickness was initially heated to the es- with a distance of 3cm be- Statgraphics Centurión (ver- ~1.5 ±0.1cm using a Hobart® tablished frying temperature, tween them. A third parallel sión 16.2.04; StatPoint Tech- cylindrical cutter (model FP100- then eggplant slices were axis made from the same sup- nologies Inc., EEUU) was 1B; Hallde Co., Sweden). Palm placed into the stainless-steel port material was displaced used to perform the analysis. oil, purchased at a local super- basket, the system was cov- vertically, exerting a stress un- Levels of each variable were market the day before the ex- ered, and the vacuum pump til completely breaking the established from preliminary periments, was used in this re- was activated. The basket was structure of the fried samples. tests and coded as Table I search because of its resistance submerged into the oil once Essays were carried out six shows. A total of 15 experi- to oxidation during heating. the equipment had reached op- times for each treatment, al- mental runs were carried out erating pressure. A 1:20 w/v ways on the central area in to study the effect of these Pretreatment of eggplant ratio of product/oil was used. order to avoid variation in the variables on the responses of slices Once the frying time was results due to anatomic loca- Y1: moisture loss (%), Y2: oil reached, the basket was re- tion (Da Silva and Moreira, uptake (%), Y : lightness (L*), Control. The samples were not 3 moved from the oil and the 2008). Y : change of color (∆E), Y submitted to pretreatment prior 4 5: vacuum pump ran for an addi- breaking force, Y : color, Y : to the vacuum frying process. 6 7 tional minute. Vacuum was Sensory analysis odor, Y8: flavor, and Y9: Blanching. Samples were sub- then interrupted and the equip- greasiness. Regression equa- merged in a temperature con- ment stopped in order to re- Panelists were encouraged to tions were obtained by adjust- trolled water bath (Tectron- move the samples. The egg- use a 5 point hedonic scale to ing experimental data to sec- Bio-20; Instrumentation Scien- plant slices were drained in a rate their perception of the ma- ond-level polynomial models tific Technologies), keeping metal net packing them in low terial according to color, odor, by least squares. Statistical the temperature at 90ºC for density polyethylene bags for flavor and greasiness. The significance of each term in 2min. The electronic device further analysis. scale consisted of categories the regression equations was was coupled to a pair of ther- ranging from ‘I like it very examined by ANOVA for each mocouples (type J, stainless Quality analysis of vacuum- much’ (5), going through neu- response. All the variables steel) of 0.25mm diameter to fried eggplant slices tral (3), until ‘I do not like it at from the vacuum frying pro- control water temperature and all’ (1). In order to rate the cess were optimized using the Moisture and oil content. The the center of eggplant slices samples, a panel of 30 tasters RSM numerical method, based average moisture and oil con- (±0.05ºC). Before vacuum fry- was used. Ages of the panelist on the convenience concept to tent of vacuum fried eggplant ing, water on the surface of group were 19 to 45 years and obtain the best slices of pro- slices were calculated accord- the product was removed us- the group included 15 women cessed eggplants: i) minimiza- ing to procedures 930.15 and ing absorbent paper. and 15 men. Panelists cleaned tion of moisture content, oil 920.39 (AOAC, 2005), respec- their palates by deionized wa- uptake, color change and Drying. It was carried out in a tively. Measurements were car- ter prior to proceeding to the breaking force and ii) maxi- vacuum oven with a tray load ried out by triplicate and ex- -2 next sample. They were sup- mization of lightness and of 1.75kg·m . The samples were pressed on a dry basis. plied with whole samples of sensory perceptions quality. then submitted to a temperature pretreated and control vacu- Prediction models were used of 60ºC with an air velocity of Instrumental evaluation of col- -1 or. Portions <5g of vacuum- um-fried eggplant slices to generate surface plots. 1.8 ±0.11m·s . Weight loss was (Yuksel and Kayacier, 2016). Results of each response vari- periodically controlled until the fried eggplant slices were cut and used for this stage. A Data was gathered in a spread- able were expressed as the slices achieved an average final sheet and transformed into nu- mean and standard deviation, moisture content of 0.7kg water/ colorimeter CR-5 (Konica Minolta Sensing, USA), with merical values for further and were in turn compared kg dry solid. analysis. using Tukey’s HSD test with a illuminant D65 and a tone an- Vacuum frying. Frying of pre- gle 10º was used under the level of significance of 5%. treated and control eggplant CIEL*a*b* scale. Complete Experimental design and Eq. 1 represents the model slices was performed using a details of this procedure were statistical analysis used to analyze responses of Gastrovac® (International Coo- described in a previous report the experimental design as a king Concepts International, (Torres et al., 2017). A randomized Box-Behnken function of the three indepen- Spain) with dimensions of experimental design with three dent variables: 40×26×46cm, with a maximum Breaking force. The vacu- capacity of 10.5L and 220V. um-fried eggplant slices were TABLE I The equipment reached a max- submitted to breaking-force LEVELS OF VARIABLES USED IN THE imum vacuum pressure of test. Slices were first weighted BOX-BEHNKEN EXPERIMENTAL DESIGN 30kPa. Water boils at 70ºC at in an electronic balance ma- this pressure. According to chine (model Gr-200; A&D Co. Variable -1 (low) 0 (central) 1 (high) Mariscal and Bouchon (2008), Ltd., Japan) and then a textu- Temperature (ºC) (X1) 120 130 140 3 deltas were used to define rometer (model TA.TX2i® Time (s) (X2) 120 210 180 Pretreatment (X ) Control Blanching Drying temperatures of the frying plus; Stable Micro System Co. 3

OCTOBER 2017 • VOL. 42 Nº 10 685 Y = β +β X +β X +β X + However, slices evidenced a ficial crust could have formed condensation within its pores 0 1 1 2 2 3 3 significant decrease in mois- and could have prevented exces- after being vacuum-fried. The β X X +β X X +β X X + 12 1 2 13 1 3 23 2 3 ture at 300s. It should be noted sive dehydration of samples. pressure difference between 2 2 β11X1 +β22 X21 +ε that moisture was significantly Finally, Figure 1c represents the surrounding material and reduced (p<0.05) in control the moisture content in eggplant the pore led to the oil uptake samples after vacuum frying at slices dried at 60ºC prior to that was adhered to the sur- where Y: response; β0: inter- 130 and 140ºC, regardless of vacuum frying. Moisture con- face. In the same way, Garayo cept; X1: temperature; X2: time; frying time. A comparison tent in dried samples before the and Moreira (2002) indicated X3: pretreatment; β1, β2 and β3: with samples processed at frying process (70.50%) was that moisture loss in the vacu- linear effect coefficients; β11, 120ºC evidences that tempera- lower than control (86.54%) and um frying of potato chips led β22, β33: quadratic effect coeffi- ture is the most influential fac- blanched (89.50%) samples. to higher oil adherence in the cients; β12, β13, β23: coefficients for the interaction of factors; tor on water loss in the fried However, dehydration was low- surface of the chips. They also and ε: random error. product. This phenomenon est, i.e., the slices had more found that when the product could be caused by the evapo- final moisture content than the reached its free water limit, a Results and Discussion ration of free water from the control samples at 140ºC and lower amount of oil was ab- product due to high tempera- 300s. This was probably sorbed. It has been found that Effect of vacuum-frying tures, which promotes diffusion caused by the fact that thermal lower vacuum pressures pro- temperature and time on (Tirado et al., 2013, 2015). It is treatment contributed to a par- duce higher drying rates quality also worth highlighting the fact tial cooking on surface of the during vacuum frying of pre- that eggplant presents a much eggplant slices, allowing the treated vegetables slices. This Moisture loss. Figure 1 shows softer and less dense histologi- formation of a barrier, which could be attributed to the fact the behavior of the moisture cal structure than other vegeta- prevented water loss during the that by lowering pressures the content of the vacuum-fried bles (Miraei-Ashtiani et al., frying process. These results boiling point of water is re- eggplant slices at different ex- 2016). The soft texture of this were similar to those obtained duced and, therefore, water perimental conditions (tempera- structure might have influ- by Nunes and Moreira (2009), within the vegetables slices tures, times and pretreatments). enced the final moisture con- who reported that moisture loss starts vaporizing in a faster It was found that moisture de- tent of the slices after being during vacuum frying in pre- way (Moreira et al., 2009). creased with increasing frying submitted to vacuum frying. treated mango chips caused Results from the present work temperature, which presented On the other hand, Figure 1b contraction and volume loss; it agreed also with those report- a significant effect (p<0.05), shows the moisture data for also produced changes in tex- ed by Shyu and Hwang (2001) as frying time did. Figure 1a bleached eggplant slices at dif- ture, such as crunchiness and regarding the vacuum frying illustrates control samples be- ferent experimental conditions. hardness, which in turn influ- of apple chips, by Shyu et al. havior. As Figure 1a shows, The final moisture content of ence dehydration rates of pro- (2005) in relation to vacu- dehydration was larger in the these samples was found to be cessed food. On the other um-fried carrot chips and control samples, with values greater at 140ºC and 300s, hand, Diamante et al. (2012) by Esan et al. (2015) regard- <10% at 140ºC and 300s, with compared to control samples. found that moisture content in ing vacuum-fried yellow statistically significant differ- This indicated that blanching gold kiwi fruit slices should be fleshed sweet potato (Ipomoea ences (p<0.05). No statistically caused water retention in the controlled in order to guaran- batatas L.) significant differences were sample in the form of residual tee a suitable formation of observed at temperature of moisture. Partial cooking crust during deep fat frying. Oil uptake. Figure 2 presents 120ºC regarding moisture of the product structure after Similarly, Oginni et al. (2015) the results of oil uptake in pre- content in samples processed bleaching may have caused reported that cassava-based treated eggplant slices and con- at 120 and 210s (p>0.05). such a phenomenon. A super- snacks experienced vapor trol samples. Oil content was

Figure 1. Effect of temperature and frying time on moisture loss of eggplant slices. : 120ºC, : 130ºC and :140ºC. Means ±standard deviation of three replicates.

686 OCTOBER 2017 • VOL. 42 Nº 10 Figure 2. Effect of temperature and frying time on oil uptake during vacuum frying of eggplant slices. : 120ºC, : 130ºC and :140ºC. Means ±standard deviation of three replicates. found to be always higher with temperature and frying time. temperatures and frying times, statistically significant (p<0.05), increasing temperatures and They indicated that oil uptake the lowest levels of L* were especially for slices processed frying times (p<0.05). Final oil was significantly related to the observed in dried samples, fol- during 120s at 120ºC. Results content in control samples was final moisture content. lowed by control slices and obtained with respect to ΔE in higher than in slices submitted However, this conclusion does then those that experienced control slices and vacuum pre- to blanching and drying at all not match the results obtained blanching. Eggplant slices fried treated slices agreed with those temperatures and frying times. by Zhang et al., (2016), who at 140ºC for 300s under all reported by Garayo and This could have been caused studied the potato frying pro- treatments exhibited the lowest Moreira (2002) and Esan et al. by the high porosity of egg- cess and reported that the final L* values. These results indi- (2015). These authors indicated plant, and the greater availabil- moisture content exerted no cate that non-enzymatic dark- that development of color ity of spaces in which free statistically significant effect ening reactions took place at a during vacuum frying takes water was kept in the samples. on oil uptake. The present higher rate when the process place after the product has ex- This water was vaporized by study provided new, convincing time and temperature was in- perienced a proper drying pro- the high temperature and then information that aims to clarify creased, regardless of whether cess. On the other hand, Dueik escaped allowing higher oil the existing relation between or not pretreatment is applied. and Bouchon (2011) found that inlet. Blanched and dried slices initial moisture and oil uptake On the other hand, Figure 4 time and temperature are the exhibited lower oil absorption during the frying process. So illustrates color changes (ΔE) factors with the greatest influ- rate under all experimental far, this relation has not been experienced by control and ence on ΔE. This phenomenon conditions of time and tem- completely elucidated and it pretreated eggplant slices. The is caused by Maillard reac- perature. In the case of drying, represents the basis for intro- most notorious ΔE were ob- tions, which depend on the the low initial water content in ducing a specific pre-drying served in both control and content of reducing sugars and samples could be responsible strategy into the frying pro- dried samples fried at 140ºC amino acids or proteins on the for this behavior. During cesses of the food industry. during 300s. Slices subjected to surface. Shyu et al. (2005) ob- bleaching at 90ºC for 2min, blanching showed a lower vari- served that ΔE of carrot chips perhaps partial cooking was Color parameters. Figure 3 ation of ΔE, which was increased with increasing promoted in the eggplant struc- shows the effect of temperature ture, sealing the pores around and time on lightness (L*) the matrices, which in turn changes in vacuum-fried egg- could have created a protective plant slices. On the CIEL*a*b barrier that prevented excessive scale, Lightness represents a oil absorption. The oil uptake numerical parameter that indi- process within the eggplant cates clarity at values close to slices was similar to those re- 100º and darkness when values ported by Shyu and Hwang tend to 0º. This scale is used (2001) in apple chips, Reis as an indicator of how color et al. (2008) in potato sticks reactions develop in processed and Moreira et al. (2009) in foods. In this work, it was apple chips and vacuum-fried found that all samples and pre- potatoes. treatments involving high tem- Finally, Shyu et al., (2005) peratures and long times observed that oil content in showed a significant decrease Figure 3. Effect of temperature and vacuum frying time on lightness of carrot chips increased with in L* (p<0.05). At the same eggplant slices. Means ±standard deviation of three replicates.

OCTOBER 2017 • VOL. 42 Nº 10 687 TABLE II EFFECT OF TEMPERATURE AND VACUUM FRYING TIMES ON BREAKING FORCE OF SLICES EGGPLANT Control Time (s) 120ºC 130ºC 140ºC 120 0.96 ±0.07 a 1.73 ±0.36 c 1.85 ±0.22 c 210 1.29 ±0.12 b 1.77 ±0.17 c 2.07 ±0.18 c 300 1.47 ±0.21 b 1.88 ±0.23 c 2.19 ±0.15 c Blanching Time (s) 120ºC 130ºC 140ºC 120 1.01 ±0.17 a 1.43 ±0.12 b 1.85 ±0.14 c 210 1.55 ±0.13 b 1.71 ±0.09 c 1.87 ±0.11 c 300 1.88 ±0.16 c 1.92 ±0.25 c 1.96 ±0.06 c Drying Time (s) 120ºC 130ºC 140ºC 120 1.24 ±0.08 b 1.39 ±0.02 b 1.91 ±0.03 c 210 1.42 ±0.05 b 1.64 ±0.19 c 2.26 ±0.11 c 300 1.76 ±0.14 c 1.94 ±0.27 c 2.43 ±0.07 c Data indicate the mean ±standard deviation of the six replicates. Different letters in the same column indicate statistically significant differences (p≤0.05), according to Tukey’s HSD test.

Figure 4. Effect of temperature and vacuum frying time on color chan- tistically significant increase variables in the experimental ges of eggplant slices. Means ±standard deviation of three replicates. (p<0.05) with increasing time design: moisture loss, oil con- and temperature. This was at- tent, lightness, color change, tributed to changes experienced breaking force and sensory temperature (between 70 and 2012). Šumić et al. (2016) re- by the product, such as partial perception (color, odor, flavor 110ºC) and frying times (be- ported that the evaporation pro- cooking of structure and a high and greasiness) after the vacu- tween 300 and 1800s). Varia- cess of water molecules causes moisture loss. Harder, crunchier um frying process of eggplant tion of these parameters be- a volume reduction by dehydra- eggplant slices were obtained at slices. On the other hand, came evident at 100ºC. This tion of the microstructure and the end of the process thanks to Table IV shows the regression was attributed to the instability increases structural porosity. this set of conditions. The ob- coefficients obtained for each of carotenoids above 100ºC. Numerous holes, clefts and tained results are consistent second-order polynomial to cracks appear at the crust ma- with those observed by Dueik which the experimental data Breaking force. Table II inclu- trix, and they are responsible of et al. (2010) when they studied was adjusted. Their respective des the variations of breaking the higher surface hardness of texture changes in carrot chips. coefficients of determination force (N) in the vacuum-fried the product. Garayo and Mo- They reported an initial soften- (R2) are also shown, as calcu- eggplant slices. It was observed reira (2002) and Da Silva and ing, which was followed by a lated by analysis of variance. that control, blanched and Moreira (2008) observed that final hardening as a product of R2 ranged from 0.75 for ∆E to dried samples exhibited a sig- vacuum-fried potato chips pre- the progressive development of 0.99 for the breaking force. nificant increase in the re- sented higher contraction than a dehydrated crust. At the be- The statistical analysis indicat- sponse of breaking force atmospherically fried potato. ginning of the frying process, ed that regression models were (p<0.05) at high values of time This seems to indicate that the texture could become softer due adequate to explain the data. and temperature. Such phenom- crust is easily formed at atmo- to the combined effects of loss These models are of a predic- enon could have been caused spheric pressure, and it prevents of cell integrity, free diffusion tive type and can be used to by a large formation of crust the achievement of higher con- of cell content in the tissue and foretell responses according to due to microstructural changes traction levels in potato slices. reduction of cell adhesion. Once different changes in factors in the tissues of fried products The determination of the a certain amount of time had levels. (Garcia-Segovia et al., 2016). breaking force represents a passed, food would increase its The moisture content of egg- Also, these changes in hard- measure of how crunchy a chip hardness due to the dehydration plant slices was significantly ness are attributed to dehydra- is. Low values of breaking of external cells and formation affected by temperature (X1), tion (Oginni et al., 2015). force indicate a high level of of a crust. A final stage of op- followed by pretreatment (X3) The texture resulting after this texture attribute. In their timization was performed to the and frying time (X2). This in- the vacuum frying process is a research, Shyu and Hwang processing conditions in order dicated that temperature chang- consequence of changes experi- (2001) and Shuy et al. (2005) to establish the best physical es and previous treatments enced by food composition observed that by increasing the and sensory responses of vacu- were the main factors, which (Dueik and Bouchon, 2011). oil temperature, lower values of um-fried eggplant slices. need to be controlled during Proteins and carbohydrates were breaking force were obtained. vacuum frying eggplant slices, mainly modified due to the heat The same happened when in- Numerical optimization of the since the product experiences transferred to the sample (Esan creasing vacuum level. This vacuum frying of eggplant dehydration at a high rate. The et al., 2015; Zhang et al., 2016). tendency was not evidenced in slices estimated effects of the model This caused the formation of a the results obtained in the cur- also indicated that the quadrat- crust, which provided crunchi- rent study. In fact, the values of Table III presents the averag- ic interactions of tempera- ness in the product (Pedreschi, breaking force showed a sta- es obtained for the response ture-time (X1 and X2) and

688 OCTOBER 2017 • VOL. 42 Nº 10 TABLE III CONDITIONS USED FOR THE BOX-BEHNKEN EXPERIMENTAL DESIGN PERFORMED ON VACUUM FRYING PROCESS AND RESPONSE VARIABLES Run Temp Time Pre-treatment Moisture (%) Oil (%) Lightness ∆E Breaking force (N) Color Odor Flavor Greasiness 1 -1 -1 0 66.19 32.62 87.39 19.41 1.01 4.98 4.15 4.51 4.26 2 1 -1 0 23.96 15.29 42.49 21.92 1.85 4.82 4.02 4.38 4.09 3 -1 1 0 35.89 22.95 67.37 18.77 1.88 4.76 4.34 4.78 4.15 4 1 1 0 24.79 17.61 44.37 24.93 1.96 4.75 4.29 4.14 3.99 5 -1 0 -1 68.02 36.59 82.19 12.96 1.29 3.78 4.83 3.95 3.22 6 1 0 -1 21.96 27.77 39.45 27.68 2.17 3.63 4.72 3.74 3.06 7 -1 0 1 31.59 15.88 73.28 18.87 1.42 3.29 3.28 3.33 4.46 8 1 0 1 19.79 17.28 46.45 26.12 2.26 2.92 3.03 4.04 4.32 9 0 -1 -1 33.24 25.86 50.54 15.87 1.73 3.45 4.29 3.72 3.11 10 0 1 -1 31.63 28.95 53.55 16.61 1.68 3.15 4.61 3.61 3.35 11 0 -1 1 29.98 15.86 71.08 15.91 1.39 2.08 3.28 3.33 4.69 12 0 1 1 22.23 10.75 70.12 25.65 1.84 2.58 2.86 3.27 4.58 13 0 0 0 54.33 12.27 86.23 13.45 0.19 4.54 3.98 4.78 3.83 14 0 0 0 52.02 16.66 74.65 21.65 1.71 4.48 4.09 4.63 4.21 15 0 0 0 41.19 14.22 56.56 21.15 1.49 4.26 3.76 4.06 4.12

TABLE IV ADJUSTED REGRESSION COEFFICIENTS Moisture Oil Breaking Coefficients (%) (%) Lightness ∆E force (N) Color Odor Flavor Greasiness β0 494.06 664.52 -599.17 406.96 50.02 84.14 43.36 23.84 -5.05 β1 -1.17 -12.39 13.18 -6.48 9.79 -1.29 -0.63 -0.34 0.12 β2 -1.03 0.56 -0.57 -0.15 1.54x10 -2 -5.08x10 -3 3.82x10 -4 2.62x10 -2 -4.66x10 -3 β3 -136.53 4.17 -22.55 20.88 -1.43 5.92 0.97 0.24 1.76 β11 2.0x10 -2 6.04x10 -2 -6.52x10 -2 2.73x10 -2 3.35x10 -3 4.95x10 -3 2.38x10 -3 1.22x10-3 -4.92x10 -4 β12 1.02x10 -2 -5.05x10 -3 6.08x10 -3 1.01x10 -3 -2.11x10 -2 4.16x10 -5 2.22x10-5 -1.41x10 -3 2.77x10 -6 β13 1.31 -7.01x10 -2 0.39 -0.18 -1.03x10 -3 -5.52x10 -3 -3.53x10 -3 2.32x10 -2 5.41x10 -4 β22 -1.51x10 -2 6.23x10-4 -6.84x10-4 -2.42x10 -5 2.59x10 -5 -1.16x10 -5 3.18x10 -6 -1.97x10-4 1.46x10 -5 β23 1.72x10 -2 -2.27x10 -2 -1.12x10 -2 0.025 1.38x10 -3 2.22x10-3 -2.05x10 -3 1.38x10 -4 -9.72x10 -3 β33 -10.55 1.58 -5.61 -4.51x10 -2 0.32 -1.51 -0.21 -0.85 -0.24 R2 0.87 0.82 0.79 0.75 0.99 0.98 0.96 0.86 0.97

temperature-pretreatment (X1 Temperature-time interaction to the second-order regression oil content of 3.89%), ∆E of and X3) were highly significant exerted a highly significant model. Therefore, changes of 16.67, and breaking force of for moisture (p<0.05). R2 of effect on the change of color breaking force that are experi- 0.96N. On the other hand, the the quadratic polynomial was in the pretreated and control enced by the product can be maximized responses were 0.87, indicating a suitable ad- slices. This suggested that properly described by using lightness (82.59), color (4.65), justment of the experimental eggplant tended to rapidly the obtained regression. odor (4.06), flavor (4.25) and data. This model was appropri- change color during deep fat Changes in sensory properties greasiness (4.35). ate for describing the behavior frying. It is therefore neces- such as color, odor, flavor and Finally, Figure 5 shows the of water loss within eggplant sary to implement an adequate greasiness were generally behavior of response surfaces slices. Oil content of vacu- control of time and tempera- highly influenced by tempera- for eggplant slices submitted to um-fried eggplant slices was ture conditions, in order to ture and pretreatments (X1 and blanching during the optimiza- significantly affected by both avoid unwanted sensory X3). The best conditions were tion of vacuum frying. Results linear and quadratic interac- changes in the product. The obtained at 130ºC and a of frying optimization of egg- 2 2 tions of temperature (X1, X12; obtained determination R was blanching pretreatment. R plant are similar to those re- p<0.05). It was also observed 0.79 for lightness and 0.75 for were 0.98, 0.96, 0.86 and 0.97, ported by Akinpelu et al. that quadratic time (X22) exert- ∆E. Acceptable levels of cor- respectively. This showed that (2014), who based on the conve- ed a highly significant effect, relation can thus be achieved data gathered from the senso- nience found frying conditions which indicated that an in- when describing the behavior ry panel successfully adjusted of 133ºC and 6min for banana crease in frying time affected of these responses. Brea- to the quadratic model. The chips. On the other hand, Esan oil uptake. A R2 of 0.82 was king force was significantly process of multiple-response et al. (2015) found optimal obtained, indicating that the affected by both linear and numerical optimization yielded conditions for frying at 108ºC implemented model successful- quadratic correlations of tem- a maximum convenience score and 9min, with maximum con- ly described behavior of oil perature (X1, X12), which indi- of 0.73 for the following set of venience of 0.61. Similarly, uptake in vacuum-fried slices. cated that eggplant slices re- experimental factors: frying Abtahi et al. (2016) found con- Lightness and color changes quire higher values of strength temperature (130ºC), frying ditions to be 153.46ºC and were significantly affected by to be broken after losing large time (210s) and blanching. 1.03min. In general, most of temperature and pretreatment amounts of water. A R2 of Optimum values of the mini- the work relating to the opti- (X1 and X3) in both linear 0.99 revealed a suitable adjust- mized responses were ob- mization of vacuum frying and quadratic interactions. ment of the experimental data tained at moisture of 64.77%, processes focuses on deciding

OCTOBER 2017 • VOL. 42 Nº 10 689 (NUSCA) through the project Monzó J (2016) Effects of pro- 110766441899 (Nº 664 of 2014). cessing conditions on the quality of vacuum fried cassava chips (Manihot esculenta REFERENCES Crantz). LWT-Food Sci. Techn. 69: 515-521. Abtahi MS, Hosseini H, Fadavi A, Mirzaei H, Rahbari M (2016) Lo Scalzo R, Fibiani M, Francese G, The optimization of the deep-fat D’Alessandro A, Rotino GL, frying process of coated zucchi- Conte P (2016) Cooking in- ni pieces by response surface fluence on physico-chemical methodology. J. Culin. Sci. fruit characteristics of eggplant Techn. 14: 176-189. (Solanum melongena L.). Food Chem. 194: 835-842. Akinpelu OR, Idowu MA, Sobukola OP, Henshaw F, Sanni SA, Mariscal M, Bouchon P (2008) Com- Bodunde G, Munoz L (2014) parison between atmospheric and Optimization of processing con- vacuum frying of apple slices. ditions for vacuum frying of Food Chem. 107: 1561-1569. high quality fried plantain chips Miraei Ashtiani SH, Golzarian MR, using response surface methodo- Baradaran Motie J, Emadi B, logy (RSM). Food Sci. Biotech- Nikoo Jamal N, Mohamma- nol. 23: 1121-1128. dinezhad H (2016) Effect of AOAC (2005) Official Methods of loading position and storage Analysis of AOAC International. duration on the textural proper- 17th ed. Association of Official ties of eggplant. Int. J. Food Analytical Chemists. Maryland, Prop. 19: 814-825. USA. Moreira R, Da Silva P, Gomes C Aramendiz-Tatis H, Espitia M, (2009) The effect of a de-oiling Cardona C (2010) Análisis de mechanism on the production sendero en berenjena (Solanum of high quality vacuum fried melongena L.). UDCA Actual. potato chips. J. Food Eng. Divulg. Cient. 13: 115-123. 92: 297-304. Correa E, Araméndiz H, Azeredo L, Nunes Y, Moreira R (2009) Effect of Figure 5. Response surface plots by vacuum frying optimization of Pombo C, Cardona C (2010) osmotic dehydration and va- blanched eggplant slices: (a) moisture (%), (b) oil (%), (c) lightness, Tipificación de comercializado- cuum frying parameters to pro- (d) ∆E, (e) breaking force, (f) color, (g) odor, (h) flavor, (i) greasiness, res de berenjena en zonas produce high-quality mango chips. and (j) convenience. ductoras del Caribe colombiano. J. Food Sci. 74: 355-362. Temas Agrarios 15: 1-10. Oginni OC, Sobukola OP, Henshaw Da Silva PF, Moreira RG (2008) FO, Afolabi WAO, Munoz L Vacuum frying of high-quality (2015) Effect of starch gelatini- on conditions that allow the Pretreatments exerted a signif- fruit and vegetable-based zation and vacuum frying con- snacks. LWT-Food Sci. Techn. ditions on structure development development of products with icant effect on sensory quality, 41: 1758-1767. and associated quality attributes good physical characteristics change of color and breaking Diamante LM, Savage GP, Vanhanen of cassava-gluten based snack. and high acceptability. force of the vacuum-fried egg- L (2012) Optimisation of va- Food Struct. 3: 12-20. plant slices. The combination cuum frying of gold kiwi fruit Pedreschi F (2012) Frying of pota- Conclusions of factors levels that produced slices: application of response toes: Physical, chemical, and the best vacuum-fried eggplant surface methodology. Int. J. microstructural changes. Dry. Results of the optimization slices were: frying temperature Food Sci. Technol. 47: 518-524. Technol. 30: 707-725. procedure indicated that mois- (130ºC), frying time (210s) and Dueik V, Bouchon P (2011) Deve- Reis FR, Masson ML, Waszczynskyj ture content and lightness blanching pretreatment. Va- lopment of healthy low-fat snacks: N (2008) Influence of a blan- levels of eggplant slices expe- cuum frying combined with Understanding the mechanisms of ching pretreatment on color, oil quality changes during atmosphe- uptake and water activity of rienced a significant decrease pretreatments represents an ric and vacuum frying. Food Rev. by increasing temperature (X ) alternative to produce fried potato sticks, and its optimi- 1 Int. 27: 408-432. zation. J. Food Proc. Eng. and frying time (X ) during purple creole eggplant slices 2 Dueik V, Robert P, Bouchon P (2010) 31: 833-852. vacuum frying. These respons- with low oil uptake, suitable Vacuum frying reduces oil up- Shyu SL, Hau LB, Hwan LS (2005) es were also affected (p<0.05) physical features and sensory take and improves the quality Effects of processing conditions by the pretreatment fac- acceptability. The same ap- parameters of carrot crisps. Food on the quality of vacuum-fried tor (X3). On the other hand, proach could be applied to Chem. 119: 1143-1149. carrot chips. J. Sci. Food Agri. oil content, color change and different vegetable food with Esan TA, Sobukola OP, Sanni LO, 85: 1903-1908. breaking force increased signi- similar characteristics. Bakare HA, Munoz L (2015) Shyu SL, Hwang LC (2001) Effects ficantly with increasing fac- Process optimization by respon- of processing conditions on tors. Linear and quadratic in- ACKNOWLEDGEMENTS se surface methodology and the quality of vacuum fried teractions of temperature had quality attributes of vacuum apple chips. Food Res. Int. fried yellow fleshed sweetpotato 34: 133-142. the highest influence. A lower Authors thank to the Admi- (Ipomoea batatas L.) chips. oil uptake and moisture loss nistrative Department of Scien- Food Bioprod. Proc. 95: 27-37. Šumić Z, Vakula A, Tepić A, was observed in blanched and ce, Technology and Innovation Čakarević J, Vitas J, Pavlić B Garayo J, Moreira R (2002) Vacuum (2016) Modeling and optimiza- dried eggplant slices under all of Colombia (COLCIENCIAS) Frying of Potato Chips. J. Food tion of red currants vacuum experimental conditions of for financial support to the Eng. 55: 181-191. drying process by response sur- time and temperature com- Research Group Nutrición, Garcia-Segovia P, Urbano-Ramos face methodology (RSM). Food pared to control samples. Salud y Calidad Alimentaria AM, Fiszman S, Martínez- Chem. 203: 465-475.

690 OCTOBER 2017 • VOL. 42 Nº 10 Tirado DF, Acevedo D, Guzmán LE Tirado DF, Acevedo D, Montero PM on the attributes of quality of cochemical and chemical pro- (2012) Freído por inmersión de (2015) Transferencia de calor y Arepa con Huevo. Inf. Tecnol. perties of wheat chips. LWT- los alimentos. Reciteia 12: 69-82. materia durante el proceso de 28: 99-108. Food Sci. Technol. 67: 89-98. Tirado DF, Acevedo D, Guzmán LE freído de alimentos: tilapia Yuksel F, Kayacier A (2016) Utili- Zhang T, Li J, Ding Z, Fan L (2016) (2013) Coeficientes convectivos (Oreochromis niloticus) y fruta zation of stale bread in fried Effects of Initial Moisture de transferencia de calor duran- de pan (Artocarpus communis). wheat chips: Response surface Content on the Oil Absorption te el freído de láminas de tilapia Inf. Tecnol. 26: 85-94. methodology study for the cha- Behavior of Potato Chips Du- (Oreochromis niloticus). Inf. Torres JD, Acevedo D, Montero PM racterization of textural, mor- ring Frying Process. Food Tecnol. 24(6): 41-46. (2017) Effects of vacuum frying phologic, sensory, some physi- Bioproc. Technol. 9: 331-340.

OCTOBER 2017 • VOL. 42 Nº 10 691