REGULAR DRYING KINETICS, FOURIER-TRANSFORM INFRARED SPECTROSCOPY ANALYSIS AND SENSORY EVALUATION OF SUN, HOT-AIR, MICROWAVE AND FREEZE DRIED MANGO LEATHER

Mango leather produced from dried mango pulp is a traditional fruit product in India. Traditionally it is processed through natural convective drying or sun drying of ripe mango pulp. Mango leather was produced through sun, convective (hot air), freeze, and microwave drying. Drying kinetics was studied with the help of nine empirical models used extensively in food industries. Root mean square error (RMSE), correlation coefficient (R), and the sum of square errors (SSE) these four statistical measures were examined for nine different models to learn the best-fitted model. The Fourier-transform infrared spectroscopy study was conducted for compositional analysis of differently dried samples in the wavelength range of 400-4000 cm at ambient temperature. For sensory analysis, the test panel was constructed as per ISO 8586-1. ISO 4120:2004, ISO 5496:2005, ISO 10399:2004 standards were maintained for selecting panel members. Quantitative descriptive analysis of different mango leather was estimated as per the protocols defined in ISO 11035:1994.


INTRODUCTION
Mango leather is cherished as a snack food in India, for its consummate organoleptic properties (Sarkar et al., 2018). Being a tropical country, sun drying is a widely practiced method for mango leather preparation, though it is a climate-dependent process. Due to uniform drying characteristics, hot air oven drying is adopted by different food industries (Fernando et al., 2016). Freeze drying is not generally employed to produce mango leather, however, it is assumed as a superior method to maintain the nutritional quality of the product (Acar et al., 2014). Being a rapid and controllable process microwave drying is considered as an energy-saving method as well as cost-effective (Zarein et al., 2013). To design mango leather processing for industrial applications, mathematical modeling is indispensable. Several studies have been available for drying kinetics of mango slices dried under tunnel dryer (Goyal et al., 2006), convective dryer (Dissa et al., 2014), Osmo-dehydrated mango (Kumar et al., 2014), solar drying (Dissa et al., 2009), though an inadequate number of studies have been found for drying kinetics of mango leather processing concluded by drying of mango pulp in different drying techniques. It is the first approach to study the sun drying (traditional approach) and freeze drying kinetics of mango leather processing.
Fourier-transform infrared spectroscopy (FTIR) progressively gains popularity in the field of food analysis due to its fast and non-destructive fingerprint modus operandi (Erwanto et al., 2016, Ojha et al., 2018. Few  predicted maturity stage of mango by FTIR analysis. Though as per our knowledge, it is the first approach to study the FTIR spectrum for differently dried mango. Sensory perception of consumer and product characteristics like colour, texture, and flavour profile can be correlated with quantitative descriptive analysis (QDA) of concerned food material. Though various studies are available in the field of sensory science (Lestringant et al., 2019), inadequate results are there in the zone of QDA study for ethnic food like mango leather, as per our literature survey this is the first QDA study for mango leather.
The aim of this study is to determine the drying kinetics for mango leather processing under sun, hot, microwave and freeze drying; FTIR spectroscopic characteristics of different mango leathers and QDA sensory estimation of the final product.

MATERIALS AND METHODS
Sun drying (30±5 °C), hot air oven (70 °C) (Concepts International, Kolkata, India), microwave (200 W) (Samsung, Combi CE1031LAT, Mumbai, India) and freeze drying (-40 °C, 0.1 mBar pressure) (FDU 1200, EYELA, Japan) was adapted as described in . Puree load was 0.5 g/cm 2 . The total soluble solid of the pulp was maintained at 25 °B with addition of sugar. In order to describe the most effective model for mango leather (Langra variety) processing thin layer drying models of various types were taken into consideration (Table 1). In calculating the Moisture Ratio (MR) during drying of mango pulp an equation stated below is used- Where, M = moisture content of mango leather at a different time interval (t) (kg moisture/kg dry leather), Min = moisture present in mango pulp initially (kg moisture/kg dry leather), Meq = moisture in mango leather at equilibrium (kg moisture/kg dry leather) (Arslan et al., 2010). In determining the best fitting model parameters like Root Mean Square Error (RMSE), Correlation Coefficient (R 2 ), and Sum of Square Errors (SSE) have been taken into account.
Where, MRe-MR from experiments, MRp-model predicted MR, N-observation number, MRt-moisture ratio at time t, MRm-mean moisture ratio (Acar et al., 2014). Fick's law of diffusion defines the drying behaviour of food material during falling rate period. Throughout the process it was assumed that effective moisture diffusivity and temperature remained constant, hence shrinkage was negligible and change in moisture content was only due to diffusion, the equation for unsteady-state moisture diffusion of an infinite slab (Demiray et al., 2017) can be represented as follows- Where De -effective moisture diffusivity (m 2 /s) and L -(sample thickness)/2 (3mm).
The logarithmic form of the first portion of the equation (5)  Bruker Fourier-transform infrared spectroscopy (FTIR) spectrophotometer (Bruker Tensor 27, Bruker Corp, Massachusetts, USA) was used in recording the FTIR spectra in duplicate at a temperature of 25 0 C, the wavelength range was 400-4000 cm -1 and sixteen scans were taken at 2 cm -1 resolution. Analysis of spectral data was carried out with the help of BRUKER software version 7.5 (1991) (Origin Lab Corporation, Northampton, MA, USA).
Four male and same number of female participants in the age bracket of 25-40 years were considered to construct the panel, as per ISO 8586-1 (1993) for sensory analysis. ISO 4120:2004 was the standard considered for triangular discrimination; ISO 5496:2005 was the standard for odour and taste identification proficiencies. ISO 10399:2004 was the standard used for ability of discrimination between two food materials with respect to sensory qualities. The sensory attributes of mango leather prepared under traditional (SD) and contemporary (HAD, MWD, and FD) dehydration techniques was estimated according to ISO 11035:1994. The evaluations of panelists and panel have been carried out through Panel Check software (Norwegian Food Research Institute and the Technical University of Denmark) version V1.4.2.  Considering the thin layer drying models and assuming an even distribution of temperature and moisture within mango pulp during the different drying processes, the endured parameters were modeled and thereafter fitted to all the nine equations (Table 1). Model convolutions can be ascribed by the constants considered for each model, though the best-fitted model is described by various statistical measures and not by the constants that exist within the model (Onwude et al., 2016). The experimental results fitted well with the entire models ( From initial moisture content of 5.45 kg moisture/kg dry leather to attain the final moisture content of 0.163 kg moisture/kg dry leather, irrespective of drying methods used moisture amputation of mango pulp was transpired mainly in the falling rate period and the constant rate period was not observed for any of the drying methods. During the thin layer drying of mango pulp, the temperature distribution was assumed to be uniform, as the thickness of the layer was uniform (Onwude et al., 2016). In the initial phase of drying due to the fast diffusion of water molecules the rate of drying was fast as cell wall pathway and transmembrane transport way define the moisture movement in this phase, while the moisture vapour diffusion became dominant in the later phase of drying through the symplastic transport mechanism, which ensued in a slowdown in the drying rate. Effective moisture diffusivity (De) is the key parameter to describe the falling rate period.  et al., 2002. The amount of moisture present within the mango pulp responsible chiefly for its dielectric characteristics. The electromotive energy absorbs by the polar components of mango pulp, mainly the water molecules, fasten the rate of evaporation, therefore the absorption of electromotive energy absorption is directly proportional to leftover moisture. In contrast to the sun and hot drying, microwave can infiltrate the outermost inert dry layer and produce an outward flux of vapour from the inner side of the mango leather (Wang et al., 2007). An extended falling rate was observed for SD. Similar result was observed for sun drying of tomato slices by Rajkumar et al., 2007. Due to the sluggish rate of heat transfer, sun drying was the leisureliest drying process with drying time 32 hours followed by freeze drying with a drying time of 22 hours and hot air drying with drying time of 10 hours, where convective heat transfer was responsible for moisture removal. During freeze drying, due to sublimation, moisture reaches to surface of mango leather through the capillary action. As the partial vapour pressure of mango pulp was higher than that of lyophilizer condenser, the moisture at the surface front got dried out (Acar et al., 2014).

FOURIER-TRANSFORM INFRARED SPECTROSCOPY ANALYSIS (FTIR)
Each spectrum containing more or less fifteen to sixteen peaks of ranges lies in between 448-3557 cm -1 (Fig. 2). Carboxylic groups (-COO) of pectin shows symmetric stretching at peak value 1336 cm -1 (SD).On the other hand, two bands appearing at 1422-1425 cm -1 (SD, HAD, MWD, FD) and 1638-1642 cm -1 (HAD, SD, MWD) (Ajila et al., 2010) represent the asymmetric and symmetric stretching frequency of ionic carboxylic group. Conjugation of the carbonyl group (C=0) with C=C and C=O stretch can be represented by the peaks appear close to 1638 cm -1 . In the case of primary and secondary amides, the amide (N-H) bending may be related to the peak at 1642 cm -1 (MWD) (Leopold et al., 2011). Partial overlapping of the absorption band is noticed for carboxyl (C=0) and amide (N-H), at peak value 1680-1630 cm -1 and 1640-1620 cm -1 respectively. C-O bond shows stretching vibration at 2123-2126 cm -1 because of the non-ionic carboxylic group or their ester acid (-COOH, -COOCH3) (Rambla et al., 1998). Symmetric stretching frequency of -CH2 lie at the peak value of 2931 cm -1 (HAD) whereas either of symmetric or asymmetric C-H stretching frequency of aliphatic acids is represented by a peak at 2932 cm -1 (SD, HAD, MWD, and FD). Absorption frequency of hydroxylation lies in the region of 3700 to 3000 cm -1 . It has been observed that stretching frequency of -OH mostly retains at 3377 cm -1 (SD) (Ajila et al., 2010). Amide (N-H) stretching frequency is observed in the range of 3500-3000 cm -1 . From the study, it can be noted that this stretching frequency is noticed for all products dried under four different methods ( The panelists and panel performance was evaluated towards the sample attributes. Panelists significantly perceive the variance within samples (Fig. 3a, 3b, 3c, 3d), with respect to orange-yellow colour, rigidity, surface finish, chewiness, and aftertaste. The panelists attained a low p-value as well as low MSE values for all the sensory attributes, by the development of constellation adjacent the lower-left portion of the plot, inferring the samples are categorically dissimilar from each other. The multivariate approach of analysis produces the Tucker-1 plot (Fig.4.), where specific incongruity was observed for statistically significant (p < 0.5) attribute aftertaste (P5) and insignificant (p > 0.5) attribute sweetness (P1, P3, P5, P6, P7, and P8). For all the other attributes no such discrepancies have been observed. Regardless of these disagreements, most of the panelists were accommodated in the outer eclipse, recommending 100% of explained variance for each attribute. It was also found that the explained variances accommodated themselves within the first two principal components (PC), by predicting the pronounced systematic variation of analyzed data set proposing the panel is well-trained and static. Instead of a panel, samples are considered during consensus analysis of the data set. The standardized PCA Bi-plot (Fig.5) describe that mutually axis 1 and 2 showed a very high correlation (70.6% and 26.3%) among all sensory attributes. The attributes like citrus flavour and sweetness reside on the positive side of the PC1, while orange-yellow colour, chewiness, surface finish, rigidity, and aftertaste belong to the negative side of PC1. On a similar fashion, sweetness, rigidity, and aftertaste were observed at the positive coordinate of PC 2, though four other attributes namely orange-yellow colour, surface finish, citrus flavour, and chewiness observed at the negative side of PC 2.  The SD (sample 1) was more correlated with orange-yellow colour, surface finish and chewiness, HAD (sample 2) and MWD (sample 3) were found to be more correlated with citrus flavour and sweetness respectively. FD (sample 4) was found to be correlated with sensory attributes like rigidity and aftertaste. SD was appeared to be preeminent regarding orange-yellow colour, surface finish, and chewiness, whereas HAD and MWD found to be superlative with respect to sweetness and citrus flavour respectively, while the attributes like rigidity and aftertaste found to be superior for FD. Spider plot (Fig. 6) showed the overall representation of the sensory attributes which was in accordance with information acquired Fig. 5. Surface finish, orange yellow colour, and chewiness were the attributes where SD (sample 1) showed its superiority. While FD (sample 4) showed the maximum score in terms of rigidity and aftertaste. MWD (sample 3) showed the highest citrus flavour development. HAD (sample 2) was scored maximum in terms of sweetness.

CONCLUSION
Thin-layer drying models were studied for mango leather and Midilli model was fitted best with the maximum value for R 2 of 0.999479, 0.999567, 0.999998, 0.99877 along with minimum RMSE and SSE value of 0.006283, 0.005867, 0.000425, 0.010534 and 0.00122, 0.0006539, 0.000003, 0.00233 for sun, hot air, microwave and freeze drying respectively. The effective diffusivity was found maximum for MWD (2363×10 -7 m 2 /s) followed by HAD (193.7×10 -7 m 2 /s), FD (44.63×10 -7 m 2 /s) and sun drying (26.37×10 -7 m 2 /s). SD was the lengthiest (32 hours) process, while MWD being the shortest (96 minutes) method for mango leather production. The FTIR analysis implied that different drying methods insignificantly affect the carbohydrate composition. SD was found better with respect to its distinguished orange-yellow appearance, smoothness of the surface, and chewability, however, HAD and MWD were exceptional with compared to sweetness and characteristics citrus flavour of mango leather respectively, whereas, in terms of rigidity and aftertaste FD was preferred most.