Far infrared assisted refractance window drying: Influence on drying characteristics and quality of banana leather

Abstract The market demand for fruit-based healthy snacks in dried form is increasing rapidly. Novel and efficient drying techniques are being explored to meet market demand of fruit-based snacks with better nutritive value and sensorial attributes. In this study, suitability of far infrared assisted refractance window (FIR + RW) drying to obtain banana leather was ascertained and drying behavior, bioactives, flavor, microstructure, and sensory attributes were compared with RW and hot air (HA) drying. FIR + RW and RW reduced the drying time by 60–75% and energy consumption by 38–45% as compared to HA. RW drying preserved color and retained higher phenolics (19%), ascorbic acid (22%) and antioxidant capacity (16–47%) compared to HA. Among the studied methods, HA resulted in higher browning index and hydroxymethylfurfural content. FIR + RW and RW dried banana samples had maximum flavor compounds, better overall consumer acceptance and improved cellular structure with widened pores. The study indicated that RW and FIR + RW have good potential to be considered as alternative drying techniques to HA for producing high-quality fruit leather.


Introduction
Moisture being a major factor in food spoilage/deterioration needs to be reduced to enhance shelf-life and drying is one of the primary methods for moisture reduction.In current scenario, novel methods are recommended for drying to meet consumer demands for product with better quality, productivity, process economics and environmental sustainability. [1]In this regard, hybrid drying method is being attempted, wherein two or more methods are combined. [2]ccording to Chua & Chou, [3] hybrid drying technique can be classified into combined, multiple stage and multiple process drying.In hybrid drying, electromagnetic radiation (EMR) based far infrared (FIR) heating with other drying methods is advantageous over conventional or single stage drying in terms of drying characteristics, energy efficiency and product quality due to its synergistic effect, particularly for heat sensitive foods.The FIR heating limits its application in surface heating/drying due to its inefficiency in penetrating the food matrix. [4,5]Hence, the application of FIR along with other methods, i.e.FIR assisted pulsed vacuum drying, [6] FIR assisted heat pump, [7,8] FIR assisted refractance window (RW), [9] and ultrasound and IR assisted RW [10] has shown to enhance the drying efficiency and product quality.
RW is a fourth-generation innovative, emerging drying technique that is getting momentum in the food dehydration sector. [11]The RW drying system consists of heating medium (hot water), water circulation system, and polyester film. [12]During drying, food material is placed on the polyester film which is in contact with the medium.The thermal energy of heating medium, circulated through reservoir, is transferred to material to be dried, by means of conduction and radiation, as the polyester film is partially transparent to infrared radiation. [11]In comparison to conventional drying, RW facilitates higher drying rate with minimal damage to bioactive/nutritional components and thus producing better quality product in a shorter duration. [11,13]So far, studies on RW have been majorly focused on drying of fruits and vegetables such as apple and carrot [14,15] in slice form, while a few reports on producing leather from puree/pulp/paste are available such as mango and pomegranate. [16,17]These reports revealed that RW drying retains heat labile compounds (vitamins and antioxidant) to a greater extent with better product appearance, as compared to that of HA dried.One of the limiting factors of RW drying is the thickness of material being dried, indicating that this technique is suitable for drying thin layer of material. [9,14]dditionally, water temperature, properties of polyester film, and type of food material to be dried also affects the efficacy of RW drying process. [9,12]owever, there are no detailed studies on FIR assisted RW (FIR þ RW) drying of banana puree to produce dried sheet/leather and its comparison with other drying methods in terms of drying characteristics and product quality.The present investigation focuses on comparison of FIR þ RW drying with RW, and HA drying on drying behavior and product quality of banana leather in terms of color characteristics, retention of bioactive and flavor components, sensory attributes and microstructure have been studied.The mass transfer parameters and energy requirement for the process are also estimated.

Preparation of banana puree
Banana puree was prepared using ripe Cavendish bananas (Musa acuminata), procured from local market of Mysuru, India.The banana puree was prepared as per the procedure described by Rajoriya et al. [18] Ripe bananas (20-23 Brix) were washed, peeled, sliced and dipped for 2 min in potassium metabisulfite (KMS) solution (0.8% w/v) to prevent discoloration and grinded into paste.Further, to prevent browning, ascorbic acid (AA) was incorporated at the concentration of 0.2% (w/w) into the prepared banana puree and mixed thoroughly.This resultant puree was considered as control and referred as treated banana puree (BP).

Drying studies
The drying studies were carried out in FIR þ RW, RW, and HA dryers with batch sizes of nearly 200 g puree, in order to maintain a bed thickness of 2 mm.All the drying experiments were continued until the moisture content of product or banana leather reached $12-13% (wb).

FIR assisted RW and RW drying
An in-house FIR þ RW drying system (batch-type laboratory-scale) was developed and used in the present study.Briefly, water was heated to a desired temperature 90 ± 2 C (selected based on our previous studies [15,18] ) using water bath with a thermostat (Shital Scientific Industries, Bombay, India) and circulated through the water reservoir with help of peristaltic pump (Ravel Hiteks Pvt., Ltd., Chennai, India).The BP was spread on 250 mm thick Mylar TM polyester film (food-grade) with its bottom surface kept in contact with hot water.The water vapor produced while drying was eliminated by a top-mounted exhaust fan (JiGO India Pvt. Ltd., India).In addition, on either side of the exhaust fan, two FIR heaters (ACE HEAT TECH, Mumbai, India) made up of ceramic were attached at a distance of $10 cm from the polyester film.The FIR heaters were used for FIR þ RW drying and, required temperature (50 and 60 ± 2 C) during drying was monitored and controlled with the help of temperature sensor (HTA Instrument (P), Ltd., India).The FIR heaters temperature, and the distance between FIR heater and polyester film were selected based on our previous study. [9]The above drying setup was used in RW trials, without FIR heaters.
The samples were collected at 5 and 10 min interval during FIR þ RW and RW drying, respectively, for moisture content estimation.

HA drying
A hot air drying system (Labline Instruments, Cochin, India) of laboratory-scale was employed in the present study.The BP was spread on the polyester film and kept on perforated stainless-steel trays in the drying chamber with maintained temperature (90 ± 2 C) and air velocity (0.5 ± 0.02 m/s).For moisture content estimation, samples were collected at 10 min interval.

Drying characteristics
2.3.1.Water activity, moisture content and moisture ratio Water activity meter (AquaLab, Decagon Devices, Washington, USA) was used to measure water activity of samples at 27 ± 2 C. Moisture content was determined according to Ranganna [19] using oven method and expressed as g H 2 O/g db.
Equation ( 1) was used to convert experimental moisture content to a dimensionless moisture ratio (MR): where M t is the moisture content at time t (min); M o is the moisture content at t ¼ 0; M e is the equilibrium moisture content.In the current study, M e is substantially lower than M 0 and M t , thus Equation (1) can be simplified as follows Weibull distribution model (Equation ( 3)) was fitted to the drying curve of banana leather, according to the method stated by Dai et al. [20] MR where t is drying time (s); a is the scale parameter (s); b is the shape parameter (dimensionless).

Estimation of effective diffusion coefficient
Effective diffusion coefficient (D eff ) can be calculated with Weibull distribution model using Equation (4) [20,21] where D cal is the estimated diffusion coefficient (m 2 /s); L is slab half-thickness; R g is the physical dimension constant which ranges from 13.1 to 18.6 m 2 /s for flat or globular shaped agricultural produces. [21]In the present study, R g value of 13.1 m 2 /s was used to estimate the D eff . [6]

Specific energy consumption
Electrical energy consumed during drying process was recorded using digital energy meter (CL1, L&T, India) and, specific energy consumption (SEC) was calculated using Equation (5) as mentioned by Shewale et al. [22] Specific energy consumption kWh=kg À Á quantity of moisture removed during drying kg ð Þ (5)
where L i , a i, and b i refer to the values of dried banana samples, whereas L 0 , a 0 and b 0 refer to the values of the BP (control).

Bioactive components
The extraction of bioactives from BP and banana leather was done according to Wojdyło et al. [23] Briefly, sample (1 g) was well mixed for 15 min using a magnetic stirrer in 80% aqueous methanol (30 mL) containing 1% HCl solution and then was sonicated twice (15 min each).Further, the suspension was incubated for 24 h in the dark at room temperature (27 ± 2 C) and later centrifuged for 20 min at 4 C and 5488 g.The resultant supernatant was used for the assessment of bioactive components.Total phenolic content (TPC) was analyzed using Folin-Ciocalteu method and the results were expressed as mg gallic acid equivalents per g dry matter (mg of GAE/g dm) as reported by Shewale & Hebbar. [24]Antioxidant capacity in terms of Ferric reducing antioxidant power (FRAP) and ABTS (2,2 0 -azino-bis(3-ethylbenzothiazoline-6-sulfonic acid)) radical-scavenging activity was determined using the methodology of Benzie & Strain [25] and Re et al. [26] , respectively and expressed as mM Trolox equivalent per g dry matter (mM TE/g dm).
Ascorbic acid (AA) extraction was carried out according to Bhavya et al. [27] $1 g of sample was macerated with 3% metaphosphoric acid þ 8% acetic acid (20 mL) in a mortar and pestle.The supernatant was recovered immediately after centrifuging the suspension at 5000 Âg and 4 C for 20 min.The AA content in supernatant was analyzed using high performance liquid chromatography (HPLC) procedure as mentioned by Bhavya & Hebbar [28] and results were expressed as ascorbic acid retention (AAR) in percentage (%) which is calculated with respect to BP.

Hydroxymethylfurfural content
The Hydroxymethylfurfural (HMF) content in banana leather was estimated as mentioned by Rufi an-Henares & Delgado-Andrade [29] with slight modification.Briefly, $1 g of sample was macerated with 10 mL of distilled water using mortar and pestle.Then, the mixture was transferred into 15 mL amber tube and centrifuged at 5488 g for 10 min at 4 C.The resultants were mixed thoroughly using the vortex and clarified with 0.25 mL each of Carrez I (potassium ferrocyanide, 15% w/v) and Carrez II (zinc acetate 30% w/v) solutions followed by centrifugation.The volume of resultant was made up to 10 mL with distilled water and 2 mL of sample was filtered through 0.2 mm PTFE syringe filter for the estimation of HMF content using HPLC method.A 10 mL filtered sample was passed through a C-18 Ascentis column (5 mm; 4.6 mm 250 mm ID) of Shimadzu HPLC system (LC-8A), using a mobile phase (0.005 N H 2 SO 4 ) with a flow rate of 1 mL/min.A photodiode array (PDA) detector was used to measure absorbance of HMF at 284 nm.Furthermore, HMF was quantified by drawing a graph between concentrations (2-40 ppm) vs peak area of the standard, and the results were expressed as mg/kg dm.

Flavor analysis
Headspace-solid phase microextraction (HS-SPME) method was used for the extraction of banana volatiles.In a hermetically sealed screw-top vial (15 mL) having a polypropylene cover with PTFE septa, $ 2 g of sample was added along with distilled water (5 mL) and sodium chloride (1 g).The volatiles of banana samples were extracted using 50/30 mm DVB/CAR/PDMS (Supleco) fiber at 50 C for 40 min, after equilibration for 5 min.Further, gas chromatography-mass spectroscopy (GC-MS) analysis was carried out using a GC (PERKLIN ELMER AUTOSYSTEM XL) with electron impact (EI) ionization mode and Capillary ELITE-1 column (100% dimethyl polysiloxane -30 m Â 0.32 mm Â 0.25 mm).The followings were the GC-MS operating conditions: the injector and detector were set at 250 C, carrier gashelium (99.9999%), flow rate À 1 mL/min, the interface was maintained at 160 C, and the split was 0:1.
The oven temperature was kept at 50 C for 3 min before being raised to 250 C at a rate of 5 C/min and held for 5 min.All mass spectra were acquired using a PERKLIN ELMER TURBOMASS GOLD MASS SPECTROMETER in EI mode with a 70 eV ionization voltage and a temperature of 150 C ion source.The Kovats indices were computed using cochromatography and an n-alkanes mixture (C 7 -C 40 ) as internal standards.By matching mass fragments and Kovats indices, all of the chromatogram peaks were identified using a computer library search.

Scanning electron microscopy analysis
Microstructural analysis was carried out as mentioned by Shewale & Hebbar. [24]Briefly, the fixation of banana leather was performed using glutaraldehyde solution (3%) for 12 h at 4 C and the leather was washed with phosphate buffer solution (0.1 M, pH 7.2).The postfixation of the samples was carried out using osmium tetroxide solution (1%) at 4 C for 2 h and then serially dried using 30, 50, 70, and 90% ethanol solution for 15 min each.Finally, samples were freeze dried after immersing them in absolute ethanol for 20 min.The obtained samples were gold coated and examined at 100Â magnification using a scanning electron microscope (SEM) (LEO 435VP SEM, Cambridge, UK).

Sensory analysis
The sensory analysis of banana leather was carried out by 40 semi-trained panelists (postgraduates and Ph.D. scholars), of which 20 were women, and 20 were men of age ranging from 23 to 35.The intensity of all sensory parameters rated by assessors were evaluated using a 9-point Hedonic scale from 1 to 9 (1 and 9 indicate dislike and like extremely, respectively).The banana leather was evaluated for characteristics such as visual appearance, color, texture, aroma, chewiness, taste and overall acceptance.

Statistical analysis
The results obtained from the experiments carrid out in triplicates were represented as mean ± standard deviation.Analysis of Variance (one way-ANOVA) was used to compare different means, followed by a post-hoc Duncan's multiple range test (significance level, P < 0.05).The Weibull distribution model parameters, a and b described in Equation (3), were determined using least-squares curve fitting procedure with the help of Solver tool of Microsoft Excel 2016 (Microsoft Corporation, Washington, USA).The principal component analysis (PCA) was performed to evaluate the sensory and quality attributes of dried samples and their correlation (Pearson method) matrix was obtained using the software XLSTAT version 2022.1.2.

Moisture content, drying time and water activity
The average moisture content of BP was found to be 77.01 ± 0.8% (wb) and it was dried till it reached 12-13% (wb).The drying curves for different methods employed in the study are presented in Figure 1.It is clearly seen that HA required longer time (100 min) for drying as compared to other methods.Nearly, 60% reduction in processing time was noticed in case of RW drying in comparison to HA.This could be attributed to development of porous structure leading to a faster moisture diffusion through the material dried by RW drying. [14]Furthermore, FIR þ RW drying lowered the duration required for drying by 25  1), which falls in the limits of safe storage (a w <0.6). [30]Water activity range of 0.42 to 0.55 was also reported in fruit leathers obtained from RW drying. [16,17]

Weibull distribution modeling of drying kinetics and effective diffusion coefficient
The a and b values of the Weibull distribution model were determined using least square method and are presented in Table 1.The b is related to the mass transfer rate which occurs at the initial period of drying and b > 1 indicates the existence of lag phase during this stage. [6]The values of b in the present study were found to be in the range of 1.33-1.51,which was similar (1.05-1.54) to that calculated by Dai et al. [20] in dried apricot halves.The a values were in the following order of HA 1).Higher a value during HA drying indicated longer drying time as a represents the rate of drying process. [20]The results of the current study suggest that both a and b parameters were significantly influenced by drying methods.Based on R 2 (>0.99) and RMSE (<0.0138) values (Table 1), it may be deduced that the experimental and calculated MR values (obtained from the Weibull distribution model) were in good agreement (Figure 1), that can efficiently define the drying behavior of banana puree during different drying methods.Previous reports also suggested the application of Weibull distribution model to describe the drying behavior of various food products such as apricot halves [20] , and wolfberry. [6]he D eff values of different drying methods were estimated using Equation ( 4) and values are presented in Table 1.The D eff values ranged from 2.75 Â 10 À11 to 1.04 Â 10 À10 m 2 /s, which lie within the standard range (10 À11 to 10 À9 m 2 /s) for agricultural commodities. [18]It can be seen from Table 1 that FIR þ RW drying had higher D eff values compared to that of HA and RW.This could be due to opening of pressureinduced pores facilitating easier moisture diffusion, as a result of absorption of FIR and internal heat generation. [31]In the current study, when FIR temperature was increased from 50 to 60 C, D eff values also increased, which could be due to enhanced activity of water molecules by accumulating more heat energy that accelerate moisture diffusion from the interior to the surface. [6]

Specific energy consumption
The influence of different drying methods on the SEC is presented in Table 1.As expected, HA drying  required higher energy (4.49kWh/kg moisture removal) due to longer processing time.While in case of RW drying, energy requirement was lower than HA by $38% due to significant reduction in drying time.Furthermore, addition of FIR heaters along with RW reduced the SEC by 7.5-11.4% and 42.3-44.8%compared to RW and HA, respectively.Similar observations were reported by Afzal et al. [32] and Rajoriya et al. [9] during FIR assisted HA and FIR assisted RW drying of barley and apple, respectively.An increase in FIR temperature from 50 to 60 C showed no significant difference in energy consumption.Among the studied drying methods, minimum SEC value (2.48 kWh/kg of water removal) was observed for FIR 60 C þ RW.

Color characteristics
The effect of method of drying on lightness (L Ã ), browning index (BI) and total color change (DE) of BP and banana leather was examined, and the values are presented in Table 2.It can be noticed that significant (P < 0.05) change in BI values were observed only in HA dried samples with highest a Ã (7.87 ± 2.35) and b Ã (29.54 ± 2.42) and least L Ã (61.61 ± 3.02) values among the drying conditions.This could be the result of non-enzymatic browning during HA due to prolonged drying at elevated temperature. [33]Even though, additional heating of FIR was applied along with RW, no significant difference in L Ã and BI values between RW and FIR þ RW were noticed.This prevention of color degradation during FIR þ RW could be a result of faster drying.It can be clearly seen that FIR temperature had no significant effect on studied color parameters of banana leather.Surprisingly, the DE values for the banana leather were found to be in the order: FIR 60 C þ RW < RW < FIR 50 C þ RW < HA, which was reverse order for L Ã values.The correlation studies (Table S1) also justify this opposite relationship as L Ã values were negatively related to DE (-0.996).

Total phenolic content
The effect of different drying methods on TPC of banana leather was determined and are represented in Table 2.The TPC in BP was estimated to be 13.19 ± 0.3 mg GAE/g dm.It was noticed that HA drying significantly lowered TPC by $34% compared to BP. Whereas, in case of RW, TPC was higher compared to HA by $19%, due to the release of bound  phenolics from the cellular matrix of dried samples as a result of rapid heating during RW drying.Likewise, Hern andez-Santos et al. [14] also observed increase in TPC of RW dried carrot slices than HA dried ones.Surprisingly, in the current investigation, hybrid (FIR þ RW) drying negatively affected TPC by $5-12% compared to RW, as a result of additional heating with FIR.Similarly, Zeng et al. [34] also noticed a decreasing trend in TPC of kiwi slices treated with FIR heating due to accelerated thermal degradation of phenolic compounds.In the present investigation, FIR þ RW dried banana leather had higher TPC than HA ($ 5-13%), which could be attributed to the synergetic effect of FIR and RW in retaining phenolics as a result of reduced drying time.Similarly, FIR treatment increased TPC compared to HA in peanut hulls and mulberry leaves. [35,36]

Antioxidant capacity
The effect of drying methods on antioxidant capacity was measured in terms of ABTS and FRAP assays and results are presented in Table 2.It can be noticed that both the assays showed a similar trend for banana puree dried by different methods.Drying process decreased the antioxidant capacity of banana leather in all the cases, while HA and RW dried leather had least and highest retention of antioxidant capacity, respectively.This could be attributed to the fact that RW required lesser drying time and thereby preserving antioxidant compounds like phenolics, and flavonoids.It was also confirmed that phenolic compounds majorly contribute to antioxidant capacity as reflected by correlation studies (Table S1), showing a positive correlation of TPC with FRAP (0.976) and ABTS (0.779).RW drying retained higher antioxidant capacity ($16-47%) as compared to HA.Similarly, RW dried carrot slices exhibited higher antioxidant capacity compared to HA by 43.3%. [14]FIR þ RW had significantly higher (P < 0.05) antioxidant capacity compared to HA (Table 2), which was similar to the results reported by Wanyo et al. [36] in FIR þ HA dried mulberry leaves.However, FIR þ RW dried leather showed a considerable decrease in antioxidant capacity, as compared to RW, which followed the same trend as TPC.No significant difference in antioxidant capacity was noticed between hybrid (FIR þ RW) drying methods.

Ascorbic acid retention
The effect of drying methods employed in present study on AAR has been presented in Table 2.It can be clearly seen that drying adversely affected the AAR in banana leather.However, among these methods, RW and FIR þ RW had better AAR (77.5-80.6%) in comparison to HA which had significantly lower AAR (66.1%).Similar trend was noticed in TPC and antioxidant capacity of banana leather as AA contributes to phenolic and antioxidant capacity which is also supported by the Pearson's correlation matrix (Table S1) showing strong positive correlation of AAR with TPC (0.964), FRAP (0.908) and ABTS (0.887).Further, higher AAR in novel drying methods can be credited to the protective effect of RW against the thermal and oxidative deterioration of AA due to shorter drying time. [15,16]This may be attributed to the uniqueness of the RW drying technique creating the thermal "window" on high moisture food which gradually closes as drying progresses, reducing the overheating of the product that prevents the thermal degradation of ascorbic acids. [37]In the present investigation, although, FIR þ RW showed a slight decrease (3.0-3.8%) in AAR compared to RW, it was not significant (P < 0.05).

Hydroxymethylfurfural content
Hydroxymethylfurfural is a measure and intermediate product of the non-enzymatic browning reactions. [16]mong the drying methods, HMF content was detected in banana leather produced by HA and RW drying.Higher HMF in HA dried samples could be due to non-enzymatic browning reactions promoted by longer drying at elevated temperature, which is also evident from the BI values (Table 2).This is also clearly corroborated from the correlation matrix presenting a strong positive relation (0.978) between HMF and BI (Table S1).In support to our findings, Tontul and Topuz [16] and Pekke et al. [38] also reported higher HMF during HA drying of pomegranate leather and banana slices, respectively.In the current investigation, RW dried banana samples showed $99% reduction in HMF content compared to HA (Table 2).Similarly, Tontul and Topuz [16] observed $74% reduction in HMF content in pomegranate pestil/leather dried using RW at 90 C in comparison to HA samples dried at 70 C.In the current investigation HMF content was not detected in FIR þ RW dried banana samples due to significant reduction in drying time.Pekke et al. [38] also reported similar observation during infrared (60 C) drying of banana slices.

Flavor analysis
The HS-SPME-GC-MS analysis identified a total of 35 volatile compounds in banana samples.It includes twenty esters, five alcohols, four acids, two aldehydes, two ketones, and two benzenes (Table 3).41] Alcohols, ketones, terpenes and aldehydes are said to directly impart ripe fruit flavor, while esters are considered as the odor-impact components. [39]The key esters such as isoamyl acetate, isobutyl acetate, butyl acetate, isoamyl butyrate, 2-pentyl acetate and isoamyl isovalerate attributed to fruity and sweet notes.Esters (amyl and butyl esters) are odor-active compounds of mature/ripe banana fruit, possessing fruity-banana note accounting for more than 50% of the compounds identified. [40]able 3 reveals that drying significantly affected the flavor profile of the banana leather.The volatile fractions (especially esters) of HA, RW, FIR 50 C þ RW and FIR 60 C þ RW were dominated by isoamyl butyrate (10.80%, 25.18%, 18.64%, 29.60%), followed by 2-pentyl butyrate (9.88%, 7.19%, 6.81%, 8.50%), isoamyl isovalerate (5.77%, 8.54%, 5.97%, 11.40%), isoamyl acetate (3.78%, 7.66%, 8.47%, 8.01%), and isobutyl butyrate (1.53%, 3.66%, 3.13%, 3.69%) respectively.Among esters, isoamyl butyrate, isoamyl acetate and isobutyl butyrate (sweet and fruity) were well preserved in RW and FIR þ RW dried banana samples compared to HA.Interestingly, following sequence/order of FIR 60 C þ RW > RW > FIR 50 C þ RW was noticed regarding the preservation of major esters, alcohols, acids and benzenes by novel drying methods (Table 3).Methyl ketones like 2-pentanone and 2octanone contributing to fruity and banana-like notes were detected in all the samples.Typical banana aroma imparting compounds according to Pine & Fables [42] such as isoamyl butyrate, isoamyl acetate, 2-pentanol acetate, 2-pentyl butyrate, butyl acetate, (Z)-3-Octen-1-ol, and (Z)-5-Octen-1-ol were also detected in the present study.Eugenol and elemicin associated with the full-bodied, mellow, sweet, phenolic, spicy aroma, while n-hexadecanoic acid and cisvaccenic acid exhibiting fatty aroma in ripe banana samples were higher in FIR 60 C þ RW, RW, and HA.Noticeably, isovaleraldehyde was only found in HA samples, which could be formed during Maillard reaction as a result of prolonged heating. [43]Overall, it can be suggested that combination of temperature and duration of drying affected the volatile profile of banana samples.The HS-SPME-GC-MS results indicate that flavor compounds, especially esters and alcohols in banana leather can be preserved significantly by RW or FIR þ RW drying.

Microstructural analysis
The microstructural analysis of banana leather produced by different drying methods are illustrated in Figure 2(a).A comparison of the micrographs revealed that the particles of the banana puree were uniformly and homogenously distributed.Hot air dried samples had a smoother cell surface with compact intercellular substances and micropores than that of other drying methods (Figure 2(a)).Development of tiny micro-channels/micropores in HA samples as a consequence of relatively longer drying time was not favorable for the migration and diffusion of water molecules.RW and FIR þ RW drying expedited the formation of spacious pores as a result of evolution of the water vapor inside the samples at faster rate.
These pores might facilitate better diffusivities of gases and liquids in the sample.The bottom surface of RW dried samples were more smoother than the upper surface as it was in direct contact with the polyester film.Additionally, synergistic effect of FIR heating and RW widened the pores and enhanced the mass and heat transfer process which was evident from higher D eff values.Similarly, bigger voids were observed by Nathakaranakule et al. [7] in hybrid dried (Heat Pump þ FIR) longan fruit as a result of FIR enhanced evaporation.

Sensory analysis
Generally, sensory evaluation of fruit leather is performed using preference test, where panelists express their responses in terms of liking or disliking. [17]The average scores analyzed using 9-point hedonic scale for all studied sensorial attributes of banana leather produced from different drying methods are depicted in Figure 2(b).Among all the treatments, the highest and lowest scores for the color and visual appearance were given to FIR 60 C þ RW (7.55) and HA dried samples (6.63), respectively, which can also be seen from Figure 2(c).The least preference for HA dried leather can be attributed to the fact that HA samples had higher BI values (Table 2 and Figure 2(c)).Further, texture of banana leather was smooth and glossy, and the sensory scores were almost similar for all the dried samples.HA dried leather had the least chewiness score (6.0) among the dried samples which could be due to case hardening effect generally observed during HA drying. [44]Noticeably, the taste preference of HA dried leather was almost near to FIR 60 C þ RW dried samples due to partial caramelization during HA drying.In case of aroma, RW dried leather showed highest score followed by FIR 50 C þ RW and FIR 60 C þ RW and HA.A higher score for overall acceptability of banana leather obtained from RW and FIR þ RW drying was recorded because of their good color, chewiness, texture, taste and aroma.It can also be justified that there was strong positive correlation among the above mentioned sensory parameters with overall acceptability in the range of 0.84-0.99(Table S2).Thus, the sensory panelists preferred banana leather produced by RW drying indicating better consumer acceptability than other studied drying methods.

Principal component analysis
Principal component analysis (PCA) is a multivariate modeling and dimensionality-reduction technique which compress the dimensionality of large data set by transforming its variables into smaller set (called principal components) without losing much of information and identify the underlying patterns in data.In the current investigation, PCA was performed separately for quality parameters, flavor profile and sensorial attributes and the obtained biplots are depicted in Figure 3(a), (b) and (c), respectively.For quality parameters, first two principal components (F1 ¼ 74.65% and F2 ¼ 23.17%) accounted for 97.82% of the variance (Figure 3(a)) with F1 mainly contributed by L Ã (0.921), a Ã (0.703), BI (0.806), DE (0.899), TPC (0.687), FRAP (0.533), ABTS (0.972), AAR (0.837) and HMF (0.785), and F2 mainly by b Ã (0.654).From Figure 3(a), it can be seen that BP is located in right side of biplot along with quality parameters such as FRAP, TPC, AAR, L Ã and ABTS indicating the higher content of these parameters as expected.On contrary, HA is positioned on the opposite side of BP along with HMF, BI, a Ã , b Ã and DE demonstrating the loss of bioactive compounds.Figure 3(a) showed higher F1 score for FIR þ RW and RW dried banana leather indicating higher contents of bioactive compounds.The biplot obtained for flavor profile (Figure 3(b)) showed that the first two principal components (F1 ¼ 70.10% and F2 ¼ 20.49%) accounted for 90.59% of the variance.Figure 3(b) revealed that the retention of esters and alcohols associated with sweet and fruity (banana like) aroma was higher in RW and FIR þ RW dried banana samples compared to HA, which is confirmed by the positioning of novel drying methods along with the majority of esters and alcohols.For sensory attributes, first two principal components (F1 ¼ 86.99% and F2 ¼ 12.05%) accounted for 99.04% of the variance (Figure 3(c)) with all the variables majorly contributing to F1.From Figure 3(c), it can be seen that RW and FIR þ RW were positioned in right quadrant along with all the studied sensory attributes indicating better consumer preference compared to HA, which was positioned in the opposite side of the biplot.Overall, from the PCA results, it can inferred that FIR þ RW and RW dried banana leather exhibited better quality parameters, flavor profile and sensorial attributes.

Conclusion
The influence of novel drying method (FIR þ RW) on drying behavior, bioactive compounds, energy consumption, flavor, microstructure, and sensorial attributes were investigated in comparison with RW drying and conventional HA drying.RW and FIR þ RW drying significantly decreased the drying time and energy consumption with higher D eff values compared to HA.Based on the results of color characteristics (L Ã , BI, and DE) and bioactive compounds (phenolics and antioxidants), it can be concluded that RW drying preserved these parameters as the samples neither experienced enzymatic degradation nor thermal hydrolysis.RW and FIR þ RW drying enhanced the cellular microstructure with widened pores facilitating improved mass and heat transfer.RW and FIR þ RW drying preserved better flavor compounds (esters and alcohols) compared to HA. Sensory analysis suggested better consumer preference for banana leather produced by RW drying due to higher score for taste, aroma, color, texture and overall acceptance, among the studied drying methods.To summarize, RW and FIR þ RW can be employed as a potential and alternative drying method for the production of high-quality fruit leathers.The findings of the present work suggest that this work could be extended to other heat-sensitive fruits, vegetables and herbs for developing commercially viable dried products and ready to eat snacks.

Figure 1 .
Figure 1.Effect of drying techniques on drying kinetics and comparison/variation between experimental and predicted moisture ratio (using Weibull distribution model).

Figure 2 .
Figure 2. Influence of drying methods on (a) microstructural matrix, (b) sensorial attributes, (c) appearance of banana leather samples.

Table 1 .
Effect of drying techniques on water activity, Weibull distribution model parameters, effective diffusion coefficient, and energy requirement.
HA: hot air; RW: refractance window; FIR: far infrared; a: scale parameter; b: shape parameter; R 2 : coefficient of determination; RMSE: root mean square error; D eff : effective diffusion coefficient; SEC: specific energy consumption.a,b,c Mean values with different superscripts within a column indicate significant differences (P < 0.05).

Table 2 .
Effect of drying techniques on color parameters and bioactive compounds of banana samples.

Table 3 .
Effect of different drying methods on the volatile composition (relative peak area in %) of banana samples.