**Preface to "Advances in Food Processing (Food Preservation, Food Safety, Quality and Manufacturing Processes)"**

The aim of this e-book and Special Issue is to compile advances in the area of food manufacturing including packaging to address issues of food safety, quality, fraud, and how these processes (new or old) could affect the organoleptic characteristics of foods, with the aim to promote consumers' satisfaction. Moreover, food supply issues are explored. This Special Issue shows that new and improved technologies have been employed in the area of food manufacturing to address consumer needs in terms of quality and safety. The issues of research and development should be taken into account seriously before launching a new product onto the market. Finally, food fraud and authenticity are very important issues, and the food industry should focus on addresssing them.

> **Theodoros Varzakas, Panagiotis Tsarouhas** *Editors*

### *Editorial* **Advances in Food Processing (Food Preservation, Food Safety, Quality and Manufacturing Processes)**

**Theodoros Varzakas 1,\* and Panagiotis Tsarouhas 2,\***


The aim of this special issue was to bring about advances in the area of food manufacturing, including packaging, addressing issues of food safety, quality, fraud and how these processes (new and old) could affect the organoleptic characteristics of foods, with the aim of promoting consumer satisfaction. Moreover, food supply issues have also been explored.

In this direction, packaging is the last step in the manufacturing process and makes the product attractive. In addition, it can extend the shelf life of the product. Hence, edible coatings have been used extensively. Acevedo-Correa et al. [1] report on the effect on cassava chips of coatings consisting of pectin and whey protein films. The coating type affected all sensory parameters except crispness, and temperature only influenced the color of the control chips.

Ishkeh et al. [2] employed chitosan as an edible coating to increase the storage of raspberries, and nanoparticles of chitosan were used to increase chitosan efficiency. This methodology was considered effective, safe and environmentally friendly, with healthpromoting effects.

In regard to the development of new technologies and the advancement of existing technologies, Negara et al. [3] developed high-frequency defrosting, superheated steam and quick-freezing treatments to improve the quality of seafood home meal replacement products. They examined the chemical, microbiological and organoleptic properties of the products and saw no statistical changes over 90 days. The optimal temperature and heating time for this technology were also determined.

Rana et al. [4] also employed superheated steam roasting (270 ◦C for 4 min) and hot smoking (70 ◦C) of chub mackerel fillets to extend their shelf-life at the market level. Oak sawdust for 25 min of smoking time offered the best organoleptic characteristics, along with the optimal physiochemical and microbiological parameters.

Ultrasonic pretreatment was employed in the work of Taghinezhad et al. [5] for drying kiwi fruits under hybrid hot air–infrared conditions. The effective moisture diffusivity coefficient (Deff) was estimated, along with other parameters, and different models were used. The results revealed the effect of temperature and ultrasonic pretreatment time, along with sample thickness, on these parameters.

The study by Liu et al. [6] enabled calcium alginate ball encapsulation in fruit-flavored drinks (Boba milk tea made of tapioca). This technology showed that the organoleptic control and microbiological quality of the products were not affected if the product was stored at a temperature less than 10 ◦C.

Diaz-Bustamante et al. [7] contributed to the design of costeño-type artisan cheeses with specific functionalities through multiscale modeling by means of the substitution and reduction of NaCl and an increase in cooking temperature. The microstructure was not affected by the reduction in salt or by modifications in the cheese-making process.

The next two studies, which were carried out in Greece, investigated safety and quality aspects. Voidarou et al. [8] screened raw unprocessed honeycombs filled with oregano

**Citation:** Varzakas, T.; Tsarouhas, P. Advances in Food Processing (Food Preservation, Food Safety, Quality and Manufacturing Processes). *Appl. Sci.* **2021**, *11*, 5417. https://doi.org/ 10.3390/app11125417

Received: 31 May 2021 Accepted: 10 June 2021 Published: 10 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

honey from Epirus, Greece, for bacteriocinogenic lactic acid bacteria and *Bifidobacterium* spp. exerting inhibitory action towards some pathogens and spoilage microorganisms isolated from fresh fruits and vegetables. The results suggested that bacteriocin-like substances are involved in this process.

Skiada et al. [9] evaluated and discriminated between monovarietal extra virgin olive oils obtained from olive cultivars cv. Lianolia Kerkyras and cv. Koroneiki, produced in Greece, based on their chemical characteristics, followed by statistical and chemometric analysis. They showed that possible authenticity tools could include sterol and fatty acid composition markers.

The work of Agustinisari et al. [10] deals with research and development, and more specifically with the formulation of protein and polysaccharides (whey proteinmaltodextrin) in eugenol encapsulation and the determination of the effect of eugenol and chitosan concentrations on the characteristics of emulsions and spray-dried powders. The presence of chitosan resulted in more stable emulsions.

The next study, by Cucci et al. [11], reports on a new methodology for the rapid measurement of the redox potential as an indicator of color changes for meat juice in carcasses without slowing down the slaughter line to avoid food waste. A symmetry was highlighted between these two parameters.

Food fraud is very frequent nowadays and could become very complex. However, analytical tools have been employed to address this phenomenon. MALDI-TOF MS has been used in different matrices and has a lot of advantages; hence, a review by Zambonin [12] demonstrates all the existing applications for the detection of food fraud.

Supply chain management needs to take into account issues of corporate social responsibility. A case study was employed here, representing the dairy sector in Vietnam. The study shows the close relationship between the company and consumers, which is reflected in its financial performance [13].

Xu et al. [14] proposed an advanced online-to-offline (O2O) strategy for a singlevendor single-retailer integrated system with coordination mechanisms consisting of revenue-sharing, buy-back and quantity flexibility.

This special issue has shown that new and improved technologies have been employed in the area of food manufacturing, addressing consumer needs in terms of quality and safety. The issues of research and development should be taken into account before launching a new product into the market. Finally, food fraud and authenticity are very important issues and the food industry should focus on these.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Review* **MALDI-TOF Mass Spectrometry Applications for Food Fraud Detection**

**Carlo Zambonin**

Dipartimento di Chimica, Università degli Studi di Bari "Aldo Moro", Via E. Orabona 4, 70126 Bari, Italy; carlo.zambonin@uniba.it

**Abstract:** Chemical analysis of food products relating to the detection of the most common frauds is a complex task due to the complexity of the matrices and the unknown nature of most processes. Moreover, frauds are becoming more and more sophisticated, making the development of reliable, rapid, cost-effective new analytical methods for food control even more pressing. Over the years, MALDI-TOF MS has demonstrated the potential to meet this need, also due to a series of undeniable intrinsic advantages including ease of use, fast data collection, and capability to obtain valuable information even from complex samples subjected to simple pre-treatment procedures. These features have been conveniently exploited in the field of food frauds in several matrices, including milk and dairy products, oils, fish and seafood, meat, fruit, vegetables, and a few other categories. The present review provides a comprehensive overview of the existing MALDI-based applications for food quality assessment and detection of adulterations.

**Keywords:** MALDI-TOF; applications; food; fraud; adulteration; quality

**Citation:** Zambonin, C. MALDI-TOF Mass Spectrometry Applications for Food Fraud Detection. *Appl. Sci.* **2021**, *11*, 3374. https://doi.org/10.3390/ app11083374

Academic Editors: Theodoros Varzakas and Tsarouhas Panagiotis

Received: 16 March 2021 Accepted: 7 April 2021 Published: 9 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Introduction**

Defining and identifying the types of food fraud is still an open issue since the relevant literature is characterized by a considerable variability of the related terms [1]. The European Commission defines food frauds as "intentional actions by businesses or individuals for the purpose of deceiving purchasers and gaining undue advantage therefrom, in violation of the rules referred to in Article 1(2) of Regulation (EU) 2017/625 (the agri-food chain legislation)" [2]. The most common foods subjected to illegal manipulations [3] include oil, fish, honey, milk and dairy products, meat, grain-based foods, fruit juices, wine and alcoholic beverages, organic foods, spices, coffee, tea, and many others.

According to the European Commission [2,4], the most common fraud is adulteration, which can occur in food products in different ways: "replacing a nutrient, an ingredient, a food or part of a food with another one with lower value" (substitution), "mixing an ingredient with high value with an ingredient with a lower value" (dilution), "adding unknown and undeclared compounds to food products in order to enhance their quality attributes" (unapproved enhancement), and "hiding the low quality of food ingredients or products" (concealment). Food frauds can also take others forms, including mislabeling, the process of putting false claims on packaging for economic gain, counterfeiting, when intellectual property rights are infringed, and grey market, referring to unauthorized sales channels for products.

Based on the above consideration, a chemical analysis on a given food can be carried out for many reasons, such as individuation of adulteration, authentication and traceability, confirmation of geographical origin assessment of toxicity, and many others. Food sample analysis, then, constitutes a crucial challenge for analytical chemistry and numerous scientists are focused on the development of reliable, fast, cost-effective analytical processes to solve analytical problems related to food frauds. Food-based matrices are complex and characterized by a wide range of chemical composition that influence the performance of chemical analytical measurements, while the nature of the manipulation is often unknown, making the analysis even more complicated. Moreover, as technologies develop to detect deceptions, they become more sophisticated since fraudulent suppliers also adapt to finding new ways to circumvent the controls.

Targeted and non-targeted analytical approaches are mainly used for food controls [5–7]. A targeted analysis, generally laborious and time-consuming, is based on the knowledge of the contaminants and focused on the detection of one or a few classes of compounds that represent the markers of the fraud. Since in most cases the type of fraud is unknown, a non-targeted analysis is often necessary. This approach relies on the fast instrumental acquisition of the chemical profile of the whole foodstuff sample, usually represented by spectra, which should provide a unique fingerprint as a reference for suspect samples, often with the help of chemometric data handling [8].

Spectroscopic techniques [5,9–16], namely nuclear magnetic resonance (NMR), nearand mid-infrared (NIR, MIR), Raman, ultraviolet–visible (UV-VIS), and X-ray fluorescence spectroscopy (XRF), have been largely used to successfully carry out non-targeted analysis. Different mass spectrometry (MS) techniques and ionization approaches, including matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF), electrospray (ESI), ambient mass spectrometry (AMS), and high-resolution mass spectrometry (HRMS), have been also widely employed for the same purpose [6,17–23], and their constant advances have enabled the development of numerous new methodologies for food quality and safety controls.

Among them, MALDI-TOF represents an ideal option for fast, reliable, and accurate detection of food frauds due to some intrinsic characteristics such as ease of use and speed with which data can be collected even from complex samples. MALDI MS is based on the use of a pulsed laser beam that hits the sample, represented by co-crystals of the so-called "matrix," present in a vast excess, and the analyte. The laser energy is absorbed by the matrix that vaporized carrying intact analyte molecules into the vapor phase. During this process, ions (mainly H<sup>+</sup> and Na<sup>+</sup> ) are released by the matrix leading to the formation of charged analyte molecules. Moreover, anions can also be generated by abstracting H<sup>+</sup> or Na<sup>+</sup> from the analyte. After being accelerated in an electric field and passing a charged grid, the ions are separated in a TOF mass analyzer based on their *m*/*z* ratios (low mass ions arrive at the detector in a shorter time than high mass ions). A "linear" geometry of TOF analyzers is used for the analysis of high molecular weight molecules. Smaller molecules can be analyzed using a "reflectron" configuration able to balance the different initial velocities of the ions during the vaporization process, leading to a consistent improvement of resolution. MALDI mass spectrometers are capable to rapidly generate spectra, profiles, and/or fingerprints from food matrices, often with simple preparation or no sample preparation at all. The potential of the MALDI approach applied to the fight against food fraud, sometimes coupled to multivariate statistical methods to extract information from complex analytical data such as mass spectra, has been clearly understood, as demonstrated by a considerable number of related publications. The purpose of the present review is to provide a comprehensive overview of the relevant literature. Based on the research carried out, most of the existing MALDI applications dedicated to the field of food fraud have concerned the analysis of milk and dairy products, oils, fish and seafood, vegetables, fruit, meat, and, to a minor extent, a few other categories. Most of the works have been focused on the evaluation of food quality and detection of adulterations, while only few applications have been devoted to the investigation of the geographical origin.

#### **2. MALDI Applications**

#### *2.1. Milk and Dairy Products*

Due to their high nutritional value, milk and derived products are largely consumed all over the world. According to the Food and Agriculture Organization of the United Nations (FAO), worldwide milk production has grown from 522 to 798 million tons in three decades (1986–2016) [24] and was expected to grow in 2020 to 859 million tons [25]. The increase in consumption also increases the number of dishonest producers willing to

commit fraud to increase their earnings, making the safety and authenticity of milk and derived products an area of growing attention and concern, as demonstrated by several regulations and governing bodies [26,27]. Traditionally, milk dilution by water has been the most common illegal practice, together with the selling of skimmed milk and semiskimmed milk instead of whole milk. As detection methods technologies has improved, fraudulent suppliers evolved [28] towards most sophisticated fraudulent approaches, such as the addition of milk anhydrous products (caseins and caseinate, milk protein concentrate, whey proteins) to liquid milk and the mixing of high-quality milk, such as buffalo, sheep, and goat, with less expensive products, such as cow milk. Many MALDI-based analytical methods have been developed and improved in the last years to assess the authenticity of milk and dairy products.

Buffalo milk has been extensively investigated by different MALDI proteomic approaches searching for possible adulteration. Cozzolino and co-workers have developed two different methods for the identification of adulteration of water buffalo and sheep milk and water buffalo mozzarella, respectively. In the first work [29], the investigation was performed on raw buffalo and sheep milk samples. The determination of the presence of bovine milk or the addition of powdered milk to fresh raw milk was accomplished by evaluating the protein profiles coming from the most abundant whey proteins, lactalbumin, and lactoglobulins. In the second work [30], patterns of the same whey proteins were obtained from the direct analysis of water buffalo mozzarella cheese, which was able to differentiate between mozzarella cheeses made from pure water buffalo milk or from mixtures of bovine-sheep-buffalo milks. Interesting quality information on the partial resistance of the detected proteins to the thermal and enzymatic processes involved in the cheese production was also inferred. In both methods, mixtures with different percentages of less expensive milk added to water buffalo milk were prepared, and detection limits below 5% were always obtained in adulterated milk or cheese, respectively. The strategies were also fast, accurate, and practically did not require sample pre-treatment.

Sassi at al. have successfully optimized an integrated MALDI-TOF platform [31] for obtaining milk peptidomic/proteomic profiles to detect the addition of either nondeclared bovine material to water buffalo, goat, and ovine milks, or of powdered bovine milk to the fresh bovine milk. Milk samples were directly analyzed permitting a fast individuation of illegal adulterations at protein and peptide level and allowing the identification of unique diagnostic ions of thermal treatment in different types of commercial milks.

A further MALDI proteomic study for the detection of ricotta buffalo cheese adulteration with bovine milk by obtaining peptide profiles for both matrices was developed [32], followed by the search for signals that could represent specific markers for each type of dairy product. A peptide marker corresponding to the region 149–162 of β-lactoglobulin was correlated to the univocal presence of bovine milk in ricotta buffalo cheese at a 5% level.

According to the European reference method, the fraudulent addition of bovine milk in water buffalo milk can be individuated by concomitant isoelectric focusing detection of bovine γ2- and γ3-casein fragments after plasminolysis. However, this approach can produce false positive results due to a water buffalo β-casein peptide, which is also formed after plasminolysis of water buffalo milk and comigrates in isoelectric focusing with bovine γ2-casein. Thus, Caira at al. have developed a proteomic approach [33] to obtain the unambiguous detection of bovine milk in water buffalo milk and derived products (LOD of 0.8% *v/v*) based on the MALDI determination of specific bovine and water buffalo β-casein phosphopeptide markers. The procedure was proposed as an integrative/alternative to the reference method.

Other studies were focused on the development of MALDI methods for the detection of the addition of sheep and goat milk with the less expensive bovine milk. Contrary to some of the above-mentioned works, which were focused on the identification of specific signals that can work as adulteration markers, one study, divided into two parts [34,35], proposed a different approach. The authors suggested that, in the analysis of a complex matrix such as milk, the simultaneous use of many variables reveals more information

compared to the use of single or few variables and proposed a method for quantitative and multivariate use of the whole MALDI mass spectra of milk samples to determine adulteration. The first part of the work [34] was focused on the optimization of sample preparation and instrument setup, the second on the quantitative determination of cow, goat, and sheep milk in mixed milk samples [35]. The results obtained seemed to confirm the original hypothesis since, with optimal parameter settings, multivariate regression on whole spectra allowed to determine the concentrations of milk in mixtures with good accuracy. A further study to detect the presence of cow milk in goat and sheep milk using whole MALDI mass spectra in combination with multivariate techniques was reported by Nicolau at al. [36]. Binary and tertiary mixtures of cow, sheep, and goat milk were easily and rapidly analyzed and accurate information on the amount and type of adulteration were obtained, with typical errors in the range 2–10% for cow milk.

The potential of proteomic approaches to fight illegal manipulations of milk was again demonstrated [37] by the detection of cow milk-specific peptide markers in sheep and goat milk and goat milk-specific peptide markers in sheep milk, by analyzing whole milk samples mixtures subjected to in-solution tryptic digestion. Seven peptide markers of cow milk and two peptide markers of goat milk were identified by MALDI-TOF analysis and the approach was able to detect adulteration up to a 5% level. Moreover, the same markers were found in cheese samples, demonstrating the applicability of the procedure even to milk-derived products. The authors stated that the use of the α-cyano-4-chlorocinnamic acid matrix was essential to reach the high sensitivity observed for the marker peptides. Moreover, the results were found in good agreement with those obtained with a traditional approach based on SDS-PAGE/in-gel digestion. The same research group reported another application [38] for the extraction and MALDI determination of phospholipids in milk, using the same matrix. The method was able to provide peculiar milk phospholipid profiles that were used for the detection of cow milk in sheep and goat milk. The abundance ratio of specific ions (*m*/*z* 703 and 706) was found to be species-specific and was used for the identification of the adulteration.

Very recently, liquid atmospheric pressure—MALDI mass spectrometry—was used [39] to optimize a very accurate approach to classify goat and sheep milk and sheep milk containing 10% of goat milk, to evaluate colostrum quality and postnatal stages, through the recording of the milk lipid/protein profiles or the detected orthologs of single proteins.

Donkey milk is a safe alternative for individuals which are allergic to cow milk and is often the object of adulteration, the detection of which is of critical importance for the safety of the consumers. The MALDI mass spectra profiles of α-lactalbumin and β-lactoglobulin were used to detect the fraudulent addition of goat or cow milk in donkey milk [40] in a protocol developed for routine analysis and potentially extendable to other milk species. Detection limits for the analysis of defatted milk samples were in the range 0.5–2.0%, comparable to those obtainable with traditional complex approaches. Despite the typical MALDI quantitative issues, the measured values were in good agreement with the actual composition of the analyzed mixtures. Another study [41] investigated donkey and goat milk adulteration by cow and sheep milk using MALDI in combination with unsupervised hierarchical clustering, principal component (PCA), and Pearson's correlation, focusing on the mass spectra profiles in the *m*/*z* mass range 2000–25,000 Da. The approach was shown to be rapid, robust, and sensitive, with detection limits of 0.5%.

The adulteration of fresh milk by reconstituted milk and the selling of reconstituted milk as fresh product is economically advantageous when either a surplus of milk powder exists, or when the importation of dried milk powder is subsidized. Moreover, the addition of powdered derivatives is difficult to detect because the adulterant materials have almost the same chemical composition of liquid milk. Specific peptide markers, attributable to modified whey proteins and/or caseins, formed by thermal degradation during milk powder production, were identified and used to detect the presence of powdered milk in liquid milk samples [42]. Whey and casein fractions of milk samples were directly digested in solution and analyzed by MALDI using α-cyano-4-chlorocinnamic acid as a matrix, which enhanced the detection of more acid peptides. The approach was able to detect the peptides diagnostic for the presence of powder in liquid milk even at a 1% level. The results were in good agreement with those obtained with a reliable but time-consuming 2D gel approach.

On the contrary, milk powder itself could be the object of adulteration when non-milk fat such as vegetable oils and fats are added. Garcia and co-workers [43] successfully developed a MALDI-qTOF method for the fast individuation of non-milk fats and oils in milk powder. Samples were subjected to a simple n-hexane extraction prior to MS analysis, which provided rapid and unambiguous profiles of the triacylglycerols (TAGs) composition capable to characterize the adulterant and estimate the adulteration level.

Other MALDI applications on the detection of frauds were focused on different kind of milks and adulterations, as well as to the analysis of cheese samples. Hinz at al. [44] compared the principal proteins in bovine, caprine, buffalo, equine, and camel milk, finding interesting differences between the species that could be used to identify sources of hypoallergenic alternatives to bovine milk and detect adulteration of milk samples and derived products. A proteomic approach for the MALDI-TOF analysis of commercial bovine milk, based on in-solution digestion of the whole samples, was also suggested [45] for the routine analysis of raw and processed foods and to detect adulterations.

Due to similar properties to bovine milk, soya milk is added to bovine milk for revenue maximization. In a recent work, England and colleagues [46] used MALDI to develop a solventless, sensitive, and cost-effective method for the discrimination of bovine milk from soya and coconut milk as well as for the individuation of milk adulteration. Samples were simply diluted in water, combined with the matrix, and subjected to instrumental analysis, which allowed to obtain unique lipid profiles (mainly phosphatidylcholine and TAGs).

Magenis at al. reported a qualitative method to check the authenticity of a typical Brazilian cheese using β-lactoglobulin as an adulteration marker [47], showing good precision and sensitivity (7 mg/g). The procedure involved SDS–PAGE and in-gel trypsin digestion followed by MALDI analysis. The applicability to real samples was demonstrated by the analysis of 42 commercial samples, 18 of which were found to be adulterated.

Direct Imprinting in Glass Surface Mass Spectrometry (DIGS-MS) for qualitative cheese analysis in a MALDI instrument was also successfully used [48] to identify complex lipids to be used as quality and/or adulteration marker in different cheese samples. The integration of analytical and statistical data could also be employed for the control of productive stages.

In another recent work [49], Rau and co-workers developed an original method to identify the dairy animal species of mozzarella and white brined cheese by MALDI-TOF in combination with direct protein extraction without tryptic digestion and a small in-house reference spectra database.

Some further MALDI works were oriented towards the quality evaluation of milk and dairy products. The selective extraction of phospholipids from dairy products such as milk, chocolate milk, and butter by micro-solid phase extraction (µ-SPE), based on homemade titanium dioxide (TiO2) microcolumns, followed by MALDI-TOF MS, was developed by Calvano at al. [50]. Since α-lactalbumin and β-lactoglobulin are among the main cow milk allergens, Gasilova at al. [51] developed a sensitive quantitative method for their determination by immunoaffinity capillary electrophoresis, using magnetic beads functionalized with opportune antibodies, coupled to MALDI. A transient isotachophoresis preconcentration step allowed to obtain LOD values of 0.02 and 0.03 µg/mL for β-lactoglobulin and α-lactalbumin, respectively, suitable for allergen detection. The method was then successfully tested on cow milk and fortified soy milk, proving the capability to determine the analytes at both high and low concentration levels.

To identify and characterize oxidized and glycated phospholipids in heat-treated food, namely milk powders, pasteurized milk, ultra-high-temperature milk, and soy flour, an extraction protocol [52] was developed by means of a methanolic solution of 1,8-bis(dimethylamino) naphthalene (DMAN), used as both extraction medium and matrix

for the successive MALDI detection, performed in negative ion mode. Thermally modified lipid products were first characterized by heating representative standards and eventually determined in real samples. MALDI-TOF, combined with C18-stage tip extraction, was also used [53] to rapidly obtain peptide profiles from different commercial milk products in view of the identification of possible changes induced by different heating regimens and storage conditions. Several peptide ions showed relative abundance variations following specific treatments, such as heating or proteolytic activity of enzymes during storage, suggesting their potential use as markers to draw information on milk types and freshness. A similar approach permitted [54] the generation of polypeptide profiles from buffalo milk samples subjected to different freezer storage times, to assess their freshness through the identification of specific markers. The statistical evaluation of data relevant to the analysis of several fresh and frozen samples allowed to identify 28 polypeptide markers of freezing storage originated from the breakdown of buffalo proteins, being α-lactalbumin, β-lactoglobulin, γ2-, γ3-, and γ4-caseins, β-casein-derived phosphopeptides, and GLY-CAM1 phosphorylated peptides as the most significant components. Their progressive formation even in freezing conditions was attributed to an unknown protease stable at low temperatures. MALDI profiling of milk proteins, in combination with multivariate data analyses, also proved to be an optimal mean to discriminate between different milk types [55]. Protein fingerprints of mammalian (various species), human (at different lactation stages), and formula (different brands) milks were rapidly obtained, permitting to obtain key information on the matrices under study.

#### *2.2. Oils*

Olive oil is one of the main ingredients in the Mediterranean diet [56]. The International Olive Council defines two categories of olive oil [57], the first comprising oils suitable for consumption, such as extra-virgin, virgin, and ordinary olive oil, the second comprising oils that must be further processed prior to ingestion. Extra-virgin olive oil (EVOO) is the most valuable, since it is obtained from olive fruits using mechanical processes or other physical means [58] that lead to a product of unmatched value, characterized by unique features, such as nutritional quality, health benefits, and pleasant flavor. These features make EVOOs expensive products that are continuously counterfeited in many countries [59] with oils produced from cheaper fruits/seeds or with other lower quality olive oils or even mislabeling virgin/refined olive oils. Of course, analytical methods are crucial for detecting EVOO frauds [60] and many scientists are committed to the development of MALDI-TOF MS applications.

A matrix-less laser desorption/ionization (LDI) approach using a stainless-steel target plate was proposed [61] for the fast analysis of diluted soy, sunflower, and extra-virgin olive oil samples. MS spectra characterized by oil-specific profiles and free of MALDI-matrix peaks were obtained, allowing the easy discrimination of the oil under study. Most of the *m*/*z* ions present in the spectrum were easily attributed to tri/diacylglycerols compounds, some of which were found to be diagnostic ions for olive (*m*/*z* 907.77) and sunflower oil (903.79 and 901.78), thus permitting the detection of adulteration of olive with sunflower oil. All the diagnostic ions were potentially attributable to different compounds, namely OOO, SOL, SLO, LSO (907.77), OLL, LOL, SLLn, LSLn, SLnL, OOLn, OLnO (903.79), LLL, OLLn, OLnL, and LOLn (901.78), with O being oleic acid, S, stearic acid, L, linoleic acid, and Ln, linolenic acid. Over the following years, the same research group published three more papers focused on the development of MALDI methods for the challenging detection of hidden hazelnut oil in extra virgin olive oil, each time using different classes of compounds as markers for the adulteration. In the first work [62], the polar fraction of oils was enriched and characterized using hydrophilic liquid chromatography micro-columns coupled with MALDI. Lysophosphatidylcholine (LPC) (16:0/0:0), LPC (18:1/0:0), and LPC (18:2/0:0) were identified and used as diagnostic compounds for the presence of hazelnut oil in EVOO to a level of 5%. The second application [63] exploited a modified Bligh–Dyer method for the selective extraction of phospholipids, present in seed oils at much higher concentration

levels and then used as markers for hazelnut in olive oil. The solution resulting from the combination of α-Cyano-4-hydroxycinnamic acid (176 mM) and tributylamine (equimolar) was used as both extraction solvent and MALDI matrix. The method was capable to detect adulterations at a 1% contamination level. The last work [64] used cold acetone precipitation followed by in-solution tryptic digestion and final MALDI analysis for the revealing of hazelnut peptide markers arising from the main hazelnut proteins Cor a 9, Cor a 11, and Cor a 1. These markers can potentially be used for the determination of hazelnut traces in oils and processed foods. SDS–PAGE analysis confirmed the presence of hazelnut proteins in hazelnut extracts with molecular masses in the range of 10–60 kDa.

Arlorio and colleagues [65] reported that the adulteration of extra virgin olive oil with solvent-extracted hazelnut oil can be directly traced down to a 1% level by SDS– PAGE analysis of hazelnut proteins, even if a MALDI analysis was indeed necessary to confirm the identity of the alleged allergens, i.e., two oleosin isoforms and Cor a 9. MALDI-TOF, combined with unsupervised hierarchical clustering, principal component analysis, and Pearson's correlation analysis, was also successfully adopted [66] to detect very low amounts (0.5%) of corn oil in EVOO.

As described in the work of Calvano at al. [61], when the fraud to be revealed concerns the presence of seed oils in EVOO, it may be sufficient to target their TAGs profile to characterize the samples and eventually determine the presence of the adulterant product. In fact, two more applications based on MALDI in conjunction with statistical approaches were successfully developed for the detection of sunflower (or refined) [67] and canola [68] oils in EVOO, respectively, using the TAGs profiles present in the relevant mass spectra.

A hardly detectable adulteration is represented by the addition of sunflower oil to poppy seed oil because of the similar fatty acid ratios. In fact, a further application [69] reported the successful MALDI-TOF MS detection of mixtures of sunflower oil with high levels of triolein (high-oleic acid type) down to the 5–10% level, but the same approach failed to discover adulteration of pure poppy seed oil by sunflower oil.

Several works were focused on the quality evaluation of edible oils (mainly EVOO). Taking into account that squalene concentrations in olive oil vary considerably depending on the cultivar, its determination in this matrix should be able to discriminate between varieties. Thus, Zambonin at al. [70] reported on the LDI-TOF MS determination of squalene and derived oxides involved in the biosynthetic pathway of cholesterol in extra virgin olive oil. The same research group used the same matrix-less approach to characterize olive and sunflower oils before and after thermally assisted oxidation [71]. The obtained MS profiles provided information about the identity of thermally induced oxidation products, such as TAGs epoxy/hydroxy, hydroperoxy derivatives, and β-scission products, making the method a tool to rapidly check the quality of cooking oil. TAGs thermal oxidation of sunflower and olive oils was also studied by Picariello at al. two years later [72]. To increase the detection of oxidized components, a chromatographic separation of polar and non-polar compounds was performed on silica gel before MALDI analysis. Tri- and diacylglycerols, TAGs oxidative dimers, oxidized TAGs, and TAGs fragments arising from the β-scission of linoleyl, peroxy, and alkoxy radicals were observed in the relevant spectra.

For authentication and characterization purposes, detailed MALDI TAG profiles of olive oils were also successfully obtained from samples coming from six different cultivars [73] and of two different varieties grown in the same area at different olive ripening stages [74].

Shen and co-workers [75] optimized a matrix solid-phase dispersion (MSPD) procedure to extract several phospholipids from olive fruit and oil samples, exploiting the ability of the sorbent (TiO<sup>2</sup> nanoparticles) to selectively interact with the phosphate group of the analyte by a chelating bidentate bond. After elution, MALDI-TOF analysis in both positive and negative ion modes was performed. The method proved to possess a great potential in lipidomic fingerprinting of olive samples for quality control.

Different analytical instrument, namely MALDI, GC, and LC-MS, were synergically used to perform a study of the lipid composition of commercial extra virgin and virgin olive oils [76], in an attempt to define specific lipid patterns and/or establish a set of lipid markers representing the identity of a specific oil. Significant differences between EVOOs and VOOs were found, and five classes of phospholipids were identified in the polar lipid fraction, with remarkable variations in phosphatidylcholines.

MALDI-SpiralTOF was employed [77] to obtain TAG fingerprints relevant to sesame, sunflower, and olive oils to draw traceability and authenticity information. The application of PCA allowed the discrimination of Istrian olive oils from those coming from other Croatian coastal regions. Furthermore, high energy collision-induced dissociation (CID) MALDI-TOF/TOF analysis of TAGs was suggested as a mean that could correlate oil TAGs to the geographical origin, analyzing a high number of samples that can provide statistically significant data. A multi-instrumental analysis (MALDI for TAGs, GC-MS for fatty acids, and NIRS for non-selective analysis) coupled to PCA and partial least square-discriminant analysis (PLS-DA) was also used [78] for geographical origin sample grouping evaluation and for testing of predictive capabilities of measured variables on accurate classification of EVOO regional category.

A synergic MALDI-TOF/GC-MS approach was proposed to study the different components of pomegranate oil [79] and of several nut oil varieties [80], respectively. In particular, MALDI confirmed its ability and reliability to profile the TAG component, showing unique fingerprints that allows to differentiate pomegranate from most edible oils and nut oils exhibiting quite similar fatty acid composition (hazelnut, pistachio, and beech oil) between each other.

Unique TAGs profiles were also displayed [81] by different Amazonian oils and fats, characterized by MALDI-TOF without prior separation. The variable combinations of fatty acids provided detailed information that could permit the development of a database from spectra to be used for fast and reliable typification, screening, and quality control. The triacylglycerol content of shea butter fat, palm kernel oil, and peanut oil was also rapidly characterized [82]. The potential of the method as a tool for quality control of the matrices under study was demonstrated by the detection, in addition to intact and specific triacylglycerols, of oxygenated and fragmented TAGs, likely arising from poor handling and production processes.

#### *2.3. Fish and Meat*

Fish represents and is perceived by consumers as a healthy and nutritious food resource. However, a common problem in the fish processing industry is adding or replacing cheap fish instead of expensive ones, as an accidental event due to the lack of experience or as a deliberate fraud. An adequate level of protection is then required to ensure seafood quality and safety, making the development of rapid, effective, accurate, and reliable analytical methods the key to effectively supervising the market. This need has prompted the issue of directives and regulations for quality control and encouraged the development of innovative MALDI analytical methodologies, almost always based on proteomic techniques.

Highly specific mass spectrometric profiles from 25 different fish species were obtained using a new MALDI method [83] meant for the fast assessment of authenticity and fraudulent substitutions. Specific protein signals at *m*/*z* values around 11 kDa were selected as markers to discriminate the various species, while a structural characterization permitted to identify some major fish allergens arising from parvalbumin.

In a pilot study [84], three freshwater fish species were discriminated based on the cluster analyses of the overall profiles generated by the MALDI analyses of muscle and liver tissues, after a simple single-step extraction procedure. Each tissue provided speciesspecific profiles, even if muscle is more suitable for routine controls since fillets are more commonly commercialized. The authors pointed out that the clear discrimination of the species was possible even though the settings and statistical algorithms used were originally designed for the analysis and identification of bacteria and fungi. Then, they suggested that performing an optimization of mass spectral analysis and data processing

targeted on fish species could lead to exhaustive evaluation of the potential of MALDI analysis for studies on freshwater fishes.

The analysis of muscles taken from two closely related fish species was undertaken by two-dimensional gel electrophoresis (2-DE) [85] and the interspecies differences between the protein spots, which could be useful for their differentiation, were visually identified. MALDI-TOF MS and/or LC-MS were then necessary to identify 19 proteins, some of which resulted to be specific for each species. Another MALDI method [86] for the profiling of protein extracted from fish muscle was reported in a book chapter by Siciliano at al. for fish authentication purposes and to detect fraudulent substitutions.

Since universal sample preparation protocols prior to MALDI analysis were available to other species but not to fish, Spielmann at al. [87] tested the performances of five preparation protocols to verify the capability to produce reproducible and high-quality spectra dependent on storage conditions and food processing, and eventually differentiate between different fish species. After the optimization of the best protocol, it was concluded that MALDI could be used for the purpose as soon as a valid species database is available. This need was evidently shared by Stahl and co-workers, who almost simultaneously used MALDI to establish a database [88] of protein patterns from 54 fish species susceptible to fraud, in order to detect and prevent substitution, characterized by low intraspecies but high interspecies variability.

MALDI-TOF combined with multivariate analysis was exploited [89] for the development of a method to directly analyze fish skin or muscle tissues surface, for authentication purposes. Samples were quickly discriminated, hinting at potential future developments for the authentication of seafood in general and/or other protein food products.

Another study centered on the comparison of several one-step sample pre-treatment protocols of fish muscle tissues, prior to MALDI determination of the relevant extracts, was recently carried out [90] by Wang and colleagues. The best results in terms of repeatability and spectral resolution were obtained by boiling samples in 0.1 M trifluoroacetic acid (TFA) for 5 min, permitting to discriminate between different fish samples, when combined with similarity coefficient-based analysis for their mass spectra.

MALDI also turned out to be a good system for assessing fish freshness. For instance, the effect of storage time on fish muscle proteins was investigated using 2-DE coupled to MALDI [91], permitting the identification of three altered proteins. More recently [92], the same approach was involved in the analysis of fish samples, both fresh and stored in a refrigerator for different days, looking for protein markers for its freshness. It was found that the comparison of three specific proteins, l-lactate dehydrogenase, adenylate kinase isoenzyme 1, and myosin heavy chain, permitted to trace the freshness of the products. In a different study, the vitreous fluid of the eyes was taken from fishes subjected to different days of post-mortem storage, which was immediately analyzed by MALDI-TOF to evaluate fish spoilage [93]. Software spectra processing allowed the identification of four *m*/*z* ions able to differentiate between the tested days of storage, although a limited applicability was shown within the end of the tested period.

A further study [94] of a comparison of different protein extraction protocols from seafood (marine mussel) prior to 2-DE was conducted by Campos and co-workers. Then, MALDI-TOF/TOF analysis of the gels was performed and proteins with several functions, such as energy metabolism, cell signaling and regulation, and stress response, were identified. The method was a precious tool to investigate the protein expression in the species under investigation and potentially in other affiliated species.

Several analytical methods are used in the meat industry for detecting contamination, adulteration, and authenticity, to safeguard safety, and to reassure consumers, who are increasingly aware and attentive to the quality of the food they consume. Two MALDI methods to determine the origin of raw and processed meat (pork, beef, horse, veal, and chicken) and of gelatin (pork or beef), respectively, were developed by Flaudrops and colleagues [95]. In the first method, intact proteins were determined in linear mode prior to cluster analysis, allowing the separation of meat into distinct mass spectra clusters depend-

ing on the origin. In the second method, gelatin was digested with trypsin and analyzed in reflectron mode. The relevant spectra showed specific profiles that permitted to distinguish pork from bovine gelatin (1% of gelatin in spiked candies and 20% of pork gelatin in beef gelatin). Although less sensitive compared to other approaches, these methods are fast and easy to perform and can be conveniently used for fast screening controls.

An approach to reveal the differences in protein expression between young and old buffalo meat looking for biological markers of tenderness was performed by 2-DE coupled to MALDI-TOF/TOF [96]. Structural proteins with expression levels associated with meat tenderness were successfully identified through digestion and MS/MS analysis of selected gel spots. Moreover, the study demonstrated that through the ageing process, it was possible to reduce the variation in tenderness between young and old buffalo meat that, consequently, requires different processing strategies for their effective utilization. The 2-DE-MALDI combination was also exploited [97] to compare meat quality traits and to identify different protein expression between low and high pH pig muscles. Fourteen proteins that differed in spot density between the two groups were identified and nine of them involved in meat quality attributes significantly increased in high-pH muscles. The whole results obtained in the work provided useful information to understand the molecular mechanism at the base of meat quality.

Furthermore, the 2-DE-MALDI-TOF/TOF approach was again used [98] to assess the protein modifications occurring in the duck breast muscle collected form three breeds during the early post-mortem storage period. Evidence of changes in the protein expressions for each breed were found in several spots, some of which (10 for each breed) were subjected to MS/MS analysis, which allowed the identification of at total of 22 proteins.

#### *2.4. Fruits and Vegetables*

Intake of fruits and vegetables is highly recommended to promote good health due to their concentrations of bioactive compounds, including vitamins, minerals, antioxidants, and fiber. Indeed, the quality of these precious foods must be safeguarded and monitored using suitable analytical methods. MALDI-TOF was adopted for the development of many related applications. For instance, fruit skins were analyzed in two different works. It was demonstrated [99] that anthocyanins profiles of the berry skins of 23 red grape varieties can be easily and rapidly obtained by MALDI-TOF, enabling the identification of several varietal traits useful for the differentiation of authentic cultivars from hybrid ones on a molecular basis. Very recently, proanthocyanidin profiles of different fruit skins were obtained [100] by MALDI-TOF coupled to multivariate analysis, which proved to be a useful tool to evaluate authenticity in mixtures of different proanthocyanidin.

Three different MALDI studies were focused on the assessment of the authenticity of saffron, an expensive spice made from *Crocus sativus* L. dried stigmas, which is often the object of frauds. MALDI MS imaging (MSI) was successfully applied [101] to the determination of different crocins species in saffron. This original approach was also compared to a traditional one, i.e., MALDI analysis after solvent extraction, and the relevant results were found in good agreement. The other two studies were both performed by Aiello at al., using MS or MS/MS as mass analyzer and curcumin as the non-isotopic isobaric internal standard. In the first one [102], powdered saffron was subjected to a onestep sample pre-treatment and subjected to MALDI analysis. Crocins C-1–C-6, flavonols, unknown highly glycosylated crocins C-7, C-8, and C-9, and carotenoid-derived metabolites were detected. The method was shown to be fast, sensitive, reproducible, and suitable for routine quality controls, including the assessment of adulterations, as ascertained through the analysis of commercial samples. In the second one [103], picrocrocin was used as a saffron authenticity marker. Again, a rapid and simple method was developed, and very good quantitation parameters were obtained, with LOD (47.63 ppm) comparable or even lower than those obtained with other approaches.

The high demand for cereals, a major part of the diet in most countries providing nutrients such as vitamins, minerals, carbohydrates, fats, and proteins, makes relevant

frauds a severe global problem. Sequenom® MassARRAY® (Agena Bioscience, San Diego, CA, USA) MALDI-TOF MS, a reliable platform for detection and validation of single nucleotide polymorphisms (SNPs) for varietal analysis, was used to target specific genes in order to genotype and differentiate 35 barley varieties [104]. Information was drawn from the majority (33) of the tested loci (45), permitting the generation of a unique barcode of SNPs for each barley cultivar. This reliable and fast approach can potentially identify more than 150,000 SNPs per day, making the platform a suitable alternative for cereal varieta identification.

Jin at al. [105] used 2-DE followed by MALDI-TOF to compare the proteome extracted from barley malts of two selected cultivars, to find correlations between the protein content and malt qualities. Almost 700 total spots were detected; most were shared by the two cultivars, 377 of which were attributed to 192 proteins, mainly enzymes and enzyme inhibitors. The same research group performed another comparative study on green malts from two cultivars using two-dimensional fluorescent difference gel electrophoresis (2D-DIGE) [106]. Several metabolic proteins were identified by MS/MS and significant differences between cultivars were detected, suggesting potential applications for malt quality assessment. MALDI-TOF and two-dimensional difference gel electrophoresis (2D-DIGE) were also used by Fernando at al. [107] in a differential proteomic study on wheat grain, grown under normal CO<sup>2</sup> concentration and Free Air CO<sup>2</sup> Enrichment, respectively. It was found that a decrease of 9% in the protein content, in particular glutenin high molecular weight subunits associated with lower flour rheological properties and bread quality, occurred at higher CO<sup>2</sup> concentrations.

Since soybean is a known source of allergens that can also be hidden in other food commodities, a MALDI-TOF/TOF study was developed [108] to identify potential markers of its presence. Soybean protein extracts were digested with trypsin, separated using an LC system and eventually subjected to MS/MS analysis. Peptides arising from G1 glycinin and β-conglycinin were identified and proposed as potential markers for the detection of soybean protein traces in processed foods. Furthermore, these peptides were found to be stable when subjected to simulated food processing reactions, such as denaturation, Maillard reaction, and oxidation.

Rapid, simple, and reliable MALDI-TOF methods has been also proposed for cultivar characterization of hazelnut kernels through proteomic analysis [109], for geographical origin discrimination (coupled to multivariate analysis) of fermented salted vegetables comparing the relevant mass fingerprints [110] and to understand the effect of organic and inorganic crop nourishments on the nutritional quality of yardlong bean through twodimensional proteome profiling [111]. MALDI-TOF/TOF was used by Pinto at al. [112] for the untargeted analysis of dark chocolate from cocoa beans of different geographical areas to characterize the relevant metabolites (including sucrose, choline, sphingolipids, phospholipids, peptide material, and polyphenols) based on molecular weights and fragmentation patterns.

#### *2.5. Other Matrices*

Product quality, authentication, and identification of adulterants are ongoing challenges facing several other food commodities. Hence, other MALDI-based methods have been developed for the detection of different frauds in various matrices. MALDI coupled to capillary zone electrophoresis was successfully exploited [113] to rapidly detect cheap sweeteners (sucrose from cane or beet, starch hydrolysates) in orange juices by monitoring the variation of concentration ratio of low molecular mass saccharides, such as glucose, fructose, and sucrose, and/or the presence of compounds that are absent in the sugar profiles of citrus fruits.

Two other proteomic studies were devoted to the assessment of authenticity and traceability of wine. A fast approach proposed [114] the direct tryptic digestion and subsequent MALDI analysis of extracts of white wine produced from different *V. vinifera* cultivars. The detailed and peculiar peptide profiles present in the relevant mass spectra were converted

into simulated images that represented unique "mass codes" able to display differences between samples, suggesting their potential use for food quality assessment. The comparison of anthocyanin concentration, main organic acids and sugars, and proteomic profiles of berry with different geographical origin, grape variety, and ripening stage was performed by Fraige at al. [115]. In particular, 2-DE combined with MALDI was employed to obtain information about the proteome changes subjecting 128 gel spots to MS analysis, 80 of which were identified as proteins mainly expressed as a response to defense/stress or related to carbohydrate metabolism. A multivariate analysis of protein abundance was able to divide the analyzed samples in different groups according to the considered variables, clearly showing the potential of the approach to provide useful information for wine characterization.

Two strategies for the analysis of honey targeting different compounds were successfully developed to differentiate various types of honey and to detect adulterated samples, respectively. Won and colleagues [116] exploited the molecular weight differences (56 and 59 kDa, respectively) of major proteins of honey produced from different bee species to discriminate between two honey types by 2-DE coupled to MALDI analysis, while Qu at al. [117] targeted oligosaccharide and polysaccharide profiles to reveal adulterations by MALDI MS or MS/MS.

Edible insects are available on the European market in ground form, in which any visual identification is impossible, thus causing concrete risks of adulteration that imposes quality and safety controls of such products. A reliable MALDI method [118] was then developed for the detection of proteins in whole insect powder extracts. Species-specific mass spectra were obtained permitting the easy differentiation of the different species under investigation (buffalo worms, mealworms, crickets, and grasshoppers).

A further common fraud is represented by truffle counterfeiting, since different varieties are almost impossible to recognize visually, and less expensive species absorb the aroma of higher prized ones when closely stored together. MALDI analysis and clustering permitted [119] to distinguish seven *Tuber* species confirming misidentifications in 26% of commercial specimens. To the same purpose, MALDI-TOF was applied [120] for the generation of specific mass spectra from 73 truffle tubers belonging to eight different species. Both studies demonstrated the potential of MALDI-TOF to provide useful information for quality assurance and fraud control in the truffle market in a fast, reliable, and inexpensive way.

The entire food industry, as well as the field of dietary and sport nutritional supplements, suffers the occurrence of several frauds. *Echinacea* species (*E. purpurea* and *E. angustifolia*) are widely used in dietary supplements and their adulteration is a great concern. Greene and colleagues [121] developed a rapid automated sample-preparation system combined with MALDI analysis for the differentiation of *Echinacea* species by the obtained mass fingerprints, which was even able to confirm the published composition of a commercial product and identify signatures suggestive of additional product components. An approach involving in-solution digestion followed by MALDI-TOF was proposed for the fast and simple screening of food supplements declaring high value protein components [122]. In this case, contrary to the products label statements, proteins were absent or present at very low concentrations in most products, clearly indicating the potential of the method for quality control purposes. The results obtained by in solution digestion were validated by a comparison with SDS–PAGE followed by tryptic in-gel digestion.

#### **3. Conclusions**

The range of analytes typically targeted in MALDI-TOF MS analyses has expanded over the years from macromolecules, such as peptides and proteins, to smaller molecules, such as lipids, hydrocarbons, sugars, phenols, and many others, widening the possible applications of the technique, which has been projected towards fields not initially contemplated, including food quality evaluation and adulteration detection. Consequently, several MALDI-based applications for the fight against food frauds have been reported in the

literature over the years. The higher number of papers have been focused on milk, oils, fish and seafood, meat, fruits, and vegetables. In the case of milk, the target analytes have been mainly proteins and peptides and, to a minor extent, lipids and phospholipids. Oils were mostly characterized by means of triacylglycerols profiles, even if some studies focused on proteins/peptides and phospholipids were also reported. Again, proteomic studies were the most performed in matrices such as fish and meat as well as in fruit and vegetables samples, where smaller polyphenolic and carotenoid compounds were also determined. The MALDI approach has proved to be an excellent system for the determination of food fraud, allowing in many cases the development of reliable, fast, and cost-effective analytical methods, often requiring very simple sample pre-treatment protocols, and sometimes assisted by chemometrics. On these bases, also considering the noticeable number of recent papers available in the literature, a further increase of these applications is likely to be observed in the next future.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **The Effect of Edible Coatings on Selected Physicochemical Properties of Cassava Chips**

**Diofanor Acevedo-Correa 1,\*, José Jaimes-Morales <sup>2</sup> and Piedad M. Montero-Castillo <sup>1</sup>**


**Abstract:** The objective of this research was to study the effect of edible coatings on the physicochemical properties of cassava chips. The oil and moisture absorption in fried cassava chips that were not coated and in chips that were coated with pectin and whey protein films were determined using a completely randomized experiment design with a 3<sup>3</sup> factorial arrangement. The multifactorial ANOVA analysis of variance showed that all factors had significant statistical differences for moisture loss and oil absorption (*p* < 0.05). The coating type, the control, and the whey protein-coated chips presented a 321% greater oil content on average at 180 ◦C and 180 s than the pectin-coated chips. The density, heat capacity, and thermal diffusivity had statistical differences at all temperatures (*p* < 0.05). The sensory analysis showed that the coating type affected all sensory parameters, except crispness, as indicated by significant statistical differences (*p* < 0.05). The temperature only influenced the color of the control chips, with statistical differences (*p* < 0.05) at all temperatures.

**Keywords:** edible coating; cassava chips; physicochemical properties

#### **1. Introduction**

The nutritional composition of cassava is important because it is the main component of the root, which is consumed in less-developed countries. Factors associated with geographic location and environmental conditions influence the nutritional value [1]. The root contains significant amounts of carbohydrates and fiber. In addition, it has significant percentages of minerals such as calcium, iron, and phosphorus; and vitamins such as thiamine, riboflavin, niacin, and vitamin C (ascorbic acid). It also contains large amounts of amino acids such as arginine, glutamic acid, and aspartic acid [2,3]. Current cassava processing methods include peeling, smoking, drying, slicing, grating, fermenting, pressing, soaking, grinding, roasting, and frying, which are carried out to remove/reduce toxic substances, impart flavor, and increase shelf-life [4–6]. The frying method stands out. Deep-fat frying is one of the conventional and most common operations in the preparation of a variety of fried foods, which is used worldwide to create desirable flavors and textures in foods [7,8]. The advantageous sensory characteristics of most deep-fried foods derive from the formation of a composite structure that provides a crispy, porous, oily outer layer and a moist, cooked interior [9–12]. Frying is commonly used in the food industry to produce a range of food products with high consumer acceptance although high-fat contents contribute to obesity and cardiovascular disease. When food absorbs fat, it can change the composition, texture, size, and shape of the food, resulting in a loss of nutrients, specifically vitamins [13]. There is growing interest in methods that could minimize oil uptake and reduce the fat content of fried foods. Hydrocolloids have been applied indiscriminately as a food coating or incorporated into chip formulation and have shown

**Citation:** Acevedo-Correa, D.; Jaimes-Morales, J.; Montero-Castillo, P.M. The Effect of Edible Coatings on Selected Physicochemical Properties of Cassava Chips. *Appl. Sci.* **2021**, *11*, 3265. https://doi.org/10.3390/ app11073265

Academic Editor: Theodoros Varzakas

Received: 2 March 2021 Accepted: 29 March 2021 Published: 6 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

promising fat reductions and moisture retention during frying [14]. Hydrocolloids are natural compounds, such as polysaccharides and proteins, which have some hydrophilic groups. They have been used as food coatings, film-forming materials, and emulsifiers. Before frying potatoes, the application of a hydrocolloid layer on the surface has been considered an effective method to prevent oil absorption and acrylamide formation [15,16].

Edible coatings are currently used as viable alternatives for frying since these substances adhere to the product and form an external barrier that prevents the absorption of fat during immersion frying processes [17]; however, these thermal processes are currently not fully controlled for temperature and processing time, which usually results in a product with inadequate sensory characteristics and a high oil content. In general, they are artisanal and inefficient, so it is not possible to have an exact control of the process variables [18]. Research by Ajo [19] reported that the application of an edible coating with xanthan gum reduced oil absorption by up to 57% and improved the overall quality of the product, including taste, flavor, and crunchiness. One of the alternatives for processing cassava is cooking it with boiling water and then frying it by immersion in oil and coating it with edible films [20] based on proteins, such as whey, isolated soybean, some carbohydrates, pectins, and hydrocolloids that have oil barrier properties [21].

This results in excessive energy and economic expense on an industrial scale and affects the low acceptability of food by consumers since the final presentation is not traditional [22]. Therefore, it is important to investigate the effects of edible coatings in reducing the absorption of oil by products during frying in order to present consumers with a quality product with adequate processing conditions that reduce fat consumption and develop more acceptable, safe, and healthy cassava products [23]. These coatings were tested on cassava chips because they have been applied on the surface of fresh, frozen, and manufactured foods to improve food quality and increase shelf-life; in addition, they present mechanical properties and a barrier to water vapor and gases [24–26]. Furthermore, coatings were applied to decrease moisture loss values during frying since properties such as crispiness are affected, which is a fundamental parameter for consumers [27,28]. Currently, research is needed to transform this product into alternatives and new presentations that allow and expand its consumption. Therefore, this study aimed to characterize the coatings and determine the effect of coatings based on pectin and whey protein on reductions in oil absorption in fried cassava chips.

#### **2. Materials and Methods**

#### *2.1. Obtaining the Raw Material*

Cassava roots (*Manihot esculenta Crantz*), variety MCol 2215, were used, which were obtained from the supply center in the city of Cartagena. Local, refined, 100% vegetable palm oil was also used.

#### *2.2. Elaboration of the Edible Films*

#### Pectin and Whey Protein Films

The films were prepared by slowly dispersing 4, 8, and 12 g of pectin in 400 mL of distilled water. The solution was heated to 90 ◦C for 10 min with constant agitation at 120 rpm. Forty milliliter aliquots were distributed in 11.8 cm diameter plates, and the films were dried at 28 ◦C for 48 h [22]. The whey protein films were prepared by slowly dissolving 18, 22, and 26 g of whey protein along with 6.0, 6.6, and 7.8 g of glycerol (respectively) in 140 mL of distilled water. The solutions were heated in a water bath at 90 ◦C for 5 min with slow agitation at 120 rpm. 20 mL aliquots were distributed in 11.8 cm diameter plates, and the films were left to dry at 28 ◦C for 24 h.

#### *2.3. Characterization of Edible Coatings*

#### 2.3.1. Permeability to Water Vapor and Thickness

The water vapor permeability (WVP) was determined according to ASTM E-96 standard method 1995. The thickness of each film was evaluated using a digital micrometer, using the arithmetic mean of 10 measurements taken at random over the entire film surface.

#### 2.3.2. Solubility in Water

The water solubility of the films was determined using the gravimetric method of Gontard et al. [29]. The films were cut in 2 cm diameter disks. The initial weight of the samples was obtained after drying for a period of 24 h at a temperature of 105 ◦C. After the first weighing, the samples were immersed in a container containing 50 mL of distilled water and kept at 120 rpm for 24 h. After this period, the samples were removed and dried at a temperature of 105 ◦C for another 24 h to obtain the final dry weight. The tests were carried out in triplicate. The water solubility (WS) was expressed with the ratio: WS = (Initial weight − Final weight)/Initial weight × 100.

#### *2.4. Experiment and Statistical Design*

For the frying process, a completely randomized experiment design (DCA) was used with 3<sup>3</sup> factorial arrangements, where the factors with their respective levels were: Temperature (140, 160, and 180 ◦C), time (60, 120, and 180 s), and% of coating (0%, 1% pectin, and 9% whey protein). Each run was done in triplicate, and the results were expressed as the mean and the respective standard deviation. Statistically significant differences were determined in each parameter with a completely randomized analysis of variance (ANOVA) and multiple comparisons using Tukey's HSD test with a significance level of 5% (*p* ≤ 0.05). Table 1 shows the experiment treatments.


**Table 1.** Factors and levels of cassava chips in the experiment design.

The responses of the experiment design were the physical-chemical parameters of moisture loss and oil absorption (Table 1).

The mathematical model of the complete experiment design with three factors (α, *β*, γ) and the interaction is schematized in Equation (1):

$$\mathcal{Y}\_{\rm ijk} = \mu + \alpha\_i + \beta\_j + \gamma\_r + (\alpha\beta)\_{\rm ij} + (\alpha\gamma)\_{\rm ir} + (\beta\gamma)\_{\rm jr} + (\alpha\beta\gamma)\_{\rm ijk} + \varepsilon\_{\rm ijk} \tag{1}$$

#### *2.5. Frying by Immersion Process*

The cassava samples were cut in cylindrical shapes (6 cm length × 3 cm diameter) and cooked for 15 min using a Laboratory Water Bath (Memmert, Germany), controlling the temperature at 100 ◦C ± 2 ◦C. For each 100 g of cassava sample, 600 mL of potable water was used. The edible coatings (1% pectin and 9% whey protein) were applied to the cassava samples with immersion. There was a control without coatings. After the coating, the samples were dried at 25 ◦C, weighed, and fried by immersion. For the frying process, 10 pre-cooked cassava pieces, 4 × 2 × 2 cm, were used. An electric MKE fryer (Indianapolis, Georgetown Rd, USA) with a capacity of 5 L was used, equipped with a thermostat (scale 0–300 ◦C) with an accuracy of ±0.1 ◦C. The process conditions included frying temperatures of 140, 160, and 180 ◦C, with times of 60, 120, and 180 s. The chips were fried in palm oil and placed in a stainless-steel basket. They were then placed in desiccators.

#### *2.6. Thermophysical Properties*

The specific heat (Cp.), density (ρ), conductivity (k), and thermal diffusivity (α) of the samples were determined using the methodology by Choi and Okos [30]. Systematized in a computer software called DEPROTER (Determination of Thermophysical Properties) [31].

#### *2.7. Physicochemical Parameters*

The moisture (g water g−<sup>1</sup> dry solids) was determined according to official method 925.09b AOAC. 3–5 g of each sample were dried in a LT04/5 convection oven at 105 ◦C to constant weight. Then, to determine the fat content, the dried samples were weighed, and placed in a Soxhlet extraction equipment with petroleum ether for 8 h at 50 ± 2 ◦C; after the extraction, a rotary evaporator was used to evaporate the solvent from the oil. Then, the fat was obtained with gravimetry and expressed in g/100 g of dry solid according to conventional method 920.39 AOAC [32].

#### *2.8. Sensory Evaluation*

For the sensory evaluation, an untrained panel of 40 panelists was used, with an age range of 20–28 years, who were habitual consumers of cassava. They were given the fried cassava chips on a plate, in an appropriate, ventilated room with controlled temperature conditions. A five-point hedonic scale was used in which the panelists indicated their degree of acceptance for the parameters of color, odor, flavor, crispness, fat and, hardness. The scale categories ranged from 1 = I dislike it very much to 5 = I like it very much. Subsequently, the data were entered into a spreadsheet and transformed into numerical scores for analysis. Uncoated cassava chips (control) and chips coated with pectin and whey protein were tested at 140, 160, and 180 ◦C for 180 s, for a total of nine experiments.

#### **3. Results**

#### *3.1. Water Vapor Permeability, Solubility and Thickness of Edible Coatings*

The pectin films showed higher solubility and lower water vapor permeability and thickness than the whey protein films. Analyzing the 2 and 3% pectin films showed that they were 45.6% more permeable to water vapor than films produced with 1% pectin. Additionally, the pectin films were 68.9% less permeable to water vapor than the whey protein films. The 1% pectin films had the lowest water vapor permeability and thickness, as compared to the 2% and 3% films. In addition, the 9% whey protein films had the lowest permeation and thickness, as compared to the 11 and 13% films. The pectin coatings were totally soluble in water since, after 24 h of immersion, the films were completely solubilized (Table 2). Similar results were reported by Freitas et al. [22], who also demonstrated that pectin films were completely solubilized in water. Pectin is a highly hydrophilic polysaccharide, which rapidly disintegrates in water. Batista et al. [33] similarly stated that all coatings from fatty acids and pectin were completely soluble.

**Table 2.** Water permeability, solubility, and thickness of edible coatings obtained from pectin and whey protein.


Different letters in the same column indicate significant statistical differences (*p* < 0.05).

The solubility of polysaccharide coatings in water is advantageous in situations where the film is consumed with the product, causing low alterations in the sensory properties

of the food. The coatings obtained from pectin showed less permeation than with whey, which was not very resistant to water vapor, mainly because of the material's porosity, that is, coatings obtained from whey protein were more porous than those from pectin; this difference was statistically significant and could be related to the thickness of the coatings.

The lower water vapor permeability of the pectin-based coatings could be explained by the lower thickness of the coatings, as compared to the whey protein coatings [34]. Matching results were obtained by Freitas et al. [22], who used pectin and whey protein coatings. A possible explanation for the increased permeability of edible coatings made from whey in this study could be the denaturation of the proteins under the temperature conditions used for their processing, i.e., when the protein was disrupted, it was denatured; the polar amino acids were possibly exposed, which increased the absorption of water.

#### *3.2. Moisture Loss and Fat Absorption*

Table 3 shows the multifactorial ANOVA analysis of variance for cassava chips processed with different coatings, frying times, and temperatures. In general, all factors had significant statistical differences for moisture loss (*p* < 0.05). Analyzing the coating type showed a statistically significant effect (*p* < 0.05) on the cassava chips; that is, the coating type had an influence on this parameter. The same phenomenon was observed at different temperatures (*p* < 0.05). Time, on the other hand, did not present statistical differences at 120 and 180 s; that is, the additional time did not affect the increase in the loss of this parameter. Significant statistical differences (*p* < 0.05) were only observed with the shortest process time (60 s), rather than the other two times (120 and 180 s).

**Table 3.** Influence of coating type, time, and temperature on moisture loss of cassava chips.


\* Different letters for the same factor in the same column indicate statistical differences.

During the chip frying, statistical differences were observed at 60 s for the control sample at all temperatures and at 180 s for the control and whey protein-coated chips (*p* < 0.05). In the first 60 s of processing time, no statistical differences were observed at temperatures of 160 ◦C 180 ◦C. The same was observed at 120 s in the pectin-coated chips and the control chips; on the other hand, at 180 s, statistical differences were observed in the control sample and the whey protein-coated chips (*p* < 0.05). When comparing the uncoated fried samples, the pectin coating treatment decreased the moisture loss of the fried chips by 44.55% at 180 ◦C and 60 s (Figure 1). The differences in fat and moisture content between the uncoated (control) and coated fried samples were possibly due to the replacement of water by oil after the evaporation process during frying [35]. Evaporation of water by heat transfer during a frying process creates empty spaces in the food that are filled with oil, increasing the oil content of the fried food [36].

**Figure 1.** Moisture loss during frying of cassava chips at different temperatures and times: The first column indicates the control sample (C), the second column indicates that pectin (P) was used, and the third column indicates that whey protein (WP) was used. Different letters indicate statistical differences (*p* < 0.05) for the same frying time at different temperatures. −

These results can be attributed to the fact that pectins in the outer layer of the samples possibly acted as a protective barrier that closed the pores of the product and prevented water from escaping as steam during the heat treatment, thus preventing the absorption of oil. These results are very important since they can help keep products in good condition in terms of moisture and oil. It is interesting to note that the samples coated with whey protein had a 452% higher oil content than the samples coated with pectin and fried at 160 ◦C for 60 s (Figure 2), which indicates that this coating was not effective in preventing the product from absorbing oil during and after the immersion frying process.

− − − − − −

**Figure 2.** Fat absorption during cassava chip frying at different temperatures and times: The first column indicates the control sample (C), the second column indicates that pectin (P) was used, and the third column indicates that whey protein (WP) was used. Different letters indicate statistical differences (*p* < 0.05) for the same frying time at different temperatures.

In general, the analyses indicated that the cassava chip samples that were treated for a longer time in both the control and the whey protein coatings obtained higher percentages of oil and lower moisture contents (Figures 1 and 2). These results were similar to those reported by Tirado et al. [37] in the immersion frying of tilapia slices and by Alvis et al. [34] during the atmospheric frying of sweet potato slices coated with carboxymethyl cellulose (CMC), where longer processing times meant heat-treated matrices absorbed more oil.

The edible coating containing 1% pectin retained the most moisture in the fried samples. In other words, the use of an edible coating retains moisture in fried foods. The coating can form barriers that prevent moisture loss and reduce fat absorption [38].

The net formed by the pectin edible coating prevented moisture from escaping, thus retaining a higher moisture content in coated samples than in uncoated ones (control) [39].

Table 4 shows the influence of the coating type and frying time and temperature on the oil absorption of the cassava chips. All factors had an influence on the oil absorption of the chips, which was corroborated by the multifactorial ANOVA analysis of variance. For the coating type, significant statistical differences (p < 0.05) were observed for oil absorption. The same behavior occurred for frying time and temperature, observing differences between chips coated with pectin and whey protein and the control.


**Table 4.** Influence of coating type, time, and temperature on oil absorption in cassava chips.

\* Different letters for the same factor in the same column indicate statistical differences.

When analyzing the oil absorption at 60 s, no statistical differences were observed in the whey protein coated chips, but differences were observed in the pectin-coated chips at 180 ◦C and in the control at 140 ◦C (*p* < 0.05) (Table 3). At 120 s, the most representative differences were evidenced in the control at all process temperatures; this was also observed at 180 s in the pectin-coated chips, the control, and the whey proteincoated chips at different temperatures. When analyzing coating type, the control and whey protein-coated chips presented 321% more oil content on average at 180 ◦C and 180 s than the pectin-coated chips (Figure 2).

On the other hand, Freitas et al. [22] investigated the influence of the use of edible coatings from three different hydrocolloids (pectin, whey protein, and soy protein isolate) during the frying of a pre-fried and frozen cassava product. They demonstrated that whey protein showed the best results with respect to fat absorption, with a 27% reduction for the mashed cassava product. The uncoated cassava chips contained the highest oil content, 52.36%. Like the coated cassava chips with a 1% pectin concentration, the oil content of the cassava chips was reduced to 22.31%. The cassava chips treated with a 3% pectin concentration solution had the best or lowest lipid content, 11.75%, where the reduction was up to 40% for the lipid content. This possibly occurred because, during water evaporation, the vapor pressure of water increases the moisture transfer, increasing the porosity of the coating and oil ingress. In addition, high temperatures modify the structure of the material, causing damage to the coating during frying.

Also, Dragich and Krochta [40] stated that chicken strips coated with 10% denatured whey protein isolate (DWPI) resulted in a surprising 30.68% reduction in fat absorption, as compared to a non-whey protein coating. The DWPI solution possibly dehydrated during the frying process to form a film on the outer surface of the coated chicken strips. Mechanisms that are independent of or in addition to the formation of the barrier film or a possible increase in interfacial tension between chicken with a DWPI solution surface coating and the frying oil may have been responsible for the reduction in fat absorption.

Different results were obtained by Aminlari et al. [41] with the application of whey protein on potato slices, resulting in a 5% reduction in fat absorption. The rate of oil absorption can be affected by changes in interfacial tension through the accumulation of surfactants in the frying oil. The formulation may alter the water holding capacity and consequently affect oil absorption. As a result of the strong hydrogen bonding between water molecules with hydrophilic materials, the displacement of water by oil/fat during the frying process is restricted.

#### *3.3. Thermophysical Properties of the Cassava Chip Control*

In general, both the thermal conductivity and diffusivity increased with processing temperature (Table 5), which was due to the fact that the food material conducted heat better, possibly because a temperature increase resulted in dehydration of the product and therefore faster heating. The density, heat capacity, and thermal diffusivity showed statistical differences at all temperatures (*p* < 0.05), indicating that this variable had an effect on these thermal properties. The conductivity increased with increasing temperatures, but no significant statistical differences were observed, which is why temperature had no effect on this parameter [42]. This property is related to the decrease in moisture loss. Sahin et al. [43] stated that the thermal conductivity of food materials depends on porosity, structure, and chemical constituents.

#### **Table 5.** Thermophysical properties of the cassava chip control.


Letters in the same row at different temperatures for each property indicate significant statistical differences (*p* < 0.05).

Several researchers have confirmed that thermophysical properties are important parameters in the description of heat transfer during the heating of solid foods, providing great advantages for information gathering, especially for energy costs and quality assurance in different products [31,44]. It was observed that temperature significantly (*p* < 0.05) affected density, a result that may be related to moisture loss and oil absorption. It was also observed that temperature had a significant influence (*p* < 0.05) on heat capacity and thermal diffusivity. These two thermal properties increased with increasing temperatures. The results of this study were different from those reported by Sosa-Morales et al. [45] during the frying of pork meat, who stated that heat capacity and thermal diffusivity decreased with increasing frying time because of moisture loss.

#### *3.4. Sensory Analysis*

Table 6 shows that the coating type affected all sensory parameters, except crispness, which was corroborated by the significant statistical differences (*p* < 0.05). It was also evidenced that temperature only had an influence on the color of the control chips, with statistical differences (*p* < 0.05) at all temperatures.

When analyzing the color of the cassava chips, the highest scores for this parameter were observed at intermediate temperatures (160 ◦C) with pectin coatings; statistical differences were also observed (*p* < 0.05), this same result was evidenced in the odor (Table 7), which is a parameter that produces a sensation because of volatile substances. At a temperature of 140 ◦C, no statistical differences were observed in the odor of the cassava chips, and the lowest scores were obtained with the whey protein coated chips. No statistical differences in chip flavor were observed at temperatures of 140 ◦C and 180 ◦C;

however, higher scores were observed at 160 ◦C with the uncoated chips. The greasiness did not show significant differences between the three temperature groups. The crispness of the cassava chips showed significant statistical differences (*p* < 0.05) at 180 ◦C between the control and the pectin-coated chips, and both coatings had differences(*p* < 0.05). The hardness had statistical differences (*p* < 0.05) at 140 ◦C between the pectin-coated chips and the other treatments. No statistical differences were observed between the control and the whey protein coated chips. The same results were observed at 180 ◦C. At 160 ◦C, no significant statistical differences were observed (Table 7).


**Table 6.** Influence of coating type and temperature on sensory parameters of cassava chips.

Different letters in the same column for each factor indicate significant statistical differences. Control chips (C), pectin-coated chips (P) and whey protein chips (WP).

**Table 7.** Sensory analysis of control cassava chips coated with pectin and whey protein at different frying temperatures.


Different letters in the same column indicate statistical differences between the three coatings tested at each temperature.

In the case of color, higher scores (4.54) were observed at intermediate temperatures (160 ◦C) (Table 7). The coating favored browning and coloring reactions in the cassava chips. Similar results were obtained by Muhamad and Shaharuddin [46], indicating that chips coated with 2% pectin had the highest score for this parameter; the pectin coating did not change the initial color during frying of the cassava chips, and the characteristic golden color of the chips can be attributed to non-enzymatic browning of the starch. On the other hand, the best rating for odor(4.52) was in the control samples, which was possibly due to the fact that, under these conditions, the chemical components responsible for flavor were enhanced, which was also found by Aguirre et al. [47] in chips of different banana varieties.

Regardless of the type of coating used and the temperature applied, higher greasiness acceptance was observed in those samples coated with pectin, with higher scores at 180 ◦C; i.e., at higher temperatures, the oil absorption in the cassava chips was reduced. Muhamad and Shaharuddin [47] found that pectin-coated cassava chips samples absorbed less oil than uncoated chips. The substantial reduction in oil absorption can be attributed mainly to the barrier properties because coatings make the surface stronger and more brittle, with fewer small pores, which reduces evaporation and leads to less fat absorption. The crispness did not present statistical differences (*p* < 0.05) in any of the treatments.

#### **4. Conclusions**

The pectin films showed higher solubility and lower water vapor permeability and thickness than the whey protein films. Moreover, the pectin films were 68.9% less water vapor permeable than the whey protein films. All factors influenced the moisture loss and oil absorption of the cassava chips. When analyzing the coating type, the control and whey protein-coated chips presented a 321% greater oil content on average at 180 ◦C and 180 s than the pectin-coated chips. The thermal properties were affected by process temperatures, except for conductivity. The sensory analysis showed that the coating type affected all sensory parameters, except crispness. Temperature only had an effect on the color of the control chips. It would be interesting to conduct future studies using vacuum frying, with different hydrocolloids and oils, to further study the physicochemical properties of cassava chips.

**Author Contributions:** D.A.-C. methodology, formal analysis, writing—original draft preparation. J.J.-M. validation, formal analysis, writing—original draft preparation. P.M.M.-C. supervision, software, writing—original draft preparation. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data used to support this study are available from the corresponding author.

**Acknowledgments:** In this section, you can acknowledge any support given which is not covered by the author contribution or funding sections. This may include administrative and technical support, or donations in kind (e.g., materials used for experiments).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### **Application of High-Frequency Defrosting, Superheated Steam, and Quick-Freezing Treatments to Improve the Quality of Seafood Home Meal Replacement Products Consisting of the Adductor Muscle of Pen Shells and Common Squid Meat**

**Bertoka Fajar Surya Perwira Negara 1,2 , Seung Rok Kim <sup>1</sup> , Jae Hak Sohn 1,2, Jin-Soo Kim 3,\* and Jae-Suk Choi 1,2,\***


**Abstract:** We developed a new seafood home meal replacement (HMR) product containing the adductor muscle of the pen shell (AMPS) and common squid meat (CSM) via high-frequency defrosting (HFD), superheated steam, and quick freezing. Test HMR products were produced by mixing defrosted and roasted AMPS, CSM, and sauce in ratios of 27.5, 27.5, and 45.0% (*w*/*w*), respectively, followed by quick freezing at −35 ◦C in a polypropylene plastic bowl covered with a plastic film. The chemical characteristics, nutritional quality, microbial and sensory properties, and shelf life of the product were examined. The response surface methodology identified the optimal temperature and heating time of the superheated steam for AMPS (220 ◦C, 1 min) and CSM (300 ◦C, 1.5 min). Chemical characteristics showed low levels of volatile basic nitrogen (9.45 mg%) and thiobarbituric acid-reactive substances (1.13 mg Malondialdehyde [MDA]/kg). No significant changes (*p* < 0.05) were observed in microbial, color, flavor, taste, texture, and overall acceptance at −23 ◦C for 90 days. After reheating, the sensory scores varied from "like moderately" to "like very much." The shelf life of the HMR product was estimated to be 24 months. In conclusion, HFD, superheated steam, and quick freezing successfully improved product quality, with little loss of nutrition and texture.

**Keywords:** HMR; pen shell; squid meat; superheated steam; high-frequency defrosting

#### **1. Introduction**

In recent decades, people have dramatically changed the way they cook food because of the changes in attitude towards food. In the last few years, with increasing lifestyles, the number of people choosing to embrace these convenient food products at the expense of taste and health has steadily increased. Consequently, the popularity of packaged food and instant meals has revolutionized the packaged food industry, with manufacturers seeking new methods to offer fresh and delicious food without inconvenience. Nowadays, home meal replacement (HMR) products that are simple meals to prepare and consume have become increasingly popular in the food and beverage industry. HMR products have the advantage of long shelf life while maintaining the nutrient content of the food. Globally, the demand for HMR products has increased due to increasing single-person households, women's social activities, and the population of elderly people.

**Citation:** Negara, B.F.S.P.; Kim, S.R.; Sohn, J.H.; Kim, J.-S.; Choi, J.-S. Application of High-Frequency Defrosting, Superheated Steam, and Quick-Freezing Treatments to Improve the Quality of Seafood Home Meal Replacement Products Consisting of the Adductor Muscle of Pen Shells and Common Squid Meat. *Appl. Sci.* **2021**, *11*, 2926. https:// doi.org/10.3390/app11072926

Academic Editor: Theodoros Varzakas

Received: 26 February 2021 Accepted: 22 March 2021 Published: 25 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Increases in HMR products have encouraged the development of new high-quality seafood HMR products with a long shelf life; these contain the seasoned adductor muscle of the pen shell (AMPS) and common squid meat (CSM). Pen shell (*Atrina pectinata*), a bivalve and common squid (*Todarodes pacificus*), a cephalopod, are two popular and commercially important species. Globally, these organisms are distributed from southeast Africa to Malaysia and New Zealand, the Indo-western Pacific region, Japan, and Korea [1,2]. However, South Korea and Japan are the largest markets for raw and processed products of these two species [3]. *Khi-Jo-Gae* and *O-Jing-Eo* are popular names for *A. pectinate* and *T. pacificus* in Korea. Due to their high nutritional value, these species are often found on the seafood menu. For instance, the AMPS and CSM, the protein-rich, highly nutritional products of pen shell and common squid, have high consumption demand.

As the quality of seafood HMR products continues to improve, it is believed that applying advanced technologies to handle raw material, cook, and freeze during the manufacturing process is likely to result in excellent quality products [4]. To this end, several technologies, including high-frequency defrosting (HFD), superheated steam cooking, and quick freezing techniques, are believed to enhance product quality. HFD, in which the amount of heat generated inside the product and the defrosting time are accurately controlled, can reduce thawing time, inhibit microbial activities, reduce drip loss, and maintain the quality of raw materials [5,6]. Furthermore, oxidation is minimized in superheated steaming due to the lack of oxygen during heating and roasting [7], rendering superheated steam cooking an effective method in the food industry. In addition, superheated steam roasting reduces energy consumption during the roasting process [8] and has been proven to reduce lipid oxidation and preserve food nutrient substances, color, and texture better than traditional cooking methods [7,9–11]. Quick freezing is a key technology for maintaining the quality and prolonging the shelf life of frozen food. This technology successfully inhibits changes in flavor, color, and texture due to oxidative, enzymatic, and microbial changes [12–14].

Though these techniques have been explored in several food products, their applications in seafood-based HMR products are limited. Therefore, we hypothesized that the development of new high-quality fresh product-equivalent seafood HMR products, especially those containing the seasoned AMPS and CSM, could be achieved by employing advanced techniques. Moreover, the shelf life of packaged seafood products is an important factor determining the consumer acceptability and saleability of the product. Therefore, it is essential to decipher the ideal conditions that could prolong the shelf life of HMR products while improving their quality.

In this study, we aimed to elucidate the best method to produce good quality, highly nutritious HMR products with an improved shelf life prepared by mixing roasted AMPS and roasted CSM. Here, we produced the seasoning-mixed AMPS and CSM as test HMR products using a high-frequency defroster, superheated steam, and quick freezing techniques and evaluated its quality characteristics of test HMR products. The findings could unravel the potential of improved technologies, which could be used in the seafood industry to produce high-quality HMR products with little loss of nutrition and texture.

#### **2. Materials and Methods**

#### *2.1. Materials*

All chemicals used in this study, including potassium carbonate, sulfuric acid, boric acid, sodium hydroxide, trichloroacetic acid, phosphoric acid, N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide with 1% Butyldimethylchlorosilane (MTBSTFA with 1% t-BDMCS), methyl red solution, methylene blue solution, acetonitrile, 2-thiobarbituric acid, ethanol, and sodium bicarbonate, were purchased from Sigma-Aldrich, Inc. (St. Louis, MO, USA). The Difco plate count agar and EC medium used for microbiological analysis were purchased from BD Co. (Franklin Lakes, NJ, USA), whereas Sanita-Kun plates were obtained from JNC Corp. (Tokyo, Japan).

#### *2.2. Experimental Sample*

Frozen AMPS and CSM were obtained from EBADA Fishery Co. Ltd. (Busan, Korea). Sample weights were 46–60 g (average 53.7 g) for AMPS and 188–266 g (average 221.8 g) for CSM.

#### *2.3. Drip Loss Analysis*

Frozen AMPS and CSM were thawed under three different conditions: room temperature, running water, and HFD. For running water and room temperature thawing, samples were placed in plastic bags. For HFD thawing, the controller of the HFD machine (CHRFT-100, Chamco Co. Ltd., Busan, Korea) was set at 27 MHz for 10 min (with an input power of 11 kW). Drip loss was analyzed using the filter-paper wetness (FPW) method following Kauffman et al. [15]. Quantitative filter paper No. 2 (55 mm; Advantech, Tokyo, Japan) was weighed (*y*), placed on the samples during thawing, and then filter paper with absorbed fluid was weighed again (*x*). The weight difference of filter paper was expressed as the weight of the absorbed exudate. Drip loss was quantified as a percentage following this formula:

$$Drip\ loss \ (\%) = \frac{x - y}{x} \times 100$$

#### *2.4. Roasting Treatment*

Thawed AMPS and CSM were cut into approximately 3.5 × 0.5 cm and 1 × 5 cm pieces, respectively. The inedible part of the CSM was removed, and the samples were then washed with tap water. Next, all samples were roasted using a superheated steam in an Aero Steam Oven (DFC-560A-2R/L, Naomoto Co., Osaka, Japan) at different temperatures and roasting times. For AMPS, the temperatures used were 192, 200, 220, 240, and 248 ◦C for 0.53, 0.67, 1.00, 1.33, and 1.47 min, respectively, while for CSM, the temperatures used were 272, 280, 300, 320, and 328 ◦C for 0.79, 1.00, 1.50, 2.00, and 2.21 min, respectively.

#### *2.5. Sample HMR Product Preparation*

First, the stock of sauce was prepared by mixing the ingredients shown in Table 1. A test seafood HMR product was then produced by mixing roasted AMPS, roasted CSM, and sauce at ratios (*w*/*w*) of 27.5, 27.5, and 45.0%, respectively. Afterward, 180 g of product was packaged in a polypropylene plastic bowl (New Ecopack Co. Ltd., Jeonju, Korea) and sealed with a plastic film using a tray sealing machine (TPS-TS3T, TPS Co. Ltd., Kyungkido, Korea) at 180 ◦C for 5 s. Next, packaged products were frozen using a quick freezer (QF-700, Alpha Tech Co. Ltd., Incheon, Korea) at −35 ◦C for 10 min, and then stored at −13, −18, and −23 ◦C in a deep freezer (DF35035, IlShin BioBase Co. Ltd., Dongducheon, Korea) for shelf life estimation.

**Table 1.** Ingredients used to prepare the sauce (stock), a key ingredient of the new home meal replacement (HMR) product.


#### *2.6. pH Measurement*

Two grams of product were added to 3rd-distilled water at a ratio of 1:9 (*w*/*v*), and then homogenized using a homogenizer (SHG-15D, SciLab Co. Ltd., Seoul, Korea). Next, the pH of each sample was measured using a pH meter (ST 3100, Ohaus Co., Parsippany, NJ, USA).

#### *2.7. Measurement of Volatile Basic Nitrogen (VBN)*

VBN was performed using the Conway micro diffusion method. Five grams of product were diluted with 25 mL of 3rd-distilled water and homogenized using a vortex for 5 min. Filter paper No. 2 (55 mm; Advantech, Tokyo, Japan) was used to filter the mixture sample. Following filtration, potassium carbonate was added to the filtrate solution at a ratio of 1:1 (*v*/*v*) in the outer chamber, while 1 mL of 0.01 M H2SO<sup>4</sup> was added to the inner chamber of the Conway unit. Conway cells were incubated at 37 ◦C for 90 min. Brunswick reagent (2–3 drops) was added to the inner chamber and titrated with 0.01 N NaOH.

#### *2.8. Measurement of Thiobarbituric Acid-Reactive Substances (TBARS)*

TBARS were measured following the method by Peiretti et al. [16] with modifications. Briefly, 5 g of each sample was homogenized (SHG-15D, SciLab Co. Ltd., Seoul, Korea) in 12.5 mL of TCA solution containing 20% trichloroacetic acid in 2 M phosphoric acid and adjusted to 25 mL with 3rd-distilled water. Next, homogenate samples were centrifuged at 1500 rpm for 10 min, and the upper layer was collected and filtered using filter paper No. 2 (55 mm; Advantech, Tokyo, Japan). The supernatant was then mixed with a 0.005 M thiobarbituric acid solution at a ratio of 1:1 (*v*/*v*) and incubated at 95 ◦C for 30 min in a water bath (JSWB-22TL, JS Research Inc., Gongju City, Korea). The samples were then left to cool to room temperature. Samples (200 µL) and the blank group (3rd-distilled water) were placed on a 96-well plate and measured at 530 nm using a SPECTROstar Nano Microplate Reader (S/N601-0618, BMG Labtech Ltd., Ortenberg, Germany). Malondialdehyde (MDA) bis (dimethyl acetal) was used as the standard.

#### *2.9. Proximate Analysis*

Proximate analysis was performed using AOAC standard methods. Moisture was measured following the methods AOAC 952.08 [17] using an oven at 105 ◦C for 24 h. Ash was determined following the AOAC 938.08 method [18] in a furnace at 550 ◦C. Sodium was measured following AOAC 971.27 method [19]. Crude protein was determined following AOAC 960.48 method [20]. N content from the test product was then multiplied with 6.26 to obtain the value of crude protein. Calories, carbohydrates, sugars, dietary fiber, crude fat content, cholesterol, vitamin D, potassium iron, and calcium were measured following AOAC 971.10 [21], AOAC 998.18 [22], AOAC 985.29 [23], AOAC 948.15 [24], AOAC 994.10 [25], AOAC 936.14 [26], AOAC 2011.11 [27], AOAC 990.05 [28], and AOAC 984.27 [29], respectively.

#### *2.10. Amino Acid Analysis*

Prior to hydrolysis, the protein was extracted following the Kjeldahl method, and the amino acid analysis was performed using the AOAC 994.12b method [30]. Briefly, a 50 µL aliquot of a solution containing a mixture of 91 µg/mL L-amino acids in 0.1 N HCl was dried. Subsequently, 100 µL of neat MTBSTFA, followed by 100 µL of acetonitrile, were added. The mixture was then heated to 100 ◦C for 4 h. Next, the sample was neutralized with sodium bicarbonate and subjected to gas chromatography-mass spectrometry (GC-MS) analysis using the GCMS-QP2020 system (Shimadzu Corp., Kyoto, Japan). An injection volume of 0.5 µL/min was used for the separations performed on an SLB™-5ms Capillary GC Column (20 m × 0.18 mm I.D., 0.18 µm; Sigma-Aldrich, Inc., St. Louis, MO, USA) using helium as the carrier gas. During the separation, the oven temperature was programmed as follows—initially, it was set at 100 ◦C for 1 min, then increased to 290 ◦C at 35 ◦C/min and held for 3 min, and finally, it was raised to 360 ◦C at a rate of 40 ◦C/min and held for

2 min. The temperature for the inlet was set at 250 ◦C, while the temperature for the mass storage device (MSD) interface was set at 325 ◦C.

#### *2.11. Fatty Acid Composition Analysis*

The fatty acid composition analysis was performed following a hydrolytic method described by Sutikno et al. [11]. Briefly, ether was used to extract fatty acids and methylate them into fatty acid methyl esters (FAMEs). FAMEs were then analyzed using gas chromatography (GCMS-QP2020) fitted with a DB-wax capillary column (30 m × 0.25 mm i.d., 0.25 mm film thickness, Agilent). Helium at a constant linear velocity of 30 cm/s was used as the carrier gas. The split ratio was set at 1/10. During the separation, the column oven was programmed as follows: initial column oven temperature was set at 100 ◦C; held for 1 min, and increased to at 25 ◦C/min to 100 ◦C and after 1 min, and finally, it was raised to 240 ◦C at 5 ◦C/min; held for 2 min. The temperature for the inlet was set at 250 ◦C, while the temperature for Flame Ionization Detector (FID) was set at 270 ◦C. FAME standard mixture (EN 14078, Paragon Scientific Ltd., Wirral, UK) was used to identify the peak and calculate the response factor. The results were expressed as g/100 g of dry matter.

#### *2.12. Total Bacterial Count (TBC) and Total Coliform Count*

TBC was determined according to the instructions of Chen et al. [31]. The sample was mixed with sterile saline at a ratio of 1:9 (*w*/*v*) in sterilized bags, and homogenized using a Stomacher 400 Circulator (Seward Ltd., West Sussex, UK) for 3 min. Three serial dilutions of an aliquot of the homogenate were plated onto a specific medium. For TBC, Difco plate count agar (BD Co., Franklin Lakes, NJ, USA) was used and incubated at 37 ± 1 ◦C for 48 h. For the coliform count, EC medium (BD Co., Franklin Lakes, NJ, USA) was used and incubated at 37 ± 1 ◦C for 24 h; if no gas was observed, results were recorded as negative (-). For *Salmonella* sp. and *Staphylococcus* sp. counts, Sanita-Kun plates (JNC Corp., Tokyo, Japan) were used and incubated at 35 ± 1 ◦C for 48 h.

#### *2.13. Texture Analysis*

A texture analyzer (CT3 4500, Brookfield Engineering Laboratories Inc., Middleboro, MA, USA) operated by TexturePRO CT software (Middleboro, MA, USA) was used to measure hardness. The texture analysis was performed by compressing the sample to 50% of its height using a cylinder glass probe (12.7 mm in diameter and 35 mm in length). The test speed was 0.5 mm/s. The textural analysis was performed at room temperature with triplicate measurements of each sample.

#### *2.14. Sensory Evaluation*

Sensory analysis, including color, flavor, aroma, texture, and overall acceptability, were evaluated. Test seafood HMR was reheated using a microwave (RE-M50, Samsung Electronics Co. Ltd., Seoul, Korea) for 1 min 30 s (700 W). Twenty-one trained and certified sensory evaluation panelists (aged 25–40 years) were employed in this test—all of them were researchers at the Industry Academic Cooperation Foundation, Silla University (Busan, Korea). The sensory test was approved by the Silla University Institutional Review Board (Approval No. 1041449-202006-HR-007; 4 October 2020). Each panelist was asked to evaluate and numerically rate all samples. A hedonic scale of 1 to 9 points was used—1 indicating 'remarkably dislike' and 9 indicating 'extreme like.' Five was considered the threshold value; any sample less than a score of 5 was considered unacceptable [32].

#### *2.15. Shelf Life Estimation*

The expiration date was established following the Ministry of Food and Drug Safety guidelines, the Republic of Korea. The accelerated experiment to define the expiration date and evaluate the shelf life qualities of the test-HMR products was conducted for 90 days. Test HMRs were stored at −18 ◦C (distribution temperature) as a control, −13 ◦C, and −23 ◦C. Samples were collected seven times over 90 days, and microbiological and physical

tests were performed. The number of common bacteria, such as *E. coli, Staphylococcus aureus*, and *Salmonella* spp., and the sensory evaluation scores were used as indicators of quality shelf life. Program simulation (https://www.foodsafetykorea.go.kr, accessed on 17 December 2020) was used to estimate the product's shelf life.

#### *2.16. Statistical Analysis*

All experiments were performed in triplicate (n = 3); the data are displayed as mean ± standard deviation (SD). Drip loss, TBC, and sensory properties were analyzed via a one-way analysis of variance (ANOVA) at a 95% level of probability (*p* < 0.05) using IBM SPSS v 23.0 (IBM, Corp., Armonk, NY, USA) software. Response surface methodology (RSM) was analyzed using Minitab v 14.0 (Minitab Inc., Birmingham, UK). Temperature and heating time were set as independent variables, while overall acceptance and hardness were set as dependent variables.

#### **3. Results and Discussion**

#### *3.1. Drip Loss*

Drip loss was analyzed via the exudate absorbed into the quantitative filter paper during the thawing process under three different conditions: room temperature, running water, and HFD. Table 2 shows the drip loss results of frozen AMPS and CSM. HFD resulted in the lowest drip loss value of both AMPS and CSM. The drip loss values of AMPS from the high-frequency defroster were significantly different from those thawed using the conventional methods (*p* < 0.05). The highest drip loss value was observed by thawing at room temperature, followed by running water.

**Table 2.** Drip loss analysis of frozen adductor muscle of the pen shell (AMPS) and common squid meat (CSM) under different conditions.


Data are presented as mean ± SD. Different superscript letters (a–c) show significantly different values according to Duncan's test (*p* < 0.05).

The thawing process for frozen fish and fishery products should be performed as quickly as possible. Water changing from its original position during the thawing process leads to drip loss, resulting in a dry, stringy, and less tasty fish. Nutrient losses, such as proteins, vitamins, and minerals, can occur with drip loss [33], with high drip loss often linked to protein denaturation. In addition, high drip loss also decreases attractiveness, nutritional value, texture, and appearance [34].

Applying technology during the thawing process may improve the quality of fishery products, also likely decreasing thawing time [35]. Furthermore, rapid thawing can maintain fish quality [36] and reduce mechanical damage to cell membranes by decreasing recrystallization [37]. Recrystallization, which can be problematic during both thawing and freezing, leads to cellular damage, increasing drip production [38].

Therefore, HFD is considered an effective option to decrease drip loss and maintain the quality of raw materials in the test HMR product. The results of the present study confirm the effectiveness of HDF compared to conventional methods.

#### *3.2. Optimum Conditions for Roasting*

The optimum conditions for roasting were analyzed using RSM. Temperature (X1) and heating time (X2) were set as independent variables, while overall acceptance (Y1) and hardness (Y2) were set as dependent variables. The model in this study was build using the actual value of independent variables. The model obtained for overall acceptance and

hardness of AMPS was Y<sup>1</sup> = 7.8097 + 0.9016X<sup>2</sup> − 0.9525X<sup>1</sup> <sup>2</sup> <sup>−</sup> 1.4762X<sup>2</sup> <sup>2</sup> with 92.1% of R<sup>2</sup> , and Y<sup>2</sup> = 471.5 + 63.588X<sup>1</sup> + 28.777X<sup>2</sup> + 79.759X<sup>1</sup> <sup>2</sup> + 46.134X<sup>2</sup> <sup>2</sup> with 95% of R<sup>2</sup> , respectively (Table 3). Three-dimensional response surface plots of AMPS showed an increase in the score of overall acceptance and decrease in hardness value with an increase in time and temperature until an optimum condition was attained (Figure 1a,b). The overall acceptance score started to decline beyond 217.8 ◦C and 1.08 min (Figure 1a), whereas the hardness value tended to increase by increasing the temperature and time beyond 217.8 ◦C and 1.08 min (Figure 1b). Therefore, the optimum temperature and time for the superheated steam roasting of AMPS were set at 217.8 ◦C for 1.08 min (Table 4).


**Table 3.** Response surface model equations of AMPS and CSM.

Y1: overall acceptance, and Y2: hardness are the dependent variables; X1: Temperature (◦C); X2: heating time (min) are the independent variables.


**Table 4.** Optimal cooking conditions of AMPS and CSM using response surface methodology (RSM)-analyzed superheated steam roasting.

The CSM response surface model is displayed in Table 3. The model obtained for overall acceptance and hardness was Y<sup>1</sup> = 8.07933 − 0.54590X<sup>2</sup> − 0.76879X<sup>1</sup> <sup>2</sup> <sup>−</sup> 1.54254X<sup>2</sup> 2 with 91.6% of R<sup>2</sup> , and Y<sup>2</sup> = 508.67 + 63.44X<sup>1</sup> + 121.91X<sup>2</sup> − 49.13X<sup>1</sup> <sup>2</sup> + 47.90X<sup>2</sup> <sup>2</sup> with 96.5% R 2 , respectively. Three-dimensional response surface plots of CSM for overall score and hardness (Figure 1c,d) showed the same pattern as those for AMPS. The overall acceptance score increased with the increasing time and temperature until the optimum condition was attained, while a reverse trend was observed for hardness. Increasing the temperature and time beyond 296.07 ◦C and 1.53 min decreased the overall acceptance scores (Figure 1c), while a temperature and time beyond 296.07 ◦C and 1.53 min increased the hardness (Figure 1d). The optimum temperature and time for CSM superheated steam roasting was 296.07 ◦C for 1.53 min (Table 4).

**Figure 1.** Three-dimensional response surface plots of overall acceptance (**a**) and hardness (**b**) of AMPS and overall acceptance (**c**) and hardness (**d**) of CSM in a superheated steam roaster using RSM.

− The optimum temperature and time of AMPS and CSM were 217.8 ◦C for 1.08 min and 296.1 ◦C for 1.53 min, respectively. These values were in the range −0.1964 to 0.2435 (Table 4). Moreover, centeral composite design results (Table 5) showed that the optimum practical temperature and time values for superheated steam roasting was 220 ◦C for 1 min for AMPS and 300 ◦C for 1.5 min for CSM. In this study, the roasting time for AMPS and CSM was faster than that reported by Mohibbullah et al. [10] and Sutikno et al. [11]. According to Mohibbullah et al. [10], the superheated steam roasting time of AMPS was 4 min at 270 ◦C to achieve a good panelist-determined sensory evaluation score, which included color, odor, texture, flavor, and overall enjoyment. Moreover, Sutikno et al. [11] proposed the optimum superheated steam roasting time for squid meat was 10 min at 240 ◦C, as evaluated by panelists based on sensory characteristics.

**Table 5.** Symbol, experimental range and values of the independent variables in the central composite design.


The optimum conditions for roasting both AMPS and CSM indicated the best combination of independent and dependent variables. Furthermore, RSM has been effectively used to improve food products [39,40]. Here, we showed RSM successfully optimized superheated steam conditions for efficient production of AMPS and CSM. Moreover, applying RSM to AMPS and CSM roasting resulted in high sensory evaluation values. Taken together, these results indicate the potential of RSM to optimize or choose operating conditions to achieve a particular set of objectives.

#### *3.3. Chemical Characteristics*

A test HMR product was prepared by thawing frozen AMPS and CSM using HFD, followed by superheated steam roasting at a particular value for both AMPS and CSM. The test product consisted of 27.5% AMPS, 27.5% CSM, and 45% sauce. Chemical properties, such as pH, VBN, and TBARS, were used to evaluate the quality of the product (Table 6). The pH, VBN, and TBARS of the test HMR were 6.30 ± 0.13, 9.45 ± 1.09 mg%, and 1.13 ± 0.27 mg MDA/kg, respectively. Consistent with the results of Sutikno et al. [11] and Mohibbullah et al. [10], who reported near-neutral pH in seafood products (squid meat and AMPS) following superheated steam roasting, the pH of the test HMR was close to neutral (7). Moreover, for fresh fish and fishery products, an ideal pH level is almost neutral. However, post-mortem period-associated nitrogenous compounds increase the pH, leading to decreased quality [41].

**Table 6.** Chemical characteristics of test seafood HMR products consisting of a mixture of AMPS, CSM, and sauce.


Data are presented as mean ± SD.

A product's VBN value indicates the amount of nitrogen derived from proteolytic bacteria and endogenous enzymes [29], with the VBN test examining the level of protein breakdown and the presence of non-protein nitrogenous compounds. Furthermore, unpleasant smells generally correlate with high VBN values. Superheated steam roasting has been observed to lead to low VBN values. Similar results were reported by Sutikno et al. [11], who found that the VBN of roasted squid meat was lower when using superheated steam roasting compared to other methods. Furthermore, Mohibbullah et al. [10] also reported a lower VBN for superheated steam roasted AMPS.

The low TBARS value of a product labels it as a perfect product according to Yildiz [41], who categorized seafood products into the following three TBARS value-based groups: perfect product (less than 3 mg MDA/kg), good product (3–5 mg MDA/kg), and consumable limit (7–8 mg MDA/kg). Using superheated steam roasting resulted in a low TBARS in squid meat [11] and AMPS [10]. This could be attributed to the reduced oxygen availability during this process resulting in higher inhibition of lipid oxidation and decreased product peroxide (POV) and TBARS values [42].

Collectively, the results of chemical characteristics of the test HMR indicated the quality of the product. According to VBN and TBARS values, test seafood HMR products can be classified as very good.

#### *3.4. Nutritional Quality*

Nutritional quality results are shown in Table 7. Proximate composition results, including moisture, crude protein, and ash, were 57.21, 11.6, and 1.62 g/100 g, respectively. Furthermore, mineral contents, such as sodium, calcium, and potassium, were 1.04, 0.19, and 0.001 g/100 g, respectively. Other parameters of nutritional quality, including calories, salt, carbohydrates, sugars, and crude fat, were 167.2 kcal/100 g, 1.03 g/100 g, 28.4 g/100 g, 14.9 g/100 g, and 0.8 g/100 g, respectively.


**Table 7.** Nutritional values of test seafood HMR products consisting of a mixture of AMPS, CSM, and sauce.

The nutritional quality of test HMR products indicated that they contained essential nutrition for humans. Wholesome food consists of macronutrients, such as protein, carbohydrates, and fat, thereby offering calories to fuel the body and energy for specific health-maintaining roles. The human body requires nutrition from food, including proteins, carbohydrates, minerals, and crude fat, to participate in daily activities and maintain good health [43]. This study suggests that the product has the essential nutrients required for healthy human body functioning, also contributing to daily nutrition needs.

#### *3.5. Amino Acid Composition*

The amino acid composition of the HMR product, consisting of mixed AMPS, CSM, and sauce, is presented in Table 8. The total amino acid content was 7.93 g/100 g, with glutamine as the predominant amino acid. Glutamine and glutamate are abundant amino acids in fish and fishery products [44,45]. In the human body, glutamine is an essential source of energy for the immune system [46].

Non-essential amino acids were dominant (53.84%) compared with essential amino acids (46.16%). Glutamine, aspartate, and alanine were the most abundant non-essential amino acids found, whereas lysine and arginine were the predominant essential amino acids (both at 8.20%). Both essential and non-essential amino acids benefit human health, such as lowering the risk of cardiovascular disease and enhancing the immune system [47]. Furthermore, following digestion, amino acids provide various benefits to human health, including repairing tissue, supporting growth, breaking down food, and providing energy [48].

This study indicates that test seafood HMR products contain amino acids essential for the human body. In addition, the product's amino acid content contributes to the overall daily amino acid requirement.


**Table 8.** Amino acid analysis of test seafood HMR products consisting of a mixture of AMPS, CSM, and sauce.

NE: Non-essential; E: Essential.

#### *3.6. Fatty Acid Composition*

Fatty acid composition results are presented in Table 9. Total saturated fatty acids (SFAs) showed the highest value (0.71 g/100 g), followed by polyunsaturated fatty acids (PUFAs) (0.26 g/100 g), and total monounsaturated fatty acids (MUFAs) (0.18 g/100 g). The three predominant fatty acids found in the sample were palmitic acid, docosahexaenoic acid (DHA), and oleic acid. Due to their nutritional value, fatty acids have important health benefits. Palmitic acid, the most common fatty acid in meat, fish, and fishery products, accounted for approximately 50–60% of total fats [49]. Furthermore, DHA is a known agent in treating primary and secondary heart disease and neurological and neuropsychiatric disorders [50]. The Food and Agriculture Organization (FAO) of the United Nations and the World Health Organization (WHO) have given special consideration to fatty acids in food as essential nutrients, and their impact on early growth, development, and nutrition-related chronic diseases in humans.

In processed seafood products, amino acids contribute to the taste of the product [51]. A sweet taste is caused by glycine, alanine, and trimethyl, whereas a bitter taste results from arginine [52]. The fatty acid content in the test seafood HMR product indicated its good nutritional value, thereby suggesting its potential to contribute to daily fatty acid requirements.


**Table 9.** Fatty acid contents of test seafood HMR products consisting of a mixture of AMPS, CSM, and sauce.

SFA: saturated fatty acids; MUFA: monounsaturated fatty acids; PUFA: polyunsaturated fatty acids.

#### *3.7. Effect of Storage Duration on Microbiological Changes*

Microbiological changes in food can indicate product safety. Table 10 shows the microbial change during storage duration. The score of TBC increased in conjunction with the storage time. At −13 ◦C, bacterial growth increased significantly (*p* < 0.05) on day 90 (3.20 ± 0.00), while at −18 ◦C, bacterial growth increased significantly (*p* < 0.05) after days 75 and 90 compared to day 0. However, at −23 ◦C, bacterial growth was maintained throughout the 90-day storage period. The optimum temperature in which to store products based on the TBC was less than −23 ◦C. In this study, the TBC of the product was less than 5 log CFU/g, which is acceptable for consumption (International Commission of Microbiological Specification for Food [ICMSF]) [53]. Overall, the findings suggest that low TBC values are influenced by the storage temperature—low temperatures, especially freezing, can decrease microbial activity.

The results of this study align with those of Alizadeh et al. [37], who reported that microbial and enzymatic activities successfully decrease in salmon fillets stored at freezing conditions, therefore maintaining a better nutritional content than those in chilled storage. Furthermore, Pohoro and Ranghoo-Sanmukhiya [54] reported that the TBC of frozen fish sold in Mauritius was lower during storage. Similar results were reported by Gornik et al. [55], who reported that low temperatures during storage result in low TBC values in lobster (Nephrops norvegicus).


**Table 10.** TBC and microbiology of test seafood HMR products, consisting of a mixture of AMPS, CSM, and sauce, stored at different temperatures.

TBC: total bacteria count; Data are presented as mean ± SD. Means in each column at each temperature group with different letters (a–c) differed significantly according to Duncan's test (*p* < 0.05).

In this study, pathogenic bacteria, such as *S. aureus*, Salmonella spp., and *E. coli*, were not detected during storage at all temperatures. These results suggest that temperature plays an important role in the growth of pathogenic bacteria in fishery products. At low temperatures, microbial activity is suppressed, and most bacteria cannot grow [56,57]; hence, the growth of many spoilage-causing bacteria is prevented via temperature reduction.

#### *3.8. Sensory Evaluation of the Product*

Five sensory parameters, including color, flavor, odor, texture, and overall acceptance, were used to evaluate changes in sensory properties following storage and reheating using a microwave. On day 0, sensory evaluation scores of the test HMR products ranged from 8.24 to 8.38, with the highest value for flavor and odor. However, during storage, the scores were decreased. When stored at −13 ◦C and −18 ◦C, the sensory evaluation score started to decrease significantly (*p* < 0.05) on day 45 and day 90, respectively, compared to day 0. In contrast, storage at −23 ◦C did not significantly alter the sensory evaluation scores throughout the 90-day storage period. Sensory evaluation scores ranged from 7.14 to 8.62 (Table 11). According to the hedonic scale, the test HMR was scored by a panelist from "like moderately" to "like very much." These results indicate that these products maintained favorable sensory properties following storage and reheating, likely due to rapid freezing. The quality deterioration of frozen food may be promoted by three factors: ice crystal formation, dehydration of protein molecules, and an increase in solid concentration during the freezing process [58].


**Table 11.** Sensory evaluation of test seafood HMR products, consisting of a mixture of AMPS, CSM, and sauce, stored at different temperatures.

Data are presented as mean <sup>±</sup> SD. Means in each column at each temperature group with different letters (a–c) differed significantly according to Duncan's test (*p* < 0.05).

> Ice crystal formation during freezing alters the physical properties of muscle tissues, thus distorting the structure of the meat [12,59–61]. Furthermore, ice crystal formation in frozen seafood reportedly decreases the sensory properties of the products [62,63]. Interestingly, the size of ice crystals is dependent on the freezing rate. Quick freezing leads to a greater number of fine ice crystal formations within muscle cells, whereas slow freezing leads to large ice crystal formations outside of muscle cells [12,37,64]. As such, the sensory properties of the products were largely preserved after reheating. Therefore, applying quick freezing to test seafood HMR products successfully maintained their sensory properties during storage or after reheating using microwaves.

#### *3.9. Shelf Life Estimation*

Changes in sensory and microbial parameters indicate product quality, potentially affecting the shelf life of fish products [65]. On day 0, the sensory characteristics score was above 8.24 for all parameters. During storage at −13 and −20 ◦C, the product began to lose its sensory characteristics; however, panelists still scored products with decreased sensory characteristics above 7. The product stored at −23 ◦C showed no significant (*p* < 0.05) differences throughout the 90-day storage period. At −23 ◦C, test seafood HMR products maintained their sensory characteristics, obtaining a score above 8 from the panelists.

The physicochemical properties of fish and fishery products correlate with their sensory properties and storage time [66]. Moreover, the microbial activity gives flavors and odors to products during the storage period [67], and the decreases in sensory properties could be caused by changes in microbial parameters (Table 10). The TBC of the HMR product at −18 and −13 ◦C significantly (*p* < 0.05) increased during storage time; however, at −23 ◦C, there was no significant change (*p* < 0.05). These results suggest that lower temperatures decrease microbial activity in the product, thereby increasing shelf life.

Moreover, the microbial quality of seafood can be classified into the following three TBC value-based groups: satisfactory (TBC < 5 log CFU/g), acceptable (≥5 TBC < 6 log CFU/g), and unsatisfactory (TBC ≥ 6 log CFU/g) [68]. Therefore, based on the TBC results of this study, the test seafood HMR product could be classified as satisfactory.

The test seafood HMR product's expiry date and overall acceptance values were established using TBC and the simulator program. It identified the shelf life of test seafood HMR products was 29.62 months when the sensory evaluation, as the main criteria, had the highest statistical value. However, the final shelf life was set to 24 months by multiplying it with the safety factor (0.82), which considers temperature changes during production, purchase, storage, and consumption by the consumer; hence, the quality of HMR products could be maintained for 24 months. The shelf life of fishery products is influenced by temperature and limited by biochemical and microbiological changes. Low temperatures may reduce microbial activity, enzymatic autolysis, and oxidation in fishery products, thereby prolonging shelf life [69–72]. Furthermore, although frozen storage shelf life has constraints, it may vary from a few weeks to years [73].

#### **4. Conclusions**

This study revealed that the technological improvements produced high-quality test HMR products by mixing AMPS and CSM. HFD maintained the nutritional quality of the raw materials with little drip loss. Optimization of superheated steam roasting using RSM successfully roasted AMPS and CSM at the optimum temperature and time, resulting in good overall acceptance and hardness. Moreover, quick freezing prevented nutrition and texture loss during storage and after reheating. Overall, the application of HFD, superheated steam, and quick freezing successfully produced test HMR products with high nutrition, good texture, and long shelf life, suggesting the potential of the methods reported here for use in the seafood industry to produce new HMR products.

**Author Contributions:** Conceptualization, J.-S.C.; methodology, J.-S.C., J.-S.K.; software, S.R.K.; formal analysis, J.H.S.; writing—original draft preparation, B.F.S.P.N.; writing—review and editing, J.-S.C.; visualization, J.-S.K.; supervision, J.-S.C., J.-S.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was funded by the Ministry of Oceans and Fisheries, the Republic of Korea, under project no. PJT201277.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data supporting reported results are available upon request.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Improved Hot Smoke Processing of Chub Mackerel (***Scomber japonicus***) Promotes Sensorial, Physicochemical and Microbiological Characteristics**

**Md. Masud Rana 1,†, Md. Mohibbullah 1,† , Na Eun Won <sup>2</sup> , Md. Abdul Baten <sup>1</sup> , Jae Hak Sohn <sup>2</sup> , Jin-Soo Kim 3,\* and Jae-Suk Choi 2,4,\***


**Abstract:** Chub mackerel (CM), *Scomber japonicus,* is a commercially important fish species in pacific countries including South Korea and its rapid quality deterioration by various spoilage mechanisms while marketed has been reported, leading to a dramatic decline of the market price. To overcome this problem, a combination of superheated steam roasting (270 ◦C for 4 min) and hot smoking (70 ◦C) on CM fillets was applied to impart extending shelf-life at the market level. Using different sawdust with time-dependent smoking revealed that Oak sawdust at 25 min of optimized smoking time significantly (*p* < 0.05) provided the highest sensory properties (appearance, odor, color, texture and overall preferences), improved physicochemical, microbial, and nutritional properties, and subsequently, promoted shelf life of processed CM during the storage period at 10 ◦C for up to 34 days. Moreover, the processed CM offered high nutritional value, especially, essential and nonessential amino acids were found to be 13.14 and 15.48 g/100 g of CM fillets, and also reduced the trimethylamine-N-oxide level to an acceptable limit, indicating its quality and safety with high nutritional standards to end-point users upon consumption.

**Keywords:** chub mackerel; smoking treatment; sensory analysis; physiochemical characteristics; microbiological quality; biochemical analysis

### **1. Introduction**

Worldwide, fish and fish-based products are exceptionally important elements of the human diet in respect to their unique taste and nutritional properties such as plethora of proteins and polyunsaturated fatty acids (PUFAs), which make them more precious than other terrestrial foods [1]. Based on the demand of healthy fish-food items, fish processing industries have been developed in the recent past to offer a variety of consumer products in which quality and safety are being prioritized [2]. The reason is that fish flesh is extremely vulnerable to spoilage [3]. The spoilage mechanisms in fish are attributed to the presence of a high amount of free amino acid, non-protein nitrogenous compounds (NPN) and a high post mortem pH, which eventually support the growth of microorganisms [4]. In addition, endogenous enzymatic activity, oxidation of lipids and proteins are responsible for the quality deterioration of fish and fish products [5,6]. Collectively, the following spoilage

**Citation:** Rana, M.M.; Mohibbullah, M.; Won, N.E.; Baten, M.A.; Sohn, J.H.; Kim, J.-S.; Choi, J.-S. Improved Hot Smoke Processing of Chub Mackerel (*Scomber japonicus*) Promotes Sensorial, Physicochemical and Microbiological Characteristics. *Appl. Sci.* **2021**, *11*, 2629. https:// doi.org/10.3390/app11062629

Academic Editor: Theodoros Varzakas

Received: 14 February 2021 Accepted: 12 March 2021 Published: 16 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

mechanisms result in the release of various metabolites that negatively affect the sensorial attributes of the product, and simultaneously, shorten the shelf-life of products, especially in storage conditions.

Chub Mackerel (CM), *Scomber japonicus,* is a migratory coastal-pelagic fish species, distributed from subtropical to temperate zones of the world ocean [7]. CM has been considered as an oily fish and one of the important commercial fish species in western North Pacific [8], and it is estimated that about 10–25% of the total annual marine fish catch are from CM in Korea [9]. CM is generally marketed as a raw, dressed or processed condition like frozen fillet, and salted fish in the retail markets of South Korea [10]. Due to the presence of a rich source of PUFAs, despite applying various processing techniques, CM deteriorates quickly especially in tropics, so many quality and safety issues have arisen on CM products while marketed or during the storage period.

Smoking is one of the traditional and common methods of preserving fish [11], which is the combination of salting, drying and cooking processes [12]. Smoked products have an attractive and unique color and flavor, which is because of phenolic compounds and other volatile components released from plant-derived sawdust materials [13]. Additionally, superheated steam technology was employed to dry CM fillets before being processed for hot smoking. Since it is colorless and transparent hot-dry gas, the result of dried products has been evident in the preparation of foods with good textural and chemical properties [14]. When a press-cake of mackerel was prepared with superheated steam, it showed low moisture and high fatty acids contents in a previous study [15]. In the present study, we hypothesized that the combination of superheated steam and hot smoking on CM could provide the final products of delicacy in taste by adding attractive and favorable color, flavor and textural attributes.

The hot smoke (70 ◦C) processed products are tastier and have longer shelf lives [16], and such treatment improves physicochemical, sensorial and microbiological qualities in processed fish fillet [17], due to the presence of many beneficial complex compounds released from incomplete combustion of wood materials such as phenols, esters, ethers, alcohols, and ketones. Despite their benefits, hot smoked processed products can lead to excessive shrinkage, buckle or even splitting, which undergo decreased yield and loss of nutrients due to denaturation of dietary proteins of fish [18]. So, the optimization of time and temperature in smoking process technology is extremely important to ensure that the products are of premium quality during storage. The physical (i.e., textural profile), chemical (i.e., pH, VBN, TBARS, and TMA), sensorial (i.e., organoleptic attributes), nutritional (i.e., proximate composition and fatty acids) and microbiological (i.e., TBC) characteristics have been assessed by the precedent research, where new or modified processed technology was applied to improve fish-based products [19,20]. However, a very limited number of studies have been investigated on CM, *Scomber japonicus*, or CM-based products. Considering the delicacy in taste and higher nutritional value, CM was selected for the current study aiming at the development of improved process technology of CM fillets with the combination of superheated steam roasting and hot smoking, for enhancing qualities of the end-product based on sensorial, physicochemical, microbiological, and nutritional aspects.

#### **2. Materials and Methods**

#### *2.1. Sample Collection and Processing*

The CM fish was bought from the local fish market, South Korea, stored with ice flakes, rinsed in the laboratory, and longitudinally cut to form a fish fillet. The size of the fillet section was 37.4 × 7.8 × 1.5 cm and the average weight of the sample was 355.2 ± 20.91 g.

#### *2.2. Pretreatment with Superheated Steam Roasting*

The CM fillets were placed in the AERO Steam Oven chamber (DFC-560A-2R/L, Naomoto Corporation, Osaka, Japan), maintaining the temperature at 270 ◦C for 4 min as followed by the previous method of Mohibbullah et al. [21]. Afterward, the roasted CM

fillets were transferred into the polyamide bags, vacuum-packed using a vacuum sealer machine (Sambo Tech. Corporation, Gyeonggi-do, Korea), and finally stored at 4 ◦C until otherwise/further use.

#### *2.3. Smoking Treatment with Variant Sawdust*

The 24-h dip treatment was conducted for CM fillets with a brine solution of 8% NaCl CM samples were then dried by placing at 30 ◦C temperature for 30 min and allowed to drain out excess water. After removing water, the fillets were submitted to smoking treatment. The samples were placed inside the smoking kiln (Braai Smoker, Bradley, Canada) and generated by combusting different sawdust (Apple, Chestnut, oak, Cherry and Walnut) materials. During the smoking process, the smoke temperature was recorded using a sensitive thermometer at different time intervals such as 20, 25 and 30 min. The desired temperature was 70 ◦C and the smoked CM fillets were stored at 4 ◦C until analysis.

#### *2.4. Sensorial Properties*

The parameters such as color, odor, flavor, and overall acceptance of treated and untreated smoked products were observed and all the CM fillet samples were impartially coded preceding the sensory evaluation. These characteristics were judged by ten panelists with ages between 25 and 40, who were previously well trained about the evaluation method. During the sensory evaluation, the changes in color, flavor, odor, and overall acceptances were assessed by a 9-point hedonic scale (1 for extremely dislike and 9 for extremely like). The value 5 was considered as the threshold limit and values less than 5 were considered as rejected [22].

#### *2.5. Physical Properties*

#### 2.5.1. Evaluation of Odor Intensity

Following the procedure of Macagnano et al. [23], a total of 5 g of sample was taken in a 50 mL conical flask, the lid closed with parafilm and fitted with an odor concentration meter (XP-329, New Cosmos Electric Co. Ltd., Osaka, Japan). The odor intensity was calculated by observing the signal (peak), which is expressed as an arbitrary unit.

#### 2.5.2. Calculation of Weight Loss

The Goulas and Kontominas [24] method was used to calculate the weight loss of CM fillets. It was done by evaluating the differences between the weight of the samples before and after oven drying.

#### 2.5.3. Texture Analysis

The texture of CM fillets was assessed by Brookfield Texture Analyzer (Massachusetts, USA), equipped with a computer software system (Texture PRO CT, Middleboro, USA). A series of parameters, explicitly hardness, chewiness, cohesiveness and springiness, were estimated by following the experiment of Ganesan and Benjakul [25].

#### 2.5.4. Evaluation of Color

The CM fillet color was assessed by the CM-700d Konica Minolta (Tokyo, Japan) instrument. Using a Hunter system value, the surface color of each sample was calculated as in previous reports [26–28].

#### *2.6. Chemical Properties*

#### 2.6.1. Determination of pH

The homogenized CM sample (4 g) was thoroughly mixed with 45 mL of distilled water (DW) for 2 min and well centrifuged (SHG-15D, SciLab, and Seoul, Korea). The pH of the sample was measured using a pH meter (OHAUS STARTER 2100, Seoul, Korea) with a glass electrode.

#### 2.6.2. Determination of Volatile Base Nitrogen (VBN)

The VBN value is measured by a Conway micro diffusion method. A total of 5 g of sample was mixed with 25 mL of distilled water, filtrated, and mixed with potassium carbonate, following the method of Ali et al. [29].

#### 2.6.3. Determination of Thiobarbituric Acid-Reactive Species (TBARS)

A total of 5 g CM fillet sample was taken and thoroughly mixed with 12.5 mL TBARS solution containing 20 trichloroacetic acid and 2 M phosphoric acid, followed by filtration, incubation at 95 ◦C for 30 min, and reading the absorbance value at 530 nm wavelength [30].

#### 2.6.4. Determination of Trimethylamine (TMA) Analysis by Gas Chromatography-Mass Spectrometry (GC/MS)

The presence of TMA in spoiling marine organisms is due to the bacterial degradation of TMAO. The TMA is used as a marker to assess the seafood quality because it is the primary component to producing fishy off-odor [31]. The GC/MS method was performed to estimate the value of trimethylamine N-oxide (TMAO) in both raw and smoked CM products, as followed by Mohibbullah et al. [21]. In this analysis, 10 g of fish sample was mixed with 10 mL of distilled water, sonicated for 20 min, filtrated, and then volatilized using GC instrument (Agilent 7890B GC, Agilent Technologies Inc., Santa Clara, CA, USA) at an oven temperature of 240 ◦C which increased from 40 to 210 ◦C following the flow rate of 10 ◦C/min. Procedures for quantification of TMAO content in raw and smoked CM fillet were followed by comparing the standard compound (Sigma Aldrich, St. Louis, MO, USA) run side-by-side.

#### *2.7. Microbiological Properties: Total Bacterial Count (TBC) and Total Coliform*

The ten-fold serial dilution method was used to determine the bacterial load of the CM fillet sample. A 1 g of sample was taken and transferred to a test tube containing 9 mL of sterile saline solution, spread quickly on the surface of plate count agar (Difco, Franklin Lakes, NJ, USA), as followed by Chen et al. [26]. The total coliform count was estimated following the preferred method of the US Food and Drug Administration (FDA) [27].

#### *2.8. Nutritional Properties*

#### 2.8.1. Analysis of Nutritional Composition

The nutritional composition of smoked CM fillets was determined by following the method of Horwitz et al. [32], as performed in the Traditional Microorganism Resources Center, Daegu, South Korea.

#### 2.8.2. Analysis of Amino Acid by HPLC Method

The samples of hot smoked CM fillets were duplicated in measurement for the estimation of amino acid content using a High-Performance Liquid Chromatography (HPLC; LC-20A, Shimadzu Corp., Kyoto, Japan) at Pukyong National University laboratory, Busan, South Korea. This method was followed by the previously published procedure of Gheshlaghi et al. [33].

#### *2.9. Statistical Analysis*

The statistical analysis was performed by IBM SPSS software, version 21.0 (IBM, Corp., New York, NY, USA). The data of different biochemical parameters were triplicated and calculated as mean ± standard deviation (std). Results were considered to be statistically significant at 95%, where the significance level was set at *p* < 0.05.

#### **3. Results**

#### *3.1. Effect of Different Sawdust Smoke on Sensory Analysis*

The color and flavor of CM were developed through the hot smoke of different sawdust. To identify the best smoke material in the processing of CM fillets, two different

smoking duration of 20 and 25 min were maintained in the present study, as followed by the previous reports [20,21]. The results significantly (*p* < 0.05) improved sensorial features when CM fillet was smoked by oak sawdust, followed by apple, cherry, and walnut sawdust at 20 and 25 min, respectively (Figure 1). Therefore, based on screening results with different sawdust materials, oak sawdust was selected as the best smoking material for further study. ‐ ‐

‐

‐

‐

‐

**Figure 1.** Effect of various sawdust (apple, chestnut, oak, cherry, and walnut) materials on instrumental odor intensity (**a**), and overall preference based on sensory attributes (**b**) in hot smoked chub mackerel (CM) fillet at two smoking times (20 and 25 min). Data were expressed as mean ± std (*n* = 10), where different letters were statistically different (*p* < 0.05).

#### *‐ 3.2. Effect of Time-Dependent Smoking on Odor and Sensory Evaluation ‐*

‐ ‐ The consumer's acceptability and preferences of combined treatment of superheated steam and hot smoked CM were evaluated by an instrumental odor analysis and sensorybased 9-point hedonic scale. The results indicated that the odor intensity significantly (*p* < 0.05) improved with the increased smoke time at 0, 10, 15, 20, 25, and 30 min. However, in the sensory analysis as performed by skilled panelists, 20 and 25 min smoke times were found to have more acceptable sensorial features such as appearance, odor, taste, and overall preferences and were optimized for the next experiment (Figure 2). ‐ ‐

**Figure 2.** Changes of odor intensity (**a**), and sensory quality (**b**) in CM fillet with different smoke times (0, 15, 20, 25, and 30 min) by oak sawdust. The data were stated as mean ± std (*n* = 10), where different letters were statistically different (*p* < 0.05).

#### *3.3. Effect of Smoke Temperature on Physical Features of CM Fillet* 3.3.1. Effect of Smoking Time on Weight Loss

The weight loss of processed CM revealed that there were no significant differences at 20, 25, and 30 min of smoking. In the present study, it was evident that smoking time being applied on processed CM had a non-significant effect on weight loss (Figure 3a).

**Figure 3.** Changes of weight loss (**a**), texture analysis (**b**), odor (**c**), and color value (**d**) in hot smoked CM fillet products with different smoking time. The data were stated as mean ± std (*n* = 10), where different letters were statistically different (*p* < 0.05).

#### 3.3.2. Effect of Smoking Time on Texture

‐ ‐ Instrumental texture analysis of processed food products is the most essential and important measurement in the assessment of quality for consumer acceptance. Here, in time-course smoking of 0, 20, 25, and 30 min, there were no significant changes in the textural properties of hardness, chewiness, cohesiveness, and springiness of hot smoked CM, as compared with non-smoking CM product (Figure 3b).

‐

#### 3.3.3. Effect of Smoking Time on Odor

‐ The odor of hot smoked CM was estimated at each 5 min and 10 min of the interval, which significantly (*p* < 0.05) decreased in the smoking time of 20 min. However, after 25 min smoking, odor intensity was similar to that of non-smoked CM, indicating that hot smoking did not influence the odor quality at this condition. (Figure 3c).

#### 3.3.4. Effect of Smoking Time on Color

‐ The color attribute is considered as one of the most important parameters of processed fish products, which directly reflects consumer choice. Instrumental color values are depicted as L for lightness (lightness/darkness: higher/lower), a for red/green (higher/lower), b for yellow/blue (higher/lower). All color parameter values obtained from the color meter showed insignificant differences with smoking time (0 to 30 min), as compared with non-smoked CM. The results suggested that an improved hot smoking process had no obvious effects on color changes of CM fillets, maintaining the original characteristics of the product appearance (Figure 3d).

#### *3.4. Effect of Smoking Treatment on Biochemical Properties of Processed CM* 3.4.1. Effect of Smoking Time on pH

pH in processed fish products is used as an indicator for determining the quality of fishery products. Here, the pH level in smoked CM fillet significantly decreased (*p* < 0.05)

‐

with smoking time at 20, 25 and 30 min, respectively, and reached its acceptability limit (pH = 6.32–6.22) (Figure 4a).

‐ **Figure 4.** Changes of pH (**a**), total bacterial count (TBC) (**b**), thiobarbituric acid reactive substances (TBARS) (**c**), and volatile base nitrogen (VBN) value (**d**) in hot smoked CM fillet with different smoking times. The data were stated as mean ± std (*n* = 10), where different letters were statistically different (*p* < 0.05).

#### 3.4.2. Effect of Smoking Time on TBC and Coliform

‐ The microbial quality of processed fish products is an important aspect of food safety to end-point consumers. Therefore, it is very important to assess the TBC and Coliform bacteria in hot smoked CM for ensuring the quality. The present study found a significant (*p* < 0.05) decrease in bacterial biomass that was assessed in smoked CM fillet with the extension of smoke time (Figure 4b).

#### 3.4.3. Effect of Smoking Time on TBARS Changes

‐ ‐ Thiobarbituric acid reactive substances (TBARS) are a convenient pathway to determine the level of fat oxidation in processed fish products and are expressed by measuring the malondialdehyde (MDA) content of the sample. The result showed that a significant (*p* < 0.05) decrease in MDA content from 0 to 25 min of smoking time and again a nonsignificant increment in MDA content at 30 min smoking time were found in processed CM fillet (Figure 4c).

#### 3.4.4. Effect of Smoking Time on VBN Changes

In this experiment, it was found that the VBN of hot smoked CM significantly (*p* < 0.05) decreased with smoking time and the lowest VBN value was observed at 30 min (Figure 4d).

#### 3.4.5. Effect of Smoking Time on Sensory Attributes

As performed by the trained panelists, sensory evaluation on appearance, odor, taste and overall preferences were found to be significantly (*p* < 0.05) increased at 25 min and further declined at 30 min smoking time (Figure 4e). Thus, we optimized 25 min smoke time for further analysis of CM fillets on the effects of storage conditions.

#### *3.5. Effect of Smoking on Storage Conditions of Processed CM*

#### 3.5.1. Analysis of Smoked CM Microbiological Quality on Storage Duration ‐

The growth of food-borne microorganisms is greatly influenced by temperature. Herein, the microbiological quality of hot smoked CM fillets was characterized by TBC and coliform count and evaluated in two different storage conditions of 10 and 15 ◦C temperature. As shown in Table S1, no bacterial growth (CFU/g) was developed during the storage period up to 34 days, indicating that shelf-life of hot smoked CM could be extended up to 34 days without invasion of bacterial biomass at both storage temperatures 10 and 15 ◦C. ‐ ‐ ‐

#### 3.5.2. Effect of Storage Duration on VBN Changes

The hot smoke processed CM fillets were kept at two storage conditions, 10 and 15 ◦C temperature for 0 to 34 days. According to Figure 5a, the VBN value significantly (*p* < 0.05) increased with the extension of storage duration. The VBN value of smoked CM was assessed in between the ranges of 13–20.3 (mg %) and 13–26 (mg %) at 10 and 15 ◦C, respectively, throughout the storage period of 0–34 days. The study indicated that the 10 ◦C storage condition had better performance for preserving hot smoked CM fillets. ‐

‐ **Figure 5.** Changes of VBN (**a**), TBARS (**b**) and overall preference (**c**) value in hot smoked CM fillet at two different temperatures (10 and 15 ◦C) in a storage period 0 to 34 days. The data were stated as mean ± std (*n* = 10), where different letters were expressed as statistically different (*p* < 0.05).

#### 3.5.3. Effect of Storage Duration on TBARS Changes

‐ ‐ The TBARS value of hot smoked CM fillet was found to be significantly (*p* < 0.05) increased during the entire storage time at both storage conditions (10 and 15 ◦C) and varied between 3.2–4.9 and 3.2–8 mg/g, respectively (Figure 5b). This observation supported thatCM fillets at 25 min smoking were shown to have extended shelf-life when stored at 10 ◦<sup>C</sup> temperature condition.

#### 3.5.4. Effect of Storage Duration on Overall Sensory Preferences

The overall sensory preferences of CM fillets were significantly (*p* < 0.05) decreased with the increase of storage time, from 0 to 34 days, at two different storage conditions 10 and 15 ◦C (Figure 5c). The overall preference of smoked CM remained at an acceptable condition up to 32 days at 10 ◦C storage temperature, whereas CM fillets at 15 ◦C storage temperature remained safe up to 13 days and then rapidly declined in quality.

#### *3.6. Effect of Hot Smoking on TMA Changes*

TMA content was found as 2.25 ± 0.29 and 1.25 ± 0.06 µg per 100 g sample of both raw and hot smoked CM, respectively (Figure 6). The study revealed that hot smoking of CM fillets had a promising suppressive effect on TMA content, which might be the cause of inhibiting the growth of specific spoilage bacteria.

μ

‐

**Figure 6.** Effect of smoking on TMA content in CM fillets. The results were compared with (**a**) standard, (**b**) raw CM, and (**c**) hot smoked CM.

#### *3.7. Effect of Hot Smoking on Preserving Nutritional Attributes*

The nutritional quality or proximate composition of processed fish products varies with processing methods (e.g., drying, salting, and smoking), handling, and storage conditions. The protein, lipid and carbohydrate content of processed CM were 23.880, 30.288, and 0.664 g/100 g, respectively. The minerals viz. potassium, sodium calcium, and iron contents of hot smoked CM were reported as 296.394, 270.803, 67.168, and 1.214 mg/100 g, respectively (Table 1).


**Table 1.** The nutritional quality of hot smoked CM fillet.

#### *3.8. Effect of Hot Smoking Treatment on Determination of amino Acid in Smoked CM Fillet*

The amino acid analysis is an important parameter for evaluating the protein quality of fish samples. This study identified 17 amino acids including 9 essential and 8 nonessential amino acids in hot smoked CM fillets (Table 2). Among essential amino acids, lysine (2252.2 mg/100 g), leucine (2153.3 mg/100 g) and arginine (1736.3 mg/100 g) were found higher, while that of glutamic acid (3691.9 mg/100 g), proline (2746.0 mg/100 g), and aspartic acid (2213.2 mg/100 g) were higher in the contents of non-essential amino acids found in hot smoked CM. These findings suggested that collective treatment of superheated steam roasting and hot smoking of CM fillets enriched amino acid profiles along with promoting physicochemical properties and antimicrobial performances, which ultimately rendered extension of shelf-life during the storage period.


**Table 2.** Amino acid content (mg/100 g) in hot smoked CM fillet.

#### **4. Discussions**

Because of the large number of nutrients, including protein, long-chain omega-3 polyunsaturated fatty acids (*n*-3 PUFAs), and micronutrients including selenium, iodine, potassium, vitamin D and B, fishes are widely considered as a healthy and balanced diet for

dietary uptake [34]. Fish is a highly perishable culinary commodity because of the fact that it becomes extremely vulnerable to oxidation resulting in the development of off-odor and -flavor and subsequent spoilage takes place while marketed [35]. Therefore, it is of need to develop or improve preservation and processing techniques more efficiently for keeping quality as well as extending the shelf-life of fish and fishery products. This study was set to the effects of combined treatment of superheated steam roasting and hot smoking (70 ◦C) of CM in improving sensory, physicochemical, nutritional, and microbiological qualities even at storage conditions, intending to achieve the best quality product for the end-point consumers.

Sensory evaluation is one of the quickest, easiest, and most common methods to evaluate the quality of fish that can be done by human senses such as sight (color), tactility (to test the flesh elasticity), and olfaction (odor) [14]. The present study conducted a sensory assessment technique to know the consumer preferences of hot smoking (70 ◦C) of Chub mackerel fillet at the different smoking time (0, 20, 25 and 30 min) with different sawdust materials (Apple, Oak, Chestnut, Cherry and Walnut). The study found a supreme sensory quality (appearance, odor, color, texture and overall preferences) product at 25 min smoke time and oak sawdust as the best smoke generating materials. This study complies with the study of hot smoked Wels catfish observed by Küçükgülmez et al. [36]. Similar results were also found in our previous studies that oak sawdust provided good sensorial attributes while smoking of adductor muscle of pen shell *Atrina pectinate* and pacific saury *Cololabis saira* [20,21]. The characteristic flavor and aroma of hot smoked CM are predominantly achieved through the absorption of many volatile components from wood smoke, among which phenolics might be the most influential.

The firmness of the fish products is considered as one of the most important quality parameters to determine consumer preference [37]. The weight loss of the hot smoked CM was found unchanged at the different smoke times (20, 25, and 30 min), which indicates good quality products, and this study was supported by Baten et al. [16,19]. Textural quality significantly affects the overall quality assessment and textural profile, which includes the assessment of hardness, springiness, cohesiveness, chewiness, adhesiveness, and gumminess of a fish product [18]. The texture attributes of hot smoked CM remained constant throughout the smoking time (20, 25, and 30 min), maintaining unique sensorial quality for the consumers, which was more or similar to the study of Mohibbullah et al. [20]. The instrumental odor and color value play an important role in the acceptance or rejection of a fish product by consumers [38]. The time and temperature of smoking treatment had no significant impact on odor and color of processed CM and the findings were similar to the previous study [39]. The pH is an important parameter for assessing the quality of the fish product, as low pH (<6.2) early in the storage indicates good nutritional status and increased shelf-life in the latter stage [18,40]. The present study found significantly lower and acceptable pH at a 25 min smoke and this is in accordance with the study of [30]. Wood smoke has antibacterial properties that retard the growth of bacteria [13,24] and this study showed a similar effect when smoking time significantly decreased bacterial load.

The TBARS is a by-product of lipid peroxidation and its assay measures the MDA level in the fish sample. The lowered level of MDA content was observed in a decreasing trend with increasing smoke time. The previous study reported that smoking treatment decreased TBARS values, where the consumability range is suggested as no more than 7-8 MDA mg/kg [30]. The VBN is considered as one of the prime indicators of raw and processed fish products to measure the degree of freshness [41]. The VBN value of hot smoked CM showed a significant decline in increasing smoke time and remained in an acceptable range as compared with the standard VBN value of fresh fish being 5–20 mg N/100 g muscle [42]. The hot smoking method significantly decreased the VBN value of Japanese Spanish mackerel, *Scomberomorus niphonius* [17]. In storage conditions (10 and 15 °C), the VBN and TBARS values were found to show an increasing trend, which is in compliance with the study of Mohibbullah et al. [21] and O ˘guzhan Yildiz [30]. However, the VBN and TBARS values of hot smoked CM of shelf-life retained an acceptable value of

17.7 mg% and 3.7 MDA mg/kg, respectively, after 32 days at 10 °C storage temperature. The results concluded that hot smoked CM provided excellent freshness conditions and did not exceed the maximum permissible limit of VBN and TBARS values up to the storage period of 32 days. The overall sensorial impacts of CM were significantly (*p* < 0.05) decreased with increasing of storage duration either at the 10 or 15 °C storage condition, which is an agreement with the reports suggesting that the sensory scores of smoked bonito, catfish, gwamegi and mackerel decreased with the extension of storage time [43–45]. The suppression of bacterial (TBC and Coliform) growth was observed throughout the storage period at both storage conditions (10 °C and 15 °C). Superheated steam roasting with hot smoking treatment prevented the growth of bacterial proliferation [13,24]. It is expected that the hot smoking process may release phenolic compounds, having antioxidant and anti-microbial actions on the surface of CM fillets.

The TMA is a very efficient index for determining the degree of spoilage in fresh, processed and lightly preserved seafood [46]. The TMAO is converted into TMA in dead fish due to the action of intrinsic enzymes and bacterial action [31,47]. Considering the TMAO content in seafood, it appears to range depending on age, species, environmental factors and harvesting time [48]. The hot smoking with oak sawdust decreased the TMA content to 1.25 µg per 100 g in smoked CM while comparing with raw CM. A similar trend of results was shown by Mohibbullah, et al. [21] and Baten et al. [17,20]. The level of TMA in hot smoked CM fillets was considered safe for human consumption, when the result was compared with the maximum consumable limit of TMA value of 10 mg/100 g [48]. The nutritional content of CM was high and was found to be comparable with Mohibbullah et al. [21]. It is evident that hot smoking treatment increased the crude protein, crude lipid, crude fiber, ash, and other minerals contents by reducing the moisture content, as can be seen in the present CM product and which is comparable to other fish and meat products [49–51]. In addition, the amino acid profile provides essential and non-essential amino acids, and it is determined that the quality of a protein is also beneficial to the human diet [52]. The amino acids and peptides are important components for the development of seafood flavor [53]. The hot smoked Chub mackerel consists of 17 amino acids, of which 9 are essential acids and 8 non-essential amino acids. The results observed in the present study are to some extent similar to Adeyeye et al. [12], who found 18 amino acids in five hot smoked processed fish. Superheated steam roasting with hot smoking treatment significantly increased the lysine and other essential amino acids in CM products, when compared with other processing techniques [54].

#### **5. Conclusions**

The improved process technology of CM, namely a combination of superheated steam roasting (270 ◦C for 4 min) with hot smoking (70 ◦C), was employed in the present study. Among different sawdust materials to smoke superheated steam roasted CM fillets with time-dependent smoking, oak sawdust showed better sensory characteristics at an optimal smoke time of 25 min. The processed CM fillets with time-dependent smoke provided desirable odor intensity, negligible impact on weight loss, insignificant color changes, and acceptable textural properties, especially, after smoking at 25 min. With this condition, the processed CM also suppressed pH, TBC, VBN, and TBARS values to an acceptable limit at the smoke duration of 25 min. At 25 min smoking time, hot smoked CM was extremely good, and able to preserve its chemical properties and inhibit microbial spoilage successfully at 10 ◦C up to 32 days in storage condition. Considering the quantity and quality, hot smoked CM posed a significant source of valuable nutrients (protein, lipid, and minerals), especially essential and non-essential amino acids. Moreover, hot smoked CM contained a lower and negligible amount of TMA compared to raw CM and may be considered as a safe item for human consumption. Therefore, this combined process technology could be effectively utilized by processors to increase the storage-ability of CM with safe and healthy conditions at the consumer level.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2076-341 7/11/6/2629/s1, Table S1: Microbial growth in storage condition of hot smoked CM.

**Author Contributions:** Conceptualization, J.-S.K., J.H.S. and J.-S.C.; methodology, J.-S.C. and J.H.S.; validation, J.H.S. and J.-S.C.; formal analysis, M.M.R., N.E.W., M.M. and M.A.B.; investigation, N.E.W. and J.-S.C.; resources, M.M.R., N.E.W. and J.-S.C.; data curation, M.M. and J.-S.C.; writing original draft preparation, M.M.R., M.A.B. and M.M.; writing—review and editing, M.M. and J.-S.C.; visualization, M.M. and J.-S.C.; supervision, J.-S.K., J.H.S. and J.-S.C.; project administration, J.-S.K., J.H.S. and J.-S.C.; funding acquisition, J.-S.K., J.H.S. and J.-S.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was funded by the Ministry of Oceans and Fisheries, Korea, under the project no. PJT200885 entitled "Development and commercialization of traditional seafood products based on the Korean coastal marine resources".

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Effect of Chitosan Nanoemulsion on Enhancing the Phytochemical Contents, Health-Promoting Components, and Shelf Life of Raspberry (***Rubus sanctus* **Schreber)**

**Shirin Rahmanzadeh Ishkeh <sup>1</sup> , Habib Shirzad 1,\*, Mohammad Reza Asghari <sup>1</sup> , Abolfazl Alirezalu <sup>1</sup> , Mirian Pateiro <sup>2</sup> and José M. Lorenzo 2,3,\***



**Citation:** Ishkeh, S.R.; Shirzad, H.; Asghari, M.R.; Alirezalu, A.; Pateiro, M.; Lorenzo, J.M. Effect of Chitosan Nanoemulsion on Enhancing the Phytochemical Contents, Health-Promoting Components, and Shelf Life of Raspberry (*Rubus sanctus* Schreber). *Appl. Sci.* **2021**, *11*, 2224. https://doi.org/10.3390/ app11052224

Academic Editor: Theodoros Varzakas

Received: 11 February 2021 Accepted: 1 March 2021 Published: 3 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Abstract:** Due to high water content and perishability, the raspberry fruit is sensitive to postharvest fungal contamination and postharvest losses. In this study, chitosan was used as an edible coating to increase the storage of raspberries, and nanotechnology was used to increase chitosan efficiency. The fruit was treated with an emulsion containing nanoparticles of chitosan (ECNPC) at 0, 2.5, and 5 g L−<sup>1</sup> , and stored for 9 d. Decay extension rate, fruit phytochemical contents, including total phenolics, flavonoids, and anthocyanin content, phenylalanine ammonia-lyase (PAL), and guaiacolperoxidase enzymes and antioxidant activity, and other qualitative properties were evaluated during and at the end of storage. After 9 d of storage, the highest amounts of phenolics compounds, PAL enzyme activity, and antioxidant activity were observed in fruit treated with ECNPC at 5 g L−<sup>1</sup> . The highest levels of total phenol, PAL enzyme activity, and antioxidant activity were 57.53 g L−<sup>1</sup> , 118.88 µmol/min *trans*-cinnamic acid, and 85.16%, respectively. ECNPC can be considered as an effective, safe, and environmentally friendly method for enhancing fruit phytochemical contents, postharvest life, and health-promoting capacity.

**Keywords:** edible coating; nanoemulsion; guaiacol peroxidase; shelf life; anthocyanins; phenylalanine ammonia-lyase

#### **1. Introduction**

Fruit and vegetables play an important role in improving human health. Among different food crops, colorful fruit and berries, including raspberries, are rich in different antioxidant and anti-stress compounds with powerful anti-cancer and anti-inflammatory attributes [1–3]. *Rubus sanctus* Schreber is a member of the family Rosaceae and its subfamily is Rosoideae. Berry fruits contain abundant phenolic compounds such as phenolic acids, flavonoids, and anthocyanins, and they are an excellent source of antioxidants such as vitamins (A, and ascorbic acid). In addition, they are useful for the treatment of various diseases, particularly diabetic patients [4].

However, this fruit is very perishable and undergoes substantial changes in antioxidant and phytochemical contents during postharvest storage and handling stages. Water loss, softening, decay extension, and metabolic activities are the main causes for a decrease in fruit quality, phytochemical contents, and marketability [5]. The content of bioactive compounds in *Rubus* depends chiefly on the sowing method, variety, harvest season, and postharvest handling. Several investigations indicate that organic farming systems followed by postharvest conditions have a considerable influence on the quality of berry

fruit production [6]. The decrease in nutritional properties and marketability of fruit and vegetables during postharvest handling and storage causes significant economic damages to the producers [7]. On the other hand, with the increase in public awareness regarding the adverse effects of the use of chemical residues on human health and environmental safety in recent years, there has been a lot of growth in the market for the demand for organic and chemical-free horticultural products [8]. However, products produced for fresh consumption are highly susceptible to major mechanisms of losses, including, enzymatic bleaching, water loss, and microbial contamination [9].

To increase the postharvest shelf life of harvested fruit and vegetables, there are some useful treatments, including modified atmosphere storage, changed atmosphere packaging, edible coatings, and the use of different natural compounds [10]. Among the treatments mentioned, the use of edible coatings is one of the promising methods for preventing water loss, decreasing metabolic and enzymatic activities and maintaining the aroma and flavor of the crops; because of a relative permeability to different respiratory gases, these coatings prevent the adverse effects of common modified atmosphere storages and packages, such as creating an unpleasant odor and smells [7].

Edible coatings are renewable compounds, including lipids, polysaccharides, and proteins, which are responsible for decreasing the exchange of water vapor, gases, etc., and many compounds used in postharvest technologies such as antimicrobial agents, antioxidants, dyes, and authorized food additives can also be added to them [11]. In addition, a new approach for enhancing fruit phytochemicals, quality attributes, and postharvest life is the use of natural compounds, such as plant growth regulators (PGR) and phytohormones as alternatives to chemical treatments during the production and postharvest stages of food crops [12,13].

Chitosan is a polymer of (4,1) β-N-acetyl-d-glucosamine derived from the chitin of crustaceans, insects, and fungi, which can play an important role in the physiology of plants and harvested crops [14]. It can be used as an edible coating for many harvested horticultural products [15]. In addition to having semi-permeability behavior against respiratory gases, resulting in decreased respiration and metabolic activities, chitosan has direct antimicrobial, bactericidal, and antivirus properties that reduce the need for the use of chemical compounds and increase the safety of the food products [15,16]. Due to its semi-permeability, chitosan has a relative permeability to water vapor and is a good inhibitor of oxygen exchange, thereby altering the internal atmosphere of the product and increasing the shelf life of the product [16,17]. As a PGR, chitosan has been shown to enhance the natural resistance mechanisms of the plants and harvested crops against different pathogens and stress conditions [18].

Moreover, the use of nanotechnology in edible coatings and packaging can increase their efficiency compared to conventional edible coatings and improve the quality of edible coatings by reducing the particle size and the pores of the coatings [19]. The small size of nanoemulsion particles has two important effects—(1) it increases the stability and physicochemical properties of the coating and (2) it can increase the biological activity of lipophilic materials by increasing the surface area to the mass unit rate [20]. Nanoemulsions enhance the bioavailability of bioactive materials [21] and the bactericidal properties of antimicrobials because they can pass as active elements across the bio-membranes [22]. Nanoemulsions are now widely used to cover, protect, and transfer lipophilic materials on foods, fruit, and vegetables [23,24].

Raspberry are rich sources of different powerful antioxidants including anthocyanins, ellagic acid, gallic acid, catechins, camphor, and salicylic acid, with high capacity in the prevention of unwanted damages of free radicals and reactive oxygen species (ROS) to cell membranes and other structures in the human body. Previously mentioned antioxidant compounds may reduce a person's risk of heart disease by preventing platelet buildup and lowering blood pressure by anti-inflammatory agents. *Rubus* antioxidant compounds and anti-inflammatory agents are correlated with cancer protection by reducing the reproduction of cancer tumors. Due to the high levels of antioxidant compounds such as

vitamin C and polyphenols, raspberries have an oxygen-radical absorbance capacity of about 30 mM to 100 mM per 100 gr [25]. Ascorbic acid or vitamin C provides extracellular and intracellular antioxidant activity mainly by scavenging reactive oxygen species [26].

Raspberries are also a great source of riboflavin, folate, niacin, magnesium, potassium, and copper. The presence of these nutrients and their synergistic role in improving human health have led to the fact that this fruit is a good source to answer daily needs for various micronutrients (anthocyanins, polyphenols, ascorbic acid, fiber, proteins, vitamins, and minerals) [27]. This fruit is one of the most popular berry fruit, but due to very high sensitivity to different postharvest losses, such as rapid water loss, mechanical damage, and pathogen sensitivity, they have a very short shelf life, leading to rapid phytochemical depletion and quality deterioration [28].

The purpose of this research was to increase the shelf life and maintain phytochemical contents and quality of this valuable fruit with the use of chitosan nanoparticle coatings as a natural safe compound. We studied the effects of this coating at different concentrations on decay extension rate, weight loss, different quality parameters of the fruit, and different resistance mechanisms, including total antioxidant activity and different antioxidant fractions. Moreover, the effects of this coating on fruit taste and flavor as important factors determining consumer acceptance were studied.

#### **2. Materials and Methods**

#### *2.1. Preparation of the Fruit for the Treatment*

Raspberry fruit (*Rubus sanctus* Schreber) were collected in August at commercial maturity from Khan Valley, Urmia, with a longitude of 45◦07′09′′, the latitude of 37◦19′16′′ and an elevation above sea level of 1392 m. Raspberry shrubs were wild-growing and organic. Sampling was carried out early in the morning. The fruit was selected for the similarity of size, maturity, and color, and the damaged and unshaped fruit samples were removed. The samples were then transferred in a fridge of 4 ± 1 ◦C to the Department of Horticulture Science of Faculty of Agriculture of Urmia University (Urmia, Iran). The *Rubus sanctus* Schreber species identification was conducted by botanist Dr. Shahram Bahadori. The voucher herbarium samples of the collected *Rubus sanctus* Schreber species have been deposited at the herbarium of Urmia University of Medical Sciences (HUPS-359). For each experimental unit, 5 g of fruit was used. Preliminary measurements of the indices of the fruit were carried out on harvest day (day 0) and the fruit was treated with different emulsions containing nanoparticles of chitosan (ECNPC) on the same day of harvest.

#### *2.2. Preparation of Chitosan Nanoemulsion*

Chitosan (degree of acetylation (85%), molecular weight (50,000–80,000 Da), and size of nano chitosan particles: <50 nm) nanoemulsion was purchased from Nano Novin Polymer Co. (Sari, Iran). Before applying the treatment, to ensure the quality of the edible coating, the coating was sent to Shahid Beheshti University (Tehran, Iran) to determine the particle size. The dynamic light scattering (DLS) device (NanoPhox 90–246 V, Sympatec GmbH, Clausthal-Zellerfeld, Germany) was used to determine the size of nanoemulsion particles. The type of device lamp was He-Ne Laser 623 nm, and the range of measurement of this device was 1–10,000 nm.

#### *2.3. Application of the Nanoemulsion in Raspberry*

For the treatment of the fruit with chitosan nanoemulsion, concentrations of 0, 2.5, and 5 g L−<sup>1</sup> were prepared. The fruit (60 g of fruit for each repetition) were immersed in containers containing chitosan nanoemulsion for three minutes, and after drying for 8–10 min placed in the pre-sterilized disposable plastic containers (polyethylene terephthalate; 23 cm × 17 cm × 8.5 cm) with four vent holes in the sides (2 mm in diameter). The lid of the containers (clamshell package) was sealed with parafilm to prevent the exchange of air and the packages were stored at cold storage of 4±1 ◦C with relative humidity (RH) of 85–95% for 9 d (Figure 1).

**Figure 1.** Effect of chitosan nanoemulsion on the shelf life of raspberry.

#### *2.4. Quality Evaluation*

#### 2.4.1. Fruit Titratable Acidity (TA), Total Soluble Solids (TSS), and pH

Total acidity was calculated by the titration method using sodium hydroxide (0.1 N) in terms of citric acid. In other words, it was titrated with soda solution (0.1 N) to reach pH 8.3. After applying, the value of the used soda was introduced in the following formula, and the acidity was calculated based on g of citric acid in 2 L of fruit extract and then, converted to percent [29]:

$$\text{TA} = \frac{100 \times \text{M} \times \text{N} \times \text{V}}{\text{S} \times \text{n}} \tag{1}$$

where TA = acidity value based on g of citric acid in 2 L sample extract, M = molecular weight of the dominant acid, n = dominant acid capacity, V = volume of the used soda, S = amount of the extract used, and N = normality of the used soda.

TSS of the sample extract was measured by a manual refractometer (ATAGO, Tokyo, Japan) at laboratory temperature. The distilled water was used to be calibrated the refractometer. A pH meter (pH-Meter CG 824, SCHOTT, Hofheim, Germany) was used to measure the pH of the fruit juice.

#### 2.4.2. Weight Loss of the Fruit

The digital scale (CANDGL300) was used to measure the value of the weight loss of the fruit. For this purpose, the difference in the weight of the fruit was calculated after 3 d, 6 d, and 9 d of storage, and the value of the weight loss compared to the first day was obtained [30].

$$\text{Weight Loss} \left( \% \right) = \frac{\text{Primary weight} - \text{secondary weight}}{\text{Primary weight}} \times 100 \tag{2}$$

#### 2.4.3. Fruit Firmness

A TA-XT plus texture analyzer (Stable Micro Systems Ltd., Godalming, UK) was used to determine the firmness of raspberries. The probe was set at a speed of 2 mm/s for test. The value of pressure (N) that was introduced into the tip of the firmness tester due to the resistance of the fruit tissue was read on the device [31].

#### 2.4.4. Decay Extension Rate and Fruit Sensory Evaluation

The effect of chitosan nanoemulsion at different concentrations on decay extension rate and consumer acceptance was evaluated by 72 evaluators (24 male and 48 females) with prior experience about the sensory attributes of this type of product. Measurement of the decay extension rate was performed by observation, and their average opinions were discussed and recorded. In this regard, score 1 was given to the samples with the lowest decay extension rate and score 10 was given to the highest rate.

The fruit panel test was also conducted for sensory evaluation. A randomized (complete) block design was conducted. The panelists tested the fruit taste and flavor and their average opinions were analyzed. Before each session, panelists were informed about the objectives of the study and the instructions to complete the test. The samples were individually labeled with aleatory numbers and randomly served to panelists situated in individual cabins during the sessions. For increasing the accuracy of sensory analysis between each testing crackers (unsalted) and water were utilized. The scores were ranged from 1 to 10, i.e., score 1 was given to samples with better taste and score 10 was given to those with the lowest sensory indices. The measurements were performed for the third, sixth, and ninth day separately [30].

#### *2.5. Measurement of Total Phenolics Content (TPC)*

TPC was determined using Folin–Ciocalteu reagent according to the method of Alirezalu et al. [32]. A total of 30 µL concentrated extract was poured into the test tube and 90 µL distilled water was added. Then, 600 µL of 10% Folin was added, and after 10 min, 480 µL of 7.5% sodium carbonate was added, and after placing it for 30 min in the dark at room temperature, a spectrophotometer (UV-1800, Shimatzu Corporation, Kyoto, Japan) was used to read the absorbance at 760 nm. Gallic acid was used as a standard. The total phenol content of extracts was expressed as g of gallic acid equivalent (GAE) per L of the fruit extract.

#### *2.6. Measurement of Total Flavonoid Content (TFC)*

To measure the total flavonoid content, 50 µL of the concentrated extract was poured into the test tube and 150 µL of 5% sodium nitrite was added. After 5 min, 300 µL of 10% aluminum chloride was added, and again, after 5–10 min, 1000 µL of 1% NaOH were mixed to the resulting solution and was brought to a volume of 5 mL with the deionized water and the absorbance of the resulting mixture at 380 nm was read, compared to the control. Quercetin was used to be drawn the standard curve. The total flavonoid content of the total extracts was expressed as g of quercetin equivalent per L of the fruit extract [33].

#### *2.7. Total Anthocyanin Content (TAC) Evaluation*

The pH-difference method was used to measure the total anthocyanin content. Firstly, two buffers were prepared with pH 1 and 4. Then, 2.5 mL of buffer 1 was poured into the test tube. Afterward, 100 µL of the extract was added to the solution poured into the test tube and the absorbance at two wavelengths of 700 nm and 530 nm was read. Then, 2.5 mL of buffer 2 (pH 4.5) was poured into another test tube and 100 µL of the

extract was added, and the absorbance at two wavelengths of 700 nm and 530 nm was read. Finally, the following formula was used to be calculated the total absorbance of each of the extracts [34]:

$$\mathbf{A} = \begin{pmatrix} \mathbf{A}530 - \mathbf{A}700 \end{pmatrix} \text{pH} = 1 - \begin{pmatrix} \mathbf{A}530 - \mathbf{A}700 \end{pmatrix} \text{pH} = \mathbf{4.5} \tag{3}$$

Total anthocyanin content was calculated by g of cyanidin 3-O-glucoside equivalent per kg fresh weight and according to the following formula:

$$\text{TAC} = \frac{\mathbf{A} \times \mathbf{M} \mathbf{W} \times \mathbf{V} \times \mathbf{DF} \times 100}{\varepsilon \times 100} \tag{4}$$

where A = Absorbance, MW = Molecular weight, DF = Dilution factor, and ε = Molar absorbance.

#### *2.8. Measurement of Total Antioxidant Activity*

#### 2,2-Diphenyl-1-Picrylhydrazyl-Hydrate (DPPH) Method

The 2,2-diphenyl-1-picrylhydrazyl-hydrate (DPPH) method was used to be evaluated the antioxidant activity. For this purpose, 2000 µL of the DPPH (pre-prepared) solution was poured into the sterilized test tubes. Then, a specific amount of the fruit extract of each of the samples was added, and the resulting solution was shaken at room temperature and in the dark for 30 min. The value of the absorbance of the resulting solution was read by a spectrophotometer at a wavelength of 517 nm. The above method was also used to be prepared the control, but instead of the extract, 80% methanol was used and calculated according to the following formula [35]:

$$\text{RSA} = \frac{(\text{Abs control})\text{t} = 30 \text{ min} - (\text{Abs sample})\text{t} = 30 \text{ min}}{(\text{Abs control})\text{t} = 30 \text{ min}} \times 100 \tag{5}$$

Ferric Reducing Antioxidant Power (FRAP) Method

In measuring the total antioxidant activity by the Ferric reducing antioxidant power (FRAP) method, 50 µL of the concentrated extract of raspberry fruit and 3 mL of fresh FRAP reagent (300 mM sodium acetate buffer with the acidity of 3.6, ferric-tris pyridyl-s-triazine 3, and ferric chloride) were mixed, and the resulting mixture was placed in a warm water bath (37 ◦C) for 30 min, and the value of the absorbance was read by a spectrophotometer at 593 nm, compared to the control. Iron sulfate was used to be drawn the standard curve and the results of the data were expressed as mol Fe+2/L extract [36].

#### *2.9. Evaluation of Enzymes Activity*

#### Phenylalanine Ammonia-Lyase (PAL) Enzyme

The method of Karthikeyan et al. [37] was used under slight modifications to measure the activity of the phenylalanine ammonia-lyase (PAL) enzyme. For this purpose, 0.5 g of fresh fruit texture was squeezed by using 1.5 mL extraction buffer (0.1 M borate buffer, 0.1% polyvinylpyrrolidone, and 1.4 mM mercaptoethanol) at pH 7. It was then centrifuged at 12 ◦C for 15 min at 12,000× *g*. After completion of the centrifuge, the supernatant was used to measure by the enzyme. To measure the enzyme, sample contents contained 30 µL of the enzymatic extract, 1 mL assay buffer (0.1 M borate buffer, 0.1% polyvinylpyrrolidone, and 1.4 mM mercaptoethanol) at pH 8.8, and 1 mL L-phenylalanine (12 mM), were placed in a warm water bath (Benmari method, 30 ◦C) for 30 min, and the absorbance was read by a spectrophotometer at 290 nm. The measurement of the PAL enzyme activity was performed by the Beer–Lambert law with an extinction coefficient of 9630 l.cm-µ in µmol/min *trans*-cinnamic acid.

#### Guaiacol Peroxidase (G-POD) Enzyme Activity

To measure the activity of the guaiacol peroxidase (G-POD) enzyme; 0.1 g of the fresh fruit texture was weighed and placed in a Chinese mortar on ice. Then, 1.5 mL of

extraction buffer (125 mM potassium phosphate buffer) at pH 7.8 was added, and after grinding, the supernatant was poured into a micro-tube and centrifuged at 4 ◦C for 10 min at 15,000× *g*. To measure the activity of the enzyme, 200 µL of the enzymatic extract was poured into the test tube, and 200 µL of 22 mM guaiacol was added. Afterward, 2 mL of 1250 mM potassium phosphate buffer was added. Then, for reading, the above solution was poured into a cell, and to prevent the rapid reaction of hydrogen peroxide with the present solution, hydrogen peroxide was added to a spectrophotometer and the absorbance was read at 0 min and 1 min after the addition of hydrogen peroxide at 470 nm, and its unit was expressed as µmol H2O2.min−<sup>1</sup> .mL−<sup>1</sup> extract [38].

#### *2.10. Statistical Analysis*

The experiment was conducted as a completely randomized design with 3 (ECNPC levels) × 3 (storage times) × 4 replications. The data analysis was performed using the SAS software (SAS Institute Inc., Cary, NC, USA). Duncan's multiple range test was used for the comparison of the mean data. The Friedman test based on a completely randomized block design was also used for the analysis of the decay and taste data.

#### **3. Results**

#### *3.1. The Coating Structure of Chitosan Nanoemulsion*

In Figure 2, the interval of the particle size of nanoemulsion chitosan is shown by the DLS device. The results of this analysis indicated that most chitosan nanoemulsion particles had a size ranging from 15 nm to 150 nm. This particle size range is approximately close to the size range of chitosan nanoemulsion particles reported in previous studies [19,39].

**Figure 2.** Particle size obtained by dynamic light scattering (DLS) analysis based on size and intensity of emulsion containing nanoparticles of chitosan (ECNPC).

#### *3.2. TA, TSS, and pH*

Fruit TA decreased during storage. TSS content showed an increasing trend during storage, and ECNPC significantly decreased the rate of increase in fruit TSS during storage (Table 1). The highest TA (1.25%) was recorded after 6 d in fruit treated with ECNPC at 5 g L−<sup>1</sup> , and the lowest (0.90%) value was recorded in control fruit after 9 d of cold storage. In both storage periods, the highest TA was seen in fruit treated with 5 g L−<sup>1</sup> ECNPC (*p* < 0.01). Moreover, the rate of soluble solids varied between 17.33% to 12.18%, which

the highest amount was observed in control on the 9 d after storage, and the lowest was observed after 3 d in samples treated with a concentration of 2.5 g L−<sup>1</sup> ECNPC (*p* < 0.01).

**Table 1.** Effect of different chitosan nano-emulsion treatment on the quality attributes of raspberry at harvest and in storage time at 4 ± 1 ◦C.


a–e Mean values with different letters indicate significant differences among samples. Data obtained from four replications, mean ± standard error, \*\* is significant at the 1% (Duncan's multiple range test).

ECNPC at both concentrations reduced the fruit pH compared to the control fruit so that the highest pH was observed in control fruit in all storage times, and the lowest was observed in treated fruit. With an increase in ECNPC concentration, the effects on decreasing the pH were increased (*p* < 0.01) (Table 1).

#### *3.3. Weight Loss*

ECNPC significantly affected fruit weight loss during cold storage (Table 2) (*p* < 0.01). The percentage of weight loss was enhanced in all treatments in storage time, and it was lower in samples treated with ECNPC. With an increase in ECNPC concentration, its effects on retaining fruit weight were increased.

**Table 2.** Effect of different chitosan nano-emulsion treatments on the quality attributes of raspberry at harvest and in storage time at 4 ± 1 ◦C.


a–f Mean values with different letters indicate significant differences among samples. Data obtained from four replications, mean <sup>±</sup> standard error, \*\* is significant at the 1% (Duncan's multiple range test). Chi<sup>2</sup> Decay = 31.93 and Chi<sup>2</sup> Panel test = 31.20.

#### *3.4. Firmness*

There was a significant difference between treatments and control fruit (*p* < 0.01). The highest firmness value was observed after 3 d of storage in fruit treated with 2.5 g L−<sup>1</sup> ECNPC (Table 2).

#### *3.5. Decay Extension and Fruit Taste*

A statistically significant difference was recorded between the fruit treated with different levels of ECNPC and the control fruit at all evaluation times (Table 2) (*p* < 0.01). The results show that the rate of decay extension in all treatments was high during storage, and the chitosan nanoemulsion treatments reduced the rate of this increasing trend. With the increase in ECNPC concentration the effects on decreasing decay extension rate were increased (*p* < 0.01) (Table 2).

The coating, in a concentration-dependent manner, significantly retained fruit taste and flavor quality (*p* < 0.01) (Table 2). The taste value of the samples varied between 3.2 and 9, which the highest value was related to the control sample after 9 d of storage indicating the lowest taste and flavor rate. On the contrary, the highest taste and flavor rate was observed in 5 g L−<sup>1</sup> chitosan nanoemulsion coating after 3 d of storage (Table 2).

#### *3.6. Total Phenol Content*

Fruit total phenol content was significantly affected by the treatment 5 g L−<sup>1</sup> chitosan nanoemulsion, and the treated fruit showed the highest phenolics at all evaluation times (Figure 3a) (*p* < 0.01). After 9 d of storage, the highest content of total phenol (57.53 g L−<sup>1</sup> ) was related to the samples treated with 5 g L−<sup>1</sup> chitosan nanoemulsion, and the lowest rate (26.78 g L−<sup>1</sup> ) was related to the samples treated with 2.5 g L−<sup>1</sup> chitosan nanoemulsion after 3 d of storage (Figure 3a).

**Figure 3.** Effect of chitosan nano-emulsion coatings treatment (at 0 g L−<sup>1</sup> , 2.5 g L−<sup>1</sup> , and 5 g L−<sup>1</sup> ) on (**a**) total phenolics content, (**b**) total flavonoid content, and (**c**) total anthocyanin content of raspberries stored for 9 d (at 4 ± 1 ◦C with 90–95% RH). Control refers to untreated raspberries. The data shown are the mean ± standard error of four replicates. Different letters indicate statistical significance (*p* < 0.01).

#### *3.7. Total Flavonoid Content*

The total flavonoid content of all fruit in all treated and control samples showed an increasing trend during storage. At all evaluation times, the highest total flavonoid content was recorded in fruit treated with ECNPC at 5 g L−<sup>1</sup> , but the increase in flavonoid content of fruit treated with 2.5 g L−<sup>1</sup> was not significant (Figure 3b).

#### *3.8. Total Anthocyanin Content*

At all evaluation times, the samples treated with chitosan nanoemulsion had more anthocyanin content than the control samples, and among the treatments, 5 g L−<sup>1</sup> chitosan nanoemulsion was more effective than 2.5 g L−<sup>1</sup> in enhancing the anthocyanin content of the fruit. The amount of this parameter varied from 1.0355 g/kg in the samples treated with 5 g L−<sup>1</sup> chitosan nanoemulsion on the third day after storage to 0.3985 g/kg on the ninth day in the control samples (*p* < 0.01) (Figure 3c).

#### *3.9. Antioxidant Activity*

#### 3.9.1. DPPH Method

Changes in antioxidant capacity in the raspberry fruit treated with chitosan nanoemulsion are shown in Figure 4a. According to the results, this index showed an increasing trend in all treatments during storage so that the highest content (85.16%) was related to the fruit treated with 5 g L−<sup>1</sup> chitosan nanoemulsion after 9 d of storage, while the lowest (56.41%) was observed in fruit treated with the same concentration of chitosan nanoemulsion after 3 d of storage. Meanwhile, there was a statistically significant difference between the treated and control samples (*p* < 0.01).

**Figure 4.** Effect of chitosan nano-emulsion coatings (at 0 g L−<sup>1</sup> , 2.5 g L−<sup>1</sup> , and 5 g L−<sup>1</sup> ) on antioxidant activity (**a**) 2,2 diphenyl-1-picrylhydrazyl-hydrate (DPPH) and (**b**) Ferric reducing antioxidant power (FRAP) of raspberries stored for 9 d (at 4 ± 1 ◦C with 90–95% RH). Control refers to untreated raspberries. The data shown are the mean ± standard error of four replicates. Different letters indicate statistical significance (*p* < 0.01).

#### 3.9.2. FRAP Method

The results of the analysis of variance of the antioxidant activity data using the FRAP method showed that there was a statistically significant difference between the samples treated with chitosan nanoemulsion and the control samples at a 1% level. In other words, there was a significant difference between chitosan nanoemulsion levels and the control samples (*p* < 0.01). However, there was no statistically significant difference between the sixth and the ninth day in any of the treatments (chitosan nanoemulsion of 2.5 and 5 g L−<sup>1</sup> and the control). Furthermore, the antioxidant activity level varied from 0.11275 mol/L to 0.03238 mol/L (Figure 4b).

### *3.10. Anti-Stress and Antioxidant Enzymes Activity*

#### 3.10.1. PAL Enzyme Activity

PAL enzyme activity in all treatments, except for the 2.5 g L−<sup>1</sup> treatment with chitosan nanoemulsion on the third day, showed an increasing trend in different during storage (*p* < 0.01). Moreover, the highest enzyme activity (118.88 µmol/min *trans*-cinnamic acid) among the samples was related to the treatment of 5 g L−<sup>1</sup> chitosan nanoemulsion in the ninth day after storage, and the lowest (49.52 µmol/min *trans*-cinnamic acid) was observed for the treatment of 2.5 g L−<sup>1</sup> chitosan nanoemulsion on the third day after storage (Figure 5a).

**Figure 5.** Effect of chitosan nano-emulsion coatings (at 0 g L−<sup>1</sup> , 2.5 g L−<sup>1</sup> , and 5 g L−<sup>1</sup> ) on antioxidant activity: (**a**) phenylalanine ammonia-lyase (PAL) enzyme activity and (**b**) guaiacol peroxidase (G-POD) enzyme activity of raspberries stored for 9 d (at 4 ± 1 ◦C with 90–95% RH). Control refers to untreated raspberries. The data shown are the mean ± standard error of four replicates. Different letters indicate statistical significance (*p* < 0.01).

#### 3.10.2. G-POD Enzyme Activity

G-POD enzyme activity varied from 0.001 µmol H2O2.min−<sup>1</sup> .mL−<sup>1</sup> extract to 0.003 µmol H2O2.min−<sup>1</sup> .mL−<sup>1</sup> extract (*p* < 0.01) in different treatments and during storage. According to Figure 4b, treatment of fruit with 5 g L−<sup>1</sup> chitosan nanoemulsion was more effective in enhancing the G-POD enzyme activity during storage (*p* < 0.01) (Figure 5b).

#### **4. Discussion**

Acidity is an important parameter in maintaining fruit quality, which is directly related to the concentration of dominant organic acids in samples [40]. Organic acids are used by the respiration reactions to provide the necessary energy for the normal activities of the cells in storage time [41]. Therefore, the acid content and the pH of the fruit reflect the status of the fruit from the senescence point of view. According to the results of this study, the total acidity content showed a decreasing trend and the pH showed an increasing trend over time, which can be due to the consumption of the organic acid in the respiration process and their conversion to sugars [31]. ECNPC, as any other coating, reduces the respiration and ethylene production rates by restricting the gas exchange resulting in elevated CO<sup>2</sup> and decreased O2, thereby reducing the consumption of organic acids and preventing an increase in fruit pH [42].

Change in soluble solids depends on metabolic activities and the activity of different cell wall degrading enzymes. An increase in metabolic activities leads to enhanced ethylene production and a subsequent increase in the activity of degrading enzymes resulting in a dramatic increase in the TSS content of the fruit [43]. Therefore, any decrease in metabolic activities will decrease the ethylene biosynthesis and action resulting in the prevention of a dramatic increase in TSS content. By restricting the gas exchange and making a modified atmosphere in the fruit, chitosan coating decreases ethylene production and senescence rate and prevents cell wall degradation processes [44]. The positive effect of the ECNPC in retaining fruit firmness in this study demonstrates the effectiveness of this treatment in decreasing the activity of cell wall degrading enzymes probably by decreasing ethylene production and action. However, the effect of chitosan on soluble solids content is different for different types of fruit; for example, in papaya fruit, the coating did not have a significant effect on soluble solids but increased the soluble solids in mangos [45].

Weight loss of harvested crops is the consequence of water loss and the respiration process. With progress in senescence, the water of the fruit is lost due to transpiration; the weight loss is also related to the consumption of fruit carbohydrates and organic acid reserves. Water loss, in turn, accelerates senescence by enhancing the production of free radicals and ROS [46]. The amount of respiration and the consumption of sugar by the cell lead to more water loss, which is an important factor in the deterioration of the products [13]. Findings demonstrate that these films/coatings act as a barrier on the surface of fruit and vegetables, which causes higher moisture and water retention, creating favorable micro-environments by optimizing the concentration of gases and delaying the ripening process [47]. The role of chitosan-based edible coating is to restrict water vapor exchange between the fruit and the environment, decrease the respiration and metabolic activities, and activate the mechanisms of cuticle formation in the tissues. In fact, as an elicitor, chitosan can activate the PAL and polyphenol oxidase (PPO) enzymes and other enzymes playing roles in the biosynthesis of polyphenols and cuticle [48]. It is well demonstrated by researchers that these additives play an important role in extending the shelf life and maintaining the nutrient profile of numerous fruit and vegetables. These functional molecules play a synergistic role along with the chitosan and alginate-based edible coatings and retain the moisture and antioxidant potential, enhance the activity of antioxidant enzymes, reduce the activity of browning enzymes, and impart antimicrobial properties in fresh fruit and vegetables. All these properties help in sustaining the appearance, lowering the senescence, and extending the life of the coated fruit and vegetables [47]. In this study, the effects of chitosan nanoparticle-containing coating on enhancing PAL, glutathione peroxidase (GPOX), and different antioxidant fractions were demonstrated. Our results show that chitosan-based coating acts as a physical barrier for the gases and water vapor [49] and enhances the anti-stress and antioxidant properties of the fruit.

The role of chitosan nanoparticles in decreasing decay extension is related both to the direct antifungal activity of the chitosan and the activation of different resistance mechanisms and anti-stress systems [50]. The quality of fruit is defined by its characteristics, such as shape, size, color, and lack of defects, including cuts and decay [42]. The appearance of the product is the most important index affecting the marketability and the presence of any signs of contamination and decay and softening of fruit reduces the market demand of the product. The decay organisms can extend and grow on the surface of shrunken, wilted, injured, and softened fruit. Therefore, reducing the senescence and deterioration rate prevents the growth of decay symptoms and maintains the appearance and market demand of the product [51]. Our results indicate that ECNPC can decrease decay extension in raspberries during storage resulting in the safety and postharvest life enhancement. The results of this study showed that with an increase in the concentration of chitosan nanoemulsion, the decay extension rate was better controlled and fruit treated with 5 g L−<sup>1</sup> chitosan nanoemulsion showed the lowest decays.

Chitosan acts as an active substance at the surface of fungal and bacterial cells, which makes them more permeable [52]. This interaction is mainly assumed to be electrostatic and occurs among the positive loads of chitosan amino acids and negative loads on the cell surface of microorganisms [53]. In general, the permeation of the cell surface causes the leakage of intracellular material and thereby causing cell death [54].

Phenolic compounds, including different polyphenols, flavonoids, and anthocyanins, are powerful antioxidants and antipathogen compounds, and a coating containing chi-

tosan nanoparticles is able to restrict decay extension directly as a natural fungicide and indirectly as a resistance mediating elicitor [55]. Phenolic compounds are one of the most important secondary metabolites and are chemically very diverse; their biosynthesis is initiated with the phenylalanine, tyrosine, and tryptophan amino acids [56]. Anthocyanins also play the most important role in raspberry fruit color and overall quality. Phenolic compounds of raspberries inhibit liposome oxidation in the body. These compounds also have significantly shown a high capacity to eliminate singlet oxygen (free radical) or act as a hydrogen supplier [57]. The powerful antioxidative and anticancer activities of raspberries phenolic antioxidants in the human body have been well demonstrated [58]. After harvest, the concentrations of phenolic and flavonoids compounds are either substantially fixed or decreased [59]. As a novel and interesting finding, in this study, we found that chitosan nanoemulsion coating substantially enhances the biosynthesis and accumulation of different phenolic contents including total phenolics, flavonoids, and anthocyanin contents of the raspberry fruit during storage, resulting in an enhanced health-promoting capacity of the fruit. Moreover, these compounds, by detoxifying the free radicals and ROS, act as anti-stress and anti-senescence agents in the fruit resulting in enhanced postharvest life and quality maintenance.

PAL plays a key role in different phenolic biosynthesis pathways called the phenylpropanoid pathway [60]. It has been demonstrated that PAL activity decreases during the maturity and postharvest stages [61]. In addition, with the increase in the activity of the PPO activity during the senescence process, the consumption of polyphenols increases, and as a result, the content of total phenolic compounds decreases with the aging of the fruit tissue [62]. The results of this study show that the treatment enhanced the activity of PAL, resulting in the production of different phenolic compounds. On the other hand, edible coatings such as chitosan, by protecting the surface of the products, have been shown to reduce oxygen and thereby reducing the oxidation of phenolics [63].

Guaiacol peroxidase is one of the important antioxidant enzymes that were increased by an edible coating containing nanoparticles of chitosan in this study. This enzyme has been reported to show high activity in biological systems against hydrogen peroxide. Hydrogen peroxide is an important ROS-attacking cell membrane, and the increase in GPOX activity is pivotal for protecting the fruit cell membranes against peroxidation of lipids and DNA hydroperoxides [64].

Different enzymatic antioxidants are the cells' weapons against the oxidation of biological molecules such as lipids, proteins, carbohydrates, and deoxyribonucleic acid by free radicals and ROS [65,66]. The free radicals and ROS are produced during normal cell metabolism and attach the biomolecules of the cells called oxidative damage. To prevent oxidative stress, plant cells activate their antioxidant systems and respond to oxidative damage through the activation of different antioxidant systems [67]. Our results indicate that the antioxidant capacity, measured with both DPPH and FRAP methods, increases in the fruit treated with 5 g L−<sup>1</sup> chitosan nanoemulsion. The use of edible coatings increases the capacity of the fruit antioxidant system and protects the cells against oxidative stress and pathogen attack [68,69].

#### **5. Conclusions**

In summary, our findings indicate that the use of chitosan nanoparticles can enhance different antioxidant fractions and the total antioxidant activity of raspberry fruit. Chitosan nanoemulsion treatments effectively prevented from the substantial loss in fruit taste and flavor during the storage period, while the control fruit exhibited off-flavors and bad taste. The taste and aroma during storage in the warehouse are significantly reduced due to increased respiration and enzymatic activity of the fruit. In the present study, the fruit treated with chitosan nanoemulsion was generally considered to be better for the sensory quality due to reduced respiration, metabolic activities, decay extension, weight loss and softening, and enhanced phytochemical contents. Competency of edible film/coatings and

nanoforms in extending the shelf life without affecting the nutritional properties and safety aspects of fruit and vegetables still requires further attention.

**Author Contributions:** Conceptualization, H.S. and A.A.; writing—original draft preparation, S.R.I. and M.R.A.; writing—review and editing, H.S., A.A., M.P., and J.M.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding authors.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Thermodynamic and Quality Performance Studies for Drying Kiwi in Hybrid Hot Air-Infrared Drying with Ultrasound Pretreatment**

**Ebrahim Taghinezhad 1,\* , Mohammad Kaveh <sup>2</sup> and Antoni Szumny <sup>3</sup>**


**Abstract:** The present study examined the effect of ultrasonic pretreatment at three time the levels of 10, 20 and 30 min on some thermodynamic (effective moisture diffusivity coefficient(*Deff*), drying time, specific energy consumption (*SEC*), energy efficiency, drying efficiency, and thermal efficiency) and physical (color and shrinkage) properties of kiwifruit under hybrid hot air-infrared(*HAI*) dryer at different temperatures (50, 60 and 70 ◦C) and different thicknesses (4, 6 and 8 mm). A total of 11 mathematical models were applied to represent the moisture ratio (*MR)* during the drying of kiwifruit. The fitting of *MR* mathematical models to experimental data demonstrated that the logistic model can satisfactorily describe the *MR* curve of dried kiwifruit with a correlation coefficient (*R* 2 ) of 0.9997, root mean square error (RMSE) of 0.0177 and chi-square (*χ* 2 ) of 0.0007. The observed *Deff* of dried samples ranged from 3.09 <sup>×</sup> <sup>10</sup>−<sup>10</sup> to 2.26 <sup>×</sup> <sup>10</sup>−<sup>9</sup> <sup>m</sup>2/s. The lowest *SEC*, color changes and shrinkage were obtained as 36.57 kWh/kg, 13.29 and 25.25%, respectively. The highest drying efficiency, energy efficiency, and thermal efficiency were determined as 11.09%, 7.69% and 10.58%, respectively. The results revealed that increasing the temperature and ultrasonic pretreatment time and decreasing the sample thickness led to a significant increase (*p* < 0.05) in drying efficiency, thermal efficiency, and energy efficiency, while drying time, *SEC* and shrinkage significantly decreased (*p* < 0.05).

**Keywords:** dry; efficiency; energy; kiwifruit; quality; ultrasound

#### **1. Introduction**

Kiwifruit with the scientific name of *Actinidia deliciosa* is a subtropical fruit belonging to the Actinidiaceae family. This fruit is rich in vitamins E and C, phenolic substances, flavonoids, various pigments, and antioxidants [1]. According to the research by the Food and Agriculture Organization (FAO) of the United Nations, kiwifruit is the fourth most popular fruit in the world after banana, citrus fruits, and apple [2].

Fruits and vegetables with a moisture content (*MC*) of about 80% are classified as highly perishable materials. The high-potential drying is one of the most common methods to preserve and increase the productivity of food [3]. The most common method for drying food is to use hot air flow, which requires hot air to provide a simultaneous transfer of heat and moisture between hot air and food to evaporate water and to bring the moisture to the desired level [4]. In addition to having advantages, such as the possibility of accurate temperature control regardless of the size and shape of the product and not needing direct contact, the drying with hot air also has some disadvantages. These disadvantages include the long time and high temperature for drying the product during the period of descending speed. In addition, high temperatures reduce the nutritional value and increase energy consumption [5]. The infrared dryers have been commonly used in recent years and have

**Citation:** Taghinezhad, E.; Kaveh, M.; Szumny, A. Thermodynamic and Quality Performance Studies for Drying Kiwi in Hybrid Hot Air-Infrared Drying with Ultrasound Pretreatment. *Appl. Sci.* **2021**, *11*, 1297. https://doi.org/10.3390/ app11031297

Academic Editor: Theodoros Varzakas Received: 12 January 2021 Accepted: 27 January 2021 Published: 1 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

received more attention in developing countries due to the low cost, simple equipment, and low price. Infrared drying is the process in which wet materials are heated by infrared radiation. Infrared radiation is a type of electromagnetic radiation with a wavelength between 0.76 and 400 µm [6]. In the hybrid method, the use of hot air and infrared waves coincide in all stages of product drying. The reports show that the hybrid method at mild air speeds can improve energy consumption and drying time while increasing product quality [7]. The application of the pretreatments that reduce the initial moisture of the food and also modify the food texture to accelerate the transfer of moisture and reduce drying time will be very useful.

Ultrasonic pretreatment is performed by immersing the fruit in water and applying ultrasound waves. Ultrasonic waves cause a series of rapid, intermittent contractions and expansions of the cell wall (sponge effect), which creates microscopic channels in the structure of the material and increases the possibility of moisture transfer. As a result, an ultrasound can be used as a pretreatment in drying heat-sensitive food products without adversely affecting the qualitative characteristics of the food, since it improves the speed and decreases the temperature required for drying [8].

Mierzwa et al. examined the effect of ultrasonic pretreatment with two types of microwaves and hot air dryers on some drying parameters (kinetics, shrinkage, specific energy consumption, and rehydration ratio) of raspberry, and stated that the use of ultrasonic pretreatment reduced specific energy consumption and shrinkage in both types of dryers [9]. Abbaspour et al. investigated the drying kinetics, *SEC*, color, shrinkage, and rehydration of carrots using three types of hot air, microwave and infrared dryers with ultrasonic pretreatment [10]. They reported that the use of ultrasonic pretreatment reduced the drying time of the sample compared to the control treatment. The use of ultrasonic pretreatment in all three types of hot air, microwave and infrared dryers also increased the *Deff* and reduced the *SEC*, color changes, and shrinkage. Dehghannya et al. studied the effect of using ultrasonic pretreatment on the drying of potato slices in hot air microwave dryers and measured the *MC*, *Deff*, *SEC*, rehydration percentage, and shrinkage [11]. They stated that the application of ultrasonic waves reduced the drying time. Additionally, the samples under ultrasonic pretreatment had lower shrinkage and *SEC* and a higher moisture diffusivity coefficient and rehydration percentage than the control sample. Ghanbarian et al. dried peppermint using the hot air dryer (at three temperature levels) with ultrasonic pretreatment (four levels of ultrasonic power) [12]. They examined *MC*, drying time, *SEC*, and energy efficiency of peppermint. The results showed that the use of ultrasonic pretreatment reduced the drying time and *SEC* and increased the energy efficiency of peppermint. Thus far, various studies have been conducted on the drying of kiwifruit using hot air dryer [13,14], hot air dryer with recirculation under heat pump [15], microwave [16], infrared-vacuum dryer [17], vacuum-hot air dryer [18], hot air, vacuum, freezer, hot airmicrowave-vacuum [19], hot air, microwave and freezer [20], *HAI* [21], freezer [22], and short and medium-wave infrared radiation with/without osmotic dehydration [23].

Due to the importance of kiwifruit in the human diet and the effect of the drying process on the maintenance of this product with high qualitative properties and low energy consumption, and given the lack of a comprehensive study on the effect of ultrasound on the qualitative and thermodynamic properties of kiwifruit, this research aimed to evaluate the effect of ultrasonic pretreatment on the changes in *Deff*, *SEC*, energy efficiency, drying efficiency, color, shrinkage, and drying time of kiwifruit slices during drying with the hybrid *HAI* dryer.

#### **2. Materials and Methods**

Fresh Hayward kiwifruit was bought from the local market of ParsAbad, Ardabil province, Iran and stored at 4 ◦C. One hour before the test, the samples were removed from the refrigerator and kept at 24 ◦C. Using an electric slicer (Celme GE300, Montebello Vicentino, Italy), kiwifruits were cut into 4, 6 and 8 mm thick slices and the slices were then cut into 24 mm discs by a circular slicer. The initial *MC* of kiwifruit samples was obtained in an oven at 105 ◦C for 24 h. The initial *MC* of kiwifruit (6.85% (d.b.)) was determined and the drying continued until reaching an *MC* of 0.2% (d.b.).

#### *2.1. Ultrasonic*

The kiwifruit samples selected for ultrasonic pretreatment were transferred to an ultrasonic bath. All samples were placed in a 6 Lultrasonic bath (Parsonic 7500S, Tehran, Iran), which was filled with distilled water at an ambient temperature of 25 ◦C in a ratio of 1:4. The samples were then exposed to ultrasonic waves for 0, 10, 20 and 30 min. The frequency and power of this test were 28 kHz and 70 W, respectively.

#### *2.2. Hybrid Hot Air-Infrared (HAI) Dryer*

A hybrid *HAI* dryer (GC 400, Grouc, Tehran, Iran) was used for the experiments. The dryer consisted of two infrared lamps (Philips, Flemish, Belgium), each with a rated power of 250 W (500 W in total), which was installed in the upper part of the drying chamber at a height of 30 cm. The flow rate of the air entering the dryer was set to 1 m/s. The kiwifruits under study were placed in the middle of the channel on the containers made from wire mesh on digital scales (AND, GF–6000, A&D Company Ltd, Tokyo, Japan) with a precision of 0.01 g embedded below and outside the channel. The drying of kiwifruit was performed at three air temperature levels of 50, 60 and 70 ◦C. To start the test, the machine was put into operation for 15 min to fix the temperature and flow rate of the dryer's air. In each experiment, 55 g of kiwifruit was placed in a single layer on a drying tray. During the drying experiments, the mean range of changes in ambient temperature and the relative humidity were 20 ± 4 ◦C and 15 ± 5%, respectively.

#### *2.3. Moisture Ratio (MR) and Mathematical Modeling*

To model the drying process, the *MR* parameter during the kiwifruit drying was obtained using the following Equation (1) [24]:

$$MR = \frac{M\_t - M\_\varepsilon}{M\_o - M\_\varepsilon} \tag{1}$$

*MR*: moisture ratio (dimensionless), *M<sup>t</sup>* : *MC* of samples at any given time on dry basis (d.b.), *M<sup>e</sup>* : equilibrium moisture content (d.b.), and *Mo*: initial moisture of samples (d.b.). According to Equation (1), the moisture ratio depends on the initial moisture, balanced moisture, and moisture of samples at any time during the drying. For long drying times, the *M<sup>e</sup>* values are very small compared to the *M<sup>o</sup>* and *M<sup>t</sup>* values. Therefore, the equation of *MR* during the drying can be simplified, as in Equation (2), and to calculate the *MR*, it is not necessary to measure the balanced moisture [25].

$$MR = \frac{M\_t}{M\_o} \tag{2}$$

#### *2.4. Modeling*

In this study, a number of experimental and proposed models were used to model the drying kinetics (fitting of *MR* versus drying time) of green kiwifruit samples (Table 1). To select the most appropriate model to describe the drying kinetics of green kiwifruit, the criteria of determination coefficient (*R 2* ), root mean square error (*RMSE*), and chisquare (*χ* 2 ) were calculated by each model and compared with other models. The higher the *R <sup>2</sup>* value and the lower the *RMSE* and *χ* <sup>2</sup> values, the better the model fitting to the experimental data and it is chosen as the best model. The *R 2* , *RMSE* and *χ* <sup>2</sup> values were calculated according to Equations (3)–(5):

$$\begin{aligned} R^2 &= \frac{\sum\_{i=1}^{N} (MR\_{\text{exp},i} - MR\_{pre,i})^2}{\sum\_{i=1}^{N} \left(\frac{\sum\_{i=1}^{N} MR\_{pre,i}}{N} - MR\_{pre,i}\right)^2} \\ &\quad \times^2 = \frac{\sum\_{i=1}^{N} (MR\_{\text{exp},i} - MR\_{pre,i})^2}{N - z} \end{aligned} \tag{3}$$

$$RMSE = \left[\frac{1}{N} \sum\_{i=1}^{N} \left(MR\_{pre,i} - MR\_{exp,i}\right)^2\right]^{\frac{1}{2}} \tag{5}$$

*N*, *z*, *MRpre,i* and *MR*exp,i are the number of observations, number of dryer constants, predicted *MR*<sup>i</sup> , and experimental *MR*<sup>i</sup> , respectively [20].

**Table 1.** Applied models to fit the experimental data *MR* = 1 + *at* + *bt*<sup>2</sup> .


*a*, *b*, *c*, *k*, *k*0, *k*<sup>1</sup> and *n* are drying constants.

#### *2.5. Determination of Effective Moisture Diffusivity Coefficient (Deff)*

The number of moisture transfer mechanisms is extensive and often complex. Transfer phenomena are usually classified into pressure diffusion, forced diffusion, and ordinary diffusion (net transfer of matter without fluid motion). According to Equation (6), the Fick's second law for unstable conditions can describe the transfer of moisture in the descending stage of the drying process [29,30]:

$$\frac{\partial X}{\partial t} = D\_{eff} \frac{\partial^2 X}{\partial^2 x} \tag{6}$$

where *X* is the local *MC* on dry basis, *t* is the time and *x* is the spatial index. The study of the Fick's second law of diffusion indicates the mass transfer during the period of the declining drying rate of agricultural products [26]. To apply Fick's law, it is assumed that the food product is one-dimensional, the initial *MC* is uniform, and it has an internal movement of moisture such as the major resistance to moisture transfer. The Fick's equation for a plane sheet is solved as follows [31,32]:

$$MR = \frac{8}{\pi^2} \sum\_{n=1}^{\infty} \frac{1}{\left(2n+1\right)} \exp\left(\frac{-D\_{eff}\left(2n+1\right)^2 \pi^2 t}{4L^2}\right) \tag{7}$$

From 1 to infinity, *t* is the drying time (s), and *Deff* is the effective diffusivity coefficient (m2/s). The effective diffusivity coefficient is obtained by calculating the slope in Equation (8) [11]:

$$MR = \frac{8}{\pi^2} \exp\left(\frac{-D\_{eff}\pi^2 t}{4L^2}\right) \tag{8}$$

The diffusivity coefficient is usually determined by plotting the experimental drying data versus LnMR over time. When plotting the LnMR value over time, place the resulting line slope in Equation (9) to obtain the *Deff*:

$$K = \left(\frac{D\_{eff}\pi^2}{4L^2}\right) \tag{9}$$

where K is the line slope.

#### *2.6. Calculation of Some Thermodynamic Parameters*

#### 2.6.1. Energy Consumption

The thermal energy consumed by the hybrid *HAI* dryer for drying kiwifruit at infrared temperature and power at constant air flow rate is obtained using Equation (10) [33]:

$$E\mathcal{U}\_{ter} = A.v.\rho\_a.\mathcal{C}\_a.\Delta t.t + K.t\tag{10}$$

where *EUter* is the total thermal energy consumption per drying period (kWh), *A* is the area of the container in which the test sample is placed (m<sup>2</sup> ), *v* is the wind speed (m/s), *ρ<sup>a</sup>* is the air density (kg/m<sup>3</sup> ), *t* is the total drying time of each sample (h), ∆*T* is the temperature difference (◦C), and *C<sup>a</sup>* is the specific heat (KJ/kg ◦C).

The mechanical energy consumed by the blower in the dryer during the drying process is obtained from Equation (11) [34]:

$$E\mathcal{U}\_{\rm mec} = \Delta P.\mathcal{W}\_{\rm air}.t\tag{11}$$

*EUmec* is the total mechanical energy consumption per period (kJ), ∆*P* is the pressure drop (Pa), and *Wair* is the airflow rate (m3/s).

The energy consumption for ultrasonic pretreatment can be obtained using the following equation [10]:

$$EL\_{ult} = \mathcal{U}I \cos\Phi \tag{12}$$

where *EUult* is the ultrasonic power (kw), *U* (V) and *I* (A) are the input voltage and current in the ultrasonic machine, respectively, and cosΦ is the power factor equal to 0.8.

The energy consumption required for 1 kg of moisture to be removed from the drying product is calculated by Equation (13):

$$\text{SEC}\_{total} = \frac{\text{EL}\_{(mec+ter+ult)}}{\text{M}\_{\text{W}}} \tag{13}$$

where *SEC* is the specific energy consumption (kWh/kg), *EU*(mec+ter+ult) is the total energy consumption during the drying process (kWh), and *M<sup>w</sup>* is the amount of moisture removed from the product (kg).

#### 2.6.2. Drying Efficiency

Drying efficiency is the division of the total energy required to heat the drying product and the energy required to evaporate moisture by the total energy consumption during the drying process (*De*):

$$D\_{\varepsilon} = \left(\frac{E\_{evap} + E\_{heating}}{SEC}\right) \tag{14}$$

where *D<sup>e</sup>* is the drying efficiency (%), *Eheating* is the energy required to increase the temperature of the dried product (kJ), *Eevap* is the energy needed to evaporate moisture (kJ), and *SEC* is the total energy consumption (kJ).

#### 2.6.3. Energy Efficiency

Energy efficiency is obtained by dividing the energy required to evaporate moisture from the drying product by the total energy *SEC* during the drying process.

$$
\eta\_{\varepsilon} = \left(\frac{E\_{evap}}{SEC}\right) \times 100\tag{15}
$$

where *η<sup>e</sup>* is the energy efficiency and *Eeva* is the energy required to evaporate moisture (kJ).

#### 2.6.4. Thermal Efficiency

Thermal efficiency is the division of the moisture evaporated from the product by the heat used to dry the product, which can be calculated from Equation (16):

$$TE = \frac{M\_{evap}}{H\_{total}}\tag{16}$$

where *TE* is the thermal efficiency (kg/kJ), *Mevap* is the weight of moisture evaporated from the product (kg), and *Htotal* is the total heat consumption (kg).

#### 2.6.5. Energy Required to Evaporate Moisture from Product

The energy required to evaporate moisture from the product during the kiwifruit drying process in the hybrid *HAI* dryer with ultrasonic pretreatment is calculated from Equation (17):

$$Q\_w = h\_{f,g} M\_w \tag{17}$$

where *Q<sup>w</sup>* is the energy required to evaporate moisture (kJ), *hfg* is the latent heat of vaporization (kJ/kg), and *M<sup>w</sup>* is the moisture evaporated from the product (kg).

The latent heat of vaporization can be obtained from Equation (18):

$$\begin{aligned} h\_{f.g} &= \left(7.33 \times 10^6 - 16T\_{\text{abs}}^2\right)^{0.5} \\ 273.16 &\le T\_{\text{abs}} \le 338.72 \\ h\_{f.g} &= 2.503 \times 10^3 - 2.386(T\_{\text{abs}} - 273.16) \\ 337.72 &\le T\_{\text{abs}} \le 533.16 \end{aligned} \tag{18}$$

#### 2.6.6. Energy Required to Increase Product Temperature

The energy required to increase the product temperature from the initial temperature (temperature before entering the drying chamber) to the latent temperature (highest product temperature) can be calculated from Equations (19)–(21):

$$E\_{\text{heating}} = \mathcal{W}\_d \mathbb{C}\_m (T\_2 - T\_1) \tag{19}$$

$$\mathcal{C}\_{m} = 1465 + 3560(\frac{M\_P}{1 + M\_P}) \tag{20}$$

$$M\_P = \frac{\mathcal{W}\_w - \mathcal{W}\_d}{\mathcal{W}\_d} \tag{21}$$

where *Eheating* is the energy required to increase the product temperature (kJ), *W<sup>d</sup>* is the dry matter weight (kg), *C<sup>m</sup>* is the specific heat of the product (kJ/kg K), *T*<sup>1</sup> is the initial product temperature (K), *T*<sup>2</sup> is the latent product temperature (K), *M<sup>p</sup>* is the *MC* of the product on dry basis (kg water/kg solid), and *M<sup>w</sup>* is the initial weight of the product (kg).

#### *2.7. Shrinkage*

Shrinkage is the ratio between the initial volume of fresh kiwifruit and the final volume of dried kiwifruit, which is expressed by the following equation:

$$S\_v = 1 - \frac{V\_t}{V\_o} \times 100\tag{22}$$

where *S<sup>v</sup>* is the percentage of shrinkage (dimensionless), *V<sup>t</sup>* is the volume at the desired time (cm<sup>3</sup> ) and *V* is the initial sample volume (cm<sup>3</sup> ). The sample volume is obtained by the toluene displacement method according to Dehghannya et al. [11].

#### *2.8. Color*

The color of the samples prior to and following the drying was measured by a color meter (Portable colorimeter, HP–200, China). To describe the color changes during the drying, the ∆*E* index (color difference of total dried samples from fresh samples) was used. This index is defined by Equation (8) where *L* represents the brightness, *b* represents the yellow-blue color and *a* indicates the red-green color [25,35]:

$$
\Delta E = \sqrt{\left(L^\* - L\_0^\*\right)^2 + \left(a^\* - a\_0^\*\right)^2 + \left(b^\* - b\_0^\*\right)^2} \tag{23}
$$

#### *2.9. Data Analysis*

In this study, the factorial experiment was used in a completely randomized design. The treatments were considered at three drying temperatures (50, 60 and 70 ◦C), three levels of thickness (4, 6 and 8 mm) and three levels of ultrasonic radiation time (10, 20 and 30 min). Additionally, the effect of treatments on the qualitative (color changes and shrinkage) and thermodynamic parameters (*Deff*, *SEC*, thermal efficiency, energy efficiency, and drying efficiency) was investigated for the drying of kiwifruit slices. To examine the significance of the factors, the data analysis of variance was performed using SPSS software and the mean comparison of treatments was made at the 5% probability level.

#### **3. Results and Discussion**

#### *3.1. Mathematical Modeling*

Table 2 shows the statistical parameters obtained from the fitting of the experimental data of the drying experiments to the 11 models listed in Table 1. According to the tables, it can be seen that, in the method of hybrid *HAI* drying with ultrasonic pretreatment, the logistic model has the highest value of determination coefficient (0.9997) and the lowest values of chi-square (*χ* <sup>2</sup> = 0.0007) and root mean square error (*RMSE* = 0.0177) compared to other models.



#### *3.2. Drying Time*

Figure 1 shows the effect of drying air temperature, thickness and ultrasonic pretreatment time on the drying time of kiwifruit slices. As shown in Figure 1, the maximum drying

time was obtained at 50 ◦C with a thickness of 8 mm (control sample) for 420 min. Also, the minimum drying time was obtained at 70 ◦C, 4 mm thickness and ultrasonic radiation time of 30 min for 70 min. As expected, the drying time is significantly reduced (*p* < 0.05) by raising the temperature due to the increased thermal gradient (as a result of higher mass and heat transfer) and greater and quicker removal of moisture from the product [36].

**Figure 1.** Drying time under different conditions for drying kiwifruit slices.

– – The results obtained in Figure 1 also show that the highest effect of using ultrasonic pretreatment on the drying time of kiwifruit is seen for 30 min of ultrasonic application and the lowest one is seen for the control sample. By applying the cavitation process (asymmetric bursting of bubbles) near the foodsurface, ultrasonic pretreatment causes the rapid and eruptive flow of sound waves to the surface and forms microscopic channels in the samples by creating successive contractions and expansions. In addition, by increasing the duration of applying this energy (ultrasound), the channels expands and the product has a spongy texture, facilitating the flow of water out of the product during the drying through the created channels [37]. Similar results were observed for carrot drying [24], banana drying [31], and blackberry drying [29]. They showed that, by increasing the ultrasonic radiation time, the drying time decreases significantly (*p* < 0.05). As shown in Figure 1, the drying time was longer as the thickness of the samples increased. The reason for this issue is the hardening mechanism of the cutting surface, increased amount of water, and also a longer path of vaporremoval, which made it difficult to transfer moisture through the texture and ultimately increased the drying time [38]. Doymaz and Ozdemir for drying tomatoes and Onwude et al. for drying potatoes showed that the drying time increases by enlarging the thickness of the samples [28,36].

#### *3.3. Deff*

–

Different values of *Deff* for different treatments of this study are reported in Figure 2. By increasing the drying temperature, the sample thickness and ultrasonic radiation time during the kiwifruit drying process *Deff* had a significantly (*p* < 0.05) increasing trend. By increasing the air temperature from 50 to 70 ◦C, it was observed that the moisture diffusivity coefficient increased from 3.09 × 10 −10 to 2.26 × 10 <sup>−</sup><sup>9</sup> m2/s. The reason for this issue is the significant effect of air temperature in creating more molecular movement and surface suction, and increasing the diffusivity coefficient. Additionally, increasing the temperature increases the enthalpy of the entering air, and increasing the enthalpy increases

– –

–

the mass and heat transfer, which increases *Deff* . Further, by increasing the thickness of the samples, the amount of *Deff* increased. Nguyen and Price reported that superficial hardening generally occurs much faster in thinner samples than in thicker samples, while the rate of moisture evaporation in thinner samples is much higher [39]. Therefore, rapid superficial hardening in the samples with less thickness causes a restriction in moisture transfer compared to thicker samples and thus leads to a decrease in moisture diffusivity coefficient in thinner samples [40,41]. Similar observations on the effect of thickness on moisture diffusivity coefficient have been reported by other researchers, including Doymaz and Gol, Azimi-Nejadian and Hoseini and Onwude et al. [28,40,42]. The *Deff* increased by prolonging the ultrasonic pretreatment time from 0 to 30 min. Ultrasonic pretreatment causes the opening of the capillary tubes due to the dispersion of surface compounds and the formation of longer microscopic channels due to the deformation of the cells, followed by the formation ofmore open capillary tubes [43]. Therefore, ultrasonic pretreatment accelerates the removal of moisture from the product by deforming the cells and destroying the cell wall. – –– –

**Figure 2.** The value of *Deff* under different conditions for drying kiwifruit slices.

Many researchers have reported *Deff* for other products with ultrasonic pretreatment. For instance, the obtained moisture diffusivity coefficient values for potatoes in the hybrid microwave-hot air dryer with ultrasonic pretreatment ranged from 1.07 × 10 −7 to 1.283 × 10 <sup>−</sup><sup>7</sup> m2/s [11]. Vallespir et al. obtained the *Deff* for kiwifruit in a hot air dryer with ultrasonic pretreatment ranging from 3.67 × 10 −11 to 9.45 × 10 <sup>−</sup><sup>11</sup> m2/s [5]. Abbaspour-Gilandeh et al. calculated the diffusivity coefficient for the drying of walnut in the microwave and hot air dryers at different ultrasonic pretreatment times between 3.12 × 10 <sup>−</sup><sup>9</sup> and 8.99 <sup>×</sup> <sup>10</sup> <sup>−</sup><sup>9</sup> <sup>m</sup>2/s and between 2.77 <sup>×</sup> <sup>10</sup> <sup>−</sup><sup>9</sup> and 5.56 <sup>×</sup> <sup>10</sup> <sup>−</sup><sup>9</sup> m2/s [43].

#### *3.4. Thermodynamic Parameters*

#### 3.4.1. *SEC*

According to Figure 3, the lowest *SEC* (36.57 kWh/kg) was obtained with a significant difference (*p* < 0.05) at an air temperature of 70 ◦C, ultrasonic radiation time of 30 min, and thickness of 4 mm. Additionally, the control sample with a thickness of 8 mm for drying kiwifruit significantly (*p* < 0.05) had the highest *SEC* (181.97 kWh/kg). This can be due to the low moisture removal from the drying product and the increase in drying time. As the air temperature grows from 50 to 70 ◦C, the thermal gradient within the drying

product increases, resulting in an increase in the moisture evaporation rate (the downward slope of the *MR* variation diagram becomes steeper as the temperature increases). Therefore, with the increasing temperature, the total energy consumption decreases [44]. Further, by enhancing the drying temperature from 50 to 70 ◦C, the difference between the drying temperature and the ambient temperature increased. As a result, increasing the temperature difference significantly decreased the drying time of the product, which is a strong reason for the reduced energy consumption. Motevali et al. and Kaveh et al., for drying *Roman chamomile* and *Pistacia Atlantica*, respectively, showed decreased energy consumption with increasing temperatures [33,44].

Increasing the ultrasonic radiation time from 0 (control sample) to 30 min significantly decreased the *SEC* (*p* < 0.05), because with increasing the ultrasonic pretreatment time, the product texture is more destroyed and the hard layer does not form in the product during the drying process, and thus, the product dries faster, and consequently, the *SEC* in the dryer is reduced [45]. The researchers showed that, with an increasing ultrasonic radiation time, the *SEC* decreases during the drying of blackberries [29].

The study of the effect of thickness on energy consumption revealed that reducing the thickness during the drying of kiwi reduces the drying time due to the increased thermal gradient and accelerated the removal of moisture from the product, thus reducing the total energy consumption. Darvishi et al., for drying kiwifruit in different thicknesses using the microwave dryer, showed that by reducing the thickness of the samples, the *SEC* is significantly reduced (*p* < 0.05) [16].

#### 3.4.2. Energy Efficiency and Drying Efficiency

–

Figures 4 and 5 show the effect of different treatments (temperature, thickness and ultrasonic radiation time) on energy efficiency and drying efficiency, respectively. The highest energy efficiency and drying efficiency values (7.69% and 11.09%) were obtained at 70 ◦C, 4 mm thickness, and 30 min ultrasonic radiation time, respectively. Increasing the temperature from 50 to 70 ◦C increases the energy efficiency and the drying efficiency. Drying temperature increases the rate of moisture removal from the product and reduces the drying time, and consequently, both efficiency values will increase with increasing temperatures. However, comparing the application times of ultrasonic pretreatment shows that the highest values of both efficiencies occur when the 30 min ultrasonic pretreatment

–

is used. According to Figures 4 and 5, it can be seen that reducing the thickness of kiwifruit samples will significantly (*p* < 0.05) increase energy efficiency and drying efficiency, because less thickness will increase the gradient of the drying product and more moisture is removed from the product, and as a result, the drying time of kiwifruit at low thicknesses is reduced. The lowest energy efficiency and drying efficiency values can be seen in the control treatment with a thickness of 8 mm (1.54% and 1.88%), respectively. The results show that increasing the ultrasonic pretreatment time improves both energy efficiency and drying efficiency. This is because, by increasing the ultrasonic pretreatment time, the changes in the product texture increases and the hard layer does not form in the process of drying this pretreated product and the product dries faster.

**Figure 4.** Energy efficiency under different conditions for kiwifruit slices drying.

**Figure 5.** Drying efficiency under different conditions for drying kiwifruit slices.

The results of this study are consistent with the results published by other researchers. The energy efficiency values for drying chamomile in hot air dryers ranged between 1.91% and 6.76% [44], for apple slices between 2.87% and 9.11% [46], for *Pistacia atlantica* in a hot air dryer equal to 5.65% [47], and for peppermint leaves in a hot air dryer with ultrasonic pretreatment varying between 1.41% and 3.69% [12]. Albini et al. obtained the drying efficiency and energy efficiency for barley drying in a hot air dryer ranging 8%–17.4% and 24%–34.2%, respectively [48]. In another study, the drying efficiency was calculated between 3.42% and 12.29% for apple drying and showed that the drying efficiency increases by increasing the temperature [46].

#### 3.4.3. Thermal Efficiency

Figure 6 depicts the effect of different treatments on thermal efficiency. It shows that the highest amount of thermal efficiency (10.58%) was achieved at 70 ◦C, ultrasonic radiation time of 30 min, and thickness of 4 mm, while the lowest value of this efficiency (1.54%) was obtained at 50 ◦C, control sample, and thickness of 8 mm. The results show that, with increasing temperatures, the changes in thermal efficiency have a significantly (*p* < 0.05) upward trend. The increase in thermal efficiency is due to the fact that, with increasing temperatures, the difference between the temperature of the drying product and the temperature of the drying air increases and leads to the accelerated removal of moisture from the product and thus reduces the time and amount of energy required. – – –

**Figure 6.** Thermal efficiency under different conditions for drying kiwifruit slices.

Thermal efficiency had an upward trend when using ultrasonic pretreatment. In the case of ultrasonic pretreatment, due to the reduced relative humidity of the drying air, the drying time is significantly (*p* < 0.05) shorter, and as a result, the share of energy consumption is smaller. The effect of thickness on the changes in thermal efficiency shows that, with decreasing thicknesses, the amount of thermal efficiency increases significantly (*p* < 0.05). This can be due to the increased drying rate as a result of the decreased thickness, because with decreasing thicknesses, the thermal gradient increases, the removal of moisture from within the product to the surface increases, and the drying time decreases.

Torki-Harchegani et al. showed that the thermal efficiency of peppermint leaf drying was 38.36%–49.33% for the microwave dryer and 3.69%–5.47% for the hot air dryer [49]. In another study by Beigi for drying apple, the thermal efficiency was calculated as ranging

3.70%–9.44%, and it was shown that, with increasing temperatures, the thermal efficiency increases [46]. Motevali et al. calculated the thermal efficiency for drying chamomile in a hot air dryer between 2.12% and 6.87% and stated that the thermal efficiency increases with increasing drying temperatures [44].

### *3.5. Qualitative Properties*

#### 3.5.1. Color

As can be seen in Figure 7, by raising the air temperature from 50 to 70 ◦C, the overall color change significantly increased (*p* < 0.05). One of the reasons is that, during the process of food drying, the pigments are decomposed due to slow heating and will cause discoloration in the products [18]. Islam et al. showed that increasing the temperature causes papaya color changes in a hot air dryer [35]. Sufer and Palazoglu achieved similar results for drying pomegranate [50]. For drying kiwifruit in a hot air dryer at different temperatures, Izli et al. showed that the color changes of kiwifruit increased by increasing the temperature [20]. Additionally, by increasing the thickness of the sample, the overall color changes increased, which was due to the fact that at high thicknesses, as the drying time increases, the kiwifruits slices were exposed to hot air for a longer period of time, and the high heat exposure to kiwifruit slices intensified the decomposition of pigments [26]. Therefore, at a thickness of 4 mm, the least amount of overall color changes was seen. The use of ultrasonic pretreatment also reduced color changes. Increasing the time of ultrasonic application reduced the rate of color changes in the ultrasonic pretreatment. Additionally, applying ultrasonic pretreatment time had less color changes than the control sample. The color changes during kiwifruit drying are due to various factors such as thermal degradation of chlorophyll, oxidation, enzymatic reactions and non-enzymatic browning reactions [51]. Taghinezhad et al. studied the drying of blackberries and concluded that the least amount of overall color change was observed at 30 min ultrasonic radiation time [37]. In another study, Dehghannya et al. examined the color changes of *Mirabelle plum* in the hot air dryer with ultrasonic pretreatment [52]. They showed that the amount of color changes decreases with increasing ultrasonic radiation time. Further, the results obtained by Wang et al. study, which was gathered on kiwifruit slices in the hot air dryer with ultrasonic pretreatment, showed that the highest amount of overall color change was 32.45 in the control sample and the lowest color change amount of 9.59 was observed in the ultrasonic radiation time of 30 min [27].

**Figure 7.** Color changes of kiwifruit slices under different drying conditions.

#### 3.5.2. Shrinkage

Figure 8 shows the percentage of shrinkage changes in the *HAI* dryer with ultrasonic pretreatment at different thicknesses of kiwifruit. The results show that, at high temperatures, high ultrasonic radiation time and less thickness, the amount of shrinkage is significantly reduced (*p* < 0.05). As kiwifruit has a high initial moisture, due to the shrinkage, which is also significant, some changes occur in the initial structure during the drying process. In this way, the fluid in the cell wall exerts pressure to the cell wall and the fluid inside the cell is put under pressure [51]. During the drying process, the removal of water from the cell reduces the stress induced by the liquid on the cell wall. This reduction in stress causes the shrinkage of the constituent texture of the substance [26]. However, air temperature, high ultrasonic radiation time and less thickness accelerate the water removal rate and do not allow the product to deform. The results of this study corroborate those obtained by Kaveh et al. [29]. They investigated the effect of ultrasonic radiation time on blackberry shrinkage in the hot air dryer, and demonstrated that with increasing air temperature and ultrasonic radiation time, the amount of shrinkage decreased. Dehghannya et al. also achieved similar results for drying potato slices [11].

**Figure 8.** Shrinkage of kiwifruit slices under different drying conditions.

#### **4. Conclusions**

0.0117 and χ – – In this study, drying of kiwifruit was performed at thicknesses of 4, 6 and 8 mm with a hybrid *HAI* dryer at air temperatures of 50, 60 and 70 ◦C and ultrasonic times of 0 (control sample), 10, 20 and 30 min. The results showed that, by increasing the temperature, ultrasonic time and decreasing the thickness, the drying time and *SEC* decreased. The logistic model was able to predict the MR with the highest value of *R <sup>2</sup>* = 0.9997 and the lowest values of *RMSE* = 0.0117 and *χ* <sup>2</sup> = 0.0007 were chosen as the best model. Additionally, by increasing the thickness of the samples, ultrasonic pretreatment time and by prolonging the air temperature, the amount of *Deff* increased from 3.09 × 10 −10 to 2.26 × 10 <sup>−</sup><sup>9</sup> m2/s. The *SEC* value decreased significantly (*p* < 0.05) by increasing the temperature, ultrasonic time and decreasing the thickness. The lowest *SEC* value was obtained at 70 ◦C, ultrasonic time of 30 min, and thickness of 4 mm. The highest values of drying efficiency, energy efficiency, and thermal efficiency were 11.09%, 7.69% and 10.58%, respectively, at air an temperature of 70 ◦C, an ultrasonic time of 30 min, and sample thickness of 4 mm. The lowest color changes and shrinkage values were 13.2%9 and 25.25%, respectively, at 50 ◦C air temperature, 30 min ultrasonic time, and 4 mm thickness.

— —

, and Wrocław University of Environmental and Life Sciences.

**Author Contributions:** Conceptualization, E.T. and M.K.; methodology, E.T. and M.K.; validation, E.T. and M.K.; formal analysis, E.T.; investigation, E.T. and M.K.; resources, E.T. and A.S.; data curation, E.T. and M.K.; writing—original draft preparation, E.T. and A.S.; writing—review and editing, E.T., A.S.; and M.K.; visualization, E.T. and A.S.; funding acquisition, E.T. and A.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the office of vice chancellor for research at Mohaghegh Ardabili University, and Wrocław University of Environmental and Life Sciences.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data for this research will available.

**Acknowledgments:** The authors are highly thankful to Department of Agricultural Technology Engineering, Mohaghegh Ardabili University, Ardabil, Iran for providing facilities to conduct this research work.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Two-Stage Production System Pondering upon Corporate Social Responsibility in Food Supply Chain: A Case Study**

**Ten-Suz Chen <sup>1</sup> , Yung-Fu Huang <sup>2</sup> , Ming-Wei Weng <sup>2</sup> and Manh-Hoang Do 2,3,\***


**Abstract:** Corporate social responsibility (CSR) has witnessed remarkable attention in academic studies as well as being widely conducted in different industries globally. This specific case was chosen as one of the biggest dairy companies that may be represented for Vietnam dairy supply chain management. This research aims to integrate CSR initiatives into food supply chain management to clarify the optimal replenishment policy, paying close attention to the relationship between midstream manufacturers and final customers. The classical economic production quantity model has been employed, relying on the two-stage assembly production system. The three parameters that contribute to the total profit formulation that have been considered consist of the social charity amount for per unit selling, the unit wholesale price of the manufacturer, and the return rate of used goods from the customer. The study has stressed that there is a significant impact from implementing CSR initiatives on the enterprise's inventory policy that leads to enhance the firm's financial performance.

**Keywords:** Two-stage production system; corporate social responsibility; supply chain; dairy industry; social charity; Vietnam

#### **1. Introduction**

The idea of conducting corporate social responsibility (CSR) has undergone phenomenal growth in recent decades and has been rolled out in diverse sectors globally, such as the European banking industry [1], Vietnamese coffee sector [2], Jordanian pharmaceutical industry [3], the U.S. restaurant industries [4], Chinese food industry [5], and so on. Indeed, this positive signal has attracted increasing interest from corporations due to the advantages of adopting CSR into their business strategy [6–8]. Some recent studies have stated the critical role of CSR activities; for example, the positive effects of CSR on the firm's financial performance [9,10], the value perception by consumers is generated [11], corporate reputation, consumer's satisfaction, or competitive advantage of the company [12–14]. Thus, two major characteristics contribute to CSR simply understanding as follows: firstly, the relationship between corporate responsibility and the firm's stakeholders, including internal and external. Secondly, toward sustainable development, which obligation has divided into three major issues: enhancing the economy, protecting the environment, and contributing to local society. Hence, CSR initiatives have been conducted and considered an excellent tool to attain sustainability [15–17].

In terms of supply chain management (SCM), the structure starts from the first stage known as upstream suppliers to the last step known as downstream customers; thus, there are many relationships among stakeholders in the supply chain that must be solved and predicted [18–20]. Additionally, the positive signal from CSR initiatives can enhance SCM

**Citation:** Chen, T.-S.; Huang, Y.-F.; Weng, M.-W.; Do, M.-H. Two-Stage Production System Pondering upon Corporate Social Responsibility in Food Supply Chain: A Case Study. *Appl. Sci.* **2021**, *11*, 1088. https:// doi.org/10.3390/app11031088

Academic Editor: Theodoros Varzakas Received: 24 December 2020 Accepted: 21 January 2021 Published: 25 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

performance and may obtain a sustainable supply chain; thus, the number of studies integrating CSR initiatives into SCM has also been gradually becoming popular and diverse in many different industries worldwide [14,21,22]. Hence, adopting CSR initiatives into SCM may have been affected by various factors or barriers [18,19]. The customer is at the last stage of the supply chain but also plays a decisive role in the sales of products and impacts the economic responsibility of the firms. However, there are only a few studies concerning the relationship between customers and manufacturers, which have an influence on CSR ideas, especially the food supply chain [16,17]. The food supply chain has several particular characteristics compared to other sectors; thereby, research integration of CSR initiatives into the food SCM is attaining much attention due to its contribution to sustainability, particularly in transitional economies [23]. From some of the existing CSR-SCM studies about stakeholder theory, they have demonstrated that businesses' efforts to adopt CSR activities into corporate governance policies to satisfy the interests of shareholders and stakeholders because SCM plays a role as the critical interface with suppliers or agency services, for instance, logistics services, transport providers, service of yards and warehouses [24,25]. Unfortunately, only a few studies consider the optimal model related to SCM stakeholders; meanwhile, the excellent role of CSR programs in SCM sustainability has not been considered properly, particularly in transitional economies [21,22]. Thus, this research aims to propose an optimal model for the food supply chain, which can adopt CSR initiatives into the food SCM. In this way, supporting the food company has a social contribution as well as obtaining superior financial performance. This study has some research questions as follows:

*RQ1. What is the optimal social charity amount for per unit product selling?*

*RQ2. What is the optimal production run time of the manufacturer?*

*RQ3. What is the optimal return rate of used products from the customer?*

This study makes several contributions to the published literature regarding CSR and SCM. Firstly, this research develops the classical economic production quantity (EPQ) model to address the stakeholders' satisfaction under the two-stage assembly production system. An analysis between two supply chain members will be conducted, including midstream manufacturers and downstream buyers, whose members are witnessing the CSR adoption. Secondly, the proposed EPQ model's efficacy has been assessed through the Vietnamese dairy supply chain case study, which is among the critical sectors contributing to the sustainable development of a developing nation like Vietnam [26]. This is the unique research considering the relationship between CSR practices and food supply chain management in Vietnam; thus, the research findings provide useful information to SCM's managers. Moreover, this model's scope can also be expanded to various sectors to determine the proposed model's efficacy. The structure of this study as follows. The authors discussed the introduction in Section 1. The literature review and the methodology of this study will be described in Sections 2 and 3, respectively. The next section will outline the notation and model formulation for this research. A numerical example of this study will be shown in Sections 5 and 6. In the last part, Section 7 outlines the research implications and limitations of the study, as well as suggestions for new study directions in the future.

#### **2. Literature Review**

To date, the definition of CSR is still controversial about consistency, and there are many different perspectives on CSR definition [7,11]. In studies of the stakeholders' theory, the scholars agreed that businesses would enhance their profit; however, they must also assume responsibilities to the environment, community, and society where the company is operating [27]. Therefore, the CSR concept has been understood under three responsibilities dimensions: economic, social, and environmental responsibility [2,27]. Regarding supply chain management, this includes many different stages with several complex relationships between the company and other stakeholders. The role of stakeholders in sustainable supply chain management (SSCM) has been described in several previous studies, such as empirical research in India of Das (2018) [12]; a review research from Sodhi

and Tang (2018) [18]; or Sarkar et al. (2018) who suggested a production model of the automobile SSCM related to environment protection [20]. Consequently, corporatations need to determine their responsibilities with goods from the first stage as forming ideas [28]. CSR initiatives have been employed as useful tools to incorporate with SCM to improve corporate decision-making effectiveness, as well as to accomplish SSCM targets [21,24,29]. In fact, the findings obtained have recently shown that various studies incorporating CSR and SCM have been increasing rapidly. For instance, Feng (2017) systematically reviewed the relationship between CSR and SCM among previous studies and pointed out that these studies are heavily concentrated in developed countries; meanwhile, just a few studies have been conducted in emerging economies [21]; Das (2018) evaluated the model to enhance SSCM performance of the Indian enterprises through competitiveness measurement that had been influenced by CSR practices [12]; Khalid et al. (2015) suggested employing threeaspect obligations to attain SSCM, however, they found a lack of attention on the social dimension of the SCM compared to other responsibilities [30]; Wu et al. (2017) indicated that supplier's CSR misconduct directly remarkably affects the economic performance in the SCM [25].

In terms of the food supply chain sector, which is always a difference between the food industry and other industries since the characteristics of food products such as food quality, safety, storage condition, and shelf life as well as the process of transporting between warehouses, and stores to end customers [5,31]. These are factors that can easily occur in incidents related to food goods. Moreover, it also may impact food supply chain management and make them more difficult and complicated, although the companies have applied risk mitigation strategies in the supply chain [5,31,32]. Evaluation of the correlation between CSR issues and the food SCM has been indicated as a hot topic and given much attention by some scholars in recent years [21]. For instance, Chkanikova and Mont (2015) highlighted some drivers and barriers for food retailers in conducting CSR in their supply chains [33]; Prakash (2018) indicated the rise of CSR and SCM studies; however, it is not really significant and worthy of research potential [34]. Hence, there has been a gap in sustainability efforts and outcomes between developed nations and emerging nations in attaining a sustainable food supply chain. The comparison research between two types of countries has been conducted and claims that in developed countries' conditions where mature supply chains already exist, the industry can adopt risk mitigation strategies and achieve the effectiveness. They also suggested that the future of research should clarify which conditions in transitional economies are needed to achieve a sustainable supply chain [35]. However, their similar findings suggested that companies should pay attention to the different stages and their roles from up-stream to down-stream [28,36].

Regarding SCM in the developing country context, especially in the perishable food supply chain, no company wants to suffer the risk of damage because of perishable products such as meat, seafood, or other perishable goods. The relationship between manufacturers and retailers is very complicated and mainly focused on their economic benefits. There are two aspects, including the price markdown costs and it is related to some parameters such as potential customer quantity and market price fluctuations. These two factors affect the optimal choice of pricing strategy of the enterprises in the perishable food supply chain [37]. Meanwhile, the perceived consumer in developing countries has changed significantly in recent years. Huang et al. (2019) stressed that Vietnamese consumers have more interest and are willing to spend more money on products from socially responsible companies [11]. In terms of the dairy SCM, the dairy quality was the top priority of all stakeholders as well as stages in the SCM; thus, the biggest issue is dairy quality assurance across the entire supply chain because the specificity of dairy products is highly perishable and it is very costly to preserve refrigeration and collect it continuously. Previous studies have shown that manufacturers and stakeholders such as farmers, logistic services, and suppliers have mutual responsibility related to dairy quality [23,38–41]. Hence, studying the optimal model of integrating CSR into SCM in the food industry context will enrich the CSR and SCM literature and suggest more new promising ideas for other scholars in the future.

#### **3. The Research Methodology**

#### *3.1. The EPQ Model*

The classical economic production quantity model is developed based on the single product and single-stage production system; furthermore, it has been widely employed in various SCM studies [42–45]. In fact, the manufacturing process's complexity recently has witnessed more involvement with different components, from raw materials procurement to distribution systems for finished products. Therefore, several articles have been interested in studying how the end product is manufactured through a multi-stage production system [46–48]. Several studies have noted that multi-component production strategies adopt opportunistic maintenance policies [49–52]. However, only a little literature has been published on the impact of CSR in the inventory field. Those recommendations have mentioned that the impact of CSR on the manufacturing process as well as the supply chain would be given new opportunities to explore. For instance, Modak et al. (2019) have considered the effect of selling price and social work donation on demand for three-channel structures [53].

The authors will illustrate the optimal model to enrich the findings of Chang et al. (2012), who have conducted that model in the automation industry. They have also suggested that other scholars may evaluate their model in various industries or another country context [47]. Consequently, many scholars have been conducting the two-stage production system approach to point out the most suitable EPQ model for their examples. This example can be demonstrated by Gupta and Mohanty (2015), Dey et al. (2019), or Sabbaghnia and Taleizadeh (2020); they have argued the two-stage production system perspective through their research [48,54,55], particularly SCM integration with CSR similarly to Nematollahi et al. (2017), and Jokar and Hosseini-Motlagh (2020) [56,57]. Furthermore, they have recommended evaluating this approach's applicability to different sectors in the real world. From these suggestions include relying on different characteristics related to the dairy industry and the emerging economy; thus, the author has employed the proposed model of Chang et al. (2012) to evaluate concordance through a case study in the Vietnamese dairy industry. Moreover, it has also contributed to filling the literature gap in theoretical inventory management in the dairy supply chain.

#### *3.2. The Research Context*

Vietnam is a transitional economy with agriculture as a majority development sector [58,59]. The Vietnamese dairy industry is one of the priority sectors for sustainable development targets and receives attention from government policies and enterprises, and the community [26]; furthermore, the Vietnamese dairy industry represents a typical case study in the Vietnamese supply chain (be presented in Figure 1). However, only a few dairy firms are involved in CSR programs and attempt to commit to stakeholders, the community, and society. Hence, their CSR programs' undeniable efforts over the years related to shareholders, employees, local-residents, environment protection, and others, have had a phenomenal contribution to the sustainability of a transitional economy like Vietnam [11]. Furthermore, the perception of Vietnamese consumers for products is also increased when they pay more attention to the origin of the products as well as the responsibilities of those companies to society instead of only being interested in price and quality [11]; therefore, it has given a positive sign and become the dynamic for Vietnamese dairy firms to CSR adoption.

**Figure 1.** Simplified Vietnam's Dairy Supply Chain. (Source: The authors).

#### **4. Model Formulation**

#### *4.1. Notation*

In this study, the authors proposed an EPQ model for a two-stage assembly production system throughout one cycle. The system illustrates the relationship for all returned products in Stage 1 (manufacturing process) and end product in Stage 2 (assembly process). The manufacturer also directly collects the returned goods through a closed-loop supply chain with a reverse dual-channel. To ease readability, the authors will adopt the same notation in the production system and formulation from Chang et al. (2012) for this paper. The notation used in this paper, including system parameters and decision variables, has been listed in Tables 1 and 2, respectively.

#### **Table 1.** System parameters of this study.


#### **Table 2.** Decision variables.


#### *4.2. Assumptions*

The aim is to determine the effects of an optimal model for this study, therefore, the authors have adopted four assumptions as below.

(1) The production cycle repeats infinitely.

(2) Following the line of Savaskan et al. (2004), the total collection cost : *C*(*r*) = *I<sup>r</sup>* + *c*0*rD* = *αr* <sup>2</sup> + *c*0*rD*, where *rD* was the total amount of goods which will be returned from customers; besides, *<sup>D</sup>* <sup>=</sup> *<sup>a</sup>* <sup>−</sup> *bp<sup>w</sup>* <sup>+</sup> *<sup>δ</sup>s*; moreover, *<sup>a</sup>* <sup>&</sup>gt; 0, is the entire size of the market, *b* is the price sensitivity of demand, *δ* is the social charity of demand.

(3) The return rate for used products from the customers to the manufacturer in the reverse channel, where *r* = √ *Ir*/*α* and *α* is a scaling parameter.

(4) Based on Ma et al. (2013) and Hosseini-Motlagh et al. (2018), *MI* is the investment function of the manufacturer which is illustrated as follows *MI* = √ *Is*/*ε*, where *ε* is a scaling parameter [46,60].

(5) The authors consider forward and reverse dual-channel in reverse logistics with a manufacturer and a retailer. In the forward dual-channel, the manufacturer sells a product on the market through the retailer through reverse logistics. In reverse dualchannel, the manufacturer collects the returned products; additionally, they also conduct CSR activities.

#### *4.3. Model Formulation*

From the notation and assumptions illustrated above, the graph of production system during the time period from 0 to T [0,T] will be demonstrated in the two-stage production system (reference in Figure A1 of Appendix A). The total profit function will be established; the model of research is a perfect cyclic process. The demands of customers with amounts of dairy products in a cycle will be met by the right number of goods from manufacturers (i.e., *p<sup>i</sup> t<sup>i</sup>* = *pnt<sup>n</sup>* = *DT*).

Therefore, *t<sup>i</sup>* and *T* can be presented in turn as follows:

$$t\_i = \frac{p\_n t\_n}{p\_i} \,\prime \tag{1}$$

with *i* = 1, 2, . . . , *n* and

$$T = \frac{p\_n t\_n}{D} \,\prime \tag{2}$$

Hence, the maximum level of inventory for raw material *i* at retailer stage is

$$Z\_{\rm i} = (p\_{\rm i} - p\_{\rm e})t\_{\rm i} \tag{3}$$

with *i* = 1, 2, . . . , *n*.

Therefore, the period when the inventory of raw material *i* depletes can be determined as:

$$t\_{id} = \frac{Z\_i}{p\_\varepsilon} = \frac{(p\_i - p\_\varepsilon)t\_i}{p\_\varepsilon} = \left(\frac{p\_i}{p\_\varepsilon} - 1\right)t\_{i\prime} \tag{4}$$

with *i* = 1, 2, . . . , *n*.

$$H\_{\varepsilon} = \left(p\_{\varepsilon} - D\right)\left(t\_{\text{n}} + t\_{\text{nd}}\right)$$

$$= \left(p\_{\varepsilon} - D\right)\left[t\_{\text{n}} + \left(\frac{p\_{\text{n}}}{p\_{\text{c}}} - 1\right)t\_{\text{n}}\right] \tag{5}$$

$$= \left(p\_{\varepsilon} - D\right)\frac{p\_{\text{n}}}{p\_{\text{c}}}t\_{\text{n}}.$$

However, the maximum inventory level of return used product as

$$Z\_{\rm rc} = \rm rt\_{rdi} \tag{6}$$

where *i* = 1, 2, . . . , *m*.

Based on six Equations (1)–(6), with the influence of these components, the total profit function will be established as follows:

(1) Sales revenue (*SR*)

$$\mathcal{SR} = (p\_w - s)$$

(2) Setup cost

$$\mathbf{C}\_s = k$$

(3) Holding the cost of the end product (*HCe*)

$$HC\_{\varepsilon} = \frac{h\_{\varepsilon} Z\_{\varepsilon} T}{2} = \frac{h\_{\varepsilon} (p\_{\varepsilon} - D)}{2 D p\_{\varepsilon}} p\_{n}{}^{2} t\_{n}{}^{2}$$

(4) Holding the cost of all materials (*HCc*)

$$H\mathcal{C}\_{\mathcal{C}} = \frac{p\_n^{\,2}t\_n{}^2}{2} \left[ \sum\_{i=1}^n h\_i \left( \frac{1}{p\_{\mathcal{C}}} - \frac{1}{p\_i} \right) \right]^2$$

(5) Disposal costs of the defective end product/all raw materials per cycle (*DC*)

$$DC = \left(d\_{\varepsilon}\theta\_{\varepsilon} + \sum\_{i=1}^{n} d\_{i}\theta\_{i}\right) DT + \left(T - \sum\_{i=1}^{m} t\_{rdi}\right) rd\_{rel}$$

(6) Return costs for used products at retailer stage (*RC*).

$$\mathcal{RC} = c\_1 r \sum\_{i=1}^{m} t\_{rdi}.$$

(7) Based on Giri's proposed model in 2005 [61], the production cost *(PC)* as:

$$PC = \left(\beta\_0 + \frac{\beta\_1}{p\_\varepsilon} + \beta\_2 p\_\varepsilon\right) p\_n t\_n.$$

In the integrated supply chain system, all upstream manufacturers are willing to conglomerate resources. In an organizational decision-making process involving the manufacturer and the retailer, a product is manufactured in a single batch. Thus, making capital investment decisions on CSR activities is given by *I<sup>s</sup>* and collecting the used products for recycling is given by *I<sup>r</sup>* . Based on Savaskan et al. (2004), both investment costs are *I<sup>s</sup>* = *βs* <sup>2</sup> and *I<sup>r</sup>* = *αr* 2 , where *α* and *β* are scaling parameters, respectively [62]. Therefore, the objective function of the proposed model consisting of seven parts to maximize the total profit per unit of time is given by optimizing *s*, *tn*, and *r*.

Thus, the total profit per unit time (denoted by *TP*(*s*, *tn*, *r*)) is given by

$$\begin{split} \text{TP}(s, t\_{\text{tr}}, r) &= \quad \left( \text{SR} - \text{SC} - \text{HC}\_{\text{c}} - \text{HC}\_{\text{c}} - \text{DC} - \text{RC} - \text{PC} \right) \\ &= \quad \left( p\_{\text{w}} - s + c\_{1}r \right) (a - bp\_{\text{w}} + \delta s) \\ &\quad - \left[ \frac{h\_{c}[p\_{\text{c}} - (a - bp\_{\text{w}} + \delta s)]}{(a - bp\_{\text{w}} + \delta s)p\_{\text{c}}} \right] + \left[ \sum\_{i=1}^{n} h\_{i} \left( \frac{1}{p\_{\text{c}}} - \frac{1}{p\_{i}} \right) \right] \right] \frac{p\_{\text{n}}^{n} t\_{\text{n}}^{2}}{2} \\ &\quad - \left[ \left( d\_{\text{c}} \theta\_{\text{c}} + \sum\_{i=1}^{n} d\_{i} \theta\_{i} \right) + \left( \beta\_{\text{0}} + \frac{\theta\_{\text{1}}}{p\_{\text{c}}} + \beta\_{2} p\_{\text{c}} \right) + \frac{rd\_{\text{rc}}}{a - bp\_{\text{rw}} + \delta s} \right] p\_{\text{n}} t\_{\text{n}} \\ &\quad + (d\_{\text{rc}} - c\_{1}) r \sum\_{i=1}^{n} t\_{\text{r}di} \\ &\quad - \beta s^{2} - ar^{2} - k. \end{split} (7)$$

Aiming to address this nonlinear programming issue, the authors first ignore the restriction and take the first-order derivation of *TP*(*s*, *tn*,*r*) with respect to *s*, *tn*,*r*, respectively. We obtain

$$\begin{aligned} \frac{\partial TP(s, t\_n r)}{\partial s} &= \left[ \delta c\_1 r + (b - \beta) p\_w + (\beta + \delta)(p\_w - 2s) - a \right] \\ &- (p\_n t\_n h\_\ell + 2rd\_{re}) \left[ \frac{\delta}{(a - bp\_w + \delta s)^2} \right] \frac{p\_l t\_n}{2}, \end{aligned} \tag{8}$$

$$\begin{aligned} \frac{\partial TP(s, t\_n r)}{\partial t\_n} &= \ -\left\{ \frac{h\_c [p\_\ell - (a - bp\_w + \delta s)]}{(a - bp\_w + \delta s) p\_\ell} + \left[ \sum\_{i=1}^n h\_i \left( \frac{1}{p\_\ell} - \frac{1}{p\_i} \right) \right] \right\} p\_n r^2 t\_n \\ &- \left[ \left( d\_\ell \theta\_\ell + \sum\_{i=1}^n d\_i \theta\_i \right) + \left( \beta\_0 + \frac{\beta\_1}{p\_\ell} + \beta\_2 p\_\ell \right) + \frac{rd\_{rr}}{a - bp\_w + \delta s} \right] p\_{n\ell} \end{aligned} \tag{9}$$

and

$$\frac{\partial TP(s, t\_n, r)}{\partial r} = c\_1(a - bp\_w + \delta s) + (d\_{re} - c\_1) \sum\_{i=1}^{m} t\_{rdi} - 2ar. \tag{10}$$

To find the optimal solution of (*s*, *tn*,*r*), let *∂TP*(*s*, *tn*,*r*)/*∂s* = 0, *∂TP*(*s*, *tn*,*r*)/*∂t<sup>n</sup>* = 0, and *∂TP*(*s*, *tn*,*r*)/*∂r* = 0, simultaneously. Solving these three equations, we obtain

$$
\delta c\_1 r + (b + \delta) p\_w = \left[ \frac{p\_n t\_n h\_\varepsilon + 2rd\_{\varepsilon}}{2(a - bp\_w + \delta s)^2} \right] \delta p\_n t\_n + a + 2s(\delta + \beta),
\tag{11}
$$

$$\begin{split} \left\{ \frac{h\_{\epsilon}[p\_{\epsilon} - (a - bp\_{w} + \delta \mathbf{s})]}{(a - bp\_{w} + \delta \mathbf{s})p\_{\epsilon}} + \left[ \sum\_{i=1}^{n} h\_{i} \left( \frac{1}{p\_{\epsilon}} - \frac{1}{p\_{i}} \right) \right] \right\} p\_{n} \,^{2}t\_{n} + \frac{rd\_{n}}{a - bp\_{w} + \delta \mathbf{s}} p\_{n} \\ = -\left[ \left( d\_{\epsilon} \theta\_{\epsilon} + \sum\_{i=1}^{n} d\_{i} \theta\_{i} \right) - \left( \beta\_{0} + \frac{\beta\_{1}}{p\_{\epsilon}} + \beta\_{2} p\_{\epsilon} \right) \right] p\_{n} \end{split} \tag{12}$$

and

$$\tau = \frac{c\_1(a - bp\_w + \delta s) + (d\_{re} - c\_1) \sum\_{i=1}^{m} t\_{rdi}}{2a}. \tag{13}$$

From Equations (11)–(13), it is clear that *s* and *r* can be uniquely determined as functions of *tn*. To analyze the inventory problem, we will show that for any given feasible (*s* ∗ ,*r* ∗ ), the optimal production run time also exists and is unique. For given *s* ∗ and *r* ∗ , the first-order necessary condition for *TP*(*tn*|*s* ∗ ,*r* ∗ ) to be maximum is

$$\frac{d T(t\_n|s^\*,r^\*)}{dt\_n} = -\left\{ \frac{h\_i[p\_e - (a - bp\_w + \delta s)]}{(a - bp\_w + \delta s)p\_e} - \left[ \sum\_{i=1}^n h\_i \left( \frac{1}{p\_e} - \frac{1}{p\_i} \right) \right] \right\} p\_n{}^2 t\_n$$

$$- \left[ \left( d\_e \theta\_\varepsilon + \sum\_{i=1}^n d\_i \theta\_i \right) + \left( \beta\_0 + \frac{\beta\_1}{p\_\varepsilon} + \beta\_2 p\_\varepsilon \right) + \frac{rd\_{tr}}{a - bp\_w + \delta s} \right] p\_n{}^2 = 0. \tag{14}$$

$$= \quad 0.$$

Let *G*(*tn*) be the left-hand side of Equation (14), i.e.,

$$\begin{split} G(t\_n) &= -\left\{ \frac{h\_\varepsilon[p\_\varepsilon - (a - bp\_w + \delta \mathbf{s}^\*)]}{(a - bp\_w + \delta \mathbf{s}^\*)p\_\varepsilon} - \left[ \sum\_{i=1}^n h\_i \left( \frac{1}{p\_\varepsilon} - \frac{1}{p\_i} \right) \right] \right\} p\_n{}^2 t\_n \\ &- \left[ \left( d\_\varepsilon \theta\_\varepsilon + \sum\_{i=1}^n d\_i \theta\_i \right) + \left( \beta\_0 + \frac{\beta\_1}{p\_\varepsilon} + \beta\_2 p\_\varepsilon \right) + \frac{r^\* d\_\mathcal{n}}{a - bp\_w + \delta \mathbf{s}^\*} \right] p\_\mathcal{n} . \end{split} \tag{15}$$

We first rewrite Equation (14) and have

$$\begin{split} \left[ \left( d\_e \theta\_e + \sum\_{i=1}^n d\_i \theta\_i \right) + \left( \beta\_0 + \frac{\beta\_1}{p\_e} + \beta\_2 p\_e \right) + \frac{r^\* d\_{\rm IF}}{a - bp\_w + \delta s^\*} \right] p\_n \\ = - \left\{ \frac{h\_e [p\_e - (a - bp\_w + \delta s^\*)]}{p\_e (a - bp\_w + \delta s^\*)} - \left[ \sum\_{i=1}^n h\_i \left( \frac{1}{p\_e} - \frac{1}{p\_i} \right) \right] \right\} p\_n ^2 t\_n. \end{split} \tag{16}$$

Because of the left-hand side of Equation (16)

$$\left[ \left( d\_{\varepsilon} \theta\_{\varepsilon} + \sum\_{i=1}^{n} d\_{i} \theta\_{i} \right) + \left( \beta\_{0} + \frac{\beta\_{1}}{p\_{\varepsilon}} + \beta\_{2} p\_{\varepsilon} \right) + \frac{r^{\*} d\_{n\varepsilon}}{a - bp\_{w} + \delta s^{\*}} \right] p\_{n} > 0, \beta\_{0} $$

then we have ∆ > 0, where

$$\Delta \equiv \left\{ \left[ \sum\_{i=1}^{n} h\_i \left( \frac{1}{p\_\varepsilon} - \frac{1}{p\_i} \right) \right] - \frac{h\_\varepsilon [p\_\varepsilon - (a - bp\_w + \delta s^\*)]}{p\_\varepsilon (a - bp\_w + \delta s^\*)} \right\} p\_n^{-2} \mathfrak{t}\_n.$$

Next, taking the first-order derivative of *G*(*tn*) with respect to *tn*, we obtain

$$\frac{dG(t\_n)}{dt\_n} = \left\{ \left[ \sum\_{i=1}^n h\_i \left( \frac{1}{p\_\varepsilon} - \frac{1}{p\_i} \right) \right] - \frac{h\_\varepsilon [p\_\varepsilon - (a - bp\_w + \delta \mathbf{s}^\*)]}{(a - bp\_w + \delta \mathbf{s}^\*) p\_\varepsilon} \right\} p\_n^{-2} = \frac{\Delta}{t\_n} > 0. \tag{17}$$

Therefore, *<sup>G</sup>*(*tn*) is a strictly increasing function *<sup>t</sup><sup>n</sup>* ∈ (0, <sup>∞</sup>).

**Theorem 1.** *For any given t<sup>n</sup>* ≥ 0, *we consider the interval t<sup>n</sup>* ∈ (0, <sup>∞</sup>),

*(a) If G*(*tn*) < 0, *then the solution* (*s* ∗ , *t* ∗ *n* ,*r* ∗ ) *which maximizes TP*(*s*, *tn*,*r*) *not only exists but also is unique*, *and t*∗ *<sup>n</sup>* ∈ (0, <sup>∞</sup>).

*(b) If G*(*tn*) ≥ 0, *then the optimal value of t<sup>n</sup> is t* ∗ *<sup>n</sup>* → 0. *The production system should not be opened.*

#### **Proof.**

(a) Firstly, we consider the interval *<sup>t</sup><sup>n</sup>* ∈ (0, <sup>∞</sup>). Because lim*tn*→∞*G*(*tn*) = <sup>∞</sup> and lim*tn*→0*G*(*tn*) <sup>&</sup>lt; 0, from the Intermediate Value Theorem, we can find a unique solution *t* ∗ *<sup>n</sup>* ∈ (0, <sup>∞</sup>) such that *<sup>G</sup>*(*<sup>t</sup>* ∗ *n* ) = 0. Substituting *t* ∗ *n* into Equations (8)–(10), the corresponding *s* ∗ and *r* ∗ can be determined. Furthermore, we also calculate that

$$\mathbf{H}\_{11} = \frac{\partial^2 T P(s, r | t\_n)}{\partial s^2} \Big|\_{(s, r) = (s^\*, r^\*)} = 2 \left\{ \left[ \frac{\delta^2 p\_n t\_n (h\_e p\_n t\_n + r d\_{re})}{\left(a - b p\_w + \delta s\right)^3} \right] - \left(\beta + \delta\right) \right\},$$

$$\mathbf{H}\_{22} = \left. \frac{\partial^2 T P(s, r | t\_n)}{\partial r^2} \right|\_{(s, r) = (s^\*, r^\*)} = -2\alpha < 0\_\lambda$$

and

$$\mathbf{H}\_{12} = \mathbf{H}\_{21} = \frac{\partial^2 T P(s, \, r | t\_n)}{\partial s \partial r} \Big|\_{(s, r) = (s^\*, r^\*)} = 0.$$

Therefore, the determinant of the Hessian matrix at the stationary point (*s* ∗ ,*r* ∗ ) is

$$\begin{array}{rcl} \det(H) &= \mathbf{H}\_{11} \times \mathbf{H}\_{22} - \mathbf{H}\_{12} \times \mathbf{H}\_{21} \\ &= -4\alpha \left\{ \left[ \frac{\delta^2 p\_n t\_n \left( h\_e p\_n t\_n + r d\_{re} \right)}{\left( a - b p\_w + \delta s \right)^3} \right] - \left( \beta + \delta \right) \right\} < 0. \end{array}$$

(b) From Equation (9), we obtain that *<sup>∂</sup>TP*(*s*,*tn*,*r*) *∂tn* < 0, which implies that *t<sup>n</sup>* causes a higher value *TP*(*s*, *tn*,*r*). Hence, the maximum value *TP*(*s*, *tn*,*r*) occurs at the point *t* ∗ *<sup>n</sup>* → 0. It seems reasonable to conclude that the production system will not be opened. This completes the proof.

Summarizing the above results, the Algorithm 1 was used to obtain the optimal solution to our problem.


#### **5. Application Example**

The empirical research context is Vietnam's dairy supply chain to demonstrate the utility and feasibility of the proposed model. ABC dairy corporation was chosen, which is currently the largest dairy company in Vietnam [63]. The ABC company currently has a complete supply chain with several farms, factories, and its own distribution system; moreover, their products have been distributed to nearly all supermarkets and retail points all over Vietnam. Towards sustainable development, ABC dairy firm's orientation would invest heavily in CSR initiatives and be directed to the five major objectives, including community development support; environment and energy; responsibility for employees; responsibility for products; and local economic development. Regarding the ABC dairy company's commitment to the stakeholders, they are committed to providing safe products to consumers with top quality and appropriate prices. Moreover, this company also declares complying with environmental regulations and integrating social responsibility into business strategies. Food products' characteristics are easily broken for many different reasons before reaching customers, combined with the ABC company's product quality assurance policy. Thus, the case study selected as the ABC dairy corporation would help the authors better assess the proposed model in the empirical research.

The authors considered an application example of CSR initiatives to demonstrate the proposed model and verify the obtained analytical results. The real data were collected from ABC dairy firm and employed for the numerical analysis as following: *k* = \$80/*cycle*, *β*<sup>0</sup> = 0.1, *β*<sup>1</sup> = 0.03, *β*<sup>2</sup> = 0.025, *p<sup>w</sup>* = \$10, *a* = 15, *b* = 9, *α* = 7, *β* = 2, *δ* = 4.

*Raw material source 1: p*<sup>1</sup> = 100/per unit time, *h*<sup>1</sup> = \$0.01/per unit, *θ*<sup>1</sup> = 0.03, *d*<sup>1</sup> = \$0.01/per unit.

*Raw material source 2: p*<sup>2</sup> = 200/per unit time, *h*<sup>2</sup> = \$0.02/per unit, *θ*<sup>2</sup> = 0.02, *d*<sup>2</sup> = \$0.02/per unit.

*Raw material source 3: p*<sup>3</sup> = 300/per unit time, *h*<sup>3</sup> = \$0.375/per unit/per unit time, *θ*<sup>3</sup> = 0.01, *d*<sup>3</sup> = \$0.03/per unit.

*End product in showrooms: p<sup>e</sup>* = 60/per unit time, *h<sup>e</sup>* = \$2/per unit/per unit time, *θ<sup>e</sup>* = 0.05, *d<sup>e</sup>* = \$0.04/per unit.

Then, we find the outcome as *s* ∗ *=* 9.58943*; t* ∗ *<sup>n</sup> =* 0.495389; *r* ∗ *=* 0.0595798 and *TP*(*s* ∗ , *t<sup>n</sup>* ∗ ,*r* ∗ ) = 287.392.

In the finding of this section, the authors evaluate the impact of implementing CSR practices on the company's inventory policy. Thus, the effect of changes in various parameters of the model, for example (in Table A1 of Appendix B), was indicated that the effects of *he* , *β*2, and *θ<sup>e</sup>* on total profit are significant. These imply that the quality of the end product is essential for the ABC dairy enterprise.

Thus, maintaining high-quality products secures a high level of demand by end customers whereas poor quality products affect the customer's confidence, reputation, and sales of the company. Aim to create a positive brand image through CSR activities of the enterprise towards the community. Hence, the dairy company was aware of the importance of the public, especially their target customers, having a positive perception of them. As Figure 2 indicated, the concave function of the total economic profit based on per unit time.

**Figure 2.** The total profit per unit time *TP*(*s* ∗ , *t* ∗ *n* , *r*) for application example.

#### **6. Discussion of Research Findings**

Toward sustainable development targets that lead to ABC dairy company's CSR programs being conducted to solve the comprehensive enterprise issue, which has a profitsharing obligation to society beyond making a profit. In fact, to contribute the profit-sharing to the nation's target through charitable activities or community development, the ABC dairy firm has an obligation that implements financial responsibilities before other duties. Therefore, it requires the ABC dairy firm to ensure their accountability to stakeholders, especially shareholders' interests. A numerical example in the above subsection is considered to study the effects of changes in the system parameters (*h*1, *h*2, *h*3, *h*e, *θ*1, *θ*2, *θ*3, *θ*e, *d*1, *d*2, *d*3, *d*e, *β*0, *β*1, and *β*2) on the optimal values of *s* ∗ , *t* ∗ *n* , *r*, and *TP*(*s* ∗ , *t* ∗ *n* , *r*). The result indicated that the effects of *h*e, *β*<sup>2</sup> and *θ*<sup>e</sup> on total profit are significant. These imply that the quality of the end product is essential for the ABC dairy enterprise. Thus, maintaining high-quality products aims to secure a high demand by end customers while struggling with the risk that poor quality products affect customers' confidence, reputation, and sales. Brand image,

reputation, as well as financial performance can be significantly enhanced through CSR activities of the enterprise towards the community; hence, ABC dairy company must be aware of the local community's importance, especially their target customers. For the present, it may be useful to look more closely at some of the more important features of the sensitivity analysis performed by changing each of the parameters by +50%, +25%, −25%, and −50%; taking one parameter at a time and keeping the remaining parameters unchanged. The analytical results were clearly demonstrated in Table A1 of Appendix B.

From the Table A1 in Appendix B, the following the management implications have been very positive as follows:

(1) When the values of parameters (*h*1, *h*2, *h*3, *h*e, *θ*1, *θ*2, *θ*<sup>3</sup> and *θ*e) are decreased, it leads to the growth for *TP*(*s* ∗ , *t* ∗ *n* , *r*). It indicates that the firm cooperation with manufacturers is based on these factors, including inventory turnover and the quality of raw material or end products. The total profit has been enhanced, leading to the firm's social donation fund also increasing. Thus, it positively directs on the ABC dairy enterprise's brand and its reputation from the local community.

(2) With decreases in the value of parameter *β*0, then *TP*(*s* ∗ , *t* ∗ *n* , *r*) increases. If the processing cost of raw materials from different sources could be reduced effectively, the total profit would be enhanced. This implies that the ABC dairy firm should invest more in new employee orientation to improve productivity.

(3) With decreases in the value of parameter *β*1, then *TP*(*s* ∗ , *t* ∗ *n* , *r*) increases. This implies that the ABC dairy firm should decrease the labor costs per unit of time (i.e., wage or salary) to increase total profit.

(4) With decreases in the value of parameter *β*2, then *TP*(*s* ∗ , *t* ∗ *n* , *r*) increases. This implies that the ABC dairy firm should decrease the tool or idle cost per unit of time to increase total profit. Furthermore, the *β*<sup>2</sup> parameter will have an influence rather than *β*<sup>0</sup> and *β*<sup>1</sup> on the total profit per unit time.

(5) With decreases in the value of parameters as *d*1, *d*2, *d*3, then *TP*(*s* ∗ , *t* ∗ *n* , *r*) can be increased. This implies that the firm should decrease scrape costs per unit of time to increase total profit.

Apart from complying with laws related to the environment and workers, improving the quality of end products and reducing their holding costs to enhance financial performance are the firms' priority targets. Based on the proposed model of Chang et al. (2012), the authors have developed and evaluated this suggestion to the Vietnamese food supply chain via collected data from the ABC dairy firm case study; thereby, the suitability and positive results have been clarified. These research findings are consistent with other studies related to evaluating the EPQ model under the two-stage assembly production system, such as Dey et al. (2019) and Sabbaghnia and Taleizadeh (2020). They indicated that relationships between SCM members could be comprehensively evaluated to point out the best decision-making; hence, an application example along with sensitivity analysis has been conducted to support this argument. In particular, the relationship between CSR initiatives and SCM of some studies has supported our findings, and the research examples can be seen by Nematollahi et al. (2017); Modak et al. (2019); and Jokar and Hosseini-Motlagh (2020). They have similar conclusions in terms of the vital role as well as the CSR initiatives that could affect decision-making and various SCM performances. Furthermore, to increase donations to local society, the economic performance would be impacted by these activities; however, there is a positive signal from the final customers, whose willingness to buy green products [11,53], or think they are making a small contribution to society generates the enhancing of economic obligation for corporations.

Although the implementation of social activities affects the economic benefits of ABC dairy enterprise, its total profits still increase, which ensures its commitment to its economic responsibility with shareholders and employees. Thus, ABC dairy corporation is a particular example for the EPQ model's suitability, which can point out some practical implications for the top managers in decision-making related to CSR programs under SCM. They need to establish a risk management mechanism to control any potential risks that

may affect their operations and profits. With such an arrangement, they can significantly lower the company's operational risk, damage, and impact.

#### **7. Conclusions**

This study is unique considering integrating CSR initiatives into the food supply chain in Vietnam, which considered some critical decision parameters together and linked the ideas and conclusions based on a particular case study. The noteworthy contributions of Vietnam's dairy firms and their CSR programs are significant to the sustainable development goal of an agricultural country like Vietnam. Aiming to demonstrate the efficiency of the proposed EPQ model, ABC dairy corporation was chosen, which is a typical example that can represent the Vietnam dairy industry in both the main issues of this study as CSR and SCM. Thus, the authors demonstrated the proposed model and clarified how to integrate CSR initiatives into the food SCM. A two-stage assembly production system involving the return rate of used goods from the customer was suggested, which has impacted the CSR policy and top manager decision-making in the dairy supply chain. The research findings are consistent with several prior studies in terms of the CSR initiatives' role, which is considered an excellent tool that may obtain a sustainable supply chain by enhancing social responsibility and the firm's financial performance. Furthermore, total profits per unit of time have been influenced by these critical parameters leading to the following research implications.

(1) Based on theories regarding the relationship between CSR and SCM, these results have contributed to the existing literature on developing the classical economic production quantity model under the two-stage assembly production system.

(2) The findings will help the top managers of dairy corporations in Vietnam better understand the integrating approach and the efficiency of adopting CSR programs into SCM.

(3) The production costs will be effectively controlled, such as material and operating costs. Hence, the firms have more dynamics to devote more money to charitable donations, which positively impacts consumer perception, and then economic performance will be enhanced.

The supply chain management includes all relationships from upstream suppliers to downstream consumers; unfortunately, in this study, the author has just focused on evaluating the relationship between producer ABC and their final customer through social activities, total profits of the dairy company, and the return of the goods issue of the customer. Thus, the authors suggested that further research might extend to different relationships between manufacturers and stakeholders based on CSR activities and SCM nexus. Considering the relationship between manufacturers and suppliers, the linkage effect between the dairy company and retailers in the dairy supply chain of an emerging economy is a promising idea for other scholars. Moreover, aiming to reduce their holding costs and enhance financial benefits, this proposed model could be conducted in various sectors to broaden and enrich the knowledge of the CSR initiatives' usefulness.

**Author Contributions:** M.-H.D. and Y.-F.H. developed conceptualization; methodology was developed by M.-W.W. and T.-S.C.; data curation and resources were undertaken by M.-H.D. and T.-S.C.; Y.-F.H. and M.-W.W. analyzed the data; wrote the manuscript, M.-H.D., M.-W.W., and T.-S.C.; writing—review and editing, T.-S.C., M.-H.D., and M.-W.W.; visualization, Y.-F.H.; supervision, Y.-F.H.; project administration, Y.-F.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** The author's research was partially supported by the Ministry of Science and Technology of the Republic of China under Grant MOST 109-2622-H-324-001. Besides, this research was supported by the Ministry of Education, Taiwan, R. O. C. (Intelligence commerce and marketing craft the social practice model for the Prionailurus bengalensis and agricultural optimized regional revitalization in Chungliao).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Due to its proprietary nature <or ethical concerns>, supporting data cannot be made openly available.

**Acknowledgments:** The authors would like to thank the editor and anonymous reviewers for their valuable and constructive comments, which have led to a significant improvement in the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A**

**Figure A1.** The graph of production system during time period [ 116 *0*, *T*].

#### **Appendix B**


**Table A1.** Effect of changes in various parameters of the model for application example.


**Table A1.** *Cont*.

#### **References**


**Ning Xu 1,2, Yung-Fu Huang <sup>3</sup> , Ming-Wei Weng 3,\* and Manh-Hoang Do 3,4**


**Abstract:** As coronavirus disease 2019 (COVID-19) continues to spread, online consumption habits in China have changed significantly. Thus, the booming online-to-offline (O2O) food ordering and delivery industry via the online bakery have been changing customers' food shopping behavior. This article proposes a comparison relying on advanced O2O strategy for a single-vendor-singleretailer integrated system. Three coordination mechanisms consist of revenue-sharing, buy-back, and quantity flexibility contracts have been employed for optimizing the order quantity. Replenishment strategies and temperature for the supply chain members are considered based on the new retailing framework. Herein, the authors suggest an algorithm for the computation of the optimal solution. Lastly, numerical examples and sensitivity analysis are also conducted to clarify the usefulness of the proposed model in the food supply chain. Sensitivity analysis revealed a number of managerial insights. For example, the results obtained under O2O operations can be compared with those obtained under online/offline operations (under various parameters settings) to determine an opportune moment for three coordination mechanisms.

**Keywords:** inventory; new retailing; baking industrial; food supply chain coordination

#### **1. Introduction**

Several recent studies have appeared to tackle the new retailing (NR) issue, which is the phrase coined by Jack Ma, Alibaba Executive Chairman [1]. Alibaba was significantly implementing NR and online, offline, logistics, and data across the individual value chain to be involved. New retailing is one sector undergoing an Internet of Things (IoT)-driven revolution, as traditional offline retail stores such as convenience stores and supermarkets have become the point of convergence for online and online integration. According to a research report [2], on "China's Online Shopping Market Research Report" published in 2012 by CNNIC, internet shopping market accounted for about RMB1.26 trillion trading volume. The online-to-offline (O2O) business model will revolutionize the customer experience in retailing, allowing companies to introduce innovative customer services. The study is still at an early stage in evaluating NR, not to mention a paucity of literature on this subject. Du and Jiang [3] evaluated that the offline service quality problems of O2O model in China. E-commerce's development in the food supply chain has played an essential role in supporting small and medium enterprises (SMEs) to deal with challenges, stimulating domestic demand for bakery, and creating jobs; thus, the term NR is considered as a new concept based on the development of O2O model. Regarding the history line of the O2O term development, which witnessed the booming of the O2O food market in China in recent years, as recommended by Maimaiti et al. [4], more than twenty percent out of China's total population has already become the customers of the O2O food delivery

**Citation:** Xu, N.; Huang, Y.-F.; Weng, M.-W.; Do, M.-H. New Retailing Problem for an Integrated Food Supply Chain in the Baking Industry. *Appl. Sci.* **2021**, *11*, 946. https:// doi.org/10.3390/app11030946

Academic Editors: Theodoros Varzakas and Tsarouhas Panagiotis Received: 7 December 2020 Accepted: 18 January 2021 Published: 21 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

market. The O2O model's impact on the optimal power structure of a retail service supply chain was studied by Chen et al. [5]. On the other hand, He et al. [6] identified the role of the competitive O2O model to adapt to multi-store service firms. In the early 20th century, Smith and Sparks [7] stressed that product quality is the most vital factor in the food supply chain. Van der Vorst et al. [8] explained that organizations can gain a competitive edge by incorporating food quality into food supply chain. In understanding quality assurance with food security, Trienekens and Zuurbier [9] presented the process of production and distribution is the major interest under food supply chain. Product quality, costs, and sustainability have been highlighted as important factors that influence the food supply chain model was dealt with by Zanoni and Zavanella [10]. In line with that, numerous studies have also investigated those issues in terms of the food supply chain [11–13]. Due to globalization, the complexity of food supply chains is increasingly becoming more complex. The concept of food supply chain traceability has become not just a way to help food manufacturers and retailers ensure food security (ex: temperature controlled), but also creating a linkage among business floors each other. Furthermore, the adoption of the IoT technology into food supply chain management (SCM) to assist with the information management process has gradually become more popular [14–17]. Early theorization of a single-vendor single-retailer integrated inventory model can be traced back to Goyal [18]. Moreover, to struggle with the rapidly changing market conditions, the supply chain system must have smooth coordination, which is considered the vital key to attaining the flexibility necessary as well as to meet superior logistics processes. Besides, several prior studies also stressed that a sounder theoretical basis for vendor-retailer integrated inventory model [19–22]. Consequently, these articles have been published that focus on a wide range of the coordination of the supply chain. Unfortunately, there is no existing article published comparing different food supply chain contracts under the new retailing framework. The channel coordination is considering as the critical key parameter of the SCM, which has been obtained by employ the contract to address related problems. The contract is a major concern in a wide variety of applications; examples of contract are wholesale price, revenue sharing, buy-back, or quantity flexibility.

#### *1.1. Revenue-Sharing Contract*

The contracting mechanism has been adopted by supply chain members and wellknown as a popular method to mitigate risks. Based on the revenue-sharing approach, which has been widely employed in the e-commerce and the manufacturing sector [23–28], Govindan and Malomfalean [29] have employed the Stackelberg game theory to propose the framework for comparing those coordination mechanisms.

#### *1.2. Buy-Back Contract*

Recently, there is an increase remarkably in the number of companies that apply a buy-back contract into supply chain management. Some firms believed that buy-back contracts can improve overall SC coordination in the manufacturing sector [30–35] and retailing sector [36,37].

#### *1.3. Quantity-Flexibility Contract*

The current market is witnessing fierce competition among the firms to obtain acceptance from consumers. Companies are making an effort to offer new goods for sale while customers struggle with so many brands for their needs. One of the roles of the new products coordinator is to track the status of all products submittals. Manufacturers expected retailers to set valuable orders for them, such as early time or large quantities due to the fact that they must spend a long time for production preparation. Quantity-flexibility contracts are efficient coordination schemes to balance the risk of each side, especially in the manufacturing sector [38–44] and retailing sector [45]. Three kinds of contracts have been more widely available and conducted in various supply chains. Based on existing literature, the authors illustrated a taxonomy that reflects accepted categories (as represented

in Table 1). On the whole, there has been relatively little progress in comparing the three coordination mechanisms until recently.

**Table 1.** The main classification of inventory models from relevant studies.


Table 1 demonstrated the comparison among existing relevant studies under different supply chain. It is hoped that this brief and necessarily oversimplified review will lead to a better understanding of three contracts in supply chain, in order to address these areas of concern by shedding light on the nature of the problems and to maximize profits of each member of a food chain made up of one manufacturer and one retailer with three contracts. Hence, the contribution of this article was to investigate the decision-making of the temperature, order quantity, number of shipments in food supply chain between manufacturer and retailer. It is important to recognize the efforts of three coordination mechanisms, but we lack empirical support.

#### *1.4. Food Security*

Temperature control is a major concern in a wide variety of applications such as preserving food (or food security), distribution food, and storage food. It is evident that a key element within these chains is temperature, also in its different forms, as it is a necessary source to guarantee quality-based processes. Shock freezing is widely used in the food industry. However, it is not popular with catering companies that produce their own bread and bakery products. Many companies prefer to use ready-made frozen semi-finished products, rather than to bake them on their own. Center kitchen (CK), also known as the distribution center, is set up as a modern food production base, equipped with complete environmental protection equipment and food processing equipment. From ordering raw materials, food production, to selling food, the center kitchen carries out unified management. CK is a new trend in the food industry where centralized preparation

and processing of fresh foods occurs in the catering chains. It is the basic function of CK to process raw food materials purchased in bulk into semi-products and/or products under the dictation of unified requirement on product qualities. However, the proper control and management of temperature is crucial in delivering perishables to consumers and ensuring that those perishables are in good condition and safe to eat. Aung and Chang [46] addressed the approach used to obtain the optimal temperature for multi-commodity refrigerated storage. Lorite et al. [47] developed a critical temperature indicator (CTI) to monitor the supply chain and storage conditions of perishable food products. Kuo and Chen [48] developed an advanced multi-temperature joint distribution for continuously temperaturecontrolled logistics. For the food manufacturer, positive food security will protect the company's brand and goodwill. However, "loss of goodwill" has been touched upon in management science textbooks of various business school curricula. Today it is widely accepted that a goodwill cost may be difficult to estimate. Lawrence and Pasternack [49] argued that many businesses do not have any idea what the long-term goodwill cost of an unsatisfied customer might be. Aksen [50] examined the role of loss of customer goodwill on the incapacitated lot-sizing problem over a finite planning horizon. Finn et al. [51] investigated whether the cost of food contamination can result in indirect losses such as lost reputation and goodwill. Mitreva and Gjurevska [52] applied the TQM tool to a bread and baking company in the quality system.

#### **2. Problem Description**

New business models have been relentlessly launched in the market due to the rapid advancement of technology and social changes. While internet shopping becomes more efficient and effective, traditional retailers need new digital innovation. An outbreak of coronavirus infection in Wuhan, China, has led to a global epidemic declared a Public Health Emergency of International Concern by the World Health Organization (WHO). COVID-19 is resistant to environment factors, has high penetration ability, and may be lethal. The food supply chain has needed to adjust rapidly to demand-sides shocks, including cake buying and changes in food purchasing patterns as well as planning for any supply-side disruptions due to potential labor shortages and disruptions to transportation and supply networks. However, few empirical studies have been done on this issue. Hobbs [53] proposed an assessment of implication of the COVID-19 pandemic for the food supply chain regarding the growth of the online grocery delivery sector and the extent to which consumers will prioritize "local" food supply chains. Singh et al. [54] investigated an economic model in helping to simulate a responsive food supply chain to match the varying demand and provided decision-making support for rerouting the vehicles during the COVID-19 pandemic. Rizou et al. [55] provided an excellent review of the possible transmission ways of COVID-19 through the food supply chain and environment before developing COVID-19 diagnostic tests. Therefore, we develop an integrated food supply chain system as well as discussing the optimal decisions found to maximize the total profit based on the new retailing framework through COVID-19 pandemic. Transportation issues caused another concern about food quality that concerned products' temperature control in customers or distribution centers from COVID hot spots. On the other hand, the traditional retailers recognize that COVID-19 will have a significant impact on their business. Online technology can support customer service staff, engage offline-to-online ordering, and gives sales assistants more opportunities to help customers explore their options. As fresh food deteriorates easily, there has been increasing consumer demand for logistics that ensure the timeliness of deliveries. He et al. [56] proposed an agent-based O2O food ordering model (AOFOM) that consists of customers, restaurants, and the online food ordering platform. Wang et al. [57] made a comparison of food shopping with four e-commerce modes: Business-to-consumer (B2C), online-to-offline delivery (O2O Delivery), online-to-offline in store (O2O In-store), and New Retail. Li et al. [58] provided an excellent review of online food delivery platforms. A fairly large body of literature exists on the supply chain. However, within that literature, there is a surprising lack of information

on O2O model. Online food delivery provided a critical lifeline during 2020 COVID-19 pandemic for the tens of millions of people quarantined at home. Such research is still in its infancy, but it may have a contribution to make to comparison the understanding of the above three contracts. Next, the empirical evidence is provided in support of the claim. In order to face the highly competitive environment, the following issues were posed by the manufacturer/retailer:

*RQ1. What is the optimal shipment size from manufacturer to the retailer?*


*RQ4. Which of three coordinated agreements attains the best outcome under new retailing framework?*

The main aim of the present article is to use the inventory model to discuss its properties in comparisons with the overall effects of three coordination mechanisms: Revenue sharing, buy-back, or quantity flexibility. The supply chain coordination proposed by Govindan and Malomfalean [29] was used the basis for replicating consumer behavior under O2O environment described above. We addressed the basic two-echelon supply chain proposed by Yang et al. [59]. They investigated single-supplier and single-retailer production and inventory models in both non-cooperative and cooperative situations. Thus, the contribution of this article is in understanding of three coordination mechanisms in the food supply chain. In actuality, however, coordination mechanism is a common problem in supply chain. Little is known about the comparison of the three coordination mechanism in the literature, with most research concentrating on single/two contracts. Moreover, the author suggests that manufacturers should focus on the effect of three coordination mechanisms in the food supply chain. Notations and assumptions will be demonstrated in Section 3, followed by Section 4, which presents the model formulation. Sections 5 and 6 will show the suggestion in terms of solutions and application example, respectively. Next, in Section 7, a numerical example will be presented as proof for the findings of Sections 5 and 6 as mentioned. Finally, conclusions are provided in Section 8.

#### **3. Notations and Assumptions**

In addition, the following assumptions are made in developing our mathematical model.

#### *Assumptions*


#### **4. Model Formulation**

The article considers a two-echelon supply chain with one manufacturer (central kitchen, CK) and one retailer (branch warehouse). The structure is developed under a coordinated case. The purpose of this study was to evaluate the effect of revenue-sharing, buy-back, and quantity- flexibility under new retailing framework.

• Temperature control

One of the major preoccupations of food processing in the past decade has been investigating the nature of temperature control. Temperature is one of the important factors to consider during the bulk production of food items. Storage temperature and time is discussed by Peleg et al. [60], as follows.

$$q(T, t) = q\_0 e^{-b(T)t^{n(T)}},\tag{1}$$

where *b*(*T*) and *n*(*T*) are temperature-dependent coefficients. In particular, *b*(*T*) can be described by the empirical model as follows.

$$b(T) = \ln[1 + e^{m(T - T\_C)}],\tag{2}$$

where *T* is the temperature (*oK*), where *m* and *T<sup>C</sup>* are constants. It is possible to empirically obtain the values of parameters *b*(*T*), *n*(*T*), m, and *T<sup>C</sup>* so as to use the equations above to calculate the expected quality of food products after storage, at given time periods and temperature levels.

• Temperature coefficient

The cost for storage and transportation depends on the temperature in central kitchen. The coefficient of performance (COP) for refrigeration is calculated as follows.

$$COP = \frac{T\_L}{T\_H - T\_L} \,\tag{3}$$

where *T<sup>H</sup>* and *T<sup>L</sup>* are higher (environment) and lower (cooling) temperature measured in Kelvin. For example, if *T<sup>H</sup>* = 293 K (20 ◦ C), and *T<sup>L</sup>* = 273.15 K(1 ◦ C), then *COP* = 13.76. It means that for each unit of energy drawn from an electrical source, the coolant can absorb as much as 13.76 units of heat from the inside of the refrigerator.

We assume the cost of electrical energy at 1 ◦ C is 1, then relative cost at higher temperatures can be calculated by multiplying the cost with the ratio of *COP* values (where we use an environment temperature *T<sup>H</sup>* of 20 ◦ C = 293 K).

• The quality degradation cost (*CQD*) at central kitchen

The loss value of a batch of size *Q* is

$$p\_2\left[Q - \int\_0^Q e^{-b(T)(Q-q)/D} dq\right] = p\_2\left[Q - \frac{D}{b(T)}(1 - e^{-b(T)Q/D})\right],\tag{4}$$

where *n*(*T*) = 1.

*4.1. Retailer's Total Profit per Unit Time*

The retailer's total profit per replenishment cycle is calculated as follows.


Base on Govindan and Malomfalean [29], considering the relationships between retailer and manufacturer, there are three possible cases: (I) Revenue-sharing contract, (II) Buy-back contract, (III) Quantity-flexibility contract.

• Case I: Revenue-sharing contract

Under a revenue-sharing contract, a retailer pays a wholesale price for each unit purchased, and a percentage of the revenue the retailer generates. Therefore, the retailer's total profit function is expressed as.

$$\begin{aligned} TPB(Q) &= SR - OC - PC - FC - GC - MC - HC \\ &= \rho [p\_2 S(Q) - g\_r (D - (\beta - \gamma)(p\_1 + p\_2) - S(Q))] - \frac{A}{n} - w'Q - (F + rQ) \\ &- \frac{1}{2} \frac{Q^2}{D} - c\_r Q, \end{aligned} \tag{5}$$

### • Case II: Buy-back contract

Under the buy-back contract, the manufacturer will buy back any unsold semi-product from the retailer at the end of period. The retailer's total profit function is expressed as.

$$\begin{array}{l} \text{TPB}(Q) &= \text{SR} - \text{OC} - \text{PC} - \text{GC} - \text{FC} - \text{MC} - \text{HC} \\ &= \left[ p\_2 \text{S}(Q) + \beta(Q - \text{S}(Q)) \right] - g\_r \left[ D - (\beta - \gamma)(p\_1 + p\_2) - \text{S}(Q) \right] \\ &- \frac{A}{n} - wQ - (F + rQ) - c\_r Q - \frac{1}{2} \frac{Q^2}{D} .\end{array} \tag{6}$$

• Case III: Quantity-flexibility contract

Under the quantity-flexibility contract, the retailer agrees to acquire (1 − *δ*)*Q* units with the option to restock during the season to optimum quantity points. The agreement constraint is *δ*, 0 ≤ *δ* ≤ 1. With the different types of demand patterns:

Demand is higher than the quantity agreed upon the retailer.

• Case 1: *<sup>D</sup>* <sup>−</sup> (*<sup>β</sup>* <sup>−</sup> *<sup>γ</sup>*)(*p*<sup>1</sup> <sup>+</sup> *<sup>p</sup>*2) <sup>&</sup>gt; ( <sup>1</sup> <sup>−</sup> *<sup>δ</sup>*)*<sup>Q</sup>*

$$TPB(Q) = p\_2S(Q) - c\_r(1 - \delta)Q - \frac{A}{n} - w(1 - \delta)Q - (F + rQ) - \frac{1}{2}\frac{Q^2}{D},\tag{7}$$

Demand is higher that the agreed upon quantity but smaller than the optimal quantity.

• Case 2: ( <sup>1</sup> <sup>−</sup> *<sup>δ</sup>*)*<sup>Q</sup>* <sup>&</sup>lt; *<sup>D</sup>* <sup>−</sup> (*<sup>β</sup>* <sup>−</sup> *<sup>γ</sup>*)(*p*<sup>1</sup> <sup>+</sup> *<sup>p</sup>*2) <sup>&</sup>lt; *<sup>Q</sup>*

$$TPB(Q) = p\_2S(Q) + w[Q - D - (\beta - \gamma)(p\_1 + p\_2)] - c\_r Q - wQ - \frac{A}{n} - (F + rQ) - \frac{1}{2}\frac{Q^2}{D},\tag{8}$$

Demand is higher than the optimum quantity.

• Case 3: *<sup>D</sup>* <sup>−</sup> (*<sup>β</sup>* <sup>−</sup> *<sup>γ</sup>*)(*p*<sup>1</sup> <sup>+</sup> *<sup>p</sup>*2) <sup>&</sup>gt; *<sup>Q</sup>*

$$\begin{array}{l}\text{TPB}(Q) = p\_2\text{S}(Q) - g\_r[D - (\beta - \gamma)(p\_1 + p\_2) - \text{S}(Q)] \\\ -\frac{A}{n} - wQ - (\text{F} + rQ) - \frac{1}{2}\frac{Q^2}{D} - c\_rQ \end{array} \tag{9}$$

*4.2. Manufacturer's Total Profit per Unit Time*

The manufacturer's total profit per replenishment cycle is calculated as follows:


The manufacturer's total profit function at in a replenishment cycle is expressed as

$$\begin{array}{l} \text{TPF}\_{ck}(\mathfrak{n},T) &= \text{SR} + \text{WV} - \text{SC} - \text{PC} - \text{GC} - \text{HC} - \text{C}\_{\text{QD}} \\ &= -\text{g}\_{ck}[(D - (\mathfrak{\beta} - \gamma)(p\_1 + p\_2) - \text{S}(\mathsf{Q}))] - \text{K}\_{\overline{\mathbb{Q}}}^{\overline{\mathbb{D}}} - \text{c}\_{\text{ck}}n\!\!Q + w^{\prime}n\!\!Q \\ &+ (1 - \rho)[p\_2\mathcal{S}(\mathsf{Q}) - \text{g}\_{\text{ck}}(D - (\mathfrak{\beta} - \gamma)(p\_1 + p\_2) - \text{S}(\mathsf{Q}))] \\ &- p\_2\left[\mathcal{Q} - \frac{\mathcal{D}}{b(T)}(1 - e^{-b(T)\mathcal{Q}/D})\right] - \frac{\hbar\_{\overline{\mathbb{D}}}n\!\!p\_{\overline{\mathbb{Q}}}\mathcal{Q}^2}{2}\left(\frac{2-n}{P} + \frac{n-1}{D}\right), \end{array} \tag{10}$$

• CASE II. Buy-back contract

The manufacturer's total profit function at in a replenishment cycle is expressed as *D*

$$\begin{array}{ll} \text{TPF}\_{\text{ck}}(\boldsymbol{\eta},T) &= \boldsymbol{\beta}[\boldsymbol{Q}-\boldsymbol{S}(\boldsymbol{Q})] - \boldsymbol{K}\_{\overline{\mathbf{Q}}}^{\underline{\mathbf{D}}} - \boldsymbol{c}\_{\text{cl}}\boldsymbol{n}\boldsymbol{Q} + \boldsymbol{\omega}\boldsymbol{n}\boldsymbol{Q} - \boldsymbol{g}\_{\text{cl}}[\boldsymbol{D} - (\boldsymbol{\beta} - \boldsymbol{\gamma})(\boldsymbol{p}\_{1} + \boldsymbol{p}\_{2}) - \boldsymbol{S}(\boldsymbol{Q})] \\ &- p\_{2} \left[ \boldsymbol{Q} - \frac{\boldsymbol{D}}{\boldsymbol{b}(T)} (\boldsymbol{1} - \boldsymbol{e}^{-\boldsymbol{b}(T)\boldsymbol{Q}/D}) \right] - \frac{\boldsymbol{h}\boldsymbol{n}\boldsymbol{k}\boldsymbol{p}\boldsymbol{Q}^{2}}{2} \left( \frac{2-\underline{\mathbf{u}}}{\underline{\mathbf{P}}} + \frac{\underline{\mathbf{u}}-\underline{\mathbf{1}}}{\underline{\mathbf{D}}} \right)\_{\text{\textquotedblleft}} \end{array} \tag{11}$$


$$\begin{array}{rcl} \text{TPF}\_{ck}(n,T) &= \textit{wn}(1-\delta)\textit{Q} - \textit{K}\frac{\textit{D}}{\textit{Q}} - \textit{c}\_{\textit{ck}}\textit{n}(1-\delta)\textit{Q} \\ &- p\_{2}\left[\textit{Q} - \frac{\textit{D}}{b(T)}(1-e^{-b(T)\textit{Q}/D})\right] - \frac{h\_{\textit{s}}n\textit{r}\_{\textit{T}}\textit{Q}^{2}}{2}\left[\frac{2-n}{\textit{P}} + \frac{n-1}{\textit{D}}\right], \end{array} \tag{12}$$

2. Case 2: ( <sup>1</sup> <sup>−</sup> *<sup>δ</sup>*)*<sup>Q</sup>* <sup>&</sup>lt; *<sup>D</sup>* <sup>−</sup> (*<sup>β</sup>* <sup>−</sup> *<sup>γ</sup>*)(*p*<sup>1</sup> <sup>+</sup> *<sup>p</sup>*2) <sup>&</sup>lt; *<sup>Q</sup>*

$$\begin{array}{ll} \text{TPF}\_{ck}(\mathfrak{n},T) &= \mathfrak{w}n\mathbb{Q} - \mathbf{K}\_{\mathbb{Q}}^{\mathrm{D}} - \mathfrak{c}\_{\mathrm{ck}}n\mathbb{Q} - \mathfrak{p}\_{2} \left[ \mathbb{Q} - \frac{\mathrm{D}}{b(T)} (1 - e^{-b(T)\mathbb{Q}/D}) \right] \\ &- \frac{\mathfrak{h}\_{\mathrm{s}}n\mathbb{R}\_{\mathbb{T}}\mathbb{Q}^{2}}{2} \left( \frac{2-n}{\mathbb{P}} + \frac{n-1}{\mathbb{D}} \right) . \end{array} \tag{13}$$

$$\text{3.} \quad \text{Case 3: } D - (\beta - \gamma)(p\_1 + p\_2) > Q$$

$$\begin{aligned} \left( \text{TPF}\_{ck}(n, T) \right) \ &= \text{wnQ} - \text{K}\_{\overline{Q}}^{\text{D}} - \text{c}\_{ck} \text{nQ} - \text{g}\_{ck} [\text{Q} - \text{S}(\text{Q})] - p\_2 \left[ \text{Q} - \frac{\text{D}}{b(T)} (1 - e^{-b(T) \text{Q}/D}) \right] \\ &- \frac{\hbar \text{xl}\_{\overline{T}} \text{Q}^2}{2} \left( \frac{2 - \underline{n}}{\overline{P}} + \frac{\underline{n} - 1}{\overline{D}} \right) , \end{aligned} \tag{14}$$

#### *4.3. The Joint Total Profit per Unit Time*

Once the manufacturer and retailer have built a long-term relationship contract, they will jointly determine the best policy for the whole supply chain system. Under this circumstance, for *j* = 1, 2, 3, the joint total profit per unit time for manufacturer and retailer is

$$\Pi\_j(n, \mathcal{Q}, T) = \begin{cases} \Pi\_1(n, \mathcal{Q}, T), \text{ if } \text{Revenue} - \text{sharing contract} \\ \Pi\_2(n, \mathcal{Q}, T), \text{ if } \text{Buy} - \text{back contract} \\ \Pi\_3(n, \mathcal{Q}, T), \end{cases}$$

where

$$\begin{array}{ll} \Pi(\mathfrak{n}, \mathbb{Q}, T)\_1 &= T P B(\mathbb{Q}) + T P F\_{\text{ck}}(\mathfrak{n}, T) \\ &= p\_2 S(\mathbb{Q}) - \frac{A}{n} + (n - 1) w' Q - (\mathbb{F} + r\mathbb{Q}) - K \frac{D}{\mathbb{Q}} - c\_{\text{cl}} n Q - \frac{1}{2} \frac{\mathbb{Q}^2}{\mathbb{D}} - c\_r \mathbb{Q} \\ &- [2g\_{\text{ck}} + \rho(\mathfrak{g}\_{\text{r}} - \mathfrak{g}\_{\text{ck}})] [D - (\mathfrak{f} - \gamma)(p\_1 + p\_2) - S(\mathbb{Q})] \\ &- p\_2 \left[ \mathbb{Q} - \frac{D}{b(\mathbb{T})} (1 - e^{-b(\mathbb{T}) \mathbb{Q} / D}) \right] - \frac{h \eta k\_{\mathbb{T}} \mathbb{Q}^2}{2} \left( \frac{2 - n}{\mathbb{P}} + \frac{n - 1}{\mathbb{D}} \right), \end{array} \tag{15}$$

$$\begin{array}{ll} \Pi(n,Q,T)\_2 &= TPB(Q) + TPF\_{ck}(n,T) \\ &= p\_2S(Q) + 2\beta[Q - S(Q)] - \frac{A}{n} - (F + rQ) - \frac{1}{2}\frac{Q^2}{D} - K\frac{D}{Q} \\ &- (c\_r + c\_{ck}n)Q + w(n-1)Q - (g\_{ck} + g\_I)[D - (\beta - \gamma)(p\_1 + p\_2) - S(Q)] \\ &- p\_2\left[Q - \frac{D}{b(T)}(1 - e^{-b(T)Q/D})\right] - \frac{h\rho k \gamma Q^2}{2}\left(\frac{2-n}{P} + \frac{n-1}{D}\right) \end{array} \tag{16}$$

Nowadays, retailers frequently make challenging decisions such as matching supply with demand. Many retailers are faced with demand uncertainty for product. There are three separate cases (i) *<sup>D</sup>* <sup>−</sup> (*<sup>β</sup>* <sup>−</sup> *<sup>γ</sup>*)(*p*<sup>1</sup> <sup>+</sup> *<sup>p</sup>*2) <sup>&</sup>gt; ( <sup>1</sup> <sup>−</sup> *<sup>δ</sup>*)*Q*, (ii) ( <sup>1</sup> <sup>−</sup> *<sup>δ</sup>*)*<sup>Q</sup>* <sup>&</sup>lt; *<sup>D</sup>* <sup>−</sup> (*<sup>β</sup>* <sup>−</sup> *<sup>γ</sup>*)(*p*<sup>1</sup> <sup>+</sup> *<sup>p</sup>*2) <sup>&</sup>lt; *<sup>Q</sup>* and (iii) *<sup>D</sup>* <sup>−</sup> (*<sup>β</sup>* <sup>−</sup> *<sup>γ</sup>*)(*p*<sup>1</sup> <sup>+</sup> *<sup>p</sup>*2) <sup>&</sup>gt; *<sup>Q</sup>*.

$$\Pi\_3(n,\mathbb{Q},T) = \begin{cases} \Pi\_{31}(n,\mathbb{Q},T), \text{ if } D - (\beta - \gamma)(p\_1 + p\_2) > (1 - \delta)Q \\\Pi\_{32}(n,\mathbb{Q},T), \text{ if } (1 - \delta)\mathbb{Q} < D - (\beta - \gamma)(p\_1 + p\_2) < Q \\\Pi\_{33}(n,\mathbb{Q},T), \text{ if } D - (\beta - \gamma)(p\_1 + p\_2) > Q \end{cases}.$$

where

$$\begin{array}{ll} \Pi\_{31}(\boldsymbol{n}, \boldsymbol{Q}, \boldsymbol{T}) &= p\_2 S(\boldsymbol{Q}) - c\_r (1 - \delta) \boldsymbol{Q} - \frac{\Lambda}{n} - w(1 - \delta) \boldsymbol{Q} - (\boldsymbol{F} + \boldsymbol{r} \boldsymbol{Q}) - \frac{1}{2} \frac{\boldsymbol{Q}^2}{\boldsymbol{D}} \\ &+ wn (1 - \delta) \boldsymbol{Q} - \boldsymbol{K} \frac{\boldsymbol{D}}{\boldsymbol{Q}} - c\_{\mathrm{ck}} n (1 - \delta) \boldsymbol{Q} - p\_2 \left[ \boldsymbol{Q} - \frac{\boldsymbol{D}}{b(\boldsymbol{T})} (1 - e^{-b(\boldsymbol{T}) \boldsymbol{Q}/\boldsymbol{D}}) \right] \\ &- \frac{\mu\_{\mathrm{c}} n \mathrm{k}\_T \mathbb{Q}^2}{2} \left( \frac{2 - n}{\boldsymbol{P}} + \frac{n - 1}{\boldsymbol{D}} \right), \end{array} \tag{17}$$

$$\begin{array}{ll} \Pi\_{32}(\boldsymbol{\eta}, \boldsymbol{Q}, T) &= p\_2 S(\boldsymbol{Q}) - c\_r (1 - \delta) \boldsymbol{Q} - \frac{A}{n} - w(1 - \delta) \boldsymbol{Q} - (\boldsymbol{F} + r\boldsymbol{Q}) - \frac{1}{2} \frac{\boldsymbol{Q}^2}{D} \\ &+ wn(1 - \delta) \boldsymbol{Q} - K \frac{D}{Q} - c\_{ck} n(1 - \delta) \boldsymbol{Q} - p\_2 \left[ \boldsymbol{Q} - \frac{D}{b(T)} (1 - e^{-b(T)Q/D}) \right] \\ &- \frac{h\_2 n k\_T Q^2}{2} \left( \frac{2 - n}{P} + \frac{n - 1}{D} \right) , \end{array} \tag{18}$$

$$\begin{array}{lcl} \text{and} & \text{and} \\ \Pi\_{33}(\boldsymbol{n}, \boldsymbol{Q}, \boldsymbol{T}) & = p\_2 \mathcal{S}(\boldsymbol{Q}) - \mathcal{g}\_r [\boldsymbol{D} - (\boldsymbol{\beta} - \boldsymbol{\gamma})(\boldsymbol{p}\_1 + \boldsymbol{p}\_2) - \mathcal{S}(\boldsymbol{Q})] - \frac{A}{n} - w\mathcal{Q} - (\boldsymbol{F} + r\boldsymbol{Q}) - \frac{1}{T}\frac{\boldsymbol{Q}^2}{D} - \boldsymbol{c}\_r \boldsymbol{Q} \\ & + w\mathcal{Q} - \mathcal{K}\frac{\boldsymbol{D}}{\mathcal{Q}} - \boldsymbol{c}\_{\mathcal{C}\mathcal{E}}\boldsymbol{n}\mathcal{Q} - \mathcal{g}\_{\mathcal{E}\mathcal{E}}[\boldsymbol{Q} - \mathcal{S}(\boldsymbol{Q})] - p\_2 \left[\mathcal{Q} - \frac{\boldsymbol{D}}{b(\boldsymbol{T})}(1 - \boldsymbol{\varepsilon}^{-b(\boldsymbol{T})\mathcal{Q}/D})\right] \\ & - \frac{h\_1 n \mathcal{k}\_\Gamma \mathcal{Q}^2}{2} \left(\frac{2-n}{\mathcal{P}} + \frac{n-1}{D}\right). \end{array} \tag{19}$$

#### **5. Solution Procedure**

*∂*

A quantization design was used to find feasible and optimal solutions. The objective is to determine the optimal the number of shipments, lot size per shipment, and temperature at central kitchen that maximizes the joint total profit per unit time of the integrated supply chain. First, the effect of n on the joint total profit per unit time will be examined. Taking the second-order partial derivative of Π*j*(*n*, *Q*, *T*) with respect to *n*, we obtain

$$\frac{\partial^2 \Pi\_j(n, Q, T)}{\partial n^2} = -\left[2An^{-3} + h\_s k\_T Q^2 \left(\frac{P - D}{PD}\right)\right] < 0,\tag{20}$$

This result identifies that Π*j*(*n*, *Q*, *T*) is a concave function in *n* for fixed *Q* and *T*. Therefore, the search for the optimal shipment number, *n* ∗ , is reduced to find a local optimal solution.

#### *5.1. Determination of the Optimal T for Given n and Q*

For given *n* and *Q*, the first-order necessary condition of Π*j*(*n*, *Q*, *T*), *j* = 1, 2, 3, with respect to *T*. This gives

$$\frac{\partial \Pi\_j(\mathbf{u}, \mathbb{Q}, T)}{\partial T} = \frac{e^{-b(T)Q/D}}{D[b(T)]^2} b'(T) \left[ Qb(T) - D(e^{b(T)Q/D} - 1) \right] = 0, \; j = 1, 2, 3. \tag{21}$$

Next, we start by taking the second derivative of Π*j*(*n*, *Q*, *T*).

$$\begin{array}{rcl} \frac{\partial^2 \Pi\_{\overline{l}}(n, Q, T)}{\partial T^2} &=& \frac{1}{D^2 \left[ b(T) \right]^3} e^{-b(T)Q/D} \left\{ D \left[ D \left( e^{b(T)Q/D} - 1 \right) - Qb(T) \right] \left[ 2(b^\prime(T))^2 - b(T)b^\prime(T) \right] \right. \\ & \left. - Q^2 \left[ b(T) \right]^2 \left[ b^\prime(T) \right]^2 \right\} . \end{array} \tag{22}$$

By fixing *n* and *Q*, we have the following result.

#### **Proposition 1.** *For given feasible n and Q*,


#### *5.2. Determination of the Optimal Q for Given n and T*

In this section, we first find the optimal lot size per shipment that maximizes Π*j*(*n*, *Q*, *T*), *j* = 1, 2, 3, respectively, for a given *n* and *T*. Then the optimal number of shipments is derived for a given *Q*. Taking the first-order partial derivative of Π*j*(*n*, *Q*, *T*), *j* = 1, 2, 3, with respect to *Q*, respectively, we have

$$\begin{split} \frac{\partial \Pi\_1(\mathbf{n}, \mathbf{Q}, T)}{\partial \mathbf{Q}} &= p\_2 S'(\mathbf{Q}) + (n - 1)w' - r + KD\mathbf{Q}^{-2} - c\_{ck}n - \frac{\mathbf{Q}}{D} - c\_r \\ &+ [2g\_{ck} + \rho(\mathbf{g}\_r - \mathbf{g}\_{ck})]S'(\mathbf{Q}) - p\_2 \left[1 - e^{-b(T)Q/D}\right] \\ &- \mathbf{h}\_{s} n k\_T \mathbf{Q} \left(\frac{2 - n}{P} + \frac{n - 1}{D}\right) . \end{split} \tag{23}$$

$$\begin{array}{lcl}\frac{\partial \Pi\_{2}(\boldsymbol{n},\boldsymbol{Q},\boldsymbol{T})}{\partial \boldsymbol{Q}} &= (p\_{2} + g\_{ck} + g\_{r} - 2\beta)S'(\boldsymbol{Q}) + 2\beta - r - \frac{\boldsymbol{Q}}{D} + K\frac{\boldsymbol{D}}{\boldsymbol{Q}^{2}}\\ &- (\boldsymbol{c}\_{r} + \boldsymbol{c}\_{ck}\boldsymbol{n}) + w(n-1) - p\_{2} \left[1 - e^{-b(\boldsymbol{T})\boldsymbol{Q}/D}\right] \\ &- h\_{s}n k\_{T} \left(\frac{2-n}{P} + \frac{n-1}{D}\right),\end{array} \tag{24}$$

$$\begin{split} \frac{\partial \Pi\_{31}(\boldsymbol{n}, \boldsymbol{Q}, \boldsymbol{T})}{\partial \boldsymbol{Q}} &= p\_2 \boldsymbol{S}'(\boldsymbol{Q}) - c\_r (1 - \delta) - w(1 - \delta) - r - \frac{\boldsymbol{Q}}{\boldsymbol{D}} + wn(1 - \delta) + \boldsymbol{K} \frac{\boldsymbol{D}}{\boldsymbol{Q}^2} \\ &- c\_{ck}n(1 - \delta) - p\_2 \left[1 - e^{-b(\boldsymbol{T})\boldsymbol{Q}/\boldsymbol{D}}\right] - h\_s n k\_T \boldsymbol{Q} \left(\frac{2 - \boldsymbol{\mu}}{\boldsymbol{P}} + \frac{\boldsymbol{n} - 1}{\boldsymbol{D}}\right), \end{split} \tag{25}$$

$$\begin{array}{ll} \frac{\partial \Pi\_{32}(\boldsymbol{\mu}, \boldsymbol{Q}, \boldsymbol{T})}{\partial \boldsymbol{Q}} &= p\_2 \boldsymbol{S}'(\boldsymbol{Q}) - \boldsymbol{c}\_r - \boldsymbol{r} - \frac{\boldsymbol{Q}}{\boldsymbol{D}} + \boldsymbol{w}\boldsymbol{n} + \boldsymbol{K} \frac{\boldsymbol{D}}{\boldsymbol{Q}^2} - \boldsymbol{c}\_{ck}\boldsymbol{n} - p\_2 \left[1 - e^{-b(\boldsymbol{T})\boldsymbol{Q}/\boldsymbol{D}}\right] \\ &- \boldsymbol{h}\_{\boldsymbol{S}}\boldsymbol{n} k\_{\boldsymbol{T}} \boldsymbol{Q} \left(\frac{2-\boldsymbol{n}}{\boldsymbol{P}} + \frac{\boldsymbol{n}-1}{\boldsymbol{D}}\right), \end{array} \tag{26}$$

$$\begin{array}{c} \frac{\partial \Pi\_{33}(\mathbf{n}, \mathbf{Q}, T)}{\partial \mathbf{Q}} \ \ = (p\_2 + g\_I)S'(\mathbf{Q}) - w - r - \frac{\mathbf{Q}}{\mathbf{D}} - \mathbf{c}\_r + w\mathbf{n} + \mathbf{K}\mathbf{Q}\mathbf{Q}^{-2} - \mathbf{c}\_{\mathbf{c}k}\mathbf{n} \\ \ \quad - g\_{c\mathbf{k}}[1 - S'(\mathbf{Q})] - p\_2 \left[1 - e^{-b(T)Q/D}\right] - h\_{\mathbf{c}}nk\_TQ\left(\frac{2-\mathbf{n}}{\mathbf{P}} + \frac{\mathbf{n}-1}{\mathbf{D}}\right), \end{array} \tag{27}$$

Next, we take the second-order partial derivative of Π*j*(*n*, *Q*, *T*)

$$\begin{array}{rcl} \frac{\partial^2 \Pi\_1(n, Q, T)}{\partial Q^2} &=& -2KDQ^{-3}[2g\_{ck} - \rho(g\_r - g\_{ck})]S''(Q) - \frac{1}{D} \\ & - \frac{p\_2}{D} \left[ b(T)e^{-b(T)Q/D} - 1 \right] - h\_s \eta k\_T \left( \frac{2-n}{P} + \frac{n-1}{D} \right), \end{array} \tag{28}$$

$$\frac{\partial^2 \Pi\_2(\mathfrak{n}, \mathbb{Q}, T)}{\partial \mathbb{Q}^2} = (p\_2 + g\_{c\mathbb{k}} + g\_r - 2\beta) \mathbb{S}''(Q) - 2KD\mathbb{Q}^{-3} - p\_2 \frac{b(T)}{D} e^{-b(T)\mathbb{Q}/D},\tag{29}$$

$$\frac{\partial^2 \Pi\_{31}(n, Q, T)}{\partial Q^2} = p\_2 \left[ S^\nu(Q) - b(T)e^{-b(T)Q/D} \right] - K \frac{D}{Q^3} - 1 - h\_s n k\_T \left( \frac{2 - n}{P} + \frac{n - 1}{D} \right),\tag{30}$$

$$\frac{\partial^2 \Pi\_{32}(n, Q, T)}{\partial Q^2} = p\_2 \left[ S''(Q) - \frac{b(T)}{D} e^{-b(T)Q/D} \right] - \frac{1}{D} - 2\frac{KD}{Q^3} - h\_\delta n k\_T \left( \frac{2-n}{P} + \frac{n-1}{D} \right), \tag{31}$$

*∂* <sup>2</sup>Π33(*n*,*Q*,*T*) *<sup>∂</sup>Q*<sup>2</sup> = (*p*<sup>2</sup> + *<sup>g</sup><sup>r</sup>* + *<sup>g</sup>ck*)*<sup>S</sup>* ′′(*Q*) − *p*<sup>2</sup> h *b*(*T*)*e* <sup>−</sup>*b*(*T*)*Q*/*<sup>D</sup>* <sup>−</sup> <sup>1</sup> i 1 *<sup>D</sup>* <sup>−</sup> <sup>2</sup>*KDQ*−<sup>3</sup> −*hsnk<sup>T</sup>* 2−*n <sup>P</sup>* <sup>+</sup> *<sup>n</sup>*−<sup>1</sup> *D* . (32)

By fixing *n* and *T*, we have the following results.

#### **Proposition 2.** *For given n and T*,


#### **6. Application Example**

The selected bakery is 2017′ s fastest-growing company (Ding-Dang convenience store) in Huizhou, Guangdong Province. The company's brand is derived from the "Internet +" wave and delivers high-quality services to consumers under the new retailing framework. Considering that it is a small and medium-sized enterprise, its founding team originated from Alibaba's and Sf-express's CEO. This company owns a new retailing framework with online marketing (digital marketing) and offline marketing (convenience store marketing), the number of platform users reached 600,000, and the number of service outlets reached 100 in Huizhou. The case company raised more than CN¥200 million of annual revenue, and the plan is to expand convenience stores in 16 new cities. In this section, we constructed two scenarios that consist of new retailing and food supply chain.

#### *6.1. New Retailing Framework*

This section presents an implementation of the proposed model in a real-life bakery logistics network operating in China. Nowadays, using artificial intelligence and big data to change the business model of the bakery industry in China, Hsiung Mao (Panda) Pu Tsou (called "case company") uses social commerce platforms to offer online group buying on a social commerce platform. The main products, five-star cakes, were made of the best raw materials in the medium and high-end market. The case company will be a Personal Service Provider (PSP) via Panda dressed up, Panda distribution. The customers make e-commerce purchases through the Wechat, Meituan, and Elema app ordering systems. Correct handling of cakes, ingredients, and packaging materials from material purchasing occurred through the production process and cold storage in the center kitchen. During Panda distribution to the consumer, it is essential to optimize cake quality. The traditional O2O models are advanced to a new retailing framework by integrating online and offline logistics. The core of the new retailing framework is online mobile payment, product standardization in the central kitchen, distributed personalization, and providing big data services.

• Online mobile payment

The formation of community-owned stores through online–offline channel integration. The case company has opened more physical stores to attract young customers to achieve the new retailing framework. Online Quick Reference Guide, an easy access tool, provides customers to order. Besides that, online customer satisfaction or product categories provide customers more flexible options.

• Product standardization in central kitchen

Four-Tier model of food supply chain consists of central kitchen, cold chain, branch warehouse, Panda distribution, and authorized store (Self-raising). Then, customer orientation plays an important role in market success achievement. Product customization helps brands boost sales on new retailing framework, EX: Sculpture cakes or fictionalized cakes. A rise in cold chain with data warehouse and central kitchen system will enhance the quality of cake, reduce the waiting time. Mercier et al. [12] discussed a significant challenge for an efficient clod chain is the different requirements of perishable food product categories (dairy, eggs, fruits, cakes, and vegetables) to maximize shelf-life and commercial potential, including different optimal temperature ranges (ambient = 15 ◦C to 20 ◦C; cool = 2 ◦C to 15 ◦C; cold = −9 ◦C to 2 ◦C; and frozen ≤ −10 ◦C).

• Immediate distribution for personal services

On-time 3 h delivery services include online order, materials handling, finished product manufacturing, branch's self-storage, and surprise services. The case company provides customers with activity planning, including surprise events. Traditional Hakka culture is incorporated into a personal-service activity ex: Traditional Hakka cake, the unique Hakka cultural experience.

− −

• Big data as a service (BDaaS)

In order to find the target customers through big data analytics, online platform could not only help case company to better discriminate the most profitable customers, but also encourage more customer to return. Offline services can attract customers online and offer precision marketing based on a customer's exact age and birth date. Since marketing is all about reaching the right customers at the right time, big data can be used to understand the behavior of bakery customers.

• Customer image capture

For Big data as a Service (BDaaS) and Internet Presence Provider (IPP), customer data analytics refer to the analysis of customer characteristics and behaviors so as to support the organization's customer management strategies. Customer image, online marketing, and site selection are used to revolutionize offline shopping experiences. Data collected from e-payment system can develop the marketing strategy to attract customers to return.

In view of the preceding decision variables, there were few empirical studies of new retailing framework. Figure 1 shows the cake supply chain summary. The food supply chain-case study is then presented, with a thorough description of cold chain Hub.

**Figure 1.** New retailing business model.

#### *6.2. Food Supply Chain-Case Study*

The food supply chain (FSC) contains several members, including one central kitchen and numerous branch warehouses. The FSC consists of a four-tier structure such as a central kitchen, cold distribution services, branch warehouse, food-panda delivery, and customer. For quick service retailers, the distribution of cakes from the central kitchen to branch warehouses is throughout the cold chain distribution process, which is normally at a constant temperature between 0 ◦C and 5 ◦C. To improve distribution efficiency, the case company owns three reefer vehicles to transport cake and cake products from central kitchen to branch warehouse (72 cakes per vehicle; 8 rounds per day; transport unit at 1700 units per day). Moreover, Panda distribution hit a 90% completion rate from branch warehouse to customer and 10% completion rate from branch warehouse to an authorized store (self-raising). Four branches can own 30 Panda vehicles and 8 electric vehicles. Maximum quantity is 800 units per delivery, 24 units per delivery for one Panda vehicle,

12 units per delivery for one electric vehicle. The simple Figure 2 seeks to capture the cake supply chain summary.

**Figure 2.** Cake supply chain summary.

Chen et al. [62] demonstrated the reliability changes of performance parameters with the importance measures in inventory systems. Few empirical studies have been done on this issue. The examination to carry out this study was using an algorithm, which solve optimal solutions between a single manufacturer and single retailer.

#### *6.3. Alogorithm*


#### **7. Numerical Example**

\* \*\* (, , ) In this section, we consider the numerical example with different situations to verify the obtained analytical results. The base settings of parameters for the example are listed in Table 2. Then the sensitivity analysis is also tabulated for exploring the variety trend of optimal polies.

**Table 2.** The value of parameters for Example 1. with new retailing problem.

\*\* \* \*


#### *7.1. Comparison among Decisions in Online and Offline Stratedgy*

The decisions made in the two O2O strategies are compared in this section. A coordination mechanism was designed to clarify the relative contribution on the supply

chain, and the three known contracts were compared to explore the application of O2O strategy. Assessment of performance measures occurred under mixture policy (online and offline) and single policy (offline). The results represented in Table 3 present the result of comparison of Revenue-sharing agreement (Case I), Buy-back contract agreement (Case II), and Quantity flexibility contract (Case III).


**Table 3.** Computation results of Example 1. for online and offline.

The retailer, the manufacturer, and total supply chain profit in the three O2O policies are presented in Figures 3 and 4. Figure 3 presents revenue-sharing agreement larger than other polies. It implied that revenue-sharing contracts in a general supply chain model with revenues are determined by each retailer's purchase quantity and price.

**Figure 3.** Mixture situation measures under three policies.

Obviously, Figure 4 presents the performance of offline policy less than mixture policy. It implied that the channel members would like to adopt new online technologies in a smaller, safe environment with the retailer and manufacturer before they scaled out to retail partners.

\* \*\* (,,) − −

<sup>n</sup> TP<sup>B</sup> TPFck \* Q

*g*

\* \* \* \* \*\* (,,)

\* T

\*

n TP<sup>B</sup> TPFck

\* T

\*

\* Q \* T

− −

− − \*

<sup>n</sup> TP<sup>B</sup> TPFck \* Q

**Figure 4.** Offline situation measures under three policies.

#### *7.2. Sensitivity Analysis*

*g* \* \* \* \* \*\* (,,) \* \*\* (,,) − − To perform data analysis, the relative contributions of these parameters (*gck*, *g<sup>r</sup>* , *F*, *r*, *c<sup>r</sup>* , *ρ*, *Tc*, *m*) were calculated on the values of *Q*<sup>∗</sup> , *T* ∗ , *n* ∗ , *TPB*(*Q*<sup>∗</sup> , *T* ∗ , *n* ∗ ) and *TPFck*(*Q*<sup>∗</sup> , *T* ∗ , *n* ∗ ). Each parameter was adjusted separately by +50%, +25%, −25%, or −50%. Our analytical results are in Table 4.


**Table 4.** Effect of changes in various parameters of the model.


**Table 4.** *Cont*.

Note: Para. = Parameter; % = Change %.

#### *7.3. The Discussion of Results*

This section will compare and analyze the three contracts. The data analysis has highlighted a most interesting possibility: Reputation management and transportation performance. Several management implications can be drawn from Table 5. Marketing policies are changes in sales or other results that can be expected from a particular strategy. For example, ensuring food quality and security during the transportation and postponing strategy.

**Table 5.** Effect of changes in various parameters of managerial insights.



**Table 5.** *Cont*.

#### **8. Conclusions**

Due to the evolution of O2O technologies, the coordination mechanism will be incorporated into the classic inventory model. The authors develop a food supply chain among one manufacturer and one retailer under the O2O model in this research. The study also compared the three coordination mechanisms to fill the gap knowledge between theoretical and empirical in terms of inventory management. Relying on the existing theoretical results, this study has attempted to point out the optimal solution through examining decision-makers with an algorithm. Therefore, the authors can assert based on having understood the shortcomings of O2O technology that our theoretical claim about how a manufacturer and its compromising retailer evaluate a suitable ordering process through a comparison of three commitment contracts. Meanwhile, there are some potential bottlenecks in the development of the O2O model, including (1) users and offline sellers lack rational knowledge of mobile e-commerce; (2) the security of e-payments; and (3) the lack of innovations. Moreover, the findings have been proved by several examples to clarify the proposed model and solution. A total of the proof examples highlighted that (1) Mobile payment lays the foundation for online/offline integration strategy. With advancements in e-payment systems and face recognition/fingerprint recognition, both companies and consumers discover new technology to pay online quickly; (2) the IoT-Driven cold chain tracking is reducing logistics cost and improving business effectiveness, (3) Instant distribution impact on customer online satisfaction. Among the many topics to be explored in future research, some important ones are as follows: Estimating an effective after-seller's services mechanism, strengthening consumers' loyalty to offline sellers, and improving the innovative capacity of the O2O platform.

**Author Contributions:** Conceptualization, Y.-F.H. and N.X.; methodology, M.-W.W.; software, M.- W.W.; validation, Y.-F.H., N.X., M.-W.W., and M.-H.D.; formal analysis, N.X.; investigation, N.X.; resources, N.X.; data curation, N.X.; writing—original draft preparation, N.X., M.-W.W., and M.-H.D.; writing—review and editing, N.X., M.-W.W., and M.-H.D.; visualization, Y.-F.H.; supervision, Y.-F.H.; project administration, Y.-F.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Due to its proprietary nature <or ethical concerns>, supporting data cannot be made openly available. Further information about the data and conditions for access are available at http://www.xiongmaodangao.com/.

**Acknowledgments:** The authors would like to thank the editor and anonymous reviewers for their valuable and constructive comments, which have led to a significant improvement in the manuscript. The author's research was partially supported by the Ministry of Science and Technology of the Republic of China under Grant MOST 109-2622-H-324-001.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**

In addition: the following notations and assumptions are used throughout this article.

#### **System Parameters**



#### **References**


**Ying Liu 1,2, Huan Cheng <sup>2</sup> and Dan Wu 2,\***


**Featured Application: The preparation technology of fruit-flavoured "boba" balls can be applied by food ingredient companies. Through the further optimization of the fruit ball core formula, healthy substances (such as flavones, active proteins, sterols, enzymes, probiotics, and vitamins) can be embed-ded in fruit balls and applied to drinks, overcoming the traditional technical bottleneck, en-hancing the stability and compatibility of the substances, and effectively extending the shelf life of the products.**

**Abstract:** Boba milk tea is very popular around the world. The "boba" balls in milk tea are usually made of tapioca. Reports on calcium alginate ball encapsulation in fruit-flavoured drinks have rarely been seen. The preparation method for this kind ball was studied. The "boba" balls were obtained by membrane formation on the interface through the addition of calcium chloride fluids into a sodium alginate solution. The operation conditions were studied, including drop height, flow velocity, sodium alginate and calcium chloride solution concentration. The diameter, mechanical strength, loading ratio and encapsulation rate of the "boba" balls are discussed. The optimized preparation conditions were as follows: the diameter of adding tube was 8 mm, the drop height was 25 cm, the drop flow rate was 60 mL/min, 1.0% sodium alginate, 1.0% calcium chloride. The prepared "boba" balls were stored at different temperatures. No microorganisms were detected in 90 days, and the sensory quality decreased with storage time. Shelf life was predicted using the Arrhenius equation; when the storage temperature was less than 10 ◦C, it could be stored for more than 1 year. This preparation technology of "boba" balls has potential for application by milk tea ingredient companies or relevant beverage manufacturing factories.

**Keywords:** boba milk tea; calcium alginate ball; preparation method; shelf life

#### **1. Introduction**

Boba milk tea originated in Taiwan in the 1980s, became very popular in Asia in the 1990s, and has been increasingly popular around the world since 2000 [1,2]. The "boba" balls, which are also called "pearl" balls, are the central ingredient of the milk tea. It is usually made of tapioca, or tapioca mixed with other ingredients, such as konjac powder, sweet potato, chamomile, and so on [3–5]. There have been some studies on the calory content and obesity risk of milk tea [6], but there have been few reports related to the improvement of milk tea technology, including that related to the "boba" balls.

Orange is rich in nutrition. It not only contains nutrient elements needed by the human body, it also contains plant phenols and flavonoids, containing an especially high content of Vc [7,8]. It is one of the most popular fruits in the world. Neves, Trombin, and Marques et al. analysed the global consumption of frozen concentrate orange juice from 2013 to 2018. They summarized the consumer behaviour of three kinds of orange juice in the main market worldwide [9]. A large number of studies have shown that eating

**Citation:** Liu, Y.; Cheng, H.; Wu, D. Preparation of the Orange Flavoured "Boba" Ball in Milk Tea and Its Shelf-Life. *Appl. Sci.* **2021**, *11*, 200. https://doi.org/10.3390/app11010200

Received: 10 December 2020 Accepted: 24 December 2020 Published: 28 December 2020

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/ licenses/by/4.0/).

a large amount of citrus fruits is beneficial to anti-cancer functions and prevention of cardiovascular diseases [10–13].

In this project, orange juice was wrapped in calcium alginate film to obtain fruit balls, which were used in milk tea, forming a new type of fruit ball drink. This technique was reported in cell immobilization during the last century. For example, Rickert et al. used calcium alginate gel to embed bacteria for the fermentation of organic acids [14]. By dropping Calcium ions (Ca2+) into sodium alginate solution, Ca2+ diffused to the periphery and reacted with sodium alginate to form a gel film to wrap the liquid core, and hollow micro-capsules were prepared. Some scholars have performed studies on the influence of processing conditions of calcium alginate gel (CAG). The physical properties of CAG beads change with heat treatment, including rupture strength, size, sphericity and porous structures [15,16].

In general, calcium alginate gel is a solid gel; the CAG studied in this paper has an orange-flavoured drink core. When you bite it, the juice flows out of the inner core, like jam, giving people a sweet and sour, delicious, unprecedented eating experience. It is also a calcium alginate immobilized embedding technology, which has the potential to maintain the stability of sensitive food ingredients, especially some natural health ingredients, such as flavones, active proteins, sterols, enzymes, probiotics and vitamins, etc. By applying calcium alginate ball embedding methods, important breakthroughs could result in some key food industry technologies, thus achieving ideal sensory, physical and chemical indicators, and shelf life, while also achieving functional food design. For example, Lin et al. wrapped astaxanthin in the CAG. The results showed that after storage at 25 ◦C for 21 days, the contents of astaxanthin in the CAG both remained above 90% of their original amount [17]. This provides an effective way of developing the stability of sensitive compounds. Jiang et al. fixed bifidobacteria in carboxymethyl chitin–calcium alginate gel, the bifidobacteria could be delivered to the gut without being dissolved in the gastric juice [18]. Here, "boba" balls made of calcium alginate for use in milk tea were studied. Operational conditions such as drop height, flow velocity, and sodium alginate and calcium chloride solution concentration were investigated. The diameter, mechanical strength, loading ratio and encapsulation rate of the "boba" balls are discussed. The sensory and microorganisms changes of the "boba" balls at different temperatures were studied, and their shelf life is predicted by using the Arrhenius equation.

#### **2. Materials and Methods**

#### *2.1. Materials*

Orange juice (sugar: 66◦Bx, acidity: 4.16%), modified starch (hydroxypropyl distarch phosphate from potato, stable to temperature, acidity and shear force), calcium chloride anhydrous and sodium alginate were provided by Hangzhou Bodo Industry and Trade Co., LTD., Hangzhou, China. Sucrose and table salt were bought from a market; citric acid anhydrous was purchased from Shandong Zhongchuang Lemon Biochemical Co., LTD., Anqiu, Shandong, China. Sunset yellow, lemon yellow, and orange essence were purchased from Hangzhou Lvjing Flavor Co., LTD., Shandong, China, Xanthan gum and Guar bean gum were purchased from Hebei Jinfeng Chemical Co., LTD., Hengshui, Hebei, China. The companies above were located in China. All of the above materials were of food grade.

#### *2.2. The "boba" Ball Preparation*

The composition of the orange-flavoured drink, which was the core of the "boba" ball, was carried out with reference to Wu et al. [19] and was improved for processing. It included sucrose (22.80%), citric acid anhydrous (0.50%), salt (0.04%), sunset yellow (0.0012%), citric yellow (0.006%), orange essence (0.07%), Xanthan (0.1%), and guar gum (0.25%). All the ingredients were put into pure boiling water (100 ◦C) to be fully dissolved, after which the mixture was cooled to 50–60 ◦C, and concentrated orange juice (5%), modified starch (3.75%), calcium chloride anhydrous (1%) were added and blended

together. Homogenizer (Shanghai Zollo Instrument Co., Ltd., Shanghai, China) was used to homogenize the orange flavoured drink at the pressure of 18–22 mpa for 30 min. process of the "boba" ball is shown in

–

–

The preparation process of the "boba" ball is shown in Figure 1. The orange fruitflavoured drink was added to the sodium alginate solution drop by drop using the Automatic liquid filler (Newking Pump Co., Ltd., Shangdong, China). After 1 min of reaction, calcium alginate balls were filtered out with a stainless-steel filter spoon, and were washed with distilled water and then transferred into 5% calcium chloride solution to react for 15 min. Finally, the residual calcium chloride outside of the "boba" ball was washed off with distilled water again, and the preparation was completed. All the above operations were carried out at room temperature. The prepared "boba" balls were stored in 0.9% sodium chloride solution for property detection. react for 15 min. Finally, the residual calcium chloride outside of the "boba" ball was prepared "boba" balls were stored

Schematic diagram of "boba" ball preparation. **Figure 1.** Schematic diagram of "boba" ball preparation.

#### *2.3. Preparation of "boba" Ball Storage Solution*

*Preparation of "boba" Ball Storage Solution* dimensional network structure, known as an "egg box" strucpH greater than 5.5 [21]. Therefore, "boba" balls must be stored in environments with pH soluble substances in the "boba" balls. To prevent the loss of solute and maintain the pH of the environment, the formulation of the "boba" ball storage solution was CAG showed a three-dimensional network structure, known as an "egg box" structure, that enables small molecules of water-soluble substances to pass through freely [20]. The swelling property of CAG was affected by pH value and stabilised at solution pH < 6. This indicates that calcium alginate would gradually decompose in salt solution with pH greater than 5.5 [21]. Therefore, "boba" balls must be stored in environments with pH values less than 5.5. There were some sugar, acid, pigment, essence and other small molecule water-soluble substances in the "boba" balls. To prevent the loss of solute and maintain the pH of the environment, the formulation of the "boba" ball storage solution was prepared as follows: sucrose 22.80%, citric acid 0.50%, salt 0.04%, sunset yellow 0.0012%, citric yellow 0.006%, orange essence 0.07%. The pH value of the storage solution was measured to be about 2.05 ± 0.05 using a pH meter (INESA Scientific Instrument Co., Ltd., Shanghai, China).

#### *2.4. Determination of "boba" Ball Properties*

*Determination of "boba" Ball Propertie* Measurement of diameter: 20 capsules were randomly selected, the diameters were measured one by one using Vernier calipers (Shanghai Meinite Industrial Co., Ltd., Shanghai, China), and the average value was taken [22].

measurement. A "boba" ball was placed on the tray of the analytical balance (Mettler Toledo, LLC, Shanghai, China), and positive pressure was applied to the "boba" ball until Determination of mechanical strength: The single-axis pressing method was used for measurement. A "boba" ball was placed on the tray of the analytical balance (Mettler-Toledo, LLC, Shanghai, China), and positive pressure was applied to the "boba" ball until it burst. The balance showed a sudden change of pressure from low to high, and then to low; the maximum positive pressure was recorded, namely the mechanical strength of the "boba" ball. Twenty "boba" balls were randomly selected for each detection, and the mechanical strength was represented by the average value [20].

The orange fruit-flavoured drink loading ratio and encapsulation rate:

Loading ratio = Total weight of orange fruit-flavoured drink in the"boba" ball/Total weight of the "boba" ball

(1)

Encapsulation rate = Total weight of orange fruit-flavoured drink in the "boba" ball/Total weight of orange fruit-flavoured drink (2)

> The encapsulation rate is an important parameter in the preparation of "boba" balls. The higher, the better.

#### *2.5. Shelf Life Determination and Prediction*

The "boba" balls were placed in a sealed glass jar by adding the storage solution, pasteurized at 70–80 ◦C for 30 min, and were then stored at 20 ◦C, 25 ◦C, 30 ◦C, 35 ◦C and 40 ◦C, respectively.

Total bacterial colony determination was conducted with reference to GB 4789.2— National Standard Food Microbiology Examination for Food Safety [23].

The sensory evaluation indexes included colour, aroma, taste and appearance, and the standards are shown in Table 1. The panel of 12 assessors was aged from 20 to 36 years and possessed expertise in the sensory evaluation of food. Approximately 20 "boba" balls stored at a constant temperature were taken out, kept in a plate at room temperature for 30 min, and served to each panellist along with the questionnaire. The samples were presented one at a time, with about a 5 min wait between samples. The intervals for sensory evaluation of the "boba" balls varied at different storage temperatures: balls stored at 20–30 ◦C were evaluated every 5–9 days, and balls stored at 35 ◦C and 40 ◦C were evaluated every 1–2 days. The intensities of the sensory attributes are presented as the mean of the scores provided by the 12 panel assessors. If the sensory score was below 80, the product was considered to be unable to meet consumers' expectations and reach the shelf-life limit.


**Table 1.** Standard of sensory evaluation.

According to [24,25], the sensory evaluation response function F(X) at a certain temperature can be expressed as follows:

$$\mathbf{F}(\mathbf{X}) = \mathbf{k} \,\mathrm{d}\mathbf{t} \tag{3}$$

where k is the sensory evaluation change reaction rate, and t is the period for which the food was stored. Here, the k value at a certain temperature can be derived from the slope of the regression equation between the response value F(X) and time. The effect of temperature on the change reaction of sensory evaluation can described by the Arrhenius relationship. Taking the logarithm on both sides of the Arrhenius function:

$$
\ln \mathbf{k} = -\; \text{Ea} / \text{RT} + \ln \text{k}\_0 \tag{4}
$$

where Ea is the activation energy in J/mol, R is the gas constant in J/(K mol) and is equal to 8.314, k<sup>0</sup> is a pre-exponential factor, and T is absolute temperature in K (273 ◦C).

Here, the Ea value and k<sup>0</sup> can be derived from the slope of the regression equation between ln k and 1/T.

#### *2.6. Statistical Analysis*

The Data Processing System (DPS) software v13.5 was applied to fix the experiment data and establish the mathematical model [26]. The Origin software (OriginLab Corporation, Northampton, MA, USA) was applied to paint the response surface.

"boba" –

−

#### **3. Results and Discussion** "boba" ball pass

#### *3.1. Selection of Hose Diameter*

The diameter of the straws used for "boba" milk tea is generally about 10.00–12.00 mm. To allow the "boba" ball pass through the straw smoothly, the diameter of the prepared balls generally needs to be 1.00–2.00 mm lower than this. The objective of this project was to prepare "boba" balls with a diameter of about 8–10 mm; therefore, it was advisable to use a hose with a diameter of 8 mm. – "boba" balls –

#### *3.2. Influence of Drop Height* shows the influence of drop height on the "boba" balls. The velocity of the

Figure 2 shows the influence of drop height on the "boba" balls. The velocity of the automatic liquid filler was 60 mL/min, the sodium alginate solution was 0.8%. It was found that if the drop height wasn't high enough, calcium alginate bal would easily float on the surface, and the final ball was often elliptical (see the drop height of 5 cm in Figure 2); with increasing drop height, the probability of elliptical balls gradually decreased (see the drop height 5–20 cm in Figure 2). When the drop height was greater than 20 cm, the "boba" balls were basically spherical. Therefore, it is advisable to choose a drop height greater than 20 cm. The drop height selected here was 25 cm. found that if the drop height wasn't high enough, calcium alginate bal would easily fl – "boba" balls were basically spherical. Therefore, it is advisable to choose a drop height

the "boba" ball. **Figure 2.** Effects of drop height on the "boba" ball.

#### *3.3. Influence of Flow Velocity*

present the effect of flow velocity on the "boba" ball properties. on the diameter of the "boba" balls (Table 2). When the flow rate on alginate micro spheres [27]. These balls wouldn't be applied to the milk tea, as they "boba" with suitable diameters of between 8– – Table 2 and Figure 3a present the effect of flow velocity on the "boba" ball properties. The diameter of the hose was 8 cm, the drop height was 25 cm, and the concentration of sodium alginate solution was 0.8%. It was found that the flow velocity had a positive effect on the diameter of the "boba" balls (Table 2). When the flow rate was less than 40 mL/min or greater than 100 mL/min, the produced ball diameters were sometimes smaller than 8 cm or larger than 10 cm. With the increasing flow velocity, there was a directly proportional increase in particle size distribution. This was in good agreement with the report on alginate micro spheres [27]. These balls wouldn't be applied to the milk tea, as they would result in a bad consumption experience. Therefore, in order to obtain the required "boba" with suitable diameters of between 8–10 mm, the flow velocity should be controlled at 60–80 mL/min. Figure 3a shows that the loading ratio reached its optimal value (76.15 ± 1.51) when the flow velocity was 60 mL/min. Under this preparation condition, the flow velocity had little effect on encapsulation efficiency and mechanical strength. The encapsulation efficiencies of the "boba" balls were all 100%, and the mechanical strength maintained stability.


**Table 2.** Influence of flow velocity on the "boba" ball diameter. the "boba" ball diameter.

encapsulation efficiencies of the "boba" balls were all 100%, and the mechanical strength

Effects of the preparation situation on the characteristics of the "boba" balls. ( **Figure 3.** Effects of the preparation situation on the characteristics of the "boba" balls. (**a**) Flow velocity; (**b**) Sodium alginate solution concentration; (**c**) Calcium chloride solution concentration.

#### *3.4. Influence of Sodium Alginate Solution Concentration*

"boba" ball properties. The diameter of the hose was 8 cm, the drop height "boba" Figure 3b shows the influence of different concentrations of sodium alginate on the "boba" ball properties. The diameter of the hose was 8 cm, the drop height was 25 cm, the flow rate was 60 mL/min, and the concentration of calcium chloride solution was 1.0%. It was found that the sodium alginate solution concentration had positive effects on the diameter, mechanical strength, loading ratio and encapsulation efficiency of the "boba" balls. when the sodium alginate solution concentration was greater than 0.6%, the diameter maintained stability. However, if its concentration was below 0.5%, there were some "boba" balls with diameters <8.00 mm. The hinge structure formed by sodium alginate and

calcium ions changed with the concentration of sodium alginate, and affected the diameter, mechanical strength, loading ratio and encapsulation efficiency of the "boba" ball [27–29]. The encapsulation efficiency of the "boba" ball was 74.07 ± 0.87% (0.5% sodium alginate) and 81.48 ± 0.75% (0.6% sodium alginate). When the concentration of sodium alginate was higher than 0.8%, the encapsulation reached 100%. However, when the concentration of sodium alginate reached' 1.2%, the "boba" balls tended to stick together and were not conducive to being cleaned separately; this situation would increase their preparation time. The mechanical strength increased exponentially with increasing concentration of sodium alginate. The relationship between the concentration of sodium alginate and the strength of the "boba" balls was as follows, and correlation coefficient was 0.9786:

$$\mathbf{y} = 0.0132 \mathbf{e}^{7.0139 \times} \tag{5}$$

When the concentration of sodium alginate increased from 0.8% to 1.0%, the strength increased from 4.83 ± 0.50 to 20.14 ± 2.60, and the crushing resistance increased by more than four times. To give the "boba" balls a certain crushing resistance and 100% encapsulation efficiency in production, 1% sodium alginate was deemed to be the best.

#### *3.5. Influence of Calcium Chloride Solution Concentration*

Figure 3c shows the influence of different concentrations of calcium chloride on the "boba" ball properties. The diameter of the hose was 8 cm, the drop height was 25 cm, the flow rate was 60 mL/min, and the concentration of sodium alginate solution concentration was 1.0%. It was found that the calcium chloride concentration had positive effects on the diameter, mechanical strength, and the loading ratio of the "boba" balls. The encapsulation efficiency was 100% under the condition of 1–5%. With the increase of calcium chloride concentration, the loading ratio decreased, and the mechanical strength first went up, then went down, and then went up again. When the calcium chloride solution was 3%, the strength reached its maximum. The calcium ion/alginate ratio here was found to present an optimized ratio. The increase of calcium content could increase the crosslinking density and shrinkage of matrix [30], but also produced more carbon dioxide, leading to a more porous and weak matrix [31]. The "boba" balls' average diameter gradually increased from 8.70 ± 0.70 to 10.73 ± 1.30, and was then maintained at a certain level. When the calcium chloride concentration reached or exceeded 2%, the "boba" balls stuck together easily, showed the astringency of calcium chloride and presented a poor sensory evaluation experience. Taking the above factors into consideration, the 1% calcium chloride concentration was determined to be the best.

#### *3.6. Microbial Colonies/Sensory Changes and Prediction of Shelf Life for the "boba" Ball*

The "boba" balls were prepared according to the optimized experimental conditions described above, and stored at different temperatures. During the 90 days of storage, no microorganisms were detected. This was due to pasteurization and sealed filling. Therefore, the microbiological safety of the product could be guaranteed. The product's shelf life is dependent on changes in sensory properties. Colour, aroma, stratification and capsule breakage rate were the main factors affecting sensory indicators.

Figure 4 shows the response of the sensory evaluation of "boba" balls stored at different temperatures. The sensory evaluation values of the product were linearly correlated with storage time, the correlation coefficient R<sup>2</sup> of the fitting regression curve was above 0.94 (Table 3), and the k value was derived, which exhibited a constant value at a constant temperature. The linear equation of ln k against 1/T was obtained according to Equation (4) (Figure 5). The activation energy Ea (93.45 kJ/mol) and the pre-finger factor k<sup>0</sup> (7.46 <sup>×</sup> <sup>10</sup>15) were calculated. Then, the mathematical model equation reflecting the change of the quality of the "boba" balls was established, as follows:

$$F(\mathbf{x}) = 100 - 7.46 \times 10^{15} \times \exp\left(-93.45 \times 10^3 / \text{RT}\right) \times \text{t} \tag{6}$$

− −

" "

" "

uality of the "boba" balls was established, as follows:

sidered to be unable to meet consumers' exp

evaluation of "boba" balls **Figure 4.** Response of the sensory evaluation of "boba" balls stored under different conditions.

evaluation of "boba" balls

**Table 3.** The Arrhenius kinetic parameters of "boba" ball quality.


<sup>a</sup> Correlation coefficients of regression lines of the sensory evaluation value against time. <sup>b</sup> Ea, activation energy; CI, confidence intervals. <sup>c</sup> Correlation coefficients of regression lines of ln k against 1/T.

"boba" ball quality. **Figure 5.** Arrhenius behaviour curve of "boba" ball quality.

The response surface of the sensory evaluation changed with time and temperature was determined (Figure 6). When the sensory score was below 80, the product was considered to be unable to meet consumers' exp ectations and reach the shelf-life limit. Therefore, product shelf life could be predicted for different storage temperatures (Table 3). When the storage temperature was lower than 10 ◦C, the shelf life of the "boba" balls was 465 days, exceeding one year.

"boba" balls **Figure 6.** The response surface of the sensory evaluation changes with time and temperature of the "boba" balls.

#### The Arrhenius kinetic parameters of "boba" ball **4. Conclusions**

The "boba" balls used in milk tea, which are made of calcium alginate, were studied here. The optimal technological parameters for the preparation of "boba" balls were discussed. The diameter of the liquid adding tube was 8 mm, the drop height was 25 cm, the sodium alginate solution concentration was 1.0%, the flow velocity was 60 mL/min, and the concentration of calcium chloride concentration was 1%. The prepared "boba" balls were stored under different temperature conditions. No microbes were detected after storage for 90 days. The sensory evaluation value of the "boba" ball decreased gradually with storage time at different temperatures. The mathematical model for predicting shelf life was as follows Equation (6).

According to the above mathematical model, the shelf life of the "boba" ball developed could reach more than 1 year in a refrigerated environment at temperatures lower than 10 ◦C.

**Author Contributions:** Carrying out the experiment and writing the manuscript, Y.L. and H.C.; Designing the experiment and assisting in editing the manuscript, D.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by Hangzhou Science and Technology Development Project (Grant Number: 20180432B34), Provincial Key R&D project (Grant Number: 2020C02033) and the national research program of China (Grant Numbers: 2017YFD0400704 and 2014BAD04B01).

**Data Availability Statement:** Data sharing not applicable.

The "boba" balls used in milk tea, which are made of here. The optimal technological parameters for the preparation of "boba" balls were dis-**Acknowledgments:** The authors thank the Teaching Reform Project of Zhejiang University for the research funding. Thanks to the teachers Li Yang, Pro. Jianchu Chen and Xingqian Ye for offering technical assistance and the experimental facilities for our experiment.

the concentration of calcium chloride concentration was 1%. The prepared "boba" balls **Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


*Article*
