Retracted: Effects of major ingredients (grape syrup, milk powder, and walnut paste) levels on physiochemical, rheological, and sensory attributes of walnut spread

Effects of major ingredients (grape syrup, milk powder, and walnut paste) levels on physiochemical, rheological, and sensory attributes of walnut spread, Food Science & Nutrition, 2022 (https://onlinelibrary.wiley.com/doi/full/10.1002/fsn3.3107). The above article, published online on October 20, 2022 in Wiley Online Library (https://wileyonlinelibrary.com), has been retracted by agreement between the authors, the journal Editor in Chief Y. Martin Lo, and Wiley Periodicals LLC. The retraction has been agreed due to an error in which the incorrect manuscript was sent to the production team and published instead of the manuscript that the journal had reviewed and accepted.

milk spread because it is melted on their tongue. Additionally, the gradual movement of mixed ingredients on the sequential rollers remove their undesirable flavors and develop their pleasant aroma in the final spread. In other words, the heterogeneous, flaky, and relatively dried paste of each ingredient is transformed into a freeflowing suspension of solid particles dispersed in a dominant phase of used fat (Acan et al., 2021). Milk spreads are classified as dark, white, or chocolate, depends on the major ingredients used in their manufacturing. The inclusion of 10%-20% milk powder in any kind of nut-spread affects its physiochemical, rheological, and sensory characteristics (Monteiro et al., 2018). Most frequently, the milk powder (with or without fat) used in spread production is made by either roller or spray drier (Coutinho et al., 2019;Crawford & Running, 2020).
Whereas crystal sugar is used in most of different spreads (as a sweetening agent), it was our interest to substitute it the Persian grape molasses (PGM). The PGM with ~76% monosaccharide sugars (mainly glucose and fructose), minerals, organic acid and considerable bioactive compounds is a sweet liquid that is made from grape or raisin (Tavakolipour et al., 2020). Since PGM has interesting rheological properties and organoleptic (texture, color, and flavor) specifications, it will enhance the quality aspects of different products, if is used as an ingredient (Azizi Tabrizzad, 2020).
It was our objectives to combine roasted walnut paste (RWP), high protein milk powder (HPMP), and Persian grape molasses (PGM) each one at three levels and apply ball mill for making various kinds of walnut spreads. Then study their chemical, physical, rheological, and sensory characteristics during storage. Later find out the best levels of ingredients for making a novel walnut spread.
Linseeds and flaxseeds nutritionally are similar; the only difference can be seen in the plant itself. While the people in the UK and Australia distinguish between linseed and flax, in the United States and Canada, they refer to both as flax.
While the world production of walnut was 4.5 million tons in 2019, the major producers (in the order of decreasing harvest) were China, the United States, Iran, and Turkey (FAOSTAT, 2019). The botanical structure of walnut (Juglans resia) is compared with human brain, so it is considered as the brain nutrient . Walnut is rich in monounsaturated fatty acids (Omega 3), arachidonic acid, phyto-chemical substances, and bio-actives (melatonin, ellagic acid, vitamin E, carotenoids, and polyphenols). These compounds have potential health effects against: aging, cancers, inflammations, and neurologic illnesses (Şen & Karadeniz, 2015).
Additionally, the 1 oz of walnut meal obtained after cold press provides 2.6 g of Omega-3 antioxidants. Linseed cake or meal resulted from cold pressing of its seeds has 40% protein, 6% minerals (ash), and 12% oil it is the most effective material for roasting and making linseed paste with interesting sensory attributes (Farag et al. 2021). The insoluble dietary fiber component of linseed (~7%) is effective in constipation relief, colon health and may even protect against colon cancer (Condori et al. 2020).
Different milk spreads are produced using most of the times in ball milling system to reduce particle size of diverse ingredients (different nuts, milk &/or cocoa powder, and cocoa butter) (Amevor et al., 2018).
Many people like the excellent taste of milk and nuts spreads because they are melted on their tongue due to their fine particles.
Additionally, the gradual breaking of mixed ingredients removes the undesirable flavors of used ingredients and develop pleasant aroma in the final spread. In other words, the heterogeneous, flaky, and relatively dried paste of each ingredient in LS product is transformed into a freeflowing of solid and fine particles dispersed in a dominant fat-phase of spreads (Acan et al., 2021). The inclusion of 10%-20% milk powder in any kind of nut-spread affects its physiochemical, rheological, and sensory characteristics (Monteiro et al., 2018). Furthermore, the method of milk drying (roller or spray dehydration) for its conversion to powder has effects on sensory attributes of final spread when it is used as an ingredient (Coutinho et al., 2019;Crawford & Running, 2020).
Since there is trend to reduce crystal sugar (as a sweetening agent) in most of food products, it was our interest to substitute crystal sugar with the native and local Persian grape molasses (PGM). While PGM has more than 50% sugar (mainly glucose and fructose) and depends on its Brix, it has considerable minerals, organic acid, bioactive compounds that is made from grape or raisin (Azizi Tabrizzad, 2020). Since PGM has interesting rheological properties and organoleptic (texture, color, and flavor) specifications, it will enhance the quality aspects of different products, if is used properly as an ingredient in food product (Azizi Tabrizzad, 2020).
It was our objectives to study the effects of mixing different levels of three valuable ingredients including roasted linseed paste (RLP), high protein milk powder (HPMP), and Persian grape molasses (PGM) on the physiochemical and sensory attributes of resultant linseed spreads (LS). Then evaluate the peroxide value (as a most important deteriorative index) of LS during long time (90 days) cold (4°C) storage. Then apply respond surface method and central composite design on the generated data and find out the best levels of mixing ingredients for making a highly nutritional and novel linseed spread with minimum deteriorating indexes. Then evaluate the physical, rheological, and sensory characteristics of optimized sample of LS before and after storage.

| Materials
Grape seedless and walnut were prepared from local markets in Miandoab, West Azerbaijan Province, Iran. Skim milk powder was prepared from Pegah company of Tehran, Iran, and milk protein concentrate of 85% was prepared from Alinda company, Greece. Liquid lecithin was prepared from Behpak company, Behshahr Mazandaran Province, Iran. Wheat flour was provided from Taravat flour company, Miandoab, West Azerbaijan Province, Iran. The chemical compounds were purchased from Merck Corporation (Germany).

| Preparation of high protein milk powder
The high protein milk powder produced by mixing Skim milk powder and milk protein concentrate (1:1) and the uniform powder mixture was packed in sterile plastic bags until consumption.

| Preparation of Persian grape molasses
First, grapes seedless arrived at Nikan Pajouhan-Bonab laboratory, the leaves and seeds were separated and grape juice was prepared by a press, and grape juice was extracted by a water intake device.
The pH, acidity, and brix of grape juice were measured. To reduce the acidity and remove the turbidity of grape juice, bentonite (3 g per 100 ml of grape juice) was used. The sap soil is first dissolved in a quantity of grape juice, and then the production treatment is added and thoroughly mixed. After about 3 h and at the end of the effect of the clarifying compounds and the help of the filter, the surface of the juice is cracked, at which time the existing foam must be removed from the surface of the juice and smoothed by a suitable filter. In the next step, grape juice with a certain amount of calcium carbonate is neutralized until it reaches a pH of 5.6. After neutralization, sap soil was used to remove turbidity. Finally, after 30 min and filtration, the production sample was subjected to vacuum cooking and concentration in the evaporator to a degree of 69 ± 1 brix (Tavakolipour et al., 2020). Production samples (ash 1.75%, pH 5.6, acidity 0.162% OA) were packaged and stored in sterile containers until the test.

| Preparation and processing
The walnut paste (Ash 3.14%, protein 21.7%, fiber 18.22%, fat 37.86%, and pH 6) was made by roasting its seeds at 121°C for 1 min followed by its uniform mixing with ̴ 0.85% lecithin (as emulsifier) at 45°C for 10 min (Turner & McNiven, 2011). Then, the null-flour of wheat roasted for 140 s at 180°C (Germishuys et al., 2020) and mixed at rate of ̴ 8.5% with HPMP. Later, the PGM was added evenly to the prepared mixture, and the contents was stirred for 10 min at 25°C to develop a liquor phase (Turner & McNiven, 2011). The production of walnut spread was completed using a ball mill (Sepehr machine, Iran) equipped with an agitator (600 RPM) containing 6 mm diameter stainless-steel balls using the technique of Bolenz et al. (2014). The mass of ultimate mixture was flowing through the balls (as a bed) using a recycling pump at medium current of 1 kg/ min. The temperature of refining (particle size reduction) process was stabilized at 45°C and checked every 5 min during 3-h milling time. If the temperature of mixed materials goes above 45°C, the vessel was cooled with tap water before proceeding. A micrometer caliper was used to keep track of particle size until it reached 30 μm.
With this setup, there was no way to reduce the size anymore. To prevent contamination, the molded product was left out for 3 h and wrapped in aluminum foil (Bolenz et al., 2014).

| Chemical characterization
The following tests were carried out to measure the chemical speci- c. Ash content or mineral residue by incineration of its total organic matter in a muffle furnace at 550°C, as described in the official AOAC international method (AACC, 2000;AOAC, 2000). e. Free fatty acids (as free oleic acid percentage) and peroxide value by applying the AOCS methods (AOCS, 1989).
f. pH was measured using a pH Meter (IKA, Germany) which was calibrated before the test.

| Sensory evaluation
After selecting 20 members of sensory panel (from the Department of Food Science and Faculty in the University of Tehran), they were trained for sensory evaluation. The evaluators assessed 8-10 g of prepared WS samples in nibble forms and in separate areas with air circulation under bright light (recommended by ISO Standards No. 8589 [2007]) 2 h after breakfast (ISO, 2007). We used a 5point hedonic scale to score different sensory attributes (appearance, aroma, sweetness, mouthfeel, and aftertaste) of each sample (Mahato et al., 2021). The final sensory acceptance score of the samples is calculated by summing the average scores of five different sensory characteristics obtained from the evaluators. Although this approach does not represent accurate customer perception, but it strongly verified the required characteristics that the high-quality WS product should have it.

| Experimental design, model fitting, and variables optimization
Central composite design (CCD) system was used to study the effects of three (PGM, HPMP, and RWP) independent parameters on the peroxide value, a w , acidity, and sensory scores of the resulting samples of LS. The design included 20 experiments, which consisted of six center points in a cube. The operating conditions were conducted at five levels coded as −2 (−α), −1, 0, +1 and +2 (+α). The actual and coded values of the independent variables are listed in Table 1. A second-order polynomial model expressed the relationship between the three independent variables and their responses for dependent variable (α): Where α represents the response variable, β 0 is the constant coefficient, x represents the independent variable (factor), i is the i th factor, β i β j β l , β ii β jj β ll and β ij β il β jl are the linear, quadratic, and second-order interaction coefficients, respectively. The statistical software Design-Expert®, version 11, was used for experimental design, analysis, graphing, and optimization.
Validation of the optimized conditions was carried out in triplicate to generate the optimum model using Duncan's new multiple range test (MRT). MRT is a multiple comparison procedure developed by David B. Duncan in 1955. The significance level used for this study was 95% (p < .05) for all the statistical analyses.

| Physical specifications (color values and texture profile analysis)
The surface color of optimized WS samples was measured in tripli-

| Rheological description
The rheological properties of the prepared WS were determined using a stress/strain and temperature-controlled rheometer (Anton Paar GmbH, Austria) equipped with parallel plates geometry (each one with 50 mm diameter), and according to Mantihal et al. [23] method, with some modification. In the first, the gap between the two plates of rheometer was adjusted to 0.5 mm. After loading samples, its probe was descended and compressed the two plates with a gap of 2 mm. Then, viscosity measurement was started after 2 min for equilibration of the temperature and relaxation of the samples.
The applied shear rate range was between 1 and 100 s −1 radian frequency at 25°C. A power-law model (Fanari et al., 2020) used to show the dependency of the measured complex viscosity data on the frequency of LS.
where σ is shear stress (Pa), K is the consistency coefficient (Pa·s n ), γ is the shear rate (s −1 ), and n is the flow behavior index.
(1) TA B L E 1 Actual and coded independent variables employed in experimental design to find the specifications of optimized linseed spread after using response surface methodology 4 | RE SULT AND D ISCUSS I ON

| Chemical characterization (peroxide value, water activity [a w ], and acidity)
The peroxide value (PV) is still the most important chemical index for determining how quickly the walnut spread oxidize and influences its shelf life due to its oil portion and to some extent its moisture content. When the percentage of PGM, HPMP, and RWP in formulation of final WS; its PV changed from 0.51 to 1.70 meq O 2 /kg when it stored at 4°C On 1st day of production.
Additionally, the effects of three ingredients on peroxide formation of WS (shown in Figure 1) confirmed that the contents of RWP and PGM had more effects than HPMP on increasing this deteriorative index.
Beside controlling PV, it was our intension to select appropriate ingredients along with suitable processes to minimize a w in the final WS and protect it from microorganisms' activities. When the proportions of PGM and RWP changed, its a w changed from 0.51 to 0.76 in the final WS at 4°C On 1st day of production. Additionally, the effects of three ingredients on water activity of walnut spread (shown in Figure 2) confirmed that the contents of PGM and RWP had impacts on increasing this index (a w ).

| Sensory evaluation
When nutritive edible ingredients are processed properly and combine with safety and health concerns, there is a good chance to receive good consumers' responses. Because customer opinion is a practical quality-level evaluation. This is the reason profiling a new food product using a trained panel's judgment is necessary to certify correctness and reproducing it with uniform sensory attributes.
Despite the growing popularity of high-quality cocoa and chocolate products, literature is limited on the sensory attributes of different spreads.
The total sensory score, which obtained the acceptance levels in walnut spread, changed from 45.3 to 90.9 (depends on the proportion of PGM, HPMP, and RWP used in the resulting LS).
As Figure 4 shows, increasing the amount of the three major ingredients to some extent improved the sensory evaluation score of walnut spread. However, the negative taste of RWP was noticed when it increased >30 g in formulation. The higher content of RWP in the final WS provided more unsaturated fatty acids for oxidation, and therefore, more taste of rancid fat (bitterness) was recognized. Table 2 shows the effects of different levels of major ingredients (PGM, HPMP, and RWP) on sensory acceptance, PV, a w , acidity and acceptance scores of final samples of walnut spread.

| Model fitting
The independent (RWP, GS, and HPMP) and dependent (PV, aw, acidity, and sensory score) variables were fitted to the second-order model equation and examined for their goodness of fit (Table 3). The analyses of variance were performed to determine the lack of fit and the significance of independent variables on linear (first order) and interaction effects along with quadratic (second order) impacts on dependent variables (specifications) of WS.
The regression results showed that different models tested for all the response variables were highly adequate because they had satisfactory levels of R 2 (>80%) and that there was no significant lack of fit in all the response variables. The regression coefficients are shown in Table 3

F I G U R E 4
Response surface plots showing the interaction impacts of independent variables (PGM, HPMP, and RLP) on the sensory acceptance score of LS.

TA B L E 2
The level effects of major ingredients on peroxide value, aw, acidity, and total sensory scores of LS (linseed spread) out of maximum samples along with central composite design arrangement right after production  Table 3). Equation 7 also showed that (RWP) 2 and (HPMP) 2 had more impacts than (PGM) 2 on increasing peroxide value. While (RWP) 2 and (HPMP) 2 had positive coefficients, the coefficient of (PGM) 2 was negative.
On the contrary, the intercept values for model of water activity (Equation 8) confirmed the increasing and decreasing effects of, respectively, two ingredients (PGM and HPMP) and one ingredient of RWP on a w (see Table 3). Equation 8 also showed that (HPMP) 2 and (PGM) 2 had more impacts than (RWP) 2 on increasing a w . While (PGM) 2 and (HPMP) 2 had positive coefficients, the coefficient of (RWP) 2 is negative.
Furthermore, Equation 9 showed that PGM, HPMP, and RWP had a significant impact on the acidity of the final WS. Additionally, the acceptance score of the WS was significantly influenced by the percentage of its three major ingredients, as Equation 10 shows this matter clearly (Table 4). Figure 5 display the high correlation between walnut spread sensory acceptance and the process variables in the selected model. This model was able to predict the total sensory score of linseeds spread with negligible error (from its actual number) using the selected process parameters.

| Optimization the levels of independent and dependent variables for long storage
RSM method used to obtain fitted regression model and find the optimum values of PGM, HPMP, and RWP for making the walnut spread with high quality and sensory scores as shown in Table 5.
In RSM, the desirability function is widely used to determine best levels of combining three major independent variables and provide goal for desirable product was set within the ranges used for each independent variable. Simultaneously, it was our aim to produce WS with the minimum values of PV, a w , and acidity, and maximum sensory acceptance score. The desirability 67.6% was obtained when the optimum levels of PGM, HPMP, and RWP were 47.402, 7.176, and 29.385 g, respectively. The maximum value of 67.6% means that the suggested proportion of the three major ingredients had good potential to produce an attractive walnut spread for consumers. As Table 5 shows, a small percentage error (<10%) was notified between the experimental and predicted values for the dependent variables. In other words, verification experiments demonstrated the adequacy of responses for the final product.
Beside the discussed dependent variables (PV, aw, and acidity), other chemical parameters of WS including the fiber, protein, fat, and ash contents along with pH can influence the sensory attributes and textural behavior of final product. Table 6 shows the results of chemical analyses performed in this study after 90 days of storage (see Table 6). The PV and acidity of the optimized sample stored for 90 days at 4°C, increased, respectively, from 0.72 to 1.07 meq O 2 /kg and 0.09% to 0.196% mainly because it had low content of high-quality fat. However, the PV and acidity of hazelnut-cocoa spread stored for 90 days at similar conditions aThe theoretical values of PGM, HPMP, and RLP numbers were gained when the actual PV, aw, acidity, and acceptance sensory score (obtained by experiment) inserted in equations of 7-10.
bThe three chemical indexes and acceptance sensory score were obtained when the appropriate levels of PGM, HPMP, and RLP inserted in equations of 7-10.

TA B L E 4
The practical ranges and appropriate levels of three independent variables for making LS with highest sensory score and permissible chemical indexes (PV, a w , and acidity) along with comparison between the predicted and actual sensory scores F I G U R E 5 The predicted linseed spread sensory scores (obtained from a regression model) versus actual sensory scores along with 95% upper and lower confidence intervals for the predicted scores.
were much higher and, respectively, changed from 4 to 6 meq O 2 / kg and 0.71% to 0.77% mainly because it had 10% palm oil and 11% sunflower oil (Tarakçi & Yildirim, 2021). Although the protein content of some nuts (such as peanut) spreads such as peanut spread reaches to >20% (Mazaheri-Tehrani et al., 2009) and higher than WS, but our suggestive spread has considerable and unique health benefits mainly attributed to its omega-3 fatty acids, fiber, and lignans. Additionally, 3 months' storage of WS did not make significant changes on its protein (~13%), fat (~20%), and ash (~2%).
Nevertheless, its acidity and PV values increased significantly during this storage time.

| Physical properties (color values and texture analysis)
In terms of customer thought and quality perception, the appealing temperatures, and these reactions may reduce the astringency and bitterness during storage, resulting in a more stable color, pleasant smell, and superior flavor in different (nuts and chocolate) spreads (Baptista et al., 2021;Lončarević et al., 2018). Storage conditions (mainly temperature and relative humidity) significantly altered the scales of color values and parameters (Nightingale et al., 2011).

| Rheological description
The storage modulus (G') shows the elastic response (or stored energy) of food and reflects its solid-state behavior. However, loss modulus (G") reveals the viscous response (or dissipated heat energy) of a food (such as spread) and shows its liquid-state behavior (Mantihal et al., 2019). Furthermore, internal friction between the components (molecules and particles) of a moving fluid causes viscous behavior and production of frictional heat. This friction is always associated with the conversion of deformation energy to heat energy. When the food material is released, the unused stored energy is a driving force to reconstruct into its original shape (Pajin et al., 2013). the storage-modulus values were considerably more than those in loss modulus (G' > G"). This is owing to the internal (chemical bonds connections, physiochemical) interactions between the molecules and particles of the materials in the resulting spread (Janmey & Schliwa, 2008). However, the situation of these two moduli was reversed when the frequency became higher than 1 Hz and subsequently, the loss modulus rose more than storage modulus.
Figure 8 also shows the loss tangent values (tan δ = G"/G') as a function of frequency. This is also known as the gel point or the sol/gel transition point and indicates that the sample moved from a liquid (or sol state) to a solid (or gel state) and vice versa during the viscoelastic measurement at different frequencies. When the tan δ > 1, the optimized walnut spread behaves like a viscus, whereas tan δ < 1 it behaves like an elastic (Manasi, 2019;Qaiser et al., 2021).
These findings can be explained by the fact that the RWP plays a dual role in the formulation of WS and acts as a filler in the complicated network of the resulting spread. In other words, significant interactions take place between crystallized HPMP and PGM, and their outcomes make hydrodynamic polymers with the oil base of walnut paste. Researchers believe that low total elasticity of spreads is related to their high particle concentrations (Taylor et al., 2009).
Lecithin has been used in nuts (and specifically chocolate) spreads to control their viscoelasticity in the confectionery industry.
This is the reason that the cocoa butter has been replaced with lecithin because lecithin has strong emulsification power and is much less expensive than cocoa butter. Moreover, usage of soy lecithin F I G U R E 8 The effects of frequency (applied forces) on storage modulus (G'), loss (G") modulus, and tan δ at 25°C in the linear viscoelasticity region of linseed spread. The units for both moduli are pressure or energy/unit volume of LS after production.
at 0.1%-0.3% level in every spread can reduce the same viscosity as over 10 times this amount of cocoa butter (Karnjanolarn & McCarthy, 2006). However, if the level of lecithin exceeds 0.3%-0.5%, unsuitable thickening of the chocolate occurs (Karnjanolarn & McCarthy, 2006). This is the reason we used lecithin below 0.5% to control the rheology of the final WS.

| CON CLUS ION
A RSM design was used to combine different levels of RWP (roasted walnut paste), PGM (Persian grape molasses), and HPMP (high protein milk powder) and develop a valuable walnut spread. Consumer acceptability, a w , acidity, and PV (peroxide value) were all determined to create a prediction model. To select the best areas, predictive algorithms were employed to create contour maps. Overlapping areas aided in the development of a walnut spread with optimized (high quality) compositions (containing 30 g RWP, 7.5 g HPMP, and 50 g PGM). The optimized walnut spread has a good potential to combine it with cocoa powder and enhance its consumer acceptance attributes (color and texture) specially for the international markets.

ACK N OWLED G M ENT
We hereby thank the services of the laboratory of food emerging technologies and rheology of Tehran and Tabriz universities.