3.1. Moisture content and water activity
The stability of spray dried powder is a function of its physical properties such as MC and aw which can be adjusted using different carriers, alone or in the mixture [21]. The changes in MC and aw as a function of wall material concentrations and the temperature are illustrated in Figs. 1 and 2. The MC of produced powder ranged between 3.44 and 6.05% respectively for the samples of run 10 (MD = GA = 3.75, Temp.=190oC) and run 1 (MD = 1.875, GA = 5.625, Temp.=110oC), respectively (Table 1). This variation was because of the wall material characteristics, feed properties, and dryer variables including air temperature, flow rate, and airflow [22]. Generally, the spray-dried powders are described through their low MC to be in the range of 2.9–4.66% [23]. On the other hand, it is reported that if MC and aw of food powder be < 10% and 0.6, respectively, it could be classified as a hygienic product [24]. Therefore, it can be concluded that all samples were acceptable concerning MC and aw.
Table 1
The values of different responses for the spray-dried samples
Run | A: Maltodextrin (%) | B: Gum Arabic (%) | C: Inlet air temperature (°C) | Product yield (%) | Particle size (D43) | TMA (mg/kg dry matter) | Moisture content (%) | aw |
1 | 1.875 | 5.625 | 110 | 43.2855 | 13.247 | 1940.27 | 6.05 | 0.22 |
2 | 3.75 | 3.75 | 150 | 42.2872 | 13.5885 | 1989.21 | 4.1 | 0.13 |
3 | 7.5 | 0 | 110 | 48.0629 | 15.373 | 1716.08 | 4.42 | 0.17 |
4 | 3.75 | 3.75 | 150 | 41.5674 | 13.028 | 1983.95 | 3.9 | 0.14 |
5 | 3.75 | 3.75 | 190 | 34.7961 | 12.1013 | 1822.51 | 3.84 | 0.13 |
6 | 3.75 | 3.75 | 110 | 47.8723 | 13.383 | 1946.05 | 5.6 | 0.17 |
7 | 0 | 7.5 | 190 | 33.3434 | 11.249 | 1848.28 | 3.58 | 0.11 |
8 | 7.5 | 0 | 150 | 47.8723 | 15.681 | 1979.28 | 3.82 | 0.15 |
9 | 0 | 7.5 | 150 | 42.9965 | 12.512 | 1844.7 | 4.15 | 0.15 |
10 | 3.75 | 3.75 | 190 | 33.8014 | 12.4 | 1886.98 | 3.44 | 0.1 |
11 | 7.5 | 0 | 190 | 37.4397 | 16.1193 | 1717.06 | 3.72 | 0.1 |
12 | 1.875 | 5.625 | 150 | 32.8059 | 14.475 | 1921.42 | 4.4 | 0.18 |
13 | 5.625 | 1.875 | 150 | 47.5177 | 14.32 | 2004.76 | 3.81 | 0.12 |
14 | 0 | 7.5 | 110 | 43.695 | 12.343 | 1968.35 | 4.91 | 0.18 |
15 | 0 | 7.5 | 150 | 36.7855 | 12.9635 | 1841.88 | 4.49 | 0.17 |
As shown in Fig. 1, the MC of powders decreased at higher concentrations of MD. This decreasing trend was more pronounced in lower temperatures because in high temperatures the impact of drying air temperature on lowering the MC is more significant than the carrier type. A similar trend was observed by Ferrari et al. [25] who spray dried blackberry juice using a mixture of MD and GA. They stated that this result is because MD is more effective than GA in decreasing the MC because GA is a complex heteropolysaccharides having a ramified structure with shorter chains and hydrophilic groups. This findings were also confirmed by T.A. Tran, V.H. Nguyen [26] who used a mixture of MD and GA for SD of lemongrass leaf extract. So, it can be concluded that producing powders with high amounts of GA in the formulation might result in a more hygroscopic product that results in problems in further processing and storage systems.
It was also shown that the MC content of powder increased at lower air temperatures in the spray dryer chamber due to the lack of sufficient thermal energy to remove water from atomized droplets [26, 27]. The same trend was observed for aw towards role-playing of temperature, in which the aw of produced powder decreased by increasing inlet air temperature (Table 1). The aw of samples varied in the recommended range of 0.1–0.22, as reported by Tonon et al. [28], who claimed that the food powders are stable if their aw was < 0.3. Similar results were reported for SD of different plant-based extracts such as saffron extract [23], blueberry extract [29], and myrtle extract [18].
The result of model fitting is also presented in Table 2. A reduced linear X quadratic model and a reduced cubic X linear model were the best models for MC and aw, respectively. All of the statistical results showed a good fitting of the model to the experimental results (Table 2). As it is obvious from the models, the linear effects of MD and GA, the interaction of temperature and MD, temperature and GA, and also the quadratic term of temperature were significant in the model proposed for MC. For aw, all of the mentioned terms for MC plus the term including the interaction of MD and GA multiplied by (MD – GA) was significant. The linear term of temperature was not significant in both models, while the interactions of temperature and both carriers were significant in both models.
Table 2
Results of model fitting for different responses
Response | Fitted model | Coded equation | Statistical parameters |
Moisture content | Reduced Linear x Quadratic model | MC = 3.83A + 4.29B-0.5346AC-1.01BC + 0.3529C2 | P value for model | 0.0012 |
Std. Dev. | 0.3813 |
Mean | 4.28 |
C.V. % | 8.91 |
R² | 0.8106 |
Adjusted R² | 0.7349 |
Predicted R² | 0.3413 |
P value of Lack of fit | 0.1644 |
aw | Reduced Cubic x Linear model | aw = 0.1389A + 0.1517B-0.0323AC-0.0342BC-0.3251AB(A-B) | P value for model | < 0.0001 |
Std. Dev. | 0.0123 |
Mean | 0.1480 |
C.V. % | 8.34 |
R² | 0.9061 |
Adjusted R² | 0.8686 |
Predicted R² | 0.8133 |
P value of Lack of fit | 0.7946 |
Product yield (%) | Reduced Cubic x Linear model | Yield = 44.12A + 38.95B-6.07AC-6.4BC + 54.77AB(A-B) | P value for model | 0.0001 |
Std. Dev. | 2.35 |
Mean | 40.94 |
C.V. % | 5.75 |
R² | 0.8773 |
Adjusted R² | 0.8282 |
Predicted R² | 0.6863 |
P value of Lack of fit | 0.6585 |
Particle size (D43) | Reduced KCV model | Particle size = 16.08A + 12.80B-3.25AB-0.8157BC-0.7012C2 | P value for model | 0.0003 |
Std. Dev. | 0.6369 |
Mean | 13.52 |
C.V. % | 4.71 |
R² | 0.8546 |
Adjusted R² | 0.7964 |
Predicted R² | 0.7035 |
P value of Lack of fit | 0.0993 |
TMA (mg/kg dry matter) | Reduced Quadratic x Quadratic model | TMA = 1979A + 1839.99B + 304.94AB-60.61BC-260.45AC2 + 65.47BC2 | P value for model | < 0.0001 |
Std. Dev. | 19.68 |
Mean | 1894.05 |
C.V. % | 1.04 |
R² | 0.9717 |
Adjusted R² | 0.9560 |
Predicted R² | 0.9396 |
P value of Lack of fit | 0.8840 |
3.2. Powder yield
The product yield of spray dried samples varied in the range of 31.81% (run12: MD = 1.875, GA = 5.625, Temp.=150oC) to 48.06% (run 3: MD = 7.5, GA = 0, Temp.=110oC). Similar results were reported by Asik et al. (2021) who asserted that the powder yield of spray dried myrtle extract by MD and GA was 40.18%. The wall material ratio and process temperature had a meaningful impact on powder yield (Table 1). As shown in Fig. 3, by increasing the ratio of GA in formulation along with decreasing temperature, the powder yield was maximized. Other studies also reported higher MD levels decreased powder yield due to increased mixture viscosity and consequently adhering the drying solids to the dryer wall [30]. Furthermore, the MC of powder increased at lower inlet air temperature resulting in increasing the spray dried powder weight collected in the cyclone which translates into a higher rate of product yield. The negative effect of high drying temperatures on powder yield might be also due to melting and more adhesion of the powders onto the wall [31]. The product yield in SD is controlled by different factors including dryer conditions, wall material type and its combined ratio, and the nature and concentration of core material in feed [23].
The result of the model fitting of powder yield is presented in Table 2. A reduced cubic X linear model was best fitted to the experimental data with suitable statistical values. The linear effects of MD and GA, the interaction of MD and GA, temperature and MD, temperature and GA, and also the quadratic term of temperature were significant.
3.3. Particle size of final powders
One of the most important properties of powder formulations is their application including the texture of products which can be determined by the particle size [32]. The particle size variations as a function of wall material ratios and temperature are shown in Fig. 4. The PSD of samples varied between 11.249 and 16.12 µm. The lowest and highest values were obtained for run 7 (MD = 0, GA = 7.5, Temp.=190oC) and run11 (MD = 7.5, GA = 0, Temp.=190oC), respectively. These observations could be identified through the fact that the particle size of spray dried powder increased by boosting the viscosity of feed emulsion [27]. On the other hand, it is reported that several factors influenced PSD such as physicochemical properties and wall materials along with their ratios in feed and the spray dryer operational conditions [23].
As shown in Fig. 4. increasing the MD concentration had a more significant effect on particle size than drying temperature and led to larger particle size. Ferrari et al. [25] also reported that the PSD of powders produced with MD was significantly higher than the powders produced using GA. Increasing the temperature also led to a smaller particle size which was in accordance with the results of Tonon et al. [33] and Aghbashlo et al. [34]. Different PSDs were reported by researchers for the powder produced by the spray dryer as follows. 0.69-16.175 µm for myrtle extract powder [18], 1–15 µm for turmeric oleoresin powder [35], 5–30 µm for blue corn anthocyanins powder [36], and 0.56–99.80 µm for grape skin extract powder [37]. The PSD obtained in this study (Fig. 5) was different from the values reported by other researchers. This difference can be interpreted due to using different wall materials, ratios, the operational condition of the spray dryer as well as the nature of core materials [23].
The result of the model fitting of PSD is presented in Table 2. A reduced KCV model was best fitted to the experimental data with suitable statistical values. The linear effects of MD and GA, the interaction of temperature and MD, temperature and GA, and also the quadratic term of temperature were significant in this model.
3.4. Total monomeric anthocyanins
Anthocyanins are highly unstable and sensitive to processing conditions which are usually influenced by drying time and temperature [31]. As shown in Table 1 and Fig. 6, the TMA content of encapsulated powder significantly changed as a function of wall material ratios together with inlet air temperature. The highest and lowest TMA were recorded after run13 (MD = 5.625, GA = 1.875, Temp.=150) and run3 (MD = 7.5, GA = 0, Temp.=110oC), that were 2004.76 and 1716.08 (mg/kg dry matter), respectively. It was observed that the combination of GA and MD was efficient toward encapsulation of anthocyanins better than other carriers due to the optimum interaction that occurred between these two biopolymers [38]. The formation of a protective film layer around the atomized droplets that contain GA in their formulation was facilitated due to the presence of protein residue in this biopolymer [23]. Accordingly, the lack of GA in the formulation of run3 brought about the loss of TMA during the drying process. On the other hand, the presence of AG in run13 resulted in the formation of an effective barrier layer around the core materials.
At very high inlet temperatures, loss of heat-sensitive TMA is expected due to oxidation and thermal degradation [39] while in moderate temperatures, a protective layer forms faster than those produced at a lower temperature, that likely resulted in the high retention rate of TMA in run 13. Do, Nguyen [40] also reported similar findings. They used a mixture of GA and Microcrystalline Cellulose for SD mulberry juice at different temperatures (120–160 oC) and found out the TMA of spray dried mulberry juice was highest at 140 oC and declined when higher temperatures were used. They also concluded that at higher inlet temperatures, a stable and more smooth layer was formed due to higher evaporation rates that protected the anthocyanins from thermal degradation. However, at severely high temperatures, loss of anthocyanins is observed due to thermal sensitivity. These findings were also confirmed by Tonon et al. [30, 41], Ersus, Yurdagel [42]. In contrast to our results, Asik et al. [18] reported that the optimum conditions for achieving the highest encapsulation efficiency were an inlet temperature of 120oC, MD = 3.75, and GA = 11.25 g/100 mL extract. It is proved that the SD process should be conducted in a way that a balance creates between the inlet and outlet air temperature and the time of the process. This balance should design toward forming the protective film as quickly as possible in parallel to reduce the residence time of droplets in the dryer chamber [23]. Due to the differences between the nature of core material, wall material, wall material ratio, SD operating conditions, and its configuration, it is predictable that temperature/time process balance is case-exclusive.
The result of the model fitting of TMA is presented in Table 2. A reduced quadratic X quadratic model was best fitted to the experimental data with suitable statistical values. The linear effects of MD and GA, the interaction of temperature and MD, temperature and GA, and also the quadratic term of temperature multiplied by GA were significant in this model.
3.5. Optimization
Optimization was performed with the aim of all input parameters in the experimental range while maximizing the powder yield and TMA and keeping the PSD in range. Based on these criteria, two optimized solutions were proposed by the software as presented in Table 3. Among these, the first solution was preferred because of its higher desirability and also because MD is a less expensive carrier than GA; therefore it would be more economical to use this compound in the SD process. The validation experiment was performed and a high correlation (R2 > 0.9) was observed between the experimental and predicted values at the optimum point which shows the adequacy of the models to predict the experimental data.
Table 3
Optimization solutions to maximize the encapsulation yield and TMA and keep the PSD in range
Number | MD | GA | Temperature | Yield | Particle size | TMA | Desirability | |
1 | 5.915 | 1.585 | 138.548 | 50.057 | 14.833 | 1988.790 | 0.492 | Selected |
2 | 0.000 | 7.500 | 110.000 | 45.352 | 12.911 | 1966.072 | 0.402 | |
3.6. Volatile components and particle size distribution at the optimum point
Aroma analysis of purple basil-lemon powder and reconstituted purple basil-lemon powder (0.5 g powder + 50 mL boiled water) at the optimum point was performed by GC-MS using the SPME method (Table 4 and Fig. 7). The amount of powder and water for the reconstituted product was determined by sensory analysis. Limonene and γ-terpinene in the final powder; limonene (for lemon juice), γ-terpinene, linalool, and 1.8 cineole were the predominant aroma components in the reconstituted product. A previous study reported that linalool and 1,8 cineole are the main volatile components of the basil leaves [43]. More aroma components were determined in the reconstituted product compared to the powder product. It is thought that this difference is due to the fact that the aroma components are encapsulated in the powder.
Table 4
GC-MS results of purple basil-lemon juice powder
Retention time | Area | Area (%) | Volatile component name |
1.275 | 167477 | 1.69 | Trans-β.-Ionon-5,6-Epoxide |
1.357 | 61219 | 0.62 | 1-Propyne (CAS) Propyne |
1.543 | 54152 | 0.55 | (Z)-(1RS,2RS,3RS,4SR)-2-hydroxy-3-(7-hydroxy-2-heptenyl)cyclopentane-1,4-dithiol |
1.975 | 16076 | 0.16 | Butanal, 2-methyl- (CAS) 2-Methylbutanal |
2.918 | 46275 | 0.47 | 4,5-Difluoroctane isomer |
4.490 | 35946 | 0.36 | 4-Benzyloxy-1-bromobutane |
6.140 | 47043 | 0.47 | 6-[(2-Oxocyclohexyl)methyl]-4-oxa-5-azaspiro[2.4]hept-5-ene |
12.058 | 59911 | 0.60 | 2-β-Pinene |
12.481 | 41356 | 0.42 | 2-methoxy[1]benzothieno[2,3-c]quinolin-6(5H)-one |
12.947 | 43855 | 0.44 | Trans-Sabinenehydrate |
13.241 | 354721 | 3.57 | l-Phellandrene |
13.396 | 5130156 | 51.63 | Limonene |
13.490 | 737370 | 7.42 | Eucalyptol (1,8-Cineole) |
14.512 | 731027 | 7.36 | γ.-Terpinene |
15.331 | 3919 | 0.04 | 1,1,2-Trifluoro-2,5-bis(trifluoromethyl)hexane |
15.587 | 122188 | 1.23 | α-Terpinolene |
15.998 | 1048561 | 10.56 | Linalool |
16.157 | 31701 | 0.32 | Nonanal (CAS) n-Nonanal |
16.484 | 21846 | 0.22 | Cyclohexene 3-(tert-butyl)peroxide |
18.068 | 13377 | 0.13 | Trimethyl[o-[.α.-(p-tolylsulfonyl)phenylethyl]benzyl]silane |
19.194 | 115572 | 1.16 | β. Fenchyl Alcohol |
19.498 | 18175 | 0.18 | 1-Methylbutyl nitrite |
19.687 | 5088 | 0.05 | Propane, 2-nitro |
22.979 | 73686 | 0.74 | 2-Propenoic acid, 3-phenyl-, methyl ester (CAS) Cinnamic acid methyl ester |
23.735 | 11319 | 0.11 | 2,3,4,4-Tretrapropyl-1-(trimethylsilyl)-1-(trimethylsilyloxy)-1,3-diaza-2,4-diborabutane |
25.398 | 495954 | 4.99 | (Z)-3-Phenyl-2-Propenoic Acid, Methyl Ester |
25.882 | 9096 | 0.09 | Di-n-octyl disulfide |
26.628 | 16746 | 0.17 | β-elemene |
27.046 | 130759 | 1.32 | Trans-.α.Bergamotene |
28.863 | 10974 | 0.11 | 1,3-Diphenyl-1-((trimethylsilyl)oxy)-1(Z)-heptene |
28.967 | 2796 | 0.03 | 2-Methyl-N-Phenylpropanoic Acid Amide |
29.146 | 239153 | 2.41 | Cyclohexene, 1-methyl-4-(5-methyl-1-methylene-4-hexenyl)-, (S)- |
29.255 | 7476 | 0.08 | 1,2(trans),2,3(trans),3,4(trans),-2,4-bis(p-cyanophenyl)-1,3-diphenylcyclobutane |
31.836 | 9549 | 0.10 | Butanoic acid, anhydride (CAS) Butyryl oxide |
39.885 | 10834 | 0.11 | 1-(Allenylsulfonyl)-2,5,5-trimethyl-3-vinyl-2-cyclohexene |
42.017 | 8475 | 0.09 | trans-1-(phenylthio)-6-oxo-4-oxahept-1-ene |