The cross-sectional diameter of the samples was measured randomly by a Vernier caliper about 20 times. The ratio of the average sample diameter to die diameter, ER, was then calculated (diameter of the circular die was 3.5 mm) (Ding et al., 2006).

SV was determined by the millet volume displacement method using a graduated cylinder (Spinello et al., 2014).

A CR-400 chroma meter (Konica Minolta, Japan) was used for color determination, and color calibration was performed on the standard whiteboard (L*, 97.13; a*, 0.21; b*, 1.87). The results are shown as ΔL, Δa, Δb (the differences in brightness, redness, and yellowness between samples and the whiteboard, respectively) (Spinello et al., 2014).

The products were poured into randomly numbered containers and the overall acceptability of products on a 9-point hedonic scale was evaluated by a ten-member panel (all from the Food Science Department, Shanghai University, Shanghai, China). The results were expressed as sensory scores (SS), from 9 (extremely like) to 1 (extremely dislike) (Wichchukit and O’Mahony, 2015).

WSI and WAI of samples were analyzed by a modification of the method of Spinello et al., (2014) according to Equation (1) and Equation (2). 2 g of sample (m₀) were placed in a centrifuge tube (m₁), and dispersed into 25 mL distilled water. After being kept in the water bath for 30 min (with intermittent shaking every 5 min), the sample was centrifuged at 4,000 rpm for 10 min. The supernatant was decanted into a glass pan (m₂) and dried at 105 ºC to constant weight (m₃). Lastly, the weight of the centrifuge tube and gel of the precipitate (m₄) were measured.

β-Carotene was analyzed by UV-1800PC ultraviolet spectrophotometer (Mapada Instrument Co., Shanghai, China) (Pan et al., 2016).

Squalene was determined by LC-20A High Performance Liquid Chromatography (HPLC) with SPD-M20A UV detector (Shimadzu Co., Japan) (Pan et al., 2016). The unsaponifiable matter of a 3-g samples were extracted, and then dissolved and diluted with 25 mL 0.22 μm membrane filtered n-hexane. The injection volume was 25 μL. An Inertsustain C18 column (250 mm × 4 mm, 5 μm; Japan, Shimadzu Corporation) thermoset at 30 °C was used. The mobile phase was acetonitrile-methanol (4:6 v/v) at a flow rate of 1 mL/min. Peaks were detected at 204 nm. Squalene was identified and quantified using standard squalene (Aladdin Industrial Co., Shanghai, China) as external standard.

Tocotrienols and tocopherols were measured by a LC-20A HPLC with RF-10AXL fluorescence detector (Shimadzu Co., Japan) programmed for excitation at 290 nm and emission at 330 nm. 2 g of sample were dissolved in methanol and sonicated for 20 min, centrifuged at 3000 rpm for 10 min, after which 5 μL of the dissolved sample were injected into the HPLC. The separation was achieved on an Inertsustain C18 column (250 mm × 4 mm, 5 μm; Shimadzu Co., Japan) thermo-set at 30 °C. The mobile phase was methanol-water (98:2 v/v) at a flow rate of 1 mL/min. The identification and quantification were based on external standards (α-, β-, γ-, and δ-isoforms, Solarbio Co., Beijing, China).

The PV and p-AV were determined according to methods Cd 8-53 and p 2.4 (AOCS, 1997), respectively.

A Rancimat 743 model (Metrohm Co., Switzerland) was used to determine the IP of the samples according to a previous method with slight modifications (Olajide et al., 2020). 2.5 g of sample were carefully weighed into rancimat tubes and subjected to accelerated oxidation at 100 °C under an air flow of 20 L/h. The IP were printed automatically by the apparatus.

ER is an important index related to the consumer perception of extruded products. The expansion concludes the longitudinal expansion, which is difficult to measure, and the radial expansion determined in Figure 2a as ER. Looking at Figure 2a through Table 1, ER was significantly affected by extrusion conditions. Better expansion can be found at a high temperature and low moisture, for example, P₁ and SP₁ (2.584 mm/mm and 2.35 mm/mm, respectively). Furthermore, similar to the results of Ding et al., (2006), Spinello et al., (2014) and Thymi et al., (2005), a decrease in temperature and an increase in moisture could reduce the ER - a phenomenon which is unsatisfactory in the extrusion industry. During the extrusion process, a high temperature and shear force in the barrel induced order-disorder transformations of starch and bubble nucleation, which meant viscoelastic starch mass was caught and entrapped in the small air bubbles. Then, the pressure difference between the pressure inside the die and the atmospheric pressure led to water evaporation and bubble growth. At a high temperature, starch molecules gelatinized to form a stable expanded structure. The thermal degradation of starch led to a lower melting viscosity, accompanied by less resistance to bubble growth, and resulting in larger ER. Moreover, because of the change in amylopectin molecular structure under high moisture content conditions, the melting elasticity reduced, and thus decreased the porosity values and expansion ratio (Thymi et al., 2005).

ER depends not only on extrusion parameters and equipment parameters, but also on the performance of raw materials during extrusion. The ER of SP products was lower than that of P products. On the one hand, the lower starch and higher fiber contents in SP (Table 2) could decrease the ER (Wan et al., 2018); on the other hand, SP contains amylase, which hydrolyzes starch into sugar. Consequently, the existence of sugar may yield poor starch conversion, encourage shrinkage of the bubble walls and cause a reduction in expansion. It has also been reported that extrudability is slightly different among different food raw materials (Corn starch > P starch > SP starch) (Bai et al., 2018; Wan et al., 2018).

SV, the volume of sample per unit mass, is a measurement of overall expansion. In general, a higher SV results in higher porosity and better texture of the products. Significant differences were observed among the extrusion parameters and starch types. As shown in Figure 2b, the SV of samples ranged from 0.81 to 4.84cm³/g and products with elevated SV were generated from high temperatures and low moisture. The extrusion of cassava and potato flours showed comparable results (Spinello et al., 2014).

The influence of extrusion variables on the color of the samples was determined (Figure 2c). The variables (temperature and moisture content) displayed a significant effect on the ΔL (lightness), Δa (redness), and Δb (yellowness) of the samples. The ΔL and Δa increased, while Δb decreased as barrel temperature increased (P₁-P₃, SP₁-SP₃), indicating that the product was redder and yellower. However, Δa also increased with the temperature decrease and moisture content increase in P₃-P₅ and SP₃-SP₅. It is known that high temperature and low humidity extrusion processes are conducive to Maillard reaction between free amino groups of protein and carbonyl groups of reducing sugars. The more starch degraded, the more reactive small molecule reducing sugar generated. In a previous study with corn rich in lysine as the main raw material, the two amino groups of lysine enhanced the Maillard reaction (Singh et al., 2007). Owing to the melanoids produced by the Maillard reaction and the tocoquinones formed by the thermal oxidation of tocopherols at a high temperature, the redness and yellowness of P₁ and SP₁ were higher. Besides, the β-carotene in corn, SP and RPO gathered in smaller products prepared by high moisture extrusion. More β-carotene in SP also contributed to higher Δa of products. This color phenomenon was also observed in SP raw materials. In brief, P₁ and SP₁ with preferred expansion and proper color showed better product characteristics.

The maximum and minimum sensory scores corresponded to P₁, SP₁ and P₅, SP₅, respectively (Figure 2d). The sensory properties of extruded snacks are associated with appearance, texture and flavor characteristics. Apart from similar shapes, the main difference in appearance between the samples was color (Figure 1). The texture, which is one of the most important qualities consumers look for, is closely connected to expansion (ER in Figure 2a). Looking at product ER with color in Figure 2 and product appearance in Figure 1, the sensory score gradually decreased with the shrinkage and deepened the color of the products. Furthermore, in Figure 2, SP showed poorer extrusion behaviors and darker color than P, thus making their sensory scores slightly lower. Generally, the results for ER, SV, color and product appearance affirm the changes in the sensory scores in Figure 2.

WSI and WAI are the interaction between samples and water. WSI is the solubility of extrudates in water, characterizing food digestibility; while WAI is the volume of swelled starch in water. Higher WAI improves handling characteristics and prevents the drying of products during storage (Ding et al., 2006). It was observed that WSI and WAI decreased obviously with a decrease in expansion (Figure 2e, Figure 2f), which was mainly attributed to the less conversion of starch at a low temperature. A similar observation was reported by Ding et al., (2006) and Ali et al., (2017). High temperature extrusion aggravated starch degradation. Soluble polysaccharides and small molecule sugars were released, causing a positive effect on WSI. In addition, crystalline starch is transformed into amorphous starch at high temperatures, inducing the binding of hydroxyl groups with water, thereby resulting in high WAI (Spinello et al., 2014). Expectedly, the process performance of SP was lower than P in terms of WSI and WAI based on a lower starch content. The changes in ER, SV, WSI and WAI are mainly related to starch conversion in materials as evidenced in this study.

A considerable amount of β-carotene remained in the samples after extrusion. Presented in Table 3, the β-carotene content in RPO before extrusion was 524.3 mg/kg, after extrusion it was 414.7-498.5 mg/kg, and the retention rate was 79.1-95.1%. The concentration of β-carotene in the extruded food was higher than that of baked food. RPO has also been reported in other functional foods to improve malnutrition in children (Sidhu et al., 2004). Therefore, it is an indisputable fact that red palm oil helps to improve the nutritional quality of foods.

However, the β-carotene retention properties of the samples were quite different from product properties. For instance, P₁ and SP₁ had better product properties; while their β-carotene contents were lower than those of P₃ and SP₂ (Table 3). The phenomenon can be explained such that the extrusion temperature affected the degradation and oxidation of the highly unsaturated β-carotene. A lower retention of carotene at high temperatures was also suggested by Qian et al., (2012). Moreover, the actual dissolved oxygen content in the materials would increase with the increase in water. According to Buton et al., (1984), the oxidation of carotene accelerates under high oxygen partial pressure. Hence, water was the main factor that affected the loss in β-carotene when the temperature was below 100 ºC and the water flow rate was above 70 mL/min (P₃-P₅, SP₃-SP₅) (Table 3). Furthermore, P samples showed better product characteristics, but lower β-carotene content because of the inherent β-carotene in SP.

Extruded products with RPO have potential as a functional food alternative for the intake of β-carotene, since it is a precursor of vitamin A, which cannot be synthesized in the human body. Due to the rupture of the links with other macromolecules, thermal treatment is beneficial for the absorption of carotenoids. Oleic acid, which is abundant in RPO also increases the accessibility and bioavailability of β-carotene (Victoria-Campos et al., 2013). For provitamin A activity, β-carotene plays an important role in cell differentiation, embryonic development, vision and prevention of certain types of cancer (Qian et al., 2012).

Similar to β-carotene, the squalene content in the SP samples was also higher than in the P samples, which was in the range of 147.5-180.5 mg/kg, and the retention rate was 76.5-97.2% (Table 3). By contrast, the effect of extrusion conditions on squalene was not exactly the same as the effect on β-carotene. Squalene in extruded products showed a general trend of descending first and then ascending. According to Ogrodowska et al., (2014), high temperature and high-pressure treatments contributed to squalene losses, which is consistent with the results obtained in this research. The extrusion processing system is complex and changeable. Therefore, the mechanism is far from clear.

Using RPO as the source of squalene may increase the content in squalene in the human body because squalene in food can be well absorbed, with the absorption rate reaching 60-85% (Reddy and Couvreur, 2009). In addition, squalene as a nutritional substance is attributed to the absence of cancer in sharks and the low incidence of human cancer among the population with high squalene intake. The anti-cancer effects of squalene include the protective role against carcinogens of hair-cell leukemia and skin cancer, and synergistic effect of antineoplastic drugs. Moreover, squalene has been shown to have powerful antibacterial and antifungal activity, and to protect against coronary heart disease (Reddy and Couvreur, 2009).

Tocotrienols and tocopherols in RPO before extrusion and in different oil samples extracted from extruded snacks were identified (Table 3). The contents in tocotrienols and tocopherols in RPO without extrusion were 348.1 mg/kg and 299.8 mg/kg, respectively, and their sum was 647.9mg/kg. Extrusion caused a loss in these substances as evidenced in the samples. Among the several active isomers, the highest contents and the most serious losses were observed in β/γ-tocotrienol and α-tocopherol. Such phenomenon was in agreement with those reported for baking, deep-frying, and other thermal processing. Hamid et al., (2014), and Schroeder et al., (2006) reported the loss sequence in vitamin E as: γ-tocotrienol > α-tocotrienol ≈ α-tocopherol > δ-tocotrienol. However, the retention of the isomers was not identical in previous reports. The stability of the different isomers is related to concentration, temperature, oil system and other factors (Rossi et al., 2007).

The retention of tocotrienols and tocopherols was under the influence of process parameters. It is obvious that the tendencies of β/γ-tocotrienol, α-tocopherol, and total tocotrienols and tocopherols resembled β-carotene. The loss in vitamin E was at its maximum at high temperatures and low moisture contents (P₁, SP₁ in Table 3, Table 1). The thermal degradation of vitamin E in oils may be aggravated by the increase in temperature (Hamid et al., 2014). Moreover, the influence of water on the degradation of vitamin E was greater than temperature. High water content caused the loss in vitamin E easily by increasing the fluidity of metal catalysts (Klibanov, 1986). The contents in tocotrienols and tocopherols in products was also related to the formula of raw materials. Tocotrienols and tocopherols in SP products were higher than those in extruded P products (Table 3), which were also consistent with β-carotene. It is easy to conclude that SP samples were more nutritious.

The tocotrienols and tocopherols are efficient antioxidants in edible oil and are major bioactive compounds in vivo. Their presence retards the autoxidation of oils and fats greatly and protects fatty foods from off-flavors. In addition, they possess gene regulatory functions (α-tocopherol), natriuretic, anti-inflammatory, antitumor activities (γ-tocopherol), neuroprotective properties (especially α-tocotrienol), preventive effects on cholesterol biosynthesis, and anticancer effects (δ- and γ-tocotrienol) (Kamaleldin and Appelqvist, 1996). In general, considerable tocotrienol and tocopherol retention in food is beneficial to both food storage and human health.

PV and p-AV were used to evaluate the deterioration of oil in the extrusion process. PV represents the main primary products of oil oxidation, like hydrogen peroxide; while p-AV represents the small molecular compounds such as aldehydes and ketones produced by further decomposition of hydrogen peroxide. High PV and p-AV indicate high oxidation of oil samples. However, as shown in Table 3, the PV and p-AV of RPO after extrusion (4.7 meq O2/kg oil and 3.4, respectively) were quite low, thus implying less deterioration and good applicability of RPO in extrusion.

There were minor differences in PV and p-AV among the different samples (Table 3). It was expected that the PV and p-AV of oil would increase at high temperatures and low moisture. PV and p-AV generally elevate with increases in temperature and processing time (Wan et al., 2018). In addition, as early as 1987, studies have shown that a very small content in water (0.2%) can reduce the catalytic activity of metal ions and benefit the stability of lipids; whereas an increased amount of water may promote the hydrolysis of lipids and accelerate oxidation (Chan, 1987). This result was further supported by the retention of β-carotene, tocotrienols and tocopherols. These substances are considered to be antioxidants in food systems which can inhibit the oxidation chain reaction by reacting with peroxyl radicals. Nevertheless, the reduction in the PV value of RPO-mixed products after extrusion as seen in this study could be attributed to the degradation of peroxides in RPO by heat, which increased the amount of secondary product like p-AV (Ekwenye, 2006).

Under the same extrusion conditions, the PV of SP and P products showed no significant difference, yet the p-AV of P products was slightly higher than that of SP (Table 3). This difference could be attributed to the different composition of P and SP. P starch or SP starch were about 79 or 62%, respectively; different types and amounts of starch may respond differently to thermal and mechanical energies during extrusion (Wan et al., 2018). RPO extruded with SP might reduce the heat transfer between barrel-oil and the contact of oil with oxygen, leading to the decreased oxidation of SP samples. The oxidation of oil can be influenced by many factors, such as unsaturated fatty acid content, antioxidants, metals and photosensitive substances, as well as the moisture, temperature, and oxygen contents in storage containers (Chan, 1987).

Rancimat assay characterizes the oxidation stability of oils. Higher IP reflects a longer shelf-life of products. At 100 ºC, the IP of RPO was 41.4 h (Table 3). The oxidative stability of RPO was higher than that of many oils as reported by Kowalski et al., (2004). Even though extrusion accelerates the oxidation reaction, resulting in the decline of IP, in this study (Table 3), the IPs of oil after extrusion processing were substantial (11.4-29.0 h).

The RPO extruded with P showed no obvious regularity in IP, but the opposite was seen in the RPO extruded with SP. The IP of SP samples decreased with the increase in temperature and water contents. And water can be the dominant factor when water and temperature change simultaneously. These phenomena are well supported by the changes in β-carotene, tocotrienol and tocopherol contents, as well as PV and p-AV. The samples with high micronutrient retention showed less oxidation and greater stability. Endogenous antioxidant components are considerable factors affecting oil stability (Mba et al., 2015). The existence of double bonds in β-carotene and hydroxyl groups in tocopherols can inhibit lipid oxidation by capturing free radicals. Previous studies have also shown that carotenoids have a synergistic effect with tocotrienols and tocopherols (Schroeder et al., 2006). Thus, RPO is a high-quality oil for thermal extrusion processing.