Processing and optimisation of complementary food blends from roasted pearl millet (Pennisetum glaucum) and soybean (Glycine max) using response surface modeling

Abstract

Adequate nutrition is vital during infancy but the high cost of supplemented infant formulae has forced inhabitants of Central and West Africa to depend solely on low-nutrient gruels. Response Surface modelling was used to process a complementary food from roasted pearl millet and Soybean flour. A central composite design was adopted to study the effects of feed composition X₁ (5.86–34.14%) and roasting temperature X₂ (126–154 °C) on the micronutrients, functional, and sensory profiles of the different blends. The responses were significantly (p < 0.05) affected by the independent factors. For the vitamins in mg/100 g, the thiamin, riboflavin, folate, and β-carotene content ranged from 0.17–0.33, 24–53.50, 1.32–2.29, and 7–22.98, respectively. For the minerals in mg/100 g, the zinc, calcium, potassium, and iron content ranged from 0.35–0.54, 39.5–62.75, 1.2–1.8, and 0.017–0.18, respectively. The viscosity, bulk density, swelling capacity, water absorption capacity, and pH ranged from 1577.5–942.5 cP, 0.74–0.79 g/cm³, 0.10–0.30 ml/g, 1.2–1.4 ml/g, and 4.70–5.70, respectively. The sensory scores were rated highly by the panelists. The optimum processing conditions with a desirability of 0.50 gave 29.28% and 130.39 °C feed composition and roasting temperature, respectively.

Introduction

Adequate nutrition during infancy and childhood is vital to the proper growth of a child to a healthy adult. According to a publication by Hayashi et al. (2020), 144 million children under the age of five are globally stunted, 47 million are wasted, and 38.3 million are overweight. A greater proportion of these indecencies are reported to be in Africa where 57.5 million children in this age group are stunted, 12.7 million wasted while 9.3 million are overweight. High incidences in Africa and other developing countries have been attributed to the high cost of supplemented infant formulae, forcing inhabitants to depend solely on low nutrient gruels to complement the infant diet (Abeshu et al. 2016). This affects the development of the infant with consequences being irreversible stunted growth, low brain development, reduction in educational performance, and increased mortality and morbidity rates (Hayashi et al. 2020; Keyata et al. 2021). This can be prevented by developing nutritious and cheap complementary foods from nutrient-dense locally available crops (Joel et al. 2019).

Complementary foods are nutritious liquids, semi-solid and/or solids foods that are usually given to older infants and young children between the ages of 6 to 23 months once breast milk can no longer meet the recommended nutritional needs of these children (Abeshu et al. 2016). With an increase in age from 6 months, there is an increase in the inability of breast milk to provide the infants with the nutritional requirements for growth and development. Complementary foods bridge the gap between this limitation of breast milk and the increasing energy needs of infants (Abeshu et al. 2016). It can be produced from grains such as millet, sorghum, cowpea and soybean, vegetables, fruits, meat and other protein-rich foods modified to a texture appropriate for infant development (Keyata et al. 2021).

A proper blending of grains such as millet, sorghum, cowpea and soybean, with vegetables, fruits, meat, and other protein-rich foods have been demonstrated from previous studies to have a tremendous potential to provide the desired nutrient requirements of complementary foods (Keyata et al. 2021). Millets like any other cereals are deficient in lysine which is an essential amino acid needed for muscle protein synthesis, hormones, calcium absorption, synthesis of antibodies, and a wide range of actions in the infant brain (Moya 2016). Supplementation of cereals with legumes such as soybean in complementary foods have become a very popular attempt to complement nutrient deficiency and compensate for the limiting lysine in cereals (Ayele et al. 2022). Soybean is rich in proteins, starch, minerals, and vitamins and have also important health-protective compounds (phenolics, inositol, and phosphates) with about 3% lecithins which are helpful for the infant brain development (Kim et al. 2021; Yalcin and Basman 2015, 2016).

The energy and nutrient requirements of infants differ for the age group and from one individual to another. Breast milk can cover all the nutritional needs of infants up to 6 months, but after 6 months there is an energy gap that needs to be covered by complementary foods (Abeshu et al. 2016). The overall energy needed for optimal functioning for infants between 6 and 8 months is about 200 kcal per day, infants aged 9–11 months is 300 kcal per day, and infants between the ages of 12–23 months require 550 kcal per day. This discrepancy has made it difficult to use the locally available food crops to formulate complementary foods to cut across these ages (USDA 2019). However, this problem can be addressed by using the response surface methodology (Myers et al. 2009).

Hence, the objective of this research was to use the response surface modelling to process and optimize a complementary food from blends of roasted pearl millet (Pennisetum glaucum) and Soybean (Glycine max) and generate models that can be used to produce other formulations depending on different nutritional needs of infants.

Materials and methods

Procurement of raw materials

The soybean and pearl millet were procured from Ogige Main Market in Nsukka, Enugu State, Nigeria.

Sample preparation

Preparation of soybean flour

The soybean flour was produced using the modified method of Mariam (2005). The soybean with different grain sizes was winnowed, hand-picked and sorted to remove bad grains. The cleaned soybean was washed using cold water and soaked at 28 °C for 12 h. The water was drained and the soybean was dehulled manually by macerating in between the palms, washed severally with enough cold water and air-dried at 28 °C for 12 h to remove the excess moisture absorbed during soaking. The dried cotyledons were roasted in a grain roaster at different temperatures (126, 130, 140, and 154 °C) and allowed to cool. It was then ground using a hammer mill (Viking Hammer mill, Horvick manufacturing model No. IG.01.09.2H.H3) and sieved into fine flour using a 4.2 µm mesh sieve.

Preparation of millet flour

The millet flour was produced using the modified method of Badau et al. (2006). The millet grains were cleaned to remove extraneous materials, winnowed and sieved with a mesh to removed stones. The millet was soaked in water at 28 °C for 92 h using lime juice of pH 3.7 in order to remove the ash colour of the grains and also to reduce the amount of anti-nutrients present in the grains. After soaking, the grains were washed thoroughly and dried in an oven at 50 °C for 8 h. It was milled using a hammer mill (Viking Hammer mill, Horvick manufacturing model No. IG.01.09.2H.H3) and sieved using a 4.2 µm mesh sieve.

Experimental design and sample formulation

A two-factor central composite design (CCD) (Myers et al. 2009) was adopted to study the effect of feed composition (${\text{X}}_1$) and roasting temperature (${\text{X}}_2$) on the chemical composition, functional properties and acceptability of the complementary foods. A quadratic polynomial regression model was assumed for predicting individual responses as presented in Eq. 1.

$$\text{Y}={\mathrm{\beta }}_0+{\mathrm{\beta }}_1{\text{X}}_1+{\mathrm{\beta }}_2{\text{X}}_2+{\mathrm{\beta }}_{12}{\text{X}}_1{\text{X}}_2+{\mathrm{\beta }}_{11}{\text{X}}_1^2+{\mathrm{\beta }}_{22}{\text{X}}_2^2+\mathrm{\epsilon }$$ 1

where Y = The Response, ${\text{X}}_1$ = Feed Composition, ${\text{X}}_2$ = Roasting temperature, ${\mathrm{\beta }}_0$ = intercept, ${\mathrm{\beta }}_1$, and ${\mathrm{\beta }}_2$ are linear terms, ${\mathrm{\beta }}_{11}$ and ${\mathrm{\beta }}_{22}$ are the quadratic regression coefficient terms, ${\mathrm{\beta }}_{12}$ is the linear interaction term and $\mathrm{\epsilon }$ = random error.

Determination of selected micronutrient content of samples

Determination of minerals

The mineral content was determined using the Atomic Absorption Spectrophotometer (AAS) (Buck Scientific Atomic Absorption Emission Spectrophotometer model 205, manufactured by Norwalk, Connecticut, USA) using standard wavelengths. Two grams of the samples were ashed following standard AOAC (2010) methods. The ashed samples were digested with 2.5 ml of 0.03 N hydrochloric acid (HCl). The digest was boiled for 5 min, allowed to cool to room temperature and transferred to a 50 ml volumetric flask and made up to the mark with distilled water. The resulting digest was filtered with ashless Whatman No. 42 filter paper. The filtrate from each sample was analyzed for mineral content (iron, zinc, calcium, magnesium, potassium) using an AAS. The real values of the minerals were extrapolated from the respective standard curves and values obtained were adjusted for HCl-extractability for the respective ions. The absorbance for zinc, calcium, magnesium, iron, and potassium were read at 258, 422.7, 285.2, 372.0, 766.5 nm, respectively.

Determination of vitamin content

Vitamins B1, B2, B9, and beta carotene was determined using the AOAC (2010) methods while beta carotene was determined using the method of Pearson (1976). One gram (1 g) of the sample each was extracted by mixing with 20 ml of petroleum ether. The extract was evaporated to dryness and the residue dissolved with 0.2 ml chloroform–acetic anhydride mixture. Two millilitres (2 ml) of trichloro acetic acid (TCA) was also added to the extract and mixed thoroughly and the absorbance read at 620 nm within 15 s. With the absorbance value, beta-carotene was calculated using Eq. 2.

$$\text{Concentration}\left(\text{mg}\right)=\frac{\text{Abs}\times \text{volume}\text{of}\text{cuvette}\times {\text{D}}_{\text{f}}}{\text{E}}$$ 2

where Abs = Absorbance, D_f = Dilution factor, E = Extinction coefficient.

Determination of some selected functional properties

Determination of bulk density (BD)

The bulk density of the flour blends was determined according to the method of Okaka (2005). A previously weighed measuring cylinder was filled to the 10 ml mark with the sample. The bottom of the cylinder was tapped gently but repeatedly on a laboratory bench until there was no further reduction of the sample level. The cylinder with the sample was weighed. The bulk density was calculated as:

$$\text{BD}\left(\frac{\text{g}}{{\text{cm}}^3}\right)=\frac{{\text{W}}_2-{\text{W}}_1}{\text{V}}$$ 3

where BD = bulk density (g/cm³), W₁ = weight of the empty cylinder (g), W₂ = weight of cylinder + sample (g), V = volume of cylinder occupied by the sample (cm³).

Determination of water absorption capacity (WAC)

Water absorption capacity was determined using the method of Kanu et al. (2009). One gram of the sample was introduced into a weighed centrifuge tube with 10 ml of distilled water and mixed thoroughly. The mixture was allowed to stand for one hour before being centrifuged at 350 rpm for 30 min. The excess water (unabsorbed) was decanted. The weight of water absorbed was determined by difference. The water absorption capacity was calculated using Eq. 4.

$$\text{WAC}=\frac{\text{Volume}\text{of}\text{water}\text{used}-\text{volume}\text{of}\text{free}\text{water}}{\text{Weight}\text{of}\text{the}\text{sample}}\times 100$$ 4

Determination of swelling capacity (SWC)

This was determined using the method described by Jongaroontaprangsee et al. (2007). Ten grams of each sample was measured into a 300 ml measuring cylinder. Then 150 ml of distilled water was added to each sample and allowed to stand for four hours. The final volume after swelling was recorded and percentage swelling capacity was calculated using Eq. 5.

$$\text{Swelling}\text{capacity}\left(\text{\%}\right)=\frac{\text{Final}\text{volume}-\text{initial}\text{volume}}{\text{Initial volume}}\times 100$$ 5

Determination of viscosity

The viscosity was determined using a torsson viscometer (Gallenkamp, England). Ten percent (10%) (w/v) of each of the flour blends was heated for 10 min at 100 °C to form a gruel, each gruel was cooled to 28 °C before viscosity measurement was taken from the viscometer.

Determination of pH

The pH of the flour blends was measured in a 10% (w/v) dispersion of the samples in distilled water. Each suspension was mixed thoroughly. A standard pH meter (Hanna meter model H196107) was used for pH determination. The pH electrode was dipped into the solution and after a few minutes of equilibration, the pH of the samples was taken.

Sensory evaluation

The nine-point Hedonic scale was used to evaluate the sensory parameters which were colour, appearance, texture, taste, consistency, flavor and overall acceptability of the complementary food. Where 9 = like extremely, 8 = moderately like, 1 = extremely dislike. The evaluation was done by 20 semi trained panelists through random selection of students from the Department of Food Science and Technology, University of Nigeria, Nsukka. The samples were coded and presented in coded plastic plates to the panelists randomly. Table water was presented to the panelists to rinse their mouth after testing each sample. The panelists were also instructed not to make comments during evaluation to avoid influencing other panelists. They were also asked to make comment freely on samples on the questionnaires given to them.

Optimization of the complementary food

The desirable goals in the optimization process was to; maximize the beta carotene content, keep B₁ between 0.2 and 0.4 mg/100 g, keep B₂ between 0.3 and 0.5 mg/100 g, minimize the vitamin B9 content, maximize calcium content, maximize iron content, maximize zinc content, minimize potassium content, minimize the antinutritional factors, maximize the viscosity from 1000 cP, maximize the sensory scores of the complementary food samples, water absorption capacity, and the swelling capacity.

Statistical analysis

Statistical software (Design Expert Version 11 -STATEASE) was used to analyze the generated data from the experiments. The significance of the generated models were accepted at p < 0.05.

Results and discussion

Effects of feed composition and roasting temperature micronutrients

The micronutrient composition of the complementary food is presented in Table 1.

Table 1.

The effects of temperature and feed composition on the micronutrient content of complementary food produced from blends of pearl millet and soybeans

Sample	Factor		Responses
	X1 (%)	X2 (°C)	Iron (mg/100 g)	Zinc (mg/100 g)	Calcium (mg/100 g)	Potassium × 103 (mg/100 g)	Vitamin B1 (mg/100 g)	Vitamin B2 (mg/100 g)	Vitamin B9 (mg/100 g)	β-Carotene (mg/100 g)
1	10.00	130.00	0.10 ± 0.00	0.40 ± 0.00	55.75 ± 0.91	1.37 ± 0.00	0.17 ± 0.01	24.00 ± 0.91	1.46 ± 0.00	11.42 ± 0.02
2	30.00	130.00	0.16 ± 0.00	0.43 ± 0.00	62.50 ± 0.23	1.68 ± 0.01	0.23 ± 0.00	42.00 ± 0.32	1.96 ± 0.01	22.70 ± 0.91
3	10.00	150.00	0.15 ± 0.00	0.54 ± 0.01	52.50 ± 1.02	1.20 ± 0.00	0.18 ± 0.01	30.00 ± 1.90	1.32 ± 0.00	11.37 ± 0.12
4	30.00	150.00	0.06 ± 0.00	0.36 ± 0.01	58.65 ± 2.00	1.76 ± 0.02	0.28 ± 0.02	53.00 ± 0.02	1.87 ± 0.00	22.24 ± 0.88
5	5.86	140.00	0.18 ± 0.01	0.44 ± 0.02	55.75 ± 0.12	1.08 ± 0.09	0.16 ± 0.00	23.50 ± 0.23	1.37 ± 0.03	7.00 ± 1.01
6	34.14	140.00	0.15 ± 0.00	0.34 ± 0.01	62.75 ± 3.21	1.82 ± 0.02	0.33 ± 0.00	53.50 ± 1.11	2.29 ± 0.01	23.30 ± 0.02
7	20.00	125.86	0.04 ± 0.00	0.41 ± 0.00	52.50 ± 4.90	1.76 ± 0.00	0.22 ± 0.01	34.50 ± 1.87	1.73 ± 0.05	19.07 ± 0.21
8	20.00	154.14	0.04 ± 0.00	0.40 ± 0.00	49.00 ± 3.21	1.48 ± 0.00	0.25 ± 0.02	43.00 ± 0.90	1.82 ± 0.00	19.65 ± 0.22
9	20.00	140.00	0.10 ± 0.00	0.31 ± 0.02	40.50 ± 0.23	1.65 ± 0.02	0.20 ± 0.02	44.00 ± 1.99	1.77 ± 0.09	12.14 ± 0.01
10	20.00	140.00	0.10 ± 0.00	0.32 ± 0.08	42.50 ± 0.94	1.63 ± 0.00	0.21 ± 0.01	47.00 ± 0.51	1.78 ± 0.01	12.26 ± 0.01
11	20.00	140.00	0.10 ± 0.00	0.33 ± 0.02	39.50 ± 0.34	1.74 ± 0.04	0.23 ± 0.01	43.00 ± 0.02	1.90 ± 0.00	12.44 ± 0.09
12	20.00	140.00	0.10 ± 0.00	0.31 ± 0.00	43.40 ± 2.90	1.72 ± 0.01	0.24 ± 0.00	42.00 ± 0.01	1.75 ± 0.00	12.05 ± 0.11
13	20.00	140.00	0.12 ± 0.01	0.35 ± 0.04	45.50 ± 2.11	1.68 ± 0.00	0.22 ± 0.00	43.50 ± 1.44	1.80 ± 0.04	12.63 ± 0.04

X₁, Feed composition (%), X₂, Roasting temperature (°C), values are presented in means ± standard deviation

Iron content

Results of the iron content as presented in Table 1 showed values ranging from 0.017 to 0.18 mg/100 g. An increase in both feed composition and roasting temperature led to a decrease in the iron content (Fig. 1a). Increased substitution of millet with soybean also reduced the iron content. This could be attributed to the dilution effect of soybean due to its lower iron content compared to millet. A similar observation was made by Bolarinwa et al. (2016) for a complementary food from malted millet, soybean, and plantain flour blends. The recommended dietary allowance (RDA) of zinc for infants between the ages of 7–12 months is 11 mg/100 g (USDA 2019). Hence, the formulated samples could provide just about 1.64% of RDA for infants between 7 and 12 months. The quadratic modelling equation (Eq. 6) proposed to model the responses was significant (p < 0.05) and could describe the iron content by 90%. The model also had an insignificant (p > 0.05) lack-of-fit with an adequate precision greater than 4. For a model to be deemed adequate, the model should be significant (p < 0.05), R² > 0.8, lack-of-fit should be insignificant (p > 0.05) and adequate precision should be greater than 4 (Myers et al. 2009).

$$\text{Zinc}\text{content}=0.08-0.01{\text{X}}_1-0.01{\text{X}}_2-0.04{\text{X}}_1{\text{X}}_2+0.05{\text{X}}_1^2-0.02{\text{X}}_2^2$$ 6

Fig. 1.

3D Effect of feed composition and roasting temperature on a zinc, b calcium, c potassium, and d iron content, e Vitamin B₁, f Vitamin B₂, g Vitamin B₉, and h Beta carotene

Zinc content

An increase in feed composition reduced the zinc content while an increase in temperature increased the zinc content (Fig. 1b). This could be due to high temperatures reducing moisture content in the complementary food and increasing the concentration of zinc and other nutrients. Results obtained for the zinc content were in the range of values (0.23 to 0.68 mg/100 g) reported by Mbaeyi-Nwaoha and Obetta (2016). The RDA of zinc for infants between the ages of 7–12 months is 3 mg/100 g (USDA 2019). The values obtained for the complementary food could provide about 11% of the RDA in infants. The quadratic model (Eq. 7) proposed was significant (p < 0.05) and could describe the zinc content by 91%. This model had an insignificant (p > 0.05) lack-of-fit, with an adjusted R² of 0.85 and an adequate precision of 10.9.

$$\text{Zinc}\text{content}=0.32-0.04{\text{X}}_1+0.01{\text{X}}_2-0.05{\text{X}}_1{\text{X}}_2+0.04{\text{X}}_1^2+0.05{\text{X}}_2^2$$ 7

Calcium content

The values ranged from 39.5 to 62.75 mg/100 g with significant (p < 0.05) differences observed with changes in the formulation, feed composition, and processing temperature. Holding temperature constant and increasing the feed composition as presented in Fig. 1c led to a steady increase in the calcium content. On the other hand, holding the feed composition constant and increasing the temperature did not significantly (p > 0.05) affect the calcium content. A similar effect of temperature on the calcium content was observed by Asefa and Melaku (2017) and within the range of values reported by Mekuria et al. (2021) for complementary foods from staple grains and honey larvae. The adequate intake value of calcium for infants aged between 7 and 12 months is 260 mg/100 g per day (USDA 2019). This implied the formulated complementary food samples could provide about 43% of the required values. The quadratic model (Eq. 8) proposed was significant (p < 0.05), R² > 0.80, lack-of-fit insignificant (p > 0.05), Adjusted R² > 0.80, and the adequate precision was greater than 4.00.

$$\text{Calcium}\text{content}=42.28+2.85{\text{X}}_1-1.51{\text{X}}_2-0.15{\text{X}}_1{\text{X}}_2+9.07{\text{X}}_1^2+4.82{\text{X}}_2^2$$ 8

Potassium content

The potassium content ranged from 1.2 to 1.82 mg/100 g with a significant (p < 0.05) difference among the samples (Table 2). As presented in Fig. 1d, holding temperature and increasing the feed composition led to a steady increase in the potassium content. On the other hand, holding the feed composition and increasing the temperature reduced the potassium content. The RDA of potassium is 600 mg/day for infants between the ages of 4 to 12 months (USDA 2019). Hence, the formulation could provide the RDA for infants within this age group. The modelling equation (Eq. 9) proposed to model the potassium content was found significant (p < 0.05) and adequate to navigate the design space. This model could explain the variation in the potassium content by 96%.

$$\text{Potassium}\text{content}=1.68+0.24{\text{X}}_1-0.06{\text{X}}_2+0.06{\text{X}}_1{\text{X}}_2-0.13{\text{X}}_1^2-0.04{\text{X}}_2^2$$ 9

Table 2.

The effects of temperature and feed composition on the physicochemical properties of complementary food produced from blends of pearl millet and soybeans

	X1 (%)	X2 (°C)	Bulk density (g/cm3)	WAC (mg/g)	Swelling capacity (mg/g)	Viscosity (cP)	pH
1	10.00	130.00	0.76 ± 0.01	1.30 ± 0.00	0.10 ± 0.00	1577.50 ± 9.01	5.00 ± 0.02
2	30.00	130.00	0.76 ± 0.00	1.30 ± 0.00	0.30 ± 0.01	1200.00 ± 10.01	5.70 ± 0.00
3	10.00	150.00	0.75 ± 0.02	1.30 ± 0.02	0.10 ± 0.00	1412.50 ± 9.09	5.00 ± 0.01
4	30.00	150.00	0.76 ± 0.00	1.30 ± 0.00	0.30 ± 0.01	937.50 ± 1.09	5.00 ± 0.00
5	5.86	140.00	0.76 ± 0.08	1.40 ± 0.01	0.10 ± 0.02	1522.50 ± 11.09	4.70 ± 0.00
6	34.14	140.00	0.75 ± 0.00	1.40 ± 0.01	0.30 ± 0.00	967.50 ± 2.09	5.60 ± 0.01
7	20.00	125.86	0.76 ± 0.01	1.20 ± 0.00	0.30 ± 0.02	1457.50 ± 2.10	4.90 ± 0.00
8	20.00	154.14	0.74 ± 0.02	1.10 ± 0.02	0.20 ± 0.01	1052.50 ± 12.89	5.40 ± 0.12
9	20.00	140.00	0.79 ± 0.04	1.20 ± 0.00	0.30 ± 0.00	942.50 ± 0.90	5.40 ± 0.08
10	20.00	140.00	0.79 ± 0.00	1.20 ± 0.02	0.20 ± 0.01	962.50 ± 10.19	5.40 ± 0.10
11	20.00	140.00	0.78 ± 0.00	1.10 ± 0.02	0.30 ± 0.00	912.50 ± 2.36	5.50 ± 0.00
12	20.00	140.00	0.79 ± 0.03	1.20 ± 0.01	0.20 ± 0.00	985.00 ± 3.88	5.30 ± 0.02
13	20.00	140.00	0.79 ± 0.01	1.20 ± 0.01	0.20 ± 0.00	892.50 ± 7.12	5.70 ± 0.01

X₁, Feed composition (%), X₂, Roasting temperature (°C), values are presented in means ± standard deviation

Vitamin B₁ (thiamin)

There was a significant difference in the formulation with Vit B1 content ranging from 0.17–0.33 mg/100 g. From the 3D plot in Fig. 1e, an increase in soybean flour led to increased vitamin B₁ content of the samples. The daily recommended intake for infants within the ages of 7–12 months is 0.3 mg/day (USDA 2019). A change in roasting temperature had a significant (p < 0.05) negative influence on the vitamin B₁ content. This could be because vitamin B₁ is unstable and subject to change its structure at high temperatures (Devi 2015). Values obtained for thiamin content were within the range of values (0.24–0.81 mg/100 g) reported by Mekuria et al. (2021). The modelling equation proposed to navigate the design space was found adequate as this model showed significance (p < 0.05) and could explain variation in the thiamin content by 88%. The adjusted R² and the adequate precision were 0.79 and 9.90, respectively.

$$\text{Thiamine}=0.22+0.05{\text{X}}_1-0.01{\text{X}}_2+0.01{\text{X}}_1{\text{X}}_2+0.01{\text{X}}_1^2+0.00{\text{X}}_2^2$$ 10

Vitamin B₂ (riboflavin)

Vitamin B₂ was found to vary in the blends of the complementary food ranging from 24 to 53.5 mg/100 g. There was an increase in vitamin B₂ with an increase in both feed composition and temperature, with feed composition being the most influencing factor. This could be due to the high level of vitamin B₂ in soybean flour compared to millet flour. The recommended daily intake of vitamin B₂ for infants (7–12 months) is 0.4 mg/100 g (USDA 2019). This implied that the RDA for Vitamin B₂ could be met with the formulation. Values obtained for riboflavin were lower than the values of Mekuria et al. (2021). A synergistic interaction was observed between the feed composition and the temperature (Eq. 11). The model R², adjusted R², and adequate precision values were 0.98, 0.96, and 20.45, respectively. The model was significant (p < 0.05) while the lack-of-fit was insignificant (p > 0.05), indicating its adequacy.

$$\text{Riboflavine}=43.90+10.43{\text{X}}_1+3.63{\text{X}}_2+1.25{\text{X}}_1{\text{X}}_2-3.04{\text{X}}_1^2-2.92{\text{X}}_2^2$$ 11

Vitamin B₉

The folic acid (vitamin B₉) of the complementary food blends ranged from 1.32 to 2.29 mg/100 g. Folic acid is vital for producing red blood cells, as well as aiding in rapid cell division and growth of an infant with a recommended daily intake for infants within the age of 6–12 months being 80 µg/day (USDA 2019). From the 3D plot in Fig. 1g, keeping feed composition constant and increasing the temperature, a steady decrease in the folate content was observed. This could be due to the high instability of folic acid which is reported to degrade with high processing temperatures (Devi 2015). Values obtained for the folate content were higher than the values reported by Mekuria et al. (2021). The quadratic modelling equation (Eq. 12) had a significant (p < 0.05) effect and the lack-of-fit was insignificant (p > 0.05). The R², adjusted R², and adequate precision were 0.90, 0.82, and 11.26, respectively.

$$\text{Vitamin}\text{B}9=1.80+0.29{\text{X}}_1-0.01{\text{X}}_2+0.01{\text{X}}_1{\text{X}}_2-0.02{\text{X}}_1^2-0.05{\text{X}}_2^2$$ 12

Beta-carotene

The value for beta-carotene content ranged from 7 to 22.9 8 mg/100 g with an increase in feed composition resulting in an increase in the beta carotene content. Keeping the feed composition constant and increasing the temperature, a reduction in the beta carotene content was observed (Fig. 1h). Up to 50%, 25%, and 35% reduction in the beta carotene content has been reported during drying, cooking, cook + drain processes, respectively, for food products (Devi 2015). The RDA for beta carotene for infants aged 7–12 months is 500 µg/RAE per day with an Upper Limit of 600 µg/day of preformed vitamin A (USDA 2019). Values obtained for the complementary food indicated that about 4.66% adequate intake per 100 g could be provided. The quadratic model (Eq. 13) had a significant (p < 0.05) effect and was adequate to be used for the optimization process since R², adjusted R², and adequate precision values were 0.99, 0.98, and 0.97, respectively. The model also had an insignificant (p > 0.05) lack-of-fit.

$$\text{Beta}\text{carotene}=12.30+5.65{\text{X}}_1-0.04{\text{X}}_2-0.10{\text{X}}_1{\text{X}}_2+1.34{\text{X}}_1^2+3.45{\text{X}}_2^2$$ 13

Functional properties

The functional properties are presented in Table 2.

Bulk density (BD)

The bulk density of the complementary food as shown in Table 2 ranged from 0.74–0.79 g/cm³. From the 3D graph presented in Fig. 2a, and increase in either temperature or feed composition led to a steady decrease in the bulk density. Even though the single effects of feed composition and temperature reduced the bulk density, a significant (p < 0.05) synergistic interaction effect was observed (Eq. 14). Values obtained for BD were within the values (0.6–0.9) g/cm³) reported by Ikegwu et al. (2021). The lower the bulk density the smaller the packaging material required for the product. The modelling equation proposed to model the BD could explain about 94% variation in the design space. This model was significant (p < 0.05) with an insignificant lack-of-fit, and an adequate precision of 10.12.

$$\text{Bulk}\text{Density}=0.79+0.00{\text{X}}_1+0.00{\text{X}}_2+0.00{\text{X}}_1{\text{X}}_2-0.02{\text{X}}_1^2-0.02{\text{X}}_2^2$$ 14

Fig. 2.

3D Effect of feed composition and roasting temperature on a Bulk density, b WAC, c Swelling capacity, and d Viscosity

Water absorption capacity (WAC)

Water absorption capacity values varied significantly (p < 0.05) among the various blends of the complementary food samples. Feed composition negatively (− 2.68) influenced the water absorption though the effect was insignificant (p > 0.05). The least feed composition (5.86%) had the highest water absorption capacity value (140) which could be as a result of higher quantity of starch from millet. The water absorption capacity of starch granules is a function of the amylose content, amylose/amylopectin configuration, and molecular structure of the amylopectin (Singh et al. 2007). The observed variation in different flour blends could be due to different protein concentration, their degree of interaction with water and conformational characteristics (Butt and Rizwana 2010). The quadratic model was significant (p < 0.05) and could be describe variation in the WAC by 90%.

$$\text{WAC}=1.18+0.01{\text{X}}_1-0.02{\text{X}}_2+0.00{\text{X}}_1{\text{X}}_2+0.12{\text{X}}_1^2-0.01{\text{X}}_2^2$$ 15

Swelling capacity (SWC)

The swelling capacity of the complementary foods ranged from 0.10 to 0.30%. The Feed composition significantly (p < 0.05) influenced the swelling capacity while roasting temperature had an insignificant (p > 0.05) effect. From the 3D plot presented in Fig. 2c, an increase in feed composition resulted to an increase in the swelling capacity with a synergistic interaction between temperature and the feed composition. The synergistic interaction on the swelling capacity could be due to weakening of bonds to facilitate water penetration in the amorphous zones and leaching of amylose from the starch granules during the heating process. The swelling capacity of food depends on the size of particles, types of variety and type of processing methods/unit operations applied in the flour production (Oyeyinka et al. 2020).

$$\text{Swelling}\text{capacity}=0.24+0.09X_1-0.02{\text{X}}_2+0.00{\text{X}}_1{\text{X}}_2-0.03{\text{X}}_1^2-0.00{\text{X}}_2^2$$ 16

Viscosity

The value of viscosity as presented in Table 2 ranged from 892.5 to 1577.5 cP. The viscosity of the complementary food decreased with an increase in both the feed composition and temperature (Fig. 2d). A reduction of viscosity with increase in feed composition could be attributed to the high protein content associated with soybean flour that may have a dilution effect on the starch content. The viscosity of an infant’s formula increases as the baby gets older and should range from 1700 to 2900 cP (The viscosity of most commercial gruels). Values obtained for the viscosity were slightly lower below this range. Values obtained were higher than values reported by Asefa and Melaku (2017). The quadratic model (Eq. 17) proposed could describe variation in the viscosity by 99%. The model was significant (p < 0.05) with an insignificant lack-of-fit and an adequate precision value of 24.74.

$$\text{Viscosity}=939.00-204.69X_1-125.04{\text{X}}_2+24.38{\text{X}}_1{\text{X}}_2+161.00{\text{X}}_1^2+166.00{\text{X}}_2^2$$ 17

pH

The pH of the complementary foods ranged from 4.70 to 5.70 with the least feed composition (5.86), having the highest acidity level which could be a result of soaking of the millet grains in water before processing. An increase in feed composition led to steady increase in the pH values. On the other hand, an increase in temperature led to a decrease in the pH. This could be due to reduction of moisture at higher temperatures leading to a more concentrated product. The quadratic model proposed to monitor the pH level was insignificant (p > 0.05). The R³, adjusted R², and the adequate precision were 0.53, 0.20, and 3.35, respectively. This implied the inadequacy of the model and its limitation to predict other values within the experimental domain.

$$\text{pH}=5.08+0.121X_1-0.12{\text{X}}_2+0.15{\text{X}}_1{\text{X}}_2+0.11{\text{X}}_1^2+0.21{\text{X}}_2^2$$ 18

Sensory scores of the complementary food

The sensory scores of the complementary food is presented in Table 3.

Table 3.

The effects of temperature and feed composition on the sensory scores of a complementary food produced from blends of pearl millet and soybeans

Sample	Factor		Responses
	X1 (%)	X2 (°C)	Colour	Flavour	Taste	Texture	Appearance	Consistency	Overall acceptability
1	10.00	130.00	6.31 ± 0.23	6.81 ± 0.12	6.32 ± 1.22	6.63 ± 1.21	6.12 ± 1.22	7.31 ± 1.20	7.12 ± 1.19
2	30.00	130.00	7.22 ± 0.43	6.44 ± 0.43	6.43 ± 2.32	7.30 ± 0.98	6.81 ± 2.12	6.22 ± 2.12	6.71 ± 1.43
3	10.00	150.00	6.91 ± 1.23	6.52 ± 0.44	6.27 ± 1.22	6.80 ± 2.21	6.31 ± 0.90	6.53 ± 0.21	6.54 ± 1.33
4	30.00	150.00	6.50 ± 2.43	6.08 ± 0.54	5.91 ± 1.09	7.20 ± 2.11	6.32 ± 0.32	6.90 ± 1.22	6.90 ± 2.12
5	5.86	140.00	6.23 ± 2.43	6.91 ± 1.23	6.41 ± 0.98	6.71 ± 1.90	5.93 ± 0.33	6.71 ± 2.13	6.30 ± 1.90
6	34.14	140.00	6.93 ± 0.34	6.43 ± 1.09	6.14 ± 1.11	7.33 ± 1.32	6.71 ± 1.23	6.42 ± 0.21	6.50 ± 1.32
7	20.00	125.86	6.96 ± 0.32	6.61 ± 2.00	6.31 ± 2.09	7.14 ± 1.78	6.50 ± 1.76	7.10 ± 1.23	7.30 ± 1.03
8	20.00	154.14	6.91 ± 1.09	5.93 ± 0.23	6.05 ± 0.12	7.12 ± 0.89	6.12 ± 2.10	6.80 ± 1.22	7.12 ± 1.33
9	20.00	140.00	7.62 ± 2.09	6.90 ± 0.32	6.71 ± 0.94	7.22 ± 0.21	7.01 ± 0.91	7.00 ± 1.98	6.93 ± 1.59
10	20.00	140.00	7.01 ± 1.33	6.84 ± 0.19	6.54 ± 1.32	7.14 ± 0.91	6.94 ± 1.33	7.11 ± 1.32	6.83 ± 1.30
11	20.00	140.00	7.21 ± 0.32	7.11 ± 1.11	6.63 ± 1.89	6.91 ± 1.67	7.02 ± 2.11	7.13 ± 2.01	6.92 ± 0.32
12	20.00	140.00	7.44 ± 0.90	6.94 ± 2.12	6.74 ± 2.01	7.04 ± 2.12	7.12 ± 0.34	6.84 ± 1.22	7.21 ± 0.90
13	20.00	140.00	7.31 ± 2.22	7.11 ± 1.67	6.81 ± 1.22	7.12 ± 1.43	6.80 ± 0.51	7.01 ± 0.23	7.04 ± 0.52

X₁, Feed composition (%), X₂, Roasting temperature (°C), values are presented in mean ± standard deviation

Colour

An increase in feed composition let to an increase in colour scores while an increase in temperature at constant feed composition let to a reduction in sensory scores (Fig. 3a). Scores obtained for colour were higher than values of Asefa and Melaku (2017). The quadratic model proposed to model the colour scores (Eq. 19) was found to be significant (p < 0.05). The model could describe variation in colour score by 88%. The adjusted R² and the adequate precision were 0.81 and 8.70, respectively, with an insignificant lack-of-fit. The pictorial representation of the complementary food is presented in “Appendix 1”.

$$\text{Colour}=7.30+0.19X_1-0.01{\text{X}}_2-0.33{\text{X}}_1{\text{X}}_2-0.38{\text{X}}_1^2-0.20{\text{X}}_2^2$$ 19

Fig. 3.

3D Effect of feed composition and roasting temperature on a Colour, b Flavour, c Taste, and d Texture, e Appearance, f Consistency, g Overall acceptability

Flavour

The flavour scores ranged from 5.9 to 7.1. From the 3D graph in Fig. 3b, an increase in both feed composition and roasting temperature resulted to a decrease in flavour scores. This could be attributed to the inherent beany flavour of soybean flour and browning of the product due to a high roasting temperature. A similar observation was reported by Lou et al. (2022). The quadratic model proposed (Eq. 20) could explain variation in flavour by 95%. This model was significant (p < 0.05) with an insignificant (p > 0.05) lack-of-fit and an adequate precision was 13.37 to indicate a good signal to noise ratio.

$$\text{Flavour}=6.96-0.20X_1-0.21{\text{X}}_2-0.03{\text{X}}_1{\text{X}}_2-0.16{\text{X}}_1^2-0.36{\text{X}}_2^2$$ 20

Taste

The taste ranged from 5.9 to 6.8 for samples 13 and 4, respectively. From the 3D plot in Fig. 3c, increased feed composition and roasting temperature led to a significant (p < 0.05) decrease in taste scores. Values obtained were higher than values reported by Asefa and Melaku (2017) for a finger millet, soybean and carrot-based complementary food. The model proposed was significant (p < 0.05) with an R², adjusted R², adequate precision values of 0.93, 0.89, and 11.97, respectively.

$$\text{Taste}=6.66-0.08X_1-0.13{\text{X}}_2-0.10{\text{X}}_1{\text{X}}_2-0.21{\text{X}}_1^2-0.26{\text{X}}_2^2$$ 21

Texture

The texture of the product ranged from 6.6 to 7.3. An increased feed composition and roasting temperature resulted to an increase in texture scores of the complementary food as rated by the panelists. An insignificant (p > 0.05) antagonistic interaction was observed between the roasting temperature and feed composition. Score obtained were higher the those reported by Lakshmi Devi et al. (2014). The modeling equation presented in Eq. 22 could explain variation in texture by 88%. This model was found adequate to navigate the design due to its high level of significance.

$$\text{Texture}=7.06+0.24X_1+0.01{\text{X}}_2-0.08{\text{X}}_1{\text{X}}_2-0.05{\text{X}}_1^2+0.00{\text{X}}_2^2$$ 22

Appearance

The appearance ranged from 5.9 to 7.1 with the highest score observed in sample 12 while the least score was found in sample 1. From the 3D plot in Fig. 3e, an increased feed composition had a positive influence on the appearance score while increased in roasting temperature resulted to a decrease in appearance of the complementary food. The proposed modelling equation had a high R² (0.95), an adjusted R² of 0.91, and adequate precision of 12.92. This implied the adequacy of this model to be used for the optimization process.

$$\text{Appearance}=6.96+0.23X_1-0.11{\text{X}}_2-0.18{\text{X}}_1{\text{X}}_2-0.31{\text{X}}_1^2-31{\text{X}}_2^2$$ 23

Consistency

Increased in feed composition and roasting temperature decreased consistency scores. The negative influence of the feed composition could be attributed to the high protein content in soybean flour. Sensory score obtained for the consistency were higher than those reported by Martin et al. (2010) for a soybean-based complementary food. The interaction between the feed composition and temperature resulted to a significant (p < 0.05) synergistic effect. The modeling equation proposed was significant and could describe variation in the consistency by 93%.

$$\text{Consisency}=7.00-0.14X_1-0.07{\text{X}}_2+0.38{\text{X}}_1{\text{X}}_2-0.23{\text{X}}_1^2-0.03{\text{X}}_2^2$$ 24

Overall acceptability

The overall acceptability ranged from 6.3 to 7.3. An increased in feed composition increased the overall acceptability while an increased in temperature led to a reduction in the overall acceptability. Values obtained for the overall acceptability were higher than those of and Mekuria et al. (2021). The modelling equation (Eq. 25) proposed to navigate the design domain was found significant (p < 0.05). This model could explain variation in the overall acceptability by 90%. The adjusted R² and the adequate precision were 0.83 and 11.74, respectively.

$$\text{Overall}\text{acceptability}=6.96+0.04X_1-0.09{\text{X}}_2+0.20{\text{X}}_1{\text{X}}_2-0.28{\text{X}}_1^2+12{\text{X}}_2^2$$ 25

Optimized formulation

The optimum roasting temperature was 130 °C while the optimum feed composition was 29.18%. This implied that if the complementary food is produced at these conditions, some of the recommended daily allowances for infants between the ages of 6–12 months will be met. The desirability function of this optimal formulation was 0.50.

Conclusion

Production of complementary foods from millet and soybean could provide the RDA of vitamins B1, B2, and B9 in infants between the ages of 6–12 months. The food could also provide the RDA values of potassium, 43% calcium, and 11% zinc. It was also observed in the study that an increase in feed composition plays a very important role in the nutritional, functional, and sensory properties while an increase in roasting temperature reduces the desirability of the complementary food and reduces it nutritional and sensory quality. The panelists also rated the samples high in terms of its sensory attributes with an overall acceptability of 6.37. The quadratic model that was proposed to monitor the responses was found to be significant accept for the pH. This quadratic model was successfully used to optimize the formulation and processing conditions and the optimal values of the independent factors that gave the desired formulation was 130 °C for the roasting temperature and 29.18% for the feed composition. This optimum conditions had a desirability of 0.5.

Appendix 1: Pictorial representation of the complementary foods

Author contributions

OEC: Conceptualization (lead); Data curation (equal); Methodology (equal); Resources (equal); Software (equal); Validation (equal); Writing–original draft (equal). Supervision (lead); ECO: Investigation (lead); Project administration (lead); Visualization (equal); Writing—original draft (equal); Resources (lead) ECR: Project administration (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing—original draft (equal); Writing—review and editing (equal). UEC: Methodology (equal); Writing–original draft (equal): Resources (equal): Writing—review and editing (equal). AMM: Writing–review and editing (lead): Data curation (equal); Validation (equal); Software (lead).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Code availability

Not applicable.

Declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Ethics approval

Not applicable.

Consent to participate

Not applicable.