Synthesis of natural starch from Elaeis guineensis trunk biomass applying bisulphite steeping method: Optimization by RSM

ABSTRACT

A massive quantity of Elaeis guineensis (oil palm) trunk biomass, containing a significant amount of natural starch, is available in Malaysia as biowaste because of annual replantation. The efficient extraction of this starch (carbohydrate polymer) would be worthwhile concerning the environmental sustainability and economy through conversion to bioresources. This study investigated the effectiveness of the bisulfite steeping method for starch synthesis from oil palm trunk (OPT) biowaste. The central composite design (CCD) of Design-Expert software executed an experimental model design, data analysis, evaluated the impacts of process variables and their interaction through response surface methodology to optimize the bisulfite steeping method for starch synthesis. The developed quadratic models for four factors (strength of sodium bisulfite solution, steeping hour, mixing ratio with the bisulfite solution, and ultrapure water) and one response (%Yield) demonstrated that a significant starch yield (13.54%) is achievable employing 0.74% bisulfite solution, 5.6 steeping hours, for 1.6 and 0.6 mixing ratio with the bisulfite solution and ultrapure water respectively. Experimental outcomes were consistent with the predicted model, which eventually sustains the significance of this method. Malvern Zetasizer test revealed a bimodal granular distribution for starch, with 7.15 µm of hydrodynamic size. Starch morphology was determined by scanning electron microscopy. X-ray diffraction investigation exhibits an A-type model, specifying persistent characteristics, while FTIR confirms the presence of hydroxyl, carboxylic, and phenolic groups like other cereal starches.

Implications: Malaysia is the 2nd largest palm oil exporter in the world. About 110 million tons of palm oil trunk (OPT) biomass is available annually during replanting activities. Modification of bio-wastes into a beneficial form (only 22% presently) like starch extraction would ensure potential reuse as a natural coagulant for wastewater and leachate treatment, food source, adhesives towards boosting the country’s economy by sustainable waste management. The current study achieved better starch yield (13.54%) than previous, from the OPT biomass through the novel bisulfite steeping method. Therefore, this method will ascertain the effective implication of numerous economic activities.

Introduction

Oil palm (Elaeis guineensis), a well-known monocotyledon perennial oilseed crop, is native to the Guinea Coast of West Africa and widely planted across 42 tropical countries, especially Malaysia and Indonesia (Murai and Kondo 2010). In 2019, Malaysia was the second among the global major oil palm producer countries, produced 20.5 million metric tons of crude oil (Kushairi et al. 2019) with 2.13 metric tons of palm kernel oil, while the ultimate plantation area of the palm oil trees was around 6.6 million hectares (Tan and Ho 2019). Malaysia continuously holds its position for palm oil production and export, attributing about 28% of oil production with 33% of export trade currently (Malaysian Palm Oil Industry – MPOC 2021). Besides this, the oil palm industry offers employment and livelihood to a massive number (around one million) of people. In tandem with increasing oil palm production and because of its unique application as well as incredible requirement, the corresponding increase of generated residues also inevitably happens. According to the study, the oil palm residues will reach up to 110 million tons by 2020, which is almost 86% of the total national biomass production (Dalton, Mohamed, and Chikere 2017). If not converted to bioresource, these huge biowastes will be a severe environmental burden and barrier to achieving sustainable development goals (SDG). Meanwhile, the UN already linked directly almost 12 goals to solid waste management globally, among the 17 SDGs (Rodić and Wilson 2017).

After an average economic life span, oil palm trees require replacement essentially with higher-yielding young palms. Although agrowastes are commonly used for animal feed (Kumar et al. 2017), open burning of OPTs is a common phenomenon for rapid discarding at replantation time. Formation of a heavy haze is a well-known impact of open burning ultimately, which is responsible for significant impact on atmospheric chemistry, global warming through increasing carbon emission (Wei et al. 2019), and human health, while remaining oil palm biowastes are discarded to mulch and replenish the land without any further usage (Ahmed, Yusoff, and Mokhtar Kamal 2020). Recently about only 22% of OPT biomass is used in veneer manufacturing for plywood, medium-density fiberboards (MDF), low-quality kiln-dried lumber (Abdullah et al. 2012), and shockingly the remaining volume of biomass is causing severe environmental degradation.

Numerous studies affirmed the existence of a significant quantity of starch in OPT biomass. O starch (OPTS) is a carbohydrate polymer with general formula (C₆H₁₀O₅)_n, stored inside the highly lignified parenchymatic cells and vascular bundle (Murai and Kondo 2010). These are rich in holocellulose (around 80%) and lignin content (around 15%) (Yusoff et al. 2019). Amylose (AM), amylopectin (AP) (holocellulose) are the two major elements of starch, which stipulate the basic nutrition for humans as well as animals. Efficient extraction of natural starch from oil palm biomass will ensure a sustainable environmental cleanup through reducing of huge postharvest processing cost (approximately RM 1000 per MT) (Syamsul Bahri 2016) for the discarded OPT biowaste.

OPTS can effectively be used as a food item, as an adhesive to prepare particle boards, or as a natural coagulant or adsorbent for leachate and wastewater treatment through the coagulation process (Yusoff et al. 2019). Precisely, OPTS can adsorb toxic cationic heavy metal ions (atomic density around 4.5 gm/cm⁻³ or above) (Liu et al. 2021), which are abundant in leachate and industrial wastewater. Even it is also possible to apply broadly in the chemical, pharmaceuticals or cosmetics sector, paper and corrugation industries (Wu et al. 2020) because of its extraordinary composition eventually, which will convert this waste to beneficial bioresources. Extracted starch usually consists of 20–36% linear and slightly branched AM with 0.41%-0.9% protein (H’ng et al. 2011). Figure 1 displays the typical bonding arrangement of OPTS.

Figure 1. Bonding arrangement of OPTS (Nadhari et al. 2013)

The remarkable reported starch yield from OPT was 7.15% of OPT dry weight by using Modified Dos Method (Mohd et al. 1999), beating C5/C6 sugars steeping method with lactic acid (1.7% yield) and conventional C5/C6 sugars steeping method (H’ng et al. 2011) although OPT contains more than 18% of extractable starch (Nadhari et al. 2013). Even though sodium meta-bisulfite is very commonly used for starch synthesis, to date no comprehensive explanation is available on implementing the bisulfite steeping method for the extraction of starch from oil palm biomass or its optimization applying modern analytical method. Thus, about one million tons of OPTS could be achieved annually from the OPT waste through synthesis process optimization. Currently, OPTS yield is about 6–7 MT per hectare, sago starch 4–5 MT (Zhu 2019), and cassava about 2 tonnes.

Therefore, the current research emphasized investigating the convenience of the bisulfite steeping method focusing on the key experimental factors for optimizing the synthesis of natural starch from the discarded OPT biomass. The optimization of the starch extraction method was performed by response surface methodology (RSM), an advanced technique of mathematical analysis. The CCD of Design-Expert software (version 7.0) conducted the experimental design and generated the mathematical models for optimizing the process parameters. Analysis of variance (ANOVA) from RSM offered the statistical outcomes with diagnostic checking, which made it possible for the researchers to justify the capability of the models. The starch yield was evaluated through this study as a sole response in terms of four experimental parameters, such as the strength of bisulfite solution, steeping duration, mixing ratio with bisulfite solution, as well as ultrapure water (UPW). Therefore, the implementation of the bisulfite steeping method to extract natural OPTS and its optimization through RSM are the novel approaches of the current study since these are as yet undocumented.

Experimental methodology

Collection of OPT (OPT)

Discarded trunks were obtained during replantation activities (between September 2019 and July 2020) for starch extraction after proper slicing, shredding and debarking. The Engineering Campus of University Sains Malaysia (USM), situated at Nibong Tebal, Pulau Pinang State, Malaysia, was the OPT collection spot. The collected OPT was transported immediately to the Environmental Engineering Laboratory, School of Civil Engineering, to preserve the samples initially at room temperature. After debarking, the freshly chopped trunks were sliced again into stiff blocks of around 8 cm to 12 cm in size (Mohd et al. 1999).

Reagents and chemicals

Analytical grade chemicals and reagents were applied throughout the study complying with the standard method (APHA 2012). This study applied collected reagents and chemicals throughout the experimental phase of extracting the natural starch from the palm oil trunk. AR Bendosen, PA, HmbG, and Rashaki Venture Sdn Bhd supplied sodium bisulfite (NaHSO₃) of 99.8% purity, acetone, and 16 mm polypropylene tubes, respectively as per experimental requirement. Veolia Water Solutions and Technologies Ltd, UK, produced distilled water (DW), while PURELAB Option-Q provided UPW for this study. Phong Heng Sdn Bhd was the provider of the filtration apparatus (1000 mL flask, funnel, filter holder, clamp), while the Bio flow company supplied glass fiber filter paper (1.6 μm, 47 mmϕ) (GF/A).

Extraction of natural starch from OPT (OPT)

After washing properly, 500 gm of freshly shredded and uniformly chipped OPT meal were steeped in sodium bisulfite (NaHSO₃) solution at room temperature. The strength of bisulfite solution varied from 0.2% to 1% (w/v) (Madruga et al. 2014), while the variable steeping duration was 2 h to 10 h (Maniglia and Tapia-Blácido 2016) at a different mixing ratio (w/v) of 1:1, 1:1.5 and 1:2 (with the initial weight of OPT). The accumulation of sodium bisulfite solution crumbles the protein–starch matrices and restrains the growth of microorganisms (Sulaiman et al. 2013). Sodium bisulfite produces sulfur dioxide in the presence of water, which eventually inhibits microbial growth and anionic bisulfite ion (HSO₃⁻) influence on breaking the starch–protein bond (Öztürk and Mutlu 2018). Once the steeping duration was over, the OPT meal was macerated with bisulfite solution homogenously in several batches (5 min each batch) using an industrial grinder, and the slurry was placed into a nylon screen for separation. After squeezing the slurry fairly, the filtrates were placed in a plastic dish. For extracting the residual starch, the remainder was passed through a similar procedure. The ultimate filtrates were screened through a 212 μm sieve and allowed for two-hour settlement. After the settlement period, the supernatant was removed by leaning the dish, and two liters of aquatic bisulfite solution was mixed thoroughly with the precipitate and allowed to settle again for two hours. Discarding the floated matter as before, the residue was mixed with UPW at a variable mixing ratio (w/v) (from 1:0.3 to 1:0.7 based on the initial weight of OPT), for washing out the soluble impurities from the starch.

An optimum mixing with UPW eventually washes the starch particles and enhances the centrifugation performance through the better accumulation of the starch particles. Later, this mixture was centrifuged using Hettich Zentrifugen EBA 270 at 3500 rpm up to 10 min (Yusoff et al. 2019). After centrifugation, the remaining starch precipitates were filtered through vacuum filtration attired with a 1.6 μm opening sized fiberglass filter for rinsing and refining the yielded starch. One hundred milliliters of acetone was applied twice for further washing and removing the impurities (lipid, lignin) from the refined starch sediment. Finally, the washed starch was sun-dried for three days (@8 hr/day) instead of oven drying for ensuring better quality and crushed lightly to get powder-like starch (Figure 2). After passing through a 70-mesh sieve, the weight of ground starch was measured to calculate the ultimate yield (%) with respect to the initial weight of OPT biomass and stored in airtight containers until use.

Figure 2. A brief display of the extraction of OPTS

Design of experiment, modeling and data analysis

CCD coupled with the RSM from Design-Expert carried out the experimental design, data analysis and optimization of the four influential experimental factors, i.e. the strength of bisulfite solution, steeping duration, mixing ratio of OPT with bisulfite solution, and mixing ratio with UPW coded as A, B, C, D, and one response, i.e. starch yield (% initial weight of OPT). Meanwhile, RSM is an advanced procedure for experimental designing, building empirical mathematical models, assessing the impact of numerous factors to consider the interacting influences. This method is remarkably effective in achieving significant optimum conditions for essential responses within reduced experimental run even in the presence of complex interaction. Preliminary batch studies were performed continuously until the process response (starch yield) had gained a satisfactory outcome, to achieve a leaner range of operating parameters in advance of applying central composite design for the experimental runs (Sharifi, Zabihzadeh, and Ghorbani 2018). Following the batch studies, the preferred ranges of experimental parameters were; strength of bisulfite solution 0.2% to 1%, steeping duration 4 hr to 8 hr, mixing ratio with bisulfite solution 1:1 to 1:2 and mixing ratio with UPW 1:0.4 to 1:0.6 accordingly for optimizing the process response.

Table 2 displays the CCD for the operating parameters in terms of their primary unit and 2k factorial design considering nF factorial runs. CCD offers the possible combinations of the experimental parameters with maximum and minimum values. The values of experimental parameters altered within three levels, between −1, 0, and +1 (Table 1). This experimental study conducted 30 runs in total following the equation CCD = 2^k+2k+6, where the number of parameters is k. For attaining the optimum values of strength of bisulfite solution (A), steeping duration (B), mixing ratio with bisulfite solution (C) and mixing ratio with UPW (D), this study analyzed the starch yield (%) (Y) as a dependent parameter with 24 experiments and six repetitions (Karimifard and Alavi Moghaddam 2018).

Table 1. Actual values with coded values in parenthesis for the experimental parameters

Experimental Parameters	Actual Value (Coded value)
A. Strength of bisulfite solution	0.2% (−1)	0.6% (0)	1% (+1)
B. Steeping time (hr)	4 (−1)	6 (0)	8 (+1)
C. Mixing ratio with bisulfite solution	1:1 (−1)	1:1.5 (0)	1:2 (+1)
D. Mixing ratio with UPW	1:0.4 (−1)	1:0.5 (0)	1:0.6 (+1)

Table 2. Central composite design outcome for experimental parameters and response (actual and predicted)

	Experimental Design for Process Parameters				Response (%Yield)
Run	A	B	C	D	Actual	Predicted
1	1	6	1	0.6	11	11.34
2	1	4	1.5	0.5	12	11.13
3	0.6	4	1.5	0.4	12.65	12.92
4	1	2	2	0.6	9.5	9.32
5	0.6	6	1.5	0.5	12.8	13.02
6	0.6	4	1.5	0.5	13	12.92
7	1	6	1	0.4	10.5	10.68
8	0.2	6	2	0.6	7.35	6.69
9	0.6	4	1.5	0.5	13	12.92
10	0.6	4	2	0.5	12.8	12.56
11	0.2	6	2	0.4	7.2	7.35
12	0.6	4	1.5	0.5	13	12.92
13	0.6	4	1	0.5	11.8	12.08
14	0.2	2	2	0.4	7.45	7.35
15	0.6	4	1.5	0.5	13	12.92
16	0.2	2	2	0.6	6.15	6.69
17	1	2	2	0.4	8	8.67
18	1	6	2	0.4	11	11.15
19	0.6	4	1.5	0.5	13	12.92
20	0.6	4	1.5	0.5	13	12.92
21	0.2	6	1	0.6	6.65	6.22
22	1	6	2	0.6	12.15	11.81
23	1	2	1	0.6	8.5	8.85
24	0.6	4	1.5	0.6	12.8	12.92
25	1	2	1	0.4	8.5	8.19
26	0.2	2	1	0.4	7.5	6.88
27	0.2	6	1	0.4	6.5	6.88
28	0.2	4	1.5	0.5	7	7.91
29	0.2	2	1	0.6	6.4	6.22
30	0.6	2	1.5	0.5	11.95	11.77

For predicting the optimized experimental conditions, eq 1(1) $Y={\beta }_0+\sum _{i=1}^k{\beta }_i{X}_i+\sum _{i=1}^k{\beta }_{ii}\cdot X_i^2+\sum _{i=1}^k\sum _{i<j}^k{\beta }_{ij}\cdot X_i\cdot X_j+\dots \dots \dots \dots \dots \dots \dots .+\epsilon$(1) exhibits the mathematical quadratic equation model (Sabour and Amiri 2017):

(1) $Y={\beta }_0+\sum _{i=1}^k{\beta }_i{X}_i+\sum _{i=1}^k{\beta }_{ii}\cdot X_i^2+\sum _{i=1}^k\sum _{i<j}^k{\beta }_{ij}\cdot X_i\cdot X_j+\dots \dots \dots \dots \dots \dots \dots .+\epsilon$(1)

In this equation, the desired response is Y, j is the quadratic coefficient, i is the linear constant, Xj and Xi are the parameters, k indicates the number of operating parameters investigated as well as optimized through this study. β₀ represents the regression coefficient, and β_i, β_ii, β_ij are the interacting coefficients for linear, quadratic as well as second-order terms, while ε indicates the random error.

Analysis of variance (ANOVA) from the Design-Expert software performed the data analysis in this study as used similarly for other starch isolation process (Wang et al. 2020) to attain the interrelationship amid the process parameters and the desired response (Aziz et al. 2011). ANOVA provides a comprehensive statistical evaluation pointing to response variance (Bakraouy et al. 2017). This analysis illustrates simultaneously the significance as well as the influence of the respective factors over the desired response. F-value (Fisher’s variation ratio) from ANOVA ensures the direct influence of factor variance over response. The values of coefficient of determination (R²) along with adjusted R² identify the fitting excellence and quality of the polynomial model. An R² value closer to one is desirable for obtaining satisfactory agreement between predicted and experimental values. Adequate precision (AP) of the ANOVA evaluates the statistical significance of linear and quadratic terms. Predicted R² illustrates the prediction capability of the suggested model through PRESS (predicted residual error sum of squares). P-value (probability) associated with a 95% confidence level evaluated the model parameters, and their interactions (<0.05) indicates the factor significance). Finally, Design-Expert Software provided three-dimensional (3D) response surfaces and their corresponding contour plots based on the outcomes.

Characterization of OPT starch (OPTS)

pH and moisture content of OPTS

About 1.5 gm of starch was mixed up with 25 mL of purified water in an extraction bottle (50 mL) for determining the pH following the previous study (Abd Karim et al. 2020). After proper mixing with a portable shaker for 3 min, the mixture became stable within 15 min before pH measurement.

For achieving the moisture content (%) of OPTS, 2 gm of the sample was positioned within a natural convection oven with 105°C temperature for 24 h. After optimum drying in the oven, the starch sample was set in a desiccator for about 15 min to cool down before measuring the moisture content from weight loss. From the initial weight (W₁) of starch before drying and oven-dried weight (W₂), moisture content was calculated.

Particle size and polydispersity analysis

After properly dispersing the OPTS sample, Malvern Zetasizer version 6.01 equipped with a dry powder feeding device, analyzed the particle size distribution, as well as polydispersity index (PDI), which indicate the broadness of molecular weight distribution. Particle size measurement was expressed in terms of the micrometer (μm), while PDI is a numerical value only.

Analysis of surface morphology

The surface morphological structure and elemental analysis with composite homogeneity of OPTS were investigated applying digital electron microscopic analysis (SEM) and energy-dispersive X-ray (EDX) spectroscopy. An ophthalmic microscope FEI-Quanta 450 FEG was implemented to conduct SEM and EDX analyses following the standard conditions within an expediting voltage of 20 kV. A portion of starch dispersion was put on the aluminum heel, coated with a thin layer (30 angstroms) of gold (sputtering) through the Polaron SC515 SEM coating system and then fixed on a sample table utilizing conductive carbon glue.

X-ray diffraction analysis (XRD)

Siemens (Germany) provided KRISTALLOFLEX D-5000 X-ray diffraction system performed the X-ray diffraction analysis for dried starch powder comprising around 10.8% moisture to detect the proportion of crystal-like arrangement in starch. After dehydration in a desiccator and removing the dust or any contaminants, the starch samples were laid on an aluminum specimen stub as densely as possible. Then the X-ray diffraction shape was documented through the operation with a monochromatic filter, Cu Kα radiation (λ = 154.0) with 40 kV of opening voltage as well as 30 mA of current. The scanning procedure was undertaken at a 2θ diffraction angle within a range of 10° to 70° with a varying scanning speed (0.02°/min to 2°/min).

FTIR spectroscopy

Fourier-transforms infra-red (FTIR) analysis was conducted to determine the bonding structure and the existence of the organic functional groups in the extracted OPTS. Spectrum IR Tracer-100 Series FTIR (Shimadzu, Tokyo, Japan) outfitted with a diamond ATR mechanism performed the FTIR analysis of dried powder like starch sample (Wu et al. 2020). The acquired spectrum from the sample analysis shows the ultimate outcome. One milligramof starch sample (previously prepared through enough grinding) was mixed up with KBr (100 mg, pure spectrum), and pressed into a pellet suitable for FTIR analysis. FTIR spectroscopic analysis was within the range of 4000 cm⁻¹ to 400 cm⁻¹. Transmittance peaks and variance in the acquired data were recognized following the published articles for FTIR spectral peaks (Bolyard et al. 2019).

Solubility index (SI) and swelling power (SP)

The solubility index (%) along with swelling power (SP) of the starch samples was undertaken following the procedure demonstrated by previous researchers (Lai et al. 2016). About 10 mL of deionized water was supplemented in the centrifuging bottle after measuring its weight and placing 100 mg (W_sa) of the starch sample in it. For more authentication, samples were prepared in triplicate. After 2 min of shaking with a handheld shaker, the centrifuging tubes were heated in the water bath (from 50°C to 90°C). Heating (inside the water bath) was performed until 35 min for each varying temperature with a whirl mixing at every 5 min intervals. Tubes were centrifuged at 4500 rpm until 35 min after heating, and the supernatant was oven dried at 105°C for up to 24 h. Once the weight of starch sediments (sticking portion to the tube wall) (W_se), and dried supernatant (W_su), were measured, then water solubility (SI) along with swelling power (SP) were determined according to the following eq2(2) $\mathrm{S}\mathrm{I}=\frac{{\mathrm{W}}_{\mathrm{s}\mathrm{u}}}{W_{sa}}\times 100$(2) and 3(3) $\mathrm{S}\mathrm{P}=\frac{{\mathrm{W}}_{\mathrm{s}\mathrm{e}}}{W_{sa}\left(1-WS\right)}$(3) .

(2) $\mathrm{S}\mathrm{I}=\frac{{\mathrm{W}}_{\mathrm{s}\mathrm{u}}}{W_{sa}}\times 100$(2)

(3) $\mathrm{S}\mathrm{P}=\frac{{\mathrm{W}}_{\mathrm{s}\mathrm{e}}}{W_{sa}\left(1-WS\right)}$(3)

Results and discussion

Analysis of experimental data from RSM

RSM analyzed the impact of experimental parameters over the process response starch yield (Y) for applying the bisulfite steeping method.

Table 2 displays the CCD to develop the mathematical model equation applying ANOVA, with corresponding results (Y) evaluated in terms of experimental parameters (A, B, C, D). To relate the experimental variations with predictable random error is the basic concept of ANOVA (Sabour and Amiri 2017). The properties of the unrestrained parameters were minimized by randomizing the experimental sequences. Similarly, Table 2 displays the actual with predicted responses of starch yield concerning the operating parameters. ANOVA analyzed the obtained results for assessing the “goodness of fit” of the model. The first equation from ANOVA analysis was modified, while eq 4(4) $\mathrm{\%}\mathbf{Y}\mathbf{i}\mathbf{e}\mathbf{l}\mathbf{d}\left(\mathrm{Y}\right)=12.92+1.61\mathrm{A}+0.62\mathrm{B}+0.24\mathrm{C}-3.40{\mathrm{A}}^2-0.53{\mathrm{B}}^2-0.60{\mathrm{C}}^2+0.62\mathrm{A}\mathrm{B}+0.33\mathrm{A}\mathrm{D}$(4) depicts the ultimate quadratic models in terms of coded factors after excluding the insignificant terms. The negative sign in the equation shows an adverse effect, while a positive sign ahead of the terms specifies synergistic consequence (Azmi et al. 2015). The equation calculated the parameters as the combination of second-order (A², B², C²), first-order (in terms of A, B, C), interlinked effects (AB, AD) with a constant, according to eq 1(1) $Y={\beta }_0+\sum _{i=1}^k{\beta }_i{X}_i+\sum _{i=1}^k{\beta }_{ii}\cdot X_i^2+\sum _{i=1}^k\sum _{i<j}^k{\beta }_{ij}\cdot X_i\cdot X_j+\dots \dots \dots \dots \dots \dots \dots .+\epsilon$(1) .

The second-order effects in this equation show a negative impact on the starch yield.

Ultimate Model Equation Concerning Coded Factors:

(4) $\mathrm{\%}\mathbf{Y}\mathbf{i}\mathbf{e}\mathbf{l}\mathbf{d}\left(\mathrm{Y}\right)=12.92+1.61\mathrm{A}+0.62\mathrm{B}+0.24\mathrm{C}-3.40{\mathrm{A}}^2-0.53{\mathrm{B}}^2-0.60{\mathrm{C}}^2+0.62\mathrm{A}\mathrm{B}+0.33\mathrm{A}\mathrm{D}$(4)

Table 3 shows the consolidated results of analytical parameters in accordance with ANOVA. Analytical data displaying in the mentioned table indicate that at the 95% confidence level the models are significant since p-values are less than 0.0001.

Table 3. ANOVA outcome for modified quadratic model of response surface

Source	Sum of Squares	DF	Mean Square	F-value	p-value
Regression model	193.4	8	24.17	114.91	<0.0001
Residual	4.42	21	0.21
Lack of Fit	4.42	16	0.28

Standard deviation = 0.46; R² = 0.98; adjusted R² = 0.967; predicted R² = 0.94; adequate precision = 27.053; coefficient of variation = 4.49; PRESS = 11.73.

The coefficient of determination along with adjusted R² assessed the model’s capability in the fitting. Meanwhile, predicted R² and adequate precision specify the prediction capability of the model. The lack of fit (LOF) term is nonsignificant while, F-test defines the data discrepancy around the fitted model and indicates the significance of a regression. Table 3 represents the value of LOF (4.42) and PRESS (11.73) which specifies the sum of squares of prediction error.

The coefficient of determination (R²) provides the total predicted response variation by the model, representing the proportion of the regression sum of squares (SSR) to the overall sum of squares (SST). The greater values of R² (0.95–0.98) and adjusted R² (0.91–0.97) are favorable and the indicator of the satisfactory adaptation between experimental results with the obtained quadratic model. Moreover, a sensible correspondence between the R² and adjusted R² (closeness in values) is essential since it designates the lower impact of R² enhancement because of the insertion of insignificant variables (Sharifi, Zabihzadeh, and Ghorbani 2018). The values of the coefficient of determination (0.98) and adjusted R² (0.967) in this study (Table 3) strongly reveal the significance of the model according to the statistical rules mentioned above.

Predicted R² and adequate precision (AP) illustrate the prediction capability of the models, while AP relates the sequence of the projected values to its average standard error. AP is indeed a sort of signal-noise fraction, and its value should be four or above to ensure an appropriate calculation for a model. The value for AP remaining between 15.0 and 30 confirms the stable prediction aptitude of the models. Conceptually, predicted R² could be evaluated after modification from the remaining values of a regression model, and it reflects the success of prediction from excluded values and residual sum of squares (Myers, Raymond, and Cook 2016). The predicted R² value, which ranges from 0.8 to 0.92, reveals a strong model prediction. The values of predicted R² (0.94) and AP (27.053) of the achieved model (Table 3) from this study are firmly in agreement that it can traverse the design space illustrated by central composite design. The coefficient of variation (CV) is the most potent method for determining the validity of a sample, which specifies the proportion of the expected standard error to the mean value of the practical response. A model’s CV value of no more than 10% (5.86% to 10.66% more precisely) is more consistent because the lower CV values, the closer the predicted values are (Ghani et al. 2017). The obtained model is firmly designated as reproducible concerning the value of CV (4.49), following Table 3.

ANOVA diagnostic plots showing the correlation between actual experimental values and predicted values guide us to justify the model competence. Figure 3a,b displays the scattering amid the data points for predicted against actual values of the starch yield obtained by the model, and the diagonal line presented satisfactory agreement.

Figure 3. DesignExpert originated plot (a) actual versus predicted; (b) normal probability plot of the residuals

Process analysis

Figure 4a–d display the 3D response surface plots for starch yield based on ANOVA and numerical optimization, correspondingly. The displayed plots in figures are almost well-proportioned in outline with round and closer contours. The response surface figures exhibit very sharp peaks signifying the optimal operating conditions for the highest response (starch yield) value regarding the experimental parameters in the design space. RSM generated 3D response surface plots (Figure 4a,b) display the consequences of interaction amid the strength of bisulfite solution and steeping duration. It is strongly evident that the starch yield increases with the enhancement of these parameters up to a certain level. Activation of the bisulfite solution at room temperature accelerates the chemical bond breakdown process within the substantial steeping duration inside biomass. The breakdown process is also recognized as the reaction of elimination (Azmi et al. 2015). Response surface plots in Figure 4 designate the optimal points are 0.74% bisulfite solution, 5.60 hr of steeping duration, mixing ratio 1:1.60 and 1:0.6 for bisulfite solution as well as UPW, respectively. Moreover, 3D response surface plots (Figure 4c,d) also reveal that the response depicts a reduction for increasing or decreasing of the tested parameters. Maintaining a constant value for the bisulfite solution strength and steeping duration according to the optimization solution, change in the values of both mixing ratios showed a decline in starch yield.

Figure 4. 3D response surface plot, (a) model graph from ANOVA; (b), (c), (d) numerical optimization based on interrelation of parameters

Research findings are consistent with the previous results (Sulaiman et al. 2012; Zhou et al. 2015). In Figure 4b the response surface exhibits 13.54% starch yield at the optimized condition for applying this bisulfite steeping method, which is remarkably higher than in previous studies.

Process optimization

Design master programming of Design-Expert software assessed the standard situations for ensuring maximum yield of starch where all parameters coincidentally address the necessary criteria concerning higher and lower limits, associating the quality capacity (Gupta et al. 2017). The preferred response limit was from 7% to 14%, which is comparatively adjacent to the attained value. The goal was to maximize the yield in terms of the lowest allowable values of the experimental parameters to obtain a reasonably specific optimum zone. The achieved outcomes are consistent with the operating parameters, which lead to the ideal conditions for starch yield measurement. Figure 5 displays the optimum conditions, (i) strength of bisulfite solution is 0.74%, (ii) steeping duration is 5.6 hr, (iii) mixing ratio with bisulfite solution is 1:1.6, (iv) mixing ratio with UPW is 1:0.6 for the highest starch yield of 13.54%, with desirability function 0.93.

Figure 5. Numerical optimization for four parameters and one response with desirability function

The optimization results were reviewed at room temperature under the corresponding setting up. To verify the proposed optimum condition from the predicted model three experiments were performed further following the same values. Based on the verification study outcomes (Table 4), the average starch yield extracted 13.87%, which is quite a satisfactory agreement from the regression model, with a comparatively lower error of 2.43%.

Table 4. Outcomes of verification study at optimum condition

Strength of Bisulfite Solution	Steeping Time (h)	Ratio with Bisulfite Solution	Mixing Ratio with UPW	Model Response	Average Verification Value	%Error	Standard Deviation
0.74	5.6	1.6	0.6	13.54	13.87	2.43	0.165

However, the outcome of the experiments confirmed the model soundness satisfactorily, ensuring the existence of the optimal point, and at the same time, recognized the significance of this method for increasing the starch yield successfully.

Outcomes of physical characteristics of starch

Starch yield from OPT

According to the central composite design, batch experimental outcomes showed that this bisulfite steeping method could ensure a maximum starch yield of 13% at 0.6% bisulfite solution, 6 hr steeping duration with the mixing ratio of 1:1.5 and 1:0.5 for bisulfite solution as well as UPW, respectively. Although extracting the starch from the cells of the abrasive vascular bundles of OPT is quite hard, rather this combination of proposed method augmented the yield remarkably higher than that demonstrated by previous researchers regarding starch extraction from OPT (Abd Karim et al. 2020), sago trunk (Aziz and Sobri 2015) and cassava peel or potatoes (Waterschoot et al. 2015). Furthermore, it was also observed that the changes in the experimental parameters reduce the starch yield. Besides this, extracted OPTS showed off-white wheat brown color because of the enzymatic phenolic compounds that produce polyphenolic pigments through some chemical reactions (Lattanzio, Cardinali, and Linsalata 2012).

pH and moisture content

The nature of extracted starch was found acidic since the measured pH value was 4.86, which is also almost similar to the earlier research outcomes (Abd Karim et al. 2020).

The extracted starch moisture content value was 10.74%, which is firmly consistent with the previous outcomes (Abd Karim et al. 2020) for starch extraction from OPT. According to the scientific report, typically, starch moisture content remains around 9% to 15% (Zhu and Guo 2017). For more authentication, this study tested the samples in triplicate.

Particle size and polydispersity index

Malvern Zetasizer test confirmed the Z-average hydrodynamic particle size distribution of extracted starch in this study is about 7.152 µm while the normal range of granular starch size is 3 µm to 25 µm. OPTS particles showed a bimodal distribution of granular size as explained by the previous researchers, which is also similar to the other cereal starches like wheat, and barley (Gilbert et al. 2010), but the granule size of sago starch (15–50 µm) is higher than OPTS (Aziz and Sobri 2015). Starch granular deformation is responsible for the reduction of its molecular weight significantly. The average molecular weight of OPTS is approximately 865 kDa, which recognizes the OPTS as a high molecular weight (greater than 100 kDa) polymer (Zamri, Mohd Akhiar, and Halim Shamsuddin 2019).

The PDI of polymer is another significant term that indicates the broadness of its molecular weight distribution. According to the Zetasizer test, the PDI verified the OPTS suspensions as monodispersing and uniform since the value is comparatively higher (0.869). However, IUPAC ideally considered OPTS as a uniform polymer rather than a monodisperse polymer.

SEM-EDX investigations outcomes

OPTS granule morphology was scrutinized using scanning electron microscopy (SEM), and Figure 6a,b shows the achieved images of OPTS at 4000 and 6000 magnifications from SEM analysis, while Figure 6c represents the EDX analysis result of OPTS. SEM images display the greater granular sizes (10–100 µm ϕ). This phenomenon was coherent with other OPTS granular architecture characteristics, featuring a much more mature or possibly fully mature storage starch. The micrographs show that the granular architecture of OPTS is almost like sago palm starch, representing ovular and elliptical patterns with condensed ends. Bell-shaped granules were also observed.

Figure 6. (a) SEM micrograph of native starch at 4000 magnifications; (b) SEM micrograph of native starch at magnification of 6000s; (c) EDX analysis report for OPTS

This outcome was almost similar to the explanations by Nadhari et al. (2013). Furthermore, a minor fraction of OPTS particles from the apparent particle fractions represented an hourglass structure with an almost areolate shape and several trumpet-like swellings. In contrast, OPTS granule surface seemed a little less smooth than previous studies and marked by more wave-like creases. Simultaneously, SEM outcomes unveiled the presence of radial pattern surface openings along with some grooves or hollows on large-sized OPTS granules, which was not available in early research.

The conspicuous openings of 0.5–2.0 µm diameters could be openings to channels that stabbed over various rings of starch growing as well as the hilum, while a portion remained on the outer layer only. Figure 6c displayed the presence of several minerals, such as Na, K, Mg, with a major portion of carbon (C) and oxygen (O) in OPTS according to the Energy Dispersive X-Ray Analysis (EDX). An ample existence of oxygen has followed in OPTS due to the extravagant carbohydrate combination (Aziz and Sobri 2015). Furthermore, carbohydrates are broken down in the form of energy by using oxygen.

XRD analysis of OPT starch

X-ray diffraction analytical outcome (Figure 7a) of OPTS samples shows the peaks of higher intensity at 2θ values (Bragg angles) of approximately 22.5° according to its crystalline arrangement. Secondary peaks are observed at 2θ = 14.75° and 2θ = 21.7° exhibiting diffraction pattern nearer to A-type crystallinity pattern, which eventually declares its similarity with other representative A-type starches such as cereals. Rather than relating to Noor et al. (1999), OPTS displayed significant structural disruption of typical A-type crystallinity in the current study, probably because of the granular disintegration during starch crushing.

Figure 7. (a) XRD analytical shape of OPTS; (b) FTIR spectrum for extracted starch (OPTS)

The peaks, for the intensity of the amorphous portion, were observed at 2θ = 19.25°. The crystallinity index of OPTS was 48.5%, which remains in the range of the relative crystallinity standard of 15–48% for common native starch, while the lower crystallinity of OPTS indicates the higher amylose existence (Qin et al. 2016).

FTIR spectrum analysis outcomes

Figure 7b displays the FTIR spectra of OPTS. The band shape of OPTS is pretty consistent with the exclusive spectral pattern of starch. The outcome exhibits significant peaks in the starch impression zone (970–1200 cm⁻¹) at 1020 cm⁻¹, 1083 cm⁻¹, and 1153 cm⁻¹, which overlays the C-O stretching in the C-O-H side group (Ghosh Dastidar and Netravali 2012). The spectrum indicates the presence of effective groups like hydroxyl, carbonyl, carboxyl, methoxy, and amino (amine and amide) groups in OPTS. Peaks at 2858 cm⁻¹, 2926 cm⁻¹, and 3358 cm⁻¹ represented the hydroxyl (O-H) active group, commonly available in carboxylic acids, phenols, or alcohols (Moharrami and Motamedi 2020). A band at 1635 cm⁻¹ represents the structural vibration of aromatic double bonds C=C (Ferraz et al. 2016). Peak appearances at 1411 cm⁻¹ and 1452 cm⁻¹ attribute to the aromatic groups (C-C stretch) (Moharrami and Motamedi 2020) and alkanes group (C-H-C bend), while another peak 1242 cm⁻¹ represents robust C-N stretching. Functional group amines in OPTS displayed traceable peaks to N-H wagging (709 cm⁻¹, 767 cm⁻¹, and 860 cm⁻¹), at the same time peaks, at 530 cm⁻¹, 578 cm⁻¹, and 615 cm⁻¹ reveal the presence of alkyl halides (C-Br stretch).

Moreover, the peaks at 1242 cm⁻¹, 1153 cm⁻¹, 1083 cm⁻¹, and 1020 cm⁻¹ indicate the existence of esters and carboxylic acids (C-O stretching), which is comparable with polysaccharides. The peak at the wavelength 860.25 cm⁻¹ showed the highest intensity following the orders of the intensities at wavelength 449.41 cm⁻¹ > 767 cm⁻¹ > 933 cm⁻¹ > 709 cm⁻¹ > 530 cm⁻¹ > 578 cm⁻¹.>615.29 cm⁻¹ > 1242.16 cm⁻¹ > 1452 cm⁻¹ > 1411 cm⁻¹ > 1153 cm⁻¹. In brief, all these FTIR analysis outcomes are similar to cereal starches like rice, which eventually declare the success of starch extraction from discarded OPT through this study (Limpongsa and Jaipakdee 2020).

Swelling power (SP) and solubility index (SI)

Table 5 displays the water solubility index (SI) as well as swelling power (SP) of OPTS at five differing temperatures from 50°C to 90°C at 10°C intervals. Swelling power and solubility index is the evidence of internal action of the amorphous as well as crystalline areas (Abd Karim et al. 2020), but there is no open relationship among these properties. In addition, amylose, amylopectin characteristics, and bond strength among the molecules are also influential over these properties (Kusumayanti, Handayani, and Santosa 2015).

Table 5. Solubility index (SI) and swelling power (SP) of extracted starch

Temperature (°C)	Swelling Power (SP) (g/g)	Solubility Index (SI) (%)
50	2.52	17.17
60	2.78	19.05
70	6.9	7.46
80	6.53	5.23
90	5.19	2.89

The outcomes revealed a steady augmentation in swelling power until the temperature reached 75°C. OPT starch indicates the maximum swelling power as well as solubility at 72°C and 60°C, respectively. The solubility values vary from 2.89% to 19.05%, while the swelling power values remain in the range of 2.52 to 6.9 (g/g). The smaller value of the OPTS swelling power denotes the existence of better amylose content and a higher degree of intermolecular relationship in comparison with other starches.

Costing of starch synthesis

OPTS synthesis costing is also a considerable issue regarding the further application. Research study reveals that OPTS synthesis is much less costly here in Malaysia because of the continuous and abundant availability of OPT. Moreover, OPTS synthesis will reduce the high postharvest management cost of OPT and save the environment from the waste burden. Although the costing of OPTS synthesis is related to several factors, such as collection source of raw materials, fluctuation of chemical prices, processing inconsistency, overhead costs, Table 6 shows a typical costing for the synthesis process. Therefore, the pricing exhibited in Table 6 should be considered as a tentative one rather than accurate values.

Table 6. A typical costing outline for OPTS synthesis (based on 500 gm OPT processing)

Items	Required Quantity	Price (RM)
OPT collection (depends on the distance of the source)	500 gm	2
Sodium bisulfite	2.8 L	1.1
Ultrapure water	300 mL	0.9
Fiberglass filter 1.6 µm	1p	1.1
Acetone	100 mL	0.9
Total variable cost for 500 gm OPT or about 70 gm OPTS synthesis		6
Total variable cost for 1000 gm or 1 kg OPTS production		85
Nylon screen (1 m*3 m)	1 for 1000 gm starch production	12
Total tentative costing for 1 kg OPTS production		97

Conclusion

The research findings have confirmed the significance of the bisulfite steeping method for a better yield of OPTS. According to CCD and ANOVA, Design-Expert suggested the optimum conditions are, 0.74% strength of bisulfite solution, 5.60 hr steeping duration, the mixing ratios are 1:1.6 and 1:0.6 for bisulfite solution as well as UPW, for the highest yield (13.54%) of OPTS with a desirability function of 0.934. The yield is remarkably higher than the identical method described by Mohd et al., (7.15% yield) (Mohd et al. 1999). Furthermore, the correlation investigation identified the significance of the process mechanisms to optimize the starch yield. A high value of the coefficient of determination (R² = 0.98) and adjusted R² (0.97) along with a nearer value of predicted R² (0.93), confirmed the justifications of the developed model based on experimental design. Starch granules indicated the existence of high amylose content and selective functional groups with a standard crystallinity index (48.5%) as well as an A-type pattern. Subsequently, it can also be suggested that future study should focus on the enhancement of OPTS yield and its various properties (physicochemical, structural, pasting, and rheological) with diverse application in a large context for environmental sustainability.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This study was funded by the Ministry of Higher Education Malaysia under Fundamental Research Scheme (FRGS) (Grant No. 203/PAWAM/6071415) for research associated with the Solid Waste Management Cluster, Engineering Campus, Universiti Sains Malaysia.

Notes on contributors

Zaber Ahmed

Zaber Ahmed is a Ph.D. researcher of Environmental Engineering at the School of Civil Engineering, Universiti Sains Malaysia. He obtained his M.Sc. Engineering (Environmental) from the Department of Civil Engineering of Dhaka University of Engineering and Technology (DUET), Dhaka, Bangladesh, in 2014. He is working as Assistant Professor of the Department of Civil Engineering in Model Institute of Science & Technology, Gazipur, Bangladesh, under the Bangladesh Technical Education Board. He has published 10 articles in international journals and proceedings. He continued to serve as a peer reviewer for 6 international journals. To date, he has reviewed 25 international papers. His research focuses on environmental impact assessment, Industrial wastewater treatment, sustainable solid waste management, landfill technology, landfill leachate treatment. He is also involved in several consultancy works concerning his expert area.

Mohd Suffian Yusoff

Mohd Suffian Yusoff is a Professor of Environmental Engineering at the School of Civil Engineering, Universiti Sains Malaysia. He obtained his Ph.D. in Environmental Engineering from the Universiti Sains Malaysia in 2006. He has been graduated with five (5) Ph.D. and ten (10) MSc students from USM. Currently, he has published over 100 refereed articles in international journals, proceedings, and chapters in refereed international books. He continued to serve as a peer reviewer for more than 15 international journals. To date, he has reviewed 200 international papers. His research focuses on solid waste management, landfill technology, landfill leachate treatment as well as wastewater treatment. He has been involved in numerous consultancy works for international and local projects concerning his expert area.

Mokhtar Kamal N.H.

Mokhtar Kamal N.H. is a Senior Lecturer at the School of Civil Engineering, Universiti Sains Malaysia. She started her academic carrier after obtaining her Ph.D. in Environmental Engineering from Imperial College London in 2015. Her research topics cover drinking water disinfection by-products, as well as drinking water treatment technologies, and rainwater harvesting and treatment. Up to date, she has graduated a Ph.D. and an MSc student. She has published several refereed articles in international journals and proceedings. Apart from that, she is actively involved in several consultancies work on wastewater impact to the environment for local clients as well as for international companies.

Hamidi Abdul Aziz

Dr Hamidi Abdul Aziz is a Professor in Environmental Engineering in the School of Civil Engineering, Universiti Sains Malaysia. Professor Aziz received his Ph.D. in civil engineering (environmental) from the University of Strathclyde in 1992. He is currently the Head of the Solid Waste Management Cluster (SWAM), Universiti Sains Malaysia. He has 29 years of teaching and research experiences in the field of environmental engineering, mainly related to solid waste management and landfill technology, water and wastewater treatment, leachate treatment, bioremediation, pollution control, and environmental impact assessment. To date, he has graduated over 100 PhD and MSc students and has published over 200 ISI papers, a few books and has become an editor and sits in editorial board members of a few international journals. Malaysia Academy of Sciences awarded him a Top Research Scientist of Malaysia in 2012. In 2020 he was listed as among the top 2% Scientist in his field in a global list compiled by the prestigious Stanford University from the United States (US) for the year 2019.