Catalytic Evaluation of an Optimized Heterogeneous Composite Catalyst Derived from Fusion of Tri-Biogenic Residues

Abstract

This study analyzes the elemental and oxide compositions of three selected agricultural residues—Dried Pawpaw Leaves (DPL), Kola Nut Pod (KNP), and Sweet Orange Peel (SOP)—for their potential as heterogeneous catalysts. Energy Dispersive X-ray (EDX) analysis identified calcium (25%) and potassium (29%) as the primary elements in DPL and KNP, with calcium oxide (CaO) and potassium oxide (K₂O) as the dominant oxides. SOP had a similar composition but lacked vanadium. Calcined residues were analyzed at temperatures ranging from 500 °C to 900 °C using X-ray Fluorescence (XRF), revealing stable silicon dioxide (SiO₂) content and temperature-dependent variations in CaO and K₂O, indicating their catalytic potential for transesterification processes. Scanning Electron Microscopy (SEM) showed non-uniform, spongy microstructures, enhancing the surface area and catalytic efficiency. Fourier Transform Infrared Spectroscopy (FTIR) identified functional groups essential for catalytic activity, such as hydroxyls, methyl, and carboxyl. X-ray Diffraction (XRD) confirmed the presence of crystalline phases like calcium carbonate and calcium oxide, crucial for catalytic performance. Experimental biodiesel production using a mixture of the calcined residues (33.33% each of KNPA, SOPA, and DPLA) resulted in the highest biodiesel yield at 65.3%. Model summary statistics, including R² (0.9824) values and standard deviations (0.0026), validated the experimental design, indicating high precision and prediction accuracy. These results suggest that the selected agricultural residues, when calcined and mixed properly, can serve as effective heterogeneous catalysts, with significant implications for biodiesel production, supporting previous research on the importance of calcium in catalytic processes.

1. Introduction

Energy is a key driver of economic development in any nation, impacting various sectors such as agriculture, industries, and transportation. The global energy demand has increased significantly, growing by 2.9% from 2017 to 2018, twice the 1.5% recorded in the previous decade [1]. This heightened demand is closely linked to the world’s population growth [1,2,3,4].

Fossil fuels, the primary global energy source, have raised environmental concerns due to associated issues like carbon emissions and greenhouse gases (GHGs), leading to climate change and adverse effects on human health [4,5]. Climate change exacerbates challenges in agriculture, contributing to risks like desertification, rising temperatures, changing rainfall patterns, and sea-level rise [6,7]. To address these challenges, diverse energy sources are imperative. An example of alternative energy sources are biogas [8], solid biofuels obtained through densification, torrefaction, and by extension pyrolysis.

Bioenergy researchers advocate for biodiesel as a viable alternative to fossil fuels, citing its non-toxic, energy-efficient, biodegradable, and environmentally friendly nature [9,10,11]. However, the cost of biodiesel production, mainly attributed to expensive feedstocks like refined vegetable oils, remains a significant hurdle [9,11]. Transesterification, particularly with chemical catalysts like KOH, NaOH, H₂SO₄, and HCl, is the preferred method for biodiesel production [12]. Despite their efficiency, these chemical catalysts pose challenges in terms of reusability, negative environmental impact, and complex downstream purification steps [13].

Recent research has shifted focus to low-cost feedstocks such as Waste Frying Oil (WFO) and catalysts derived from agricultural residues, contributing to environmental pollution control [1,14]. Bio-based catalysts, being heterogeneous and sourced from natural materials, present advantages in terms of easy recovery and reusability across multiple catalytic cycles [14,15].

Utilizing waste biomass feedstocks for biofuel production significantly reduces GHG emissions by displacing conventional fossil fuels in the transportation sector. For instance, the conversion of agricultural residues, municipal solid waste, and used cooking oil into biofuels can result in up to 80% fewer emissions compared to gasoline and diesel, depending on the production process and feedstock used [14]. Moreover, biofuels from waste biomass typically do not compete with food crops, avoiding the ‘food vs. fuel’ conflict that can arise with first-generation biofuels sourced from edible crops like corn and soybeans [15].

From an economic standpoint, waste-derived biofuels offer cost savings by repurposing waste materials that might otherwise require disposal. This approach also supports circular economy principles by transforming waste streams into valuable energy sources, reducing landfilling and associated costs [16]. The use of locally available waste materials for biofuel production can also stimulate rural development and create jobs, especially in agricultural regions where residues are abundant [17].

Recent advancements in biofuel production technologies, such as the development of green heterogeneous catalysts, have improved the efficiency of converting waste biomass into biofuels. These catalysts, which can be derived from agricultural residues, offer a more sustainable option for biofuel production, lowering the environmental impact of the production process [18]. However, challenges remain, such as scaling up production, improving feedstock collection systems, and enhancing the energy density of biofuels to compete with fossil fuels.

This study investigates the potential of a heterogeneous composite catalyst derived from three selected agricultural residues—Dried Pawpaw Leaves (DPL), Kola Nut Pod (KNP), and Sweet Orange Peels (SOP). The aim is to explore alternative, environmentally friendly catalysts for biodiesel production, contributing to sustainable energy practices.

2. Materials and Methods

2.1. Materials Collection and Preparation

The selected agricultural residues are Dried Pawpaw Leaf (DPL), Kola Nut Pod (KNP), and Sweet Orange Peel (SOP). The dried fallen pawpaw leaves were gathered from their parent trees at the Research Farm of the Department of Agricultural and Bio-Environmental Engineering, Federal Polytechnic Ado Ekiti. Manual sorting was conducted to eliminate foreign matter, and the leaves were sun-dried for three days until a constant weight was achieved. Subsequently, they were ground to a granular consistency using a pestle and mortar. The ground sample was sieved to a uniform particle size of 150 µm using a standard sieve with a mechanical shaker. The sieved sample was stored in airtight containers for further analysis. The sample preparation method aligns with that reported by [19,20,21].

The KNP was freshly sourced from a farm in Ile Ife (7.49° N, 4.55° E), Osun State, Nigeria, where it is widely available. To prepare it, manual sorting removed dirt, immature pods, and fresh husks. The pods were then cut, rinsed with clean tap water, and sun-dried. Further drying in an oven at 105 °C continued until the sample reached a stable weight, ensuring all moisture was eliminated. The dried sample was milled into a fine powder with a disk mill and sieved to a uniform particle size of 150 µm using a mechanical shaker. This powdered sample was stored in airtight containers for subsequent analysis, following the preparation method outlined by [19,20,21]. Additionally, sweet orange peels, commonly discarded by vendors, were collected, sorted manually to remove dirt and foreign materials, then washed with clean water and sun-dried for three weeks. The dried peels were ground into granules using a pestle and mortar, sieved to a 0.15 mm particle size, and stored in a sealed container for later analysis.

WFO was sourced from local food restaurants. The collected WFO underwent centrifugation and filtration to remove any suspended matter and burned food bits. Subsequently, it was heated at 105 °C for 2 h to eliminate water through evaporation, following the suggestion of [22]. The oil was stored at room temperature in a closed container for characterization. The WFO characterization was carried out at the Petroleum Engineering Laboratory of Afe Babalola University, Ado Ekiti. The oil was characterized for acid value, free fatty acid value, density, flash point, pour point, Kinematic Viscosity and saponification.

Catalyst Preparation

The raw powdered samples were divided into two parts. One part was reserved for further analysis, while the remaining portion of each DPL, KNP, and SOP was calcined at varying temperatures from 500 to 900 °C at 100 °C intervals for 4 h per sample in a muffle furnace (Galenkamp). This process aimed to eliminate carbonaceous and volatile matter, obtaining ash products. The resulting calcined ash samples, designated as Calcined Dried Pawpaw Leaf Ash (CDPLA), Calcined Kola Nut Pod Ash (CKNPA), and Calcined Sweet Orange Peel Ash (CSOPA), were placed in a desiccator and then stored in airtight containers to prevent moisture interaction before further usage. Catalyst preparation procedures followed those detailed by [19,20,22].

2.2. Catalyst Characterization

Characterizations involved Scanning Electron Microscopy/Energy Dispersive X-ray (SEM/EDX), Fourier Transform Infrared (FTIR), and X-ray diffraction analysis (XRD). These analyses aimed to assess elemental composition, morphology, crystalline structures, and active surface functional groups present in the raw powdered samples and calcined ashes (catalysts). Procedures followed those reported by [10,23].

2.2.1. XRF Analysis

The chemical composition of the raw samples was determined through XRF analysis using a PANalytical England Philips Fluorescence Machine at the National Geological Research Laboratory (NGRL), Barnawa Kaduna, with a 15 kV acceleration voltage and 10 nA beam current.

2.2.2. Scanning Electron Microscopy (SEM) Analysis

SEM-EDX analysis was employed to observe the physical morphology of sample surfaces and analyze the element compositions, including visible light elements such as carbon, nitrogen, and oxygen. The EDX detector, equipped with ultra-thin element light windows detecting elements with atomic numbers greater than four, was utilized to analyze contamination, which may impact content quality [3,22].

2.2.3. Fourier Transform Infrared (FTIR) Analysis

Catalysts were characterized by FTIR to identify functional groups and components, ensuring their quality falls within acceptable ranges. FTIR spectra were obtained using a spectrophotometer (Thermo-Nicolet iS10, Thermo Fisher Scientific Inc., Waltham, MA, USA), in the wavelength range of 4000–400 cm⁻¹ [21,24].

2.2.4. X-Ray Diffraction Analysis

X-ray Diffraction (XRD) analysis of each catalyst sample was conducted using a Rigaku-binary diffractometer (SmartL Ab, Japan, Thermo Fisher Scientific Inc., Waltham, MA, USA). The Rigaku-binary instrument measured XRD over Cu-Kα2 radiation with a wavelength of 1.5444 Å, a voltage of 45 kV, and a current of 40 mA. The samples underwent scanning at a speed of 5°/min within the 2θ range of 5–76° to determine the crystalline structure of the catalyst samples [25].

2.2.5. Analysis Method of Loss on Ignition (LOI)

The method is commonly used to measure the organic or volatile content in a sample by calculating the mass difference before and after heating it to a specified temperature as described by [26]. This method is often applied in soils, sediments, clays, and some food samples to assess organic matter, carbonate content, or moisture. The sample was dried to remove moisture at 105 °C before starting the LOI analysis. The dried sample was heated in a Muffle furnace at a high temperature, 550 °C for organic matter for 2 h. The sample is cooled in a desiccator to avoid moisture absorption, then weighed again. The LOI is calculated by taking the difference between the initial and final weights, representing the weight of material lost during heating.

2.3. Development of the Composite Catalyst

Samples of CDPLA, CKNPA, and CSOPA at 900 °C, 900 °C, and 500 °C, respectively, were selected and combined based on the highest level of alkaline metal oxide to develop a heterogeneous catalyst composite from the three selected residues. The simplex lattice design from Design Expert software (12.0.1) was employed to generate mixing ratios. Component levels of 10% and 80% for low and high levels, respectively, were chosen and input into the software to produce a wide range of mixture levels (

Table 1. Selected levels for generation of predicted mixing ratio.

Name	Components		Levels
	Code	Unit	Low	High
CKNPA	A	%	10	80
CDPLA	C	%	10	80
CSOPA	B	%	10	80

). CDPLA, CKNPA, and CSOPA were mixed at different ratios following the fourteen experimental runs generated by the software.

2.4. Catalytic Testing of the Developed Heterogeneous Composite Catalysts

The effectiveness of the developed heterogeneous composite catalysts was assessed through the transesterification of WFO with ethanol. Laboratory experiments were conducted to produce biodiesel with each of the fourteen mixing ratios generated. The reaction conditions included a 4.5 wt.% catalyst loading, 6:1 ethanol-to-oil molar ratio, 65 °C reaction temperature, and 120 min reaction time, all under constant agitation at 700 rpm. Biodiesel yield served as the response variable [27,28].

The transesterification of WFO was conducted in a 500 mL three-necked round-bottom flask fitted with a reflux condenser, a heating controller, a thermometer, and a magnetic stirrer [22]. Upon completion of the reaction, the catalyst was recovered by filtering with Wartman filter paper (100 microns). The reaction mixture was then poured into a separating funnel, allowing it to settle into two distinct layers. The upper layer contained biodiesel, known as WFO Ethyl Ester (WFOEE), while the lower layer consisted of excess ethanol and glycerol. After separation, excess ethanol was removed by evaporation at 78 °C. The biodiesel yield was determined using Equation (1). This transesterification process followed the methods described by [9,21,29].

$$B_y\left(\%\right)={\displaystyle \frac{W_b}{W_o}}\times 100,$$

(1)

where $B_y$ = biodiesel yield, $W_b$ = weight of biodiesel produced, $W_o$ = weight of oil used.

3. Results and Discussion

3.1. Elemental Composition of the Raw Agricultural Residues

Chemical analyses were conducted on the raw residues (DPL, KNP, and SOP) to determine their elemental compositions using EDX analysis. For DPL, the analysis identified the following elements: O, Mg, Al, Si, P, S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Sr, Zr, Nb, Ag, Sn, Ba, W, and Pb. Ta was not detected (

Table 2. The elemental composition of the raw samples’ residues.

S/N	Element	Concentration (%)
		DPL	KNP	SOP
1	O	33.83	29.21	28.90
2	Mg	1.18	7.39	0.13
3	Al	2.75	4.15	3.73
4	Si	7.21	1.64	4.17
5	P	0.59	0.74	0.31
6	S	2.12	1.98	0.98
7	Cl	2.42	2.03	1.72
8	K	7.39	31.18	25.18
9	Ca	38.31	17.09	29.53
10	Ti	0.22	0.21	0.40
11	V	0.01	0.01	Nil
12	Cr	0.01	0.01	0.01
13	Mn	0.30	0.46	0.19
14	Fe	2.93	2.28	2.71
15	Co	0.01	0.03	0.04
16	Ni	0.00	0.01	0.01
17	Cu	0.18	0.21	0.29
18	Zn	0.11	0.12	0.15
19	Sr	0.15	0.08	0.18
20	Zr	0.02	0.04	0.09
21	Nb	0.03	0.05	0.06
22	Mo	0.01	0.01	0.02
23	Ag	0.02	0.06	0.08
24	Sn	0.08	0.96	0.62
25	Ba	0.10	0.06	0.43
26	Ta	Nil	Nil	Nil
27	W	0.01	0.01	0.01
28	Pb	0.02	0.03	0.05

). The top five elements identified in DPL were calcium (38.3%), oxygen (33.8%), potassium (7.4%), silicon (7.2%), and iron (2.9%). Similarly, the elemental composition of KNP was comparable, with potassium being the most abundant element at 31.7%, followed by calcium (17%), and magnesium (7.4%). These results suggest that DPL and KNP share similar compositions, with calcium and potassium being the predominant elements in both samples. Vanadium (V) was absent from SOP, and the elemental composition analysis of SOP revealed the same components as those identified in DPL and KNP. The composition percentages of Ca (25%) and K (29%), respectively, were the highest in DPL and KNP. The most prevalent elements were Ca and K based.

3.2. The Basic Oxide Composition of Selected Raw Agricultural Residues

The corresponding basic oxides of the elements identified by XRF are presented in

Table 3. The oxides composition of the raw residues.

S/N	Compound	Concentration (%)
		DPL	KNP	SOP
1	SiO2	16.10	4.05	10.44
2	V2O5	0.00	0.01	0.00
3	Cr2O3	0.01	0.01	0.01
4	MnO	0.35	0.58	0.25
5	Fe2O3	1.64	1.41	1.70
6	Co3O4	0.00	0.01	0.02
7	NiO	0.00	0.01	0.01
8	CuO	0.18	0.23	0.33
9	Nb2O3	0.01	0.02	0.02
10	MoO3	0.01	0.01	0.01
11	WO3	0.00	0.00	0.00
12	P2O5	0.59	0.82	0.35
13	SO3	4.15	4.27	2.15
14	CaO	59.92	29.52	51.81
15	MgO	3.04	21.04	0.39
16	K2O	5.92	27.61	22.65
17	BaO	0.05	0.03	0.22
18	Al2O3	3.19	5.33	4.86
19	Ta2O5	Nil	Nil	Nil
20	TiO2	0.28	0.31	0.59
21	ZnO	0.10	0.129	0.16
22	Ag2O	0.01	0.02	0.03
23	Cl	4.28	3.96	3.42
24	ZrO2	0.01	0.03	0.07
25	SnO2	0.04	0.56	0.37
26	SrO	0.12	0.06	0.14
27	PbO	0.01	0.01	0.02

, including SiO₂, V₂O₅, Cr₂O₃, MnO, Fe₂O₃, Co₃O₄, NiO, CuO, Nb₂O₃, MoO₃, WO₃, P₂O₅, SO₃, CaO, MgO, K₂O, BaO, Al₂O₃, Ta₂O₅, TiO₂, ZnO, AgO₂, Cl, ZrO₂, SnO₂, SrO, and PbO. In raw DPL, the five major compounds are CaO (59.9%), SiO₂ (16.1%), K₂O (5.9%), Cl (4.3%), and SO₃ (4.1%). The lowest oxide detected was NiO (0.001%). The results reveal that calcium oxide (CaO) is the most prominent compound in DPL, followed by silicon oxide (SiO₂). Other trace compounds include Fe₂O₃, SO₃, MgO, K₂O, and Al₂O₃.

3.3. Characterization of the Selected Calcined Agricultural Residue

Chemical composition analysis of the calcined residues at different temperatures for each of the three selected residues (DPLA, KNPA, and SOPA) was conducted using XRF analysis. The basic oxide composition at different calcination temperatures (500 to 900 °C) identified the presence of SiO₂, Al₂O₃, CaO, Fe₂O₃, MgO, K₂O, Cl, P₂O5, SO₃, TiO₂, and MnO. The Loss of Ignition (LOI) was also estimated for each sample (

Table 4. Chemical composition of the individual calcined residue at different temperatures.

S/N	Oxide	Calcined Temperatures for DPLA (°C)					Calcined Temperatures for KNPA (°C)					Calcined Temperatures for SOPA (°C)
		500	600	700	800	900	500	600	700	800	900	500	600	700	800	900
1	SiO2	21.94	20.19	24.08	23.32	23.13	24.39	27.11	27.08	27.41	27.53	23.00	23.28	22.95	23.93	23.42
2	Al2O3	6.70	5.31	4.66	4.60	4.71	5.37	4.31	5.44	5.63	5.72	1.78	2.41	2.38	2.24	1.87
3	CaO	25.24	26.62	27.21	27.08	27.00	42.81	42.81	41.53	41.72	41.46	30.02	30.74	29.42	30.32	31.62
4	Fe2O3	6.02	4.79	4.62	4.75	4.68	2.00	1.43	2.46	2.50	2.31	1.26	2.17	1.68	1.63	1.87
5	MgO	4.25	4.03	5.48	5.68	5.36	1.26	1.65	3.14	2.45	2.29	3.40	3.64	3.79	4.21	4.08
6	K2O	22.73	21.28	21.78	21.12	21.27	12.08	10.03	10.26	10.00	10.31	22.42	22.73	23.26	24.07	25.93
7	Cl	1.30	1.27	1.31	1.44	1.53	1.65	1.41	1.38	1.42	1.48	0.05	0.04	0.03	0.05	0.03
8	P2O5	2.00	2.13	2.35	2.61	2.50	2.44	2.29	2.32	2.40	2.32	2.07	2.58	2.4	2.52	2.27
9	SO3	1.32	1.30	1.63	1.49	1.32	4.00	4.21	2.16	2.03	2.11	0.3	0.61	0.53	0.33	0.86
10	TiO2	2.10	2.10	2.02	2.00	2.05	0.85	0.80	0.63	0.74	0.61	0.61	0.59	0.38	0.35	0.39
11	MnO	0.56	0.24	0.33	0.31	0.34	1.40	1.40	1.51	1.59	1.38	0.27	0.29	0.32	0.30	0.32
12	LOI	5.30	8.34	4.52	5.54	6.02	1.70	1.84	2.06	2.00	2.02	13.98	10.54	12.63	9.97	7.31

Note: LOI means Loss of Ignition.

). The prominent oxides present are calcium oxide (CaO), silicon dioxide (SiO₂), and potassium oxide (K₂O), similar to their composition in their raw forms. The percentage of SiO₂ remains relatively stable across all temperatures for all three samples. There is a slight increase in SiO₂ content from 500 to 700 °C, followed by a slight decrease at higher temperatures (Table 4). The composition of CaO initially decreases from 500 to 700 °C for DPLA and KNPA but significantly increases for SOPA. At temperatures above 700 °C, the CaO content remains relatively stable for all samples. The percentage of K₂O remains relatively stable across all temperatures for DPLA and KNPA samples.

However, in the SOPA sample, there is an increase in K₂O content with increasing temperature. These major oxides found in the three samples establish the good potential of the selected residues as heterogeneous catalysts for the transesterification process. This aligns with past studies [3,30] where a major compound found in the heterogeneous catalyst is crucial for the transesterification process. The Loss of Ignition (LOI) represents the weight loss due to the release of volatile components during the calcination process. The LOI values vary significantly across different samples and temperatures. In general, the LOI decreases with increasing temperatures, indicating the release of volatile components.

3.4. EDX Analysis of the Composite Heterogeneous Catalyst (CHC)

The elemental composition of CHC obtained from the EDX analysis is detailed in

Table 5. Elemental composition of calcined composite heterogeneous catalyst.

Element	Composition (%)
Silicon (Si)	4.81
Carbon (C)	24.56
Oxygen (O)	7.24
Calcium (Ca)	59.03
Magnesium (Mg)	2.74
Sodium (Na)	1.29

. The results showed that calcium (Ca) had the highest mass fraction at 59.03%, aligning with the EDX spectrum of CHC’s elemental composition displayed in Figure 1 Magnesium (Mg) and sodium (Na) were detected in trace amounts, with mass fractions of 2.7% and 1.3%, respectively. This is consistent with the analysis of the raw samples (Table 2), where calcium was identified as the predominant element. These findings are also in agreement with previous studies [18,23,31]. The results indicate that the composite, after calcination, retains the active element and is therefore well suited for use as a heterogeneous catalyst [31,32,33].

3.5. Scanning Electron Microscopy (SEM) for CHC

The SEM image of CHC, viewed at a 500,000 resolution in Figure 2, shows the agglomeration of non-uniform particles, with clusters of varying shapes and sizes and spongy, fibrous microstructures formed during calcination. The image (Figure 2) reveals a vesicular crystal structure, with some particles bonded into aggregates. These aggregates likely enhance the specific surface area, allowing for optimal contact between the catalyst and reactants. Similar SEM characteristics were observed in eggshell-based heterogeneous catalysts studied by [34]. The particles display an irregular arrangement, and the porous, spongy texture of the ash results from high calcination temperatures that sinter small mineral aggregates and cause particle agglomeration. The non-uniform particle size distribution may stem from the merging of particle structures from the three residue materials during catalyst synthesis. After calcination, the catalyst forms a fine powder, which is indicative of its enhanced surface area and promising catalytic performance [35].

3.6. Functional Group Composition of the Raw and Calcined Composite Residue

3.6.1. FTIR of Raw Composite Residue

The major peak from the FTIR spectra for the raw composite of the selected residue is presented in Figure 3. The peak at 3437.88 cm⁻¹ corresponds to the OH stretching vibration, observed in alcohols, phenols, and carboxylic acids containing hydroxyl groups. The specific peak at this wavelength is attributed to inter- and intra-molecular hydrogen bonding [36]. The peak at 3050.13 cm⁻¹ corresponds to the -CH stretching vibrations of -CH₃ and -CH₂ functional groups, commonly found in compounds containing methyl (-CH₃) and methylene (-CH₂) groups [37]. The peak at 2573.11 cm⁻¹ corresponds to C-H stretching vibrations, characteristic of alkanes, and hydrocarbons consisting of only carbon and hydrogen atoms, where the stretching of C-H bonds gives rise to this peak [38]. The peak at 1950.04 cm⁻¹ corresponds to the C=O stretching vibration of the carbonyl group, characteristic of compounds like ketones and aldehydes [39]. The peak at 1748.21 cm⁻¹ corresponds to the -CO stretching vibration of ether functional groups, present in ethers with an oxygen atom bonded to two alkyl or aryl groups, resulting in the stretching of the C-O bond [39]. The peak at 1592.53 cm⁻¹ corresponds to C=C stretching vibrations observed in aromatic rings, such as benzene derivatives, due to the stretching of the carbon–carbon double bonds in the aromatic ring [39].

The peak at 1400.62 cm⁻¹ corresponds to C-O stretching vibrations in carboxyl functional groups. Carboxyl groups contain both a carbonyl group (C=O) and a hydroxyl group (OH). The stretching of the C-O bond in the carboxyl group gives rise to this peak [38]. The peak at 1009.52 cm⁻¹ corresponds to C-N stretching vibrations in aliphatic amines. Amines are organic compounds containing a nitrogen atom bonded to one or more alkyl or aryl groups. The stretching of the C-N bond in amines leads to this peak [38].

3.6.2. FTIR of Composite Calcined Heterogeneous Catalysts

The major peaks for FTIR analysis of the composite calcined heterogeneous catalyst development are shown in Figure 4. The peak at 3450.47 cm⁻¹ corresponds to the O-H stretching vibration of hydroxyl groups. In this case, the vibration mode also corresponds to the presence of moisture. The stretching of the O-H bond in hydroxyl groups, along with the moisture effect, contributes to this peak [40]. The peak at 2446.20 cm⁻¹ corresponds to the -CH stretching vibrations of -CH₃ and -CH₂ functional groups. These vibrations are commonly observed in alkanes, which are hydrocarbons consisting of carbon and hydrogen atoms. The stretching of the -CH bonds give rise to this peak [40]. The peak at 1601.26 cm⁻¹ corresponds to C=C stretching vibrations observed in aromatic rings. Aromatic compounds, such as those with aromatic rings, exhibit this peak due to the stretching of the carbon–carbon double bonds in the aromatic ring [40]. The peak at 1528.40 cm⁻¹ corresponds to the asymmetrical stretching vibrations of C=O and C-O groups in carbonate ions (CO₃²⁻). Carbonates contain a carbon atom bonded to three oxygen atoms and exhibit specific vibrations corresponding to the stretching of C=O and C-O bonds [39].

3.7. XRD Analysis of Composite Heterogeneous Catalyst (CHC)

The XRD diffractogram in Figure 5 illustrates the crystalline structure of the calcined composite catalyst (CHC), showing multiple peaks that confirm the presence of crystalline phases. These peaks, located at 2θ angles of 16.0°, 21.0°, 24.0°, 29.0°, 35.0°, 38.5°, 42.0°, 46.0°, 49.5°, 56.0°, and 67.0°, are characteristic of calcium carbonate. The prominent peak at 2θ = 35.0° indicates the formation of CaO, with additional calcium oxide peaks appearing at 2θ = 21.0°, 35.0°, 38.5°, and 56.0°. The composite catalyst contains heterogeneous catalysts such as CaO, MgO, and K₂O, suitable for biodiesel production. The higher intensity of CaO peaks compared to other compounds suggests a greater CaO yield. This finding aligns with the FTIR analysis results and is consistent with previous studies [21,34].

3.8. Responses from Experimental Data

D-Optimal Design (DOD) under the Mixture Methodology (MM) of Design Expert Software (version 12.0.1) was adopted for the analysis of data obtained in this study [41]. The results obtained show that Run 5 of the catalyst mixture, comprising 33.33% Kola Nut Pod Ash (KNPA), 33.33% Sweet Orange Peel Ash (SOPA), and 33.33% Dried Pawpaw Leaves Ash (DPLA), respectively, gave the highest biodiesel yield of 65.3% (

Table 6. Biodiesel yield of different mixing ratios of the catalysts.

Run	Component (%)			Response
	A: KNPA	B: SOPA	C: DPLA	Biodiesel Yield (%)
1	45.00	10.00	45.00	16.0
2	21.67	56.67	21.67	32.0
3	21.67	21.67	56.67	61.0
4	10.00	10.00	80.00	47.0
5	33.33	33.33	33.33	65.3
6	80.00	10.00	10.00	52.0
7	10.00	45.00	45.00	54.0
8	45.00	45.00	10.00	36.0
9	80.00	10.00	10.00	44.0
10	10.00	10.00	80.00	60.0
11	56.67	21.67	21.67	37.3
12	10.00	80.00	10.00	44.0
13	10.00	80.00	10.00	47.7
14	45.00	45.00	10.00	30.7

). In contrast, the mixing ratio in Run 1, with a mixture of 45.0% KNPA, 10.0% SOPA, and 45.0% DPLA, yielded the least biodiesel values at 16.0% (Table 6). The biodiesel yield from this study (65.3%) is higher than the 46.3% reported for the use of waste vegetable oil and ethanol, with Epobond Pseudomonas cepacia as a catalyst [42], but less than the 98% reported by [43] from the use of WFO with KOH as the catalyst.

3.8.1. Model Summary Statistics for the Responses

The model summary statistics obtained from the software were utilized to determine the most suitable model for the selected study responses. Properties such as ‘standard deviation’ (Std. Dev.), indicating the degree of error between experimental and predicted values, were examined. The coefficient of determination (R²) reveals the efficiency of the experiment, with higher values expected (≈1), while the adjusted R² (Adj R²) and predicted R² (Pred R²) represent R² values adjusted and predicted by the Design Expert software (https://www.statease.com/software/design-expert/, accessed on 10 May 2024), respectively.

The Std. Dev. values obtained from the available models (linear, quadratic, special cubic, and cubic) embedded in the software were 0.0127, 0.0102, 0.0068, and 0.0026, respectively. Corresponding R² values were 0.04455, 0.5495, 0.8269, and 0.9824 (

Table 7. Model summary statistics for the responses from developed catalyst mixture.

Source	Std. Dev.	R2	Adjusted R2	Predicted R2	PRESS
Linear	0.0127	0.0445	−0.1292	−0.3848	0.0026
Quadratic	0.0102	0.5495	0.2679	−1.3850	0.0044
Special cubic	0.0068	0.8269	0.6785	−0.6881	0.0031
Cubic	0.0026	0.9824	0.9542	0.7417	0.0005 *
Special quartic	0.0026	0.9824	0.9542	0.7417	0.0005
Quartic	0.0027	0.9840	0.9481

* Suggested.

). Similarly, their Adj R² values were −0.1292, 0.2679, 0.6785, and 0.9542; linear, quadratic, special cubic, and cubic had Pred R² values of −0.3848, −1.3850, −0.6881, and 0.7417, respectively, while the PRESS was 0.0026, 0.0044, 0.0031, and 0.0005, respectively.

The cubic model demonstrated the highest R², lowest Std. Dev., and minimum difference between Adj R² and Pred R², with values of 0.9824, 0.0026, and 0.2125, respectively. However, the software suggested ‘cubic’ as the suitable model for the study. Moreover, the special quartic model was considered based on its relatively high R² (0.9824), small differences (0.2125) between the Adj R² (0.9542) and Pred R² (0.7417), the smallest PRESS value of 0.0005, and the lowest standard deviation (0.0026) of the data obtained compared to the mean values [44,45].

3.8.2. ANOVA for the Developed Catalyst Composite

The biodiesel yield response, presented as a mixture quadratic model in

Table 8. ANOVA for biodiesel yield response for the development of catalyst composite.

Source	Sum of Squares	Df	Mean Square	F-Value	p-Value
Model	0.0018	8	0.0002	34.88	0.0006 *
Linear mixture	0.0001	2	0.0000	6.32	0.0428 *
AB	0.0001	1	0.0001	15.84	0.0105 *
AC	0.0015	1	0.0015	224.22	<0.0001 *
BC	2.348 × 10−6	1	2.348 × 10−6	0.3575	0.5760
A2BC	0.0001	1	0.0001	21.62	0.0056 *
AB2C	0.0001	1	0.0001	11.39	0.0198 *
ABC2	0.0003	1	0.0003	49.93	0.0009 *
Residual	0.0000	5	6.567 × 10−6
Lack of fit	3.045 × 10−6	1	3.045 × 10−6	0.4088	0.5573
Pure error	0.0000	4	7.448 × 10−6
Cor total	0.0019	13

Note: * Significant at 0.05 < (prob > F) < 0.1. A = KNPA, B = SOPA and C = DPLA.

, has a model F-value of 34.88, indicating a significant model. There is only a 0.06% chance that a model with such a large ‘F-value’ could occur due to noise. The linear mixture components are equally significant because their ‘Prob > F’ values (p = 0.0428) are less than 0.05 (p < 0.05). Similarly, the model terms such as AB, AC, A²BC, AB²C, and ABC² are significant with p-values of 0.0105, 0.0001, 0.0056, 0.0198, and 0.0009, respectively. The only insignificant model term is BC with a p-value of 0.5760. This result implies that the combination of SOPA and DPLA does not influence the composite catalyst, while other model terms determine the output of the composite catalyst.

3.8.3. Regression Statistics for the Development of Catalyst Composite

The significance levels of the model terms for yield, along with the ANOVA results for the regression model, are presented in

Table 9. Regression statistics for the development of composite catalyst.

Properties	Biodiesel Yield
Standard deviation	0.0026
Mean	0.0253
C.V	10.11
PRESS	0.0005
R2	0.9824
Adjusted R2	0.9542
Predicted R2	0.7417
Adequate precision	22.9195

. The R² and adjusted R² values demonstrate a strong correlation between the observed and predicted response values, as reported in previous studies [21]. A high value of R² indicates that the variables in the model are in agreement, while a low value of R² implies that the variables in the model are poor [46]. The R² obtained for biodiesel yield was 0.9824, while the adj R² value was 0.9542 (Table 9), implying that the observed and predicted values in the model have a reasonable agreement. The value of the R² = 0.9824, being a measure of goodness of fit to the model, indicates a high degree of correlation between the observed value and predicted values. The R² = 0.9824 suggests that more than 98.24 percent of the variance is attributable to the variables, indicating a high significance of the model. Thus, 1.76 percent of the total variance cannot be explained by the model.

Adequate precision, having a ratio greater than 4, is desirable because it indicates an adequate signal of the model, which could be used to navigate the design space [47]. The PRESS is used in determining the suitability of the model in predicting the responses in new experiments; however, small values are desirable [39]. The PRESS value for the biodiesel yield was 0.0005 (Table 8), which shows a desirable PRESS. The coefficient of variations (CV) obtained for the biodiesel yield response of this model is 10.11. CV is the ratio of the standard error of estimate to the mean value of the observed response. It determines the reproducibility of the model; this can be estimated when the value is not more than 10 percent [48]. Hence, low values of CV and SD obtained for the biodiesel yield showed the adequacy with which the experiment was conducted. The model has high R² values, a significant F-value, a low coefficient of variation, and a low standard deviation. Therefore, the results showed a high precision in predicting the biodiesel yield using the developed heterogeneous composite from KNP, DPL, and SOP residues.

3.8.4. Model Equations of Responses for the Development of Catalyst Composite

The final model equation in terms of real components for the biodiesel yield is presented in Equation (2). The model equations show the relationship between the biodiesel yield and the components used (KNPA, SOPA, and DPLA). The terms ‘A, B, and C’ represent KNPA, SOPA, and DPLA, respectively. Each quadratic response of biodiesel yield is discussed based on the two factors’ interaction that can be applied in all experimental regions. This allows one factor to be varied while the other factor is kept constant [49,50].

The quadratic equation generated by the software is represented in Equation (2). The equation can be used to predict the response for given levels of each factor. The high levels of the mixture components are coded as +1, and the low levels are coded as 0. The coded equation is useful for identifying the relative impact of the factors by comparing the factor coefficients. The positive and negative coefficients of a model equation usually indicate the positive and negative effects of the independent variables on the selected responses [51]. The coefficients +47.62, +45.66, and +52.63, obtained for model terms A (KNPA), B (SOPA), and C (DPLA), suggest that the biodiesel yield of WFOB was highly influenced by the three residues used [52]. It is also suggested that the order of influence is C > A > B. Similarly, the coefficient obtained for the mixture of KNPA and SOPA has a positive coefficient of +28.33, suggesting that the mixture of the two residues has a stronger influence on biodiesel yield than the mixture of KNPA and DPLA with a positive coefficient of +5.85. The coefficient of +1.17 also suggests that the mixture of KNPA and DPLA with the square of SOPA influences the biodiesel yield.

$$\mathrm{B}\mathrm{i}\mathrm{o}\mathrm{d}\mathrm{i}\mathrm{e}\mathrm{s}\mathrm{e}\mathrm{l} \mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}=47.62\mathrm{A}+45.66\mathrm{B}+52.63\mathrm{C}+28.33\mathrm{A}\mathrm{B}+5.85\mathrm{A}\mathrm{C}-147.06\mathrm{B}\mathrm{C}-0.85{\mathrm{A}}^2\mathrm{B}\mathrm{C}+1.17\mathrm{A}{\mathrm{B}}^2\mathrm{C}-0.55\mathrm{A}\mathrm{B}{\mathrm{C}}^2$$

(2)

where A is the coded variable for KNPA, B is the coded variable for SOPA and C is the coded variable for DPLA.

The findings in this study show improvement in the catalyst performance over the work of [53] that used coconut shells as a heterogeneous catalyst to esterified waste cooking oil. The reported elemental compositions of CaO (17.32%) and SiO₂ (21.40%) are lower than 42.81% and 27.53%, respectively, obtained in this study. The predicted R² (0.7705) and actual R² (0.9541) values found by [53] are in close agreement with 0.7417 and 0.9542, respectively, obtained in this work.

3.9. Physicochemical Properties of WFO for Testing the Composite Catalyst Developed

The result of the characterization of WFO used in this study is presented in

Table 10. Results of the physicochemical properties of WFO.

Properties	Values Obtained
Density (g/cm3)	910.4
Kinematic viscosity (mm2/s) at 40 °C	32.83
Cloud point (°C)	−6
Pour point (°C)	−11
Flash point (°C)	164
Saponification value (mgKOH/g)	186.27
Acid value (mg/g)	3.48
Free fatty acid (mg/g)	1.74

. The oil properties show a density of 910.4 kg/m³ and flash point of 164 °C; these values are close to 886 kg/m³ and 181 °C, respectively, as used by [54] to achieve an optimum biodiesel yield of 86.5%, and [55] reported 918.5 kg/m³. This implies that the oil meets the requirement for biodiesel production. The saponification level of the oil is 186.27 mgKOH/g. The result of the analysis is also similar to those of [50,56].

3.10. Physicochemical Properties of the Waste Frying Oil Biodiesel (WFOB) Produced

The result of the first three highest biodiesel (WFOB) yields produced, (Run 5 (R₅), Run 3 (R₃), and Run 10 (R₁₀)) (

Table 11. Physicochemical properties of characterized WFOB.

Properties	RUN 5	RUN 3	RUN 10
Specify gravity @40 °C	0.8752	0.8770	0.8804
Kinematic viscosity (@40 °C (mm2s−1)	4.32	4.90	3.75
Moisture content (%)	0.02	0.02	0.02
Saponification (mgKOH/g)	82.46	80.26	80.18
Iodine value (g I2/100 g)	0.41	5.70	3.95
Peroxide value (meq. O2 kg−1)	7.4	6.8	6.2
Refractive index	1.46	1.46	1.45
Flashpoint (°C)	137	160	172
Pour point (°C)	6	8	6
Cloud point (°C)	−4	−2	−2
Cetane number (Ignition quality)	46	45	45
Calorific value (MJ/kg)	37.42	37.38	37.33

), reveals the properties of biodiesel obtained from the present work. The findings show that the values of specific gravity, moisture content, refractive index, and calorific value for the three runs are equal to 0.88, 0.02%, 1.46, and 37 MJ/kg, respectively. Also, the values of 82.46, 80.26, and 80.18 mgKOH/g for saponification are closely the same for the three runs. This trend is also similar for peroxide values, pour points, cloud points, and cetane numbers for the three runs 7.4, 6.8, and 6.2; 6, 8, 6; −4, −2, and −2; 46, 45 and 45, respectively. However, the kinematic viscosity (4.32, 4.90, and 3.75); iodine values of 0.41, 5.70, and 3.95, and flash points 137, 160, and 172 °C obtained for R₅, R₃, and R₁₀, respectively, indicate a little disparity.

The higher the value of the specific gravity of the fuel, the denser the fuel will be. This will affect a number of the fuel’s properties, particularly the flow and the volatility (Foroutan et al., 2018). The specific gravity observed for WFOB was 0.88 (Table 11) which is comparable with petroleum diesel and other biodiesels. Viscosity is another fuel property that will measure the ease of fuel in engines. A less viscous fuel will flow easily in the engine. Thus, a drop in the viscosity is desirable and improves the fuel for application in engines. The kinematic viscosity of 4.32 mm²s⁻¹ for WFOB perfectly fitted into 1.9–6.0 as recommended by ASTM D6751 for biodiesel. This is because the transesterification has converted the triglyceride to methyl esters of the fatty acids which have shorter chains than the triglyceride [57].

The results of these three runs were subjected to Analysis of Variance (ANOVA) using SPSS software to establish if the properties of the WFOB are not statistically different. The result of the ANOVA shows that there is no significant difference between the three runs (R₅, R₃ and R₁₀) as the p-value (0.354) is greater than the 0.05 level of confidence (

Table 12. Tests of between-subjects effects.

Dependent Variable: Observation
Source	Type III Sum of Squares	Df	Mean Square	F	Sig.
Corrected Model	74751.601 *	13	5750.123	211.335	0.000
Intercept	29043.124	1	29043.124	1067.425	0.000
Sample	59.338	2	29.669	1.090	0.354
Properties	74692.263	11	6790.206	249.561	0.000
Error	598.589	22	27.209
Total	104393.314	36
Corrected Total	75350.190	35

* R² = 0.992 (adjusted R² =0.987).

). However, there is a highly significant difference among the properties of the WFOB as the p-value of 0.000 is less than a 0.05 level of confidence. The validity of the result is further shown by the high value of R² (0.992) which is approximately equal to the adjusted R² of 0.987 (Table 12).

The post hoc test sample for multiple comparisons was also conducted using the Least Significant Difference (LSD) of the ANOVA, as shown in

Table 13. Multiple comparisons of sample using LSD.

(I) Sample	(J) Sample	Mean Diff. (I – J)	Std. Error	Sig.	95% Confidence Interval
					Lower Bound	Upper Bound
R5	R3	−2.41932	2.129498	0.268	−6.8356	1.9970
	R10	−2.94960	2.129498	0.180	−7.3659	1.4667
R3	R5	2.41932	2.129498	0.268	−1.9970	6.8356
	R10	−0.53028	2.129498	0.806	−4.9466	3.8860
R10	R5	2.94960	2.129498	0.180	−1.4667	7.3659
	R3	0.53028	2.129498	0.806	−3.88603	4.94659

, which also confirmed that there is no significant difference among the first three runs of WFOB. It is observed from Table 13, that comparing, R₅. R₃, and R₁₀ to one another at a level of 95 confidence interval, all the p-values (0.180, 0.268, and 0.806) are greater than 0.05. The implication of this result reveals that any of the biodiesel properties among the three highest yields well represent the properties of WFOB in further analysis. However, one of the interests in biodiesel production is to obtain an optimum condition that gives the highest yield [19,58]. Therefore, the run (R₅) that gave the highest WFOB yield of 65.3% was chosen for further analysis.

The properties were further compared with the biodiesel standards (ASTM D6751, ASTM D975, EN 14214 and EN590) [59,60]. as shown in

Table 14. The properties of the Waste Frying Oil Biodiesel (WFOB) produced with biodiesel standards.

Properties	Biodiesel Standards				Present Work (WFOB)
	ASTM D6751	ASTM D975	EN 14214	EN590
Physical state	Liquid	Liquid	Liquid	Liquid	Liquid
Specify gravity @15 °C	0.88	NA	0.86–0.9	NA	0.8752
Kinematic viscosity (@40 °C (mm2s−1)	1.9–6.0	1.3–4.1	3.5–5.0	2.0–4.5	4.32
Moisture content (%)	0.050	0.52%	0.5	0.02	0.02
Saponification (mgKOH/g)	<500	NA	NA	NA	82.46
Iodine value (g I2/100 g)	<115	NA	Max 120	NA	0.41
Peroxide value (meq. O2 kg−1)	NA	NA	NA	NA	7.4
Refractive index	NA	NA	NA	NA	1.46
Flashpoint (°C)	100 to 170	60–80	Min 120	55	137
Pour point (°C)	−15 to 10	−35 to −15	NA	NA	6
Cloud point (°C)	−3 to 12	−15 to 5	L and S dependent	L and S dependent	−4
Cetane number (Ignition quality)	48–65	40–45	Min 51.0	51.0 min	46
Calorific value (MJ/kg)	42	NA	35	NA	37.42

. The results of the characterization of the WFOB show that all the properties are within the biodiesel specification described by ASTM D6751, except for the flash point, cetane number, and calorific value. The flash point of 137 °C from this work is higher than 93, 60–80, and 55 °C recommended by ASTM D6751, ASTM D975, and EN590, respectively, but falls into the limit of >130 °C recommended by ASTM D93 [59,60] The result is also in agreement with the 120 °C minimum described by EN 14214 (Table 14). The calorific value of the present work seems to be slightly lower than the one recommended by [59,60] ASTM D6751, but higher than 35 from [59,60] EN 14214. The cetane number of 46 for biodiesel produced in this work is slightly lower than the 48 minimum recommended by ASTM D6751; however, it falls within the range given by [59,60] ASTM D975.

4. Conclusions

A composite catalyst from the ashes of three selected agro-residues of Dried Pawpaw Leaves (DPL), Kola Nut Pods (KNP), and Sweet Orange Peels (SOP) was developed and used for the transesterification of Waste Frying Oil (WFO) with ethanol. The composite developed with D-Optimal Design under the Mixture Methodology (MM) of the Design Expert software (12.0) gave the best mixing ratio of the three residues as 1:1:1 w/w. The application of the catalyst in the transesterification of WFO with ethanol produces WFO Biodiesel (WFOB) with a yield of 65%. The physicochemical analysis of the WFOB produced was found to meet the biodiesel standards. The feasibility of combining the three selected residues (KNPA, SOPA, and DPLA) to develop a composite catalyst for the transesterification of WFO is established. The developed model equation can be used to predict biodiesel yield during the transesterification of WFO using the composite catalyst.