Introduction

Biofuels demand, as an alternative energy source, has been increasing in recent years. The environmental concerns not only to reduce the emissions of greenhouse gases, GHG, but also to improve the security of the energy supply took changes in the environmental policies [1]. Biodiesel, a mixture of fatty acids alkyl esters, it is pointed out as a promising renewable substitute for conventional fossil diesel [2].

Nowadays biodiesel is produced through catalyzed transesterification of vegetable oils and animal fats with short-chain alcohols, such as methanol or ethanol [3]. The mild reaction conditions needed, the fast alcoholysis rate and the easy phase separation, in parallel with its low price and industrial availability, make the methanol the most used alcohol in biodiesel production [4, 5]. However, the use of methanol has some drawbacks due to its toxicity and lack of sustainability. Methanol is majorly obtained from a fossil source such as natural gas reforming [6] thus, the biodiesel produced is not a fully renewable fuel. On the other hand, ethanol is made from agricultural products such as potatoes, grain, and corn, allowing the production of renewable fuel [7]. Ethanol has a higher oil dissolving capacity compared with methanol allowing one phase reaction mixture, whereas oils methanolysis proceeds in a two-phase reaction mixture with mass transfer limitations issues. Due to the extra carbon atom, fatty acid ethyl ester (FAEE) cloud and pour point are lower than that of fatty acid methyl esters (FAME) [8]. Also. the combustion heat and the cetane number are higher for FAEE than for FAME. The storage properties of FAEE fuel are also improved [4]. The main drawbacks of ethanolysis for biodiesel production are its lower reactivity compared with methanol as well as the more difficult separation of biodiesel from the coproduced glycerin [9].

Many studies have been carried out to compare the effect of methanol and ethanol on biodiesel production from different feedstocks [10,11,12,13].

Recently, Bolonio et al. [14] reported the production of fully renewable biodiesel using wastes from Spanish winery. The authors used bioethanol obtained by distillation of grapes leftovers and oil obtained by soxhlet extraction of grapeseeds. The ethanolysis was accomplished using a homogeneous sodium ethoxide catalyst.

All these studies reported that the yield obtained by ethanolysis is lower and more time is needed to complete the reaction than for methanolysis. The separation of FAEE from glycerin is more difficult too. However, it allows achieving completely renewable biodiesel. Garcia et al. [11] also studied the influence of alcohol on biodiesel production from animal fats. The results showed that although all biodiesel samples presented lower yield values when the ethanol is used, FAEE presented bigger resistance to oxidation than FAME.

Transesterification reactions are catalyzed by homogeneous basic catalysts although the interest in the use of heterogeneous basic catalysts is growing (Table 1).

Table 1 Heterogeneous basic catalysts for oil ethanolysis and reactions conditions from literature
Full size table

The main advantages of heterogeneous catalysis are the easy catalyst separation from the reaction mixture and the ability to be reused [21]. As can be seen from the data presented in Table 1, calcium oxide is extensively used as a heterogeneous basic catalyst for the oils ethanolysis. Several attempts have been made to modify surface basicity by doping with an alkali metal and other elements. The conventional homogeneous basic catalyst (KOH) is always more active than heterogeneous CaO raw or modified catalysts.

Since the transesterification reaction is affected by several parameters, such as temperature, reaction time, Wcat/Woil, alcohol/oil ratio, among others, the research surface methodology (RSM) is often used to optimize the reaction conditions which allows to maximize the biodiesel production [22,23,24,25]. Neural networks, with genetic algorithms, are used less frequently for the same purpose [4].

In this work, the ethanolysis reaction of soybean oil was carried out over CaO heterogeneous catalyst obtained by calcination of calcium-rich alimentary wastes (scallop shells) and the reaction conditions were optimized using the research surface methodology.

Material and methods

Materials and catalyst preparation

Biodiesel (FAEE) was produced by soybean oil (alimentary grade) ethanolysis over heterogeneous Ca-based catalyst. The catalyst was prepared by calcination of alimentary wastes scallop shells. The as received shells were washed, dried, and coarsely crushed and then calcined in a muffle at 900 °C for 3 h. The heating step was performed at 5 °C/min. The calcined Ca material was ground in a ceramic mortar to obtain a fine powder (< 750 μm). More details on the catalyst preparation are given elsewhere [26].

Characterization of the catalyst and ethanolysis products

The catalysts, fresh and post-reaction, were characterized by X-ray diffraction (XRD) to identify the Ca crystalline phases and the degree of crystallinity of each phase. The diffractograms were recorded with a Rigaku Geigerflex diffractometer with Cu Kα radiation at 40 kV and 40 mA (2° min−1). ATR-FTIR spectra were collected for fresh and post-reaction catalysts to evaluate catalyst surface species, which can participate in the reaction mechanism. The FTIR spectra were recorded with a spectral resolution of 4 cm−1 on a Perkin Elmer Spectrum Two IR Spectrometer equipment with an ATR accessory from Pike using a ZnSe crystal. Liquid products from ethanolysis, oily phase containing FAEE and glycerin, were also characterized by ATR-FTIR.

Ethanolysis reaction tests

The ethanolysis tests were carried out in a 500 mL three-neck round-bottom pyrex flask equipped with a condenser and a magnetic stirrer. The reaction temperature was kept using a nest-shaped heating jacket with temperature control. The temperature sensor was in contact with the reaction mixture. All the reaction tests were performed at ethanol reflux temperature (75 °C) for ethanol:oil molar ratios of 10:1, 12:1, and 14:1. Tests using catalyst loading of 10.0, 12.5, and 15.0% (oil basis weight) were accomplished for three different reaction time (6, 8, and 10 h). For each reaction batch, the catalyst was previously contacted with methanol for 0.5 h and then the pre-heated oil (100 g, 60 °C) was added to the reaction flask. After the reaction period, the slurry was filtrated to remove the solid catalyst. The liquid effluent was transferred into a decantation funnel for glycerin phase separation. The oily phases were characterized, without further purification by ATR-FTIR to quantify the FAEE yield. The quantification was performed considering the reflectance bands in the range 1780–1690 cm−1 (Fig. 1) being the band centered at 1736 cm−1 attributed to biodiesel (FAEE) as reported by Zagonel et al. [27].

Fig. 1
figure 1

ATR-FTIR spectrum of oily phase containing FAEE and unreacted soybean oil

Full size image

RSM analysis and optimization

Research surface methodology (RSM) was applied to evaluate the influence of reaction parameters on the FAEE yield. The Minitab® software (Version 18, Minitab Inc. USA) was used to evaluate the parameters influence, and interaction, as well as to optimize the reaction conditions (maximization of FAEE yield). Graphical analysis of the data was also obtained using Minitab software. The experimental design applied to this study was a full 33 factorial design consisting of 27 experiments. The coded and uncoded levels of the independent variables evaluated are presents in Table 2.

Table 2 Independent variables and their levels for RSM modeling and optimization
Full size table

The experimental data achieved was analyzed by response surface regression to fit the following polynomial equation [28]:

$$ {\text{Y}} = \upbeta_{0} + \mathop \sum \limits_{{{\text{i}} = 1}}^{{\text{k}}} \upbeta_{{\text{i}}} {\text{X}}_{{\text{i}}} + \mathop \sum \limits_{{{\text{i}} = 1}}^{{\text{k}}} \upbeta_{{{\text{ii}}}} {\text{X}}_{{\text{i}}}^{2} + \mathop \sum \limits_{{{\text{i}} > {\text{j}}}} \sum \upbeta_{{{\text{ij}}}} {\text{X}}_{{\text{i}}} {\text{X}}_{{\text{j}}} + \upvarepsilon $$
(1)

Here Y is FAEE yield (response), Xi and Xj are the independent process variables (ethanol:oil molar ratio, catalyst loading, and reaction time). β0 is a constant, βi, βii, and βij are the regression coefficients, and k is the number of factors studied and optimized in the experiment. In this case, k is equal to 3. ε is the error term.

Results and discussion

Characterization of the catalyst

The fresh and post-reaction catalysts were characterized by XRD diffraction (Fig. 2). As expected, the scallop shells calcined at 900 °C present XRD patterns ascribable to lime (CaO) slightly contaminated with small crystallites (broad XRD lines) of calcium hydroxide (portlandite).

All the post-reaction catalysts presented XRD patterns similar to that depicted in Fig. 2, showing XRD lines belonging to calcium diglyceroxide and calcium hydroxide. Calcium diglyceroxide is reported to have enhanced catalytic activity in oil alcoholysis [29] but is also highly unstable leading to Ca leaching into reaction medium [30, 31]. For methanolysis the Ca(OH)2 phase is believed to be the catalyst active phase [30] and for oil ethanolysis Watcharathamrongkul et al. [8] reported an improved catalytic behavior for CaO catalyst loaded with Ca(OH)2. From XRD patterns it seems that CaO suffer deactivation during oil ethanolysis may be due to the long reaction periods used.

Fig. 2
figure 2

XRD patterns of fresh and post reaction (10 h; MeOH/Oil = 14 molar ratio; 15% Wcat/Woil) catalysts

Full size image

XRD data, in Fig. 2, show that the catalyst dynamics during ethanolysis is similar to that reported to occur during methanolysis [30]. Thus, catalyst reactivation can be achieved by calcination to remove hydroxide and diglyceroxide calcium species.

To evaluate the species, from the reaction medium, which remains adsorbed on the post-reaction catalysts ATR-FTIR spectra were acquired and compared with that from fresh catalyst [30]. IR spectra in Fig. 3 show for fresh catalyst reflectance bands to CaO surface with adsorbed carbonate species (1700–1100 cm−1 wavenumber) [32] whereas post-reaction catalysts show IR bands belonging to adsorbed esters species (1740 cm−1) and oil and glycerin species. The features in the range 3000–2800 cm−1 are attributable to alkyl species from adsorbed oils and esters.

Fig. 3
figure 3

ATR-FTIR spectra of catalyst (fresh and post-reaction)

Full size image

The co-produced pale yellow, low viscosity glycerins were characterized by ATR-FTIR in order to assess the main functional groups of contaminants (MONG- organic matter non glycerin). The spectrum of a commercial glycerin (purum) was used as a reference. FTIR spectra, displayed in Fig. 4, show infrared features of glycerins slightly contaminated with ethanol and ethyl esters (1740 cm−1). As reported by Dias et al. [30], for oil methanolysis over calcium catalysts the purity of the obtained glycerin can be used as a measure of the extent of homogeneously catalyzed processes. The author stated that homogenous processes produce a brownish glycerin whereas pure heterogeneous processes allow to a clear glycerin phase. Spectral data and co-produced glycerin allowed to conclude that homogeneous catalyzed contribution can be discarded.

Fig. 4
figure 4

ATR-FTIR spectra of co-produced glycerins and commercial glycerin was used as standard [spectra labels are ethanol:oil molar ratio; Wcat/Woil (%); reaction time(h)]

Full size image

Ethanolysis reaction

The experimental and predicted values for biodiesel yield responses and all three variables coded form are given in Table 3. The maximum and minimum of FAEE yield achieved was 98.2% and 56.8% respectively. The RSM was applied to the experimental data of the FAEE yield to fit the polynomial equation, Eq. 1, using the coded levels of variables:

Table 3 Experimental and predicted FAEE yield (%) values achieved in each experimental test
Full size table

\({\text{Y}} = - 147 + 8.6A + 29.4{\text{B}} - 0.8{\text{C}} - 0.632A^{2} - 1.036B^{2} - 1.585C^{2} - 0.732{\text{AB}} + 1.685{\text{AC}} + 0.806{\text{BC}}\)

The graph between predicted and experimental biodiesel yield (%), Fig. 5, shows that predicted values are quite close to experimental ones. The correlation between the experimental and predicted FAEE yields was 0.921. This evidence validates the achieved equation.

Fig. 5
figure 5

Predicted vs experimental FAEE yields values for ethanolysis

Full size image

The equation along with the Pareto chart, Fig. 6, shows that ethanol:oil molar ratio (A), the catalyst loading (B), the interaction of ethanol:oil molar ratio with time (AC) and catalyst loading with time (BC) have positive effects on the FAEE yield. Although the time, other interactions and square interactions influenced the FAEE yield negatively. In Fig. 6, the bars that represent variables C (time), AC, A (ethanol:oil molar ratio), BB, CC, and BC cross the reference line at 2.110. These parameters are statistically significant at the 0.05 level with the current model terms. Of all these parameters C (time) shows the most significant effect.

Fig. 6
figure 6

Pareto chart of the standardized effect of FAEE yield (α = 0.05)

Full size image

Influence of process variables on biodiesel yield

Interactions of process variables on the FAEE yield were also analyzed. They are exhibited graphically in response surface and contour plots, Fig. 7.

Fig. 7a shows the interaction between ethanol:oil ratio and amount of catalyst on FAEE yield achieved. The results show that independently of the amount of catalyst, yield decreases with the increase of ethanol:oil. On the other hand, an initial increase in the amount of catalyst increases the yield but after achieving the maximum value the yield decrease; except for 10:1 ethanol:oil ratio this decrease does not happen.

Fig. 7
figure 7

Response surface and contour plot of a ethanol:oil ratio (A) and catalyst loading (B) interactions impact on FAEE yield with 8 h of reaction time; b ethanol:oil ratio (A) and time (C) interactions impact on FAEE yield with 12.5% of catalyst loading; c catalyst loading (B) and time (C) interactions impact on FAEE yield with 12:1 of ethanol:oil ratio

Full size image

Fig. 7b shows the interaction between ethanol:oil ratio and time on the FAEE yield achieved. The results show that for lower ethanol:oil ratio the increase of reaction time increases the yield but after reaching the maximum begins to decrease. However, for large ethanol:oil ratio the increase in time increases the yield.

The interaction between the amount of catalyst loading and time on FAEE yield is shown in Fig. 7c). For a small amount of catalyst, the increase of reaction time slight increases the yield but for a large amount, the increase of the reaction time increases significantly the FAEE yield.

Optimization of process variables

The optimization of independent variables was done to maximize de FAEE yield based on the polynomial equation obtained. The optimal reaction parameters achieved were: 11:1 ethanol: oil molar ratio; 13.8% catalyst loading and 9.1 h with 99.2% maximum FAEE yield predicted.

For a similar catalytic system, Avramović et al. [4] obtained slightly different optimal conditions (Table 4) because they used higher catalyst loading and higher ethanol/oil molar ratio which allowed them to decrease the reaction time.

Table 4 Optimal reaction conditions for oil ethanolysis over CaO catalysts using RSM (adapted from [4])
Full size table

Three replicas of the reaction test in the optimal conditions were performed to validate the above model (Eq. 1). The three replicas allowed an average FAEE yield of 99.0% with a variation of 0.19% between the predicted and experimental yield.

Conclusions

Fatty acids ethyl esters (FAEE) were successfully produced by oil ethanolysis using a basic heterogeneous Ca catalyst prepared by the calcination of scallop shells. During ethanolysis, the CaO catalyst was converted into Ca(OH)2 which reacted with produced glycerin forming the calcium diglyceroxide, thus showing a catalyst dynamics analogous to that reported for methanolysis.

The research surface methodology was successfully applied to the experimental data of the FAEE yield and the obtained analytic function was used to optimize the reaction conditions. The optimal parameters achieved were 11:1 ethanol:oil molar ratio; 13.8% catalyst loading with 9.1 h of reaction time having reached a maximum FAEE yield of 99.0%.