An innovative microreactor approach for sustainable biodiesel production: process design, continuous purification and comparative LCA

Abstract

This study presents a fully integrated continuous-flow approach for the production and purification of fatty acid methyl esters (FAME), addressing key inefficiencies in conventional batch processes. Unlike traditional methods that require prolonged reaction times, excessive methanol use, and large volumes of water for purification, our microreactor-based system optimizes transesterification efficiency while significantly reducing waste generation. This study aims to seamlessly integrate microreactors with downstream purification steps—including extraction and biphasic liquid–liquid separation—to achieve continuous production. The optimized system achieved a FAME yield of 91.14 ± 13.85%, with an acid value as low as 0.268 mg KOH per g, well below the standard limit. The phase separation purities ranged from 73.15% to 95.11%. Life cycle assessment at an industrial scale demonstrated a 35% reduction in water consumption and a lower carbon footprint compared to batch production. This scalable and automated continuous approach advances sustainable biodiesel manufacturing by enhancing efficiency, reducing environmental impact, and promoting decentralized energy production.

---

Cheng-You Yang

Mr Cheng-You Yang joined Prof. Ya-Yu Chiang's lab in 2023 and is currently pursuing a PhD degree in mechanical engineering at National Taiwan University (Taiwan). His research focuses on the continuous process and purification of fatty acid methyl esters and sustainable aviation fuel.

Cheng-Yu Wang

Mr Cheng-Yu Wang joined Prof. Ya-Yu Chiang's lab in 2021 worked on the novel liquid–liquid extraction and separation strategies project and graduated in 2024. His research focuses on continuous biodiesel manufacturing process development using a combination of unit operators.

Ni Ni Myint

Dr Ni Ni Myint is a researcher in the Department of Chemical Engineering at Kasetsart University, Thailand. She received her doctoral degree in Chemical Engineering from Kasetsart University, where she was granted a full scholarship for her PhD studies (2019–2022). Following her doctoral studies, she received a prestigious one-year postdoctoral fellowship (2023–2024) from the Faculty of Engineering at Kasetsart University. Her research focuses on life cycle assessment (LCA) and techno-economic analysis (TEA).

Yi-Chun Chen

Dr Yi-Chun Chen received her bachelor's and master's degrees from the Department of Forestry at National Chung Hsing University, Taichung, Taiwan, in 2003 and 2005, respectively, and here Ph.D. degree from the Department of Chemical Engineering, National Tsing Hua University, Hsinchu, Taiwan, in 2012. In 2013–2014, she was a visiting researcher at the Department of Materials Engineering of The University of Tokyo in Japan. Then, she joined InnoCept Biomaterials, Inc. In February 2015, she joined the Department of Forestry, National Chung Hsing University in Taiwan as an assistant professor. She got promoted to associate professor in August 2019. Her research interests include bio-based materials utilization, functional polymers and nanomaterials.

Penjit Srinophakun

Professor Dr Penjit Srinophakun graduated in chemical engineering from the University of Queensland, Australia, in 1996. She is now the director of the KU-biodiesel Project. She was the director of the Center of Excellence for Jatropha (2010–2024), the president of Thai Society for Biotechnology (2010–2016 and 2020–2022), and Vice President of the Asian Federation of Biotechnology (AFOB) (2010–2022). Her research covers waste valorization, biofuels, biochemicals and biorefineries, techno-economic feasibility, life cycle assessment, etc.

Ya-Yu Chiang

Prof. Ya-Yu Chiang is an associate professor at the Department of Mechanical Engineering of National Taiwan University. She obtained her doctoral degree from the Department of Biochemical and Chemical Engineering, Technische Universität Dortmund, Germany. Her current research interests are fluidic manipulation, biomimicry, and multiphase mass transfer and mechanics. Dr Chiang has been awarded the Ta-You Wu Memorial Award, Excellent Young Scholars, and Taiwan Future Tech Awards from the National Science and Technology Council. She has also received several other awards, including Corning Outstanding Female Researcher 2023 and a gold medal in the Taiwan Innotech Expo Innovation Patent Competition.

---

Green foundation

1. Our work advances green chemistry by developing a continuous flow microreactor system integrated with novel purification technologies. This process minimizes reliance on traditional batch systems that are resource-intensive and environmentally taxing. This method offers a more sustainable and efficient approach for biodiesel production, promoting reduced environmental impact and increased scalability for industrial applications.

2. The continuous system achieved a %FAME value of 91.14 ± 13.85%, while significantly lowering the acid value and methanol residues to acceptable levels. Water consumption was decreased by 35% compared to traditional batch methods. Additionally, the helix wire separator enabled a novel and efficient method for biphasic liquid separation, minimizing energy usage.

3. Future research could scale the helix wire separator, explore enzymatic catalysts, adopt renewable energy, and optimize methanol recovery to further reduce the carbon footprint and enhance sustainability in industrial applications.

1. Introduction

With increasing global attention on environmental issues, biodiesel has emerged as a promising alternative due to its low carbon emissions and renewable nature.^1–4 However, commercial conventional biodiesel production not only suffers from inefficiencies such as large reactor volumes, extended diffusion distances, uneven heating, and prolonged reaction times but also operates as a large-scale, high-risk, centralized energy production model. This centralized nature requires extensive infrastructure, occupies significant land area, and poses inherent safety risks associated with bulk chemical processing. To overcome these limitations, there is growing interest in transforming centralized plant generation into distributed generation systems that prioritize energy security, operational flexibility, and regional power autonomy. In this context, microreactor-based full-flow systems offer a transformative solution, as their compact footprint and potential for complete automation enable decentralized, modular installations that can be tailored to meet localized or mission-specific energy needs. Such systems offer particular advantages for biodiesel production in residential zones, off-grid applications, or safety-critical environments where conventional plant-scale approaches are impractical or hazardous. While various enhancements such as ultrasound-^5–7 or microwave-assisted transesterification^8–10 have been employed to accelerate mass transfer and improve yields, or supercritical alcohol transesterification techniques^11–13 have been developed to eliminate catalyst usage and reduce purification steps—especially for high free fatty acid feedstocks—these approaches typically demand elevated temperatures and pressures above the critical point of methanol or ethanol. As such, they impose significant challenges in equipment maintenance, system control, and safety management, particularly in community-scale or decentralized settings where fluid storage and pressure regulation are constrained by public safety standards.

As the chemical industry advances toward microreactor-based continuous flow processes, these challenges are being addressed through process intensification.^14–17 Microreactors offer a higher surface-area-to-volume ratio, significantly enhancing reaction efficiency and product purity.^18–21 Additionally, microreactor systems can operate beyond solvent boiling points and even under supercritical conditions, enabling reaction acceleration through increased flow rates and significantly boosting overall production efficiency.^22–24 Beyond these fundamental advantages, transitioning biodiesel production from batch to a continuous microreactor approach represents a paradigm shift in energy production. This transformation improves process efficiency and aligns with a broader movement toward decentralized and modular energy systems, reducing dependence on large-scale facilities and enhancing flexibility in distributed energy generation. Such advancements hold significant implications for carbon footprint reduction, energy autonomy, and disaster resilience—particularly when integrated with self-healing microgrids and flow-reactor-based biodiesel systems for secure and adaptable emergency power solutions. The production of biodiesel involves transesterification processes; a chemical reaction in which fatty acid compounds react with alcohols in the presence of enzymes and acidic or alkaline catalysts, exchanging carboxylic and alcohol groups to form new alkyl esters and glycerol.^25–29 Batch production can achieve FAME yields above 95%, but the low mass transfer efficiency necessitates longer reaction times (>30 minutes), and the reactors require significant space, which impacts environmental sustainability. However, one of the significant challenges in continuous biodiesel production is the effective handling of downstream biphasic liquid processing, including recovery, separation, extraction, and purification. Despite substantial advancements in feedstock handling and reactor design over the past decades—widely documented in research and patents—downstream separation still heavily relies on batch-based methods, such as gravitational settling,^30,31 centrifugation,^32–34 filtration,^35,36 and membrane separation.^37–39 Conventional filtration and membrane separation techniques frequently suffer from membrane fouling issues,^40,41 particularly when dealing with high-viscosity or oil-rich biphasic liquids, limiting their practicality in continuous flow operations. The lack of micro-scale biphasic reactors capable of continuous separation and purification of viscous or oil-based flows remains a critical bottleneck in automating continuous biodiesel production.

Life cycle assessment (LCA) provides us with a comprehensive method used to evaluate the environmental impacts of two biodiesel production processes from soybean oil throughout its entire life cycle, from raw material extraction to end-of-life disposal. LCA is critical in determining biodiesel viability as a sustainable alternative to fossil fuels. Biodiesel production offers advantages such as cost reduction, improved waste management, and potential environmental benefits. However, the environmental impact of bio-based products is highly dependent on specific feedstock characteristics and production pathways. Conducting a case-specific LCA ensures a thorough understanding of the ecological trade-offs and helps avoid unintended shifts of impacts between different life cycle stages.⁴²

This study aims to leverage the benefits of flow chemistry to address the current challenges in downstream purification in continuous processes, transitioning from batch biodiesel production to continuous flow systems. The optimization parameters in this investigation were methanol-to-oil ratios, total flow rates, and reaction times to enhance transesterification efficiency. Additionally, specialized helical coil structures with a unique biphasic flow interface control technology will effectively separate biodiesel from the extracted wastewater. Finally, a life cycle assessment compares the potential environmental impacts of the developed continuous FAME process and the conventional ones.

2. Experimental principle and method

This study proposes a continuous flow green transesterification process for biodiesel using soybean oil and methanol, with 1%NaOH by weight as the alkaline catalyst (Fig. 1). The reaction was carried out in a low-flow microreactor designed for high-efficiency mixing. The low-flow microreactor features a unique heart-shaped structure, allowing immiscible oil and methanol feed liquids to interact effectively. One liquid maintains its minimum surface energy through its surface tension, forming dispersed droplets within the other liquid. This design significantly enhances the interfacial area between the two phases and reduces mass transfer distances, thereby increasing the efficiency of the transesterification reaction.

Fig. 1 The reaction schematic of continuous biodiesel production.

Upon the transesterification, the initial feedstock of soybean oil and methanol was converted into the target product, biodiesel, and the byproduct, crude glycerol. At this stage, introducing pure water or a mildly acidic aqueous solution extracts excess methanol, crude glycerol, and the unreacted NaOH catalyst from the mixture; this step, known as washing or acid washing, represents the first phase of downstream purification. In this phase, biodiesel and wastewater remain as two immiscible liquid phases. The process then proceeds to the second downstream purification step: continuous separation, completely separating biodiesel from the wastewater to achieve a high-purity product.

This study will investigate various operational conditions, including different flow rates and methanol-to-oil molar ratios of the feedstocks, to optimize the transesterification and separation efficiencies to the fullest extent. The goal was to develop a fully integrated and optimized continuous-flow green production process for biodiesel production. In addition, this new process's 18 midpoint environmental impacts were compared with the conventional batch operation, leading to higher process efficiency and more sustainable biodiesel production.

2.1 Transesterification reaction

Transesterification reactions are crucial for converting vegetable oils or animal fats into biodiesel.⁴³ The transesterification reaction involves a three-step reversible process where triglycerides (TGs) in the oil phase react with methanol in a suitable catalyst.⁴⁴ The first step produces diglycerides (DGs) and the first fatty acid methyl ester (FAME) molecule. In the second step, the diglycerides react with methanol to form monoglycerides and a second molecule of FAME. Finally, the monoglycerides react with methanol to produce the third molecule of FAME and glycerol as a byproduct. Stoichiometrically, when 1 mole of triglyceride reacts with 3 moles of methanol, it makes 3 moles of FAME and 1 mole of glycerol. While the theoretical methanol-to-oil molar ratio is 3, in practical applications, excess alcohol (6, 9 and 11 moles) is used to drive the reaction towards the product side due to the reversibility of the reactions, enhancing FAME production.⁴⁵ Catalysts also play a vital role in transesterification by effectively facilitating the reaction. An alkaline catalyst, 1 wt% NaOH, was used in this study.^30,46–48

2.2 Continuous flow green production of FAME

The continuous flow green biodiesel production process in this study differs from the traditional methods described earlier. This process efficiently carries out high-flow transesterification reactions, followed by effective acid washing and extraction of the products, and then directly and continuously separates the immiscible biodiesel and wastewater in downstream stages (Fig. 2B). This approach eliminates the need for gravity settling for phase separation, reducing reactor residence time and spatial requirements.

Fig. 2 (A) Batch and (B) continuous flow green biodiesel production. (C) Photo of a continuous flow production system.

A successful transesterification reaction depends on the efficient mixing of triglycerides (TGs) and alcohol, which allows for a reduction in the amount of alcohol required and thus decreases costs. This study used a Corning low-flow reactor (AFR, Corning, France) microreactor to mix commercial soybean oil and methanol. The heart-shaped microstructures within the reactor facilitate overcoming capillary limitations, breaking the immiscible liquids into finer droplets, enhancing the surface area-to-volume (SA/V) ratio, and reducing mass transfer distances between the phases.^49–51 However, excessively tiny droplets can lead to emulsion formation, which hampers subsequent phase separation; thus, achieving a balance between droplet size and separation efficiency is critical. The transesterification reaction produces two immiscible liquid phases: FAME and glycerol. However, due to the excess methanol used in the reaction, the mixture often contains residual methanol and alkaline catalysts, necessitating extraction to remove non-FAME components. Liquid–liquid extraction operates on the principle of the partition law, which utilizes the differential solubility of compounds in two immiscible solvents to transfer the target compound from one phase to another. For liquid–liquid extraction to be effective, two conditions must be met: first, the selection of two immiscible solvents—one as the original solvent phase and the other as the extractant, and second, allowing the mixture to contact the extractant to form an interface where the target compound, due to its differential solubility, moves into the phase where it has higher solubility.⁵² This study used a 150 mM hydrochloric acid aqueous solution as the extractant. The product and acid solution were introduced together via a T-junction, forming droplets or slug flow within the tubing due to capillary and shear forces, thus increasing the contact area between the product and the extractant. This process enables the transfer of the byproduct glycerol, residual methanol, and alkaline catalysts into the aqueous phase, achieving a similar effect to batch extraction with stirring.

After extraction, the resulting product consists of biodiesel and an aqueous solution containing glycerol, methanol, and neutralized catalyst, referred to as wastewater in subsequent sections. Biodiesel and wastewater remain as two immiscible liquid phases. As the transesterification and extraction efficiency improves, the SA/V ratio of the droplets increases, and the droplet size decreases, approaching emulsion. Relying on gravity settling or centrifugation under these conditions would significantly prolong separation time, increasing spatial and operational costs. Thus, a continuous biphasic liquid separation and purification system addresses these challenges. The continuous purification system employed in this study has been extensively reported in our previous research.^23,53–57 It flows the mixed droplets through a helix wire, where one liquid phase exits through the coil's exterior gaps while the other remains in the wire's core. This configuration successfully separates the immiscible liquid phases. As the mixed droplets flow into the helix wire, the liquid–liquid–air interface between the two phases and the external air is stabilized by the Laplace pressure, ΔP_Laplace, generated between the liquid and the microstructures and by the system's pressure drop. By adjusting the Laplace pressure of the helix wire and the pressure drop during sample introduction, the separation of the phases is controlled, allowing the low surface tension liquid to either exit or remain within the helical coil, achieving effective separation.⁵⁵ This continuous flow process not only improves the production yield but also represents a greener and more sustainable method of FAME production by integrating highly efficient separation techniques into the transesterification workflow. The integrated system is shown in Fig. 2C.

2.3 Principle of biphasic liquid flow behavior control

To achieve a complete oily fluid separation, we direct the oily phase (biodiesel phase) and wastewater into different flow paths, the core position within the helix wire and, dripping out of helix wire, towards its own outlets. One liquid pressure must be lower than ΔP_Laplace to retain a liquid within the helix wire. However, as liquids flow, they experience pressure loss due to viscous forces, denoted as ΔP_ddrop. The interaction between ΔP_Laplace and ΔP_ddrop affects the stability of the liquid interface. By controlling ΔP_ddrop to be greater than the ΔP_Laplace of one of the liquid phases relative to the air, selective separation of a single-phase liquid can be achieved. The study assumes that introducing liquid at the inlet of the helix wire generates an input pressure, P_in. When the sum of the Laplace pressure ΔP_laplace between the biphasic droplets and the outlet pressure, P_out, is less than P_in, the liquid flows towards the outlet. If the inlet flow rate is maintained constant and the outlet pressure drop is increased (e.g., by extending the flow path length), P_in will also increase. Since the composition of the biphasic droplets remains unchanged, ΔP_laplace remains constant, resulting in increased pressures within each phase, P_waste (wastewater phase) and P_BD (biodiesel phase). For effective separation, the pressures must satisfy the conditions outlined in eqn (1) and (2):

P_BD > ΔP_Laplace between biodiesel and air (1)

P_waste < ΔP_Laplace between waste water and air (2)

Calculating the pressure balance at the interface of the two immiscible liquids involves numerous complex parameters and formulations, detailed in the equations and methods presented in our previous studies.^54,58 For simplicity, this paper illustrates the pressure balance between the two phases using a free-body diagram (Fig. 3A), with blue and yellow colors representing the wastewater phase and the biodiesel phase, respectively.

Fig. 3 (A) A free-body diagram of pressure balance between the biphasic liquid. (B) A schematic of the phase separation process.

These relationships indicate that when the absolute pressure of the biodiesel phase exceeds the interface pressure between biodiesel and air, the biodiesel phase droplets will be expelled from the helix wire (Fig. 3B). Simultaneously, to ensure that the wastewater phase droplets remain within the helix wire, their absolute pressure must be less than the interface pressure between the wastewater phase and air. Factors such as interfacial tension, polarity, and the wire gap width of the helix wire biphasic continuous separator influence the separation purity. The most distinctive feature of this study is the use of a helix wire as a continuous separator for biphasic oily fluids, which has not been previously reported in the literature. This innovative approach leverages the precise control of ΔP_Laplace and ΔP_ddrop dynamics to selectively separate immiscible liquid phases in a continuous flow, offering a novel and efficient solution for the challenges of biphasic fluid separation in constant flow systems.

2.4 Experimental procedure and system specifications

The complete continuous flow green biodiesel production process can be divided into three main stages: the front-end transesterification reactor, the first-stage purification step using a T-junction liquid–liquid extractor, and the second-stage purification step involving a continuous biphasic liquid separation and purification system. The process begins with commercial soybean oil (supplied by Taiwan Sugar Corporation, Taiwan), methanol (HPLC grade, REAGENTS DUKSAN, Korea), and 1 wt% sodium hydroxide (NaOH, Honeywell Fluka™, USA) as the alkaline catalyst. These reactants are introduced into a temperature-controlled low flow microreactor set at 60 °C using an HPLC pump (LC-40D, SHIMADZU, Japan) at flow rates ranging from 1500 to 6000 μL min⁻¹. Efficient mixing within the microreactor facilitates high-efficiency transesterification. A T-junction is installed at the reactor's outlet, allowing the transesterified products to mix with a 150 mM hydrochloric acid aqueous solution, forming a mixed droplet flow. This mixture is then introduced into a PFA tube with an outer diameter of 1/8 inch, an inner diameter of 1/16 inch, and a total length of 1000 mm, serving as the liquid–liquid extractor for the first-stage purification step.

Following extraction, the biodiesel and wastewater phases are separated using a continuous biphasic liquid separation and purification system in the second-stage purification step. The inlet and outlet tubes of the separator are made of 304 stainless steel with an outer diameter of 3.2 mm and an inner diameter of 1.59 mm. The helix wire inside the biphasic liquid separation system is also made of 304 stainless steel. For this study, two helix wire sizes were tested: one with a wire diameter of 0.2 mm and another with a diameter of 0.3 mm, both with an inner coil diameter of 1.7 mm. The connections between the stainless steel tubes and the helix wire were fabricated using custom-machined Teflon fittings. The overall structure of the biphasic liquid separation system, including collection funnels and supports, was fabricated using a 4K UV-curable 3D printer (Phrozen Sonic Mini 4K, Phrozen, Taiwan) with TR-300 resin (TR-300, Phrozen, Taiwan).

We investigated the effects of controlling the flow rate ratio and total flow rate of the soybean oil and methanol phases using the LC pump inputting into the low flow reactor. The total flow rates of the two-phase system were set at 1500 μL min⁻¹, 3000 μL min⁻¹, and 6000 μL min⁻¹, with three different stoichiometric ratios of methanol-to-oil (6, 9, and 11). Additional lengths of PTFE tubing were used to adjust the reaction times to 1.9 min and 3.9 min, allowing us to study the impact of reaction time on transesterification efficiency. The flow rate ratios corresponding to methanol-to-oil molar ratios of 6, 9, and 11 at different total flow rates and the corresponding reaction times and additional tubing lengths required can be found in Table S1.

The transesterification products include FAME, crude glycerol, residual methanol, and NaOH catalysts. Post-reactor, the mixture is routed through a T-junction component. A 150 mM hydrochloric acid aqueous solution, introduced at approximately one-third of the product flow rate, serves as the extractant, minimizing wastewater generation. The product and acidic solution form a complex mixed droplet flow, where FAME remains immiscible with the acidic aqueous phase. At the same time, crude glycerol, residual methanol, and NaOH have higher solubility in the aqueous phase, creating wastewater and completing continuous liquid–liquid extraction. The extraction time is controlled by connecting PTFE tubing with an outer diameter of 1/8 inch and an inner diameter of 1/16 inch downstream of the T-junction. The extraction retention times, calculated based on the total flow rate and tubing length, are 0.169 min, 0.084 min, and 0.042 min for flow rates of 2000 μL min⁻¹, 4000 μL min⁻¹, and 8000 μL min⁻¹, respectively.

After passing through the T-junction liquid–liquid extraction system, the output consists of two immiscible liquid phases: biodiesel and extracted wastewater (containing the acidic aqueous solution, unreacted methanol, and byproduct glycerol). These immiscible phases are continuously separated using an adjustable helix wire separator. The separated phases are collected at their respective outlets: biodiesel at outlet #1 and the extracted wastewater phase at outlet #2.

2.5 Evaluation of reaction and purification performance

2.5.1 Fatty acid methyl ester (FAME). The mixed product from the outlet of the Low Flow microreactor was collected and allowed to settle for 10 minutes to evaluate the transesterification efficiency at different flow rates and methanol-to-oil molar ratios. The upper and lower liquid layers were separately extracted and quenched by mixing with an equal 150 mM hydrochloric acid aqueous solution. Alternatively, biodiesel or wastewater samples were collected directly from the continuous extraction and separation system. This study mixed 4 mg of transesterified biodiesel FAME with methyl heptadecanoate (5 mM). This internal standard does not react with the experimental chemicals and is analyzed by gas chromatography (GC). The FAME was calculated using eqn (3):⁵⁹where ∑A is the total area of all fatty acid methyl esters detected by GC after transesterification; A_EI is the area of the GC peak corresponding to the internal standard; C_EI is the concentration of the internal standard (mg mL⁻¹); V_EI is the volume of solvent used to prepare the GC sample (mL); and W is the weight of FAME in the GC sample (mg).

The capillary column used for FAME analysis was model CP FFAP CB, with a total length of 25 m, an internal diameter of 0.32 mm, and a film thickness of 0.3 μm. The GC temperature program for FAME analysis involved initially setting the column temperature to 40 °C and holding for 0.2 minutes, followed by a temperature increase at a rate of 25 °C min⁻¹ to 100 °C, holding again for 0.2 minutes, and finally increasing the temperature at 10 °C min⁻¹ to 240 °C, where it was held for 4 minutes to complete the setup. The transesterified FAME sample was prepared by dissolving 0.004 g of FAME in 5 mL of hexane. A 5 mM solution of methyl heptadecanoate was added as the internal standard in the FAME–hexane mixture, which was then injected into the gas chromatograph for compositional analysis. Standards of each fatty acid methyl ester component from soybean oil were injected separately to confirm the identity of each peak in the GC chromatogram. Biodiesel comprises five primary FAME components, appearing in the chromatogram at the following retention times: methyl oleate at 13.35 minutes, methyl palmitate at 15.20 minutes, methyl stearate at 15.39 minutes, methyl linoleate at 15.81 minutes, and methyl linoleate at 16.37 minutes. The internal standard, methyl heptadecanoate, appears at 14.33 minutes. This detailed compositional analysis ensures the accurate determination of FAME and provides insight into the efficiency of the continuous flow green production process under various operational conditions.

2.5.2 Acid value. The effectiveness of the extraction was evaluated by measuring the acid value of FAME and the residual methanol content. To verify whether the extracted products were fully processed and neutralized, the acid value of FAME was determined using the phenolphthalein titration method.⁶⁰ A 0.1 M potassium hydroxide (KOH, Honeywell, USA) solution in ethanol (Ethanol, Kingming Chemical, Taiwan) was prepared for titration to neutralize the acidic FAME. Additionally, 0.12 g of benzoic acid (Thermo Scientific, USA) was dissolved in 50 mL of ethanol. Benzoic acid serves as a standard acid for calibrating the concentration of the KOH solution to ensure the titration results’ accuracy, with the KOH solution's normality calculated using eqn (4). Before preparing the benzoic acid solution, benzoic acid must be dried in an oven at 105 °C overnight to remove moisture. A phenolphthalein indicator solution was also prepared by dissolving 1 g of phenolphthalein in 60 mL of ethanol and diluting it to 100 mL with water.where g is the weight of benzoic acid (g), f is the normality factor of the KOH solution, and V is the volume of the KOH solution used in titration (mL).

Five grams of the sample were placed in a 250 mL Erlenmeyer flask to measure the acid value, followed by adding 100 mL of toluene (Macron, USA) and ethanol mixture in a 2 : 1 volume ratio. Approximately three drops of the phenolphthalein indicator were added using a plastic dropper, and the sample was titrated with the prepared 0.1 M KOH–ethanol solution until the solution turned light pink, indicating the endpoint of the acid–base titration. The titration volume was recorded, and the acid value was calculated using eqn (5):²

where V is the volume of KOH solution used in titration (mL); N is the normality of the KOH solution (M); f is the normality factor of the KOH solution; and G is the weight of the sample (g). These steps ensure an accurate assessment of the acid value, providing insights into the completeness of extraction and neutralization in the continuous flow process.

2.5.3 Separation purity. The effectiveness of the separation was evaluated using a mass fraction to assess the separation purity. Purity is defined as the weight fraction of the aqueous or organic liquid collected from the appropriate outlet, independent of its solubility in water. A purity value of 100% (ϑ₁) indicates that outlet #1 contains only biodiesel, with no wastewater present. However, complete phase separation is only achieved when both ϑ₁ and ϑ₂ are equal to 100%. The formulas used to calculate separation purity are as follows:where X_BD is the weight of biodiesel collected at the biodiesel outlet (outlet #1) in grams, and X_w is the weight of the extracted wastewater collected at the wastewater outlet (outlet #2) in grams. These calculations provide a quantitative measure of the separation purity of the continuous flow system, indicating how effectively the biodiesel and wastewater are separated in the process.

2.5.4 Methanol residual. The upper biodiesel layer was collected from the batch or continuous extraction liquids to measure the residual methanol concentration and diluted to a 1% volume concentration in isopropanol (2-propanol, IPA, Kingming Chemical, Taiwan). The diluted sample was then analyzed using gas chromatography (GC; Shimadzu GC 2014, Japan). The capillary column used for GC was model CP-WAX 57 CB, with a total length of 25 m, an internal diameter of 0.32 mm, and a film thickness of 0.3 μm. The GC analysis method involved initially setting the column temperature to 35 °C and holding it for 7 minutes. The temperature was then ramped up at a rate of 25 °C min⁻¹ to 120 °C and held for 5 minutes, followed by a final ramp at 5 °C min⁻¹ to 200 °C, where it was held for 15 minutes. After various extraction residence times (t_e), the residual methanol concentration in FAME was determined by comparing the GC results against a methanol concentration calibration curve prepared using the same GC method.

2.6. Life cycle assessment (LCA)

2.6.1 LCA approach for biodiesel production: continuous method vs. batch method. The last section of this study focuses on the life cycle assessment (LCA) of biodiesel production, comparing its impacts with traditional methods.⁶¹ It assessed the life cycle impact of two biodiesel production methodologies: continuous flow microreactor technology and the conventional system, using soybean oil as the feedstock, sodium hydroxide as the catalyst, and methanol as the alcohol. The continuous system features a continuous reactor for efficient transesterification, where the feedstock, catalyst, and alcohol are introduced simultaneously, leading to the effective separation of biodiesel using hydrochloric acid and deionized water in a core-annular liquid–liquid phase separator. This method is characterized by its streamlined approach and low energy consumption. In contrast, the conventional batch system involves several steps, including an initial glycerol separation, extensive washing, and dehydration to ensure biodiesel purity. While this traditional approach aims for high-quality output, it requires more processing stages and high energy consumption. This comparison reveals the continuous process's efficiency versus the conventional batch system's thoroughness in biodiesel production. In this impact assessment report, we evaluate the environmental performance of two products, scenario 1 (SC-1) and scenario 2 (SC-2), each weighing 1 kg.

The assessment was conducted using the SimaPro 9.1.1.1 software, applying the ReCiPe 2016 Midpoint (H) V1.04 methodology. The results provide a detailed characterization of the environmental impacts associated with each product. The comparative analysis highlights the strengths and weaknesses of both products across various impact categories, allowing for informed decision-making in product selection and life cycle management. This analysis aims to support sustainable practices and minimize ecological footprints within the relevant production sectors. Further exploration of the data will reveal specific impact areas that may benefit from optimization and improvement strategies.⁶²

2.6.2 Transesterification design considerations for linoleic acid biodiesel production. In this study, the continuous process for producing biodiesel from trilinolenin or 1,2,3-tri-linolenoyl glycerol or triglyceride (TG) of linoleic acid was evaluated at a small scale from the experimental result and scaled up for 1 kg of biodiesel. For LCA purposes, only the consumed methanol was considered, assuming residual methanol could be reused, and currently, we ignore the reused methanol impacts. The transesterification reaction proceeded according to the following equation:

Triglyceride + 3 methanol → glycerol + 3 linoleic acid methyl ester (FAME) (8)

One trilinolenin reacts with three methanol molecules, yielding one glycerol molecule and three methyl linoleate (linoleic acid methyl ester). The reaction mixture was maintained at 60 °C and stirred continuously for 10 minutes to ensure complete conversion of triglycerides to methyl esters (batch operation). Upon completion of the reaction, the mixture was allowed to cool and settle, forming two layers. The upper layer, comprising FAME or biodiesel (methyl linoleate), was carefully separated from the lower layer, which contained glycerol, unreacted methanol, and catalyst residues. The biodiesel layer was washed several times with warm deionized water to remove any remaining catalyst and impurities, followed by drying at 105 °C to eliminate residual moisture.⁶²

2.6.3 LCA comparison of biodiesel production processes from soybean oil. LCA was employed to access and compare two biodiesel production processes from soybean oil to identify a sustainable and environmentally responsible biofuel solution while emphasizing the efficiency of continuous flow micro-reactors technology by setting the functional unit as 1 kg of product biodiesel with gate-to-gate scope. The first scenario (SC-1) utilized mass balances based on experimental data for a continuous flow micro-reactor system. In contrast, the second scenario (SC-2) applied mass balances from published research⁶¹ for the conventional batch operation.

The LCA covered the entire life cycle of biodiesel production, from raw material acquisition to end-of-life disposal, carefully considering critical environmental indicators such as greenhouse gas emissions, energy consumption, and resource depletion. The system electricity consumption is measured using an AC voltage and current power meter (JL24WB, ATORCH, China). The detailed setup and schematic diagram (Fig. S1) of the electricity consumption measurement have been included in the SI to provide further clarity on the methodology used for SC-1. By systematically comparing the environmental impacts of both esterification pathways, the study aims to determine the most sustainable approach for biodiesel production, enabling data-driven decisions that promote environmental stewardship and resource conservation.

3. Results & discussion

3.1 Transesterification yield

We first investigated the transesterification yield under two different total inlet flow rates of 1500 μL min⁻¹ and 3000 μL min⁻¹, with a fixed transesterification reaction time of 1.9 minutes. Methanol (HPLC grade, REAGENTS DUKSAN, Korea), soybean oil (supplied by Taiwan Sugar Corporation, Taiwan), and 1 wt% NaOH as the alkaline catalyst (sodium hydroxide, Honeywell Fluka™, USA) were introduced into the Low Flow microreactor. Three different methanol-to-oil molar ratios (6 : 1, 9 : 1, and 11 : 1) were used for the transesterification reaction. A higher inlet flow rate required a longer tubing length to maintain the fixed reaction time.

As shown in Fig. 4, at a total flow rate of 1500 μL min⁻¹, the %FAME values for methanol-to-oil molar ratios of 6, 9 and 11 were 63.04 ± 4.06, 76.38 ± 1.30% and 75.37 ± 9.23%, respectively. These results indicate that the %FAME increased as the methanol-to-oil molar ratio increased. However, in the experiment conducted at a higher flow rate of 3000 μL min⁻¹, the %FAME for a methanol-to-oil molar ratio of 6 was 51.48 ± 3.53%; for a methanol-to-oil molar ratio of 9, it was 60.32 ± 0.73%; and for a methanol-to-oil molar ratio of 11, it was 51.17 ± 7.12%. Similarly, the %FAME increased with the methanol-to-oil molar ratio, but all %FAME values were lower than the results at the total flow rate of 1500 μL min⁻¹. Notably, the %FAME at a methanol-to-oil molar ratio of 11 decreased by 18.2%. This decline is likely because, at the higher flow rate of 3000 μL min⁻¹, the same reaction time of 1.9 minutes was maintained, necessitating the addition of a 1.5-meter length of 1/8′′ PFA tubing as a tube reactor to increase the volume by approximately 3.0 cm³ between chips. However, this longer external tube reactor increased pressure on the low flow microreactor, leading to suboptimal droplet splitting at the reactor inlet.

Fig. 4 (A) %FAME and (B) collected product photos of two different material input flow rates and methanol-to-oil molar ratios of 6 to 11. N ≥ 3.

3.1.1 Effect of adding a T-junction droplet generator on micro-scale transesterification. Subsequent experiments incorporated a T-junction droplet generator at the inlet of the microreactor to address the issue, as mentioned earlier. The T-junction allowed the two feed materials to form a two-phase droplet flow before entering the reactor, enhancing droplet splitting and improving the transesterification yield. In the experiment with a total flow rate of 3000 μL min⁻¹ and a methanol-to-oil molar ratio of 11, the %FAME increased to 86.57 ± 10.67%, representing an improvement of approximately 29.4% compared to the experiment without the T-junction droplet generator. Additionally, for methanol-to-oil molar ratios of 6 and 9, the %FAME values were 68.68 ± 8.97% and 79.44 ± 3.85%, respectively, showing increases of 17.2% and 19.12%. These results demonstrate that adding the T-junction droplet generator enhances droplet splitting during the initial phase of the reaction, providing a substantial benefit to the transesterification process.

Under the conditions where the T-junction droplet generator was added at the inlet and the reaction time was maintained at minutes, the total flow rate for transesterification was increased to 6000 μL min⁻¹. At this higher flow rate, the %FAME values for methanol-to-oil molar ratios of 6, 9 and 11 were 74.57 ± 9.15%, 73.94 ± 4.34%, and 81.33 ± 3.83%, respectively (Fig. 5). Notably, at the lowest methanol-to-oil molar ratio of 6, the %FAME improved from 51.48% to 74.57%—a difference of over 23.1%—compared to the low-flow-rate experiment without the T-junction droplet generator. This significant improvement is mainly attributed to the enhanced droplet-splitting effect of the T-junction droplet generator. At the higher flow rate of 6000 μL min⁻¹, the droplet splitting within the low flow chip was more pronounced than at the lower flow rate of 1500 μL min⁻¹, enabling the system to achieve higher transesterification efficiency even at low methanol-to-oil molar ratios.

Fig. 5 (A) %FAME and (B) collected product photos of the production process of adding a T-junction droplet generator before entering the reactor. N ≥ 3.

It is important to note that when the T-junction droplet generator was not added, the immiscible oil and methanol phases were still influenced by capillary forces and flow shear forces, forming unstable droplets. However, the selection of either the oil or methanol phase as the carrier fluid significantly affected the efficiency of droplet formation.^63,64 When soybean oil was used as the carrier fluid, it was expected to break up the methanol, causing it to become the dispersed phase. However, we observed that this arrangement failed to generate droplets in the reactor during the first few heart-shaped microstructures (Video S1). Instead, a parallel flow of the soybean oil and methanol was observed, and droplet formation did not occur until around the tenth heart-shaped microstructure, delaying the advantage of utilizing active circulation within and around the droplets to enhance oil–methanol mixing. When the inlet configuration was reversed, with methanol serving as the carrier fluid, droplet formation was observed in the first few heart-shaped microstructures. However, the exact position of droplet formation was inconsistent compared to the previous setup. These two configurations exhibited unstable mixing, making it challenging to define reaction times precisely. This instability led us to conclude that using the T-junction droplet generator upstream of the reactor was a more stable and optimized approach for enhancing the transesterification reaction. The generator effectively promoted splitting the two-phase liquids and improved mass transfer, ensuring a more reliable and efficient reaction.

Furthermore, in the experiment with a reaction time of 1.9 minutes and a total flow rate of 1500 μL min⁻¹, the slower flow rate resulted in less effective droplet splitting. Consequently, the glycerol byproduct produced after the reaction formed larger droplets, allowing more time for the glycerol to dissolve into the unreacted methanol and settle in the lower layer. However, in the transesterification experiments conducted at flow rates of 3000 μL min⁻¹ and 6000 μL min⁻¹ with the same reaction time of 1.9 minutes, the higher flow rates resulted in finer fragmentation of the glycerol byproduct. This made it less likely for the glycerol to dissolve into the unreacted methanol within a short time. As a result, when samples were collected and analyzed, the glycerol byproduct had not fully dissolved into the methanol. Instead, it remained as tiny suspended droplets floating in the upper FAME layer.

3.1.2 Increasing reaction time. We also found that increasing the reaction time from 1.9 minutes to 3.9 minutes had a minimal effect on the transesterification efficiency and, at lower flow rates, even showed a trend of decreased yield (Fig. 6). When the reaction time was extended to 3.9 minutes with a total transesterification flow rate of 1500 μL min⁻¹, the yields for methanol-to-oil molar ratios of 6, 9, and 11 were 44.90 ± 33.64%, 67.21 ± 10.09%, and 75.86 ± 8.48%, respectively. Compared to the reaction time of 1.9 minutes, the %FAME values decreased by 20.75%, 13.58%, and 9.32% on average. Similarly, when the reaction time was 3.9 minutes with a total flow rate of 3000 μL min⁻¹, the yields for methanol-to-oil molar ratios of 6, 9, and 11 were 64.91 ± 11.26%, 62.85 ± 15.5%, and 66.78 ± 21.01%, respectively, representing yield decreases of 3.77%, 16.59%, and 19.79% compared to the 1.9-minute reaction time. This decrease in %FAME may be due to the reverse reaction that can occur during extended reaction times, reducing the overall FAME production. At a flow rate of 6000 μL min⁻¹ and a reaction time of 3.9 minutes, the yields for methanol-to-oil molar ratios of 6, 9, and 11 were 71.25 ± 19.09%, 85.11 ± 9.42%, and 91.14 ± 13.85%, respectively. Compared to the 1.9-minute reaction time, the yield for the molar ratio 6 decreased by 3.32%, but the yields for methanol-to-oil molar ratios 9 and 11 increased by 11.17% and 9.81%, respectively. These results suggest that increasing the reaction time and molar ratio improves the %FAME at a higher transesterification flow rate of 6000 μL min⁻¹. However, the increase is not particularly significant.

Fig. 6 %FAME, input flow rates, and methanol-to-oil molar ratios as a function of residence times.

The highest fatty acid methyl ester content in this study was 91.14 ± 13.85%, which falls slightly short of the Taiwan CNS 15072 standard requirement of 96.5%, representing a 5.36% difference (Fig. 7A). We hypothesize that the inability to significantly improve the transesterification efficiency by increasing the reaction time at different flow rates may be due to the additional PFA tubing required to extend the reaction time between the Low Flow microreactor chips. The lack of structural features in the tubing to promote efficient mixing between the droplets may have limited the reaction to droplet diffusion and internal circulation convection, reducing the interfacial area and, consequently, the effectiveness of the transesterification reaction.

Fig. 7 Comparison of (A) %FAME and (B) acid values between this study and the CNS 15072 standard. GC chromatogram of the produced FAME with (C) 100 mm and (D) 1000 mm long extraction tubing. The pink area highlights the residual of methanol.

In summary, the factors influencing %FAME in continuous processes are (1) the effectiveness of methanol and feed oil mixing, which increases the surface-to-volume ratio between the droplets in the two-phase flow, and (2) providing sufficient and appropriate reaction time. If the process optimally satisfies these two variables, then (3) a higher methanol-to-oil ratio will result in higher transesterification efficiency. However, considering the relatively high cost of methanol, achieving similar high transesterification efficiency with less methanol would reduce the complexity of downstream purification and recovery steps.

3.2 Purification

3.2.1 Extraction and acid value. The extraction step involves connecting the reactor outlet to another T-junction component, where the product is mixed with a 150 mM hydrochloric acid aqueous solution. This creates a droplet flow with a high SA/V ratio, where biodiesel in the product remains immiscible with the acidic solution. At the same time, crude glycerol, residual methanol, and the NaOH catalyst have higher solubility in the hydrochloric acid solution, forming wastewater. This completes the continuous liquid–liquid extraction process. The flow rates of the hydrochloric acid solution were set at 500 μL min⁻¹, 1000 μL min⁻¹, and 2000 μL min⁻¹ to minimize wastewater generation, which is precisely one-third of the transesterification reaction flow rates. As a result, the total flow rates in the downstream purification stages became 2000 μL min⁻¹, 4000 μL min⁻¹, and 8000 μL min⁻¹, with the of 0.169 min, 0.084 min, and 0.042 min, respectively. After the T-junction liquid–liquid extraction system, the output consisted of two immiscible liquid phases: FAME and the extracted wastewater, which contained the acidic solution, unreacted methanol, and byproduct glycerol.

To ensure adequate extraction, the acid value of the FAME after extraction must meet the Taiwan CNS 15072 standard. We sampled FAME produced at a total downstream purification flow rate of 8000 μL min⁻¹ (t_e = 0.042 min), and the acid values of FAME produced from transesterification with methanol-to-oil molar ratios of 6, 9, and 11 were 0.214, 0.268, and 0.214 mg KOH per g, respectively (Fig. 7B). These values are well below the Taiwan CNS 15072 standard limit of 0.5 mg KOH per g, demonstrating that the T-junction liquid–liquid extraction system used in this experiment effectively reduced the acid value of FAME.

However, the residual concentration of methanol and its need for recovery are critical factors in evaluating whether the continuous flow process can be considered greener. In the product obtained with a methanol-to-oil molar ratio of 11 and an extraction flow rate of 8000 μL min⁻¹ after acid washing, a methanol peak was observed at 3.9 minutes in the GC chromatogram (Fig. 7C and D). The calculated residual methanol concentration was 0.8% (w/w), slightly higher than the 0.2% (w/w) limit specified by CNS 15072. We also tested shorter (100 mm) and longer (1000 mm) extraction tubing lengths, resulting in 0.042 minutes and 0.84 minutes, respectively, within the T-junction liquid–liquid droplet extractor. The corresponding methanol residual concentrations were 1.38% (w/w) and 2.94% (w/w), indicating no significant linear relationship between residence time and methanol concentration.

We hypothesize that while extending the t_e value within the extractor may slightly reduce the residual methanol concentration in biodiesel, in high methanol-to-oil ratio systems, an excessively long extraction residence time might lead to a reverse reaction between the wastewater and biodiesel, resulting in the precipitation of small amounts of methanol and thus increasing the residual concentration. This concentration would still exceed the CNS 15072 standard. Therefore, another potential approach for future exploration would be to increase the concentration of the hydrochloric acid aqueous solution used as the extractant, thereby enhancing the extraction efficiency.

3.2.2 Direct coupled biphasic flow separator leads to sustainable material production. The immiscible phases of biodiesel and wastewater were continuously separated using a two-phase flow separator, with each phase collected at its respective outlet. Biodiesel was collected at outlet #1, while the extracted wastewater phase was collected at outlet #2. The continuous two-phase flow separator employed here has been the subject of several foundational studies in our laboratory over the past few years.^54,55 Based on the machine learning predictions by Chang et al. (2023),⁵⁵ we selected a helix wire with a diameter of 0.3 mm, an original length of 60 mm, and stretched by 1 mm (with a gap width of 306.186 μm) for the continuous separation of FAME and wastewater.

The separation results at total flow rates of 2000 μL min⁻¹, 4000 μL min⁻¹, and 8000 μL min⁻¹ are shown in Fig. 8. At a total flow rate of 2000 μL min⁻¹, with methanol-to-oil molar ratios of 6, 9, and 11, the purity of FAME collected at outlet #1 (ϑ₁) was 89.59%, 88.02%, and 73.15% ± 6.94%, respectively, while the purity of wastewater collected at outlet #2 (ϑ₂) was 100%, 96.62%, and 83.08% ± 6.55%. At a total flow rate of 4000 μL min⁻¹, ϑ₁ values were 94.04%, 95.11%, and 80.23% ± 11.81%, and ϑ₂ values were 75.89%, 75.6%, and 74.25% ± 3.56% for the same molar ratios. For a total flow rate of 8000 μL min⁻¹ with methanol-to-oil molar ratios of 6, 9, and 11, ϑ₁ was 90.15%, 87.25%, and 74.28% ± 8.63%, and ϑ₂ was 86.45%, 82.65%, and 72.65% ± 7.99%, respectively. These data show that the best separation efficiency (ϑ₁ of 95.11%) was achieved at a total flow rate of 4000 μL min⁻¹ with a methanol-to-oil molar ratio of 9. Under the influence of both flow rate and methanol-to-oil molar ratio, the lowest ϑ₁ (73.15 ± 6.94%) was observed at the low and high molar ratios, though it maintained a separation efficiency above 70%. These experimental results demonstrate that the continuous two-phase flow separator developed in this study effectively separates FAME and wastewater, achieving high purity at both the FAME and the extracted wastewater outlets (Video S2).

Fig. 8 The separation results. (A) The purity plot of biodiesel and wastewater as a function of molar ratios and total flow rates. (B) Photos of the collected sample at outlets #1 (left-hand side bottle) and #2 (right-hand side bottle).

Beyond its high separation efficiency, the developed system provides added value by enabling the repurposing of separated streams. While FAME serves as biodiesel, the wastewater phase retains a significant quantity of crude glycerol. This crude glycerol can be further processed by adding castor oil at varying volume ratios, followed by the incorporation of 1,6-hexamethylene-diisocyanate dimer diurette (HDB) with an NCO content of 21.0%. The resulting reaction produces polyurethane (PU) foam, a versatile polymer with the elasticity of rubber, the strength of plastics, and exceptional processing capabilities (Fig. 9). PU foam finds applications across seven major fields: plastics, rubber, foam, fibers, coatings, adhesives, and functional polymers. The materials and methods for converting biodiesel-derived crude glycerol into PU are detailed in the SI (Table S2), demonstrating the broader utility and sustainability of the system.

Fig. 9 Increasing the proportion of crude glycerol can notably reduce the polyaddition reaction time of PU resin. (A) Appearance of PU foams. (B) Start of rise time, (C) end of rise time, and (D) tack free time of PU preparation.

3.3 LCA

3.3.1 Mass balances evaluation for LCA approach. In our analysis, data from the experiment were used as scenario 1 (SC-1) for the continuous flow micro-reactor system, and secondary data were referred to as scenario 2 (SC-2) for the conventional batch process (Saydut et al., 2016).⁶¹ The methanol-to-oil molar ratio is 11 for the continuous flow micro-reactor system but 6 in the conventional one. Higher methanol was needed to drive the reverse reaction to the product side, allowing better flow and enhancing suitable reactant mixing in the micro-reactor and traditional systems. The processes were designed according to the experimental steps, as shown in Fig. 10, and the production was scaled up to 1 kg of biodiesel in both scenarios. Table 1 shows the reacted soybean oil and methanol and their residues from the stoichiometric calculation, which are needed for mass balances.

Fig. 10 Process diagram for biodiesel production using (A) continuous flow micro-reactors, and (B) conventional batch system.

Table 1 The experimental conditions and data for biodiesel production from soybean oil of a continuous flow micro-reactor system (SC-1) and a conventional system (SC-2)

	SC-1	Note	SC-2	Note
Methanol-to-oil molar ratio	11	Experiment	6	Saydut et al. (2016)^61
Catalyst (NaOH) loading	1% in MeOH	Experiment	1%	Saydut et al. (2016)^61
Reaction temperature (°C)	60	Experiment	60	Saydut et al. (2016)^61
Reaction time (min)	10	Experiment	120	Saydut et al. (2016)^61
Oil conversion (%)	97	Experiment	98	Estimation
m (g)	3.4553	Experiment	300	Saydut et al. (2016)^61
m added methanol (g)	1.45	Experiment	68.46	Calculation
m reacted methanol (g)	0.39	Calculation	34.23	Calculation
m methanol residue (g)	1.05	Calculation	34.23	Calculation
m NaOH (g)	0.03	Experiment	3.00	Saydut et al. (2016)^61
m converted oil (g)	3.39	Calculation	294	Calculation
m unreacted oil (g)	0.07	Calculation	6.00	Calculation
m produced glycerol (g)	0.37	Calculation	32.385	Calculation
m produced FAME	3.555	Calculation	301.399	Calculation
% FAME	91.14%	Experiment	92.17%	Saydut et al. (2016)^61
Consumed electricity	0.0397 kWh	Experiment

---

As shown in Fig. 10A, the final product was 1 kg of biodiesel, achieving a FAME of 91.14%. The process generated byproducts, including 0.37 kg of glycerol and 1.96 kg of wastewater. Notably, the total energy consumption for this biodiesel production was measured at 0.186 kWh, underscoring the system's efficiency calculated based on the experimental consumption as follows.

Electricity per g per min = 0.0112 kWh

Electricity per kg per h = 0.1861 kWh

However, in the case of conventional processing (Fig. 10B), to produce 1 kg of biodiesel with 92.17% FAME, 294 kg and 32.385 kg of glycerol and wastewater were generated. Both scenarios used the same chemicals but in different amounts. Scenario 1 has a smaller reactor size than SC-2 (batch) due to its continuous process.

Table 2 outlines the inventory data of input and output of biodiesel production processes from soybean oil of SC-1 and SC-2. The functional unit is 1 kg of biodiesel product (output). Therefore, all data must be recalculated and normalized to reflect this reference amount. Establish system boundaries that include all stages, from raw material extraction through the transesterification and purification stage to waste management. SC-2 uses more water (input) in the washing step than SC-1, discharging more wastewater (output). As SC-1 and SC-2 are based on the same transesterification, the reacted soybean oil and methanol are the same amount due to the stoichiometric ratio. Assumptions were made for the complete reaction of soybean oil, and excess methanol was reused.

Table 2 The inventory data for biodiesel production from soybean oil of SC-1 (continuous flow micro-reactors) and SC-2 (batch)

Input				Output
Item	Unit	SC-1	SC-2	Item	Unit	SC-1	SC-2
From Nature				Product
-Water	kg	0.560	3.610	-Biodiesel	kg	1	1
From Techno-sphere (Materials)				Emission to water
-Reacted soybean oil	kg	0.972	0.972	-Glycerol residue in w/w	kg	0.104	0.104
-Reacted methanol	kg	0.111	0.111	-Moisture	kg	0.018	0.108
-NaOH	kg	0.010	0.010	-Wastewater	kg	0.534	3.491
-HCl	kg	0.003	—
Electricity consumption	kWh	1.316	3.35
-Reactor	kWh	0.186	1.22
-Dyhydration	kWh	1.13	1.13
Total mass	kg	1.656	4.703	Total mass	kg	1.656	4.703

---

Meanwhile, HCl was used in the microchannel process to neutralize the catalyst during washing. Scenario 2 in batch operation involved processing a significantly larger volume of reactants in a single batch, leading to increased energy consumption. The detailed estimation of the electricity consumption is as follows. Supposing the 50 L reactor consumes 1.1 kWh for its motor and works for 2 h of reaction retention time, the transesterification reaction for 2 h requires about 2.2 kWh. Similarly, 1.1 kWh would be required for 1 h of dehydration process. Details of the calculations are given in the SI (Table S3).

For the mass balance calculation, these molecular weights were used as follows:

MW_{linoleic acid TG} = 841.35 g mol⁻¹; 3(C₁₈H₃₂O₂)

MW_methanol = 32 g mol⁻¹; (CH₃OH)

MW_{linoleic acid methyl ester} = 294.5 g mol⁻¹; (C₁₉H₃₄O₂)

MW_glycerol = 92 g mol⁻¹; (C₃H₈O₃)

When comparing the two methods, the flow micro-reactor process stands out for its continuous operation and lower energy consumption, producing biodiesel steadily with a total energy usage of just 0.186 kWh. The conventional system, while effective, involved a batch process with more steps, including washing and dehydration, leading to higher energy demands. The traditional system consumed 3.35 kW (calculated) for transesterification, including dehydration. The biodiesel yield was comparable between the two methods. Still, the continuous flow micro-reactor process offered a more streamlined and energy-efficient approach, making it advantageous for large-scale biodiesel production.

3.3.2. Life cycle assessment and comparison. After finalizing the mass balance, the subsequent steps for conducting an LCA using SimaPro 9.1.1.1 were to compare the environmental impacts of the micro-channel continuous process with those of the conventional batch process. Next, perform a life cycle inventory (LCI) by compiling comprehensive data on all inputs (materials, energy) and outputs (emissions, waste) associated with the production of 1 kg of biodiesel for each process, where the inventory data were translated into potential environmental impacts. This comprised impact assessment methods, such as global warming potential and acidification potential, to evaluate critical environmental indicators like greenhouse gas emissions, energy use, water consumption, and potential toxicity for both the continuous flow micro-reactors and conventional batch processes. The results were implemented by comparing the environmental impacts of the two processes and identifying the specific impact categories where each process either excels or falls short. This comparison was performed for an assessment of the relative environmental performance of the micro-channel continuous process against the conventional batch process, emphasizing any significant differences as results in Table 3.

Table 3 The overall LCA comparison of 18 environmental impacts for SC-1 and SC-2

	Impact categories	SC-1	SC-2
1	Global warming (kg CO2 eq.)	7.79	8.05
2	Stratospheric ozone depletion (kg CFC11 eq.)	0.00	0.00
3	Ionizing radiation (kBq Co-60 eq.)	0.12	0.12
4	Ozone formation, human health (kg NOx eq.)	0.01	0.01
5	Fine particulate matter formation (kg PM2.5 eq.)	0.01	0.01
6	Ozone formation, terrestrial ecosystems (kg NOx eq.)	0.01	0.01
7	Terrestrial acidification (kg SO2 eq.)	0.01	0.01
8	Freshwater eutrophication (kg P eq.)	0.03	0.03
9	Marine eutrophication (kg N eq.)	0.00	0.00
10	Terrestrial ecotoxicity (kg 1,4-DCB)	11.20	11.29
11	Freshwater ecotoxicity (kg 1,4-DCB)	0.20	0.20
12	Marine ecotoxicity (kg 1,4-DCB)	0.24	0.24
13	Human carcinogenic toxicity (kg 1,4-DCB)	0.14	0.14
14	Human non-carcinogenic toxicity (kg 1,4-DCB)	2.58	2.60
15	Land use (m^2 a crop eq.)	7.16	7.17
16	Mineral resource scarcity (kg Cu eq.)	0.02	0.02
17	Fossil resource scarcity (kg oil eq.)	0.43	0.43
18	Water consumption (m^3)	0.18	0.28

---

The impact assessment results for products SC-1 and SC-2 revealed notable differences across various environmental impact categories. While both scenarios exhibited minimal contributions to stratospheric ozone depletion and maintained consistent levels of terrestrial acidification and eutrophication, SC-1 demonstrated a slightly lower global warming potential (7.795 kg CO2 eq.) compared to SC-2 (8.044 kg CO2 eq.). Although the differences are minimal, SC-1 showed marginally lower terrestrial and freshwater ecotoxicity values. A significant distinction arose in water consumption, where SC-2 required substantially more water (0.276 m³) than SC-1 (0.181 m³). These findings suggest that while both scenarios perform comparably across several impact categories, SC-1 might be favored for its lower carbon footprint and reduced water usage, highlighting the importance of selecting products that minimize environmental impacts in sustainable practices.

As shown in Table 3, the graph compares two biodiesel production processes, SC-1 and SC-2, using multiple environmental impact categories. SC-1, likely representing the continuous flow micro-reactor process, showed a slight advantage in most categories, particularly in reducing water consumption, fossil resource use, and ecosystem toxicity. Both processes perform similarly in terms of global warming potential and ozone formation. Overall, SC-1 demonstrated better environmental sustainability, making it a more efficient option for biodiesel production, especially regarding resource conservation and ecological impact. Individual comparisons with the significant 10 impact values gave the results shown in Fig. 11. The results show the most impact of terrestrial ecotoxicity in both scenarios. High terrestrial ecotoxicity showed the harmful effects of a chemical substance used in the soybean plantation and biodiesel process on terrestrial organisms and plants. Terrestrial ecotoxicity was about 1.4 times the global warming. At the same time, the impacts of global warming and land use were quite similar. Land use was the effect of soybean plantations, with the same value for SC-1 and SC-2 because of the same amount of soybean. It can be interpreted that the chemical used in this study (cradle to gate) is more relevant than the biodiesel conditions (methanol-to-oil molar ratio, reaction time, etc.). As a result, terrestrial ecotoxicity was the highest, and land use was the third impact category (almost the same as the second impact). Different continuous and batch operations affected the global warming category (Fig. 11A) and fossil resource scarcity (Fig. 11B).

Fig. 11 The LCA comparison for SC-1 and SC-2 of (A) global warming, terrestrial ecotoxicity, human non-carcinogenic toxicity, land use, (B) ionizing radiation, freshwater ecotoxicity, marine ecotoxicity, human carcinogenic toxicity, fossil resource scarcity, and water consumption.

While Table 3 reflects only a slight advantage of SC-1 across most categories when analyzed on a per-kilogram basis, scaling up to industrial production levels reveals much more pronounced differences. These scaled-up LCA simulation results, listed in the SI (Table S4), further emphasize the advantages of SC-1. At the same production capacity—for example, 1 million liters per day or approximately 1000 tons daily—SC-1 and SC-2 would use the same amount of raw materials. SC-2 requires a significantly larger reactor size because of batch operation. Despite the continuous conventional system, the size of reactors in SC-2 is still larger due to a longer residence time than SC-1. This larger reactor consumes more energy for heating and mixing than the smaller, more efficient continuous flow system (SC-1). Furthermore, SC-1 demonstrates a clear advantage in the efficiency of the washing and purification steps, using a much smaller volume of water during the washing process and producing substantially less wastewater. Specifically, at this production scale, SC-1 demonstrates a water consumption impact of 180 840 m³, compared to 276 542 m³ for SC-2, highlighting a significant improvement in resource efficiency. Water deficiency should be the disadvantage of SC-2 for input water, and well water management and water reclamation should be implemented for water reservation of the SC-2 process.

To holistically assess the impact of the proposed system, the experimental results from the continuous flow transesterification and purification units were integrated. The implementation of a T-junction-assisted microreactor significantly promotes a highly homogeneous mixture of methanol, soybean oil, and catalyst; excellent temperature distribution is maintained throughout the process, which enables a reduction in the reaction time from 120 minutes in batch mode to merely 10 minutes under continuous operation, as shown in Table 1. This intensified mass transfer, combined with the high surface-to-volume ratio inherent to microreactors, resulted in FAME yields exceeding 90% under optimized conditions. Downstream, the entire separation and purification workflow operates in a fully automated, continuous fashion. The development of a continuous biphasic liquid separation unit using a helix wire structure addressed the long-standing bottleneck of viscous phase disengagement. In this configuration, immiscible droplets exiting the reactor were passively separated by interfacial tension-driven mechanisms, wherein one phase was guided to the exterior of the coil while the other remained in the core. The separator maintained high phase purity (73–95%) across various flow conditions without requiring centrifugal force, membranes, or external fields. Furthermore, acid removal via integrated liquid–liquid extraction achieved acid values as low as 0.268 mg of KOH per g, which is well below standard fuel specifications. This ensures product quality and regulatory compliance. These results validate the technical viability of the full-flow system and suggest its strong potential for safe, compact, and modular deployment in distributed biodiesel production. In addition to improvements in reaction and separation efficiency, the overall system configuration, when evaluated at industrial scale, also demonstrated notable environmental advantages. This study also reveals interesting points. In the case of new equipment and system design, a reactor, separation, and purification system still significantly affect the environment. The new microreactor system design reduced global warming potential and terrestrial ecotoxicity, based on its performance and efficiency, when using the same type and amount of chemicals. The life cycle assessment results further validate the sustainability of the proposed continuous approach.

4. Conclusion

We successfully demonstrated a complete continuous flow production process for biodiesel by integrating a series of innovative downstream purification reactors for seamless extraction and biphasic liquid–liquid separation. Compared to traditional batch production, the continuous flow micro-reactor method achieves stable methanol–oil biphasic mixing, provides sufficient and appropriate reaction time and molar ratios, and effectively suppresses the occurrence of reverse reactions, resulting in improved %FAME. The biphasic continuous separator, seamlessly connected to the AFR reactor and extraction tubing, utilizes an adjustable helix wire design to achieve high-purity separation of biodiesel washed with acidic aqueous solutions from wastewater. This represents the first application of a non-gravitational settling, non-centrifugal, non-filtration, and non-membrane-based method in FAME continuous flow production. The continuous flow micro-reactor method produces FAME with yield, acid value, and residual methanol levels that are comparable to the Taiwan CNS 15072 standard. Future enhancements, such as increasing the number of reactors and incorporating additional distillation units, are expected to improve product yield further and facilitate the recovery of unreacted methanol, enhancing both efficiency and sustainability. Furthermore, the LCA analysis indicates that SC-1 was the better choice due to its innovative use of continuous flow micro-reactor technology, significantly enhancing efficiency while minimizing energy and water consumption. Therefore, SC-1 had a lower global warming impact category than SC-2 owing to less energy consumption and volume of wastewater. Despite utilizing the same feedstock, soybean oil, and methanol with sodium hydroxide, both scenarios had quite the same terrestrial ecotoxicity and land use impact categories. These results proved to be more environmentally friendly of newly designed continuous systems for the future production of biodiesel. While not explicitly discussed in this study, such a shift holds profound implications for reducing carbon emissions from raw material transportation, improving energy resilience, and enhancing disaster preparedness. This approach could provide a highly adaptive, mobile, and secure emergency power source if integrated with self-healing microgrids and microreactor-based biodiesel production systems.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the SI.

Supplementary information, including chemicals used in biodiesel production, electricity consumption measurement and calculation, the flow rate ratios correspond to molar ratios, material and method of the PU foam preparation, the inventory data for biodiesel production, the overall LCA comparison and continuous purification video are available. See DOI: https://doi.org/10.1039/d4gc06538g.

Acknowledgements

We thank all members (both NTU and NCHU sites) of Dr Ya-Yu Chiang's group for the useful discussions. Thanks to Dr Nutchapon Chiarasamran for the Simapro software and valuable discussion of the LCA investigation. This study was financially supported by the National Taiwan University Laurel Research Projects, Core Consortiums and Competitiveness Programs, the Taiwan National Science and Technology Council (NSTC) and the ENABLE (ENgineering in Agriculture Biotech LEadership) Project funded by Prof. Charles W. Tu. National Taiwan University Project numbers: 114L7731, 113L893802 and 114L892002; NSTC Project numbers: 111-2221-E-002-204-476 MY2, 111-2628-E-002-023-MY3, and 113-2221-E-002-098-MY3.

References

1. M. Tabatabaei, M. Aghbashlo, M. Dehhaghi, H. K. S. Panahi, A. Mollahosseini, M. Hosseini and M. M. Soufiyan, Reactor technologies for biodiesel production and processing: A review, Prog. Energy Combust. Sci., 2019, 74, 239–303,  DOI:10.1016/j.pecs.2019.06.001.
2. J. V. Gerpen, Biodiesel processing and production, Fuel Process. Technol., 2005, 86(10), 1097–1107,  DOI:10.1016/j.fuproc.2004.11.005.
3. P. T. Benavides and U. Diwekar, Optimal control of biodiesel production in a batch reactor: Part II: Stochastic control, Fuel, 2012, 94, 218–226,  DOI:10.1016/j.fuel.2011.08.033.
4. W. Du, Y. Xu and D. Liu, Lipase-catalysed transesterification of soya bean oil for biodiesel production during continuous batch operation, Biotechnol. Appl. Biochem., 2003, 38(2), 103–106,  DOI:10.1042/BA20030032.
5. C. Stavarache, M. Vinatoru, R. Nishimura and Y. Maeda, Fatty acids methyl esters from vegetable oil by means of ultrasonic energy, Ultrason. Sonochem., 2005, 12(5), 367–372,  DOI:10.1016/j.ultsonch.2004.04.001.
6. V. G. Gude and G. E. Grant, Biodiesel from waste cooking oils via direct sonication, Appl. Energy, 2013, 109, 135–144,  DOI:10.1016/j.apenergy.2013.04.002.
7. A. Nikseresht, A. Daniyali, M. Ali-Mohammadi, A. Afzalinia and A. Mirzaie, Ultrasound-assisted biodiesel production by a novel composite of Fe(III)-based MOF and phosphotangestic acid as efficient and reusable catalyst, Ultrason. Sonochem., 2017, 37, 203–207,  DOI:10.1016/j.ultsonch.2017.01.011.
8. P. Lidström, J. Tierney, B. Wathey and J. Westman, Microwave assisted organic synthesis—a review, Tetrahedron, 2001, 57(45), 9225–9283,  DOI:10.1016/S0040-4020(01)00906-1.
9. J. Yang, W.-j. Cong, Z. Zhu, Z.-d. Miao, Y.-T. Wang, M. Nelles and Z. Fang, Microwave-assisted one-step production of biodiesel from waste cooking oil by magnetic bifunctional SrO–ZnO/MOF catalyst, J. Cleaner Prod., 2023, 395, 136182,  DOI:10.1016/j.jclepro.2023.136182.
10. H. M. Khan, T. Iqbal, M. A. Mujtaba, M. E. M. Soudagar, I. Veza and I. M. R. Fattah, Microwave Assisted Biodiesel Production Using Heterogeneous Catalysts, Energies, 2021, 14(23), 8135 . Available: https://www.mdpi.com/1996-1073/14/23/8135.
11. S. Saka and D. Kusdiana, Biodiesel fuel from rapeseed oil as prepared in supercritical methanol, Fuel, 2001, 80(2), 225–231,  DOI:10.1016/S0016-2361(00)00083-1.
12. H. He, T. Wang and S. Zhu, Continuous production of biodiesel fuel from vegetable oil using supercritical methanol process, Fuel, 2007, 86(3), 442–447,  DOI:10.1016/j.fuel.2006.07.035.
13. K. T. Tan, M. M. Gui, K. T. Lee and A. R. Mohamed, An optimized study of methanol and ethanol in supercritical alcohol technology for biodiesel production, J. Supercrit. Fluids, 2010, 53(1), 82–87,  DOI:10.1016/j.supflu.2009.12.017.
14. A. Manz and J. C. T. Eijkel, Miniaturization and chip technology. What can we expect?, Pure Appl. Chem., 2001, 73(10), 1555–1561,  DOI:10.1351/pac200173101555.
15. F. Fanelli, G. Parisi, L. Degennaro and R. Luisi, Contribution of microreactor technology and flow chemistry to the development of green and sustainable synthesis, Beilstein J. Org. Chem., 2017, 13, 520–542,  DOI:10.3762/bjoc.13.51.
16. J. Sun, J. Ju, L. Ji, L. Zhang and N. Xu, Synthesis of Biodiesel in Capillary Microreactors, Ind. Eng. Chem. Res., 2008, 47(5), 1398–1403,  DOI:10.1021/ie070295q.
17. R. Gopi, V. Thangarasu, A. Vinayakaselvi M and A. Ramanathan, A critical review of recent advancements in continuous flow reactors and prominent integrated microreactors for biodiesel production, Renewable Sustainable Energy Rev., 2022, 154, 111869,  DOI:10.1016/j.rser.2021.111869.
18. I. Ambat, V. Srivastava and M. Sillanpää, Recent advancement in biodiesel production methodologies using various feedstock: A review, Renewable Sustainable Energy Rev., 2018, 90, 356–369,  DOI:10.1016/j.rser.2018.03.069.
19. J. Kobayashi, Y. Mori and S. Kobayashi, Multiphase Organic Synthesis in Microchannel Reactors, Chem. – Asian J., 2006, 1(1–2), 22–35,  DOI:10.1002/asia.200600058.
20. F. Reichmann, A. Tollkötter, S. Körner and N. Kockmann, Gas-liquid dispersion in micronozzles and microreactor design for high interfacial area, Chem. Eng. Sci., 2017, 169, 151–163,  DOI:10.1016/j.ces.2016.10.028.
21. M. Rahimi, B. Aghel, M. Alitabar, A. Sepahvand and H. R. Ghasempour, Optimization of biodiesel production from soybean oil in a microreactor, Energy Convers. Manage., 2014, 79, 599–605,  DOI:10.1016/j.enconman.2013.12.065.
22. G. S. Calabrese and S. Pissavini, From batch to continuous flow processing in chemicals manufacturing, AIChE J., 2011, 57(4), 828–834.
23. Z.-X. Yu, X.-L. Wang, Y.-L. Chang, C.-Y. Lin, Y.-C. Chang and Y.-Y. Chiang, A core-annular liquid–liquid microextractor for continuous processing, Chem. Eng. J., 2021, 405, 126677.
24. M. Yang, L. Yang, J. Zheng, N. Hondow, R. A. Bourne, T. Bailey, G. Irons, E. Sutherland, D. Lavric and K.-J. Wu, Mixing performance and continuous production of nanomaterials in an advanced-flow reactor, Chem. Eng. J., 2021, 412, 128565,  DOI:10.1016/j.cej.2021.128565.
25. H. Noureddini and D. Zhu, Kinetics of transesterification of soybean oil, J. Am. Oil Chem. Soc., 1997, 74(11), 1457–1463,  DOI:10.1007/s11746-997-0254-2.
26. Y. Li, B. Ye, J. Shen, Z. Tian, L. Wang, L. Zhu, T. Ma, D. Yang and F. Qiu, Optimization of biodiesel production process from soybean oil using the sodium potassium tartrate doped zirconia catalyst under Microwave Chemical Reactor, Bioresour. Technol., 2013, 137, 220–225,  DOI:10.1016/j.biortech.2013.03.126.
27. R. K. Elango, K. Sathiasivan, C. Muthukumaran, V. Thangavelu, M. Rajesh and K. Tamilarasan, Transesterification of castor oil for biodiesel production: Process optimization and characterization, Microchem. J., 2019, 145, 1162–1168,  DOI:10.1016/j.microc.2018.12.039.
28. M. Hosseini, A. M. Nikbakht and M. Tabatabaei, Biodiesel production in batch tank reactor equipped to helical ribbon-like agitator, Math. Models Methods Appl. Sci., 2012, 6, 40–45.
29. R. Metawea, T. Zewail, E.-S. El-Ashtoukhy, I. El Gheriany and H. Hamad, Process intensification of the transesterification of palm oil to biodiesel in a batch agitated vessel provided with mesh screen extended baffles, Energy, 2018, 158, 111–120,  DOI:10.1016/j.energy.2018.06.007.
30. I. Contreras-Andrade, E. Avella-Moreno, J. F. Sierra-Cantor, C. A. Guerrero-Fajardo and J. R. Sodre, Purification of glycerol from biodiesel production by sequential extraction monitored by ¹H NMR, Fuel Process. Technol., 2015, 132, 99–104.
31. L. Schlieper, M. Chatterjee, M. Henschke and A. Pfennig, Liquid–liquid phase separation in gravity settler with inclined plates, AIChE J., 2004, 50(4), 802–811,  DOI:10.1002/aic.10075.
32. A. Eggert, S. Sibirtsev, D. Menne and A. Jupke, Liquid–liquid centrifugal separation—New equipment for optical (photographic) evaluation at laboratory scale, Chem. Eng. Res. Des., 2017, 127, 170–179,  DOI:10.1016/j.cherd.2017.09.005.
33. Q. Liu, S. Liu, Y. Li, Y. Liu, N. Wang, Q. Yang and H. Lu, Microbubble-based process for the enhancement of microfine and heavy oil droplets swirl separation in axial inlet hydrocyclone, Sep. Purif. Technol., 2024, 332, 125642,  DOI:10.1016/j.seppur.2023.125642.
34. G. N. Kraai, F. van Zwol, B. Schuur, H. J. Heeres and J. G. de Vries, Two-Phase (Bio)Catalytic Reactions in a Table-Top Centrifugal Contact Separator, Angew. Chem., Int. Ed., 2008, 47(21), 3905–3908,  DOI:10.1002/anie.200705426.
35. S. Wang, M. Li and Q. Lu, Filter Paper with Selective Absorption and Separation of Liquids that Differ in Surface Tension, ACS Appl. Mater. Interfaces, 2010, 2(3), 677–683,  DOI:10.1021/am900704u.
36. M. J. Alves, S. M. Nascimento, I. G. Pereira, M. I. Martins, V. L. Cardoso and M. Reis, Biodiesel purification using micro and ultrafiltration membranes, Renewable Energy, 2013, 58, 15–20,  DOI:10.1016/j.renene.2013.02.035.
37. A. Adamo, P. L. Heider, N. Weeranoppanant and K. F. Jensen, Membrane-Based, Liquid–Liquid Separator with Integrated Pressure Control, Ind. Eng. Chem. Res., 2013, 52(31), 10802–10808,  DOI:10.1021/ie401180t.
38. M. C. S. Gomes, N. C. Pereira and S. T. D. d. Barros, Separation of biodiesel and glycerol using ceramic membranes, J. Membr. Sci., 2010, 352(1), 271–276,  DOI:10.1016/j.memsci.2010.02.030.
39. I. M. Atadashi, M. K. Aroua and A. A. Aziz, Biodiesel separation and purification: A review, Renewable Energy, 2011, 36(2), 437–443,  DOI:10.1016/j.renene.2010.07.019.
40. D. Dimitrijević, M. Bösenhofer and M. Harasek, Liquid–Liquid Phase Separation of Two Non-Dissolving Liquids—A Mini Review, Processes, 2023, 11(4), 1145 . Available: https://www.mdpi.com/2227-9717/11/4/1145.
41. W. Guo, H.-H. Ngo and J. Li, A mini-review on membrane fouling, Bioresour. Technol., 2012, 122, 27–34,  DOI:10.1016/j.biortech.2012.04.089.
42. J. F. Sanz Requena, A. C. Guimaraes, S. Quirós Alpera, E. Relea Gangas, S. Hernandez-Navarro, L. M. Navas Gracia, J. Martin-Gil and H. Fresneda Cuesta, Life Cycle Assessment (LCA) of the biofuel production process from sunflower oil, rapeseed oil and soybean oil, Fuel Process. Technol., 2011, 92(2), 190–199,  DOI:10.1016/j.fuproc.2010.03.004.
43. W. M. Antunes, C. d. O. Veloso and C. A. Henriques, Transesterification of soybean oil with methanol catalyzed by basic solids, Catal. Today, 2008, 133–135, 548–554,  DOI:10.1016/j.cattod.2007.12.055.
44. I. Noshadi, N. A. S. Amin and R. S. Parnas, Continuous production of biodiesel from waste cooking oil in a reactive distillation column catalyzed by solid heteropolyacid: Optimization using response surface methodology (RSM), Fuel, 2012, 94, 156–164,  DOI:10.1016/j.fuel.2011.10.018.
45. R. Mythili, P. Venkatachalam, P. Subramanian and D. Uma, Recovery of side streams in biodiesel production process, Fuel, 2014, 117, 103–108,  DOI:10.1016/j.fuel.2013.09.008.
46. Y. Wang, S. Ou, P. Liu, F. Xue and S. Tang, Comparison of two different processes to synthesize biodiesel by waste cooking oil, J. Mol. Catal. A: Chem., 2006, 252(1), 107–112,  DOI:10.1016/j.molcata.2006.02.047.
47. D. Y. C. Leung and Y. Guo, Transesterification of neat and used frying oil: Optimization for biodiesel production, Fuel Process. Technol., 2006, 87(10), 883–890,  DOI:10.1016/j.fuproc.2006.06.003.
48. K. Narasimharao, A. Lee and K. Wilson, Catalysts in Production of Biodiesel: A Review, J. Biobased Mater. Bioenergy, 2007, 1(1), 19–30,  DOI:10.1166/jbmb.2007.1976.
49. M. J. Nieves-Remacha, A. A. Kulkarni and K. F. Jensen, Hydrodynamics of Liquid–Liquid Dispersion in an Advanced-Flow Reactor, Ind. Eng. Chem. Res., 2012, 51(50), 16251–16262,  DOI:10.1021/ie301821k.
50. X. Wang, Y. Wang, F. Li, L. Li, X. Ge, S. Zhang and T. Qiu, Scale-up of microreactor: Effects of hydrodynamic diameter on liquid–liquid flow and mass transfer, Chem. Eng. Sci., 2020, 226, 115838,  DOI:10.1016/j.ces.2020.115838.
51. K.-J. Wu, V. Nappo and S. Kuhn, Hydrodynamic Study of Single- and Two-Phase Flow in an Advanced-Flow Reactor, Ind. Eng. Chem. Res., 2015, 54(30), 7554–7564,  DOI:10.1021/acs.iecr.5b01444.
52. J. P. Brody and P. Yager, Diffusion-based extraction in a microfabricated device, Sens. Actuators, A, 1997, 58(1), 13–18,  DOI:10.1016/S0924-4247(97)80219-1.
53. C.-Y. Lin, Y.-Y. Chen, P.-Y. Chen, M.-C. Chen, T.-F. Su and Y.-Y. Chiang, Scale-out production in core-annular liquid–liquid microextractor, J. Flow Chem., 2021, 11, 569–577.
54. X.-L. Wang, C.-H. Ni, J.-W. Chang and Y.-Y. Chiang, Continuous low surface tension and high viscosity two phases liquid separation, Sens. Actuators, B, 2022, 365, 131957.
55. Y.-C. Chang, Y.-J. Chen, P.-Y. Chen, Y.-C. Chen, F. Maqbool, T.-Y. Ho and Y.-Y. Chiang, Machine Learning for Two-Phase Flow Separation in a Liquid–Liquid Interface Manipulation Separator, ACS Appl. Mater. Interfaces, 2023, 15(9), 12473–12484.
56. Y.-C. Chen, B.-C. Hsueh, G.-Y. Lu, Y.-H. Chien and Y.-Y. Chiang, Continuous high viscosity biphasic liquid separation, Sep. Purif. Technol., 2024, 343, 127111.
57. B. Oldach, Y.-Y. Chiang, L. Ben-Achour, T.-J. Chen and N. Kockmann, Performance of different microfluidic devices in continuous liquid-liquid separation, J. Flow Chem., 2024, 1–11.
58. B. Oldach, Y.-Y. Chiang, L. Ben-Achour, T.-J. Chen and N. Kockmann, Performance of different microfluidic devices in continuous liquid-liquid separation, J. Flow Chem., 2024, 14(3), 547–557.
59. P. D. Patil, V. G. Gude, H. K. Reddy, T. Muppaneni and S. Deng, Biodiesel production from waste cooking oil using sulfuric acid and microwave irradiation processes, J. Environ. Prot., 2012, 3(1), 107–113.
60. H. Becker and C. Gärtner, Polymer microfabrication methods for microfluidic analytical applications, Electrophoresis, 2000, 21(1), 12–26,  DOI:10.1002/(SICI)1522-2683(20000101)21:1<12::AID-ELPS12>3.0.CO;2-7.
61. A. Saydut, A. B. Kafadar, F. Aydin, S. Erdogan, C. Kaya and C. Hamamci, Effect of homogeneous alkaline catalyst type on biodiesel production from soybean [Glycine max (L.) Merrill] oil, Indian J. Biotechnol., 2016, 15, 596–600.
62. A. V. Miniakhmetova, O. I. Sergienko, V. S. Ilina, A. I. Lepeshkin and D. A. Baranenko, A comparative life cycle assessment of phytosterol and meadowsweet (Filipendula ulmaria) oncoprotective functional food ingredients, Food Biosci., 2022, 50, 102076,  DOI:10.1016/j.fbio.2022.102076.
63. K. S. Elvira, F. Gielen, S. S. H. Tsai and A. M. Nightingale, Materials and methods for droplet microfluidic device fabrication, Lab Chip, 2022, 22(5), 859–875,  10.1039/D1LC00836F.
64. A. Fergola, A. Ballesio, F. Frascella, L. Napione, M. Cocuzza and S. L. Marasso, Droplet Generation and Manipulation in Microfluidics: A Comprehensive Overview of Passive and Active Strategies, Biosensors, 2025, 15(6), 345 . Available: https://www.mdpi.com/2079-6374/15/6/345.

---