Comparative Carbon Footprint Analysis of Alumina-Based Multichannel and Hollow Fiber Ceramic Membranes for Microfiltration

Abstract

This study aims to compare the carbon footprints associated with the fabrication of two types of alumina-based tubular ceramic membranes used in microfiltration (MF): a multichannel membrane produced by extrusion and dip-coating, and an asymmetric hollow fiber membrane fabricated via phase inversion. The multichannel process involves two sintering steps but uses no organic solvents, whereas the phase-inversion method simplifies production through single-step shaping and sintering but requires organic solvents that increase environmental burdens. Using a functional unit of 1 m² effective membrane area, carbon emissions were quantified from raw material extraction to waste disposal. The results showed total emissions of 8.57 kg CO₂-eq/m² for the multichannel membrane and 10.67 kg CO₂-eq/m² for the hollow fiber membrane. Although the hollow fiber process consumed less energy, its extensive use of solvents, particularly NMP, led to significantly higher emissions. This study provides the first quantitative comparison of these two common ceramic membrane fabrication routes and underscores the importance of considering both energy use and solvent impacts when evaluating the environmental sustainability of membrane production. A sensitivity analysis further evaluated the influence of key parameters, including alumina emission factor, regional electricity carbon intensity, alumina recycling, and solvent substitution or NMP recycling. The analysis demonstrated that each factor could significantly influence the total carbon footprint and, under favorable conditions, narrow or even reverse the gap between the two fabrication routes. This study provides the first quantitative comparison of these two common ceramic membrane fabrication methods and highlights the importance of considering energy use, solvent impacts, and potential mitigation strategies when assessing the environmental sustainability of ceramic membrane production.

1. Introduction

Ceramic membrane technology is increasingly attracting attention across diverse environmental and industrial fields, especially in water treatment, owing to its excellent thermal, chemical, and mechanical stability under harsh conditions [1]. Ceramic membranes exhibit outstanding resistance to extreme conditions, such as high temperatures, high pressure, and strong acidic or alkaline environments, thus overcoming the physical and chemical limitations of polymeric membranes [2]. These advantageous properties have driven the widespread use of ceramic membranes in various membrane filtration technologies, including microfiltration (MF), ultrafiltration (UF), and nanofiltration (NF) [3]. Particularly in microfiltration applications, ceramic membranes are expanding their applicability to areas requiring long-term durability and reliability, such as wastewater treatment, food and beverage processing, petrochemical pretreatment, and membrane bioreactors (MBRs) [4].

Among the various ceramic materials, alumina (Al₂O₃) has been widely adopted as a representative ceramic membrane material due to its excellent sintering properties, stable pore size controllability, mechanical strength, and cost-effectiveness, making it suitable for commercial applications [5]. Alumina-based ceramic microfiltration membranes can be broadly categorized into two main structural types: multichannel tubular membranes, and hollow fiber membranes [6]. Each type exhibits distinct structural characteristics and fabrication methods.

Multichannel ceramic membranes are typically fabricated by mixing ceramic powders and binders with water to form a paste, which is then shaped into a robust porous tubular support through extrusion processes [7]. After drying, the support undergoes high-temperature sintering to create a mechanically strong and porous substrate. Subsequently, an additional microfiltration (MF) functional layer is coated onto the support using a water-based slurry via dip-coating, followed by a second sintering process. A notable advantage of this method is that it employs entirely aqueous systems without organic solvents throughout the entire fabrication process, significantly enhancing its environmental compatibility amidst increasingly stringent environmental regulations. Nevertheless, the requirement of two high-temperature sintering steps inevitably leads to higher energy consumption and longer manufacturing times [8].

In contrast, ceramic hollow fiber membranes fabricated via the phase-inversion method involve preparing a dope solution containing ceramic powder, polymeric binders, and organic solvents such as N-methyl-2-pyrrolidone (NMP) [9]. The dope solution is extruded through a wet spinning process, followed by phase inversion, resulting in asymmetric hollow fiber membranes characterized by a finger-like and sponge-like morphology. This process offers distinct advantages, such as simplified manufacturing due to a single shaping and sintering step, thus enhancing productivity and process efficiency [9]. However, the necessary use of organic solvents like NMP raises significant environmental and health concerns due to their poor biodegradability and potential emission of volatile organic compounds (VOCs). Consequently, recent trends include regulatory restrictions on NMP use and efforts to develop alternative, more sustainable solvents [10].

Although ceramic membranes are generally more expensive in terms of initial production costs compared to polymeric membranes, their excellent durability and extended lifetime significantly reduce long-term operational costs and environmental burdens, making them increasingly attractive from a sustainability perspective [11]. Conversely, conventional polymeric membrane manufacturing still depends heavily on organic solvents, presenting substantial environmental impacts and occupational safety challenges, despite recent attempts to introduce ‘green solvents’ to reduce their environmental footprint [12]. In this context, the extrusion and dip-coating manufacturing process of multichannel ceramic membranes, which entirely eliminates organic solvent usage, is inherently eco-friendly and sustainable [13].

Given these fundamental differences between the two ceramic membrane manufacturing methods, a quantitative evaluation and comparison of their environmental impacts from a life-cycle perspective is essential. However, a detailed life-cycle impact assessment (LCIA) comparing multichannel ceramic membranes fabricated via extrusion and dip-coating against hollow fiber ceramic membranes produced by phase-inversion methods remains lacking [14]. Specifically, detailed inventory analyses accounting for raw material use, sintering energy requirements, and waste treatment in these fabrication methods have not been sufficiently addressed in the existing literature.

Life-cycle assessment (LCA) has become an essential tool in environmental technology to quantitatively evaluate the sustainability of processes and products [15]. Its application to membrane separation processes is continuously increasing, particularly in the assessment of membrane bioreactors (MBRs) and reverse osmosis (RO) systems, where extensive studies have examined operational-phase impacts such as energy use and carbon emissions [16]. However, most LCA studies in the membrane field have concentrated primarily on operational stages, while evaluations focusing on membrane material production and manufacturing phases remain relatively limited. The complexity associated with the diverse chemicals and processes involved in membrane fabrication further contributes to the paucity of studies addressing this manufacturing stage.

Recently, several studies have started to fill this gap by investigating the environmental impacts explicitly associated with membrane manufacturing. For instance, Yadav et al. [17] performed an LCA on hollow fiber polymeric membranes and identified the use of toxic organic solvents, such as NMP, DMF, and DMAc, as the primary contributor to environmental burdens. They reported that substituting these polar aprotic solvents with green alternatives could reduce total environmental impacts by up to 35%, emphasizing that solvent toxicity is a critical barrier to achieving environmental sustainability in polymeric membrane manufacturing.

Similar efforts have been extended to ceramic membranes. Kanth et al. [18] conducted an LCA on ceramic membranes derived from Fuller’s earth clay, revealing that energy consumption, mainly associated with high-temperature sintering, dominated the overall carbon emissions of the ceramic membrane manufacturing process. They demonstrated that transitioning from fossil fuels to renewable energy sources could potentially reduce these impacts by approximately 97%. While ceramic membranes require significant initial energy input, their superior durability and long service life—reportedly over 20 years, compared to approximately 6 years for polymeric membranes—can substantially reduce waste generation and associated environmental burdens over their full life cycle. Thus, the extended lifespan of ceramic membranes effectively offsets the high initial manufacturing impacts, highlighting their long-term environmental sustainability advantage.

Despite these insights, existing LCA studies have predominantly focused on comparing different materials or operational-phase impacts within the same membrane type, with minimal attention to comparing different fabrication routes for ceramic membranes [19]. To the best of our knowledge, a detailed comparative LCA of multichannel ceramic membranes fabricated via extrusion and dip-coating and hollow fiber ceramic membranes produced by phase-inversion methods has not yet been conducted.

The framework and principles for conducting LCA are standardized in ISO 14040 and ISO 14044. ISO 14040 outlines the general principles and framework for life-cycle assessment [20], whereas ISO 14044 specifies the requirements and guidelines for defining the goals and scope, compiling the life-cycle inventory (LCI), evaluating environmental impacts, and interpreting the results [21]. These standards provide an internationally recognized methodological basis for ensuring consistency, transparency, and comparability among different LCA studies. While a comprehensive life-cycle assessment can encompass a variety of environmental impact categories, such as water consumption, ecotoxicity, and resource depletion, this study focuses exclusively on global warming potential (GWP100), expressed as CO₂-equivalent emissions. Carbon emissions were selected as the primary metric because they are the most widely used and standardized indicator in environmental performance evaluations, and they directly align with ongoing industrial decarbonization and climate policy targets. Furthermore, reliable and consistent inventory data for other categories—particularly for ceramic membrane manufacturing—remain limited, which could introduce significant uncertainty. This targeted scope allows for a robust and quantitative comparison of the two fabrication routes, while recognizing that future work should expand the analysis to additional impact categories to provide a more comprehensive environmental profile.

Therefore, this study aims to address this research gap by quantitatively comparing the carbon footprints of two alumina-based ceramic membrane fabrication methods: a multichannel membrane produced via extrusion and dip-coating, and a hollow fiber membrane fabricated by phase inversion. Laboratory-scale experiments were conducted to obtain precise data on raw material use, energy consumption, and solvent requirements for both membrane types. Using a functional unit of 1 m² membrane area, this study quantifies carbon emissions from raw material extraction through membrane manufacturing and waste disposal. In addition, a sensitivity analysis was performed to evaluate the influence of key parameters—including alumina emission factor, regional electricity carbon intensity, alumina recycling, and solvent substitution or NMP recycling—on the total carbon footprint. The findings provide valuable insights into the relative environmental sustainability of these manufacturing methods and offer guidance for the development of environmentally friendly and resource-efficient ceramic membrane production strategies.

2. Materials and Methods

2.1. Functional Unit and System Boundary

In this study, a comparative carbon footprint assessment was conducted following the guidelines of the ISO 14040 and ISO 14044 standards to quantify and compare the environmental impacts associated with two representative fabrication methods for alumina-based ceramic membranes: the multichannel membrane produced via extrusion and dip-coating, and the asymmetric hollow fiber membrane manufactured via a phase-inversion process. The main objective of the analysis was to identify the environmental burdens associated specifically with the membrane fabrication phase. The analysis focused exclusively on the fabrication stage of the membranes, encompassing raw material extraction and processing, energy consumption during production, and the management of wastewater and hazardous chemicals generated throughout the manufacturing processes.

The functional unit selected for this study was defined as 1 m² of effective membrane area suitable for microfiltration applications. Given that multichannel and hollow fiber membranes possess different structural characteristics and fabrication processes, selecting a consistent functional unit allowed for an objective comparison of the environmental impacts between these two distinct membrane types on an equal basis. Thus, the environmental burdens were quantified and evaluated uniformly per unit membrane area, irrespective of structural differences.

The system boundary was established based on a cradle-to-gate approach, covering raw material extraction, processing, energy consumption during manufacturing, and waste treatment, including wastewater disposal and handling of hazardous materials resulting from the fabrication processes. Specifically, in the raw material phase, the analysis accounted for the production and supply of alumina powder, polymeric additives (such as methyl cellulose, polyvinyl alcohol (PVA), polyethersulfone (PESf), and polyvinylpyrrolidone (PVP)), solvents (water, isopropanol (IPA), N-methyl-2-pyrrolidone (NMP)), and other minor additives.

In the manufacturing phase, the processes involved in multichannel membrane production included paste extrusion, drying, support sintering, dip-coating of the MF layer, subsequent drying, and final sintering [22]. In contrast, the fabrication of hollow fiber membranes consisted of dope preparation, spinning, phase inversion, drying, and a single-step sintering process [23]. The energy consumption for each of these processes was individually quantified, with particular emphasis on accurately estimating the energy used for sintering, which varied depending on the sintering temperatures and the number of sintering steps.

Additionally, due to the significant environmental impacts associated with wastewater and organic solvent disposal (particularly NMP in the phase-inversion process), wastewater treatment and hazardous waste disposal processes were explicitly included within the system boundary.

Figure 1 provides a schematic overview of the two membrane fabrication processes assessed in this study. As illustrated in Figure 1a, the multichannel ceramic membrane manufacturing process involves the preparation of an alumina-based support paste, which is extruded to form a tubular support structure. High-purity α-Al₂O₃ powder (average particle size = 4.8 μm, AM-210, Sumitomo Chemical Co., Tokyo, Japan) was employed to prepare the ceramic support paste. The powder was first dry-blended with methyl cellulose (5 wt%, Sigma-Aldrich, St. Louis, MO, USA) as a binder, followed by the addition of deionized (DI) water (11 wt%) as the main solvent and polyethylene glycol (PEG, MW = 400, 1 wt%, Sigma-Aldrich, St. Louis, MO, USA) as a plasticizer. The mixture was allowed to equilibrate at ambient temperature for 48 h before being processed through a twin-screw extruder (KTE-50S, Kosentech, Gyeonggi-do, Republic of Korea). The resulting green bodies had a tubular geometry (outer diameter = 24 mm) containing 30 parallel channels (each = 2.7 mm inner diameter). These specimens were dried at room temperature for 24 h and subsequently sintered at 1500 °C for 2 h. After drying and initial sintering (support sintering), a porous support with sufficient mechanical strength was obtained. Subsequently, a water-based MF slurry containing alumina particles and polymeric binders was prepared and applied to the support via dip-coating. For the microfiltration (MF) coating layer, sub-micron α-Al₂O₃ powder (average particle size = 0.27 μm, AKP-30, Sumitomo Chemical Co., Tokyo, Japan) was dispersed in a mixture of 2-propanol (Sigma-Aldrich, St. Louis, MO, USA) and DI water, together with polyvinyl alcohol (PVA, MW = 500, Junsei Chemical, Tokyo, Japan) as an organic binder. The resulting slurry was applied to the inner channel surfaces of the tubular supports, which were masked on the exterior, using a laboratory-scale dip-coating unit (EF-4300, E-flex, Gyeonggi-do, Republic of Korea) for 60 s at a withdrawal rate of 1 mm/s. The coated membranes were air-dried for 24 h, followed by final sintering at 1300 °C for 1 h with a controlled heating rate of 5 °C/min. After another drying step, the membrane underwent a second sintering process to achieve a final multichannel ceramic membrane. Although this method requires two separate sintering steps, its major environmental advantage lies in completely avoiding the use of organic solvents throughout the entire fabrication process.

Figure 1b illustrates the phase-inversion spinning process for the hollow fiber ceramic membrane. Initially, a dope solution consisting of alumina particles, polymeric binders (such as PESf), dispersant (e.g., PVP), and the organic solvent NMP was prepared. Commercial α-Al₂O₃ powder (average particle size = 1.1 μm, Sumitomo Chemical, Tokyo, Japan) was employed as the ceramic precursor for hollow fiber membrane fabrication. Polyethersulfone (PESf, Ultrason^® E6020P, BASF, Ludwigshafen, Germany) served as the polymer binder, while polyvinylpyrrolidone (PVP, Sigma-Aldrich, St. Louis, MO, USA) acted as a dispersant. N-methyl-2-pyrrolidone (NMP, 99.5%, Samchun Pure Chemical Co., Ltd., Gyeonggi-do, Republic of Korea) was used as the organic solvent. The formulation of the spinning dope was prepared with an Al₂O₃:NMP:PESf:PVP weight ratio of 120:67:12:1. The raw materials were blended under continuous mechanical stirring for approximately 36 h at room temperature to obtain a homogeneous suspension. The resulting dope solution was subsequently degassed under vacuum for 1 h to eliminate air bubbles generated during mixing. The dope solution was then extruded through a spinneret into a coagulation bath (typically water), causing rapid phase inversion and solidification of the fiber, resulting in a characteristic asymmetric structure featuring a finger-like macrovoid interior and a sponge-like porous exterior. For spinning, the dope solution was fed from a pressurized reservoir using N₂ gas, while ultrapure water (10 mL/min) was simultaneously introduced as the bore fluid through a syringe pump (Fusion 100, Chemyx, Stafford, TX, USA). The hollow fiber extrusion passed through an air gap of approximately 10 cm before entering the external coagulation bath (tap water). Fibers were maintained in the bath for 24 h to enable solvent–non-solvent exchange and complete phase inversion. The as-spun fibers were thoroughly rinsed with water to remove residual solvent, and then they were dried in an oven at 100 °C overnight. Finally, sintering was carried out in a muffle furnace at 1450 °C for 2 h to produce mechanically robust ceramic hollow fibers. This fabrication approach simplifies the process by requiring only one sintering step but mandates the use of considerable amounts of the hazardous organic solvent NMP, thereby increasing the potential environmental and health impacts.

Thus, the two membrane fabrication processes, despite using the same base material (alumina), exhibit distinct environmental profiles in terms of material usage, energy consumption, and waste generation. This study aims to quantitatively evaluate how these process differences influence the overall environmental sustainability of ceramic membrane production, thereby providing practical insights for environmentally preferable ceramic membrane fabrication methods.

2.2. Membrane Fabrication

These two fabrication processes not only influence the resulting pore structure and mechanical properties of the membranes but also differ significantly regarding raw material usage, energy consumption, and environmental impacts, all of which have been thoroughly analyzed in this study.

Table 1. Raw material consumption and sintering condition per unit membrane area (1 m²).

Category	Material	Multichannel (g/m2)	Hollow Fiber (g/m2)
Ceramic	Alumina	2711	1159
Solvent	Water	921	16,100
	IPA	188.6	-
	NMP	-	647.2
Polymer	MC	214	-
	PVA	17.8	-
	PESf	-	115.9
	PVP	-	9.7
Sintering	Temperature (°C)	1500 (support) 1300 (MF)	1450

summarizes the consumption of raw materials and the sintering conditions required to produce a unit membrane area of 1 m². Raw material usage data were based on actual laboratory-scale production and normalized per square meter of effective membrane area. The area-based weights reported in Table 1 were determined from the mass of the membranes after sintering, when only alumina remained. During the sintering process, all polymeric binders and solvents are completely removed; therefore, the final mass corresponds solely to the alumina content. The dry mass of each sintered membrane was measured using an analytical balance (CAS, Gyeonggi-do, Republic of Korea) and divided by the total effective membrane surface area. For multichannel ceramic membranes, the effective surface area was calculated as the sum of the internal wall areas of all channels, based on the measured channel diameters and lengths. For hollow fiber membranes, it was calculated from the measured outer diameter and length of each fiber, multiplied by the total number of fibers. Since the formulation ratios of polymeric additives and solvents relative to alumina in the green body are fixed for each fabrication method, the corresponding initial masses of these organic components can be calculated in reverse from the measured alumina mass. This approach ensures that the calculated area-based weights inherently account for the contribution of all pore walls, as the effective membrane surface area calculation includes all internal channel surfaces for multichannel membranes and the full surface for hollow fiber membranes.

For the multichannel membrane, alumina usage was the highest at approximately 3073.1 g/m², which appears greater than that of the hollow fiber membrane due to its lower surface area per unit volume, rather than excessive material consumption. In contrast, the hollow fiber membrane has a much smaller diameter, leading to a higher surface area-to-volume ratio and, therefore, requiring less alumina (approximately 1159.2 g/m²) per unit membrane area. In the multichannel process, methyl cellulose (MC) was used as a binder, while PVA and isopropanol (IPA) were employed in the preparation of the MF dip-coating suspension to enhance the coating uniformity and particle adhesion. The entire fabrication process used only water and IPA as solvents, without involving any toxic organic solvents. In contrast, the hollow fiber membranes were fabricated via a phase-inversion process using NMP as the primary solvent, with approximately 647.2 g/m² consumed. The polymeric additives PESf and PVP were used at 115.9 g/m² and 9.7 g/m², respectively, to facilitate membrane formation and stabilize the pore structure. Although a large volume of water was used in the coagulation bath, this water was contaminated with NMP through solvent exchange during phase inversion, requiring subsequent wastewater treatment within the system boundary.

Regarding the sintering conditions, the multichannel ceramic membranes required two separate sintering steps: an initial sintering of the support layer at 1500 °C, followed by a second sintering step for the MF layer at 1300 °C. In contrast, hollow fiber membranes were manufactured through a single sintering step at 1450 °C. These differences in raw material consumption and sintering conditions directly impact energy consumption and associated carbon emissions, providing a critical basis for the environmental sustainability comparison of the two ceramic membrane fabrication routes.

2.3. Membrane Characterization

Various characterization methods were employed to evaluate and compare the structural properties and filtration performance of the fabricated multichannel and phase-inversion hollow fiber ceramic membranes.

First, the cross-sectional microstructures of the membranes were observed using scanning electron microscopy (SEM, SNE-4500 M, SEC Co., Ltd., Gyeonggi-do, Republic of Korea). For the multichannel ceramic membranes, cross-sectional samples were prepared by cutting the tubular membranes to expose both the inner and outer surfaces. These samples were coated with platinum prior to SEM analysis. Similarly, hollow fiber membranes were sectioned to obtain cross-sectional views under identical SEM conditions. The SEM images were used to qualitatively evaluate the porous structure, coating layer thickness, and structural asymmetry of each membrane type.

The pore size distributions of the membranes were quantitatively measured using a capillary flow porometer (CFP-1100, PMI, New York, NY, USA) with Galwick^® (surface tension = 15.9 dyn/cm) as the wetting liquid. Prior to measurement, the instrument was factory-calibrated by the manufacturer using certified reference membranes. In this analysis, membrane samples were mounted under identical conditions and saturated with a wetting liquid. Gas was then applied incrementally, and the pressure required for gas penetration through membrane pores was measured to determine the pore size and pore size distribution. The average pore sizes for both the multichannel and hollow fiber membranes were found to be in the range of approximately 100–150 nm, confirming their suitability for microfiltration (MF) applications. Pore size distribution curves were primarily used to compare the dominant pore size and the breadth of pore size distributions between the two membrane types.

Filtration performance, specifically water permeability, was assessed using a constant-pressure filtration apparatus (SepraTek, Daejeon, Republic of Korea). Membrane samples were mounted in a filtration cell, and pure water flux was measured at a constant applied pressure of 2 bar and at a controlled temperature of 25 °C. Permeate flux values, representing the amount of water permeating per unit membrane area per hour, were determined after initial stabilization of the flux rate. These measurements enabled direct comparison of water permeability between multichannel and hollow fiber ceramic membranes.

Through this comprehensive characterization approach, the structural similarities and differences between the two ceramic membranes, along with their fundamental filtration performance, were systematically analyzed. These results served as a crucial reference point for subsequent environmental impact evaluations.

2.4. Carbon Footprint Analysis

In this study, the environmental assessment is limited to global warming potential (GWP100), expressed as CO₂-equivalent emissions. This focus was chosen for two main reasons: (i) CO₂-equivalent emissions are the most standardized and widely reported indicator for environmental performance, enabling robust comparison between different manufacturing routes; and (ii) consistent and high-quality life-cycle inventory data for other environmental categories, such as water usage, ecotoxicity, and resource depletion, are scarce for ceramic membrane fabrication processes. Expanding the scope to multiple impact categories would require additional primary data collection and a broader inventory, which is beyond the scope of the present work. Nevertheless, the methodological framework established here can be readily extended in future studies to include additional LCA indicators for a more holistic evaluation.

In this study, an inventory analysis was conducted based on the raw materials and energy consumption involved in the fabrication processes of multichannel ceramic membranes and phase-inversion hollow fiber membranes. The inventory data were derived from actual laboratory-scale fabrication experiments and normalized per 1 m² of effective membrane area to enable a direct comparison of environmental impacts between the two different membrane manufacturing processes.

Table 1 summarizes the types and quantities of raw materials consumed in each fabrication route. For the multichannel ceramic membrane, the primary raw materials included alumina powder, methyl cellulose (MC), polyvinyl alcohol (PVA), isopropanol (IPA), and water, with alumina accounting for the largest portion as the main component of both the support and MF coating layers. In contrast, the hollow fiber membrane fabricated via the phase-inversion method utilized relatively smaller amounts of alumina but required substantial quantities of the organic solvent N-methyl-2-pyrrolidone (NMP) and polymeric additives, including polyethersulfone (PESf) and polyvinylpyrrolidone (PVP). The environmental impacts arising from raw material production and energy consumption were evaluated based on carbon emissions. The carbon emission factors for the raw materials employed are presented in

Table 2. Carbon emission factors of materials used in membrane fabrication.

Material/Energy	Carbon Emission Factor (kg CO2-eq/kg or kWh)	Note	Ref.
Alumina	(0.89~2.4)	Representative value: 1.65 (average)	[24]
	2.4	Coal-fired process	[25]
	0.89	Target reduction scenario
	1.72	Thermal-energy process	[26]
	1.44	Hydropower-based process
	1.52	Hydropower-based process	[27]
MC	3.69		[28]
PVA	2.47		[29]
PVP	7.5	Estimated from similar polymer emission factors (mid-value of 7–8)	[30]
PESf	7.96		[30]
Water	0.00091	Tap water	[31]
IPA	1.85		[32]
NMP	9.84		[33]
Electricity	0.473	Global average	[34]
	0.794	Average of 5 highest-intensity countries *
	0.042	Average of 5 lowest-intensity countries **

* Kosovo, Kazakhstan, Mongolia, South Africa, India; ** Norway, Switzerland, Sweden, France, Costa Rica.

.

These emission factors were obtained from the relevant literature and publicly available databases, reflecting cradle-to-gate values. For alumina (Al₂O₃), the literature reports a wide range of cradle-to-gate carbon emission factors, from 0.89 to 2.40 kg CO₂-eq/kg, depending on the production route, particle size, powder flowability, and the energy source used in the Bayer process [25]. For example, hydropower-based refining processes have been reported at the lower end of the range (~0.89 kg CO₂-eq/kg), while coal-fired processes can reach values as high as ~2.40 kg CO₂-eq/kg. Intermediate values, such as 1.44–1.72 kg CO₂-eq/kg, are associated with natural gas-based or thermal-energy Bayer processes [26]. In this study, we adopted a representative value of 1.65 kg CO₂-eq/kg, which lies within the reported range and reasonably reflects the average for commercially available Al₂O₃ powders. This representative value was applied to both membrane fabrication routes to ensure a consistent basis for comparison. To account for the variability in this parameter, a sensitivity analysis was conducted by applying the lower- and upper-bound values from the literature range. For NMP, the carbon emissions associated with wastewater treatment were also considered. The wastewater stream was assumed to consist of the water used as a coagulation bath and the residual NMP from the dope solution, which was treated by incineration [33]. For PVP, a direct cradle-to-gate carbon emission factor could not be found; therefore, we referred to the literature reporting emission factors of similar polymeric materials in the range of 7–8 kg CO₂-eq/kg and adopted the mid-value of 7.5 kg CO₂-eq/kg [30]. Since PVP was used only in trace amounts as a dispersant, contributing less than 0.1% to the total carbon footprint, the use of this approximate value was deemed acceptable. For electricity, the 2024 emission factor was used, and the global average value was applied for the base scenario [34]. In the sensitivity analysis, we also evaluated the impact of electricity emission factor variability by considering the average values from the five lowest-emission countries (Norway, Switzerland, Sweden, France, and Costa Rica; primarily renewable-based energy mixes) and the five highest-emission countries (Kosovo, Kazakhstan, Mongolia, South Africa, and India; fossil fuel-dominated energy mixes). This approach enabled us to assess the influence of extreme variations in electricity grid carbon intensity on the overall carbon footprint results.

The electricity consumption for the sintering process was estimated based on the rated power of the muffle furnace (28 kW, ST-1600, SentroTech, Strongsville, OH, USA; chamber volume ≈ 100 L) and the heating schedule for each firing cycle. For the support sintering process (1500 °C), the heating and holding durations were 295 min and 120 min, respectively. For the MF membrane (1300 °C), the heating and holding times were 255 min and 60 min, respectively, while for the hollow fiber membrane (1450 °C) they were 285 min and 60 min, respectively. All processes were followed by natural cooling without active heating.

While a simple nameplate-based calculation (rated power × total heating time) assumes continuous full-power operation and, thus, overestimates the actual consumption, the furnace typically operates at full power during the heating ramp and at a reduced duty cycle during the holding stage. In this study, we assumed 100% of rated power during the heating stage and 55% of rated power during the holding stage, which reflects typical operating conditions for laboratory-scale muffle furnaces.

The total electricity consumption (kWh) for each sintering cycle was calculated as follows:

E_total = (P_rated × t_heating) + (P_rated × 0.55 × t_holding)

(1)

where P_rated is the furnace’s rated power (kW), and t_heating and t_holding are the heating and holding times (h), respectively. Based on this approach, the total energy consumption per sintering cycle was approximately 606 MJ (168 kWh) for the support sintering, 484 MJ (134 kWh) for MF membrane sintering, and 534 MJ (148 kWh) for hollow fiber membrane sintering.

Under our loading conditions—less than 50% of the chamber volume filled per firing cycle—each sintering batch yielded approximately 51 m² of membrane area. The total energy consumption per unit membrane area (MJ/m²) was therefore obtained by dividing the total batch energy (MJ) by the processed membrane area (51 m²). Based on this approach, the specific energy consumption values were approximately 11.9 MJ/m² for the support sintering, 9.5 MJ/m² for MF membrane sintering, and 10.5 MJ/m² for hollow fiber membrane sintering.

Based on the inventory analysis and carbon emission factors, carbon emissions resulting from the consumption of raw materials and energy were calculated for each manufacturing process. This provided a quantitative basis for comparing the relative environmental impacts between the two ceramic membrane fabrication methods.

3. Results and Discussion

3.1. Membrane Structure and Performance

The geometric specifications of the multichannel ceramic membrane and the hollow fiber ceramic membrane are summarized in

Table 3. Geometric specifications of multichannel and hollow fiber ceramic membranes.

Specification	Unit	Multichannel	Hollow Fiber
Outer diameter	mm	24	2.7
Inner diameter (channel)	mm	2.7	1.9
Number of channels	ea	30	-
Length	m	1.0	0.5
Effective membrane area per piece	m2	0.2545	0.0030
Porosity	%	40	40

. These specifications were derived from actual measurements of the membranes fabricated in this study. These geometric differences are directly related to the distinct manufacturing processes of the two membranes—extrusion and dip-coating for the multichannel membrane, and phase-inversion spinning for the hollow fiber membrane—and critically influence both the mechanical characteristics and the environmental impact calculations in the subsequent LCA analysis.

The surface and cross-sectional morphologies of both the multichannel and phase-inversion hollow fiber ceramic membranes were analyzed using scanning electron microscopy (SEM), as presented in Figure 2.

Figure 2a shows the surface and cross-sectional SEM images of the multichannel ceramic membrane. The surface image clearly reveals a dense, homogeneous microfiltration (MF) layer characterized by uniform micropores of approximately 0.1 μm in size. In the cross-sectional image, a thin and uniform MF functional layer is evident on top of a highly porous alumina support, with a continuous and well-integrated interface between the two layers. The support itself comprises multiple channels within a tubular structure of several millimeters in outer diameter. These internal channels exhibit large macropores (several micrometers in size) optimized for high permeability. This structural arrangement results from the initial extrusion of the alumina support paste, followed by a dip-coating step and subsequent high-temperature sintering at 1450 °C. Notably, AKP-30 alumina powder was employed in the MF coating process, and identical sintering conditions were applied for both membranes to ensure consistency in manufacturing conditions.

Figure 2b illustrates the surface and cross-sectional SEM images of the hollow fiber membrane produced by the phase-inversion process. The surface morphology is quite similar to that of the multichannel membrane, showing uniform micropores with an approximate size of 0.1 μm. However, the cross-sectional image clearly displays a characteristic asymmetric structure formed by the phase-inversion method, comprising distinct finger-like macrovoids on the inner surface and transitioning gradually into a spongy outer region. This specific asymmetric structure results from differences in diffusion rates between the solvent (NMP) and the coagulation bath (water) during the phase-inversion process. The phase-inversion method thus provides the advantage of achieving high porosity and effective microfiltration performance in a single fabrication step.

In summary, both membranes exhibit similar surface pore structures suitable for MF applications, but they differ notably in their support structures and manufacturing routes. The phase-inversion method naturally produces an asymmetric hollow fiber structure in a single step, whereas the extrusion and dip-coating method requires separate steps for support fabrication and coating application, leading to distinct structural features and environmental implications.

The pore size distributions of the multichannel and hollow fiber ceramic membranes were quantitatively evaluated using capillary flow porometry (CFP), with the results shown in Figure 3.

Both membrane types exhibited distinct pore size distribution peaks within the range of approximately 100–150 nm, aligning well with the intended microfiltration (MF) application. Specifically, the multichannel membrane displayed a somewhat broader distribution, with additional minor peaks observed below 100 nm. This wider distribution is likely due to minor variations in shrinkage between layers during the multiple sintering steps, as well as subtle differences in the coating layer thickness resulting from the extrusion and dip-coating processes.

In contrast, the hollow fiber membrane exhibited a relatively sharp, single pore-size peak around 140 nm, reflecting a narrower and more uniform distribution. This characteristic can be attributed to the formation of a homogeneous skin layer after the single-step sintering in the phase-inversion process, despite the underlying finger-like macrovoid structure formed during the initial phase-inversion step. Overall, both membranes shared similar average pore sizes, but the hollow fiber membrane demonstrated a more uniform pore size distribution compared to the multichannel membrane. These findings underscore that both fabrication routes are well suited for producing effective microfiltration membranes, although they differ slightly in their pore structure uniformity due to specific fabrication processes.

Water permeability tests were performed on the multichannel and hollow fiber ceramic membranes at a constant applied pressure of 2 bar, using pure water, as shown in Figure 4.

Both membrane types exhibited similar water permeability characteristics, starting from an initial flux of approximately 450 LMH/bar and gradually stabilizing at around 400 LMH/bar after approximately 60 min of operation. The nearly identical permeability values between the two membranes, despite structural and manufacturing differences, indicate that both fabrication routes produce membranes with comparable filtration performance. This similarity in permeability correlates closely with the comparable average pore size (~0.15 μm) determined by the CFP analysis. It should be noted that performance equivalence between the two membrane types in this study was established based on their similar average pore sizes (~100–150 nm) and comparable pure water permeability (~400 LMH/bar), which are characteristic of MF-grade membranes. Other performance parameters such as particle rejection efficiency, fouling resistance, and long-term operational stability were not evaluated in this work, as the primary focus was to compare the environmental impacts of the fabrication processes. Future studies will incorporate these additional performance metrics to more comprehensively validate the functional equivalence of the two membrane types in practical applications.

The equivalent permeability performance between the two membrane types suggests that they would deliver similar operational effectiveness and lifespan under practical microfiltration conditions. Consequently, in this study, the comparative environmental assessment was performed solely based on differences in raw material consumption and energy usage during membrane fabrication, without considering operational or lifespan differences. However, it should be recognized that this assumption of identical filtration performance is a simplification; in real-world applications, differences in fouling resistance, cleaning frequency, or long-term durability could influence use-phase impacts. The assumption of identical filtration performance nevertheless provides a valid and practical basis for comparing environmental sustainability across these distinct ceramic membrane fabrication routes.

Both fabrication methods produce alumina-based membranes with similar average pore sizes and comparable microfiltration (MF) performance. Therefore, they can generally be applied in the same application fields, such as water and wastewater treatment, food and beverage processing, petrochemical pretreatment, and membrane bioreactor (MBR) systems. However, structural and manufacturing differences may influence their preferred market segments. Multichannel ceramic membranes offer superior mechanical strength and fouling resistance, making them well suited for long-term operation in industrial plants and under higher-pressure conditions. In contrast, hollow fiber membranes are lightweight, allow for high packing density, and can be more cost-effective in terms of initial fabrication and module assembly, which may be advantageous in space-limited installations. As a result, while there is significant overlap in potential applications, the two membrane types may be preferentially selected for different markets or operational conditions, depending on performance requirements, installation constraints, and cost considerations.

3.2. Carbon Footprint Comparison

The total carbon emissions for the multichannel ceramic membrane and the phase-inversion hollow fiber ceramic membrane fabrication processes are presented in Figure 5. Based on a functional unit of 1 m² of membrane area, the total carbon emissions were calculated to be 8.57 kg CO₂-eq for the multichannel membrane and 10.67 kg CO₂-eq for the hollow fiber membrane.

For the multichannel ceramic membrane, the primary contributors to the carbon footprint were alumina usage (4.47 kg CO₂-eq/m²) and energy consumption from the dual sintering processes (2.81 kg CO₂-eq/m²). In contrast, the contributions from polymeric additives and water-based solvents were comparatively minor. The phase-inversion hollow fiber membrane, on the other hand, exhibited lower emissions from sintering energy (1.38 kg CO₂-eq/m²) compared to the multichannel membrane, but it demonstrated significantly increased emissions (6.38 kg CO₂-eq/m²) arising from the production, use, and disposal of the organic solvent NMP. This highlights the substantial environmental impact associated with the use of organic solvents in the phase-inversion process.

Consequently, although the phase-inversion hollow fiber membrane process required less sintering energy, the overall carbon emissions were higher compared to the extrusion and dip-coating process, due primarily to solvent-related emissions. In addition to its environmental burden, NMP is recognized as a toxic solvent with potential health risks for workers involved in membrane fabrication. Occupational exposure to NMP, primarily through inhalation or dermal absorption, has been linked to respiratory irritation, skin disorders, and potential reproductive toxicity. While the present study focused solely on the environmental carbon footprint, the occupational health risks associated with NMP use should not be overlooked. Future studies should integrate worker exposure assessment and toxicity risk evaluation into the overall sustainability analysis of membrane manufacturing processes, particularly for those involving hazardous organic solvents such as NMP.

3.3. Sensitivity Analysis

To assess the robustness of the carbon footprint results, and to explore potential pathways for emission reduction, a sensitivity analysis was performed considering four key scenarios, as summarized in Figure 6.

(a) Variation in alumina emission factor.

The total carbon footprint of membrane fabrication was recalculated using the lower-bound (0.89 kg CO₂-eq/kg) and upper-bound (2.40 kg CO₂-eq/kg) values reported in the literature for α-Al₂O₃ production. These values reflect the variability associated with different production routes (e.g., hydropower-based vs. coal-fired Bayer processes). For both membrane types, increasing the alumina emission factor from the lower bound to the upper bound resulted in a substantial rise in total carbon emissions, with a more pronounced absolute impact on multichannel membranes due to their higher alumina content. This finding demonstrates that raw material sourcing can significantly influence the overall environmental impact.

(b) Variation in electricity emission factor.

The influence of regional electricity grid mixes was assessed by applying three different emission factors: (i) the global average for 2024 (base case), (ii) the average of the five lowest-emission countries—Norway, Switzerland, Sweden, France, and Costa Rica—characterized by predominantly renewable-based generation, and (iii) the average of the five highest-emission countries—Kosovo, Kazakhstan, Mongolia, South Africa, and India—characterized by fossil fuel-dominated generation. Across this range, total emissions varied significantly, with the largest changes observed for the multichannel membranes, where high-temperature dual sintering accounted for a major portion of the carbon footprint. This analysis highlights the importance of regional electricity carbon intensity in the life-cycle performance of ceramic membrane manufacturing.

(c) Membrane recycling (alumina recovery).

A scenario was considered in which end-of-life membranes are mechanically processed to recover alumina, reducing the need for virgin Al₂O₃ in subsequent membrane production. The recovered alumina was assumed to replace an equivalent mass of virgin material, thereby reducing the associated raw material carbon emissions. Based on the literature and industrial estimates, the electricity required for mechanical crushing and reprocessing was assumed to be approximately one-tenth of that required to produce virgin alumina. Under this assumption, the net emissions from the alumina component could be nearly eliminated, leading to notable reductions in the total carbon footprint—particularly for the multichannel membranes. This finding underscores the potential environmental benefits of implementing closed-loop recycling systems for ceramic membranes.

(d) Use of environmentally friendly solvents or NMP recycling.

In addition to NMP recycling, the substitution of NMP with environmentally friendly (green) solvents in the phase-inversion process was also considered. While such substitution could substantially reduce solvent-related emissions, several technical barriers remain for ceramic membrane fabrication. Green solvents such as dimethyl sulfoxide (DMSO), γ-valerolactone (GVL), and PolarClean differ significantly from NMP in terms of polarity, viscosity, and miscibility with the non-solvent phase. These differences can affect the kinetics of phase inversion, potentially altering pore morphology and membrane performance. Maintaining stable ceramic–polymer–solvent suspensions and controlling the rheology of the casting dope also become more challenging. Furthermore, limited commercial availability and higher costs may hinder large-scale implementation. Further research is therefore needed to optimize ceramic dope formulations and phase-inversion conditions specifically for green solvents, to ensure process stability and product quality comparable to those of NMP-based systems.

For hollow fiber fabrication, a substitution scenario was modeled in which the toxic aprotic solvent NMP is either replaced with a green alternative or recovered through a recycling process. The cradle-to-gate carbon emission factor of the green or recycled solvent was expressed as a percentage relative to that of NMP. Scenario analysis showed that when the solvent emission factor is reduced to ≤65% of the NMP value (i.e., a reduction of ≥35%), the total carbon footprint of the hollow fiber membrane fabrication can become lower than that of the multichannel membrane fabrication. However, if the reduction is smaller than this threshold, the high contribution of alumina production and sintering energy in the hollow fiber process means that the total footprint remains higher than that of the multichannel process. These findings highlight that substantial solvent emission reductions—through either green solvent adoption or high-efficiency NMP recycling—are necessary for hollow fiber membranes to achieve a net advantage in total carbon emissions.

Overall, the sensitivity analysis indicates that all four factors examined—alumina emission factor, electricity grid carbon intensity, alumina recycling, and solvent-related measures (green solvent substitution or high-efficiency NMP recycling)—can significantly influence the total carbon footprint of ceramic membrane fabrication. The magnitude of their impact varies depending on the specific conditions and assumptions applied; for example, in the hollow fiber process, solvent-related measures become particularly effective when the solvent emission factor is reduced to ≤65% of the NMP baseline (≥35% reduction). These results highlight that comprehensive emission-reduction strategies should address all of these factors in parallel, rather than focusing on a single parameter.

Although the present analysis is based on laboratory-scale data, scaling up to industrial production is expected to further improve the per-unit-area energy efficiency of sintering processes through higher furnace loading, improved insulation, and potential waste-heat recovery. For solvent-using processes such as hollow fiber fabrication, industrial-scale operations typically implement high-efficiency solvent recovery systems, which can substantially reduce solvent-related emissions. Such improvements would lower the absolute carbon footprint for both membrane types; however, the relative ranking observed in this study is expected to remain unchanged, as the fundamental differences in material requirements and processing steps persist.

4. Conclusions

This study performed a comparative carbon footprint assessment of two representative fabrication routes for alumina-based ceramic membranes used in microfiltration (MF): a multichannel membrane produced via extrusion and dip-coating, and a hollow fiber membrane fabricated by phase inversion. Both processes were experimentally implemented at the laboratory scale to quantify raw material consumption, energy usage, sintering conditions, and the amount of waste and solvents generated per functional unit of 1 m² membrane area, thus enabling a systematic comparison of their associated carbon footprints.

The analysis revealed that the multichannel ceramic membrane fabrication required substantial amounts of alumina and involved two separate high-temperature sintering steps, resulting in considerable manufacturing energy consumption. However, it offered significant environmental advantages by completely avoiding the use of organic solvents through reliance on an entirely aqueous-based system. Therefore, the total carbon footprint of the multichannel membrane was primarily driven by raw material (alumina) use and sintering energy, while environmental impacts associated with solvents and waste disposal remained comparatively low.

In contrast, the phase-inversion hollow fiber membrane manufacturing required lower alumina consumption and only one sintering step, thus reducing the manufacturing energy use. However, the mandatory use of the organic solvent NMP substantially increased its overall carbon emissions. The environmental impacts associated with the production and disposal of NMP emerged as the primary factor influencing the total carbon emissions. This finding underscores the limitations of evaluating the sustainability of ceramic membrane production based solely on manufacturing energy consumption and highlights the necessity of a comprehensive life-cycle approach that includes raw material characteristics, solvent usage, and waste management.

Sensitivity analysis further demonstrated that key parameters—such as alumina emission factor, regional electricity carbon intensity, alumina recycling, and solvent substitution or NMP recycling—can each substantially alter the total carbon footprint. Under favorable conditions, mitigation strategies like alumina recovery or solvent reduction can narrow or even reverse the footprint gap between the two fabrication routes.

Overall, the environmental sustainability of ceramic membrane manufacturing cannot be evaluated based solely on process simplicity or energy consumption. A holistic perspective considering raw material characteristics, solvent use, and full life-cycle carbon emissions is essential. This study represents the first quantitative, experimental-data-driven comparison of the carbon footprints of different ceramic membrane fabrication methods and provides practical guidance for sustainable decision-making in future ceramic membrane production and commercialization. Future work should expand the analysis to include operational and end-of-life phases and explore alternative materials and greener processing strategies to further reduce the environmental burden of ceramic membrane manufacturing.