A Brief Review of Nanocellulose Based Hybrid Membranes for CO₂ Separation

Abstract

Due to the high specific surface area, high mechanical strength and broad possibility of surface modification, nanocellulose has obtained much attention as a new class of bio-based nanomaterials with promising potential in a wide variety of applications. Recently, a considerable amount of research has been aimed to the fabrication of nanocellulose based hybrid membranes for water treatment. However, nanocellulose based hybrid gas separation membrane is still a new research area. Herein, we force on recent advancements in the fabrication methods and separation performances of nanocellulose-based hybrid membranes for CO₂ separation, the transport mechanisms involved, along with the challenges in the utilization of nanocellulose in membranes. Finally, some perspectives on future R&D of nanocellulose-based membranes for CO₂ separation are proposed.

1. Introduction

Nanocellulose is a class of nanomaterials (one or more dimensions between 1 and 100 nm) composed of the biopolymer cellulose. Like the biopolymer from which these nanomaterials are composed, nanocellulose are highly versatile; they have a very high tensile strength [1], high biodegradability, low toxicity, can be made biocompatible and can readily be modified chemically. The biopolymer has a long history in human endeavors. More recently scientific curiosity and financial interest have increased the exploration of nanostructures composed of cellulose (Figure 1). Casting a wide net in Scopus with the search term nano AND cellulos* we find 6449 results at the time of writing, and 132,109 patents. By the end of 2018 this was 6131 and 130,143 respectively, suggesting a continued rapid climb. The admittedly naive search term suggests a large and growing scientific and industrial interest in nanocellulose.

Various kinds of nanocellulose are produced, many of which are available commercially, or produced in a laboratory setting. These are commonly divided into three main groups: Cellulose nanofibrils (CNF), cellulose nanocrystals (CNC) or bacterial nanocellulose (BNC), each with sub-groups with a wide range of properties and applications. In the history of these nanomaterials various names have been and to some extent still are used in literature. In the current paper, we will adhere to nomenclature suggested by TAPPI and ISO, namely CNF, CNC, and BNC, respectively. This versatility has led to applications in a wide range of fields, including but not limited to drug delivery [2], tissue engineering [3], packaging films [4], nanocomposites [5], paper and board [6], rheology modifiers [7] and photonics [8].

1.1. Chemistry and Structure of Nanocellulose

Chemically, cellulose is a homopolysaccharide, specifically β (1→4) linked D-glucose units (Figure 2). Given that all substituents of the D-glucose monomer are equatorial, the repeating unit for cellulose is quite planar, or “flat”. The rotation around the glycosidic bond ensures the polymer itself is linear. The combination of planar monomers and linear polymers enables the polymer chains to bond to one another into larger structures like fibers. Interactions between polymers are through inter-chain hydrogen bonds (within the plane) and van-der-Waals interactions (out of the plane). This amphiphilic interaction as a basis for the insolubility of cellulose in water is termed the Lindman hypothesis [9,10,11].

1.2. Fabrication of Different Nanocelluloses

The most common sources of nanocellulose are plant matter. Nanocellulose, either CNF or CNC from plant matter are prepared by the disintegration of the source material. Various methods, chemical and/or mechanical, can be employed to produce these. This section will only briefly list the commonly used methods for nanocellulose preparation. Further details on nanocellulose production can be seen in a range of review papers and books focusing on this matter [15,16,17,18].

1.2.1. CNF

CNF, previously “microfibrillated cellulose” (MFC), was first reported in the early 1980s, then produced through high pressure homogenization of softwood pulp [19]. Since the 1980s a wide range of techniques have been applied to produce cellulose nanofibers. The primary means to produce CNF by mechanical means remain high pressure homogenization, though microfluidization and grinding are also common [16,17,20]. Other techniques are also applied, though they are less common [16,17,20]. Mechanical fibrillation is energy intensive and produces a morphologically heterogenous product (Figure 3a) [21]. Chemical pre-treatments such as TEMPO-mediated oxidation [22,23] can be used to produce fibrils with narrower size distributions, smaller diameters at substantially lower energy requirements (Figure 3b).

1.2.2. CNC

CNC is also referred to as nanocrystalline cellulose (NCC), crystallites, cellulose nano-whiskers (CNW), whiskers, or rodlike cellulose microcrystals [15]. CNC can be produced by attacking and removing areas with higher disorder. Classically this is done by acid hydrolysis, though other approaches including enzymatic hydrolysis or TEMPO mediated oxidation [25] can also be used [11].

The morphology of CNCs, primarily their length, depends on the hydrolysis time [26], acid concentration [27] and on the origin of the raw material [11].

1.2.3. BNC

Bacterial nanocellulose (BNC, also referred as bacterial cellulose (BC), microbial cellulose and biocellulose) has been known since 1886 [28], when BNC pellicles produced by Gluconobacter xylinus was first identified as cellulose. Since the initial discovery of BNC we have found that several bacterial genera produce cellulose as a part of their metabolic processes [29].

Fibrils produced by G. xylinus have a rectangular cross-section, and are frequently described as ribbons rather than fibrils [30]. Nanofibrils produced by different bacteria have different morphologies and properties [29]. BNC also has a higher degree of crystallinity than most plants [31,32], leading to a higher tensile strength for BNC.

1.3. Nanocellulose Applications

Since the last decade, nanocellulose has been used in abundant various applications ranging from composite reinforcement, food packaging materials, conductive materials to rheological modifiers. However, many of the proposed applications were not marketed because of the high energy consumption and production costs [33]. Currently, due to the fact that the isolation methods to produce nanocellulose are much less costly, the perspectives on the industrial production of nanocellulose have been changed. It has been applied in a wide range of fields, including but not limited to drug delivery [2], tissue engineering [3], packaging films [4], nanocomposites [5], paper and board [6], rheology modifiers [7] and photonics [8].

Nanocellulose has been widely studied as components of materials for food packaging. It has been found out that the use of nanocellulose can extend the food shelf life and also improve the food quality as they can work as carriers for some active substances, such as antioxidants and antimicrobials [34,35,36,37,38,39].

Nanocellulose can be easily fabricated into aerogels with high porosity, large surface area and low density; thus it has been proposed as adsorbents for different applications, including removing various pollutants from waste water, such as heavy metals, dissolved organic pollutants, dyes, oil and other undesired effluents [40].

The use of nanocellulose as reinforcement in nanocomposites has also become a popular research topic [1,41,42]. The tailorability, design flexibility and processability of nanocellulose–polymer composites permit broad utilization in the various applications, like automotive, electronics, biotechnology industries, etc [43].

Besides these applications mentioned above, nanocellulose also have been used in many novel applications such as rheological modifiers [44], food science [19,45], 3D-printing additives [46], printed electronics [47], sensing and biosensing [48], energy devices [49,50], drug delivery vehicles [2,51], and membrane separation [52,53].

2. Nanocellulose Hybrid Membranes

Membranes can be used to separate different mixtures by allowing some species to pass while the others are stopped. Permeability and selectivity are the two key parameters for a membrane. Developing membrane with both high permeability and selectivity is always desired to increase the efficiency of the membrane separation processes.

The combination of unique features of nanocellulose, such as high strength, chemical inertness, hydrophilic surface chemistry, and high specific surface area, makes it a very promising material for high-performance membranes. In this section, the different methods to fabricate nanocellulose based hybrid membranes are briefly introduced, and the separation applications of the membranes are summarized.

2.1. Nanocellulose Based Membrane Fabrication

There are various basic methods to fabricate membranes of different morphologies [54,55]. In the case of nanocellulose base membranes, several membrane fabrication methods have also been developed considering the special properties of the nanocellulose fibers and the membrane casting suspension, as illustrated in Figure 4 and summarized as follows:

2.1.1. Vacuum Filtration

Vacuum filtration is a fast, simple, and accessible process to produce layered structures of nanocellulose membranes and nanopapers followed by optional hot-pressing. The thickness and pore size of resultant membranes could be controlled by the amount and concentration of nanocellulose suspensions [56]. Different porous supports have been used in this method, including cellulose paper [57] and nylon filter membranes [56,58].

2.1.2. Casting Evaporation and Coating

Self-standing membranes for gas separation is normally prepared via casting evaporation of dilute nanocellulose suspension in a petri dish [59,60]. Generally, the nanocellulose dispersion needs to be diluted to a rather low concentration (heavily dependent on surface chemistry and fibril diameter, but generally <1 wt%) to reduce possible agglomeration [61,62].

Nanocellulose based thin-film-composite (TFC) membranes can be fabricated by knife casting method [59] or dip-coating method [62]. By controlling and optimizing the operation parameters, TFC membranes with a selective layer thickness of ~500 nm can be obtained [62].

2.1.3. Electrospinning

Membranes prepared from electrospining usually consist of lower basis weight, larger effective surface area, and higher effective porosity with continuously interconnected pores [63]. Incorporating nanocellulose inside the electrospinning membranes could change the membrane surface charge density, enhance the overall effective surface area and improve functional group density [64,65]. Moreover, the CNC content could affect the mean pore size and pore size distribution of this multilayered nanofibrous system, and thus the separation performance.

2.2. Nanocellulose Hybrid Membrane Applications

Cellulose fibers have a long tradition as filter materials dating back to ancient times [66]. In the past decades, nanocellulose has been intensively studied as membrane materials for water purification, possibly due to the inherent hydrophilicity of nanocellulose combined with tunable surface chemistry, which could be promising for reducing the bio-fouling or organic fouling [67,68,69]. Nanocellulose based membranes have been also applied in many water purification processes, such as bacteria removal [70], virus removal [58], and the removal of pollutants (e.g., heavy metal species and dyes) originated from either industrial or natural sources [52], and oil/water separation [56,71,72].

In addition, adding nanocellulose into polymer electrolyte (e.g., Nafion) could not only improve the mechanical and thermal stability, but also reduce water uptake plus area and volume swelling ratios in proton conducting membrane, which is one of the essential and critical materials in both proton exchange membrane fuel cell (PEMFC) and direct methanol fuel cell (DMFC). The cross-linked structure formed by the nanocellulose could also effectively reduce the undesired methanol crossover [73,74,75,76].

The nanocellulose based membrane has also been used in other process such as solar cells [77], membrane distillation (MD) [78], organic solvent nanofiltration (OSN) [79] and hemodialysis [80]. A new application of nanocellulose membranes for CO₂ separation have been emerged and attracted great attention during the recent five years; Promising separation performances of the membranes have been reported [59,62].

3. Nanocellulose Hybrid Membranes for Gas Separation

Membrane separation has been widely accepted as a promising alternative for traditional CO₂ separation technologies (e.g., absorption and/or adsorption) due to its high cost efficiency, high modularity, small footprint, lower or no chemical emissions, and ease of operation. However, both high selectivity and high gas permeability is required to make the gas separation membranes economically favorable in industrial applications [54,81].

As previously discussed, CNC and CNF differ in length as well as crystallinity (CNC is pre-dominant crystalline; CNF is frequently described as amorphous with crystalline regions). The advantages of CNC include uniform size with nanometric dimensions in both length and width. In certain applications such as high pressure gas separations, CNF with a higher aspect ratio and tensile strength might be desirable [82]. As mentioned in Section 1.2, CNF is often produced in a two-step process starting with a chemical/ enzymatic pre-treatment step followed by a mechanical step. In addition to improve the processability as well as making the size distribution more uniform, the pre-treatment step can contribute to nanocelluloses with different properties tailored different gas separation applications (e.g., introduction of CO₂ reactive groups) [82].

Based on the previous studies on nanocellulose, nanocellulose membranes have very low gas permeability, hence they are usually used as gas barrier materials, not for gas permeation applications [83]. In 2010, Minelli et al. investigated the O₂ and H₂O transport properties in CNF (labelled MFC by the authors) films [84], and found out that dry CNF films showed excellent oxygen and H₂O barrier properties. However, a 4~5 orders of magnitude increase in both O₂ and H₂O permeability was observed when the water content inside membranes was increased.

Wang et al. summarized studies of applying nanocellulose for gas barrier application [85]. Similar results were obtained for different nanocellulose: they could work as excellent oxygen barrier materials at the dry state and the performances are still good at low relative humidity (<65%), but they are a poor moisture barrier. In addition, the O₂ permeability increases 5 orders of magnitude as the relative humidity (RH) value increases from 0 to 60% (as shown in Figure 5). Further increasing the RH value may lead to even higher gas permeability.

3.1. Facilitated Transport Membranes

Inspired by the findings from Minelli et al., which concluded that the nanocellulose membrane characteristics could be changed from barrier to highly permeable by simply controlling the RH value, Ansaloni et al. firstly investigated the possibility of applying CNF as CO₂ separation membrane materials [59]. In their work, CO₂ transport properties of membranes fabricated from neat CNF and CNF/ polyvinylamine (Lupamin®) was studied under different RH values, ranging from ~50 to ~100. In the case of pure CNF membranes, similar to the trend of O₂ reported before, a dramatic increment was found for all the tested gases (i.e., CO₂, CH₄ and N₂), which shows at least 3 orders of magnitude enhancement as the RH increases from 50% to ~100%. Under optimized conditions, a CO₂/N₂ selectivity of over 500 can be obtained. However, the maximum CO₂ permeability value obtained from the pure CNF membrane is only about 25 Barrer, which is a rather low value compared to many commercially available polymeric membranes (e.g., polyimide, polyethylene glycol-based polymers).

On the other hand, mixing CNF into Lupamine® resulted in a clear enhancement of the permeation coefficient for all the three tested gases with respect to neat CNF films in the entire humidity range. However, due to high swelling and dilation of the matrix at high water content, a major increase of methane and nitrogen permeability was obtained, which resulted in a reduction of CO₂/N₂ and CO₂/CH₄ selectivity at high humidity conditions. The gas permeation results for CO₂/N₂ and CO₂/CH₄ separation are presented in Figure 6.

In another report from the same group [60], carboxymethylated CNF was employed to mix with Lupamin®, followed by a thermal treatment at 105 °C. Different from the neat CNF, in which the CO₂/CH₄ selectivity firstly increases and then decreases at high RH condition, both CO₂ permeability and CO₂/CH₄ selectivity increase with RH in carboxymethylated CNF membranes. In the case of carboxymethylated CNF/Lupamin® membranes, quite comparable results were obtained with respect to the neat CNF/Lupamin® membranes [86].

CNF was also employed in making nanocellulose/Lupamin® hybrid membranes [86]. Self-standing films with different nanocellulose content (from 30 to 70 wt%) were fabricated. It has been found out that the membrane permeability consistently increased with water vapor, and a higher content of Lupamin in the film led to an increment of both gas permeability and selectivity. The maximum selectivity (135 for CO₂/CH₄ and 218 for CO₂/N₂) was obtained at 60 RH%, while the maximum CO₂ permeability (187 Barrer) was observed at 80% RH.

Instead of mixing nanocellulose into a polymeric matrix, Zhang et al. developed nanocellulose free standing membranes by dissolving α-Cellulose in zinc chloride/calcium chloride solution [87]. Zn²⁺ ions reduce the pristine hydrogen bonds that normally bind the cellulose chains together, and Ca²⁺ ions facilitate interactions among the Zn²⁺–cellulose chains to form nanofibrils. It is claimed that the Zn²⁺ ion could work as fixed site carriers for CO₂ gas molecule. The highest CO₂ permeability (155.0 Barrer) was obtained from the membrane with the highest Zn²⁺ content (22.2%), with selectivity of 27.2 (CO₂/N₂) and 100.6 (CO₂/O₂).

In another study, BNC (labelled bacterial cellulose, BC, by the authors) was used as membranes for CO₂ separation purpose [88]. From the FTIR results, it has been found out that there is a strong interaction of CO₂ with the BNC membrane. The BNC membrane was further modified with silk fibroin protein as well as ZnO nanoparticles to improve the interaction between the CO₂ and the membrane. However, all the three membranes show rather low gas permeability, which is 2.73, 2.69 and 2.66 Barrer for the pristine, silk fibroin-modified and ZnO nanoparticles-modified BNC membrane, respectively. As the CO₂ is the only gas tested, there is no selectivity reported.

3.2. Nanocellulose/Polymeric Hybrid Membranes Based on Solution-Diffusion Mechanism

Uribe mixed nanocellulose with two commercially available polymeric membrane materials, which are cellulose acetate (CA) and poly ether imide (PEI) [89]. Even though high variability (>60%) was observed on the gas permeation test, it can be clearly observed that adding only 0.1wt% of CNFs into the polymeric matrix leads to a significant improvement on CO₂ permeance, for both CA and PEI membranes. However, the CO₂ permeance falls in the low region: the maximum CO₂ permeance obtained is about 2 GPU ($\mathrm{GPU}={10}^{-6}\frac{{\mathrm{cm}}_{\mathrm{STP}}^3}{{\mathrm{cm}}^2·\mathrm{s}·\mathrm{cmHg}}$) for CA/nanocellulose membranes and around 50 GPU for the PEI/nanocellulose membranes. In addition, the addition of CNF improved the permeance of both gases, at the cost of a decrease in CO₂/N₂ selectivity (reduced from ~20 to 5~10).

Jahan et al. mixed CNC with polyvinyl alcohol (PVA) and fabricated them into thin-film-composite (TFC) membranes for CO₂/CH₄ separation application [90]. It is found out that increasing CNC concentration of CNC in coating solution, the thickness of the selective layer (PVA + CNC) is also increased. So 1.5 wt% CNC is the optimized content in the PVA matrix, a higher or lower content resulted in reduced CO₂ permeance and CO₂/CH₄ selectivity. Under optimized conditions, a CO₂ permeance of ~100 GPU and a CO₂/CH₄ selectivity of 39 can be obtained. The same system has also been employed to separate CO₂/CH₄ at high feed pressure conditions [91]. By introducing small amount of CNC (1~2 wt%) into the PVA matrix, followed by adjusting the pH of the resultant solution, a CO₂ permeance of 107 GPU with a CO₂/CH₄ selectivity of 3 was obtained under optimized conditions. Due to the low CNC content in the membranes, it seems the effect of pH is more important than the CNC. Increasing feed pressure from 5 to 15 bar leads to a dramatic decrease in both CO₂ permeance and CO₂/CH₄ selectivity.

Torstensen et al. synthesized several different types of nanocellulose and used them for CO₂ capture from flue gas [82]. Three different in-house CNF’s are compared to commercially CNC and the main focus of the publication was investigating which properties of nanocellulose are important for gas separation membranes. Some of the different nanocellulose properties (different charges, size and functional groups) of CNFs obtained by different pre-treatment steps (TEMPO-mediated oxidation, phosphorylation) are discussed in this paper. The nanocellulose was mixed with PVA and then cast onto a flat sheet polysulfone ultrafiltration membrane. It is a screening study where the nanocellulose content in the hybrid membranes was maintained at 4 wt% as this concentration would not significantly change the suspension viscosity. The results from mixed gas separation testing show that a high nanocellulose charge and uniform nanoscopic size are favorable nanocellulose properties. A suggested transport mechanism for CO₂ in the membrane is suggested where CO₂ is transported along the water dense nanocellulose/PVA interfaces.

Very recently, Dai et al. employed CNC and CNF as nanofillers into PVA matrix. The hybrid membrane was fabricated onto a PPO (polyphenylene oxide) hollow fiber substrate [62]. Up to 80 wt% of nanocellulose was used in the hybrid membranes. By incorporating a large amount of nanocellulose into the PVA matrix, significant improvement can be obtained: increasing CNC content from 0 to 80 wt% resulted in a dramatic increase in CO₂ permeance, from ~400 GPU to ~700 GPU. The CO₂/N₂ selectivity is almost unchanged. On the other hand, adding CNF into PVA matrix only leads to moderate enhancement in CO₂ permeance and the increased selective layer thickness counteracts the improvement raised from the nanocellulose. In addition, the long term stability test was carried out in a period of one year, the membrane maintained good stability in the whole tested range.

3.3. Nanocellulose/Metal Organic Frameworks Hybrid Membranes

Metal-organic frameworks (MOFs) have been became as a new class of materials in membrane areas due to their high surface area, controllable pore size, and versatilities in group modification [92]. However, it is challenging to fabricate pure MOF membranes without defects and voids. Thus MOF particles are normally mixed into various polymeric matrices to form hybrid membranes with reasonable flexibility needed in membrane processing [93]. Recently, nanocellulose fibers have been also employed as the flexible polymeric phase to fabricate hybrid membranes containing different MOF particles.

Matsumoto et al. developed a ZIF-90/nanocellulose hybrid membrane by in-situ growth of the MOF particles on the nanocellulose fiber surfaces [94]. The nanocellulose fibers were treated via a TEMPO-mediated oxidation process. Thus, the nanofiber surface has a high density of carboxyl groups, which could form metal–carboxylate complexes with the metal ions in the MOF particles. The MOF/nanocellulose hybrid membranes were fabricated onto a paper filter, with a thickness in the range of 20~50 μm. Gas permeation properties of the resultant membrane were tested using a CO₂/CH₄ mixture (50:50 vol%) as the feed gas. As shown in Figure 7, the fabricated MOF/nanocellulose hybrid membranes exhibited ultrahigh CO₂ permeance, which is around 3000 GPU. If taking into account of the membrane thickness (~50 μm), the gas permeance is corresponding to a CO₂ permeability of ~15,000 Barrers, which is only slightly lower than the well-known the most permeable polymeric material in the world—poly(1-trimethylsilyl-1-propyne) (PTMSP) [95].

In terms of the CO₂/CH₄ selectivity, the MOF/nanocellulose hybrid membrane also exhibited a high but unusual trend: the selectivity increases with membrane thickness. The selectivity of the same membrane material can be thickness-dependent when the membrane selective layer thickness is lower than 1 μm [96]. However, increasing membrane thickness in micrometer range normally does not change the selectivity. In another work [94], a CO₂/CH₄ selectivity value of up to 123 was obtained with a membrane thickness of ~50 μm. The combination of the gas permeability and selectivity make this MOF/nanocellulose hybrid membrane extraordinarily promising for CO₂ separation applications.

Inspired by Reference [94], Zhang et al. also carried out research work on developing MOF/nanocellulose hybrid membranes [97]. In their work, UiO-66-NH₂ particles were used instead of the ZIF-90 particles, and nanocellulose fibers with –COOH group on the surface was employed. The MOF/nanocellulose hybrid membranes were fabricated by vacuum filtration of the MOF and nanocellulose dispersion through a MCE filter with micrometer-sized pores (0.22  µm). The resultant membrane shows a thickness of ca. 24  µm, and the thickness can be controlled by changing the volume of the dispersions. Gas permeation results reveal that, under optimized conditions, adding UIO-66-NH₂ particles into the nanocellulose matrix greatly improved both CO₂ permeability and CO₂/CH₄ selectivity. A CO₂ permeability of 139 Barrer and a CO₂/N₂ selectivity ratio of 46 was obtained, which is almost 20 times of the neat nanocellulose membranes. However, even though the materials used in Zhang’s report and Matsumoto’s report, the gas permeability data obtained from Reference [97] is 2 orders of magnitude lower than the value reported in Reference [94].

3.4. Comparison of Gas Separation Performances and Transport Mechanisms

Table 1. Comparison of CO₂ separation performances of different nanocellulose based hybrid membranes.

Material	CO2 Permeability (Barrer)	CO2/CH4 Selectivity	CO2/N2 Selectivity	Operation Condition	References
CNF	0.45	262.4	-	Single gas, 1 bar, 35 °C, 65% RH	[59]
	0.94	332.0	-	Single gas, 1 bar, 35 °C, 70% RH
	2.64	428.6	614.59	Single gas, 1 bar, 35 °C, 75% RH
	7.39	424.9	591.11	Single gas, 1 bar, 35 °C, 80% RH
	17.66	363.7	515.79	Single gas, 1 bar, 35 °C, 85% RH
CNF Lup	0.82	46.07		Single gas, 1 bar, 35 °C, 40% RH	[59]
	12.82	58.42		Single gas, 1 bar, 35 °C, 50% RH
	49.22	73.98	214.75	Single gas, 1 bar, 35 °C, 60% RH
	100.55	59.12	154.23	Single gas, 1 bar, 35 °C, 70% RH
	143.56	36.99	74.31	Single gas, 1 bar, 35 °C, 80% RH
	182.86	25.52	50.83	Single gas, 1 bar, 35 °C, 90% RH
CNF G2	0.03	3.01	-	Single gas, feed pressure of 1 bar, 35 °C	[60]
	0.09	6.08	-
	0.42	23.24	-
	3.42	143.32	-
	36.19	480.09	-
CNF/PVAm	1.31	263.54	-	Single gas, feed pressure of 1 bar, 35 °C	[60]
	12.88	432.00	-
	38.33	211.72	-
	83.33	111.73	-
	243.70	36.63	-
CNF pure	4.84	105.12	223.37	Single gas, feed pressure of 1 bar, 35 °C	[86]
	8.15	109.01	200.43
	23.13	93.41	100.00
	53.89	66.75	79.08
	127.07	48.13	64.25
L30NFC70	7.01	89.27	81.25	Single gas, feed pressure of 1 bar, 35 °C	[86]
	28.82	100.46	137.17
	34.30	62.64	98.21
	42.74	35.67	63.10
	79.00	20.31	32.93
L70NFC30	67.95	120.47	186.46	Single gas, feed pressure of 1 bar, 35 °C	[86]
	91.85	133.13	221.36
	118.53	105.12	195.07
	156.54	58.25	133.50
	209.16	29.21	86.55
CM-0	155	-	27.2	Single gas, feed pressure 2 bar at 25 °C	[87]
CM-10	93.5	-	17.4
CM-20	42.4	-	8.4
CM-30	39.3	-	8
basic BC membrane	2.73	-	-	Single gas, feed pressure of 480 Pa at room temperature	[88]
silk fibroin-modified BC membrane	2.69	-	-
ZnO nanoparticles-modified BC	2.66	-	-
PEI	2.48	-	36.37	Single gas, feed pressure 100 psig, room temperature	[89] *
PEI U 0.05%	6.41	-	7.03
PEIU 0.1%	9.31	-	14.48
PEIU 0.07%	9.65	-	15.57
PEIB0.05%	11.74	-	3.06
PEIB0.07%	31.69	-	31.48
PEIB 0.1%	50.21	-	9.91
CA	0.26	-	66.17
CAU 0.05%	0.32	-	14.35
CAU0.07%	0.30	-	67.38
CAU 0.1%	0.42	-	12.95
CAB 0.05%	0.39	-	62.93
CAB 0.07%	0.52	-	3.84
CAB 0.1%	1.95	-	1.51
PCNC 0.5%	69.92	31.18	-	Mixed gas, feed pressure of 5 bar at 100RH% and RT	[90] *
PCNC1%	92.60	33.98	-
PCNC 1.5%	100.28	39.16	-
PCNC 2%	86.06	32.02	-
PCNC 4%	77.60	29.08	-
PVA	105	-	36	Mixed gas, feed pressure of 1.2 bar, 100 RH% and 23 °C	[82] *
PVA/L-CNF	92.6	-	37.5
PVA/H-CNF	90.7	-	42
PVA/P-CNF	100	-	42
PVA/CNC	128	-	39
MOF-TOCN	3000	123	-	-	[94] *
CNF/UiO-66-NH2	139	-	46	Single gases, 25 °C	[97]
PVA	407.16	-	33.74	Mixed gas, feed pressure of 2 bar at 100 RH% and RT	[62] *
PVA 20CNC	480.99	-	39.94
PVA 40CNC	549.78	-	36.16
PVA 60CNC	616.20	-	42.84
PVA 80CNC	671.95	-	43.60
PVA 20CNF	408.46	-	41.93
PVA 40CNF	412.85	-	32.61
PVA 80CNF	470.79	-	34.57

* Gas permeance (GPU) was reported instead of gas permeability (Barrer).

listed all the CO₂ separation performances of various nanocellulose based hybrid membranes.

Based on Table 1, it can be clearly seen that in most of the reported nanocellulose based membranes, the selectivity of CO₂/N₂ and CO₂/CH₄ are extraordinarily high at humid state, especially when nanocellulose was mixed with facilitated membrane materials (e.g., PVAm) [59]. However, when mixed with polymers based on solution-diffusion mechanism, the presence of nanocellulose does not resulted in a significant change in selectivity (e.g., PVA [62]). Conversely, adding different nanocellulose into non-facilitated transport polymeric matrix normally resulted in a high CO₂ permeability/permeance, while the selectivity of CO₂ over other gases are almost unchanged [62,82].

The gas separation performances were also plotted in Robeson upper bound, as shown in Figure 8. For both cases, i.e, CO₂/N₂ and CO₂/CH₄ separation, extremely high selectivity can be obtained in pure nanocellulose membranes or nanocellulose based facilitated transport membranes [59,86]. However, for membranes with high selectivity, the CO₂ permeability were almost all always in the low region (<200 Barrer). In addition, the maximum selectivity were normally obtained in the RH range of 65% to 80%, further increasing the RH resulted in a higher CO₂ permeability but lower selectivity. Based on the gas transport results, a possible gas transport mechanism in nanocellulose based membranes were proposed as shown in Figure 9.

As discussed before, due to the low gas permeability at dry state or low RH condition, nanocellulose has been proposed as food packaging material. In the case of CO₂ separation application, nanocellulose based membrane normally show rather low gas permeability for both CO₂ and other gases (e.g., N₂) at dry state. The densely packed nanocellulose fibers work as gas barrier, which stopped most of the gas molecules from the feed side. When the RH was increased to a medium level (e.g., 60~80 %RH), the CO₂ gas permeability could be greatly improved, while the gas permeability of other gases (e.g., N₂ and CH₄) were still maintained at a rather low level, thus resulting in extraordinarily high CO₂ selectivity over other gases. In this phase, the polymer phase swells with the presence of water vapor, thus the distance between the nanocellulose slightly increased. Due to the size difference (CO₂ ~ 0.34 nm, N₂ ~ 0.36 nm and CH₄ ~ 0.38 nm), the nanocellulose worked as molecular sieve, the CO₂ molecules could diffuse through the space between the nanocellulose fibers whereas N₂ and CH₄ could not, thus resulted in high CO₂ selectivity. Further increase the RH value resulted in a higher swell degree of the polymeric matrix and consequently larger space between nanocellulose fibers. Therefore, these spaces are big enough for all the gases to pass through, causing the nanocellulose based membrane lose their high selectivity. At high RH conditions, the selectivity of the nanocellulose based membrane are normally dominated by the host polymeric matrix [62,86].

4. Conclusion and Perspectives

Nanocellulose holds many different unique properties, such as high mechanical strength and stiffness, high surface reactivity (with numerous hydroxyl groups), specific organization, and nano-sized dimensions. In addition, nanocellulose is abundant, renewable, nontoxic, and biodegradable. All these properties make it a promising material for a diverse range of applications such as paper and packaging products, adsorbents, additives, electronics, and membrane separation processes. The application of nanocellulose based membranes for waste water treatment has also been intensively studied. However, as described in this review, using nanocellulose hybrid membranes for gas separation is a relatively recent application.

According to aforementioned literatures, using nanocellulose in gas separation membranes are quite promising. The presence of nanocellulose in the hybrid membranes leads to either high selectivity or high permeability. In facilitated transport membranes (e.g., (Polyvinylamine, PVAm)), the presence of nanocellulose results in ultrahigh selectivity of CO₂ over other gases (e.g., CH₄ or N₂), while the improvement of gas permeability is not as effective. On the other hand, in nanocellulose hybrid polymeric membrane based on solution-diffusion model (e.g., PVA and PEI), the gas permeation can be greatly improved while the selectivity was maintained unchanged. Hybrid membranes fabricated from nanocellulose and (in)organic particles were also studied. Some of these membranes exhibited ultrahigh CO₂ permeance and CO₂/CH₄ selectivity simultaneously. However, big variations were obtained from membranes fabricated from similar materials. The gas transport mechanism inside these membranes should be further studied and the results should be more concrete.

We expect that the applications of nanocellulose in gas separation membranes will be a topic of intense study in the coming years. However, to make nanocellulose base membrane more competitive to other conventional polymeric membrane materials, some perspectives for future research are proposed:

(1) Fabrication of defect-free TFC membranes with feasible gas permeance is critical. Up to now, many of the reports are using self-standing membranes with a thickness of 50~100 µm. However, for practical considerations, to reach a sufficient gas flux through membranes and thus making the separation process economically feasible, membranes containing a defect-free selective layer with athickness ranging from 0.1 to 1.0 μm are usually required;

(2) Work should be done on understanding the gas transport mechanism in nanocellulose based hybrid membranes, thus providing theoretical support for developing optimal membrane materials with both high CO₂ permeability and high selectivity;

(3) Gas permeation of the nanocellulose based membranes is moisture sensitive. For most of the nanocellulose hybrid membranes, the highest selectivity was not obtained at the highest relative humidity, thus work should be done to either control the relative humidity in the feed gas, or optimize the nanocellulose hybrid membrane material structure to have the best separation performance at the highest RH.

(4) Functionalization of the nanocellulose may further improve the gas separation performances of the membranes, such as by grafting CO₂-philic compounds onto the nanocellulose fiber surface.

(5) The cost of the nanocellulose addition in the hybrid membranes should be evaluated for the possible industrial applications, by considering the balance between the gain from the improved properties and the increased expenses for the nanocellulose addition.