Surfactant-free cellulose filaments stabilized oil in water emulsions

There has been significant interest over recent years in the production and application of sustainable and green materials. Among these, nanocellulose has incurred great interest because of its exceptional properties and wide range of potential applications, including in Pickering emulsions. However, the production cost of these cellulosic materials has limited their application. In this study, the capability of a new type of cheaper cellulosic material, cellulose filaments (CFs), in formulating stable oil in water Pickering emulsions was investigated and compared with three conventional nanocelluloses, namely cellulose nanocrystals (CNCs), cellulose nanofibrils (CNFs) and TEMPO-oxidized CNFs (TEMPO-CNFs). Results showed that CFs can provide stable surfactant-free emulsions over wide ranges of salt concentration (0–500 mM) and pH (2–10), as indicated by the near-constant oil droplet size and dewatering index of the emulsions. This is due to the ability of CFs to strongly adsorb to the oil and water interface, as evidenced by Cryo-SEM and visualized through labelled CFs with engineered carbohydrate-binding module (CBM2a) conjugated with green fluorescent protein (CBM2a-eGFP) under fluorescent microscopy. Compared to the emulsions stabilized by other types of nanocelluloses, the CF-stabilized emulsion demonstrated a larger average droplet size and comparable (with CNFs) or better (than CNCs and TEMPO-CNFs) stability, which is partially attributed to the higher viscosity of continuous phase in the presence of CFs. The results of this study demonstrate the use of CFs as a novel and cheaper cellulosic material for stabilizing emulsions, which opens the door to a range of markets from the food industry to engineering applications.

(2-10), as indicated by the near-constant oil droplet size and dewatering index of the emulsions. This is due to the ability of CFs to strongly adsorb to the oil and water interface, as evidenced by Cryo-SEM and visualized through labelled CFs with engineered carbohydrate-binding module (CBM2a) conjugated with green fluorescent protein (CBM2a-eGFP) under fluorescent microscopy. Compared to the emulsions stabilized by other types of nanocelluloses, the CFstabilized emulsion demonstrated a larger average droplet size and comparable (with CNFs) or better (than CNCs and TEMPO-CNFs) stability, which is partially attributed to the higher viscosity of continuous phase in the presence of CFs. The results of this study demonstrate the use of CFs as a novel and cheaper cellulosic material for stabilizing emulsions, which opens the door to a range of markets from the food industry to engineering applications.

Introduction
Emulsions are colloidal systems in which droplets of a liquid phase are dispersed in another liquid phase (Schramm 2014). Generally, one of the liquid phases is aqueous while the other one is the hydrocarbon or oil phase. Depending on the dispersed and continuous phase, emulsions are primarily categorized into oil-inwater and water-in-oil wherein water forms the continuous and dispersed phase respectively (Schramm 2014). Emulsions are widely applicable in a plethora of industries, including food (Shao et al. 2020), membranes (Ahmad et al. 2019), drug delivery (Yan et al. 2019), cosmetic (Venkataramani et al. 2020), personal care (Marto et al. 2018), pharmaceutical (Kiss et al. 2011), energy storage , and oil and gas (Zhou et al. 2019), and their stability is of high significance.
Emulsions have a free energy of formation of greater than zero, making them fundamentally thermodynamically unstable with a tendency to destabilize (Ghosh 2009). Thus, stabilizers or emulsifiers are often used to form stable emulsions. Soluble surfactants and polymers are two conventional surface-active stabilizers that have been used for making stable emulsions (Jiang et al. 2020). However, conventional emulsions that are stabilized by either surfactants or polymers tend to show poor stability when exposed to change in pH, temperature or ionic strength. Moreover, some surfactants are toxic, have negative environmental footprints and can have adverse effects on health. Additionally, it has been demonstrated that the usage of surfactants and polymers can be prohibitively expensive for some processes (Khan et al. 2018;Gonzalez Ortiz et al. 2020;Shi et al. 2020).
Another interesting way of formulating stable emulsion is via the incorporation of solid particles as replacements for conventional surface-active materials. These types of emulsions are called Pickering emulsions, based on pioneering work by Ramsden and Pickering in the early twentieth century (Ramsden 1904;Pickering 1907). Pickering emulsions are beneficial in comparison to traditional emulsions because of their superior stability against creaming, sedimentation, coalescence and flocculation (Shi et al. 2020;Zhang et al. 2020). During the emulsification process, particles are partially wetted by both liquid phases, reducing the interfacial free energy between them and enabling the formation of a packed layer at the interface of the liquids (Shi et al. 2020;Yan et al. 2020). Additionally, solid particles can also form a strong network in the continuous phase, preventing the merging of the droplets while enhancing the viscosity and further stabilizing the emulsions (Low et al. 2020;Shi et al. 2020;Yan et al. 2020).
The wettability of the particles, which is demonstrated by the three-phase contact angle at the interface, is a crucial factor regarding the formation of Pickering emulsions (Ragesh et al. 2014;Wu et al. 2020). Besides wettability, several other factors including pH of the continuous phase, ionic strength, and the concentration and shape of the particles control the emulsification process and emulsion stability (Li et al. 2018;Mikulcová et al. 2018;Varanasi et al. 2018;Zhao et al. 2019).
Previously, several types of solid particles have been applied as stabilizers for Pickering emulsions, including (i) inorganic particles such as silica Björkegren et al. 2020), metal oxide particles (Xie et al. 2017;Fessi et al. 2019), graphene (He et al. 2013), calcium carbonate (Zhu et al. 2013), etc.;(ii) polymeric particles such as polystyrene (Jiang et al. 2019;Li et al. 2019); and (iii) food-grade particles such as starch (Zhu 2019), chitin (Jiménez-Saelices et al. 2020), whey (Lee et al. 2020), etc. With the growing demand for sustainable materials with reduced environmental footprints and with green and energy-efficient production routes, the application of natural materials has attracted increasing attention. Among such materials, nanocellulose (e.g. cellulose nanocrystals (CNCs), cellulose nanofibrils (CNFs), and TEMPO-oxidized CNFs) have received significant interest due to their sustainability, renewability, nontoxicity, biodegradability, and biocompatibility (Prathapan et al. 2016). For example, Pandey et al. (Pandey et al. 2018a, b) investigated the ability of two different kinds of CNCs with various degrees of surface charge in stabilizing oil in water emulsions, and found that CNCs with lower surface charge have faster adsorption kinetic and formulate emulsions with smaller droplet size. Bai et al. (Bai et al. 2019) employed high-energy microfluidization to stabilize oil in water emulsions using CNCs, which showed good stability over NaCl range of lower than 100 mm and pH range of 3-10. Aaen et al. (Aaen et al. 2019) demonstrated that both enzymatically treated CNFs and TEMPO-oxidized CNFs can result in stable rapeseed oil in water emulsions. Their results showed that enzymatically treated CNFs can be stable in the presence of NaCl and at low pH with only a slight increase in droplet size, while TEMPO-oxidized CNFs are unstable at these harsh conditions. As described above, there are several studies available in the literature on the application of different types of nanocellulose on stabilizing Pickering emulsions. However, these nanocelluloses are expensive and there is still strong demand to produce cheaper and more effective cellulose-based materials for formulating Pickering emulsions. Cellulose filaments (CFs) are an emerging class of ''nanocellulose'', which contains a heterogeneous combination of fine and long nano/microfibrillar materials and are produced using a relatively cheap production process. CFs can be manufactured from a variety of bleached or unbleached wood pulps using simple mechanical shearing and without chemical or enzymatic treatments (Hamad et al. 2019;D'Acierno et al. 2020). The application of CFs has been explored in many fields but limited information is available for stabilizing Pickering emulsions. Herein, this study aims to investigate the formulation and characterization of oil-water Pickering emulsions stabilized by commercially available CFs. The effect of particle concentration and environmental stresses including ionic strength and pH on the stability of the emulsion are evaluated. The emulsifying ability of the CFs is compared to other types of nanocelluloses including CNFs, CNCs and TEMPO-oxidized CNFs. The characterization of the emulsions was performed by dynamic light scattering, zeta potential measurements, optical microscopy, viscosity measurements, and fluorescent microscopy.

Materials
Cellulose filaments (CFs) at 10.0 wt% consistency was kindly given by Performance BioFilaments, BC, Canada. Cellulose nanocrystals (CNCs), cellulose nanofibrils (CNFs) and TEMPO-oxidized cellulose nanofibrils (TEMPO-CNFs) were obtained from Cel-luForce and Cellulose Lab, Canada. All the celluloses were used as supplied without any further modification. Water was purified by the Mili-Q water purification system. Dodecane (anhydrous [ 99%) were purchased from Sigma Aldrich, Canada, and were used as received for the oil phase. Oil Blue N (Dye content 96%. Sigma Aldrich) was used to label the dodecane. Enhanced green fluorescent protein fused carbohydrate-binding module 2a (CBM2a-eGFP) was synthesized in-house as described in the literature and used for fluorescent labelling of the CFs (Hu et al. 2014;Gourlay et al. 2015). Sodium chloride (NaCl), hydrochloric acid (HCl) and sodium hydroxide (NaOH) were obtained from Sigma Aldrich, Canada.

Preparation of cellulose suspensions
Cellulose suspensions were prepared by adding an appropriate amount of celluloses in water, which is then sonicated for 10 min (Branson 2800 series). The sample is then ultrasonicated in an ice bath (Qsonica Q700 sonicator) with 50% of maximum amplitude with 1 min on and 30 s off interval to receive an energy input of 10 kJ for a 20 cc sample. The pH of the suspensions was adjusted by using HCl and NaOH solutions whereas NaCl was added to modify the salt concentration.
Dynamic light scattering, f-potential The apparent hydrodynamic diameter and f-potential of the cellulose samples were measured using a Particulate Systems Nano Plus HD nanoparticle/zeta analyzer. Dynamic light scattering (DLS) measurements were performed on suspensions with 0.1 wt% cellulose concentration and measurements were repeated 15 times to obtain an average value. fpotential measurements were done on cellulose dispersions with 0.025 wt% concentration and measurements were repeated at least 5 times.

Scanning electron microscopy and X-ray diffraction
Scanning electron microscopy (SEM) was conducted using a Hitachi S 300 N system (Tokyo, Japan). CFs sample with 0.5 wt% concentration was freeze-dried to perform SEM observations. Before imaging, to build up charge on the surface, the sample was coted with Pd-Au alloy. Cellulose crystallinity (CrI) was obtained through recording X-ray diffraction (XRD) in the 2-Theta range of 5°-80°(Bruker TOPAS 4.2).

Viscosity measurements
For measuring the viscosity of 0.5 and 1.0 wt% CF suspensions with and without 1000 mM NaCl, Anton Par MCR-302 rotational rheometer with concentric cylindrical geometries was used. Measurements were done at a shear rate range of 0.01-1.0 1/s with a parallel plate (25 mm diameter) geometry of the rheometer. Also, the apparent shear viscosity of all cellulose suspensions with 0.5 wt% concentration was performed by the DV3TLVTJ0 Brookfield rheometer. The shear rate was 50 1/s and the measuring time was 3 min with 6 measurements with a 30 s interval.

Emulsion preparation and stability evaluation
Emulsion preparation 10 ml of cellulose suspension and dodecane with 2 mg Oil Blue dye were added to a 20 ml vial to prepare a 1:1 volume ratio emulsion. For CFs four different concentrations of 0.05, 0.1, 0.25 and 0.5 wt% were used. To compare the different celluloses, emulsions with 0.25 wt% of CNCs, CNFs, TEMPO-CNFs and CFs were prepared. The concentration here refers to the cellulose content in the emulsion phase. Therefore, the cellulose content in the used aqueous suspension is twice. To investigate the effect of salt and pH, modification of the salt concentration and pH were done on the aqueous phase before emulsification. Emulsification was performed using ultrasonication (Qsonica Q700) for about 1 min at 100% amplitude to reach an input energy of 3800 J. All the prepared emulsions were stored at ambient temperature for 14 days for stability analysis.

Visual assessment
The stability of the formulated emulsions in terms of dewatering/phase separation was evaluated by visual inspection of the stored emulsions. The inspection was performed after 1 day and 2, 3, 7 and 14 days after preparation of the emulsions. The serum and emulsion layer heights were recorded at these time intervals and the dewatering index was calculated based on the following formula: where H s is the serum layer height separated from the emulsions phase at the bottom of the vials and H a is the initial height of the aqueous phase in the vial.

Optical microscopy
To record the images of the oil droplets in the emulsions, a Zeiss Axio Vert.A1 optical microscope was used. A drop from each emulsion was deposited on a glass slide. Then a cover glass was laid on the top of the sample to avoid air bubbles and placed on the microscope. The obtained images were analyzed in Image J software to evaluate the droplets size distribution. The Sauter mean diameter (D 3,2 ) (Sauter 1926) of the droplets was obtained by evaluating at least 200 droplets. Sauter mean diameter (D 3,2 ) reflects the mean diameter of the droplets by taking into account the volume to surface area ratio and is calculated by the following equation: where d i is the diameter of a droplet and n i stands for the number of droplets.
To obtain the width of the droplets size distribution, the span (dispersion index) was calculated through the following formula (Mcclements 2007): where d 10 , d 50 and d 90 are the droplet diameters that are larger than 10%, 50% and 90% of the droplets, respectively.

Fluorescent microscopy
The fluorescence images of the emulsions were taken using an Olympus IX73 inverted fluorescence microscope. CFs were labelled through incubation with CBM2a-eGFP. For this purpose, CFs were washed with 50 mM sodium phosphate buffer, pH 7.4. The liquid was removed by centrifuge. Then CFs were incubated with 0.26 mg/mL CBM2a-eGFP in 50 mM sodium phosphate buffer, pH 7.4 with gentle shaking overnight at 4°C. Excessive protein was removed by centrifugation. Protein-bound CFs were rinsed with sodium phosphate buffer, then resuspended in the same buffer for future application.
Cryogenic scanning electron microscopy (Cryo-SEM) Scanning electron microscopy under cryogenic conditions (Cryo-SEM) was performed using a FEI Quanta FEG 250, to get further insight into the structure of emulsions and arrangement of the nanocelluloses at the interface and continuous phase. The cryo-SEM was equipped with a Gatan cryo-stage module and a Gatan Alto2500 cryogenic sample preparation and transfer system. A droplet of the prepared emulsions with 0.25 wt% of nanocellulose was taken and placed on the specimen holder and rapidly cryofixed in slushy nitrogen. Then the frozen sample was transferred to the preparation chamber and sublimed-etched, fractured with a blade and coated with a layer of gold-palladium (60:40 wt%). The prepared sample was then transferred to the observation chamber equipped with variable pressure environmental field SEM with a cold stage module held at about -150°C and under low-vacuum conditions.

Characterization of different cellulose samples
To compare the different nanocellulose materials in terms of their surface charge and size, zeta potential and dynamic light scattering measurements were performed at neutral pH (Table 1). As expected, cellulose nanocrystals (CNCs) showed a high negative surface charge (-30 mV) due to the existence of sulfate groups resulting from their synthesis route which includes a sulfuric acid hydrolysis step. Among the different nanocelluloses, cellulose nanofibrils (CNFs) and cellulose filaments (CFs) have a low surface charge with a zeta potential of about -10 mV which is three times lower than the CNCs. This is likely due to the mechanical refining methods used for their preparation, which does not introduce extra acid groups on the fiber surface. Negative charges on CFs and CNFs are attributed mostly to the presence of hydroxyl groups. In addition, CFs also contain a certain amount of hemicellulose which could also contribute to the surface charge. The slightly higher zeta potential of the CFs compared to CNFs is probably due to the higher amount of hydroxyl groups and some charged groups on hemicellulose that exist in their structure. The TEMPO oxidation process in which the primary hydroxyl groups are modified to anionic carboxylic groups results in cellulose nanofibrils (TEMPO-CNFs) with a medium amount of negatively charged surface (-18 mV) among others. Aside from surface charges, the size of nanocelluloses is related to their synthesis route, where CFs show the highest average apparent hydrodynamic diameter, followed by CNFs, TEMPO-CNFs and CNCs. SEM images of the CFs (Fig. 1) display that CFs are heterogeneous combinations of both long and fine nano/microfibrillar materials demonstrating that why CF average apparent hydrodynamic diameter is larger than other types of nanocelluloses. SEM images reveal that CFs have wide particle size distribution with a minimum width of * 20 nm for fine fibrils, and a width up to * 30 lm for large fibres. Besides CFs' fine fibrils have lengths in the 10-500 lm range, and large fibre with lengths ranging from 500 lm to 2500 lm. However, CNCs and CNFs typically have more homogenous particle length and width. CNCs are highly crystalline with width in the range of 3-5 nm and length in the range of 50-500 nm (Moon et al. 2011;Kalia 2016;Thomas et al. 2018). While CNFs have higher aspect ratios, with widths of 4-20 nm and lengths of 500-2000 nm (Moon et al. 2011;Thomas et al. 2018). Fig. S1 shows that CFs have the typical cellulose XRD pattern with similar crystallinity (* 71%) as cellulose fibers.

Visual inspection
Visual inspections were carried out to evaluate the stability of the prepared emulsions by observing their ability to resist physical changes over time. The photographs of the formulated emulsions at various CF concentrations immediately after preparation, 1 day after preparation and 14 days after preparation are shown in Fig. 1a-d. As seen from Fig. 1a and b, the oil in water emulsions prepared at 0.05 wt% and 0.1 wt% showed phase separation and creaming immediately after their preparation. However, emulsions with CF concentrations of 0.25 wt% and 0.5 wt% exhibited a delayed phase separation with a delay time of 1 day ( Fig. 1c and d). The emulsion volume increased with the concentration of CFs and at 0.5 wt% of CFs, there is almost complete emulsification. Increasing CF concentrations allowed for increased droplet stabilization by providing more interfaces that fill larger volumes. Additionally, higher CF concentration increases viscosity, preventing mobilization of the continuous phase between the oil droplets, resulting in a more stable emulsion. In Fig. 1a and b, the cloudy serum layer at the bottom of the vials at CF concentrations of 0.05 wt% and 0.1 wt% indicates that the CFs desorbed from the oil and water. interface and gathered in the serum layer. This is attributed to the high mobility of continuous phase and coalescence of oil droplets resulting in detachment of the CF from the interface. At CFs concentrations of 0.25 wt% and 0.5 wt%, clear water is observed in the serum layer. This indicates that all the particles are retained in the emulsion phase, either absorbed at the oil and water interface or retained in the continuous phase because of the low mobility of this phase and lack of droplet coalescence. In general, photographs in Fig. 1a-d show that CFs can stabilize oil in water emulsions with a concentration as low as 0.05 wt%, while the volume and stability of emulsion enhance with increasing concentration.

Droplets size distribution
The stability of the emulsions was evaluated by analyzing oil droplet morphology and the average diameter of droplets. In addition to the photographs of emulsions, the micrographs obtained by optical microscopy are also shown (Fig. 2a-d). It appears that increasing the CF concentration from 0.05 wt% to 0.5 wt% leads to a clear reduction in droplet diameter. At very low concentrations of CFs, there is obvious evidence of flocculation and coalescence. The degree of flocculation reduces as CF concentration is increased to 0.1 wt% and 0.25 wt%. At CF concentration of 0.5 wt% where the most stable emulsion is obtained, there is not much difference in the micrographs over time, indicating that at this concentration the emulsion is stable against flocculation and coalescence. In general, emulsions with smaller droplets demonstrate higher stability since it takes a longer time for the droplets to coalesce (Pal 1996). Quantitative analysis was performed by calculating the average diameter of the droplets (d 3,2 ) in Fig. 2e. The average diameter of the droplets significantly increased over time at CF concentration of 0.05 wt%, from a value of about 58 lm for fresh emulsions to 91 lm for emulsions stored for 14 days. At this concentration, the amount of CFs is too low to stabilize small droplets, resulting in a tendency to coalesce, reducing the total interface area of the system. By doubling the CF concentration, the average diameter of the droplets became stable at a value of about 47 lm which is about half of the average diameter of the droplets at 0.05 wt% CFs. By increasing the CF concentration both the average diameter of the droplets and the percent of change in its value with the storage period reduce. For the emulsions with 0.5 wt% CFs, there is only a negligible increase in the average diameter of droplets after 14 days, demonstrating good stability of the emulsions. The size distribution of the droplets for all the emulsions are provided in Fig. S2-S5. Additionally, the span of the emulsions reduced from a value of 2.2 at 0.05 wt% CFs to a value of 0.73 at 0.5 wt% CFs, indicating very wide droplets size distribution at low concentration as a result of significant coalescence. Increased CF concentration led to a decrease in droplet size distribution range. Span index calculations are summarized in Table S1.

Dewatering index
The stability of the emulsions was further evaluated by analyzing the dewatering index. The calculated dewatering index based on visual inspection (Fig. 2f) shows that the phase separation of the water phase from the emulsion starts shortly after the generation of the emulsions. Most of the dewatering happened within a day after emulsion preparation, independent of the amount of CFs introduced. After one day, the dewatering index slightly increased, reaching a constant value after three days for all emulsions. The most stable emulsion is the emulsion with 0.5 wt% of CFs which is stabilized at a dewatering index of 14%. In the case of emulsions with 0.05 and 0.1 wt% CFs, more than 50% of the water in the emulsion dewatered after 14 days which can be attributed to two effects: (i) larger droplet size and (ii) lower continuous phase viscosity.

Fluorescent microscopy
To better understand the mechanism of CFs stabilizing oil-water emulsion, fluorescent protein labelled carbohydrate-binding module (CBM) was employed to visualize the location of CFs under the fluorescent microscope. CBMs are the special structural protein modules of carbohydrate-active enzymes, which play the role of binding the enzymes to their targeted polysaccharides in nature. The high binding specificity and the relatively simple operation endow CBMs with excellent potential to probe the carbohydrate polymers in a complex system. We and others have developed various engineered CBM linked with different fluorescent proteins/probes to assess the location and/ or surface morphology of various cellulosic fibers in  In this study, the specific cellulose-binding carbohydrate-binding module (CBM2a) linked with enhanced Green fluorescent protein (CBM2a-eGFP) was used to track the location of CFs within the emulsion. Figure 3 displays the fluorescent microscope images of the stabilized emulsions with 0.5 wt% CFs labelled by CBM2a-eGFP under bright field and fluorescent light. The fluorescent light images in Fig. 3b (corresponding to the bright field image in Fig. 3a) display the adsorption of the CFs at the oil and water interface. The bright fluorescent light can be observed around the oil droplets while the inside of the droplets is dark. Also, no fluorescent reflection was observed in the continuous phase. This observation indicates the mechanism of stabilizing emulsions by CFs. CFs are mainly located at the oil and water interface demonstrating their adsorption at the oil droplet surfaces, leading to particle-stabilized Pickering emulsions.

Effect of ionic strength
It has been demonstrated that changing the ionic strength can greatly affect the stability of Pickering emulsions. In this study, the effect of the addition of simple electrolyte, NaCl on the CF suspension zeta potential (Fig. 4a), on the 0.5 wt% CF-stabilized emulsion stability (in terms of the average diameter of droplets and dewatering index (Fig. 4b and c)), and on the viscosity of CF suspensions (Fig. 4d) are investigated, respectively. Figure 4a displays that the zeta potential of CF suspension is reduced from a value of -18 mV to a value of about -6 mV in the presence of 60 mM NaCl, which is due to the electric double layer formed by the addition of Na ? cation counter ions. The Na ? counter ions adsorb to the CF particle surface shielding their surface charge and causing a reduction in the electrostatic repulsion, therefore reducing the zeta potential value. The electrostatic screening happens in the presence of Na ? counter ions and the Debye-Huckel screening strength expands by increasing Na ? counter ions concentration hence Debye length decreases (Prathapan et al. 2016). It should be noted that due to the device limitation, we only measured the zeta potentials values reliable to the ionic strength of up 60 mM. However, we believe that the trend is correct and zeta potential will approach zero at higher ionic strengths.
Unlike Zeta potential, the addition of NaCl (100 mM) initially increases the average diameter of oil droplets and then decreases the droplet size as salt concentration is increased to 500 mM. (Fig. 4b). The increasing droplet diameter in the presence of 100 mM NaCl could be attributed to the reduction in the electrostatic repulsion between the droplets allowing flocculation to take place, as evidenced by reducing Zeta Potential in Fig. 4a. The reduction of the droplet diameter by the addition of more NaCl to the system (500 mM NaCl) appears to be related to the enhanced viscosity at this salt concentration. Overall, from the salt concentration of 0-100 mM, the reduction in the electrostatic repulsion is the reason why droplets diameter increases, while from the salt concentration of 100-500 mM, the increased viscosity of the emulsions is the reason why droplet diameters reduced. This will be discussed in more detail later (Fig. 4d). Although the droplet size increased (Fig. 4b), the dewatering index remains constant after the addition of 100 mM NaCl to the system (Fig. 4c). However, in the presence of 500 mM NaCl, the dewatering index slightly decreased which indicates better stability of the emulsion with high salt concentration. At both CF concentrations of 0.5 and 1.0 wt%, the addition of 1000 mM NaCl to the suspension (equivalent to the presence of 500 mM NaCl in the emulsion) increases the viscosity of suspensions by a factor of two (Fig. 4d). It seems that the enhanced viscosity of the continuous phase in the presence of high salt concentration helps the stability of emulsions by immobilizing the droplets and preventing them from flocculation. In general, from Fig. 4, it can be seen that the presence of salts does not significantly affect the stability of CF stabilized oil-in-water emulsions, and these emulsions are stable over a wide range of ionic strengths.

Effect of pH
To further investigate the effect of pH variation on CF stabilized oil-in-water emulsions, the zeta potential of CF suspensions, the average diameter of droplets and the dewatering index of the emulsions are next assessed at different pH (Fig. 5). By reducing the pH from 10 to about 1.88, the zeta potential of the CF suspensions is reduced from a value of -15.5 mV to a value of near zero (Fig. 5a). This is likely because of the protonation of the negative surfaces (carboxyl and hydroxyl groups) by the addition of acid to the  Figure 5b and c show the effect of pH on the stability of the emulsion at 0.25 wt% and 0.5 wt% CF concentrations. The average diameter of the oil droplets slightly reduced as pH was increased, as the surface charge protonation at low pH reduces the repulsion between the droplets which causes more droplets interactions and results in the increased diameter. However, the formulated emulsions are fairly independent of the pH as the dewatering index remained almost constant at all pH values. This indicates that changing the pH did not trigger phase separation and further demonstrates the strong adsorption of CF particles to the oil and water interface.

Comparison of the different types of cellulose
To further demonstrate the potential of using CFs for stable emulsion forming, we compared the emulsifying capacity' CFs with other commercially available nanocelluloses including CNCs, CNFs and Tempo-CNFs (at 0.25 wt% concentrations). Even though these materials have different topologies such as particle shape, aspect ratios, etc., the goal of comparison was to investigate their ability in emulsion stabilization in terms of droplet size, dewatering rate with similar emulsification methods, regardless of their morphology and size. It appears that among different commercially available nanocelluloses, CFs provide the best results in emulsion stabilization, which could be considered as more shovel-ready technology.
As can be seen from Fig. 6 all the investigated different cellulose particles can provide stable emulsions. However, the emulsion stabilized by TEMPO-CNFs has the smallest droplet size, while the CFs stabilized emulsions have the largest droplets ( Fig. 6a and b). Dewatering index results (Fig. 6c) show that the CNFs can provide the highest emulsion volume (lowest dewatering index) followed by CFs, TEMPO-CNFs and CNCs. This may be attributed to the adsorption energy of the particles. Adsorption of CFs and CNFs to the interface appears to be stronger and irreversible, while adsorption of CNCs and TEMPO-CNFs is reversible. These celluloses desorbed from the interface leading most of the emulsion phase to become unstable shortly after preparation. Another explanation for the higher stability of emulsions stabilized by CFs and CNFs in comparison to CNCs and TEMPO-CNFs is that CFs and CNFs suspension have higher viscosity (Fig. 5d).

Cryo-SEM imaging
To characterize the arrangements of the different nanocelluloses at the oil and water interface and the continuous water phase, cryo-SEM measurements were performed at 0.25 wt% emulsions stabilized by different nanocelluloses. Cryo-SEM measurements at different magnifications are shown in Fig. 7 and Fig. S6. Closer views are provided in Fig. 7 to observe the interface and lower magnifications are shown in Fig. S6 to observe the continuous phase as well.
During the fracturing of the samples before SEM imaging, the frozen oil droplets have been removed resulting in the formation of a cavity that leads to direct observation from above of particles located at the interface. From Fig. 7, in all emulsions, oil droplets with a shell composed of nanocelluloses are observable. For CNC and Tempo-CNF emulsions, there is a dense and smooth layer of the relatively small-sized nanocelluloses at the interface forming a web-like structure. On the other hand, for the CNF emulsions, a network of mostly fine nanofibers is observable at the interface, while for the CF emulsions, there is a network of large nanofibers linked through fine fibril constituents. Even though CNCs and Tempo-CNFs seem to have a uniform and more surface coverage they have lower stability compared to CNFs and CFs. These configurations suggest that for the nanocelluloses the particle size and their morphology is the main factor that affects the stability of the emulsions. In the case of CNF and CF emulsions, the oil droplets' surfaces are associated with the highly entangled network of fine fibrils. A recent study showed that the incorporation of fine cellulosic components can highly enhance the strength of a large fibres network at the air and water interface  (Pöhler et al. 2020). We suppose that the same behaviour happens here at the oil and water interface, in which the presence of fine fibrils in CFs causes a higher degree of entanglement in CF particles at the interface acting as a steric hindrance protecting oil droplets against coalescence and flocculation. Interestingly from Fig. S6 a-d, an almost clear continuous phase with the presence of few particles is visualized for CNC, CNF and CF emulsions. Suggesting that the adsorption of the particles to the interface is the major mechanism stabilizing emulsions. This was unlike the literature studies where both particles' adsorption to interface and formation of a strong particle network at the continuous phase were contributed for emulsion stabilization (Pandey et al. 2018a;Goi et al. 2019). On the other hand, for Tempo-CNF emulsions, there is a dense presence of particles at the continuous phase (Fig. S6b). This indicates that these particles were less adsorbed to the interface and part of them remained in the water phase. The sharp increase in the dewatering index of Tempo-CNF emulsions (Fig. 5c) might be related to this configuration. The remaining Tempo-CNFs at the water phase might drained out of the emulsion phase and accumulated at the serum layer and caused instability of the emulsions. This was confirmed based on the image of the emulsions, where for Tempo-CNF emulsions a cloudy serum layer observe below the emulsion phase (images are not provided here).

Potential applications of the CF emulsions
Considering the much cheaper production cost of CFs and the stable Pickering emulsion they form at varying salt concentrations and pH, CFs can be a better choice than other nanocellulose materials for stabilizing emulsions for a wide range of applications, specifically in food and health, and innovative fields of lightweight materials and 3D printing. Nanocelluloses based advanced emulsions have been applied in the food industry where improved functional attributes, increased bioavailability of bioactive, controlled digestion behaviour, controlled food delivery, and improved sustainability and safety are required (Tan and McClements 2021). For example, CNC and CNF stabilized emulsions were used to produce functional foods for the targeted delivery of oil-soluble nutrients and vitamin D3 (Winuprasith et al. 2018;Zhou et al. 2018). Moreover, Pickering emulsion stabilized by nanocelluloses has been successfully applied as a precursor for producing food packaging materials (Deng et al. 2018). Recently published reviews have discussed the potential applications of nanocellulose Pickering emulsions in the food industry (Tan and McClements 2021;Zhu et al. 2021). Although CNC and CNF applications have been widely explored in the food industry, they are often expensive, have complicated production routes. Besides, they sometimes suffer from having stability at the desired conditions, and they don't have enough viablility for large scale applications. To these ends, CNC and CNF based emulsions in food application can potentially be replaced by CF emulsions since CFs can overcome the mentioned limitations. In addition to applications in the food industry, CF emulsions can be potentially applied for the development of aerogels and ink for 3D printing. Aerogels are the least dense solid materials with a porosity of around 99%, which can be further used in thermal insulation, energy storage devices, biomedical scaffolds, flame retardancy, etc. (Ahankari et al. 2021) Recently, CNF oil-in-water emulsions has been applied to develop strong aerogel simply by freeze-drying of the emulsions and used as thermal superinsulation material (Jiménez-Saelices et al. 2018). On the other hand, the application of CNF based emulsion in 3D printing has been recently investigated (Huan et al. 2019). Complex geometries with tunable structures were formed using a simple approach using nanocellulose based emulsions through 3D printing (Huan et al. 2019). Herein we showed that the performance of CF emulsions is comparable to CNFs and provides highly stable gellike emulsion. These gel-like emulsions can be potentially utilized in 3D printing and as templates to form aerogels with several different subsequent applications. Overal, based on the advantages of the CF emulsions, they can be a great candidate to be applied to several innovative fields as described above. However, the application of CF emulsions in these areas is relatively novel and systematic studies are required to elucidate all different potential capabilities of CF emulsions.

Conclusions
In this study, the potential application of a novel and green material, cellulose filaments (CFs) in stabilizing oil in water Pickering emulsions is investigated. Because of the strong adsorption of the CFs to the oil and water interface and the formation of a highly viscous aqueous phase, CFs successfully stabilized oil in water emulsions. In contrast to CFs, emulsions stabilized by CNCs and TEMPO-CNFs showed significant coalescence and creaming even though they have smaller average droplets diameter. Results showed that the better performance of the CFs was likely due to the high viscosity of the aqueous phase in the presence of CFs compared to CNCs and TEMPO-CNFs. Besides, Cryo-SEM images suggested that strong adsorption of the CFs at interface and formation of the entangled network due to the presence of fine fibrils is responsible for stabilizing the emulsions. The analysis of the effects of pH and salt concentration showed that the CFs-stabilized emulsion was not significantly affected by a change in the pH (2 -10) and salt concentrations (0-500 mM). From the results presented here, it can be concluded that the CFs are a promising new nanocellulose for the sustainable and cost-efficient formulation of stable Pickering emulsions without using any additives or surfactants and with potential applications in a wide range of markets.