Separation of chemically modified carbon nanotubes by surfactant free microbubble generation

A microbubble is defined as a bubble having a diameter of <50 μm, which has attracted great attention because of its superior physical chemistry properties and the differences in performance compared to ordinary macrobubbles.¹ Microbubbles have been widely used in various fields such as pharmaceutical engineering,² chemical engineering,³ and industrial technology.^4–6 Previous studies have shown that the flotation recovery of particles can be enhanced by fine bubbles (microbubbles and nanobubbles) in froth flotation systems.^7–9 Fine bubbles are stable for longer periods under water, thereby increasing the probability of collision between particles and bubbles.¹⁰ 

At present, a majority of the methods developed for microbubbles such as pressure waves,¹¹ hydrodynamics,¹² or optics are cavitation-based.¹³ For these cavitation-based techniques, chemical reagents such as surfactants and frothers are usually added to increase the quantity of generated microbubbles, reduce the size of the bubbles, and stabilize their structure.¹² In recent years, bubbles generated in surfactant solutions with the hydrophilic surface property have been reported.^14,15 However, both the small size of bubbles and the hydrophobic surface property are critical for the recovery of hydrophobic materials.

In this study, carbon nanotubes (CNTs) with different surface modifications are chosen as separate materials. A CNT is formed by rolling a graphene sheet into a hollow tube.¹⁶ It received attention in various fields since its discovery in 1991.¹⁷ Because of its excellent characteristics such as high electrical conductivity,¹⁸ good mechanical properties,¹⁹ and high thermal conductivity,²⁰ it is expected to have potential in the fields of chemical sensors,²¹ thermoelectric conversion composite materials,²² and semiconductor devices.²³ However, its tendency to form bundles because of van der Waals interactions and its high hydrophobicity property make its use limited in applications.¹⁶ Recently, the chemical functionalization of CNTs has been reported to improve their poor wettability and enhance the compatibility of target media²⁴ and to increase their potential for application in various fields, such as electrochemical sensors²⁵ and high-performance batteries.²⁶ 

In this study, we show a method that generates dense (∼10⁴ microbubbles/μl) and small diameter microbubbles (average diameter: 2 µm) in a supersaturated solution by applying impact and without addition of surfactants and frothers. This method generates microbubbles that do not contain surfactants adsorbed at the interface between the solution and the microbubble, causing the surface of the microbubble to have different properties from the surface of the surfactant microbubble. Based on the different properties of the surface, hydrophobic CNTs and hydrophilic surface-modified CNTs (aCNTs) could be separated by dense microbubbles.

In this study, microbubbles are generated by subjecting rigid surfaces to impact with free fall from a height of 1.5 m and an instantaneous velocity of 5.4 m/s on colorimetric tubes containing different solutions, as shown in Fig. 1(a). Different concentrations of ethanol solution (purity 99.5 wt. % purchased from Nihon Shiyaku Reagent), namely, 0%, 50%, 75%, and 100% (v/v), diluted with deionized (DI) water, a cationic surfactant solution [dodecyldimethylethylammonium bromide (DDAB) purchased from Tokyo Chemical Industry], an anionic surfactant solution [sodium dodecyl sulfate (SDS) purchased from Merck], a nonionic surfactant solution (Pluronic® F-127 purchased from Sigma), and a supersaturated carbonated solution are used to generate microbubbles by this method.

Carbon dioxide is introduced into the solution by a sparkling water maker so that it becomes a supersaturated carbonated solution (SodaStream International Ltd.). The maximum pressure in the CO₂ cylinder is approximately 870–1000 psi.

We designed a chamber to measure the size of microbubbles by observing the dynamics of microbubble generation. The assembly of the measuring chamber is illustrated in the inset of Fig. 1(d–II). The chamber consists of polymethyl methacrylate (PMMA) (top layer), double-sided tape (flow channel), and a transparency film (bottom layer). The 25% ethanol solution is injected to the channel [L (length): 7 mm, W (width): 1.8 mm, and H (height): 92 µm], and the channel is linked to the reservoir (L: 10 mm, W: 10 mm, and H: 92 µm). The mechanical vibration caused by the shooting of an airsoft pellet (diameter: 6 mm, weight: 0.3 g, and shooting velocity: 29 m/s) at the reservoir generates dense microbubbles. The images of microbubble generation are captured by using a CMOS camera (DMK 37BUX287, The Imaging Source) with an objective lens (20×, NA 0.4, Olympus), and microbubble sizes are measured using image analysis software (Image J).

Multi-walled CNTs (a length of 3–12 µm and diameter of 8–15 nm) are dispersed in DI water and sonicated at room temperature for 2 h to obtain a homogeneous suspension. For hydrophilic CNTs, carboxylated MWCNTs (aCNTs) are synthesized by chemical modification, as per our previous work.²⁷ aCNTs are dispersed in DI water and sonicated at room temperature for 30 min to form a homogeneous aCNT solution.

The hydrophobicities of CNTs and aCNTs are measured by the microbial adhesion to hydrocarbons (MATH) test. The MATH test is a fast and commonly used experimental technique to evaluate the relative hydrophobicity of particles or bacterial cells by the percentage of particles or bacterial cells attached to hydrocarbons.²⁸ In brief, 1 ml of n-dodecane (>99% purchased from Alta Aesar) is added to 3 ml of the CNT suspension (0.1 mg/ml) and aCNT solution (0.1 mg/ml) separately. The samples and n-dodecane are mixed for 2 min with a vortex and then allowed to separate by resting the solution for 15 min. After calculating the optical density of the suspension in the aqueous phase, the percentages of CNTs and aCNTs partitioned to the hydrocarbon phase or trapped at the interface are calculated. All samples are measured in the visible spectrum of 400–700 nm. The hydrophobicity percentage is calculated by²⁹ hydrophobicity = $ODinitial−ODresidualODinitial×100%$⁠.

Different quantities of microbubbles can be generated according to the concentration of the ethanol solution. The container surface was subjected to free-fall impact to generate microbubbles, and the images of generated microbubbles in different concentrations of ethanol solutions are compared, as shown in Figs. 1(b-I) and 1(b-II). We find that the densest microbubbles are generated in the 25% ethanol solution. As the concentration of the ethanol solution increases from 50% to 75%, the microbubbles generated by subjecting the container surface to impact gradually decrease. Almost no generation of bubbles is observed in 100% and 0% ethanol solutions. To quantify the generated microbubbles and the process of microbubbles dissipating in different concentrations of ethanol solutions, we measure the intensities of scattered light of the generated microbubbles after subjecting different concentrations of ethanol solution to impact, as shown in Fig. 1(b–II). The measurement starts 5 s after the impact, and the red rectangle in Fig. 1(b–I) is the labeled measurement position. The relative intensities of scattering light of microbubbles are measured by $I=IsolIDI$⁠, where I, I_sol, and I_DI are the relative intensities of scattering light, intensities of scattering light of different concentrations of ethanol solutions (0%–100%), and DI water, respectively. The results show that microbubbles generated in 50% and 75% ethanol solutions quickly dissipate within 20 s. However, compared with other concentrations, the intensity of scattering light of microbubbles generated in the 25% ethanol solution is higher, and the dissipation time is longer.

The dissipation of microbubbles in the 25% ethanol solution is observed by microscopy to confirm the relationship between the quantity and the size of microbubbles with the intensities of scattering light, as shown in Fig. 1(c). The measurement starts 20 s after the impact. Corresponding to the intensity of light scattering, the presence of microbubbles can still be observed in the 25% ethanol solution 140 s after the impact. During the dissipation process, the quantity of microbubbles decreases with time, while the diameter of microbubbles increases with time. To confirm that the generated microbubbles are not affected by impurities in non-high purity ethanol, we use higher purity ethanol purchased from other manufacturers (with a purity of 99.9% purchased from J.T. Baker) to generate microbubbles by the same method. The quantity and size of microbubbles that dissipate with time are observed by microscopy. The results are similar to the results of this study (data not shown).

The image of generated microbubbles in the 25% ethanol solution is shown in Fig. 1(d–I) (0.05 s after impact). It shows that the microbubble size distribution is mostly below 10 µm, which indicates that our method can generate microbubbles on the micrometer scale [Fig. 1(d–II)]. Since no surfactant was adsorbed on the surface of the microbubbles, the generated microbubbles in the ethanol solution may have different surface properties from the microbubbles generated in the surfactant solution. Based on the differences in surface properties, the collection effect of hydrophobic CNTs could be affected. Next, we show the recovery of CNTs by microbubbles generated in different solutions.

To investigate the recovery ratio of CNTs, each container containing different surfactant solutions and supersaturated solutions including 25% ethanol solution and supersaturated carbonated water was subjected to impact. The same concentration of CNTs, 0.2 mg/ml, is added to 1% SDS, 1% DDAB, 1% Pluronic® F-127, beer [classic Taiwan beer, 5% alcohol (v/v) purchased from Taiwan Tobacco and Liquor Corporation], supersaturated carbonated water, and 25% ethanol solution (all use v/v) separately. After subjecting the colorimetric tubes to impact to generate dense microbubbles, most of the CNTs rise to the solution surface in the 25% ethanol solution, and a few CNTs in supersaturated carbonated water also rise to the solution surface. Recovery of CNTs in surfactant solutions and beer is hardly observed, as shown in Fig. 2(a).

The recovery of different concentrations of CNTs ranging from 0.25 mg/ml to 1 mg/ml is measured, and the recovery ratio of CNTs after subjecting each colorimetric tube containing 1% SDS, 1% DDAB, 1% Pluronic® F-127, beer, supersaturated carbonated water, and 25% ethanol solution to impact is measured [Fig. 2(b)]. In supersaturated solutions, we observe that the recovery ratio of CNTs in the 25% ethanol solution is higher than that in the supersaturated carbonated solution. However, we observe little to no collection effects in the surfactant solutions and beer.

The microbubbles generated in the supersaturated solution may have hydrophobic surface properties different from those of the microbubbles generated in the surfactant solution, making it easier for hydrophobic CNTs to attach to their surface for separation. Based on the results, the method may be applied to the separation of hydrophilic and hydrophobic materials. Next, we show the separation of materials with different surface properties. aCNTs and CNTs having different surface properties are used as the separation materials. A mixed solution of CNTs (0.1 mg/ml) and aCNTs (0.1 mg/ml) is dispersed in DI water and sonicated at room temperature for 1 h to obtain a homogeneous suspension. Afterward, a mixed suspension of CNTs and aCNTs is added to a colorimetric tube containing 25% ethanol solution, and the pH of the mixed suspension of CNTs and aCNTs in the 25% ethanol solution is measured, which is 7.4. The dense microbubbles are generated in the solution by subjecting the colorimetric tube surface to impact. The results show that a part of the sample floated on the surface of the solution and another part remained in the solution after impact [Fig. 3(a)].

Fourier-transform infrared spectroscopy (FTIR) is performed to investigate the sample remaining in the solution and the sample floating on the surface of the solution, as shown in Fig. 3(b–I). Compared with the sample floating on the surface of the solution, the sample remaining in the solution, which has a band at 1722 cm⁻¹, is assigned to the C=O stretching vibration. C=O indicates that the carboxyl groups are bonded to the CNTs after acidification. A band at 3471 cm⁻¹ assigned to the O—H stretching vibration is also observed. Compared with the spectrum of CNTs and aCNTs, 98% CNTs and 94% aCNTs can be recovered from the upper and lower layers of the solution, based on the C=O peak intensity ratio of CNTs to aCNTs of the FTIR spectrum, as shown in Figs. 3(b-I) and 3(b-II). The results indicate that the sample remaining in the solution is aCNTs and the sample collected by the microbubbles is CNTs. The method provides a simple and effective way to separate materials with hydrophilic or hydrophobic surfaces.

The supersaturated gas in the solution is easily released by physical disturbance and causes bubbles generation.³⁰ Therefore, the change in the solubility of the gas in different concentrations of ethanol solutions becomes the main reason for bubble generation. The Ostwald coefficient is used to express the solubility of the gas in the solution and is defined as the ratio of the volume of gas absorbed to the volume of the absorbing liquid.³¹ The Ostwald coefficients of gas (nitrogen) are 0.151, 0.0622, 0.0363, and 0.0217 in the ethanol solution with mole fractions 1.0, 0.46, 0.27, and 0.1 (at 20 °C), respectively.³¹ DI water is added to the ethanol solution to reduce the mole fraction from the value of 1.0 to 0.1, which caused the Ostwald coefficient to decrease the most. This process causes excessive gas release in the solution to form a supersaturated solution (0.717 ml of gas is released).

In this method, the process of subjecting the supersaturated solution to impact may cause shear force in the solution and release gas to generate microbubbles.^32,33 To study the relation between the shear force applied to the supersaturated solution and the generation of bubbles, a disperser (agitator shaft: S10N-10 G, T10 basic ULTRA-TURRAX® IKA) that provides a stable shear force is applied to the solution to release gas and generate bubbles from the solution [Fig. 4(a–I)]. The relative intensity of the transmitted light of bubbles is used to describe the quantity of generated bubbles. Shear force is applied to DI water, 1% DDAB solution, 1% SDS solution, 1% Pluronic® F-127, and 25% ethanol solution (all use v/v) to induce the generation of microbubbles. After applying a shear force of 25 Pa for 20 s (rotational speed: 6700 rpm), the generated bubbles in the 25% ethanol solution gradually decrease with time, while the generated bubbles in the surfactant solutions (surfactant trends follow 1% SDS solution) increase with time [Fig. 4(a–II)]. After applying shear force, the 25% ethanol solution is then subjected to free-fall impact to induce microbubble generation. We apply shear force for 15 s, 40 s, and 80 s durations to 25% ethanol solutions separately. The microbubbles generated by the impact decreases as the duration of the shear force increases [Inset of Fig. 4(b)]. Dense microbubbles cannot be generated by subjecting the solution to impact because the gas in the supersaturated solution is driven out during shear force.

To compare the collection effects of CNTs under different degrees of supersaturation of the gas, shear force is separately applied to 25% ethanol solutions by the disperser for 15 s, 40 s, and 80 s durations. After shear force is applied to the solutions, the CNT suspension (0.2 mg/ml) is added to the solutions. The container is then subjected to impact to generate microbubbles to collect CNTs with recovery ratios of 94%, 64%, and 22%, respectively [Fig. 4(b)]. The results show that by applying shear force, the supersaturation state can be changed to reduce the microbubbles generated by the impact, thereby reducing the CNT collection amount.

The microbubbles generated in 25% ethanol could effectively collect CNTs, which may be due to the high hydrophobicity of CNTs. Therefore, the hydrophobicities of CNTs and aCNTs are measured by the MATH test. CNTs show high hydrophobicity compared to aCNTs [Fig. 5(a)], wherein CNTs are trapped at the interface between the hydrocarbon phase and the aqueous phase while showing clarity in the aqueous phase of the CNT samples, as shown in the inset of Fig. 5(a).

It has been confirmed that CNTs can be attached to the microbubbles generated in the 25% ethanol solution and float to the surface of the solution with the microbubbles. After subjecting the mixed solution of 25% ethanol and CNT suspension to impact, the microbubbles in the upper layer of the solution are investigated. The CNTs attached on the surface of microbubbles can be observed by microscopy, as shown in Fig. 5(b).

According to Fig. 2(a), surfactant solutions can generate dense microbubbles on impact, but it is impossible to separate CNTs effectively. Hence, we investigate the properties of microbubble surfaces and their effect on collection of CNTs. After subjecting the colorimetric tube containing the surfactant solution to impact, the surfactant is dispersed in the solution and adsorbed at the interface between the microbubble and the solution, causing a microbubble surface with hydrophilic properties, as shown in Fig. 5(c–I). The microbubbles generated in the ethanol solution do not have a hydrophilic surface because no surfactant is adsorbed on the interface between the microbubble and the solution [Fig. 5(c–II)]. According to the principle of attachment between hydrophobic objects and hydrophobic bubbles,³⁴ we investigate whether microbubbles generated in the 25% ethanol solution are attached to the hydrophobic surface or the hydrophilic surface. The hydrophobic inner wall (contact angle: 93.8°) of the colorimetric tube is treated by plasma to change the property to hydrophilic (contact angle: 21.8°). After subjecting the colorimetric tube containing the 25% ethanol solution to impact, the generated microbubbles are easily attached to the hydrophobic inner wall, as shown in Fig. 5(d–I). However, the generated microbubbles do not attach to the hydrophilic modified inner wall [Fig. 5(d–II)]. These results indicate that the surface property of the microbubbles generated in the 25% ethanol solution is hydrophobic.

To confirm whether the recovery of CNTs by microbubbles is caused by the hydrophobic surface effect, we prepare microbubbles with different surface properties. 1% SDS solution is added to the 25% ethanol solution to generate dense microbubbles by subjecting the container to impact. We compare the dynamics of microbubbles in the 25% ethanol solution and those in the SDS added 25% ethanol solution. Dense microbubbles can be generated in both solutions, as shown in Fig. 5(e–I). Compared with hydrophobic microbubbles generated in the 25% ethanol solution, microbubbles generated in the SDS added 25% ethanol solution do not attach to the hydrophobic inner wall of the colorimetric tube after microbubble dissipation [Fig. 5(e–II)]. The results show that different surface properties of microbubbles are generated in the 25% ethanol solution and the SDS added 25% ethanol solution. Moreover, to investigate the relation between the collection effect of CNTs and microbubbles with or without the hydrophobic surface, the CNT suspension with the same concentration of 0.2 mg/ml is added to the 25% ethanol solution and the SDS added 25% ethanol solution separately [Fig. 5(e–III)]. In contrast to the result of collecting CNTs by microbubbles generated in the 25% ethanol solution, CNTs cannot be collected even if dense microbubbles are generated in the SDS added 25% ethanol solution, as shown in Fig. 5(e–IV).

In summary, we report a method to generate dense microbubbles with an average diameter of 2 µm by subjecting an ethanol solution without chemical reagents to impact. Moreover, the surface property of the generated microbubbles is hydrophobic. The proposed method can effectively separate hydrophilic aCNTs and hydrophobic CNTs, based on the differences in the surface property with different components on the surface. This study could provide a way for further applications such as wastewater treatment, chemical engineering, and mining treatment.

The authors gratefully acknowledge the support for this work from the NSC, Grant No. 108-2112-M-002-015, funded by the Ministry of Science and Technology, Taiwan.

The data that support the findings of this study are available from the corresponding author upon reasonable request.