Microemulsion Encapsulated into Halloysite Nanotubes and their Applications for Cleaning of a Marble Surface

Featured Application: Halloysitenanotubesforsurfacecleaning. Abstract: Halloysite nanotubes were used to incorporate anionic surfactant micelles and an organic solvent to generate a cleaning system to be applied in Cultural Heritage restoration. The targeted adsorption is driven by electrostatic interactions based on the nanotubes peculiar charge separation. Namely anionic species are driven to the positively charged inner surface while being prevented from interacting with the halloysite outer surface that possesses a positive charge density. The hybrid organic/inorganic emulsion was characterized by dynamic light scattering. Analysis of the autocorrelation function allowed us to deﬁne the presence of surfactant aggregates inside/outside the nanotube lumen as a function of the nanotube/surfactant ratio in an aqueous mixture. The application of this prepared emulsion for the controlled cleaning of a marble artifact is demonstrated. To this purpose, a membrane of nanoﬁbrous polyacrylonitrile was prepared by electrospinning and was applied between the work of art and the cleaning agent to avoid the release of residues on the marble surface. This work represents a further step toward the opportunity to extend the use of emulsions for cleaning protocols for stone-based artifacts or in technological applications where surfactant separation is required by a simple centrifugation/sedimentation method.


Introduction
Microemulsions have largely attracted attention in cleaning for Cultural Heritage restoration, replacing conventional methods that are difficult to control. In particular, the commonly used organic solvents or gels are often toxic and high evaporation rates can represent a risk for the operators. Moreover, the aggressive nature of the organic solvents may cause partial swelling and solubilization of the original artwork materials [1]. On the other hand, microemulsions allow for better control of diffusion at the surface of the substrate and their formulations require fewer amounts of solvents halloysite nanoparticles in a quantitative way [50]. For the cleaning test, we selected an Egyptian Kilga, from Cairo, dating back to the 10th or 11th century, preserved today in the exhibition space dedicated to Islamic art at the Interdisciplinary Regional Gallery of Palazzo Abatellis. The Kilgas were probably born due to human need and survival, given the scarcity of fresh and pure water in the Nile Delta villages after a long period of drought and plague.

Materials
Halloysite nanotubes (HNTs) were obtained from Imerys Ceramics, scanning electron micrographs are provided in Supplementary Materials ( Figure S2). Sodium dodecyl sulfate (SDS), tetradecane, 1-pentanol and polyacrylonitrile (PAN, Mw = 150 kDa) were obtained from Sigma Aldrich (Milan, Italy). PAN membranes were prepared using an electrospinning method after PAN was dissolved in dimethyl sulfoxide (DMSO, purchased from Fisher Scientific) as a solvent and mixing for 12 hours. The PAN solution was electrospun onto a rotating drum (300 rpm) via an electrospinning device (NB-EN1, NanoBond, China), containing a metallic needle, a feed pump and a high voltage power supply. The sample was electrospun with a solution flow rate of 1.4 mL/h, at a distance of 15 cm and voltage of 13-14 kV. A detailed description of the preparation method has been explained in our previous study [50].

Preparation of Oil in Water Emulsion
An oil in water emulsion was prepared based on literature findings [51]. Briefly, water was mixed with tetradecane in different ratios. SDS and 1-pentanol, as co-surfactants, were added and the mixture was kept under magnetic stirring overnight. The investigated concentrations are provided in Table 1, together with the tetradecane/SDS weight ratios (R oil/SDS ).

Preparation of Oil in Water Emulsion in the Presence of Halloysite
The microemulsion with R oil/SDS = 2.63 was used for incorporation into HNTs provided that the average size for the HNT lumen can be as small as~15 nm for our sample [52], based on the microemulsion size. Variable amounts of nanotubes have been explored in the range from 0.00125 to 0.043% in water.
The halloysite was added to the designed microemulsion without previous treatments, before incubation in an ultrasonic bath (5 minutes) to promote the loading of the micelles into the nanotubes.

Methods
Dynamic light scattering (DLS) experiments were carried out using a Zetasizer NANO-ZS (Malvern Instruments, London, UK) at 25 • C. The field-time autocorrelation functions were well described by a single or double exponential decay (Equation (1)), which provides the decay rate (Γ) of each diffusive mode and its intensity weight.
where B and C are pre-exponential factors, correlated to the concentration of the particles. From the decay rates, we calculated the collective diffusion coefficients at a given concentration as D t = Γ/q 2 where q is the scattering vector given by 4πnλ −1 sin(θ/2) where n is the water refractive index, λ is the wavelength (632.8 nm) and θ is the scattering angle (173 • ). The apparent hydrodynamic radii (Rh) of the equivalent (with equivalent translational diffusion coefficients) spherical particles was calculated using the Stokes-Einstein equation as Rh = k b T/(6πη), where k b is the Boltzmann constant, T is the absolute temperature, and η is the water viscosity.
From the pre-exponential factors, we calculated B/(B + C) and C/(B + C) as parameters providing the particle concentrations, proportional to a % of the two populations.
A microscope ESEM FEI QUANTA 200F (FEI Company, Cambridge, UK) was used to image the halloysite nanotubes, after gold coating in argon by means of an Sputter Coater S150A (EDWARDS, Burgess Hill, UK). Measurements were conducted in high vacuum (<6 × 10 −4 Pa), at 15 kV and a working distance of 10 mm.

Oil in Water Emulsion
The oil in water emulsions were characterized by DLS to identify the best conditions to prepare oil droplet sizes for loading into the HNT lumen. Figure 1 shows that all the hydrodynamic radii of the prepared suspensions were in the range of 4-22 nm, suggesting the possibility of an efficient entrapment of the oil droplets into the nanotubes. The hydrodynamic volume ( Figure 1) was calculated, assuming spherical-like micelles, and was reported as a function of R oil/SDS . We observed a linear increase in volume as the oil content increased, underlining the crucial role of oil in droplet size.
Furthermore, a first evaluation of the temporal stability of the prepared quaternary systems was carried out repeating the DLS measurement after 24 hours of storage at 25 • C. Retention of 99% of the initial size was obtained, showing the colloidal stability of the mixture.
The ζ potential for the microemulsions provided a negative value of~−12 mV. These findings suggest the presence of a negative charge at the micelle/water interface, which guarantees the selective electrostatic interaction with HNT positive inner surface.
where B and C are pre-exponential factors, correlated to the concentration of the particles. From the decay rates, we calculated the collective diffusion coefficients at a given concentration as Dt = Γ/q 2 where q is the scattering vector given by 4πnλ −1 sin(θ/2) where n is the water refractive index, λ is the wavelength (632.8 nm) and θ is the scattering angle (173°). The apparent hydrodynamic radii (Rh) of the equivalent (with equivalent translational diffusion coefficients) spherical particles was calculated using the Stokes-Einstein equation as Rh = kbT/(6πη), where kb is the Boltzmann constant, T is the absolute temperature, and η is the water viscosity.
From the pre-exponential factors, we calculated B/(B + C) and C/(B + C) as parameters providing the particle concentrations, proportional to a % of the two populations.
A microscope ESEM FEI QUANTA 200F (FEI Company, Cambridge, UK) was used to image the halloysite nanotubes, after gold coating in argon by means of an Sputter Coater S150A (EDWARDS, Burgess Hill, UK). Measurements were conducted in high vacuum (<6 × 10 −4 Pa), at 15 kV and a working distance of 10 mm.

Oil in Water Emulsion
The oil in water emulsions were characterized by DLS to identify the best conditions to prepare oil droplet sizes for loading into the HNT lumen. Figure 1 shows that all the hydrodynamic radii of the prepared suspensions were in the range of 4-22 nm, suggesting the possibility of an efficient entrapment of the oil droplets into the nanotubes. The hydrodynamic volume ( Figure 1) was calculated, assuming spherical-like micelles, and was reported as a function of Roil/SDS. We observed a linear increase in volume as the oil content increased, underlining the crucial role of oil in droplet size.
Furthermore, a first evaluation of the temporal stability of the prepared quaternary systems was carried out repeating the DLS measurement after 24 hours of storage at 25 °C. Retention of 99% of the initial size was obtained, showing the colloidal stability of the mixture.
The ζ potential for the microemulsions provided a negative value of ~−12 mV. These findings suggest the presence of a negative charge at the micelle/water interface, which guarantees the selective electrostatic interaction with HNT positive inner surface.

Oil in Water Emulsion in the Presence of HNT
We explored the possibility of driving the oil droplet inclusion into the HNT lumen by electrostatic interactions. On this basis, an anion surfactant (SDS) was selected because it drives interactions with the positive charges of the HNT's inner surface. Figure 2 displays the autocorrelation function, g(τ), obtained for the prepared suspensions with various HNT amounts in the microemulsion. A double exponential decay is observed and is correlated to the presence of two particle populations namely, oil in water droplets and HNTs. In particular, the initial decay is associated with fast particles (microemulsion) and as time progresses, the decay is associated with slow particles (HNTs). Typically, the particle volume is the parameter that most affects DLS behavior. We also observed an increase in g(τ) intensity for the second decay as the HNT content increased. We explored the possibility of driving the oil droplet inclusion into the HNT lumen by electrostatic interactions. On this basis, an anion surfactant (SDS) was selected because it drives interactions with the positive charges of the HNT's inner surface. Figure 2 displays the autocorrelation function, g(τ), obtained for the prepared suspensions with various HNT amounts in the microemulsion. A double exponential decay is observed and is correlated to the presence of two particle populations namely, oil in water droplets and HNTs. In particular, the initial decay is associated with fast particles (microemulsion) and as time progresses, the decay is associated with slow particles (HNTs). Typically, the particle volume is the parameter that most affects DLS behavior. We also observed an increase in g(τ) intensity for the second decay as the HNT content increased. The hydrodynamic radii, calculated from the field-time autocorrelation functions, were reported as a function of % HNT content (Figure 3a). The increase of HNT content in the investigated suspensions appears to have a negligible effect on the Rh of the two particle populations, indicating a type of interaction that does not involve a significant change in particle size. These findings may be correlated to an electrostatic adsorption of negative oil droplets into the positively charged HNT lumen. Furthermore, the g(τ) function provides information about the concentration of the dispersed particles (B and C parameters in Equation (1)). Figure 3b shows the concentration percentage of the two populations as the HNT content increases. The microemulsion decreases, and the content of nanotubes increases, with a non-linear trend. This behavior suggests that the oil droplets have been partially driven into the halloysite nanocontainer lumen. The hydrodynamic radii, calculated from the field-time autocorrelation functions, were reported as a function of % HNT content (Figure 3a). The increase of HNT content in the investigated suspensions appears to have a negligible effect on the Rh of the two particle populations, indicating a type of interaction that does not involve a significant change in particle size. These findings may be correlated to an electrostatic adsorption of negative oil droplets into the positively charged HNT lumen. Furthermore, the g(τ) function provides information about the concentration of the dispersed particles (B and C parameters in Equation (1)). Figure 3b shows the concentration percentage of the two populations as the HNT content increases. The microemulsion decreases, and the content of nanotubes increases, with a non-linear trend. This behavior suggests that the oil droplets have been partially driven into the halloysite nanocontainer lumen. The oil droplet inclusion into the HNT lumen is shown in Figure 4 and represents the B/C ratio as a function of the SDS/HNT ratio. We observe a linear increase if the interaction between particles, ideally, does not involve any changes in their concentration. On the contrary, it was found that for RSDS/HNT  915, as SDS content increases, the B factor increases at a lower rate. This can be explained by considering a possible interaction involving an adsorption of microemulsion into the HNT lumen. In this case, the loaded oil droplets do not scatter, resulting in a slower increase to the B/C ratio.
In particular, for RSDS/HNT  1000 (0.02% of HNT), it is possible to hypothesize the instauration of an equilibrium between micelles dispersed in solution and loaded into the nanotubes. The value of RSDS/HNT = 1000 can be considered as a parameter for the estimation of the saturation limit. In fact, with a higher amount of SDS, the B/C rises linearly as in absence of any interaction.

Cleaning Tests on a Marble Sculpture
The Kilgas are decorative supports for jade, formed by an octagonal trunk of hollow marble. The end of the trunk closes with a solid base that continues through a short pass inclined through a The oil droplet inclusion into the HNT lumen is shown in Figure 4 and represents the B/C ratio as a function of the SDS/HNT ratio. We observe a linear increase if the interaction between particles, ideally, does not involve any changes in their concentration. On the contrary, it was found that for R SDS/HNT ≤ 915, as SDS content increases, the B factor increases at a lower rate. This can be explained by considering a possible interaction involving an adsorption of microemulsion into the HNT lumen. In this case, the loaded oil droplets do not scatter, resulting in a slower increase to the B/C ratio.
In particular, for R SDS/HNT ≤ 1000 (0.02% of HNT), it is possible to hypothesize the instauration of an equilibrium between micelles dispersed in solution and loaded into the nanotubes. The value of R SDS/HNT = 1000 can be considered as a parameter for the estimation of the saturation limit. In fact, with a higher amount of SDS, the B/C rises linearly as in absence of any interaction. The oil droplet inclusion into the HNT lumen is shown in Figure 4 and represents the B/C ratio as a function of the SDS/HNT ratio. We observe a linear increase if the interaction between particles, ideally, does not involve any changes in their concentration. On the contrary, it was found that for RSDS/HNT  915, as SDS content increases, the B factor increases at a lower rate. This can be explained by considering a possible interaction involving an adsorption of microemulsion into the HNT lumen. In this case, the loaded oil droplets do not scatter, resulting in a slower increase to the B/C ratio.
In particular, for RSDS/HNT  1000 (0.02% of HNT), it is possible to hypothesize the instauration of an equilibrium between micelles dispersed in solution and loaded into the nanotubes. The value of RSDS/HNT = 1000 can be considered as a parameter for the estimation of the saturation limit. In fact, with a higher amount of SDS, the B/C rises linearly as in absence of any interaction.

Cleaning Tests on a Marble Sculpture
The Kilgas are decorative supports for jade, formed by an octagonal trunk of hollow marble. The end of the trunk closes with a solid base that continues through a short pass inclined through a

Cleaning Tests on a Marble Sculpture
The Kilgas are decorative supports for jade, formed by an octagonal trunk of hollow marble. The end of the trunk closes with a solid base that continues through a short pass inclined through a tricuspid bow opening in a protruding frontal basin. Four stylized legs support the octagonal trunk. A jar full of water was placed vertically on the trunk with the water flowing through the porous body of the vessel into the protruding mouth and was collected by using cups.
The exterior of the trunk was decorated with Islamic traditions, such as the muqarnas, which are found in many works in the Palermo area, including the wooden roof muqarnas of the Palatine Chapel.
The first cleaning step had been mechanical, with the use of an aspirator set to low power and a brusher for the removal of superficial deposits. The stone surface was visibly shiny and greasy to the touch. A coherent film was deposited on the marble surface and by means of a Frontier FTIR spectrometer (PerkinElmer, Italy, Milan). analysis shows that the main constituent is an oil-based varnish, likely due to a protective coating applied during a previous restoration.
To achieve fine control of the release of the product onto the stone surface, a membrane of PAN nanofibers was used. The presence of PAN nanofibers ensures the absence of direct contact between the HNTs and the stone surface thus avoiding the release of residues on the surface. Moreover, this protocol guarantees a great control of the cleaning area; the part that has first been treated with a soft bristle brush and rinsed with de-ionized water. A first evaluation of the prepared cleaning system is shown in Figure 5, displaying the treated area before and after cleaning. A second examination of the stone surface, by FT-IR, confirmed the removal of the oily layer from the same work of art (spectra are in Supplementary Materials, Figure S1). Figure 6 shows the full view of the Kilga before and after the cleaning treatment. tricuspid bow opening in a protruding frontal basin. Four stylized legs support the octagonal trunk. A jar full of water was placed vertically on the trunk with the water flowing through the porous body of the vessel into the protruding mouth and was collected by using cups. The exterior of the trunk was decorated with Islamic traditions, such as the muqarnas, which are found in many works in the Palermo area, including the wooden roof muqarnas of the Palatine Chapel.
The first cleaning step had been mechanical, with the use of an aspirator set to low power and a brusher for the removal of superficial deposits. The stone surface was visibly shiny and greasy to the touch. A coherent film was deposited on the marble surface and by means of a Frontier FTIR spectrometer (PerkinElmer, Italy, Milan). analysis shows that the main constituent is an oil-based varnish, likely due to a protective coating applied during a previous restoration.
To achieve fine control of the release of the product onto the stone surface, a membrane of PAN nanofibers was used. The presence of PAN nanofibers ensures the absence of direct contact between the HNTs and the stone surface thus avoiding the release of residues on the surface. Moreover, this protocol guarantees a great control of the cleaning area; the part that has first been treated with a soft bristle brush and rinsed with de-ionized water. A first evaluation of the prepared cleaning system is shown in Figure 5, displaying the treated area before and after cleaning. A second examination of the stone surface, by FT-IR, confirmed the removal of the oily layer from the same work of art (spectra are in Supplementary Materials, Figure S1). Figure 6 shows the full view of the Kilga before and after the cleaning treatment.  tricuspid bow opening in a protruding frontal basin. Four stylized legs support the octagonal trunk. A jar full of water was placed vertically on the trunk with the water flowing through the porous body of the vessel into the protruding mouth and was collected by using cups. The exterior of the trunk was decorated with Islamic traditions, such as the muqarnas, which are found in many works in the Palermo area, including the wooden roof muqarnas of the Palatine Chapel.
The first cleaning step had been mechanical, with the use of an aspirator set to low power and a brusher for the removal of superficial deposits. The stone surface was visibly shiny and greasy to the touch. A coherent film was deposited on the marble surface and by means of a Frontier FTIR spectrometer (PerkinElmer, Italy, Milan). analysis shows that the main constituent is an oil-based varnish, likely due to a protective coating applied during a previous restoration.
To achieve fine control of the release of the product onto the stone surface, a membrane of PAN nanofibers was used. The presence of PAN nanofibers ensures the absence of direct contact between the HNTs and the stone surface thus avoiding the release of residues on the surface. Moreover, this protocol guarantees a great control of the cleaning area; the part that has first been treated with a soft bristle brush and rinsed with de-ionized water. A first evaluation of the prepared cleaning system is shown in Figure 5, displaying the treated area before and after cleaning. A second examination of the stone surface, by FT-IR, confirmed the removal of the oily layer from the same work of art (spectra are in Supplementary Materials, Figure S1). Figure 6 shows the full view of the Kilga before and after the cleaning treatment.

Conclusions
Colloidal dispersions of HNTs in microemulsion were prepared by mixing components under ultrasonic treatment. The micelles encapsulation was characterized by DLS. The prepared microemulsion showed hydrodynamic radii in the range of 4-22 nm, which was found to be suitable for efficient entrapment into the nanotubes. From the analysis of the autocorrelation function, we have confirmed the presence of surfactant aggregates inside/outside the nanotubes lumen as a function of the nanotubes/surfactant ratio in an aqueous mixture. In particular, an equilibrium between micelles dispersed in solution and loaded into the nanotubes is confirmed up until the saturation limit of R SDS/HNT ≈ 1000. The application of the prepared emulsion for the controlled cleaning of a marble artifact is demonstrated. To this purpose, a membrane of nanofibrous PAN was prepared by electrospinning and applied between the work of art and the cleaning agent to avoid the release of residues onto the marble surface. The results obtained indicate good cleaning control and efficacy.

Conflicts of Interest:
The authors declare no conflicts of interest.