Effects of chemical dispersant on the surface properties of kaolin and aggregation with spilled oil

After oil spills occur, dispersed oil droplets can collide with suspended particles in the water column to form the oil-mineral aggregate (OMA) and settle to the seafloor. However, only a few studies have concerned the effect of chemical dispersant on this process. In this paper, the mechanism by which dispersant affects the surface properties of kaolin and the viscosity and oil-seawater interfacial tension (IFTow) of Roncador crude oil were separately investigated by small-scale tests. The results indicated that the presence of dispersant impairs the zeta potential and enhances the hydrophobicity of kaolin. The viscosity of Roncador crude oil rose slightly as the dosage of dispersant increased, while IFTow decreased significantly. Furthermore, the oil dispersion and OMA formation at different dispersant-to-oil ratio (DOR) were evaluated in a wave tank. When DOR was less than 1:40, the effect of dispersant on the dispersion of spilled oil was not obvious. With the increasing DOR, the effect became more pronounced, and the adhesion between oil droplets and kaolin was inhibited. The size ratio between oil droplets and particles is the significant factor for OMA formation. The closer the oil-mineral size ratio is to 1, the more difficultly the OMA forms.


Introduction
With the increase of global demand for petroleum products and maritime transportation, many petroleum hydrocarbon contaminants entered the marine environment through various oil spill accidents (He et al., 2018;Qi et al., 2018). Because of the toxic effect of oil on the aquatic ecology, it is necessary to clean it up in time once the oil spill accident occurs (French et al., 2004). The application of oil dispersant is a traditional means of oil spill removal (Prince and Butler, 2014). During the Deepwater Horizon spill in 2010, more than 2 million gallons of oil dispersant were sprayed onto the oil using aircraft and ships (McNutt et al., 2012;Hansel et al., 2015). The dispersant contains amphiphilic molecules ("surfactants") in a solvent base (Fernandes et al., 2019), and the surfactants break up the oil slick into micron-sized droplets, which are further carried below the water surface by waves. The oil is degraded by various microorganisms present in the water column (Li et al., 2008).
However, in estuaries and coastal waters, due to the high concentration of particulate matter in the water column and the strong shear effect of waves, dispersed oil droplets are easy to aggregate with suspended particulate matter to form Responsible Editor: Philippe Garrigues the oil-mineral aggregate (OMA), thus leading to the sinking of spilled oil (Fitzpatrick et al., 2015;Silva et al., 2019). Under appropriate conditions, up to 20% of the spilled oil can interact with particulate matter, and the oil content of the formed OMA after spraying oil dispersant accounts for 65% of the total spilled oil (Bandara et al., 2011). As polycyclic aromatic hydrocarbons (PAHs) persist much longer in the sediment phase than at the sea surface (Harayama et al., 1999), these oil-bearing aggregates will exist on the seabed for a long time, resulting in excessive petroleum hydrocarbon contaminant, especially toxic PAHs, posing a serious threat to both the benthic communities and marine environment (Gao et al., 2019;Qi et al., 2021;Suneel et al., 2021). Common oil spill responses are mostly applicable to the floating oil but ineffective for the sunken oil (IMO, 2012;Usher, 2009). Remarkably, the sunken oil cannot be easily monitored by typical visual observation or remote sensing techniques Hammoud et al., 2019), posing challenges to the cleaning operations (Jacqueline, 2008).
Except for the physical adsorption force, there may be a more potent chemical force between oil droplets and particles (Lambert and Variano, 2016); thus, the morphological structure of OMA is usually stable. Zhao et al. (2017) analyzed the 3D structure of OMA using a confocal microscope imaging technology and found that particles could be embedded into the oil droplets at depth of 2-10 μm. The factors that affect the OMA formation are very complex, such as oil characteristics (concentration, droplets size, viscosity, density, and polar component), particulate matter characteristics (concentration, particle size, density, hydrophobicity, composition, and zeta potential), and environmental conditions (temperature, salinity, and mixing energy) (Loh et al., 2014;Gao et al., 2018). Many scholars have studied the above influencing factors on OMA formation and summarized some empirical results. Hua et al. (2018) proved that oil viscosity is a key factor affecting the aggregation process of oil and suspended particulate matter, and lower oil viscosity was conducive to the stable formation of OMA. Moreira et al. (2015) found that polar components (asphaltenes and resins) in oil can improve the adhesion between oil droplets and suspended particles, thus contributing to the formation of OMA. Yu et al. (2019a) found that under the same mass concentration, more oil droplets can be trapped by smaller particles, and larger particles tend to interact more quickly with oil droplets. Strong hydrophobicity of solid particles can promote their affinity with the oil droplets, making them easier to form OMA (Loh and Yim, 2016). Unfortunately, it remains unknown which of the above factors has a more significant effect on the oil-mineral aggregation.
Besides, few studies have focused on the formation of OMA in the presence of chemical dispersant, and related conclusions have been still a matter of controversy. Khelifa et al. (2008) proposed that in the marine environment where the mixing energy and suspended particulate concentration are usually low, the use of oil dispersant would be beneficial to the formation of OMA. The surfactants in the dispersant can significantly reduce the oil-water interfacial tension and viscosity, making the spilled oil disperse into a large number of emulsified droplets and maintain stable (Cai et al., 2017). Li et al. (2007) found that the synergistic effect of dispersant and mineral fines enhances the transfer of oil from the surface downward into the water column. However, some studies have found that excessive dispersant can inhibit the formation of OMA (Yu et al., 2019a;Sorensen et al., 2014;Page et al., 2002), but the specific cause has not been confirmed. Furthermore, previous studies often ignored the effect of dispersant on the physicochemical properties of oil and minerals, so this study will start from these aspects to fill the knowledge gap.
The overall goal of this study is to investigate the effect of chemical dispersant on the surface properties of minerals and spilled oil and determine the mechanism of OMA formation in the presence of chemical dispersant. The specific objectives are to (1) examine the variation of zeta potential, cation exchange capacity (CEC) and contact angle (CA) of minerals, and viscosity of oil and oil-water interfacial tensions (IFT ow ) with dispersant-to-oil ratios (DOR); (2) evaluate the dispersion and sedimentation of oil with medium viscosity in a wave tank at different DOR; and (3) analyze the influencing factors of the OMA formation.

Materials
The Roncador crude oil from Brazil was selected in this study due to its excellent sinking performance. The density and kinematic viscosity (ν k ) were determined using a petroleum densitometer (SY-05, Guangzhou WanMu Instrument Equipment Co., Ltd, Foshan City, China) and kinematic viscosimeter (BF-03C, Dalian North Analytical Instrument Co., Ltd, Dalian, China), respectively. The IFT ow was measured by an Automatic Surface Tensiometer (BZY-101) referring to the standard (GB 6541-1986). The content of asphaltene and sulfur was measured according to the standard methods (NB/SH/T 0509-2010). Detailed physicochemical properties are summarized in Table 1.
The white kaolin purchased from Andi Metal Materials was selected in this study for the following two reasons: first, kaolin was a representative mineral that has been widely used in most previous OMA experiments (Li et al., 2007;Wang et al., 2013;Ji et al., 2021;Yu et al., 2021); second, although montmorillonite is more widely distributed in ocean, the swelling behavior of it may bring unpredictable interference when the effect of dispersant on the surface properties is studied (Yotsuji et al., 2021). The particle size was measured by a laser diffraction instrument (LISST-100X; Sequoia, Seattle, WA), and the mean size was 12.46 μm.
The chemical dispersant used in this study was GM-2, a non-ionic oil spill dispersant provided by the Qingdao Guangming Environmental Protection Technology Co. The surfactant is mainly fatty alcohol polyoxyethylene ether, and the solvent is polyethylene glycol. The pH, ignition point and υ k (30 °C) of GM-2 are 7-7.5, 70 °C and 50 mm2/s, respectively. The seawater was obtained from Xinghai Bay, Dalian, China (39°5′51.65"N and 121°50′95.19"W) with a salinity of 30‰. Before each experiment, it was filtered through 0.45 μm pore.

Modification of kaolin
Briefly, 300 mL seawater and 3 g kaolin were added into a 500-mL baffled flask, leading to a particle concentration of 10 mg/mL. After fully mixing, certain amounts of chemical dispersant were added based on a series of dispersant-tomineral ratios (DMR). DMR were set to 1:150, 1:75, 1:38, 1:30, and 1:15, respectively. Then the dispersant-kaolin-seawater mixtures were shaken at 200 rpm in a reciprocating shaker for 12 h. The supernatants were discharged after the mixtures were left for 6 h at 15 °C. The bottom modified kaolin (MK) was repeatedly washed by deionized water and dried at 50 °C for 12 h.

Modification of oil viscosity
According to different DOR (0, 1:100, 1:50, 1:10, 1:8, and 1:5), 15 g fresh oil and a certain quality of dispersant were stirred evenly in a 100-mL beaker, placed in an oscillator to shake for 30 min, and then left for 6 h at 15 °C in a standing incubator. The viscosity of oil was measured using the viscometer (BF-03C, North Dalian, China) and the capillary viscometer (φ = 2 mm, the viscometer constant is 1.024 mm 2 /s 2 ). The measurement was repeated at least three times for each sample.
After stirring for 30 min, the samples were left for 6 h at 15 °C. The IFT ow were measured using the automatic surface tensiometer (BZY-101, SHfangrui, China). The measurement was repeated at least three times for each sample.

Analysis of zeta potential
The zeta potential of kaolin and MK in seawater and deionized water was measured using the zeta potentiometer (JS94J2M, Global Hengda Scientific, Beijing, China), respectively. The measurement was repeated at least three times for each sample.

Analysis of CEC
CEC mainly characterizes the exchangeability between the cations on the surface of particles and those in solution and also plays an important role in the expansion of minerals. CEC was determined by the barium clay method: (1) 0.3 g of minerals was added in a dry 10-mL centrifuge tube, weighed as g 1 (g). Then 5 mL of BaCl 2 solution (0.5 mol/L) was added into it and vibrated for 1 min, centrifuged at 10,000 rpm for 5 min, and the supernatant was sucked out. Distilled water was added for centrifugation at 10,000 rpm for 5 min, and the supernatant was sucked out. The total weight of the centrifuge tube and wet minerals was denoted as g 2 (g); (2) 7 mL of H 2 SO 4 solution (0.025 mol/L) was added in the centrifuge tube and vibrated for 1 min and then centrifuged at 10,000 rpm for 5 min, and 5 mL of the supernatant was sucked to a conical flask; (3) three droplets of phenolphthalein indicator were dropped into the conical flask, titrated until the solution kept red with NaOH solution (0.05 mol/L). The volume of NaOH solution consumed was denoted as V 2 (mL); (4) 5 mL of H 2 SO 4 solution (0.025 mol/L) was added in a new conical flask, and step (3) was repeated. The volume of NaOH solution consumed was denoted as V 1 (mL). CEC (mmol/100 g) was calculated by: where N is the concentration of NaOH solution (0.05 mol/L), L is the humidity correction term (g 2 -g 1 , g), and m is the mass of kaolin (0.3 g). (1)

Analysis of CA
The CA of kaolin and MK was measured using the optical contact angle measuring instrument (Theta Flex, Biolin Scientific, Finland). The measurement was repeated at least three times for each sample.

Oil-mineral aggregation experiments
Wave tank facility In this study, the oil dispersion and oil-mineral aggregation experiments were individually conducted in a mesoscale wave tank. Figure 1 illustrates the schematic of the wave tank facility. The dimensions were 1.5 m long, 0.3 m wide, and 0.5 m high. The temperature of seawater was adjusted to 15 °C, and the average water depth was 0.20 m. A motor-driven push-type wavemaker was situated at one end of the tank to generate waves. Furthermore, a wave absorber was placed oppositely to control the reflection of the waves. The wave energy in this tank is quantitatively characterized by the turbulent kinetic energy (TKE) and energy dissipation rate (ε), which were calculated as follows (Shen and Maa, 2017): where u', v', and w' are velocity fluctuations in x, y, and z directions, with u' = u -U, v' = v -V, and w' = w -W, respectively, and the overbar denotes the time average within the time duration T.
where C is the reading of energy at f = 1 Hz from the least square fitted line with a fixed slope of − 5/3 in the log-log plot of E(f) vs. f. A is the Kolmogorov universal constant selected as 0.7 (Liu et al., 2011). Calculated by Eq. (2) and Eq. (3), TKE and ε were 7.9 × 10 −3 m 2 /s 2 and 9.7 × 10 −3 m 2 / s 3 , respectively.

Oil droplets size measurement
In this study, the oil droplet size distribution (OSD) and volume concentration of emulsified oil (V oil ) were measured by a laser-diffraction instrument (LISST-100X, Type C, Sequoia Scientific, Seattle, WA, USA) situated 100 cm downstream from one side of the tank with the detection windows 7 cm below the surface. It was operated in realtime mode, and the dynamic oil-droplet size distribution was acquired every 3 s. The volumetric mean diameter (VMD) of the oil droplets was calculated by: where V is the total volume concentration and v i is the volume concentration of oil droplets in a size interval with average size d i .

Experimental procedures
The dispersed oil concentration in the water column was measured by an on-line monitor (E09 TD-4100XDC, Turner Designs Hydrocarbon Instruments, Fresno, CA, USA). The sampling time was set as 1, 5, 10, 20, 30, 40, 60, 90, and 120 min, respectively. There were 6 sampling points, represented by red dots in Fig. 1 (two different depths and three horizontal locations). According to the pre-experiment, DOR for the wave tank experiment was set as 0, 1:400, 1:100, 1:40, 1:20, and 1:10, respectively. Firstly, 9 g of kaolin was added into the wave tank with a particle concentration of 100 mg/L. Twenty grams of Roncador crude oil and a 1 Schematic representation of the mesoscale tank facility certain quality of dispersant were added to the water surface, and then the wave-making plate was started to make the oil droplets and minerals fully mixed and collided. The wavemaking plate was stopped after 2 h of operation and left standing for 12 h to make the formed OMA settle to the bottom naturally. After the oil slick on the surface was removed, the upper liquid was extracted with a vacuum pump until the remaining water depth was between 3 and 5 cm. Then all the sediments were removed and transferred to the separation funnel; 50 mL of n-hexane and 2 mL of dilute sulfuric acid were added. The funnel was constantly shaken for 5 min and left standing for 10 min. The extract was filtered through micro-fiber-glass filters and dried with anhydrous sodium sulfate, analyzed by absorbance (A) at a wavelength of 225 nm with UV spectrophotometry (Epoch 2, BioTek, Winooski, VT, USA) (Yu et al., 2019b). The function of the standard curve of petroleum hydrocarbons concentration for Roncador crude oil (C) was A = 0.0325C + 0.0029 (R 2 = 0.9999). The oil trapping efficiency (OTE) was used to quantitatively assess the OMA formation: where m 1 is the mass of the sunken oil (g) and m 2 is the mass of the dispersed oil in the water column at 120 min (g). Figure 2 shows the change of zeta potential and CEC of kaolin with various dosages of dispersant. With the increase of DMR from 0 to 1:75, the value of zeta potential in deionized water and seawater decreased from − 64.36 ± 1.15 to − 42.63 ± 2.02 mV and decreased from − 34.70 ± 5.40 to − 23.20 ± 6.38 mV, respectively, while CEC had no significant change. As DMR continued to increase, the decline of zeta potential in deionized water was no longer apparent, while CEC showed an initial increase followed by a drop. When DMR increased to 1:15, the value of zeta potential in seawater decreased to − 18.20 ± 4.34 mV. Overall, the magnitude of variation in CEC was not obvious. The minimum and maximum value was 8.68 ± 0.41 and 12.11 ± 0.63 mmol/100 g, respectively. The evolution of the zeta potential versus DMR was consistent with Cai et al. (2017), in which that increasing DMR from 0 to 1:10 modestly suppressed the zeta potential of the OS sediments (collected from the Ocean Springs National Seashore Park) from − 15.5 to − 13.5 mV. The reason for this phenomenon is that the non-ionic surfactant in the dispersant is adsorbed to the surface of the particle and then has a shielding effect

Effects of dispersant on the surface properties of kaolin
on the surface potential of the particle (Yoon et al., 2014). Besides, due to the directional induction of hydrophilic groups and lipophilic groups, the compressing double-layer process of particles is promoted to some extent, which leads to the migration of surface charge, so the self-polymerization phenomenon between particles becomes significant (Kang et al., 2019). For these reasons, the space of hydration film was effectively enlarged, leading to the increase of CEC. However, when the concentration of chemical dispersant was high, the cationic surfactant would be adsorbed onto the surface of particles and form a hydrophobic interfacial film with a certain strength. Thus, the osmotic expansion process was gradually arrested, and the swelling of particles was effectively inhibited (Murtaza et al., 2020). The ion exchange between the particle surface and water phase was inhibited by the hydrophobic interfacial film. Simultaneously, the interlayer space was also compressed, and the channel for the cation exchange became narrow. Both of them directly led to the decrease of CEC.
As shown in Fig. 3, the contact angle (CA) of kaolin increased from 11.78° to 37.89° as DMR rose from 0 to 1:15, which indicated that the kaolin became more hydrophobic. This was also caused by the formation of hydrophobic interfacial film after the addition of chemical dispersant. The larger the dosage of dispersant, the greater the strength of interface film, and the higher the corresponding hydrophobicity.

Effects of dispersant on the oil viscosity and oilwater interfacial tension
The oil viscosity (υ oil ) after addition of dispersant was assessed at DOR of 0, 1:100, 1:50, 1:10, 1:8, and 1:5, respectively (Fig. 4a). As DOR increased from 0 to 1:10, υ oil increased from 208.38 ± 1.04 to 225.28 ± 4.31 mm 2 /s Fig. 2 Zeta potential and CEC of kaolin particles at different DMR at 50 °C and from 65.54 ± 1.66 to 74.75 ± 4.09 mm 2 /s at 80 °C, respectively. However, when DOR continued to grow, υ oil at 50 °C remained stable, while υ oil at 80 °C continued to increase. As DOR increased from 0 to 1:50, υ oil increased from 2298.44 ± 60.44 to 2476.21 ± 106.60 mm 2 /s at 15 °C; when DOR continued to grow to 1:5, υ oil increased to 2514.88 ± 80.64 mm 2 /s. When DOR increased from 0 to 1:5, the increase in oil viscosity ranged from 8 to 34%. This result was consistent with Khelifa et al. (2008), in which the increase in crude oil viscosity was 15% with the presence of dispersant. The oil viscosity reflects the intermolecular friction and is closely related to the size and structure of molecules (Baled et al., 2018). Asphaltene molecules in oil have a crystal structure similar to graphite. A large number of polymers can form through the intermolecular forces such as charge transfer, hydrogen bond, and dipole interaction and stably exist in the lamellar interlayer of asphaltene molecules (Shi et al., 2021;Fajardo-Rojas et al., 2020). This is also the main reason for the high viscosity of the oil. Because dispersant is mainly composed of surfactants and solvent with strong permeability, we speculate that the surface-active agent may make the polymer in the interlayer space of the asphaltene molecules more closely and orderly arranged and indirectly increase the stability of the asphaltene flake structure. The intermolecular friction increased accordingly, leading to a higher υ oil . However, the temperature has a more significant effect on the oil viscosity, and the viscosity decreases rapidly with the increase of oil temperature. Oil with good viscosity-temperature characteristic has a smaller range of viscosity varying with temperature.
IFT ow after addition of dispersant was studied at different DOR (Fig. 4b). IFT ow significantly decreased from 38.70 ± 1.01 to 10.13 ± 1.02 mN/m with the increase of dispersant dosage in the DOR range of 0-1:20. When DOR exceeded 1:20, no significant change of IFT ow was observed. It is well known that when dispersant is sprayed on the oil slick, the surfactant molecules are immediately adsorbed on the oil-water interface (Kujawinski et al., 2011). The lipophilic group extends to the oil phase, while the hydrophilic group extends to the aqueous phase, making the IFT ow greatly reduced and an interfacial film with certain strength forms. Brandvik et al. (2019) compiled data from several earlier studies and found that IFT ow decreased logarithmically with the increase of dispersant dosage, which was consistent with the results obtained in this experiment.

Effects of dispersant on the oil dispersion and OMA formation
As shown in Fig. 5, the dispersed oil concentration (C oil ) increased first and then stabilized with time in each condition with the presence of the dispersant. The more the dispersant was applied, the higher the C oil was. In the absence of dispersant, C oil changed very little and fluctuated in the range of 10-20 mg/L. When DOR was between 0 and 1:40, C oil increased slowly with time and reached the maximum value at 90 min and then remained stable. The maximum C oil at DOR of 1:40 was 2.5 times higher than that without dispersant. When DOR was 1:20 and 1:10, C oil both increased rapidly with time and reached the maximum at 90 min of 143.56 mg/L and 169.88 mg/L, respectively. This phenomenon was attributed to the chemical dispersant that stimulated the dispersion of two-ring and multiple-ring aromatic fractions of oil into the bulk aqueous phase (Li et al., 2007).
As shown in Fig. 6, the VMD of oil droplets under different conditions all decreased with time. However, the decay rate of VMD gave a slightly different result. When DOR was 1:400, the VMD decreased to 128.04 μm at 60 min and then reduced slowly to 120.65 μm at 120 min. A similar variation trend was also observed when DOR was 1:100. When DOR was over 1:100, the VMD decreased rapidly within 10 min. The more the dispersant was applied, the smaller the VMD was. When DOR was 1:10, the VMD decreased from 87.06 to 40.80 μm within 10 min and then fell to 31.05 μm at 120 min. Li et al. (2017) found that oil dispersant and suspended sediment can promote oil dispersion into the water column either individually or together through the wave tank experiment, and the presence of oil dispersant can reduce the average particle size of oil droplets from 150-200 to 50 μm. The larger the dispersant dosage, the smaller the IFT ow , making the oil droplets more easily dispersed. Figure 7 shows the OSD curve at 1 min and 120 min with different DOR. The OSD curve was unimodal at 1 min, and the higher the DOR, the smaller the peak value of VMD. When DOR was 1:400 and 1:100, the OSD curves were both bimodal at 120 min but transformed into unimodal when DOR increased. The peak values were 32.50, 27.60, and 23.40 μm at 120 min when DOR was 1:40, 1:20, and 1:10, respectively. When DOR was 1:10, the oil droplet size mainly concentrated between 12 and 32 μm at 120 min, close to the average size of kaolin (12.46 μm). Brandvik et al. (2013) observed that the peak value of dispersed oil droplets size distribution decreased significantly after adding dispersant, which was consistent with the results of this study.
Several studies have found that the addition of oil dispersant is helpful to the formation of OMA Loh et al., 2020). Unexpectedly, the OMA formation was gradually inhibited with the increase of dispersant dosage in this experiment (Fig. 8). When dispersant was not applied, the OTE was 28.59% and fell to 0.34% when DOR was 1:20. While the dispersant does not directly affect the partitioning of individual components (or groups of components) to the water phase, it does reduce the affinity of the oil droplets to the suspended particulate matter (Sorensen et al., 2014). Sorensen et al. (2014) considered that the reduction in adsorption of oil to suspended particulate matter might be attributed to the surfactants in the dispersant stabilizing the smaller oil droplets in the water column (and thus counteracting adsorption to particulates). However, other studies have found that chemical dispersant can promote the aggregation of oil with montmorillonite, quartz sand, and other particles to form OMA (Bandara et al., 2011;Khelifa et al., 2005). Therefore, Roncador crude oil and kaolin should be aggregated by adhesion force to form OMA. Although the chemical dispersant effectively promotes the dispersion of spilled oil and increases the oil concentration in the water column, it also dramatically reduces the size of the dispersed oil droplets, making the size between the minerals and the oil droplets closer. Since OMA majority performed as an oil droplet coated uniformly by small minerals (oil droplets larger than minerals) (Yu et al., 2019a;Qi et al., 2020), the closer the size of oil droplets is to that of minerals, the more difficult it is to form OMA. The decrease of zeta potential and the increase of hydrophobicity of minerals should have promoted the aggregation of oil and particle, but the opposite result was obtained. Therefore, these are not the dominant factors for the formation of OMA. CEC fluctuated with the increase of dispersant dosage, so it could not make a sustained reduction in the formation of OMA. Dispersant could increase oil viscosity, which would have inhibited the formation of OMA. However, due to the low temperature in this study, dispersant had little effect on oil viscosity, so it was not the main reason for the significant reduction of OMA formation. The decrease of IFT ow made the oil more easily dispersed into the water column and increased the chance of collision and aggregation between oil droplets and minerals, so the decrease of OMA formation was not caused by the reduction of IFT ow . The oilmineral particle size ratio tended to be the major influencing factor for the formation of OMA. In addition, with the increase of dispersant dosage, the dissolved oil content in the water column increased significantly. In contrast, the number of emulsified oil droplets decreased, and the adhesion between oil and minerals was inhibited, which significantly reduced the formation of OMA.

Conclusions
This study first investigated the effect of chemical dispersant on the surface properties of kaolin and Roncador crude oil. Then, its further influence on the oil dispersion and OMA formation was studied through a mesoscale experiment. With the increase of dispersant concentration, the zeta potential of kaolin decreased gradually, and the decay rate was from fast to slow. By comparison, cation exchange capacity increased first and then decreased, but the overall change was not significant; and the hydrophobicity increased continuously. The viscosity of Roncador crude oil increased as dispersant dosage increased, and the variation trend was more obvious at high temperature. The oil-seawater interfacial tension decreased significantly and remained stable as dispersant dosage increased. Chemical dispersant promoted the dispersion of oil spills in water column and reduced the oil droplets size. Meanwhile, it also inhibited the aggregation between oil droplets and kaolin as DOR increased, owing to the decrease of the number of emulsified oil droplets. Compared to other factors, the OMA formation is mainly affected by the size difference between oil droplets and minerals. The closer the two sizes are, the more difficult it is to form OMA.

Data Availability
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations
Ethics approval and consent to participate Not applicable.

Competing interests
The authors declare no competing interests.