Surface modification of stereolithography-based 3D printed structures utilizing ultrasonic-atomised sprays

Stereolithography (SLA)-based three-dimensional (3D) printing has become a popular tool for creating experimental models to study the two-phase flow behavior in complex flow structures. The main drawback while implementing such models is the wettability nature of the 3D printed surfaces. As non-geological materials are used while printing the porous designs, the flow mechanics do not follow similar patterns as in the reservoir. This work demonstrates the feasibility of using an SLA-based printing technique to replicate a porous structure. The porosity and pore size values of the 3D print are observed to be very close to that of the porous input image of the rock sample. A simple method to modify the surface characteristics of 3D printed surfaces using an ultrasonic-atomized fine spraying technique is developed. Here a thin layer of CaCO3 is deposited on the 3D printed surface by subjecting it to fine alternate sprays of calcium chloride and sodium carbonate. Thirty cycles of coating are observed to have altered the surface's wettability from neutral to oil-wet, resembling a carbonate reservoir. Ultrasonic assisted coating of 3D-printed surfaces.


Introduction
Understanding the complex flow mechanics inside the porous rock structures is paramount in improving oil recovery from a reservoir. For example, the reservoir rock's wettability affects the fluid's permeability. Also, porosity is another reservoir property that impacts fluid flow and distribution. As a direct pore-level observation of displacement mechanisms would improve our understanding of two-phase flow behavior, several physical flow models were developed in the past to experimentally examine the flow phenomenon inside porous structures [1]. The earlier experimental flow models did not mimic the reservoir morphology. They were mainly constructed by inserting a layer of glass or Lucite beads sandwiched between two glass plates to ensure a leakage-free system. Soda-lime glass beads were sometimes used as a simplified porous media to study the two-phase flow model [2]. To account for the physio-chemical characteristics of the reservoir, the glass beads were later replaced by rock sand grains [3]. With improvement in manufacturing capabilities, these simple physical flow models were replaced by laboratory-scale micro-models [4].
Micro-models are porous devices fabricated to mimic microstructures of reservoir rocks. Such devices offer possibilities for real-time visualizing the complex flow phenomenon at the pore scale. Most micro-models used in laboratory research are fabricated from glass or silicon wafers [5]. Laser and wet etching are generally used to carve the pore morphology onto the substrate, which is subsequently covered by a leak-proof attachment of glass [6]. The advent of three-dimensional (3D) printing technologies enabled a cost-effective mimicking of complex geometries of rock structures with an acceptable amount of accuracy [4,[7][8][9][10][11][12][13].
Stereolithography (SLA) is a 3D printing technique gaining more interest within the oil-and-gas community [20,21]. Stereolithography belongs to a family of additive manufacturing technologies that uses a photopolymerization technique. The print is achieved layer-by-layer in the build platform positioned upside down from the resin tank. As the SLA resin is subjected to a pre-determined ultraviolet light (UV, 405 nm), short molecular chains bond together, producing the highest resolution and accuracy, with the sharpest details and the smoothest surface 3D print. The integrated use of computer-aided design (CAD) with 3D printing technology permitted the production of highly complicated three-dimensional shapes that the other techniques mentioned earlier could not produce [22].
The biggest drawback in utilizing 3D printing for micro-models is the materials used. Petroleum reservoirs primarily contain minerals such as silica, calcite, and mica [23]. However, most of the experimental micro-models that were fabricated used nongeological materials such as glass [22], silicon [24], and polymers [25]. It is understood that the flow mechanisms within the rock grains heavily depend on wettability [26]. Hence there exist limitations in using the 3D printed model directly to interpret experimental observations where wettability effects are important. Glass and quartz micro-models could mimic sandstone rock formations but not the carbonate surfaces. As carbonate reservoirs account for more than 60% of global oil reserves, researchers have tried various ways to incorporate calcite surfaces in the micro-models. Porter et al. [27] fabricated models with fractures etched into thin sections of rock types using a femtosecond laser. Song et al. [28]created a micro-model from thin wafers of calcite crystal by laser etching patterns and immersion in hydrochloric acid. The channels were sealed on one side with a glass slide, creating a non-complete carbonate enclosure. Song and Kovscek [29] functionalized a 2D silicon micro-model with pore surfaces coated with kaolinite clay. Lee et al. [30] introduced a strategy to photo pattern calcium carbonate (CaCO 3 ) in situ by growing microstructures with Ca 2? and CO 2À 3 ions rich supersaturated solutions. This approach enabled the fabrication of micro-models having dynamically tunable geometries with submicrometer pore-length scales and controlled wettability. Wang et al. [5] improved the in situ CaCO 3 growth method by depositing nanolayers of CaCO 3 on glass walls of micro-model channels. They also verified the homogeneity of the CaCO 3 coating. Shaik et al. [31] presented a way to transform glass surfaces into calcite-coated surfaces using in situ layer-bylayer deposition procedures. They also reported the effects of process parameters, including pre-cleaning fluids and supersaturation on the coating density and the formation of CaCO 3 polymorphs. Alzahid et al. [26] proposed a process for functionalizing the pore space of Polydimethylsiloxane (PDMS) microfluidic chips with rock minerals and controlled wetting conditions. Quartz and calcite are used for the coating.
From the above research works, it is observed that not many studies have been performed to modify the surface characteristics of micro-models. Surface modifications are reported for glass, silicone, and PDMS substrates but not for a 3D-printed surface. As improvements in 3D printing technologies have enabled cost-effective replication of complex porous geometries with transparent materials, a step to modify its surface characteristics would be beneficial. Adjusting the surface with a thin coating layer allows surface wettability to be controlled. Growing uniform layers of CaCO 3 by chemical deposition is primarily complex within porous media due to the permeability limitation of liquids to reach the extremities in the geometry. However, if the chemical deposition can be performed using vapors or mists, the reachability of chemicals to less reachable porous corners would be easier. In other words, the coating process would be more efficient if the coating reagents could be replaced by a fine mist or spray instead of liquids. Hence in this work, we propose a surface modification strategy for 3D printed surfaces by alternately exposing them to calcium chloride (CaCl 2 ) and sodium carbonate (Na 2 CO 3 ) mists. The calcium carbonate (CaCO 3 ) particles formed by the reaction get attached to the surface, and the byproduct, sodium chloride (NaCl), is washed away. Fine mists or sprays are generated with the help of ultrasonic vibrations. Such a coating strategy is deemed superior to a dipcoating method as the latter may not exposure to all surfaces in a microporous structure. The present work provides a detailed description of the coating methodology and surface characterization using an ultrasonic-assisted spray. The effect on the wettability caused by the coated surface is reported. The first part of the paper explains the Stereolithographybased (SLA)-3D printing process of a porous structure derived from a model rock. In the second part, the strategy used to modify the 3D-printed surface by coating a thin layer of CaCO 3 is detailed. Owing to the optical transparency of the 3D-printed structures, flow visualization, together with the effects of surface interaction of various fluids on the CaCO 3 -coated surface, can be studied using the printed sample.

Stereolithography-based 3D printing
An SLA-based 3D printing is used in the present work due to its ability to produce highly accurate, isotropic, and impermeable structures with reasonably smooth surfaces. In addition, such 3D printers have low maintenance and cheaper resin costs.
Since one objective of this work is to investigate the capabilities of SLA-based 3D printing to replicate a porous structure accurately, a model rock sample is used as the parent structure. The work plan used to investigate the effectiveness of the printing technique is outlined in Fig. 1. A commercially procured ceramic disk (OFITE-#173-53) with a mean pore throat size of 50 lm, which is routinely used in mud filtrate characterization, is used as the rock model. A ceramic disk is preferred over an actual rock sample mainly due to its uniform porosity and permeability. Initially, the details of the porous structure of the ceramic disk are obtained using a micro-Computed Tomography (l-CT) scan. The X-ray l-CT from the Thermo-Fisher Heliscan lCT scanner enables volume images of medium to low-density materials to be acquired at resolutions down to * 800 nm. A computer-aided design model (CAD model) of the 3D porous structure is then reconstructed from the CT slices using Avizo software (Avizo 9.2, Thermo-Fisher). Here the 3D slices from the CT scans are bundled for volume rendering after selecting a particular region within the porous structure. Using interactive thresholding and an unconstrained smoothening value of two, 3D surfaces are extracted to form a.stl file. This file is modified using STL processing software (Ansys Discovery SpaceClaim 2019 R3) suited for 3D printing applications. The geometry is scaled up for printing and enclosing sides and inlet/exit ports in the modification process. The geometry is then printed using ANYCUBIC photon mono X 3D printer. Based on several preliminary trial prints, a magnification of 20x and an ultraviolet exposure time of two seconds was appropriate for our case. The actual 2.1 mm 9 2.1 mm 9 1 mm sub-volume extracted from the porous slab is scaled 20 folds to obtain 42 mm 9 42 mm 9 20 mm in the 3D print. The structure is enclosed with a 2 mm covering on all sides with four 20 mm diameter openings for the inlet/exit ports. Photographs of the 3D printed surface, the CAD model, and the ceramic disk are also shown in Fig. 1.
To evaluate the accuracy of the 3D printing, a l-CT scan of the 3D printed sample is taken and compared with that of the CAD model. Such a comparison would enable us to estimate the drawbacks of the selected printing technique. Ortho-slices of the CAD model and the 3D print taken at the exact locations are compared. Due to inaccuracies in placing the sample while taking a l-CT scan, getting the actual baseline image for comparisons with the CAD model is difficult. Several slices of the CAD model (separated by 1 lm) are generated and compared with a given ortho-slice from the 3D printed scan to overcome this issue. The ortho-slices are compared based on the cross-correlation value between the two. The best matching pair will have the maximum value of the cross-correlation coefficient. A maximum crosscorrelation value of 0.921 was obtained for a pair and was taken as the baseline for comparison. Forty-five such ortho-slice pairs formed from 3D Print and the CAD model are considered in the analysis. Figure 2 shows one such example of the image pair. These ortho-slices are obtained after converting the grayscale images into binary images by applying a global grey thresholding method [32]. Figure 2a shows the binary image of the CAD model, whereas Fig. 2b shows that of the actual 3D print. Black spaces represent pores, and white spaces represent the solid structure. It is observed that the 3D print is able to replicate most of the intricacies of the porous structure. Some details are missed by the 3D print and are highlighted in the blue dotted lines in Fig. 2b. Detailed comparisons of the 3D print and the CAD model on how well the porosity and pore size values matched are discussed in the results section.

CaCO 3 coating
After printing the 3D porous structure, the characteristic of the printed surface is modified to match the surface nature of the rock formation. For this, a fine coating of CaCO 3 is made using an ultrasonic-assisted spray coating procedure. A typical ultrasonic atomizer is used in the study, where an ultrasonic transducer is placed inside the bulk liquid to produce mist droplets [33]. Focused acoustic energy is transmitted to the bulk solution, where the free surface lever is kept a few centimeters above the ultrasonic transducer. A portion of the solution is transformed into mist or fine droplets, which are moved away from the bulk fluid using a compressed carrier gas. Figure 3 shows the schematic of the surface coating procedure.
A mist of CaCl 2 is generated by subjecting ultrasonic waves to an aqueous solution of 1 M CaCl 2 in a container (M represents the molar concentration in mol/liter). The generated mist is pushed out of the container with the help of compressed air flow. These fine mists or droplets are then allowed to fall freely on the surface of the 3D print. Using a similar approach, a mist of Na 2 CO 3 from an aqueous solution of 1 M concentration is generated and carried out above the surface using compressed air flow. As shown in Eq. (1), a chemical reaction occurs at the surface.
In a typical coating process, the spray of CaCl 2 is initially deposited, followed by the spray of Na 2 CO 3 . It is then dried using cold air to remove the excess water droplets from the surface. This is typically termed one coating cycle. After each coating cycle, the NaCl formed (water soluble) is gently washed using water, and the CaCO 3 formed, which is insoluble in water, gets adhered to the 3D print surface. Several such coating cycles are repeated to achieve a Identical experiments were repeated in the case of 3D-printed porous structures. The reachability of the fine mists in the porous matrix depends on the pore sizes and porosity. The usage of compressed air helps to distribute the chemicals to far-reaching corners in a manner better than a dip coating process. Since characterization studies of the coated surface are difficult in porous structures, a simple flat geometry, as shown in Fig. 3, is used for further studies.

Characterization
The surface characterization of the CaCO 3 coating is performed on the 3D-printed flat surface. The uniformity of the coating is initially visualized using an optical microscope, followed by a detailed observation with a Scanning Electron Microscope (SEM, Thermo Fisher Apero Model). Compositional analysis was investigated using elemental maps generated using Energy Dispersive Spectrometry (EDS, Thermo Fisher) equipped with the SEM. X-ray diffraction (XRD) measurement of the coated surface was performed using RIGAKU Ultima IV with Cu Ka radiation 40 kV between 0 and 80°angles. The change in wettability of the coated surface is estimated based on the contact angle of a droplet formed on the surface. The contact angle formed by a water droplet immersed in diesel placed on the coated surface is measured using a Ramé-hart model 500 goniometer.

Results and discussion
Comparison of the 3D print with the CAD model Before characterizing the CaCO 3 coating, the degree of similarities of the 3D print with that of the input CAD model is compared. This is to establish that SLA-based 3D printing technique is successful in replicating the inside porous structure. For this, the  2D slices from the CT scan of the 3D printed model are compared with that of the CAD model. Orthoslices are taken on the CAD model at the exact location as that for the 3D print model. Parameters such as porosity, average pore size, and pore throat radius are analyzed. Open-source and in-house MATLAB codes are used for this analysis.
Porosity represents the ratio of pore space available to the total volume of the rock sample. For the 2D image obtained from the CT scan, porosity is derived as the ratio of the area of pores to the total area of the CT slice. In Fig. 4a, the porosity values obtained for the 3D print and CAD model are plotted at different slice locations. The slice number on the x-axis depicts the slice location notation of the 45 ortho-slices used along the height of the CAD model and 3D print, each separated by a physical distance of * 444 lm. For the porosity analysis, the 2D image is first converted to a binary image [32]. Morphological transforms are further applied to the binary image to derive the majority features. Here the pixel values are set to one if five or more pixels in its 3-by-3 neighborhood are one; It is observed that the porosity values of the 3D print are consistently lower than that of the CAD model. The average porosity of the 3D print is at least 7.5% smaller than that obtained from the CAD model. The percentage deviation of the porosity values of the 3D print from the CAD model, calculated using Eq. (2), is shown in Fig. 4b.
The negative percentage deviation value of porosity indicates that the 3D print would have missed mimicking some of the fine pore features in the structure or would have formed thinner pores in some of the regions, as indicated in Fig. 2 Next, the average pore size or the pore radius is estimated using the method described by Rabbani et al. [34,35]. For this, the connected pore spaces are initially segmented using a watershed algorithm. The watershed image segmentation method can be compared to the physical flooding of water on the surface and finding holes, higher surfaces, and ridges. The watershed algorithm is applied to the gradient of the image obtained from the first partial derivation of an image and involves a measurement for the gray level changes. Further, a distance map is generated in the binary image to form connected topological surfaces. The contact watershed ridgeline formed by the interconnected pores is estimated. The average pore radius is calculated for each pore, and the throat size represents the size of the pore along the ridgeline between two interconnected pores. For each slice, pore sizes follow a distribution, and an average is obtained from this. Figure 5a represents the average pore sizes obtained for the 3D printed structure and the CAD model at different slice locations. It is to be noted that the sizes plotted in the figure are obtained after scaling it down 20 folds to match the case of that in the ceramic disk. The percentage deviation of the  pore size of the 3D print from the CAD model is given in Fig. 5b. It is observed that most of the percentage deviation values are trending toward the negative side (* -3.7%), indicating that the pore sizes of the 3D print are smaller than that of the CAD model.
Next, the average pore throat size, which represents the size of the ridges formed by pores, is compared. Figure 6a shows the average values of throat sizes (throat radius) obtained for different slices of the 3D print and the CAD model. It is seen that the throat sizes of the 3D print and the CD model are pretty much comparable, and the percentage deviation values are scattered in both positive and negative directions, as shown in Fig. 6b. The average value of the percentage deviation is * 1%, indicating that the throat sizes of the print are almost comparable to that of the CAD model. The experimental value of the average throat size of the ceramic disk, as provided by the manufacturer, is * 50 lm, which is comparable to our 3D print throat size (estimated after scaling down). The pore connectivity is also evaluated using the same algorithm. The pore coordination number, which indicates how much each pore is connected to the adjacent pores, is estimated. The average value of the pore coordination number from all the slices for the 3D print and the model is estimated to be 2.231 and 2.282, respectively. This shows that the pore connectivity in the 3D print is the same as that of the input CAD model.
From the above observations, it can be concluded that scaled-up 3D printing has faithfully mimicked the porous structure. The next section analyzes modifications caused to the 3D printed surface by ultrasonic-assisted coating. Figure 7a and b show the scanning electron microscope (SEM) images of the CaCO 3 deposits on the 3D printed surface. Ten cycles of deposition were performed for the surface shown in Fig. 7a, whereas 30 cycles were performed for the layer shown in Fig. 7b. Both images are provided on the same scale of magnification. In both cases, the deposits of the CaCO 3 are observed to be not homogeneous. In addition, the coverage is also not 100 percent. Analysis of the SEM images shows that the percentage coverage of deposits has increased from 10% to 28.5% as deposition cycles increased from 10 to 30 cycles. However, with the increase in the deposition cycles, the agglomeration of particles in the deposit is observed. Bigger chunks of CaCO 3 deposits are seen with a higher deposit cycle. The ImageJ toolbox is used to estimate the average particle sizes for distinguishable particles in the SEM image. It is observed that the average particle diameters for the 10-cycle case are in the range of 2.2 ± 0.8 lm, whereas for the 30-cycle case are in the range of 7.1 ± 0.9 lm. Hence, increasing the deposit cycles has also increased the sizes of the deposited particles. The above inferences are obtained after converting the SEM images into binary ones and performing image analysis.

Surface modifications
After obtaining the SEM images of the deposits, element analysis by using Energy Dispersive Spectrometry (EDS) is performed. Figure 8 shows the EDS spectrum and the concentration map distribution of major elements detected in the deposited region. The greyscale image of the selected region and the color concentration maps of the detected elements are shown in the inset. The EDS results indicate the presence of calcium, oxygen, and carbon as major elements, which suggests that the deposits may reasonably be of CaCO 3 . Carbon is fairly distributed throughout as the 3D-print (obtained from epoxy acrylate resin) also has carbon components. However, calcium and oxygen are distinctly visible mainly on the deposited region rather than the substrate. It is noteworthy that the platinum peaks seen in the EDS come from the platinum film deposition by sputtering during the sample preparation for the SEM characterization and should not be counted on the actual deposits.  To further confirm the chemical composition of the coated layer is of CaCO 3 , XRD analysis of the deposited region is performed. Figure 9 shows the XRD pattern of the deposits in the selected region. In the inset of Fig. 9, the XRD pattern of the 3D printed substrate, which is not coated with CaCO 3, is provided. The difference between the two plots is the clear appearance of the peak at 2h = 29.5°which corresponds to a reflection at the 104-lattice plane of the calcium carbonate polymorph calcite [36,37]. CaCO 3 exists in three polymorphs, calcite, aragonite, and vaterite, in the decreasing order of their thermodynamic stability. Under ambient conditions, calcite is the most thermodynamically stable phase, and the other two polymorphs transform into the calcite phase under kinetically favorable conditions [31,38]. Hence from the EDS analysis and XRD pattern, it can be concluded that the deposit on the 3D printed surface mainly contains CaCO 3 .
It is observed that 30 cycles of CaCO 3 coating created a visible layer. To confirm that this coating has affected the wettability nature of the 3D printed surface, contact angle measurements at the surface for an oil-water interface are performed. The oil considered is diesel. A droplet of * 2 lL of deionized water is metered out on top of the 3D-printed surface using a stainless-steel needle. The 3D printed surface is kept in a diesel bath through which the water droplet descent and falls on its surface. An image of the water droplet and the 3D-printed surface is recorded. Figure 10a and b show the photograph depicting the contact angle formed on the uncoated and CaCO3-coated surface, respectively. The contact angle values indicated in the figure are obtained after averaging three independent measurements. It is observed that an uncoated surface is almost neutrally wet [39], and the CaCO 3 coating makes the surface more oil-wet. Carbonate reservoirs are oil-wet [40,41], and the ultrasonic-assisted surface coating has helped make the neutrally wet 3D printed surface more oil-wet. From the contact angle measurements, it can be inferred that the wettability conditions in the CaCO 3 -coated 3D prints have turned closer to that of the carbonate reservoirs.

Conclusions
This study demonstrates the application of SLAbased 3D printing techniques for replicating complex three-dimensional geometries. The geometry's computer-aided design (CAD) is derived from a model rock and is used to create the 3D print morphology. The obtained 3D print had pore characteristics similar to the input CAD model. A simple ultrasonic-assisted coating method is applied on the 3D-printed surface. The wettability of the 3D printed surface is observed to have altered with the coating technique. Following are the specific observations made in the study.
(1) The porosity and average pore size values of the 3D print are observed to be slightly lower than that of the input CAD model. However, the average pore throat size and pore connectivity of the 3D print and the CAD model had a higher degree of agreement. (2) A thin layer of CaCO 3 successfully adheres to the 3D printed surface after subjecting to alternate mists of calcium chloride and sodium carbonate. Increasing the number of coating cycles increased the coverage and the size of CaCO 3 particles in the deposited region. (3) The presence of calcite polymorph of CaCO 3 is observed in the coating process based on XRD and EDS analysis. (4) The contact angle measurement showed that the wettability nature of the coated surface gas changed toward oil-wet, resembling a carbonate reservoir Figure 10 Contact angle measurement of -water in diesel droplet on an a uncoated 3D printed surface b 3D printed surface coated with CaCO 3 . A&M University Qatar) for his help with contact angle measurements. This paper was made possible by an NPRP award NPRP 11S-1126-170031 from the Qatar National Research Fund (a member of The Qatar Foundation). The contents herein are solely the responsibility of the authors. Open Access funding is provided by the Qatar National Library.

Funding
Open Access funding provided by the Qatar National Library.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licen ses/by/4.0/.