In Search of an Effective Workability Zone during the 3D Printing of Polymeric Periodic Open Cellular Structures Potentially Useful as Microreactors

Abstract

The question of how easy the transition is between design and manufacturing by the 3D printing of periodic open cellular structures occurs from the analysis of cases in which additive manufacturing and heterogeneous catalysis merge. The synergy between these two fields suggests that one of the great advantages that the catalysis of this manufacturing methodology can take advantage of is the obtaining of advanced designs that would allow improving the processes from the geometry of the reactors. However, not all 3D-printing techniques offer the same degree of resolution, and this uncertainty grows when using more complex materials to work with, such as ceramics or metals. Therefore, the present work seeks to answer this question by finding experimentation strategies, starting with a simple case study inspired by the additive manufacturing–catalysis combination, in which a ceramic polymer resin of high thermal resistance is used to obtain POCSs that are potentially useful in thermochemical or adsorption processes. This exploration concludes on the need to define limits for what we have called an “effective work zone” that combines both design criteria and the real possibility of printing and manipulating the pieces, making sweeps in structural parameters such as cell size and the diameter of struts in the POCS. Similarly, the possibility of coating these systems with inorganic oxides is explored, using a generic oxide (Al2O3) to analyse this scenario. Finally, a cartridge-type assembly of these systems is proposed so that they can be explored in future processes by other researchers.

1. Introduction

Since its inception in the 1980s, additive manufacturing has been consolidated as one of the pillars of the fourth industrial revolution. The concept of additive manufacturing brings together a series of methodologies that allow the obtaining of 3D objects by adding layers of a certain printing material that is dosed in a controlled manner. The main advances in this technology are based on the use of polymers as a printing material. However, new strategies have been created that allow printing using materials such as coals, ceramics or even metals. Today, in the additive manufacturing market, the great leaders in the production of equipment for 3D printing are in open competition to lead the sector with technologies that allow more precisely controlling the printing of metal parts.

One of the great advantages of additive manufacturing compared to other traditional manufacturing methods is the ability to generate geometries with a degree of complexity and detail unattainable so far. For this reason, more and more sectors are incorporating this manufacturing methodology into their production processes. This is the case of a field with enormous potential to take advantage of additive manufacturing, which is the chemical industry and, specifically, heterogeneous catalysis. The latter is one of the disciplines of chemistry that faces the challenge of improving the efficiency of the processes dealing with, among others, issues related to design engineering and the manufacture of reactors. In fact, several authors suggest that through an advanced and improved design of the reactor in which some processes occur, the levels of performance and energy use can be optimized.

Recently, we presented an in-depth review on the synergy between heterogeneous catalysis and additive manufacturing [1] and, like other reviews on this subject [2,3,4], it can be inferred that one of the great advantages that promises to be one of the most decisive is the possibility of adapting greatly advanced geometries to enhance fluid dynamics in catalysed reactions [1,2,3]. However, it is also evident that, although there is already evidence of improvements in catalytic processes thanks to structured catalysts generated with the help of 3D printing [5,6,7,8], the vast majority of printed devices used in catalysis have rudimentary shapes (such as the woodpile configuration [9,10] especially observed for ceramics and metals [11,12]) that do not exploit this potential to the fullest [1].

Among the designs of advanced geometries generated by 3D printing [13,14], one stand-out is the periodic open cellular structures (POCSs) that emulate the structure of a foam but with a regular distribution of the cavities [8,15,16,17]. This regularity is achieved by taking a unit cell that is projected in the three dimensions to generate superior shapes that can have specific shapes, such as cubes, cylinders, spheres or those that allow the best adaptation of the POCSs to the process. Among the most-used unit cells to generate POCSs by 3D printing are the cubic, diamond, gyroid and Kelvin structures [16,18], but there is a wide range of structures that are also being explored, and both polymeric materials and ceramics and metals have been tested.

In most cases in which POCSs are studied and compared, improvements in process performance are proposed thanks to the advantages offered by this type of geometries. However, it can be said that, in most cases, a first exploration is being made and there is no experimental design systematic enough to draw decisive conclusions about the real contribution of geometry. This represents an important dilemma in the growth of the synergy between catalysis and additive manufacturing because the possibility of generating advanced designs with great versatility and shapes is accepted as one of the main advantages of this manufacturing method, but there may be indications that such versatility may not be so wide.

One of the research teams that today presents one of the most complete and comprehensive strategies to explore the use of POCSs as support for catalysts is the one led by Prof. E. Tronconi in the Politecnico di Milano in Italy. In this research group, they have managed to demonstrate the advantages of structuring printed catalysts based on advanced geometric parameters that support the design of several structures [18,19,20,21,22,23,24]. This allows them not only to design and test geometries randomly, but also to draw important conclusions about design parameters that can affect the mass [22] and heat [18,24] transport phenomena in certain catalytic processes. In this way, they have managed to propose real advanced design strategies that in turn have allowed them to formulate protocols to design and predict, for example, the existence of limitations in the processes of material transport in this type of device [22].

In a similar way, the research led by Prof. H. Freund at the University of Erlangen Nuremberg in Germany presents an in-depth analysis of the correlation between geometric parameters of POCSs and their heat and mass transfer properties or pressure drop when fluid travels through these structures [25,26,27,28,29,30,31,32,33,34]. Moreover, another investigator leading a more direct exploration of the use of POCS in catalysis is C.A. Grande. This has allowed him and his team to establish improvements in the catalytic properties of some systems through the creation and control of the shape of POCS-type reactors [35,36,37].

These three examples represent the right way to focus research on structured systems obtained by 3D printing, not only because they contribute to experimental tests of the advantages of their application, but also because they allow generating prediction models thanks to the probability of iteration achieved with these devices. However, despite the existence of approaches as successful as that of Prof. Tronconi’s team, there are still variables whose effect has not yet been explored. For example, it is not clear how direct the transition between design and manufacturing is, because POCS studies usually show a geometry that works, or that is easy to print, but does not analyse sweeps of properties of the structures to find how many geometries are actually printable. This uncertainty is exacerbated if possible modifications are taken into account by the printing methodology used or by the manufacturing material.

Obviously, the number of variables involved in the design of POCS structures, such as the size of the unit cell, the diameter of the struts, the thickness of the planes, the void fraction or the exposed area, would generate a very high number of combinations and it would be very complex to predict the limits for all structures. However, when comparing structures with each other, certain values of these parameters can be restricted, which would allow us to analyse whether all geometries can really be printed or not. This is especially useful for defining not only whether the designed geometries meet the geometric requirements for the desired application, or whether or not they can be printed with the selected technique and material, but also how comparable the structures can become to each other.

Similarly, if the selection of the manufacturing limits of POCSs is already complex, the evaluation of their use as systems to structure substances such as adsorbents or catalysts opens another very broad scenario of combinations of variables that must also be addressed systematically. In this sense, it is crucial to apply strategies to study the coating of POCSs, such as the one recently presented by Balzarotti et al. [38] in their study on the influence of the operational parameters of the coating of metallic monoliths type foam. Although in this case, the authors did not study printed structures, they highlighted the need to evaluate multiple factors involved in the coating processes that can also mark the applicability of these structures.

Therefore, taking into account the scenario described above, the present work proposes exploring, in a more detailed way, the design and manufacture of a group of POCSs starting from cubic, diamond and Kelvin unit cells that have been well-explored in the literature. However, in this case, additional information will be provided that will allow establishing the real limits of work and will also show what criteria can define whether the POCSs can be compared with each other or not. For this, the manufacturing of POCSs is explored using the stereolithography technique (SLA), using a polymer resin with thermal resistance guaranteed by the manufacturer below 238 °C that could be used in structured systems such as those proposed by Chaparro-Garginca et al. [39]. In a complementary way, the coating of these structures with a generic inorganic oxide (Al₂O₃) that simulates the interaction of a catalyst or an adsorbent that adheres to the surface of the POCS is explored. Finally, a cartridge-type assembly mechanism of these POCSs is tested that can be analysed well in future research in which these systems can be applied as substrates to structure catalysts or adsorbents.

2. Results and Discussion

2.1. Analysis of the CADs (In Search of Limits of Design)

The design of the POCSs modifying the strut diameter generated a series of 3D structures whose STL files are included in the Supplementary Information (Table S1). Figure 1 presents different views of the extreme cases of strut diameter for the three unit cells used.

In the first place, it is important to define that plugging is assumed to be a situation in which some cavities of the structures present a total blockage due to overlapping of the struts. The POCSs made by cubic and diamond unit cells do not present plugging even in the case of the largest diameter used (1.0 mm). However, POCS with Kelvin unit cell exhibit plugging in the holes of the structure with smaller details. To better visualize this fact, Figure 1C shows all the structures designed with the Kelvin unit cell and it can be established that such plugging begins to be observable from strut diameters of 0.6 mm.

The plugging of some designs with Kelvin structures points out the fact that there are geometry restrictions that have to be identified for every structure depending on the selected cell size and strut diameters. These restrictions imply that there is a limited range of struts that allow generating POCSs with open cavities for every cell size. In the case of the cell sizes and strut diameters selected for the present work, plugging was observed only for the Kelvin geometry, but this does not mean that cubic or diamond geometries cannot be plugged. The behaviour observed in Figure 1 only suggests that the plugging is more easily achieved in the case of the Kelvin because of its complexity in some zones with small details compared to those of the other structures.

Among the variables that control the properties of POCS and determine their applicability, for instance as catalytic supports, are the occupied solid volume (O_SV) and the exposed area (S). In POCS structures, the O_SV represents the difference between the total volume of the structure, which in this case is the projected squircle (1405 mm³—see Section 3.1, Figure 13) and the empty volume of the cavities (V_c) that allows calculating the void fraction (ε = V_c/1405 mm³) [20]. The higher the S_OV values, the denser the POCS, and recent studies of this type of structure have shown a close relationship between its density and its capacity to transfer heat [18,20]. Regarding the S value, this represents the available surface that is in contact with any fluid that passes through the cavities of the POCS. When POCS are used for the structuring of a powder catalyst using a coating with such a catalyst on the surface of the POCS, the S value determines the thickness of the coated catalytic layer. According to this, both O_SV and S are important parameters that must be controlled during POCS fabrication. Figure 2 shows how the O_SV and the S vary with the diameter of the struts for each of the selected geometries.

For all geometries, the O_SV values depicted against the strut diameters (Figure 2A) generate trends that fit with third-degree polynomial functions, even in the case of the structures with Kelvin geometry that includes the plugged ones (depicted as empty hexagons in Figure 2). Despite this, a slight decrease in the fitting is observed in the r² value for the Kelvin structure which may be due to the superior complexity of this geometry compared to that of cubic and diamond.

The ε values also fit third-degree polynomials in all cases with a trend opposite to that of the O_sv as expected (Figure 2C), confirming that the void fraction decreases with the increment of the strut diameter. The fit of the void fraction values to a third-degree polynomial is in agreement with the analytical relationship between cell size, strut diameter and void fraction presented by Bracconi et al. [15,17,18,20] in a study carried out with this type of structure. Similarly, the complexity of the equation that describes the calculation of the void fraction for these structures grows in the following way: cubic < diamond < Kelvin. This is surely related to the slight deviation observed in the linear fit of the void fraction values obtained for the Kelvin structure.

Another important parameter in the manufacture of monolithic systems when these are used for the structuring of catalysts is the exposed area (S). This parameter strongly influences the thickness of the catalytic layer that is deposited on the walls of the monolith. Therefore, it may be directly influencing the tailoring of the mass and heat transport processes along the catalytic layer. The effect of the exposed area (S) becomes more important when comparing the thickness of a layer deposited in structured systems with different geometries, such as those analysed in this work. In that case, it would be necessary to select design conditions in which the same values of the exposed area are achieved so that the same load of deposited catalyst generates the same thickness in all cases. According to Figure 2B, the design parameters selected for this study do not allow exposed area values for the cubic structures comparable to those of the diamond and Kelvin structures. Therefore, it is important to note that this type of experimental limitation can only be visualized with representations such as those presented in Figure 2. Surprisingly, despite the relevance of this subject, when POCSs of different geometries for structuring catalysts are compared in the literature, the thickness of the catalytic layer is not usually normalised based on the exposed area in most of the published studies. Consequently, the need for a more careful experimental design in which more control parameters are included when working with this type of device for catalytic applications is evident.

In summary, what the analyses of the geometries reveal, just in the virtual stage prior to the manufacturing, is that even the digital 3D models may present zones limited by geometric parameters that generate blockages in the structure. These complex zones must be identified in order to perform a suitable selection of the control parameters and the design limits aiming to be clear about the printability of the pieces from the point of view of design. In addition, it has to be remarked that with the inclusion of more control parameters, the suitability of the pieces may be considerably reduced.

2.2. Printing and Post-Treatment Strategies

Once the design limits have been defined, it is necessary to establish the real printability limits that, in theory, should coincide with the resolution values achievable (~0.2 mm) with the high-temperature resin and the printer used in this study, according to data from the company Form Labs.

A first approach of printing the designed structures (discarding those outside the design limits—see Figure 1) was achieved following a generic printing setup aiming to observe possible issues during the generating of the pieces. This approach implied the loading of the STL file by the slicer program and the generation of a green piece that was subsequently thoroughly washed with isopropyl alcohol and cured for 1 h at 80 °C in the Form Cure unit. The results of the observed issues are summarized in

Table 1. Summary of problems observed before and after the printing of the parts.

Struts Diameter (mm)	Cubic	Diamond	Kelvin
0.1	⮹	⮹	⮹
0.2	⮹	⮹	⮹
0.3	✓	✓	✓
0.4	✓	✓	✓
0.5	✓	✓	✓
0.6	✓	✓	ø
0.7	✓	✓	ø
0.8	✓	✓	ø
0.9	✓	✓	ø
1.0	⌧	⌧	ø

⮹ Breakage of the structure during the removal of the part from the printing area; ⌧ Blockage of some cavities observed after the printing of the piece; Ø Blockage of some cavities observed in the CADs (Design Limits); ✓ Printable structures without any problem.

, and different regions may be identified.

Firstly, for the lowest strut diameters, it was observed that although most of the pieces can be printed, these cannot be subsequently properly detached from the print area. This is a mechanical process in which a tool is used to separate the printed parts, and with a small strut diameter, these do not have enough consistency to resist the mechanical stress of the tool. Therefore, it is impossible to remove an entire part without the promotion of cracking. The labelling of the pieces that present this problem with the symbol ⮹ generates an area highlighted in yellow in Table 1 that is called the “fragile pieces” area and is not suitable for obtaining any of the geometries of the POCS studied.

Another area in which problems were observed corresponds to systems with a 1.0 mm strut diameter. In this case, both the cubic structure and the diamond structure can be easily printed and there is no problem in separating them from the printing area. However, the width of the struts generates plugging in some cavities. This behaviour could not be avoided, even by modifying print parameters to maximize resolution. Therefore, these cases were marked with the symbol ⌧ and generated an area called the “plugging area” highlighted in Table 1 in orange. This “plugging area” must be differentiated from the region where plugging was detected before manufacturing in the CAD file itself, which was marked with the Ø symbol and is highlighted in red.

Combining the areas ruled out by the design criteria and those that show manufacturing problems due to fragility or clogging of the parts, an area is seen in which there are no problems from the point of view of design or manufacturing for the systems proposed in this studio. These cases, which were marked with the ✓ symbol, are highlighted in green and are located in what we have called the “effective work zone”. This effective work zone is obtained not only by evaluating the design but also the possible errors in the printing; therefore, it is much more precise and guarantees success in obtaining parts with control parameters that are within it.

According to Table 1, the Kelvin structure presents the most restricted effective work area. Therefore, its analysis can be deepened. In this regard, Ferroni et al. [40] proposed the analysis of what they call open windows that allow communication between the different unit cells. In the case of the Kelvin structure, there is a square window corresponding to what we have called the “small detail zone”. As the size of the struts increases, this square window tends to assume a circular shape due to the overlapping of the struts and then becomes blocked. Therefore, it is suggested that as the complexity of the structures increases, the open windows that can generate blockages can more quickly be detected and the dimensions of the pieces in that area can be analysed. In our particular case, the printed parts with a strut diameter of 0.3 to 0.5 mm did not generate any type of clogging.

It is important to note that the limits of the effective work zone in Table 1 cannot be extrapolated to other cell sizes that have not been included in our designs. Surely, each combination of variables will generate different effective work zones. However, what is clear is that the combination of design and manufacturing criteria, in addition to control parameters, can considerably reduce the geometries that can be used. This is an aspect very little discussed in the literature that goes against the great versatility in terms of printable geometries that are attributed to additive manufacturing. Actually, the designs can become significantly restricted, and it is necessary to carry out explorations such as the ones proposed in this study to see if certain types of geometries can actually be printed or not.

Moreover, the printing conditions generated by the characteristics of the printer can play a determining role in defining an effective work zone. In particular, the characteristics of the laser can significantly influence the thickness of the layers that are consolidated during the printing process. Although different types of printers were not compared in this study, it is very likely that each type of printer can generate particular alterations in the limits of printability, which reaffirms the importance of carrying out studies such as the one proposed in this document.

Finally, as a complementary analysis, different regions of SEM micrographs (not presented) of the printed pieces of the three types of structures within the effective work zone were analysed to establish deviations in the theoretical diameter of the struts. Averaging the values obtained, it can be said that in all cases there is a deviation in the printed parts that does not exceed 0.6% of the theoretical value of the diameter of the struts according to the CAD. This deviation value is comparable to that obtained by Ferroni et al. [40] in their study of POCSs printed with cubic, diamond and Kelvin structures.

Once the effective work zone was defined, the post-treatment stage was analysed. For this, different pieces were printed and subjected to the post-treatment strategies described in Section 3.2 of the Materials and Methods.

Taking into account that a high-thermal-resistance polymeric resin was used as the raw material for obtaining the POCSs of the present study, and considering that one of the potential applications of these structures is its use as catalyst supports in thermocatalytic processes, a thermal analysis of fragments of pieces subjected to different post-treatments was carried out and the results are presented in Figure 3.

Regarding the behaviour of weight (Figure 3A), it is practically identical in all cases. This is stable until 250 °C, and above this temperature, a progressive weight loss is observed that becomes much more pronounced from 450 °C. Therefore, from the point of view of the weight, no differences between the post-treatment strategies are observable. However, the DSC (Figure 3B) shows a considerable peak (~212 °C) only for the piece submitted to the PT1. This signal is probably due to the transformation of the polymeric resin during the thermal treatment of the thermal analysis since this post-treatment strategy did not include curing or any thermal treatment. This may be due to the incomplete consolidation of the polymer, which also generates a series of broad and intense bands starting at 350 °C.

For the PT2 and PT3, the DSC signal near ~212 °C does not appear and for the PT3, only a narrower endothermic signal than for PT2 is observed, near 457 °C. Therefore, the PT3 strategy is selected as the most appropriate because it generates a more stable thermal behaviour in the printed pieces and guarantees adequate workability with this type of material at temperatures below 238 °C, which is what is recommended by the maker.

Complementing the study of thermal effects on the printed pieces, cylinders whose dimensions appear in the experimental section were printed and treated with PT3. These cylinders were subjected to heating in a thermal microscope that allowed any change in the shape of the part to be monitored. The results of tracking the 2D image of the piece as a function of temperature are shown in Figure 4, and it is observed that changes in the angles of the piece’s silhouette begin to be observed only up to 325 °C. These results reinforce the data of the TGA/DSC analysis because the PT3 not only generates parts that do not change in weight or composition in a temperature range below 250 °C, but it also guarantees that there are no changes in the shape of printed parts by folds or curvatures generated by heating.

2.3. Coating of Printed Pieces with a Generic Inorganic Oxide

Again, considering the possible application of POCSs as substrates in which different substances can be structured, such as catalysts or mere adsorbents, it is important to analyse their ability to be coated. This ability depends on the physicochemical properties of the exposed surface of the printed pieces, which in turn can be altered differently in each of the pre-treatment strategies proposed in Section 3.2 of the Materials and Methods. For this, Al₂O₃ has been selected as a generic inorganic compound that can represent the main component, for example, of a heterogeneous catalyst well. Therefore, different concentrations of a commercial slurry of Al₂O₃ (Nyacol^® Al20) were used for the washcoating of plates (described in the Section 3.1 of Materials and Methods—see Figure 14). The monitoring of the weight gain after each coating is presented in Figure 5, and it should be noted that the coating experiments were carried out in triplicate with each of the slurry concentrations; thus, the trends depicted in Figure 5 are the averages of the triplicates. In addition, an approximate calculation of the thickness of the coated layer after the tenth coating is presented. For this, the density of Al₂O₃ (3.96 g/cm³) was used and 960 mm² was assumed as the exposed area of the plates.

The Al₂O₃ load increases as the coatings are made in the three cases. In addition, as expected, the higher the concentration of the alumina suspension, the greater the increase in the mass of loaded Al₂O₃ that is observed. Despite this, the variation in weight gain is very significant in the case of the 100% Nyacol suspension. In addition, the more concentrated suspension generates preferential accumulations towards the edges of the plates.

Different SEM micrographs of the plates were obtained after 10 coatings with each of the suspensions used, and the most representative ones are shown in Figure 6.

The micrographs confirm that the more concentrated suspension (Nyacol 100%—Figure 6c,f,i) generates not only more heterogeneous coatings, but also less-stable ones. In fact, detachment and cracking of the Al₂O₃ layer can be seen in different areas of the plate. This behaviour is repeated regardless of the type of pre-treatment.

Regarding the less-concentrated suspensions (Nyacol 33 and 66%), both seem to generate more homogeneous coatings, probably because they have a lower solid load. However, the higher magnification micrographs show that Nyacol 66% produces some cracks in the plates pre-treated with the PT1 and PT2 strategies.

Only the coatings made with the Nyacol 33% suspension produce a homogeneous coating regardless of the type of pre-treatment. Therefore, this concentration is selected as the most appropriate to carry out subsequent studies of part coatings.

Finally, regarding the pre-treatment strategies, PT3 is definitely selected, not only because it improves the thermal stability of the printed structure, but also because it seems to generate more homogeneous coatings, at least in the cases of more concentrated suspensions.

2.4. Thermal Stability of the Coatings Below 200 °C

After defining the PT3 post-treatment strategy and confirming that the suspension of less-concentrated Al₂O₃ (Nyacol^® 33%) produces the most homogeneous coatings, new plates are generated that are subjected to these experimental conditions. Subsequently, these plates are subjected to different heat treatments. In the first treatment, the plates are heated at 50 °C for 5 h. In the second treatment, they were heater for 5 h at 100 °C, and finally another 5 h at 200 °C. Plates were removed after each 5 h period, analysed by scanning electron microscopy, and the most representative micrographs are shown in Figure 7.

Although the experiment was prolonged, not only the thermal stability of the printed substrate is confirmed, but also that of the adhered alumina layer. Therefore, the interaction between the Al₂O₃ layer and the polymeric resin is not altered at temperatures below 200 °C when it is submitted to an oxidizing atmosphere (air), since no cracking or detachment is observed. This shows that in the eventual presence of oxidizing reagents from a reaction that occurs at working temperatures (<200 °C), the POCS printed with the high-temperature resin could be used as support for structuring, or as adsorbents or catalysts.

2.5. Washcoating of POCS Structures

Taking into account the argument that we have previously discussed in which we emphasize that different geometries are not usually compared with a normalized exposed area, we are aware that in the case of the designs proposed for the cubic structure, it would not be possible to generate any that would reach the area of the diamond or Kelvin systems in the effective work zone (See Figure 2B). Therefore, some design parameters were modified in order to achieve three POCSs with similar exposed areas. In the case of the diamond and Kelvin structures, the unit cell volume was kept at 8 mm³, and 0.45 and 0.35 mm struts were used, respectively, whereas for the Cubic structure, a unit cell volume of 1 mm³ was set and 0.75 mm struts were used. These new design conditions allowed the generation of POCSs with a very similar exposed area of around 2400 mm². Thus, having POCSs with comparably exposed area values, these were subjected to the PT3 post-treatment strategy and subsequently coated with the Nyacol 33% diluted suspension. The results of the washcoating process are presented in Figure 8, including the estimation of the thickness of the oxide layer generated with the eighth coating in each case. For the stated estimation, a density of 3.96 g/cm³ was assumed for the alumina, and an area of 2400 mm² in which the alumina was spread.

Firstly, it is observed that in all cases, the load of Al₂O₃ increases with the number of coatings. Despite this, the trend for the cubic structure seems to be more regular in terms of the amount of Al₂O₃ reached per coating, which looks constant, while for the diamond and Kelvin structures, the first coating generates a much higher load than that achieved with successive coatings. In both cases, after the second coating, the alumina weight gain seems to stabilize.

Taking into account these changes in trends and the fact that the final load of alumina was not the same for the three systems, despite having a comparable exposed area, it becomes clear that the geometry can influence the amount of oxide that adheres to the system. This is why the thickness of the coated layer is different in all cases. This behaviour is in agreement with what was observed in previous works on monolith coating by washcoating, in which it was observed that the angle zones promote selective accumulations of the agent with which it is being coated [41]. In our case, as the complexity of the structures increases (cubic < diamond < Kelvin), more angles are promoted in the structure, and this would justify the results between the diamond and Kelvin POCSs. However, what about the cubic structure? In this case, changing the cell volume to a smaller one probably creates a lower cell density. This means that the block has more angles, and therefore its ability to capture Al₂O₃ can be increased. In this sense, as obvious as it may seem, the fact of having normalized the exposed area allows us to conclude much more consistently about the effect of geometry on the ability to gain weight in a coating process by POCSs.

Another effect that the preferential accumulation of material may have is the increase in uncertainty in the estimation of the thickness of the oxide layer (whose values are presented for each case in Figure 8). Therefore, how homogeneous the coatings are must be evaluated.

In order to analyse the homogeneity of the coatings, some of the coated blocks were fragmented with a cutting tool in order to assess the quality of the coating, not only in the external areas but also in the interior cavities. The fragments were analysed by scanning electron microscopy, and mapping was also performed to see the distribution of the main constituent elements of the micrograph by EDX. The most representative micrographs for the three structures are presented in Figure 9.

In all cases, the micrographs (even those not shown) exhibit a homogeneous distribution of Al and O atoms (due to the layer of Al₂O₃) just in the zones where it is expected, which is in the surface of the struts. This is confirmed not only for struts in the outer zones but also those in the inner zones. Therefore, the success of the washcoating process is observable in all cases. Moreover, it has to be commented that the regions where high densities of C are detected, rather than Al or C, correspond to the cut edges generated by the cutting tool.

Despite the homogeneous appearance that is generally observed in the coatings, the observation with higher magnification confirms the fact that towards the angles of the structures there tends to accumulate a little more Al₂O₃, as in the example of some angles in the diamond structure shown in Figure 10.

What is observed in Figure 10 confirms what was previously discussed in the monitoring of the weight gain per coating in relation to the shape of the POCS and the content of angled zones. In addition, only where there is more accumulation of material, slight cracks or even the beginnings of the detachment of the alumina layer are observed.

These results are comparable with those of other types of substrates that are subjected to washcoating [41,42,43]. Therefore, it becomes clearer that in order to improve homogeneity and minimize the accumulation of material, it is possible to work on optimizing the rheological properties of the slurry of the material to be used as coating.

Finally, to evaluate the adherence of the alumina layer to the different POCSs, two procedures were applied. On the one hand, a procedure reported in the literature [44] was adopted, in which monoliths covered with a layer of a metal oxide are introduced into petroleum ether or some aliphatic organic solvent and then subjected to ultrasound in a bath for 15 min. Subsequently, the blocks are removed, dried and weighed to compare how much mass has been lost during the process. In our case, the organic solvent was replaced by distilled water to avoid affecting the polymeric resin. We have labelled the test as the “wet adhesion test”, and the results are shown in Figure 11A. However, it should be noted that this turns out to be a very aggressive adherence test. In fact, if the conditions to which this type of device is usually subjected are analysed, one thinks of subjecting them to a stream of gases. Therefore, an alternative adhesion test is proposed in which the blocks are subjected to a stream of compressed air of 1 L/min for 12 h at a temperature of 200 °C. The results of this test, which we have called the dry adhesion test, are shown in Figure 11B.

The results of the wet adhesion test are much worse than those of the dry adhesion test. In the first case, all the POCSs lose almost half of the catalytic layer load, so it can be said that from this perspective the adherence is very low. This means that in processes in which the POCSs are put in contact with a solvent, a great adherence of the species to the surface is not expected.

However, in the dry adhesion test, in which the POCSs are subjected not only to considerable airflow, but also to a temperature of 200 °C, it can be said that hardly any loss of material is observed. Therefore, under these operating conditions, one can speak of excellent adherence, which is closer to the conditions of a heterogeneous reaction or process in which the reactants or adsorbates are in the gas phase. Therefore, this type of POCS will preferably be recommended for heterogeneous processes with gaseous substances.

2.6. Proposal for Detachable Cartridge-Type Assembly of the POCS

Apart from studying how easy the transition from the design to the manufacture of POCS is, analysing the possible application of this type of structure made with a polymeric resin in adsorption processes or even catalytic processes is also sought after. Therefore, taking advantage of the versatility offered by 3D printing, it has been proposed to carry out a design in which several POCSs that have been generated in this study can be inserted into a casing. The characteristics of this design were presented in Section 2.1 of the Materials and Methods, and some examples of printed parts are shown in Figure 12.

Firstly, it is important to note that once the best post-treatment strategy has been selected, PT3 has been applied in all cases, both for the POCS and for the carcasses. Second, it should be noted that when parts are to be integrated into a single assembly, there is a certain degree of deviation in dimensions that must be optimized to ensure that the assembly mechanism can be used repeatedly, not just for the same set of POCS and case, but also for different sets printed at different times.

Therefore, a trial-and-error methodology was applied to find the best fit of the POCS inside the shell, guaranteeing its fixed permanence, but removable with a minimum application of pressure. On the other hand, it is desirable that the diameter of the assembled cylindrical cartridge is always fixed. To achieve this, a tolerance margin is given to the CAD of the carcass in the dimensions of the internal cavity, which in this case consists of reducing the ideal thickness of the carcass in the internal zone. Figure 12B shows how in the CAD, the thickness of the shell in the internal zone was reduced by 0.15 with respect to the geometrically perfect design, while the external diameter remained constant.

This mounting system could be applied in any process where POCSs are used as a system to structure an adsorbent or a catalyst. In fact, the adjustment of several complete sets was tested in a commercial reactor described in the literature in previous studies in which metallic cylindrical monoliths [45] are studied, and the coupling was perfect in all cases.

3. Materials and Methods

3.1. Design and 3D Printing

The CAD designs were performed using SolidWorks software and exported as STL files that were subsequently prepared for the printing process in the slicer software PreForm (Formlabs^®, Somerville, MA, USA). Regarding the structures, three unit cells (cubic, diamond and Kelvin) were selected for generating structures with the same shape through the regular repetition of every single unit cell in the three-dimensional space. The final shape of all the structures is a projected (10 mm) squircle with high corner smoothing. The main features and size of the unit cells or the final shape of the POCS are presented in Figure 13.

As design variables, the unit cell volume was set at 8 mm³ in all cases, and a 0.1 mm sweep was made on the diameter of the struts between 0.1 and 1.0 mm. Therefore, for each unit cell, 10 structures were generated by modifying the diameter of the struts.

Moreover, it has to be pointed out that the purpose of using a projected squircle shape instead of a cylinder is for creating a system with a better fit of three structures or units into a case. Although the casing could be included in the design itself, this is not only to explore the manufacture of individual structures, but also as a versatile system that allows different interchangeable units to be combined in a removable cartridge-type application, which is one of the advantages that microreactors can offer. This is a design strategy partially inspired by the “packed-POCS with skin” concept recently proposed by Fratalocchi et al. [24] aiming to intensify non-adiabatic catalytic processes.

For the first analysis of the post-treatment and coating strategies (whose details will be described below in Section 2.2), square plates (20 mm × 20 mm × 2 mm) such as that presented in Figure 14 were designed and printed too.

The 3D printing of all pieces was carried out via spatially resolved layer-by-layer (layer thickness 100 μm) photoinduced polymerization and cross-linking of a polymeric resin, which is the printing strategy labelled as stereolithography (SLA). A Form 2 printer (Formlabs^®) was used, equipped with a 402 nm violet laser (spot size or diameter of the focus: 140 μm), an automatized resin fill system and a wiper for the sliding peel process. The printing parameters as well as the settings of the printer were controlled by means of the PreForm software. Regarding the printing material, the high-temperature resin for heat resistance (FLHTAM02) Formlabs^® offers a heat deflection temperature of 238 °C.

3.2. Post-Treatment and Coating of the Printed Objects

Complementary units for the post-treatment of the printed pieces were used: firstly, the Form Wash unit that allows the removal of excess material thanks to an exhaustive wash with isopropyl alcohol. This procedure was applied to all the printed pieces.

Secondly, the Form Cure unit is a controlled temperature (up to 80 °C) chamber that allows for homogeneous irradiation on the printed pieces to give a better finish with light curing using LED lamps with a specific wavelength (405 nm). As a complement to the washing and curing steps, Formlabs recommends a subsequent thermal treatment of the printed pieces when using their high-temperature resin (FLHTAM02), especially to reinforce the mechanical properties. Therefore, considering that in the present study the coating of 3D-printed pieces made of this resin will be evaluated for the first time, the effect of three different post-treatment strategies were evaluated, aiming to establish how they can affect the coating. The post-treatment strategies were studied over the square plates, and the most suitable was selected for the post-treatment of the POCS.

Post-treatment strategy 1 (PT1): In this case, just the washing of the printed pieces with isopropyl alcohol was carried out in the Form Wash unit. Afterwards, the pieces were dried with pressurized air at room temperature.

Post-treatment strategy 2 (PT2): The procedure applied in PT1 was applied in PT2 too, followed by a curing treatment in the Form Cure Unit at 25 °C for 1 h.

Post-treatment strategy 3 (PT3): PT3 combined the steps described in PT2 followed by a thermal treatment of the pieces at 160 °C for 3 h.

In addition, the washcoating method was selected for studying the coating of the printed pieces (plates and POCS). For this, Al₂O₃ was selected as a generic oxide that represents the main features of a typical catalyst or an active adsorbent that can be structured in this kind of geometry. Therefore, a commercial slurry of such an oxide was used (Nyacol^® AL20), supplied by the Department of Ceramics/COMERCIAL QUÍMICA MASSÓ.

The printed pieces were submitted to several coatings, keeping them submerged in the slurry for 1 min. Immersion and emersion were conducted at a controlled speed (2 m/min). Three different concentrations of Nyacol^® AL20 (Nyacol Nanotechnologies, Inc., Ashland, MA, USA) were tested for the coatings: (a) Nyacol^®/H₂O 33 v/v%; (b) Nyacol^®/H₂O 66 v/v%; (c) Nyacol^® 100 v/v%.

3.3. Characterization Techniques

Thermal analysis: A dynamic thermogravimetric study was performed to analyse the effect of the post-treatment of the printed pieces in a Perkin-Elmer^® STA 6000 thermobalance (Perkin-Elmer^® STA 6000, Waltham, MA, USA). For the simultaneous TGA and DSC analyses, ~20 mg of sample were heated from 30 to 500 °C (20 °C/min) under 20 mL/min of O₂.

Thermal microscopy: A Heating Microscope MicrOvis (Camar Elettronica^®, Modena, Italy) was used to evaluate shape variations in the material with which the structured ones are printed, due to the sintering, softening, sphere, half sphere or melting produced with the increment of temperature. For this, a cylinder with a diameter of 2 mm and a height of 5 mm was printed and deposited in the sample holder. The mentioned sample holder is inside a heated zone that allowed the temperature to be raised from room temperature to 600 °C at a rate of 2 °C/min. The equipment has a camera that allows the silhouette of the printed piece to be digitally recorded and detects millimetric changes in the said silhouette as a function of time and the evolution of temperature.

Scanning Electronic Microscopy (SEM): SEM micrographs were taken in a Merlin microscope Carl Zeiss^® Oberkochen, Germany, equipped with a field emission gun (FEG), and an Oxford Inca Energy 350X-MAX-50 EDX analyser with linear resolution 127 eV in Mn Kα 1–1 × 10⁵ cps.

4. Conclusions

Firstly, in response to the question posed by the title of this paper and taking into account the information presented, it can be said that there is no easy transition between the design and printing of structures such as POCSs. Additionally, this is something that should be taken into account in future studies of printed structures that will later be used as systems to structure solids, to avoid conclusions about advantages in improving the process due to geometric effects without knowing the limits of these parameters.

Although the versatility in the types of achievable geometries and the high degree of detail are the two premises that are considered great advantages of additive manufacturing, they can really be limited to much narrower areas both due to aspects of the designs themselves, as well as the limits of printing and the workability of the parts after printing.

Therefore, in light of the results presented, it is suggested to continue delving into the design of experiments that allow obtaining truly normalized information and thus are able to extract information truly associated with the geometric effects of the printed pieces. This is especially recommended in the emerging field of application of printed structures as systems for structuring adsorbents or catalysts, in which many phenomena are combined simultaneously, and an inadequate experimental design can generate erroneous conclusions.

The analysis system in this work was a set of POCSs printed in a high-temperature resin, which, in light of the results, may well be used as a manufacturing material for supports that serve to structure adsorbents or catalysts applicable to thermal processes below 200 °C.

This printing material allows the generation of pieces with a high degree of detail, and the thermal stability not only guarantees the resistance of the piece but also its interaction with metal oxides such as alumina. However, it is recommended to explore the interaction with other materials more specifically and to test the stability of the stated interaction in both oxidizing and reducing environments or the presence of some solvent vapours to evaluate the integrity of the printed pieces.

Finally, taking advantage of the versatility of the fabrication method explored, a cartridge-type design that fits properly is proposed. Therefore, the STL files are included as Supplementary Material so that other researchers can explore new ways of taking advantage of this assembly mechanism.