Optimizing the Material Extrusion Process for Investment Casting Mould Production

Abstract

This study investigates the optimization of the Material Extrusion (MEX) process for producing polylactic acid (PLA) patterns used in investment casting moulds, specifically targeting the casting of non-ferrous alloys such as brass. Key MEX process parameters—layer thickness, wall thickness, infill density, and post-processing with dichloromethane vapour for surface enhancement—were systematically analyzed for their impact on mould quality. Results indicate that an optimized combination of MEX parameters yields moulds with high dimensional accuracy, low surface roughness, and minimal pattern residue within the mould cavity. These optimized moulds were subsequently used in brass casting, with the final cast parts evaluated for dimensional precision and surface finish. The study concludes that PLA patterns manufactured via optimized MEX parameters provide a precise, cost-effective, and easy-to-implement solution for industry applications. Additionally, this process is environmentally friendly and presents clear advantages over other pattern-making methods, offering a sustainable alternative for producing complex metal parts with reduced environmental impact. The findings underscore the significant role of post-processing in enhancing mould quality and, consequently, the quality of the cast parts.

1. Introduction

The casting process remains fundamental within the industrial manufacturing sector for producing complex components of varying sizes and in different metals, each with unique properties. Historically, conventional casting methods have enabled efficient mass production of standardized parts, yet the modern industrial landscape demands advanced manufacturing techniques that allow for greater product complexity, customization, cost-effectiveness, and reduced environmental impact. The integration of additive manufacturing (AM) into casting is emerging as a transformative approach to optimize and improve conventional casting processes, minimizing both production time and costs [1,2]. Moreover, AM enhances the design of casting-related components such as patterns, cores, moulds, and shells, providing a pathway toward more sustainable and efficient manufacturing practises [3,4,5,6].

One promising application of AM within the casting sector is in the production of disposable patterns for the investment casting (IC) process [7]. Traditionally, these patterns have been produced either by manual, labour-intensive processes for unique components or via moulds for serial production, which restrict geometric complexity and require the incorporation of draft angles. AM processes such as material jetting (MJT) [8] and material extrusion (MEX) [9,10,11,12] offer solutions to these limitations, with MEX showing advantages in terms of accessibility, cost-efficiency, and material versatility. Comparative analyses indicate that, while MJT enables high geometric precision and fine surface quality, the high cost of MJT equipment (ranging from USD 50,000 to USD 500,000) and the toxicity of materials commonly used in MJT make it less accessible [13,14]. MEX machines, however, are significantly more affordable, ranging from approximately USD 300 for basic models to USD 100,000 for advanced industrial systems and support a broader range of materials, particularly PLA—a biodegradable, non-toxic, and economically viable polymer [15,16].

Furthermore, the sustainability profile of MEX is bolstered by the ability to process PLA, derived from renewable and fully biodegradable sources. Unlike MJ, which utilizes photopolymers with limited recyclability and higher environmental impact, MEX reduces toxic waste generation and aligns better with environmental objectives in casting [17,18]. The ease of post-processing PLA patterns, such as the simple removal of support structures, adds to its practical advantages. Surface quality can be further improved with chemical treatments, such as a vapour bath in dichloromethane, which smoothens the PLA surface by dissolving outer layers slightly, leaving a smoother finish and reducing the visible layering effect typical of AM [19]. This process typically achieves up to 70% improvement in surface smoothness and an increased surface gloss, while minimizing dimensional alterations with controlled exposure times (30 s) [20].

Given these strengths, recent studies have examined the use of MEX patterns in various materials for IC, demonstrating that material properties significantly impact mould cavity quality, including surface and dimensional accuracy, which, in turn, affect the quality of the cast part [21,22]. PLA has emerged as an ideal material due to its low cost, straightforward MEX processing, and lack of warping or cracking, unlike Acrylonitrile Butadiene Styrene (ABS), which leaves residues that compromise part quality [23,24,25,26,27]. The thermal stability of materials such as PLA and Polyvinyl Butyral (PVB) in IC moulds has been proven to be essential, with PLA showing minimal cracking and emitting no harmful gasses, thus further supporting its use [28,29,30,31].

This study aims to explore the optimal configuration of MEX process parameters to produce high-quality PLA patterns suitable for IC. The initial phase of this research systematically investigates the impact of key MEX process parameters—including layer height, wall thickness, infill density, and post-processing with dichloromethane vapour—on the dimensional accuracy, surface roughness, and residue levels within the resulting mould cavities. After identifying the optimal MEX parameter configuration and evaluating the effectiveness of chemical surface treatments, the study validates the use of PLA patterns for brass alloy casting in IC.

The quality of the cast parts is then analyzed by correlating the characteristics of the mould cavities, such as dimensional fidelity and surface finish, with the final properties of the brass castings. Finally, the study demonstrates the practical synergy between MEX and IC through case studies involving geometrically complex parts, showcasing MEX’s adaptability, cost-effectiveness, and environmental advantages over MJ in pattern production. These findings highlight the significant potential of optimized MEX configurations with PLA as a sustainable, precise, and industry-applicable solution for advanced casting applications, positioning this method as a viable alternative in the production of high-quality metal components.

2. Materials and Methods

In this section, the equipment used in this research is presented, both for the fabrication of patterns, moulds, and casting parts, as well as for evaluating their quality in terms of dimensions, roughness, and the presence of residues. Subsequently, the methodology employed in the study is described.

2.1. Polymer Models

The patterns used in this study were obtained through additive manufacturing using the MEX process on an Ultimaker 2+ machine (Ultimaker, Utrecht, The Netherlands). Ultimaker Cura 5.3 software was employed to generate the G-code executed by this machine, allowing for the configuration of MEX process parameters.

Among the materials available for the MEX process, black PLA supplied by Ultimaker was selected for several key reasons. First, PLA’s hydrophobic nature prevents it from absorbing moisture during the wet plaster coating process, thereby minimizing the risk of pattern deformation. Additionally, PLA has a slightly higher melting point than the plaster drying temperature used in this study, with a melting range between 170 °C and 180 °C compared to a drying temperature of 150 °C. This temperature differential enables effective plaster drying without degrading the pattern material or distorting the cavity geometry. Furthermore, PLA’s sensitivity to dichloromethane (CH₂Cl₂) allows for surface enhancement through a vapour treatment, significantly improving the finish. Finally, the black colour of PLA was chosen to facilitate the detection of any potential pattern residues on the white cavity surface, aiding in quality assurance.

To analyze the influence of MEX main process parameters and the application of post-processing on the quality of the obtained patterns, a factorial design of experiments (DOE) was implemented, considering the factors presented in

Table 1. Factors and chosen levels for the analysis of MEX-printed parts for indirect casting.

Factors	Units	Low Levels	High Levels
Number of initial/final layers	Layers	2	5
Layer height	mm	0.06	0.10
Wall thickness	mm	0.40	0.80
Infill density	%	5	20
Post-processing	-	No	Yes
Factors	Units	Low levels	High levels

.

Due to the labour-intensive nature of the mould fabrication process, a resolution V factorial design was selected. This design minimizes the number of experimental combinations from 2⁵ to 2⁵⁻¹ (16 combinations) by creating aliases only between two- and three-factor interactions [32]. The specific factor combinations are presented in

Table 2. Coding and configuration of MEX parameters for each test specimen.

ID	Post-Processing	Layer Thickness	Wall Thickness	Number of Initial/Final Layers	Fill Density
		mm	mm	n°	%
F1	Yes	0.06	0.40	2	5
F2	No	0.06	0.40	2	20
F3	No	0.06	0.40	5	5
F4	Yes	0.06	0.40	5	20
F5	No	0.06	0.80	2	5
F6	Yes	0.06	0.80	2	20
F7	Yes	0.06	0.80	5	5
F8	No	0.06	0.80	5	20
F9	No	0.10	0.40	2	5
F10	Yes	0.10	0.40	2	20
F11	Yes	0.10	0.40	5	5
F12	No	0.10	0.40	5	20
F13	Yes	0.10	0.80	2	5
F14	No	0.10	0.80	2	20
F15	No	0.10	0.80	5	5
F16	Yes	0.10	0.80	5	20

.

A custom-designed test specimen was created using CATIA V5 SP21 software. This specimen consists of a 30 mm cube with a truncated cone transition from one face of the cube to a circular section with a 10 mm diameter (see Figure 1a). The truncated cone transition extends 22.5 mm in length. This geometry not only allows for effective pattern material removal in a later phase, but also facilitates precise measurement of the dimensions of both the patterns and the resulting moulds. The 16 patterns specified in the DOE (see Figure 1b) were produced in two batches of eight specimens, arranged in a 4 × 2 matrix. All specimens were fabricated consecutively using the same coil of PLA material.

For post-processing the eight printed parts, a custom-designed airtight chamber was used. This chamber has a prismatic design, equipped with a heating element at its base and a small stainless steel container to hold liquid dichloromethane. The sample is suspended inside the chamber using a hook. The heating element heats and vaporizes the dichloromethane, creating a vapour atmosphere that uniformly affects the entire surface of the sample, regardless of complex or intricate geometries. The heating element reaches a temperature of 200 °C in approximately 3 s, with a sample exposure time of 30 s. This process yields parts with a high gloss and excellent surface quality.

2.2. Ceramic Mould Manufacturing

To manufacture the ceramic moulds, ULTRA-VEST investment material from supplier Ransom & Randolph (Maumee, United States of America) was used. This material is specifically formulated for creating investment casting moulds suitable for non-ferrous alloys with melting points below 1200 °C.

The process involves coating the previously printed pattern with the ceramic slurry. This slurry is prepared by mixing the ULTRA-VEST powder with water according to the manufacturer’s specifications, then pouring it into the container holding the pattern (Figure 2). After coating, a vacuum is applied at an absolute pressure of 150 ± 1 mbar using a Griño Rotamik (Barcelona, Spain) vacuum pump and Airon ZSE30 digital vacuum gauge (Airon-Pneumatic Iberica S.L., Barcelona, Spain) to remove air bubbles trapped within the ceramic slurry. The mould is then left to dry for 8 h, resulting in a “green” mould. Finally, a thermal treatment cycle is applied to remove the pattern and impart the desired mechanical properties to the mould. During this cycle, the mould is positioned so that the pattern faces downward, facilitating the removal of the molten pattern material (Figure 2). The thermal treatments were performed using a 12PR/400 muffle furnace (Hobersal, Barcelona, Spain) equipped with a temperature programmer.

The thermal cycle begins with a 3 h hold at 150 °C to remove moisture from the coating material. This is followed by a heating phase at a rate of +100 °C/h up to 730 °C, where it is maintained for 3 h to eliminate the plastic pattern. A controlled cooling phase to room temperature is then conducted at a rate of −20 °C/h to prevent potential cracking of the material due to thermal stress. Once the moulds were cooled, they were cut as shown in Figure 2 to proceed with the quality analysis.

Mould Quality Evaluation

The quality of the obtained moulds was assessed in terms of dimensional accuracy (distance between two faces of the cubic geometry), surface quality (average roughness, Ra), and the presence of residues in the mould. To conduct this evaluation, the moulds were cut using the saw. The dimension between two opposite faces of the cubic zone was measured with a Mitutoyo two-contact micrometre following ISO 5725 [33] recommendations, taking five measurement points with five replicas between each pair of opposite faces (Figure 3a). To assess this quality, the indicator described in Equation (1) was defined, considering the deviation of the cavity dimensions from the dimensions of the pattern used to manufacture the mould in the horizontal X and Y directions (Figure 3a). This indicator, $\mathsf{\delta }$, allows assessing the fidelity of the mould-manufacturing process to achieve dimensions as similar as possible to the pattern dimensions. The lower this indicator, the better the dimensional quality of the mould.

$$\mathsf{\delta }={\displaystyle \frac{\left|\mathsf{\Delta }{\mathrm{d}}_{\mathrm{X}}\right|+\left|\mathsf{\Delta }{\mathrm{d}}_{\mathrm{Y}}\right|}{2}}={\displaystyle \frac{\left|{\mathrm{d}}_{\mathrm{X}}^{\mathrm{P}\mathrm{a}\mathrm{t}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{n}}-{\mathrm{d}}_{\mathrm{X}}^{\mathrm{C}\mathrm{a}\mathrm{v}\mathrm{i}\mathrm{t}\mathrm{y}}\right|+\left|{\mathrm{d}}_{\mathrm{Y}}^{\mathrm{P}\mathrm{a}\mathrm{t}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{n}}-{\mathrm{d}}_{\mathrm{Y}}^{\mathrm{C}\mathrm{a}\mathrm{v}\mathrm{i}\mathrm{t}\mathrm{y}}\right|}{2}}$$

(1)

Surface quality was measured with a SURFTEST SJ-500/SV-2100 (Mitutoyo, Takatsu-ku, Japan) roughness profilometer equipped with a 5 µm stylus on the four plain faces of the cavity (Figure 3b), using a cutoff length (λc) and a filter (λs) of 2.5 mm and 8 μm, respectively, following ISO 4287 recommendations [34]. Finally, the presence of residues was assessed using a matrix of 15 images for each face of the cubic cavity using a Z16 APO optical microscope (Leica, Wetzlar, Germany), for which the mould was cut into five parts using a 170/K023 (Sierras SABI, Azkoitia, Spain) saw (Figure 3c), four lateral parts, and the bottom. A total of 75 images per mould were obtained and analyzed in detail.

2.3. Metal Pouring

After analyzing the quality of the mould produced using different DOE factor levels (Table 2), a new mould was manufactured with the optimal factor configuration, as detailed in Section 3. This mould underwent a complete casting process using C85700 brass alloy, with composition specified according to ASTM B176-18 [35] (

Table 3. Chemical composition of the commercial alloy C85700.

Elements	Cu	Zn	Pb	Sn	Ni	Al	Fe	Si	Residuals
% Weight	58–64	32–40	0.8–1.5	0.5–1.5	0–1.0	0–0.8	0–0.7	0–0.05	0–1.3

). The supplier was Wisco Española, S.A., a company specialized in alloy provision. The melting temperature of this alloy is approximately 950 °C, characterized by excellent castability, ease of machining, low friction coefficient, and high corrosion resistance. Typical applications of this alloy include those requiring low friction and corrosion resistance, such as gears and locks.

The brass casting was conducted in a 12 kW Agatronic R (Argenta, Brzeziny, Poland) induction furnace designed for investment casting. The alloy was heated to 1050 °C for 5 min, followed by a 3 min holding time before being poured into the mould under vacuum conditions. Vacuum was applied using the previously mentioned pump and gauge, at an absolute pressure of 150 ± 1 mbar. This vacuum was initiated just before pouring and maintained until 90 s after pouring, allowing solidification of the piece within the mould.

After metal solidification, the mould was immersed in room-temperature water to dissolve it and facilitate the removal of the cast piece. Once the mould material was removed, the cast piece was cleaned with a brush, and the gating system was cut off using a saw. The quality of the piece was then analyzed in terms of dimensions and surface finish. The analysis procedure was similar to that applied for assessing the dimensional and surface qualities of the moulds, though sectioning of the piece was unnecessary due to its accessibility for measurements. The sequence can be seen in Figure 4.

2.4. Case Studies

After validating the proposed procedure, the casting of complex geometry parts was analyzed as a several cases studies. Two ornamental parts and an industrially applicable part were analyzed. For the ornamental figures, a bust (Figure 5a) and a silhouette (Figure 5b) were created, both with a high level of detail. As an industrial application part, a bevel gear (Figure 5c) was analyzed.

3. Results and Discussion

In this section, the obtained results from the analysis of the influence of MEX process parameters, as well as the potential impact of post-processing, on the quality of the cavity in terms of dimensional accuracy, surface quality, and residue presence from the disposable pattern are presented. Additionally, the results obtained for the method validation and those from various case study parts are discussed.

3.1. Dimensional Quality and Post-Processing Effects

Figure 6a illustrates the δ parameter value, as described in the P3 phase of the methods section, obtained after fabricating and measuring the cavities. Figure 6b presents the Pareto diagram with α = 0.05, identifying the most influential factors affecting this parameter.

As observed in the Pareto diagram (Figure 6b), the most influential aspect in the pattern–cavity dimension relationship is a combination of fill density and wall thickness. Thus, the tests that show the least difference between pattern and cavity dimensions are those manufactured with lower fill density and wall thickness (F7, F2, F4, and F16), as depicted in Figure 6a. These patterns ultimately have less mass and lower mechanical strength, preventing them from expanding within the mould cavity due to temperature effects during the thermal treatment applied to remove its material. Consequently, they avoid altering the cavity dimensions produced by the pattern at room temperature. Conversely, patterns with higher mass and strength, suggesting greater volumetric expansion (F3, F11, and F9), generate more significant cavity deformation, adversely affecting mould quality. From these results, it is evident that test F7 has the smallest pattern–cavity deviation, i.e., 0.127 mm, while Probe F9 has the largest, i.e., 0.510 mm.

Before proceeding with the analysis of other aspects of mould quality and considering that pattern F7 achieved the best dimensional results (it was post-processed with dichloromethane vapour), it is worthwhile to assess the influence of the post-processing vapour treatment on the dimensions of the pattern itself. This will provide insights into the impact of post-processing on pattern dimension changes, enabling the future attainment of cavities with dimensions closer to the nominal dimensions sought during pattern fabrication. Figure 7 illustrates the dimensional changes due to post-processing, evaluated with the δ parameter applied exclusively to the dimensions of these patterns before and after post-processing. Logically, this analysis is unique to those patterns treated with the vapour treatment post-processing.

As can be seen, the application of post-processing systematically reduces the dimensions of the pattern, since a chemical attack erodes its surface. This erosion ranges between 0.03 mm and 0.11 mm, which is 0.10–0.37% of the nominal dimension of the pattern. Specimens F7, F6, F1, and F16 show the least erosion, although the changes among them are very similar, within 0.01 mm.

3.2. Surface Quality

The average roughness value, Ra, of the moulds after pattern removal is shown in Figure 8. As observed, the moulds obtained through post-processed patterns exhibited a substantial reduction in roughness. Values in the range of 3.03–4.75 µm are evident for moulds obtained from post-processed patterns, compared to the range of 6.00–8.74 µm for moulds obtained from untreated ones. The highest surface quality was identified in pattern F16, although all post-processed patterns exhibited similar values and fell within the narrow margin mentioned above.

Similarly, to the case of dimensional quality, it was decided to analyze the influence of post-processing on the change in surface quality of the patterns. Figure 9 shows the roughness before and after post-processing for the treated patterns. All post-processed patterns exhibited very similar roughness, around 1.7 µm. On the other hand, the Ra value decreased by approximately 6 µm after applying post-treatment. Another noteworthy aspect of these results is that the pattern’s roughness does not directly transfer to the cavity; instead, the cavity exhibits a roughness around 2 µm higher (see Figure 8). This increase in roughness occurs on the surface of the coating material used to manufacture the mould. The explanation lies in alterations suffered by this material during the thermal treatment for pattern removal, resulting from both physical changes in this material and the fusion and detachment of PLA material from the melted pattern. In any case, although post-processing causes erosion in the pattern’s dimensions, and the roughness obtained in the mould is higher than that of the pattern used for its fabrication, the beneficial effect of this treatment in achieving moulds with higher surface quality, which will be transferred to the cast part manufactured with it, is undeniable.

3.3. Presence of Pattern Residues

The main objective of this study is to manufacture moulds by coating plastic patterns produced in MEX so that the cavities of these moulds are as similar as possible to the patterns themselves. Beyond that, it is essential that these cavities be as clean as possible, considering that the material of the patterns is removed through melting. The presence of these residues can lead to incomplete filling and their evaporation during the pouring process, causing porosity or other defects in the cast part due to the capture of gasses generated in the evaporation of residual plastic in contact with the liquid metal.

Upon visual inspection of the analyzed moulds, it was observed that all PLA material had liquefied without leaving physical residues. Marks on the cavity surface were also observed, attributed to the characteristic texture of parts manufactured using MEX (Figure 10a). In a more detailed analysis using optical microscopy (Figure 10b) of the cavity surfaces, the absence of residues in all analyzed parameter combinations was confirmed (

Table 4. Summary of optimal printing parameters: F7 combination.

Post-Processing	Layer Thickness	Wall Thickness	Number of Initial/Final Layers	Fill Density
	mm	mm		%
Yes	0.06	0.80	5	5

). This demonstrates the suitability of PLA for manufacturing moulds using this procedure.

3.4. Validation of the Method

Considering the results obtained in the analysis of mould quality, the conclusion is drawn that the optimal combination corresponds to the F7 specimen. While it is true that the configuration with the best surface finish is associated with the F16 configuration, the difference in roughness compared to the F7 specimen is minimal. Therefore, it is more interesting to choose the latter for achieving better overall results, i.e., a combination of dimensional and surface quality. Table 4 summarizes the optimal MEX parameters.

The complete casting process is detailed using the optimal parameter combination for the part geometry used as a pattern. The dimensions and roughness of the pattern during the process, as well as those of the obtained part, are analyzed. It is worth mentioning that the mould quality is not evaluated in this process as it requires destructive tests; refer to the mould quality evaluation section. Figure 4 presents the pattern, the mould, and the obtained part in the process.

In Figure 11, the average dimension of the pattern and the roughness before and after post-processing are shown. These results confirm what was observed in the previous analysis, namely, the post-processing with dichloromethane, causes erosion in the pattern but improves its surface quality. Regarding the obtained cast part, on the one hand, there is some contraction in comparison to the dimensions of the pattern in the condition used to fabricate the mould, i.e., post-processed pattern. This contraction is known in the casting process and is due to a combination of the metal shrinking during cooling from the melting temperature to room temperature.

As for roughness, a significant increase in roughness is observed when compared to the pattern. This increase is logical when considering the roughness increase that occurs between the pattern and the mould cavity, due to physical changes in the coating material during the thermal treatment for pattern removal. In any case, it is important to highlight that the roughness achieved with this method is good considering the typical roughness range in coating casting processes (from 0.8 µm to 5 µm Ra).

3.5. Case Studies Manufacturing

Once the method has been validated on the cubic test part, this section presents several case studies that demonstrate the procedure’s potential for use in the current foundry industry, given the advantages it offers. On the one hand,

Table 5. Examples of ornamental parts obtained with the method.

	3D Pattern	PLA Pattern	Blank Part	Final Part
1
2

shows two examples of ornamental parts obtained with the method. As seen, despite their small dimensions and high geometric complexity, the details obtained after the casting process are highly faithful.

On the other hand, in

Table 6. Example of an industrial application part obtained with the method.

3D Pattern	PLA Pattern	Blank Part	Final Part

, an industrial application part obtained with the method is shown. In this case, once again, a high quality of part reconstruction can be observed.

4. Conclusions

This research presents an optimized process for manufacturing metal casting moulds via the investment casting method, utilizing ULTRA-VEST material and PLA patterns produced through the MEX additive manufacturing process. Combining MEX with casting techniques enables a cost-effective, precise, and environmentally sustainable approach to producing moulds of complex metal geometries, making this process highly relevant for industrial applications.

The optimal MEX parameter configuration includes a lower layer height, reduced wall thickness, and low infill density, which collectively minimize the pattern’s mass. This reduces thermal stresses during the pattern removal phase, prevents deformation of the mould cavity, and ensures superior geometric fidelity. Additionally, post-processing with dichloromethane vapour substantially improves surface quality by smoothing the MEX layer texture, albeit with a slight dimensional reduction. This vapour treatment effectively removes any pattern residue, enhancing the mould’s suitability for precision casting.

Validation of the process was conducted through brass casting of complex shapes, including ornamental components and a bevel gear, achieving highly detailed reproductions of the original patterns. The study highlights the practical and sustainable benefits of using MEX-based patterns in casting applications. Future research will aim to refine dimensional compensations for shrinkage during both casting and post-processing, advancing the technique for high-precision industrial applications.