The integration of additive manufacturing (AM) with traditional foundry practices represents a significant paradigm shift, offering unprecedented design freedom and rapid prototyping capabilities. One of the most promising avenues is the direct fabrication of ceramic molds or shells for precision investment casting. Traditional methods for creating these intricate ceramic shells, such as the sequential slurry dipping and stuccoing around wax patterns, are labor-intensive and time-consuming. The advent of vat photopolymerization, particularly Digital Light Processing (DLP) technology, presents a compelling alternative by enabling the layer-by-layer fabrication of complex ceramic green bodies directly from a digital model. This work details the first-person development and parametric study of an in-house built, bottom-up DLP 3D printer specifically tailored for processing photopolymerizable ceramic slurries aimed at precision investment casting applications. Our focus was on creating a cost-effective platform to explore the feasibility of manufacturing fine-featured, high-resolution ceramic shells that could withstand the subsequent thermal processing and metal pouring stages of precision investment casting.
The core principle of DLP vat photopolymerization involves using a digital micromirror device (DMD) to project a mask image of a single cross-section onto the bottom of a vat filled with a photosensitive resin or slurry. Where the light of sufficient intensity and wavelength strikes, the photoinitiator triggers polymerization, solidifying that layer. The build platform then lifts, separating the cured layer from the vat’s transparent bottom, recoats the vat with fresh material, and the process repeats. For ceramic manufacturing, the slurry consists of a high loading of ceramic powder (e.g., silica, alumina, zircon) dispersed within a photopolymerizable resin system. After printing, the “green” part undergoes thermal debinding to remove the polymer binder, followed by sintering to densify the ceramic particles, resulting in a purely ceramic component—in this context, a mold for precision investment casting.
We selected an upside-down (bottom-up) configuration for our printer design. While this approach mitigates the severe sedimentation issues common in high-solid-loading ceramic slurries used in top-down systems and reduces material waste, it introduces challenges such as light attenuation through the vat bottom and the need for a reliable separation mechanism between the cured layer and the vat. The primary systems of our developed printer include: the DLP Exposure System, the Forming System (vat and build platform), the Z-axis Motion System, and the Integrated Control System.
The heart of the apparatus is the modified DLP projection system. We repurposed a commercial Acer H6517ABD projector. Its standard high-pressure mercury lamp emits the necessary UV spectrum, but factory-installed UV filters significantly attenuate the required curing wavelength. We carefully replaced this filter with a quartz glass pane, which offers high UV transmittance and thermal resistance. The projector’s color wheel, which would otherwise absorb UV light, was physically relocated within the housing to bypass its light path without triggering the projector’s fault detection circuitry. Furthermore, to reduce the overall machine footprint and increase irradiance at the curing plane, we adjusted the projection lens assembly to shorten the throw ratio. Based on the thin lens equation, reducing the focal length allows for a reduced image distance while maintaining a viable object distance:
$$ \frac{1}{f} = \frac{1}{u} + \frac{1}{L} $$
where $f$ is the focal length, $u$ is the object distance (from DMD to lens), and $L$ is the image distance (from lens to vat bottom). By modifying the lens setup, we achieved a high-resolution, high-intensity UV projection area suitable for desktop fabrication.
The forming system comprises a high-transparency acrylic vat and an aluminum build platform. A critical component is the flexible, UV-transparent film adhered to the vat’s bottom. This “release layer” minimizes the adhesion force between the cured part and the vat, facilitating successful layer separation during the peeling stage. We utilized a polydimethylsiloxane (PDI\*S) silicone film due to its excellent release properties, high transmittance >90% in the 365-405 nm range, and good chemical resistance. The aluminum build platform was sandblasted to enhance adhesion for the first printed layers.
Precise vertical motion is controlled by the Z-axis system. Since the DLP projector defines the X-Y resolution, the Z-axis only requires accurate, jitter-free linear motion. We employed a NEMA 23 stepper motor (57HS5630B4D8) coupled with a precision ground ball screw (diameter 10mm, lead 4mm). The motion was guided by two linear rails to ensure stability and minimize deflection. A limit switch at the home position provided accurate initialization. The control system is built around an Arduino Mega 2560 microcontroller board, which interprets G-code commands and drives the stepper motor driver. The entire machine, showcasing the integrated systems, was successfully assembled.
With the hardware operational, we conducted a series of experiments to determine the optimal processing parameters for a commercially available photopolymerizable silica-based ceramic slurry intended for precision investment casting. The key parameters investigated were Z-axis movement speed, layer exposure time, layer thickness, and minimum achievable feature size.
The Z-axis speed encompasses both the lifting speed after curing and the lowering speed for recoating. An excessively high lift speed can cause part detachment from the build platform due to high peel forces, while a very slow speed drastically increases total print time. Similarly, a very fast lowering speed can create turbulence in the slurry. Through iterative testing, we found the robust operating window to be:
$$ 150 \text{ mm/min} \leq v_z \leq 300 \text{ mm/min} $$
Speeds within this range ensured reliable layer separation and adequate slurry recoating without undue time penalty.
Determining the correct exposure time ($t_e$) per layer is paramount. Unlike pure resins, ceramic slurries exhibit significant light scattering due to the refractive index mismatch between the ceramic particles and the resin matrix. This scattering broadens the cure profile, meaning the practical cure depth ($C_d$) does not follow the classic Beer-Lambert law perfectly. The effective cure depth can be approximated by a modified formula:
$$ C_d = D_p \ln\left(\frac{E}{E_c}\right) = D_p \ln\left(\frac{I_0 \cdot t_e}{E_c}\right) $$
where $D_p$ is the penetration depth (a scattering-dependent constant), $E$ is the exposure energy density, $I_0$ is the irradiance at the window, and $E_c$ is the critical energy density for gelation. To find the suitable $t_e$, we performed single-layer exposure tests of an 80mm diameter circle at varying times. The results are summarized below:
| Exposure Time, $t_e$ (s) | Observation & Curing State | Measured Layer Thickness (mm) |
|---|---|---|
| < 20 | No solidification occurred. | N/A |
| 26 | Weak solidification; layer curled when handled. | ~0.18 |
| 30 | Good solidification, distinct shape. | 0.206 |
| 31 | Good solidification. | 0.210 |
| 32 | Good solidification. | 0.214 |
| 33 | Good solidification, slight over-cure at edges. | 0.219 |
| 34 | Good solidification, noticeable over-cure (burr). | 0.225 |
| 35 | Good solidification, significant over-cure. | 0.228 |
The data shows that a minimum threshold energy (∼30s) is required for adequate polymerization and handling strength. However, exposure times exceeding ∼33s lead to over-curing, where light scattering causes curing beyond the intended pixel boundaries, reducing feature accuracy and creating burrs. Therefore, the optimal exposure time per layer was determined to be:
$$ 30 \text{ s} \leq t_e \leq 35 \text{ s} $$
For the first few layers (the “burn-in” layers), a longer exposure time of 45s was used to ensure strong adhesion to the build platform.
The layer thickness ($L_t$) set in the slicer software must be less than the practical cure depth ($C_d$) to ensure strong inter-layer bonding. From the table above, the cure depth for $t_e$ = 30-35s ranges from 0.206mm to 0.228mm. Setting $L_t$ too close to $C_d$ risks delamination, while setting it too small increases print time and file size unnecessarily and demands higher mechanical precision. A layer thickness of:
$$ L_t = 0.1 \text{ mm} $$
was chosen as a balanced compromise, providing a safety margin below the cure depth for reliable bonding while maintaining good vertical resolution and manageable print duration.
For precision investment casting shells, wall thickness is critical. Walls that are too thin (<~0.9mm) may fracture during debinding/sintering or under the hydraulic pressure of molten metal. Walls that are excessively thick waste material and can cause casting defects. To evaluate the printer’s capability, a test model featuring cylindrical pillars of decreasing diameter (from 4.0mm down to 0.5mm in 0.1mm steps below 1mm) was printed using optimized parameters ($t_e$=32s, $L_t$=0.1mm, $v_z$=200 mm/min). The results defined the minimum printable feature size. Pillars with a designed diameter ≥ 0.9mm were successfully fabricated. Those below 0.9mm fractured during the peeling stage due to adhesive forces exceeding their mechanical strength in the green state. Consequently, the minimum reliable printable feature size for this printer-slurry combination is:
$$ d_{min} \approx 0.9 \text{ mm} $$
Measurement of the printed pillars revealed a consistent positive deviation from the designed diameter, as shown in the following summary table. This error, typically between 0.1mm and 0.2mm, is attributed to light scattering and over-curing effects, which cause lateral curing beyond the projected image boundaries.
| Designed Diameter (mm) | Measured Diameter (mm) | Dimensional Error (mm) |
|---|---|---|
| 4.0 | 4.16 | +0.16 |
| 3.5 | 3.64 | +0.14 |
| 3.0 | 3.19 | +0.19 |
| 2.5 | 2.71 | +0.21 |
| 2.0 | 2.13 | +0.13 |
| 1.5 | 1.67 | +0.17 |
| 1.0 | 1.12 | +0.12 |
| 0.9 | 1.07 | +0.17 |
In conclusion, we have successfully developed and characterized a functional, bottom-up DLP 3D printer for fabricating ceramic green bodies. The system was specifically contextualized for advancing the art of precision investment casting by enabling the direct digital manufacturing of complex ceramic molds. Through systematic experimentation, we established a foundational set of optimal process parameters: a Z-axis movement speed of 150-300 mm/min, a layer exposure time of 30-35 seconds, a layer thickness of 0.1 mm, and a demonstrated minimum feature size of 0.9 mm. These parameters provide a crucial starting point for subsequent research focused on printing actual ceramic shells for precision investment casting, including studies on slurry formulation for higher strength, dimensional accuracy during sintering, and final casting trials. The primary limitation of the current setup is the manual slurry handling, which can lead to inconsistencies. Future work will focus on integrating automated slurry feeding, continuous stirring, and temperature control systems to further enhance the reliability and repeatability of the process for industrial precision investment casting applications.