Effect of sintering temperature on feature resolution and flexural strength of ceramics fabricated through vat photopolymerization additive manufacturing

Abstract

Purpose

Although ceramic additive manufacturing (AM) could be used to fabricate complex, high-resolution parts for diverse, functional applications, one ongoing challenge is optimizing the post-process, particularly sintering, conditions to consistently produce geometrically accurate and mechanically robust parts. This study aims to investigate how sintering temperature affects feature resolution and flexural properties of silica-based parts formed by vat photopolymerization (VPP) AM.

Design/methodology/approach

Test artifacts were designed to evaluate features of different sizes, shapes and orientations, and three-point bend specimens printed in multiple orientations were used to evaluate mechanical properties. Sintering temperatures were varied between 1000°C and 1300°C.

Findings

Deviations from designed dimensions often increased with higher sintering temperatures and/or larger features. Higher sintering temperatures yielded parts with higher strength and lower strain at break. Many features exhibited defects, often dependent on geometry and sintering temperature, highlighting the need for further analysis of debinding and sintering parameters.

Originality/value

To the best of the authors’ knowledge, this is the first time test artifacts have been designed for ceramic VPP. This work also offers insights into the effect of sintering temperature and print orientation on flexural properties. These results provide design guidelines for a particular material, while the methodology outlined for assessing feature resolution and flexural strength is broadly applicable to other ceramics, enabling more predictable part performance when considering the future design and manufacture of complex ceramic parts.

Citation

Bezek, L. and Lee, K.-S. (2024), "Effect of sintering temperature on feature resolution and flexural strength of ceramics fabricated through vat photopolymerization additive manufacturing", Rapid Prototyping Journal, Vol. 30 No. 11, pp. 260-275. https://doi.org/10.1108/RPJ-04-2024-0166

1. Introduction

Advanced ceramics, which can be classified as both structural ceramics and functional ceramics (e.g. for magnetic, electronic and optical applications) (Rahaman, 2003), generally possess high thermal resistance, high hardness, high stiffness, low coefficient of friction and chemical inertness (Lakhdar et al., 2021), making them beneficial in industries including aerospace, automotive, biomedical, chemical, communications, defense, dental, electronics, energy, manufacturing and transport (Li et al., 2020; Zakeri et al., 2020; Lakhdar et al., 2021). Furthermore, using additive manufacturing (AM) to fabricate ceramics provides the opportunity to leverage the potential advantages of AM, including customizability, high resolution, design flexibility, reduction of material waste and time and cost efficiency from reduced tooling and assembly (Gao et al., 2015). However, there are many challenges in ceramic AM, including parts requiring high processing temperatures, parts exhibiting defect-sensitive and anisotropic mechanical properties and parts having poor machining characteristics (Pelz et al., 2021). Other reported challenges include cracking, poor interfacial adhesion, low strength and toughness, poor resolution and surface quality and lack of scalability (Bhuvanesh Kumar and Sathiya, 2021; Lakhdar et al., 2021).

1.1 Vat photopolymerization of ceramics

All seven AM process categories, as defined by ASTM 52900, have been used in fabricating ceramics (Pelz et al., 2021). The focus of this work is the vat photopolymerization (VPP) AM process, also commonly called stereolithography, which uses energy from light to selectively cure liquid resin stored in a vat. Compared to other AM processes, the general advantages of VPP include high accuracy, smooth surface finish, speed and grain isotropy (Li et al., 2020). To successfully use VPP for ceramics, ceramic filler is typically added to photocuring resin and post-processing heating steps are required for polymer debinding and subsequent sintering. A detailed description of the material requirements for ceramic VPP can be found in Rasaki et al.(2021), and Halloran(2016) provides a thorough explanation of the fundamentals of photopolymerization specifically for ceramic VPP.

To achieve accurate and consistent ceramic VPP parts, it is important to establish quality control and optimize process parameters (Li et al., 2020; Löffler et al., 2020). For the VPP process, relevant variables include material parameters (e.g. ceramic particle size and shape, ceramic particle density, differences in indices of refraction between ceramic and polymer and viscosity of the resin) and process parameters (e.g. ultraviolet [UV] intensity, exposure time, part orientation and layer thickness). Even if these variables are optimized to achieve successfully printed parts, end-use parts also require appropriate parameters for post-processing, which involve the cleaning of residual uncured resin, polymer debinding and sintering.

As the VPP process is only the forming process for the green (as-printed) ceramic, the properties of the final part, which depend on the composition and microstructure, are highly influenced by the post-process (Halloran, 2016; Chen et al., 2019). In traditional ceramic manufacturing, understanding the effects of sintering parameters on material and mechanical properties is a common research topic (Manikam et al., 2012; Wan et al., 2014; Al-Hasnawi and Al-Hydary, 2017; Limpichaipanit et al., 2017; Liu et al., 2017; Barrientos-Hernández et al., 2021). In a similar way, a better understanding of sintering effects is also one of the main goals for advancing ceramic VPP. Bai et al.(2021) compared different sintering approaches and found that liquid silicon infiltration was the most effective sintering approach for silicon carbide parts in terms of mechanical strength and density. Xu et al.(2021) used a Formlabs Form 2 printer with Formlabs Ceramic resin and measured the shrinkage of cubes and flexural strength through three-point bend testing when varying the maximum sintering temperature between 1121°C and 1271°C. The authors observed cracking in the parts even with the suggested sintering profile (Formlabs, 2018b), though changing the orientation of a specific geometry eliminated cracking. Others have reported similar cracking using the VPP process with silica-based (Bae and Halloran, 2011; Truxova et al., 2020; An et al., 2022) and alumina (Johansson et al., 2017; Cramer et al., 2021) specimens because of non-optimized printing or post-processing conditions. Zhang et al.(2024) provide a review of defects in ceramic VPP parts, which are commonly pores, cracks, delamination and surface defects. High process variation has also been witnessed after sintering, especially in large parts with inconsistent slurries (Li et al., 2020). As a result, there is a need for a better understanding of how sintering parameters affect part shrinkage and geometric features, such that parameters can be selected to increase the likelihood of consistently fabricating parts with minimal defects.

1.2 Additively manufactured test artifacts

One way to test geometric features is by evaluating a test artifact, which is defined in ISO/ASTM 52902:2019(E) as a benchmarking test piece geometry that is typically used to assess the performance of an AM system. Qualitative and quantitative measurements can be taken to evaluate and/or calibrate an AM system, or in the case of this research, gauge the impact of a specific set of process parameters for a selected material and printer.

There are many aspects involved when considering the design of test artifacts for AM (Rupal et al., 2018; De Pastre et al., 2020). Moylan et al.(2014) provide a review of the early test artifacts for AM; many of which were used for evaluating individual processes, such as to characterize the performance of a new material or to improve or optimize process parameters. The primary focus of their developed artifact (commonly called the NIST test artifact) was to evaluate the performance of metal AM systems and establish a standard (Moylan et al., 2014). The NIST test artifact has been used to assess part distortion (Martínez et al., 2020) and redesigned to evaluate machine performance, focusing on quicker print and inspection times (Giorgetti et al., 2019). Others have proposed unique artifacts for metal laser-based powder bed fusion AM, targeting printer differences (Moshiri et al., 2019), surface analysis (Townsend et al., 2018) and geometric tolerances (Rupal et al., 2021). Various benchmarking designs have been fabricated with VPP to evaluate stair-stepping effects (Shan et al., 2022), examine microscale features (Fritzsche et al., 2023) and optimize process parameters for novel materials (Choi et al., 2022). In a thorough review of test artifacts for assessing geometric performance (Rebaioli and Fassi, 2017), various test artifacts have been introduced for VPP, binder jetting, material jetting, material extrusion, sheet lamination and powder bed fusion AM processes. Out of 72 sources, there is only 1 source that mentions a test artifact for ceramics, and it is within a study heavily dominated by metal AM. A ceramic test artifact has been fabricated by using binder jetting AM to assess the capabilities of layerwise slurry deposition (Zocca et al., 2019). To the author’s knowledge, the evaluation of ceramic VPP through a test artifact has not been attempted. As a result, there has not been a robust assessment of the geometric aspects of a ceramic VPP part, such as feature resolution, survivability and shrinkage.

The ISO/ASTM 52902:2019(E) standard offers seven types of artifact geometries: linear artifacts, circular artifacts, resolution pins, resolution holes, resolution ribs, resolution slots and surface texture. While it is stated in the standard that it is not intended for a user to build all features to adequately examine an AM system, this study is focused more on examining the survivability of features across different sintering parameters. Recognizing that the sintering parameters are external to the AM system, the effects of sintering could change for each of the artifact geometries. As a result, the current study aims to assess how sintering affects a variety of features to aid designers with future, real-world parts.

1.3 Overview

In this work, the authors outline a methodology for assessing geometric features and strength and demonstrate how it can be applied to a particular silica-based material. Test artifacts are developed to analyze how changing sintering temperature affects feature resolution and survivability. Particle size distribution is also measured. Finally, mechanical performance is considered because a certain degree of mechanical strength is necessary for handleability and functionality. To assess the effect of sintering temperature on mechanical properties, three-point bend testing has been previously conducted on parts fabricated using VPP with iron oxide-doped zirconia (Wang et al., 2023) as well as alumina both with (Qian et al., 2023) and without (Li et al., 2021; Chen et al., 2022) sintering aids. Here, the effect of sintering temperature on flexural properties is investigated through three-point bend testing using specimens printed in multiple orientations. Improved understanding of the process-property relationships for ceramic AM at the macroscale will guide the post-processing conditions and design considerations for manufacturing future advanced ceramic parts using VPP.

2. Methods

2.1 Additive manufacturing materials and processing

The material used for this study was Formlabs Ceramic. This is a commercial material for which the full composition is not disclosed, but it includes an acrylate-based photopolymer and a silica-based ceramic (Formlabs, 2018a, 2019). According to an analysis by Nawrot and Malecha(2020), the material is composed of ∼65% ceramic. The material was printed on a Formlabs Form 2 VPP system. The laser spot size is 140 µm, and the layer thickness is 100 µm. Because the resin is prone to settling and clogging the cartridge, the material cartridge was fully emptied into a secondary container and stirred vigorously before being transferred to the vat. The build plate was sanded to improve adhesion. Once parts were loaded into the PreForm software, default print settings were used with no support. Although PreForm has the option to apply scaling of the Z dimension to account for anticipated shrinkage from sintering, no scaling was applied for this study.

After printing, parts were carefully scraped from the build plate and submerged in isopropyl alcohol for 5 min. A wash bottle was used to clean more fragile features. Parts were then dried in a convection oven at 60°C for 20 min. No UV post-curing was administered, following the suggested guidelines (Formlabs, 2018b).

A Nabertherm muffle furnace was used to administer the post-process sintering program in air, which includes a low-temperature stage to burn out the polymer binding agent and a high-temperature stage for densification. The program, shown in Table 1, was slightly modified from Formlabs’ suggested program for Ceramic material (Formlabs, 2018b). The furnace did not have controlled cooling, so parts were removed after two days when the temperature was below 50°C. The maximum sintering temperature was set to either 1000°C, 1100°C, 1200°C, 1271°C (the company’s suggested value) or 1300°C. These values were selected because although the company provides a sintering profile, prior work indicated that the profile was not optimized (Xu et al., 2021). Consequently, a wide temperature range is investigated to study the evolution of the sintering behavior and compare different sintering parameters.

2.2 Particle size analysis

To prepare the powder, 25 1 cm cubes were printed and cleaned through the same method outlined in Section 2.1. The cubes were sintered to 700°C to ensure all polymer was burned out while not initiating sintering. The cubes were then crushed into powder. Particle size distribution was measured with an Anton Paar PSA 1090 laser diffraction particle size analyzer. Dry measurements were taken over 1 min, and the resulting size distribution was volume-weighted.

2.3 Design of the test artifacts

After referencing typical criteria for test artifacts in ISO/ASTM 52902:2019(E) and (Moylan et al., 2014), a comprehensive list of characteristics was compiled. Features were then selected to demonstrate these characteristics, shown in Table 2.

The artifact was designed to have features of various sizes and shapes, including positive and negative features, be relatively quick to build, consume a relatively small quantity of material, be easy to measure and have meaningful features that could be present in functional parts. The features were divided between two artifacts because in preliminary prints, multiple attempts of a larger artifact containing all features were consistently well-adhered to the build plate and the parts broke during removal. The two artifacts will be referred to as the “wall artifact” and the “slot artifact,” as shown in Figure 1. The XY plane indicates the build plane, while the z-axis aligns with the build direction.

The intent of the artifact study is to highlight changes in the parts that are attributed to the changing post-processing parameters. As a result, quantifying the accuracy of the AM system (i.e. if the printer can produce a feature in a correct location) and thus testing for dimensional accuracy is outside of the scope of this work.

In the statements that follow, the dimensions of the features are outlined. Dimensions regarding the positioning of features are excluded here but can be determined from the provided standard tessellation language files. Feature positions were selected to enable measurement of all features and to provide space between features in the event of failed build geometries. Supports were avoided and care was taken to select dimensions that would be below the minimum feature size (and thus be failed build geometries).

The base of the wall artifact was 50 × 25 × 4.5 mm. It contained five features with the relevant dimensions described as follows:

1. Walls. Six walls were designed with thicknesses of 0.3, 0.4, 0.5, 0.6, 0.7 and 0.8 mm. Walls were 10 mm long and 5 mm tall.

2. Holes. Nine through-holes were designed with diameters of 0.8, 1.0, 1.2, 1.4, 1.6, 1.8, 2.0, 2.5 and 3.0 mm.

3. Pins. Cylinders were designed with both 1 and 2 mm diameters with heights of 4, 8, 12 and 16 mm.

4. Overhangs. Four rectangular prisms were designed with 4 × 2 mm cross sections and set to angles of 50°, 60°, 70° and 80° relative to the XY plane. The heights of the lowest edge of each prism’s uppermost plane were 12 mm.

5. Lateral features. Three circles were designed with diameters of 1, 2 and 3 mm, and the centers of each circle were 4 mm apart. Three squares with sides of 1, 2 and 3 mm were designed with sides 2 mm apart. Parts were centered in the Z direction. Each feature was debossed 2 mm.

The base of the slot artifact was 50 × 26 × 4.5 mm. It contained the following features:

• Slots. A total of 12 slots were designed with thicknesses of 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, 1.25, 1.5, 1.75 and 2.0 mm. Slots were 6 mm long and 2 mm deep.

• Steps. Eight steps were designed with heights of 0.5, 1.5, 2.5, 3.5, 5.0, 5.5, 6.0 and 6.5 mm. Each step was 3 × 4 mm.

• Tubes. Four circles with 8 mm diameters and 3 mm heights were designed with inner diameters of 0, 2, 4 and 6 mm.

• Qualitative features. Several features were added to the artifact for qualitative assessment. These include steps going into the part that is constrained on all sides, positive and negative slopes, positive and negative domes and a cone.

There were two builds for the test conditions, each sintered in a separate cycle so that the artifacts could be placed in the same location and orientation in the sintering furnace each time. Both after printing and after sintering, the test artifacts were weighed and the outer dimensions of each base were measured with calipers. A Keyence VHX-6000 digital microscope was used to measure the artifact features listed in Table 3. The light settings were selected to best highlight each feature and kept constant for that feature to ensure measurement consistency. The microscope has a resolution of 1,600 × 1,200 pixels and was calibrated to 10 µm per pixel.

In general, coaxial lighting offered the most clarity for measuring XY (build plane) features, while partial lighting was preferred for out-of-plane measurements. Owing to anisotropic shrinkage, instead of measuring the diameters of the circular lateral features, vertical and horizontal measurements were taken.

It was hypothesized that the artifact bases could restrict the shrinkage of some features. The tubes were also printed as standalone features for comparison to those on the artifact.

2.4 Flexural testing

For three-point bend testing, ASTM D790-17 was referenced for guidance on part design and testing procedure. ASTM C1161-18, which is specifically for flexural testing of ceramics, was not followed because it is meant for ceramics with strengths greater than 50 MPa, and the reported flexural strength for this material was 33.5 MPa (Formlabs, 2018a).

Rectangular specimens were designed to have a final thickness of 3 mm, width of 4.5 mm and a length of at least 60 mm. To achieve these dimensions, the part design was scaled up for anticipated shrinkage due to sintering. Parts were printed in the XY (i.e. longest dimension along the printer’s x-axis and second longest dimension along the printer’s y-axis), XZ and ZX orientations, and there were 30 specimens per trial. A Type I test was conducted following Procedure A using an Instron 3343 with a 1 kN load cell. The support span was 45 mm, and the rate of crosshead motion was 1.1 mm/min. Flexural stress and strain at break were calculated, and statistical tests were used with a 95% confidence interval to identify statistical significance. The non-parametric Kruskal–Wallis test was used because of inconsistent variance across the data, in part because of unequal sample sizes (discussed more in Section 3.6). Once the Kruskal–Wallis test identified statistical significance by comparing the p-value to α = 0.05, Dunn’s post-hoc test was run for pairwise comparisons of specific conditions. This test identifies statistical significance if the calculated p-value is lower than an adjusted α, and this is further verified by the calculated z-statistic falling higher than the critical z-value. The Real Statistics Resource Pack Add-In for Microsoft Excel was used for this analysis.

The Weibull modulus was also calculated for each test condition. Because this parameter typically exhibits statistical bias, unbiased Weibull moduli were also computed by multiplying by an unbiasing factor. Lower and upper confidence bounds were calculated by dividing the biased Weibull modulus by a q-value that was based on either a 5 or 95 percentile distribution obtained by Monte Carlo simulation. The unbiasing factors and q-values were selected from statistical tables in ASTM C1239-13 depending on the sample size.

3. Results

3.1 Particle size analysis

The particle size distribution for the ceramic powder is shown in Figure 2. The distribution is bimodal, with a majority of the particles being submicrons. The mean particle size is 0.36 µm. According to Zakeri et al.(2020), a good ceramic powder for VPP should have a mean particle size less than the layer thickness with the ideal range between 0.05 and 10 µm. Although D₅₀ is just over one-third of the layer thickness, over 10% of the particles are larger than 10 µm, which could lead to non-uniform dispersion and contribute to diminished feature resolution.

3.2 Test artifact analysis

Aside from some of the finest features failing to consistently print (discussed in subsequent sections), the artifacts showed no visual defects in the green state. After sintering, the artifacts experienced different degrees of shrinkage [Figure 3(a)]. Some of the test conditions caused warping. As an example, the part in Figure 3(b) was sintered at 1200°C and displayed warping in the base, pins and walls. Parts sintered at 1000°C did not retain sufficient mechanical strength for easy handling because the temperature was not high enough and/or long enough for densification. While some measurements could be taken on these artifacts, often, features would disintegrate into powder with minor handling. In the part sintered at 1000°C, shown in Figure 3(c), a delamination effect occurred where partial consolidation was observed, but the surface layers were cracked and easily separated from the base.

Figure 4 displays the mass loss and shrinkage of each artifact for both builds. All artifacts maintained a mass loss between 31.9% and 34.3%. The shrinkage of the artifact bases generally increased with temperature. For artifacts sintered at 1000°C, all but one dimension showed less than 5% shrinkage; in contrast, artifacts sintered at 1300°C exhibited shrinkage of ∼10%–15%. There were no distinguishable differences in shrinkage observed between the different artifacts and dimensions, although Z shrinkage was occasionally higher than that of X and Y. This is expected to be due to a lower concentration of ceramic particles between layers (Formlabs, 2018b) and/or natural settling effects.

In the initial analysis, the authors explored taking multiple measurements of each feature, such as what is shown in the steps in Figure 5(a). In preliminary measurements of multiple features, the replicate measurements were relatively consistent; consequently, the authors elected to take only one measurement per feature, as shown with the slots in Figure 5(b). If notable differences were observed in identical features between the two builds, additional measurements were taken for verification purposes. The overhang angles were often nonlinear; for these features, the measurements were taken relative to the upper region of the lower surface, as shown in Figure 5(c).

3.3 Straight features: walls and slots

Figure 6 shows how the measurements of the sintered walls deviate from the target dimensions. All deviations were within 0.14 mm of the target dimension, indicating that for this material, printer and processing conditions, thin walls do not need to be further scaled to account for shrinkage.

Many of the thinner walls warped or partially fell over; as a result, Figure 6 alone is not an accurate representation of the design considerations necessary for successful part manufacturing. It is worth noting that measurements of the walls with defects were included to provide an analysis on potential shrinkage, which is considered independent of distortion. Table 4 presents the number of each wall that survived sintering without defects. Increasing temperature and wall thickness improve the likelihood of a successfully printed and sintered wall.

The measurements of the slots from one representative build are shown in Figure 7. All slots were smaller than designed. In most cases, the slots expanded when the maximum temperature was 1000°C, while the slots shrank with higher maximum temperatures. All deviations were under 0.47 mm, but smaller slots in general were more accurate. The 0.30 mm slot did not consistently print and is thus excluded.

3.4 Round features: holes, pins and tubes

The measurements of the holes from one representative build are provided in Figure 8. Consistent with the behavior previously observed, for artifacts with the maximum temperature of 1000°C, the features were closer to target dimensions and experienced less shrinkage. Higher sintering temperatures resulted in deviations up to 0.7 mm and approximately 15% shrinkage. The smallest hole did not consistently print and/or survive sintering, so values were averaged across the eight larger holes.

The measurements of the pin diameters from one representative build are shown in Figure 9. Values were averaged from the four pins of the same diameter. In the green state, the features had high resolution, mostly underbuilt by no more than 0.1 mm. The shrinkage effects from higher sintering temperatures were more pronounced for the pins with a diameter of 2 mm, which were up to 0.3 mm smaller than desired, while the pins with a target diameter of 1 mm were up to 0.15 mm smaller than desired.

The measured green and sintered pin heights for both diameters are provided in Figure 10. There is only one build shown for maximum temperatures of 1000°C and two builds for the rest. The features were overbuilt by up to 2 mm, with the deviation increasing with target height. The 4 and 8 mm pins had no defects after sintering. For these pins, as sintering temperature increased, the sintered pin height tended to decrease with maximum shrinkage of ∼2 mm. Any shrinkage greater than this indicates the pin has likely warped. All 12 and 16 mm pins sintered at 1200°C and higher had some degree of warping. The 16 mm pins with the smaller diameter experienced the most warping. Although there was no warping for the 1000°C case, these pins were very fragile and broke with handling. The 16 mm pin with 1 mm diameter and a maximum temperature of 1000°C did not survive sintering.

Results from measuring the tubes, which were printed as standalone features in addition to being on the artifact, are shown in Figure 11. As can be seen in Figure 11(a), the standalone tube with a 6 mm inner diameter cracked when sintered at 1300°C. This was the only tube prone to cracking after sintering; no other tube features experienced defects. Figure 11(b) displays the deviations in height for the tubes. The tubes on the artifact were all overbuilt in the green state, while standalone tubes in the green state were all underbuilt. Once sintered, all tubes experienced reduced heights with a majority of the standalone artifacts being 0.6 to 0.9 mm below the target dimension. On the contrary, the tubes on the artifact usually did not deviate from the target dimension by more than 0.3 mm. The sintering temperature and the inner diameter of the tube did not have noticeable effects on influencing heights.

Figure 11(c) presents the deviations in outer diameters of the tubes. All tubes in the green state had outer diameters deviating no more than 0.2 mm from the target dimension; however, once sintered, deviations increased and increasing temperature generally increased the deviation. For almost all tubes sintered at or above 1100°C, the standalone tubes had smaller outer diameters than those on the artifact and deviations were up to 1.3 mm smaller than designed. The inner diameter of the tubes did not have a noticeable effect on influencing the outer diameters.

The deviations in inner diameters of the tubes are provided in Figure 11(d). All tubes in the green state exhibited underbuilt inner diameters, usually 0.2 to 0.4 mm below the target dimension. Once sintered, deviations generally increased. For sintering at 1200°C and higher, the deviations increased as the inner diameter increased. Only the 6 mm inner diameter condition had an increase in temperature lead to a general increase in deviation, similar to what is seen in Figure 11(c).

3.5 Other features measured in XZ

3.5.1 Steps

Figure 12 displays the deviations in step heights. Often, the smallest step (0.5 mm) was broken when the artifact was removed from the build plate. Steps were also commonly deformed or cracked because of the artifact warping. In general, with higher sintering temperatures, the larger steps exhibited larger deviations. Many of the steps were over 1 mm smaller than the target dimension, which is likely because of the curling of the base of the artifact.

3.5.2 Overhangs

As shown in Figure 13, the unsupported overhanging angles were at most 4° higher than the target dimensions in the green state. However, after sintering, all angles were more acute than their target values. The deviations increased as the target angle decreased, and deviations were at times over 15° for the target angle of 50°. Figure 13(a) provides representative data from one build, while Table 5 provides the feature survivability for each of the test conditions. Feature survivability increased with sintering temperature and increased angle. In Figure 13(b), there is an example artifact showing the 50° feature not surviving after sintering, while Figure 13(c) shows an example of a surviving 50° feature. Between these images, there is a notable difference in the warping of the base, which may indicate that Figure 13(b) used less optimized conditions, which were detrimental to the features.

3.5.3 Lateral features

The lateral features include three circles and three squares. Owing to anisotropic shrinkage, all features were measured in both X and Z. Furthermore, regardless of the target dimensions, the feature deviations were relatively consistent within a single artifact. As a result, the deviations from all six features were averaged for each build and temperature condition, and these results are shown in Figure 14. For the green state, all dimensions were underbuilt, and Z dimensions were less accurate than X dimensions. Across all sintering temperatures, X deviations were approximately 0.3 mm, and Z deviations were around 0.8 mm. This could be attributed to not using support material, where the unsupported tops of the features may have slumped. After sintering, shrinkage along both axes was observed, and there tended to be larger differences between the green and sintered states with higher sintering temperatures.

3.5.4 Qualitative features

The slot artifact includes a few features intended for qualitative assessment of surface conditions. The slope, cone and dome features at different sintering temperatures can be observed in Figure 15. The slopes exhibited a clear stair-stepping effect with low surface quality, mainly because of the layered nature of the VPP process. Although this did not disappear even after sintering at 1200°C and 1300°C, the ridge lines on the side profiles became less pronounced. Layers were visible on the cones and domes in the green state and up to 1100°C. Beyond this temperature, layers were not discernable, though there was a residual surface roughness. These images highlight the evolution of sintering behavior with temperature. Because 1000°C is not a high enough temperature to fully sinter the parts, the artifact surfaces resemble those of the green state. Once temperatures reach 1200°C for this material, the features start to densify and a visible change to the exterior surface is observed.

3.6 Mechanical properties

The results of three-point bend testing of the green specimens are provided in Figure 16(a). The flexural strength for the XY and XZ orientations was ∼12 MPa, and the company reports flexural strength for this material in the green state to be 10.27 MPa (Formlabs, 2018a). The ZX specimens had about one-third less strength, but all specimens were able to be post-processed and handled without breaking. The flexural strain at break for the green specimens was at most ∼3%. Figure 16(b) shows the flexural strength and strain of the sintered specimens. All samples were linear elastic until failure. The full results from statistical testing are provided in Supplemental Information. The 1271°C condition is excluded from statistical analysis to retain even temperature intervals for comparisons. As the sintering temperature increased, flexural strength increased. The flexural strengths for 1000°C, 1100°C, 1200°C and 1300°C were all statistically different from each other except for the comparison between 1000°C and 1100°C in XY (p = 0.13), where the 1100°C condition had higher flexural strength, but the statistical test was sensitive to lower sample size and large differences in variance. An increase in temperature above 1100°C tended to decrease the flexural strain, and the 1300°C sintering condition had a statistically lower strain than those of 1000°C and 1100°C.

Only the 1100°C and 1300°C sintering conditions were printed in multiple orientations. XZ orientation tended to yield the highest flexural strength (statistically significant for 1100°C), and the ZX orientation yielded the lowest strength (statistically significant for both). While ZX orientation tended to yield a higher flexural strain at break, there were no consistent statistical trends in strain for varying orientations.

Table 6 reports the Weibull modulus and sample size for each print orientation and sintering condition. Although all test conditions initially included 30 specimens, specimens with uncharacteristically low flexural results were excluded, since this pointed to a specimen defect. This left between 11 and 30 specimens per test condition. Unbiased Weibull moduli are also provided, which reduce the biased Weibull moduli by an unbiasing factor, which is dependent on sample size. Lower and upper bounds are calculated for a 90% confidence interval.

A higher Weibull modulus is desirable as this indicates a stronger fit and less data variability. Weibull modulus was highest for the green specimens and generally lower for the specimens with the higher sintering temperatures. There has been wide variation in Weibull moduli reported in the literature for silicon-based additively manufactured specimens. Weibull modulus has been reported as 10 for Silicon oxycarbide (O’Masta et al., 2020), 11.8 for silicon nitride (Qi et al., 2019), 3.21 for silicon carbide composites (Cramer et al., 2022) and ranging between 1.60 and 4.42 for silica glass (Toombs et al., 2022).

4. Discussion

In general, increasing sintering temperatures and/or increasing feature size decreases feature resolution while increasing feature survivability. Comparing the performance of similar positive and negative features, the walls were generally closer to designed dimensions than the slots. Likewise, the pins were generally closer to designed dimensions than the holes. However, the positive features were also prone to warping since they were less constrained.

The results from the artifact analysis can be referenced as design guidelines for future complex ceramic parts with geometries resembling similar features. For example, using Figure 6 and Table 4, a designer may conclude that when sintering at 1200°C or higher, it is best to use a wall thickness of at least 0.7 mm to increase the likelihood of no warping, and no part scaling is needed to be within 0.2 mm of the target thickness. Another example is using the data from Figure 10 to conclude that when sintering at 1200°C or higher, it is best to use pins that are at most 8 mm tall to increase the likelihood that they will only shrink in height by ∼2 mm and not warp.

This study also demonstrates how the location of features could influence the resolution and survivability. The steps were on the exterior of a flat rectangular base, and when the base warped, it affected their resolution. Even with careful use of a razor blade, the smallest step, 0.5 mm, often broke when removing the part from the build plate [such as what can be seen in Figure 11(a)]. The tubes showed different dimensions depending on whether they were constrained to the artifact base or printed separately. Finally, the lateral features suffered from a lack of support, which increased the deviations in the Z direction.

This work has highlighted a challenge in achieving part consistency for flexural testing. Occasionally, defects were observed after printing, such as a part with delaminating layers, and those specimens were discarded. However, for every test condition, at least 30 specimens with no perceptible defects were sintered, and almost every test condition had multiple specimens needing to be excluded. Most of the excluded specimens possessed a slight hairline crack after sintering, and upon flexural testing, the crack propagated to cause full delamination of a few layers of the largest planar surface. Every sintering condition used specimens from multiple print batches in case a particular batch had errors. The materials used were from the same lot, and experiments were conducted in a six-month span to reduce the chance of compositional changes. Despite all controls, the cracks arising after post-processing seemed sporadic with regard to the frequency of appearance and the location within the sintering furnace.

Defects of this nature are common in ceramic VPP (Zhang et al., 2024), and Xu et al.(2021), who observed cracks in their silica parts regardless of sintering temperature, speculate that one mechanism leading to cracking could be correlated to part geometry and orientation. They further speculate that horizontal cracking in sintered parts could be because of print errors if the adhesive forces in the VPP process exceed the cohesion between part layers, which would promote layer separation.

It is also possible that the defects arise in the debinding stage, where outgassing of the polymer during decomposition leads to stresses that cause delamination. However, this cannot be discerned for this study because debinding and sintering occurred in the same furnace. Although this study has shown that maximum sintering temperature influences feature resolution and flexural properties, an investigation of debinding and other sintering conditions will be a relevant extension of this work to enhance the understanding of crack formation and enable the fabrication of repeatable and reliable ceramic parts with the VPP process.

The data provided in this study are based on a specific material and set of processing conditions, which advances knowledge of the capabilities of ceramic VPP with silica-based material. There are opportunities to extend this work by comparing these results to similar tests using different ceramics or modified processing conditions. The artifact study was also limited to investigating feature resolution and survivability, but investigating dimensional accuracy could be a relevant extension, as aspects such as linear positioning accuracy and circularity could be of future interest, as well as print orientation of the test artifacts. Finally, imaging analysis of the internal porosity would be useful to better understand the evolution of densification.

5. Conclusions

The VPP AM process offers a versatile option for fabricating complex and functional ceramic parts, but the post-processing effects have not been well-studied. Test artifacts were designed and manufactured to evaluate aspects of resolution and survivability under different sintering profiles. Three-point bend testing was also conducted to assess mechanical properties. This study provides reference data, translating to design guidelines for a silica-based material; however, the method presented in this study for evaluating feature resolution and flexural strength can be applied to ceramics and other materials and used to achieve more predictable performance in future parts. This study contributes to qualification efforts for ceramics manufactured with the VPP AM process, enabling expanded capabilities of advanced ceramics in real-world applications.

There was a clear distinction between parts in a not-sintered state (e.g. maximum temperature of 1000°C) and parts in a sintered state (e.g. maximum temperature at or above 1200°C), where parts not fully sintered had less shrinkage and lower flexural strength. The deviations from target dimensions for the test artifacts often increased with higher sintering temperatures and larger features. In many cases, there was high feature resolution; for example, walls were within 0.2 mm of the target dimensions, and slots were mostly within 0.3 mm but never exceeding 0.5 mm. Artifact features with deviations over 1 mm were mainly because of warping effects or slumping from lack of supports. Larger overhanging features also tended to slump and had lower chance of survivability. In general, increasing feature size and temperature improved the chance of feature survivability, though certain finer features, such as pins, were prone to warping with increased temperature and height.

The flexural properties showed that increasing temperature generally yielded higher strength and lower strain at break, indicating that the higher sintering temperatures lead to mechanical properties that are more appropriate for functional applications. Specimens in the XZ orientation produced slightly higher flexural strengths than those in the XY orientation, although results were not statistically significant across all temperatures. However, specimens in the ZX orientation exhibited consistently lower flexural strength. The 90% confidence bounds for the Weibull moduli for the sintered samples ranged between 2.8 and 19.4, indicating that the current post-process conditions are not adequate for quality control since high data variability exists even when removing defects. Thus, future work will further investigate debinding and sintering parameters to better understand and prevent undesirable defects. Specifically, this work motivates a thorough investigation of the interaction between material characteristics (e.g. particle size, concentration), sintering parameters (e.g. time, temperature) and resultant properties (e.g. density, porosity, mechanical strength). A better understanding of the process-structure-property relationships with regards to sintering will enable future optimization of the parameters for different combinations of materials, geometries and process parameters.

Figures

Figure 1

Custom test artifacts to assess the effect of maximum sintering temperature on feature resolution and survivability

Figure 2

Particle size distribution (volume-weighted) of Formlabs Ceramic powder

Figure 3

Characteristics observed in the test artifacts

Figure 4

(a) Mass loss and (b) shrinkage of the bases of the test artifacts across two builds after being sintered at various temperatures

Figure 5

Sample measurements of the test artifacts

Figure 6

Deviations in wall thickness across two builds after being sintered at various temperatures

Figure 7

Deviations in slot thickness after being sintered at various temperatures

Figure 8

(a) Deviations in hole diameter and (b) shrinkage of the holes after being sintered at various temperatures

Figure 9

Deviations in pin diameter for each of the two diameters after being sintered at various temperatures

Figure 10

Pin heights for each of the two diameters after being sintered at various temperatures

Figure 11

Deviations in tube height, outer diameter, and inner diameter after being sintered at various temperatures

Figure 12

Deviations in step height across two builds after being sintered at various temperatures

Figure 13

Deviations in unsupported overhanging angles after being sintered at various temperatures

Figure 14

Deviations in X and Z dimensions of the lateral features across two builds after being sintered at various temperatures

Figure 15

Magnified images of positive slopes, cones and domes (a-b) in the green state; or with the maximum sintering temperature of (c-d) 1000°C, (e-f) 1100°C, (g-h) 1200°C and (i-j) 1300°C

Figure 16

Flexural strength and strain at break for (a) green parts and (b) parts sintered at various temperatures

Table 1

Debinding and sintering program for the ceramic material

Segment	Initial temperature (°C)	Final temperature (°C)	Rate (°C/min)	Hold time (h)
Ramp 1	25	360	1	–
Polymer debinding	360	360	–	6
Ramp 2	360	460	1	–
Polymer debinding	460	460	–	1
Ramp 3	460	Varies (1000–1300)	3	–
Sintering hold	Varies (1000–1300)	Varies (1000–1300)	–	1 / 12

Source:

Table by authors

Table 2

Characteristics of test artifacts that were explored in this study as well as the features implemented into the test artifact design corresponding to each characteristic

Characteristic investigated	Features to demonstrate characteristic
Straight features	Walls, slots, lateral squares
Round features	Tubes, pins, lateral circles, holes
Minimum feature (positive)	Walls
Minimum feature (negative)	Slots, holes
Features in multiple planes	Lateral circles and squares
XY resolution	Walls, slots, tubes, pins, holes
Z resolution	Overhangs, steps, pins, cone, dome, slopes
Unsupported overhangs	Overhangs

Source:

Table by authors

Table 3

Features that were quantitatively measured on the digital microscope and the corresponding light settings used

Feature	Measurement	Light setting
Walls	Thickness	Coaxial
Holes	Diameter	Partial
Pins	Diameter	Coaxial
Pins	Height	Partial
Overhangs	Angle	Partial
Lateral features (circles)	Length and height	Partial
Lateral features (squares)	Length and height	Partial
Slots	Thickness	Coaxial
Steps	Height	Partial
Tubes	Inner and outer diameter	Coaxial
Tubes	Height	Partial

Source:

Table by authors

Table 4

Number of thin wall features across two builds that survived sintering without defects

	Maximum sintering temperature (°C)
Wall thickness (mm)	1000	1100	1200	1271	1300
0.3	0	2	0	0	1
0.4	1	0	1	0	1
0.5	1	0	2	1	1
0.6	2	2	2	1	1
0.7	1	2	2	2	2
0.8	2	2	2	2	2

Source:

Table by authors

Table 5

Number of overhanging angle features across two builds that survived sintering without defects

	Maximum sintering temperature (°C)
Angle (°)	1000	1100	1200	1271	1300
50	0	1	1	2	2
60	1	1	1	2	2
70	1	1	2	2	2
80	1	2	2	2	2

Source:

Table by authors

Table 6

Weibull modulus of the flexural specimens for each sintering condition and print orientation

Maximum sintering temperature (°C)	Print orientation	Weibull modulus (biased)	Sample size	Weibull modulus (unbiased)	Lower bound	Upper bound
Green	XY	13.3	30	12.7	10.0	16.3
	ZX	35.8	30	34.2	26.9	43.7
	XZ	22.3	20	20.7	15.4	28.2
1000	XY	10.7	11	9.3	6.1	14.3
1100	XY	6.5	18	6.0	4.3	8.2
	ZX	15.6	25	14.8	11.3	19.4
	XZ	10.1	19	9.3	6.9	12.8
1200	XY	6.8	13	6.0	4.1	8.9
1271	XY	4.9	24	4.6	3.5	6.1
1300	XY	4.2	18	3.9	2.8	5.4
	ZX	7.7	30	7.3	5.7	9.4
	XZ	8.0	30	7.6	6.0	9.7

Notes:

The sample size of n = 30 was reduced to exclude parts with defects after sintering. Biased and unbiased Weibull modulus values are provided, as well as 90% confidence bounds

Source: Table by authors

Supplementary material

The supplementary material for this article can be found online.