Preprint for: Thermal binder removal from large ceramic parts 3D printed by the digital light processing method

1 Thermal binder removal from large ceramic parts 3D printed by the digital light processing method Premysl Stastny a, Ondrej Man a, Dominik Brouczek b, Martin Schwentenwein b, Martin Trunec a, c, * aCEITEC BUT, Brno University of Technology, Purkynova 123, Brno 612 00, Czech Republic bLithoz GmbH, Mollardgasse 85a/2/64-69, Vienna 1060, Austria cInstitute of Materials Science and Engineering, Brno University of Technology, Technicka 2, Brno 616 69, Czech Republic Corresponding author: Central European Institute of Technology, Brno University of Technology, Purkynova 123, 612 00 Brno, Czech Republic e-mail: [email protected] Abstract This paper investigates the size limit for defect-free binder removal from 3D printed bodies produced using a commercial alumina suspension by the digital light processing method. Binder removal from cylindrical bodies with a diameter of 5 mm, 9 mm, 11mm, and 15 mm was carried out in two different debinding atmospheres (nitrogen and air) using various heating schedules. In the nitrogen atmosphere, the binder could be removed without defects from bodies with a diameter of 15 mm. In contrast, bodies with a diameter of only 5 mm were prepared defect-free in the air atmosphere. Thermoanalytical methods and microstructural analyses were applied to explain the observed differences in binder removal and defect formation. The debinding mechanisms were proposed, and the critical steps for defect-free processing were identified and discussed. Keywords 3D printing, Binder removal, Microstructure, Size limit, Cracking 1. Introduction Additive manufacturing (AM) of ceramics, also known as ceramic three-dimensional (3D) printing, is experiencing a huge interest in both academic and industrial communities because of promising results in the production of high-performance ceramic components with geometrically complex architectures, which are difficult to process using classical processing 2 methods [1-3]. Among the various AM technologies, vat polymerization methods, e.g., stereolithography (SLA) or digital light processing (DLP), are considered to be the most promising technologies due to their excellent feature resolution and quality of the final parts [4]. Regardless of the vat-polymerization method, the photopolymerization of photosensitive resins with dispersed ceramic particles is the essential processing step [5]. After illuminating the photosensitive ceramic suspension with a light source of appropriate wavelength, a complex polymeric network is formed, and liquid ceramic suspension is transformed into a solid layer. As the layers are stacked on top of each other, the final 3D part is created. The 3D-printed ceramic part is then composed of ceramic particles anchored in a polymeric network [6]. Before the high-temperature densification of the particle compact, the polymeric network and all other organic additives must be removed in a separate binder removal (debinding) step. The removal of the polymeric network is often the most critical step of the process chain. Although many debinding methods have been developed in ceramics processing, thermal binder removal [7, 8] is almost solely used for photopolymerized 3D printed parts. This step usually takes more time than printing itself and often results in defect formation [9]. Once the defects are created in the body during the debinding process, their healing during sintering is not probable, and the defects are usually exaggerated during sintering. The debinding problems are especially pronounced in the case of large solid bodies [10, 11]. The mechanisms of thermal debinding include thermal degradation of the polymeric network, diffusion of degradation products and other low-molecular-weight binder additives to the body surface (or binder/atmosphere interface), and their evaporation. In the air atmosphere, an additional degradation mechanism, namely oxidative degradation, is involved. The sudden gas evolution during boiling of low molecular weight components and degradation products in the pores filled with a liquid/solid binder is considered to be the main cause of defect formation during thermal debinding [12]. The easiest way to prevent defect formation is to slow the heating rate, especially in the temperature regions with the highest mass loss rate. Thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) of the printed material are commonly used to adjust the heating profile of the binder removal [12]. Recent reports also showed the importance of a debinding atmosphere. Many authors reported stronger internal stresses and more serious defect formation during debinding in an air atmosphere than in non-oxidizing atmospheres of Ar [9], 95 % Ar + 5 % H2 [13], or N2 [14]. Zhang et al. [15] compared the effect of different aerobic and anaerobic atmospheres and vacuum on defect formation during debinding. The vacuum minimized the number of defects in the bodies after debinding and provided a higher sintered density. A higher sintered density after debinding in the vacuum compared with debinding in the air was also confirmed in another study by Wu et al. [16]. Zhou et al. [17] proposed a two-step process with vacuum debinding followed by air debinding. The vacuum prevented abrupt gas formation, and 3 subsequent air debinding removed all remaining carbon residues. Unfortunately, most of these articles did not satisfactorily explain the effect of debinding atmospheres. The discussion about the underlying mechanisms of debinding in different atmospheres was either missing, or the explanation was highly hypothetical without experimental proof [15, 18]. Another approach to preventing high peaks of weight loss rate during debinding is to add a nonreactive diluent to the binder system. This compound is not involved in the photopolymerization reaction and, due to its lower molecular weight, can be removed prior to the main decomposition of the polymeric network. The removal of a non-reactive diluent can facilitate the formation of open pores and interconnected channels, thus providing shorter pathways for the subsequent discharge of degradation products from the decomposed polymer network. This strategy resembles the debinding of a multicomponent binder in ceramic injection molding [19-21]. Moreover, the non-reactive diluent can also act as a plasticizer, controlling the rheological properties of the ceramic suspension, reducing photopolymerization shrinkage, and weakening the internal stresses by decreasing the crosslinked density of the polymeric network. The successfully applied non-reactive diluents include dibutyl phthalate [22], polyethylene glycols (PEG 200 [23] and PEG 400 [24]), or polypropylene glycol (PPG 400)[18]. Non-reactive diluents with even lower molecular weight (butoxyethyl acetate [25] and diethylene glycol [26]) were used when the removal temperature of PEG 400 coincided with the removal temperature of the main binder. This investigation aimed to describe the effect of a heating profile and debinding atmosphere on the size limit for defect-free binder removal from bodies printed by the DLP method. In this work, we tried to reveal and explain the main differences in debinding mechanisms in air and inert atmospheres and to correlate them with the maximum available size of the defectfree bodies after debinding. 2. Experimental 2.1. 3D printing of test bodies The alumina bodies were prepared using a DLP printer (CeraFab S65, Lithoz GmbH, Austria) and a commercially available alumina-based photocurable suspension (Lithalox 350, Lithoz GmbH, Austria). The test bodies were printed in the shape of a cylinder with a constant height of 15 mm and a diameter of 5 mm, 9 mm, 11 mm, and 15 mm. The printing was performed with an exposure energy of 150 mJ cm-2, a layer thickness of 25 µm, and a light intensity of 50 mW cm-2. The system provides a lateral pixel resolution of 40 × 40 µm², enabling the production of ceramic bodies with high dimensional accuracy and surface quality. All printing parameters were selected based on the material manufacturer’s recommendations to ensure optimal curing and printing behavior. 4 2.2. Thermal analysis of printed bodies Thermogravimetric analysis (TGA) and differential thermal analysis (DTA) of the printed bodies were performed using a TGA/DTA analyzer (96 Line TGA - DTA/DSC, Setaram, France) at a heating rate of 1 °C min-1 up to a temperature of 800 °C in nitrogen and air (80% N2+20% O2) atmospheres with a gas flow of 50 ml min-1. Tests were performed with three sizes of samples that were cut from the largest printed cylinder: crushed sample (crushed in a mortar), small sample (body volume of ~45 mm3), and large sample (body volume of ~90 mm3). 2.3. Debinding and sintering of cylindrical test bodies The samples for the binder removal were used as printed without any additional pretreatment before debinding. Debinding experiments were performed in both air and nitrogen atmospheres. During air debinding, the ceramic bodies were placed on a ceramic plate. The air debinding was performed in a static atmosphere. During the nitrogen debinding, the bodies were embedded in granular activated carbon (AY-5 12x30, Carbon Link, UK). The nitrogen flow of 20 L h-1 was maintained during the binder removal in the muffle furnace. Two debinding schedules with different heating rates (long and short) were applied, as shown in Fig. 1(a). The long schedule in the air atmosphere was recommended by the producer of the ceramic suspension for large bodies. Both debinding schedules were tested under the air and nitrogen atmospheres, i.e., four variants of debinding experiments were conducted. Ten samples of each size were used for each debinding experiment. All samples, after debinding, were sintered in air at 1650°C using the schedule given in Fig. 1(b). In addition, the debinding of cylinders with a diameter of 15 mm using the short debinding schedule in nitrogen or air atmospheres was interrupted at various temperatures (150 °C, 200 °C, 250 °C, 300 °C, and 350 °C) to investigate their internal structure and monitor the progress of binder removal. Fig. 1 Heating schedule for (a) debinding and (b) sintering of printed samples. 5 2.4. Characterization methods The macro defects and cracks in the printed samples were checked using an optical microscope before and after debinding. After sintering, the samples were inspected optically again, and the density of the sintered samples was measured in water using the Archimedes method. The relative densities were calculated using a theoretical density (t.d.) of 3.987 g cm−3 for alumina. The macrostructure of the samples after partial debinding was analyzed using a stereomicroscope (Stemi 508, Zeiss, Germany), and the microstructure of the green bodies and bodies after partial debinding was analyzed using a scanning electron microscope (SEM) (Verios 460L, FEI, Czech Republic). For the SEM investigation of as-printed samples and samples after partial debinding, the cross section of the samples was ion beam polished under cryogenic conditions (-120 °C, 6kV, 3 mA) using an ion beam milling system (EM TIC 3X, Leica, Germany). The sintered samples were mechanically polished for the SEM analysis. All samples for electron microscopy were carbon-coated using 15 nm of carbon (AM ACE 600, Leica, Germany) 3. Results and discussion 3.1. Thermal analysis Fig. 2(a) shows mass losses during the heating of the printed bodies in nitrogen and air atmospheres. The oxygen in the air atmosphere caused the shift of mass loss curves to lower temperatures compared with nitrogen. A pronounced effect of sample size was observed during debinding in the air. The mass loss curves of bigger samples were moved to higher temperatures, closer to the mass loss curves of samples treated in nitrogen. On the other hand, only negligible differences between samples of different sizes were found during debinding in nitrogen. After debinding in the nitrogen atmosphere, carbon residues from the organic binder and other additives (about 1% of the total mass) remained in the body structure. Fig. 2(b) gives the thermal effect of debinding in different atmospheres. The thermal degradation occurring in nitrogen resulted in almost no thermal effects. On the other hand, the oxidative Fig. 2 Thermal analysis of the printed samples showing (a) thermogravimetric curves and (b) heat flow curves. 6 degradation occurring in air exhibited three exothermic peaks. These peaks correspond to the steps with steep mass losses (see Fig. 3). With the increasing sample size, the mass loss below 200 °C rapidly decreased, and the first low-temperature exothermic peak decreased. The other two exothermic peaks increased as the mass loss was moved to higher temperatures in the large sample. 3.2. Debinding defects A set of cylindrical bodies with different diameters after debinding in nitrogen are shown in Fig. 4. Tables 1 and 2 show the percentage of defect-free bodies after debinding in nitrogen and air, respectively. It is obvious that debinding in the nitrogen atmosphere was more successful than in the air atmosphere. Debinding in nitrogen using the long schedule provided all bodies defect-free. Debinding in nitrogen using the short schedule also provided defectFig. 3 Comparison of the temperature position of peaks on heat flow curves and derivative thermogravimetric curves. Fig. 4 Set of printed cylinders after debinding in the nitrogen atmosphere. 7 free bodies, except for the largest cylinders with a diameter of 15 mm. On the other hand, debinding in the air was completely successful only when using the long schedule for cylinders with a diameter of 5 mm. In all other cases, cracks appeared in several or often in all samples. The typical debinding defect is shown in Fig. 5. This was a partial or full crack through the cylindrical sample along the printing layers. Table 1 and Table 2 also show the Table 1 Relative sintered density and percentage of defect-free sintered bodies after debinding in nitrogen N2 Long debinding schedule Short debinding schedule Cylinder diameter Relative density SD* Defect-free Relative density SD* Defect-free (mm) (% t.d.) (% t.d.) (%) (% t.d.) (% t.d.) (%) 5 97.58 0.13 100 97.28 0.18 100 9 95.98 0.20 100 96.06 0.18 100 11 95.55 0.10 100 95.51 0.18 100 15 94.42 0.22 100 94.67 0.25 80 *SD = sample standard deviation of relative density Table 2 Relative sintered density and percentage of defect-free sintered bodies after debinding in air Air Long debinding schedule Short debinding schedule Cylinder diameter Relative density SD* Defect-free Relative density SD* Defect-free (mm) (% t.d.) (% t.d.) (%) (% t.d.) (% t.d.) (%) 5 97.31 0.24 100 97.40 0.31 80 9 96.33 0.19 70 96.66 0.23 0 11 96.03 0.34 20 96.26 0.24 0 15 95.50 0.40 0 95.55 0.32 0 *SD = sample standard deviation of relative density Fig. 5 Cross-section of a sintered cylinder with a typical crack after debinding in air. 8 relative densities of sintered ceramic bodies. The sintered densities were similar for bodies processed in nitrogen using long and short debinding schedules and decreased with the increasing cylinder diameter from 97.6 % t.d. to 94.4 % t.d. The densities of bodies processed in air were similar to those processed in nitrogen. The slightly higher densities achieved in larger cylinders processed in air compared to nitrogen could result from cracking. The fully cracked parts behaved as individual smaller bodies during the remaining debinding and sintering and reached a density corresponding to these smaller bodies. 3.3. Structure characterization Fig. 6 shows photographs of longitudinal and transverse sections of cylindrical bodies with a diameter and height of 15 mm after debinding to different temperatures in nitrogen and air. This figure shows a clear difference between bodies treated in nitrogen (Fig. 6(a)) and air (Fig. 6(b)). In the nitrogen atmosphere, the bodies turned slightly grey homogeneously throughout the entire volume during debinding. After debinding in the air atmosphere, we could observe a layer of brownish-grey products of oxidative degradation, which grew with the debinding temperature from the surface to the center of the bodies. Moreover, the photos of transverse sections of the cylindrical bodies show that the binder degradation and removal were microscopically inhomogeneous, i.e., a debinding inhomogeneity was observed in the volume of a printed layer. Because the transversal cut was not perfectly perpendicular to the cylinder axis (i.e., not perfectly parallel with the 25 µm-thick printed layers), color strips Fig. 6 Longitudinal and transverse sections of cylindrical bodies with a diameter and height of 15 mm after debinding to different temperatures in (a) nitrogen and (b) air using the short schedule. 9 demonstrate cross sections of more or less degraded parts of the printed layers. This phenomenon could be clearly observed during air debinding due to the brownish-grey degradation products, but it was also observed in bodies treated in nitrogen. The graph in Fig. 7 shows binder losses of the investigated cylindrical bodies (with 15 mm in diameter) in different atmospheres. The air atmosphere sped up the binder removal compared to nitrogen. To compare structures with the same binder loss in air and nitrogen, we must compare bodies heated to different temperatures, e.g., the binder loss of ~40% was reached at a temperature of 250°C in air and 300°C in nitrogen. To better understand the debinding mechanisms, we investigated the microstructure of bodies after partial debinding. Figs. 8 and 9 show the microstructure of green printed bodies, which can serve as a standard for comparison with bodies after partial binder removal. Fig. 8 shows a cross-section of the green body with printed layers of 25 µm in thickness. The interface between layers can be clearly observed, which means that these interfaces create a certain structural inhomogeneity in the body structure. Fig. 9 shows a detailed view of the microstructure. The white arrows mark the interface between the printed layers. The printed layers consisted of dispersed ceramic particles and a binder that fully filled the pores between the particles. The interface between the layers showed discontinuity in regular particle packing but without any voids or air bubbles. Fig. 10 shows microstructures of a body after debinding to 250 °C in the air atmosphere (with a binder loss of ~40 %). A porous structure with a porosity gradient in the printed layer was found near the body surface (see Fig. 10(a)). The difference in the porosity was visible at the printed layer interface. A similar structure, but with a smaller overall porosity, could also Fig. 7 Binder mass loss during debinding of cylindrical bodies with a diameter and height of 15 mm after debinding to different temperatures in nitrogen and air. 16 [5] S.A. Rasaki, D.Y. Xiong, S.F. Xiong, F. Su, M. Idrees, Z.W. Chen, Photopolymerizationbased additive manufacturing of ceramics: A systematic review, J Adv Ceram, 10 (2021) 442471. [6] S. Zakeri, M. Vippola, E. Levänen, A comprehensive review of the photopolymerization of ceramic resins used in stereolithography, Addit Manuf, 35 (2020) 101177. [7] M. Trunec, J. Cihlar, Removal of thermoplastic binders from ceramic green bodies, Ceram-Silikaty, 41 (1997) 67-80. [8] J.A. Lewis, Binder removal from ceramics, Annu Rev Mater Sci, 27 (1997) 147-173. [9] J.X. Sun, J. Binner, J.M. Bai, 3D printing of zirconia via digital light processing: optimization of slurry and debinding process, J Eur Ceram Soc, 40 (2020) 5837-5844. [10] A.K. Hofer, J. Rabitsch, D. Jutrzenka-Trzebiatowska, C. Hofstetter, I. Gavalda-Velasco, J. Schlacher, M. Schwentenwein, R. Bermejo, Effect of binder system on the thermophysical properties of 3D-printed zirconia ceramics, Int J Appl Ceram Tec, 19 (2022) 174-180. [11] E. Schwarzer-Fischer, J. Abel, J. Sieder-Katzmann, M. Propst, C. Bach, U. Scheithauer, A. Michaelis, Study on CerAMfacturing of Novel Alumina Aerospike Nozzles by Lithography-Based Ceramic Vat Photopolymerization (CerAM VPP), Materials, 15 (2022) 3279. [12] S.X. Zhou, G.Z. Liu, C.S. Wang, Y. Zhang, C.Z. Yan, Y.S. Shi, Thermal debinding for stereolithography additive manufacturing of advanced ceramic parts: A comprehensive review, Mater Design, 238 (2024) 112632. [13] Y.R. Shen, Y. Sun, B.C. Jin, M. Li, B.H. Xing, Z. Zhao, Effect of debinding and sintering profile on the optical properties of DLP-3D printed YAG transparent ceramic, Ceram Int, 48 (2022) 21134-21140. [14] P.L.A. Alves, I.L. Camargo, J.R. Verza, A.P. Luz, Impact of debinding atmosphere and sintering additives on the fabrication of alumina ceramics via vat photopolymerization, Ceram Int, 51 (2025) 9063-9072. [15] K. Wang, M.B. Qiu, C. Jiao, J.J. Gu, D.Q. Xie, C.J. Wang, X.B. Tang, Z. Wei, L.D. Shen, Study on defect-free endebinding green body of ceramic formed by DLP technology, Ceram Int, 46 (2020) 2438-2446. [16] H.D. Wu, Y.L. Cheng, W. Liu, R.X. He, M.P. Zhou, S.H. Wu, X. Song, Y. Chen, Effect of the particle size and the debinding process on the density of alumina ceramics fabricated by 3D printing based on stereolithography, Ceram Int, 42 (2016) 17290-17294. [17] M.P. Zhou, W. Liu, H.D. Wu, X. Song, Y. Chen, L.X. Cheng, F.P. He, S.X. Chen, S.H. Wu, Preparation of a defect-free alumina cutting tool via additive manufacturing based on stereolithography - Optimization of the drying and debinding processes, Ceram Int, 42 (2016) 11598-11602. [18] J. Kim, C.W. Gal, Y.J. Choi, H. Park, S.Y. Yoon, H.S. Yun, Effect of non-reactive diluent on defect-free debinding process of 3D printed ceramics, Addit Manuf, 67 (2023) 103475. 17 [19] M. Trunec, J. Cihlar, Thermal debinding of injection moulded ceramics, J Eur Ceram Soc, 17 (1997) 203-209. [20] M. Trunec, J. Cihlar, Thermal removal of multicomponent binder from ceramic injection mouldings, J Eur Ceram Soc, 22 (2002) 2231-2241. [21] M. Trunec, J. Cihlar, S. Diethelm, J. Van Herle, Tubular La0.7Ca0.3Fe0.85Co0.15O3delta perovskite membranes, part I: Preparation and properties, J Am Ceram Soc, 89 (2006) 949-954. [22] S.X. Zhou, G.Z. Liu, A.N. Chen, J. Su, K. Liu, C.S. Wang, Y. Zhang, C.Z. Yan, Y.S. Shi, Defect inhibition mechanism of 3D-printed ceramics via synergetic resin composition and debinding processing regulation, J Am Ceram Soc, 108 (2025) e20168. [23] Z.Q. Han, S.H. Liu, K. Qiu, J. Liu, R.F. Zou, Y.Y. Wang, J. Zhao, F.T. Liu, Y. Wang, L. Li, The enhanced ZrO2 produced by DLP via a reliable plasticizer and its dental application, J Mech Behav Biomed, 141 (2023) 105751. [24] E.G. McAleer, M.K. Alazzawi, C. Hwang, J.M. LaManna, D.L. Jacobson, B. Khaykovich, R.A. Haber, E.K. Akdogan, Binder removal from ceramic stereolithography green bodies: A neutron imaging and thermal analysis study, J Am Ceram Soc, 106 (2023) 4399-4410. [25] E. Johansson, O. Lidström, J. Johansson, O. Lyckfeldt, E. Adolfsson, Influence of Resin Composition on the Defect Formation in Alumina Manufactured by Stereolithography, Materials, 10 (2017) 138. [26] D.W. Zhang, E. Peng, R. Borayek, J. Ding, Controllable Ceramic Green-Body Configuration for Complex Ceramic Architectures with Fine Features, Adv Funct Mater, 29 (2019) 1807082. [27] J.K. Wright, J.R.G. Evans, Kinetics of the Oxidative-Degradation of Ceramic InjectionMolding Vehicle, J Mater Sci, 26 (1991) 4897-4904. [28] J.W. Halloran, Ceramic Stereolithography: Additive Manufacturing for Ceramics by Photopolymerization, Annual Review of Materials Research, 46 (2016) 19-40.