Scan pattern, stress and mechanical strength of laser directly sintered ceramics

Transcription

1 Int J Adv Manuf Technol (213) 64: DOI 1.17/s ORIGINAL ARTICLE Scan pattern, stress and mechanical strength of laser directly sintered ceramics Xiaoyong Tian & Bo Sun & Jürgen G. Heinrich & Dichen Li Received: 3 November 21 / Accepted: 13 February 212 / Published online: 3 April 212 # Springer-Verlag London Limited 212 Abstract A finite element model was developed to simulate the influence of laser scan patterns (laser jump patterns, ratio of length to width and laser scan angle) on the temperature and stress distributions in the process of laser direct sintering ceramics. The effects of laser scan patterns on the stresses in the laser directly sintered ceramic samples were investigated by simulation. Different ceramic samples were prepared by the laser sintering process for the four-point bending strength measurement. The simulation results consist with the experimental ones. The results in the present research can be used to optimize process route for the laser direct sintering process and provide a guidance for selecting the laser processes parameters for high mechanical strength ceramic components. Keywords Scan pattern. Laser direct sintering. Ceramic. Residual stress. Mechanical strength 1 Introduction Solid freeform fabrication (SFF) is an automated manufacturing process that produces components with 3D complex structures layer by layer direct from CAD data without partspecific tooling and human intervention. When SFF came into being at the end of last century, the primary purpose X. Tian (*) : D. Li State Key Laboratory of Manufacturing System Engineering, Xi an Jiaotong University, Xi an 7149, China leoxyt@hotmail.com B. Sun : J. G. Heinrich Department for Engineering Ceramics, Clausthal University of Technology, Clausthal-Zellerfeld 38678, Germany was just to make prototypes and reduce the period of product development process. But now, functional components, specially using metal and ceramic as raw materials, are fabricated by some SFF processes. There are several fabrication processes to realize the net shaping of ceramic functional components, such as stereolithography of ceramic slurry [1, 2], selective laser sintering of ceramic powder [3], and extrusion deposition ceramic paste [4, 5], and other processes [6 8]. Direct laser sintering ceramics by using layer-wise slurry deposition (LSD) [9 11] is one of the selective laser sintering processes, which uses ceramic slurry as starting material instead of ceramic powder. Homogenous green ceramic tape with high density can be achieved by drying out the ceramic slurry layer deposited on a preheated ceramic tile. Consequently, laser sintering process is more activated than the ceramic powder substrate. The relationships between process parameters and properties of final ceramic parts have been already carefully investigated [12 14]. It has been proved that large hatching space and low laser energy density will increase the final bending strength of ceramic samples after being post-sintered in furnace. But lasersintered ceramic bodies with high final bending strength have looser microstructures before post-sintering in furnace than that of the samples with a low final bending strength. A stress relief mechanism has been put forward to interpret this phenomenon [15].Thereliefofstressinthelasersintered ceramic bodies during post-sintering process will cause delamination and finally reduce the bending strength. The larger residual stresses, the lower final bending strength will be. So, the mechanical properties of laser-sintered ceramic samples can be predicted by the inside residual stresses according to the proposed stress relief mechanism. During the fabrication process of ceramic components, the scan pattern of laser beam is a key factor for the residual

2 24 Int J Adv Manuf Technol (213) 64: stress distribution and distortion. It has been examined that deposition patterns had a significant influence on the resulting stresses and deflections in the laser-deposited metal parts [16]. Laser scan patterns also have effects on the distortion and the out-of-plane distortion of a layer can be minimized with a proper selection of the laser scan patterns. Many finite element thermo-mechanical models have been developed for SFF processes, such as selective laser sintering [17, 18], shape deposition manufacturing [19], laserengineered net shaping [2], and multi-material laser densification [21, 22]. These numerical researches have supplied substantial insights into how temperature gradients, thermal transient stress, residual stress and strain are developed during SFF process. But the influence of laser sintering patterns on the residual stress and final mechanical strength has never been systematically investigated. In the present research, a simplified numerical simulation model was established to investigate the influence of laser scan patterns on the residual stress distribution in the ceramic bodies. Laser jump pattern, ratio of length to width and laser scan angle were carefully studied by simulation and experimentally. Bending strength of the final ceramic samples was experimentally measured in order to find the connections between laser scan angles and final mechanical strength. 2 Numerical model descriptions The model developed for the simulation is shown in Fig. 1. The porcelain tape deposited on a ceramic tile by using LSD process is schematically shown in Fig. 1a. Grey section in the centre of porcelain tape is the laser sintering area with a size of 1 mm in length, 5 mm in width and 2 mm in height. One of the laser scan patterns is demonstrated in Fig. 1b. Incident heat flux of a Gaussian-distribution laser beam, ~f q, is related to the incident laser power, P, through the following equation [23, 24]: f q ¼ 2Pa pr 2 e 2R2 =R 2 b b ð1þ where α is the absorption of porcelain tape, R b is the radius of laser beam at which the heat flux value is e 2 times of that in the laser beam centre and it is.6 mm in present research, and R is the distance of a point on the surface of porcelain tape from laser beam centre. Before the deposition process started, the ceramic tile was preheated to a constant temperature of 2 C. For this case, the nodal temperature in the bottom face of the model was set to the preheating temperature and kept unvaried during the entire laser sintering process. The thermal properties of porcelain ceramic used in present simulation are summarized in Table 1 [24 26]. This model was executed using ANSYS commercial finite element package. The thermal element (solid 7) was used to simulate the temperature distribution. Each element near the surface has a size of.1 mm length,.1 mm width, and.4 mm height. Then stress fields were simulated by replacing the thermal element (solid 7) with an equivalent structural element (solid 45). During stress simulation analysis, the nodal thermal loads were provided by the previous thermal analysis. In the LSD process, ceramic slurry was deposited on a preheated tile and then dried out before laser sintering. Therefore, the distortion or displacement caused by thermal stress was almost completely constrained by the substrate. A full constraint condition had been applied in the bottom surface which was used for the simulation of stress distribution. Thermal convention and radiation were considered in all surrounding surfaces except for the bottom. Laser parameters used in present simulation were laser power of 4 W, scan speed of 15 mm/s and hatching space of.6 mm. Transient sintering temperature and stress distribution in the laser sintering process were attained and analysed. After laser sintering, the model was cooled down to the ambient temperature in order to evaluate the residual stress in the model. The present simulation allows for the assessment of the first laser sintering layer from the dried multilayer ceramic substrate. Sintering process of subsequent layers will be modelled in afuturework. Fig. 1 Scheme of finite element model (left) and laser scanning style (right) Laser beam a b

3 Int J Adv Manuf Technol (213) 64: Table 1 Thermal properties of porcelain Temperature (K) ,173 1,373 1,8 Thermal conductivity (W/m K) Specific heat (J/kg K) 742 1,25 1,125 1,178 1,266 1,341 1,444 1,474 Thermal expansion coefficient (1 6 /K) Elastic modulus (GPa) Emissivity.7 Density (kg/m 3 ) 2,52 Melting temperature (K) 1,573 Forming temperature (K) 973 to 1,273 Poisson's ratio.2 3 Results and discussion 3.1 Laser jump patterns There are two different kinds of substrates for the simulation, beam substrate (Fig. 2a, b) and plate substrate (Fig. 2c, d). The previous FEM model was used to investigate two different scan patterns for each substrate, raster patterns oneway (Fig. 2a) and loop (Fig. 2b) route for beam substrate, spiral patterns from inside to outside (Fig. 2c) and reverse (Fig. 2d) for the plate substrate. The included angle φ was introduced into this paper in order to explain its influence on the temperature fluctuation at the intersection of two connected scan lines, as shown in Fig. 3. Theincluded angles for the laser jumping patterns a and b were 18 and respectively and 9 for scan patterns c and d, as shown in Fig. 2. The transient temperature curves for the four kinds of scan patterns are shown in Fig. 4. Figure 4a indicates the temperature curves for the beam substrate raster-scanned by one-way and loop patterns. The difference in the average temperature for two patterns was not significant. Average a b o temperature of loop pattern (1,3.6 C) was around 1 C higher than that of one-way pattern (1,19.7 C). The temperature fluctuation at the both ends of each scan line in loop pattern may be caused by the laser re-sintering in a short period due to an included angle φ of, as shown in Fig. 2b. The temperature fluctuation induced inordinate sintering at both ends of the samples. Material sinking and vitrification were observed in the experiments, as shown in Fig. 5. The temperature curve for one-way pattern with an included angle of 18 was more stable than that of loop pattern even though there was a temperature decrease at each end of scan lines due to the laser jumping delay. Transient temperature curves for two spiral patterns are shown in Fig. 4b. For each scan pattern, the stable-state temperature in the area with short scan lines (1,28.6 C) was much higher than that in the area with long scan lines (769.5 C) due to the residual heat influence from the previous scanned area. There were temperature fluctuations at the intersection of two connected scan lines. All these scan lines had a same included angle of 9. The extent of residual thermal influence on the following transient laser sintering temperature depended on the included angles between two connected scan lines. Even though there were always some errors from modelling and simulation, it still can be concluded that small included angle will produce large temperature fluctuation. This conclusion is useful when planning the laser sintering route. Considering of the sequence in which higher temperature appeared, the spiral pattern scanned from outside to inside would produce a low i x Scan line 1 ϕ y c (i o) d (o i) Scan line 2 Fig. 2 Different laser jump patterns, raster patterns a one way and b loop for the beam substrate, spiral patterns scanned c from inside to outside and d reverse for the plate substrate Fig. 3 Scheme of the included angle between two connected scan lines

4 242 Int J Adv Manuf Technol (213) 64: a Transient sinteirng temperature ( o C) btranisent sintering temperature (oc) Oneway sintering Loop sintering From i to o From o to i Fig. 4 Temperature curves for the a beam and b plate substrates temperature gradient, and finally a low and uniform deflation which was consistent with the conclusions in a published literature [16]. The large temperature difference between two stable states in one scan pattern will cause inhomogeneous phase composition and microstructure. The properties of each part in one laser-sintered sample could be different, especially for the ceramic materials with complex phase compositions. So, closed-loop temperature control sintering strategy should be employed in the future experiments and simulation. Average transient stresses of one-way and loop scan patterns are shown in Fig. 6. Transient stresses of loop scan pattern were slightly higher than that of one-way pattern because the temperature fluctuation might produce stress concentration at the end of scan lines. The compressive stresses which were found close to the sintering zone were much higher than the tensile stresses. Z component tensions (laser incidence direction) were larger than X and Y components. X component stresses were slightly larger than Y direction (laser scanning direction), which could be explained by temperature gradient distribution and had been interpreted in a published paper [15]. The relative size of X and Y components were opposite to the results in a research paper [22] in which the stress in laser scanning direction was larger than the vertical direction. That might be caused by the different constraint conditions used in the simulation models and dual materials used in Dai and Shaw's work. Maximum transient stress curves for the plate substrates are plotted in Fig. 7. Compressions in all directions and tension of Z component changed with the sintering temperature. With higher sintering temperature, larger stresses were attained, as shown in Fig. 7b, d and dash dot lines in Fig. 7a, c. The changes of X and Y tensions were opposite to the temperature as well as the rest stresses. For both spiral scan patterns, X and Y tensions in the high-temperature area were smaller than those in the low temperature area. In another word, the X and Y tensions were in direct ratio to the length of scan lines. The increased substrate temperature in the area with short scan lines probably reduced the temperature gradient in x and y directions as well as the transient stress. Residual stresses for the four laser jump patterns are listed in Table 2. The quantities of residual stresses were small because there was just one starting material and the Transient stress (MPa) 25 5 Tension Compression Tension Compression Tension One way Loop Compression X Component Y Component Z Component Fig. 5 Material sinking at the end of the scan lines in the sample loop pattern sintered by laser beam Fig. 6 Average transient stress comparison between one-way and loop scan patterns for beam substrate

5 Int J Adv Manuf Technol (213) 64: a b 1 Stress (MPa) X Tension Y Tension Z Tension Stress (MPa) X Compression Y Compression Z Compression c 1 d 9 8 X Tension Y Tension Z Tension 3 25 Stress (MPa) Stress (MPa) X Compression Y Compression Z Compression Fig. 7 Maximum transient tension and compression for the plate substrates, a and b from inner to outer; c and d reverse bottom surface was bolted. When the model cooled down to the ambient temperature, most of the stress could be removed in the scope of elastic strain. The residual stresses in loop pattern for the beam substrate were larger than the rest due to the temperature fluctuation and stress concentration at the end of scan lines. Based on the stress relief mechanism, the higher residual stresses, the lower the final bending strength will be. So, the laser jump patterns basically had an insignificant influence on the final bending strength of ceramic samples. But different scan patterns indeed affected the surface accuracy and texture characteristic according to the experimental observation, as shown in Fig. 5. Table 2 Residual stress for different laser jump patterns Tension (MPa) Compression (MPa) X Y Z X Y Z One way Loop i o o i

6 244 Int J Adv Manuf Technol (213) 64: Ratios of length to width In the SFF processes, most of the cross-sections are complex and irregular. In the case of same laser sintering parameters, laser power, scan speed and hatching space, the sintering temperature and stresses will be different. To investigate this diversity, a simulation model had been designed to study the relationships between the geometric parameters and residual stresses as well as final mechanical strength. The geometric parameters used in present research were length and width. The ratio of length to width had been considered. The schematic is shown in Fig. 8. Models with different values of the ratios from.625 to 64 had been calculated under the condition that the product of length and width were kept to a constant quantity, 64 square units. The influence of length width ratio on the sintering temperature was very obvious, as shown in Fig. 9. With an increasing ratio, the temperature decreased approaching an inverse proportion function. The interval time between two scan lines for small ratio model was shorter than that of large ratio model. Long interval time reduced the heat influence from the previous laser-sintered area. Therefore, the larger length width ratio, the longer interval time and lower sintering temperature will be. Length width ratio also had a significant effect on the residual stresses, as shown in Fig. 1. The Z component tension was larger than the tensions in x and y directions. But the compression in z direction was much smaller than the components in x and y directions. Residual stress achieved a maximum value when length width ratio was equal to 1 and then decreased with an increasing ratio. This relationship could be considered as the results caused by the double effects of heat accumulation and dimensional effects. Therefore, when the cross-sections have a square shape (ratio of length to width1), there will be obvious residual stresses available in the laser-sintered body. Based on the stress relief mechanism, the final bending strength will be reduced by the stress relief in the post-sintering process. High residual stresses produce low final mechanical strength. 3.3 Scan angles The influence of laser scan angle on the final properties of ceramic samples had also been investigated in present research. Because the simulation of different laser scanning angle was very difficult to realize, the experimental method had been adopted to study the connections between the laser Temperature ( o C) Ratio (Length/Width) Temperature Fig. 9 Influence of length width ratios on the maximum sintering temperature scan angles and the final bending strength. The model with different scan angles (, 3, 6, 9 ) is schematically shown in Fig. 11 Residual stress (MPa) Residual stress (MPa) ,5 3, 2,5 2, 1,5 1, Ratio (Length/Width) X Compression YCompression Z Compression X Tension YTension ZTension,5 w l y x, Ratio (Length/Width) Fig. 8 Scheme for different ratios of length to width, l/w Fig. 1 Influence of length width ratios on the residual stresses

7 Int J Adv Manuf Technol (213) 64: θ Fig. 11 Scheme for different laser scan angles, θ, 3, 6, 9 Eleven samples for each scan angle were fabricated in the present research. Ceramic slurry provided by Imerys Tirschenreuth Germany, PM 95, was deposited via a doctor blade on a preheated (up to 2 C) tile by a dispenser pump (Dispenser 3NDP8, Netzsch, Germany). A worktable with high kinematic resolution, 6.25 μm per step for vertical motion and 12.5 μm per step for horizontal motion, was used to achieve these relative motions. A laser system with a Rofin Sinar SC1 1-W CO 2 laser tube (Hamburg, Germany) and a galvano scanner (Hurryscan, Scanlab AG, Germany) had been introduced into the experiments. The laser sintering parameters were laser power 5 W, scan speed 1 mm/s and hatch space.6 mm. The post-treatment process of laser-sintered bodies which had been already investigated by Tian was following to densify the microstructures and improve the mechanical strength of ceramic samples at 1,425 C for 12 min [12]. Bending strength of 11 specimens for each experiment group was determined using a four-point bending method. The specimens for bending strength test were fabricated into a standard size ( mm) without surface polishing. The fractured samples after bending strength measurement are shown in Fig. 12. The texture structures on the laser-sintered surface are obvious and have an influence on the mechanical strength of the final ceramic samples. The shrinkages during post-sintering process had been measured using a digimatic calliper. The shrinkages of samples during the post-sintering process are shown in Fig. 13. Samples with a laser scan angle of 9 had a smaller shrinkage in x and z directions than those with a angle. Bending strength of samples with 9 angle was also lower than that of sample with angle, as shown in Fig. 14. Samples with 9 angle were scanned by laser beam vertical to the length direction. The scan lines were short and sintering temperature was high by using the same laser sintering parameters, as discussed in the previous section. The microstructures of laser-sintered bodies with a higher sintering temperature were denser than those with a lower sintering temperature and the residual stresses of the former were larger than the Shrinkage (%) latter. According to the stress relief mechanism, higher residual stresses produce lower final mechanical strength after being post-sintered in a furnace, which is consistent with the experimental results. In another viewpoint, the scan lines could be considered as an enhancement phase with homogenous and continuous microstructures, just like the fibre phase in the fibre-reinforced matrix composites. The bending strength along the direction of continuous phase was always higher than that vertical to it due to the reinforcement from continuous phase. 4 Conclusions Angle (deg.) Y shrinkage X shrinkage Z shrinkage Voulume shrinkage Fig. 13 Influence of scan angles on the shrinkages during the postsintering process A finite element model was developed to simulate the process of laser direct sintering ceramics. Laser jump patterns, Bending strength (MPa) 35 3 Bending strength Polynomial Fit of Bending strength (uv) Angle (deg.) Fig. 12 Fractured samples with different laser scan angles Fig. 14 Influence of scan angles on the bending strength of postsintered ceramic samples

8 246 Int J Adv Manuf Technol (213) 64: different ratios of length to width in the model and laser scan angles had been carefully studied. Different laser jump patterns have an obvious effect on the laser sintering temperature. A temperature difference of 26 C has been attained in the spiral pattern. There is no big difference of residual stress between the different laser scan patterns. Ratio of length to width has a dramatic effect on the sintering temperature under the conditions of the same laser sintering parameters and laser scanned area. Short scan lines will produce high sintering temperature. Residual stresses achieve maximum values when the ratio is equal to 1. Bending strength of samples with long laser scan lines with a scan angle of is higher than those with a scan angle of 9 due to the sintering temperature and stresses as well as the possible reinforcement of the continuous laser-sintered phase. References 1. Hinczewski C, Corbel S, Chartier TJ (1998) Ceramic suspensions suitable for stereolithography. J Eur Ceram Soc 18: Zhou WZ, Li DC, Chen ZW (211) Influence of ingredients and laser exposure on curing behaviors of aqueous ceramic suspensions in stereolithography. Int J Adv Manuf Technol 52: Bertrand PH, Bayle F, Combe C, Goeuriot P, Smurov I (27) Ceramic components manufacturing by selective laser sintering. Appl Surf Sci 254: Mason MS, Huang T, Landers RG, Leu MC, Hilmas GE (29) Aqueous-based extrusion of high solids loading ceramic pastes: process modeling and control. J Mater Proc Technol 29: Wang J, Shaw L (25) Rheological and extrusion behavior of dental porcelain slurries for rapid prototyping applications. Mater Sci Eng A 397: Tian XY, Li DC (21) Rapid manufacture of net-shape SiC components. Int J Adv Manuf Technol 46: Friedel T, Travitzky N, Niebling F, Scheffler M, Greil P (25) Fabrication of polymer derived ceramic parts by selective laser curing. J Eur Ceram Soc 25: Wu HH, Li DC, Chen XJ, Sun B, Xu DY (21) Rapid casting of turbine blades with abnormal film cooling holes using integral ceramic casting molds. Int J Adv Manuf Technol 5: Günster J, Engler S, Heinrich JG (23) Forming of complex shaped ceramic products via layer-wise slurry deposition (LSD). Bull EcerS 1: Gahler A, Heinrich JG, Günster J (26) Direct laser sintering of Al2O3 SiO2 dental ceramic components by layer-wise slurry deposition. J Am Ceram Soc 89: Yen HC, Tang HH (28) Developing a paving system for fabricating ultra-thin layers in ceramic laser rapid prototyping. Int J Adv Manuf Technol 36: Tian XY, Günster J, Melcher J, Li DC, Heinrich JG (29) Process parameters analysis of direct laser sintering and post treatment of porcelain components using Taguchi's method. J Eur Ceram Soc 29: Yen HC, Tang HH (212) Study on direct fabrication of ceramic shell mold with slurry-based ceramic laser fusion and ceramic laser sintering. Int J Adv Manuf Technol. doi:1.17/s Dingal S, Pradhan TR, Sarin Sundar JK, Roy Choudhury A, Roy SK (28) The application of Taguchi's method in the experimental investigation of the laser sintering process. Int J Adv Manuf Technol 38: Tian XY, Sun B, Heinrich GJ, Li DC (21) Stress relief mechanism in layer-wise laser directly sintered porcelain ceramics. Mater Sci Eng 527(7): doi:1.116/j.msea Nickel AH, Barnett DM, Prinz FB (21) Thermal stresses and deposition patterns in layered manufacturing. Mater Sci Eng A 317: Bugeda G, Cervera M, Lombera G (1999) Numerical predication of temperature and density distributions in selective laser sintering processes. Rapid Prototyp J 5: Williams J, Deckard CR (1998) Advances in modeling the effects of selected parameters on the SLS process. Rapid Prototyp J 4: Amon CH, Beuth JL, Weiss LE, Merz R, Prinz FB (1998) Shape deposition manufacturing with microcasting: processing, thermal and mechanical issues. J Manuf Sci Eng 12: Wang L, Felicelli SD, Craig JE (29) Experimental and numerical study of the LENS rapid fabrication process. J Manuf Sci Eng 131: Dai K, Shaw L (26) Parametric studies of multi-material laser densification. Mater Sci Eng, A 43: Dai K, Shaw L (23) Finite-element analysis of effects of the laser-processed biomaterial component size on stresses and distortion. Metall Mater Trans A 34A: Dai K, Crocker JE, Shaw LL, Marcus HL (23) Thermal modelling of selective area laser deposition (SALD) and SALD vapor infiltration of silicon carbide. Rapid prototyp J 9: Dai K, Shaw L (25) Finite element analysis of the effect of volume shrinkage during laser densification. Acta Mater 53: Schneider SJ (1991) Engineered materials handbook. Vol. 4. Ceramics and glasses. ASM International, Materials Park, p Touloukia YS, Ho CY (197) Thermaophysical properties of matter. Vol. 12. Thermal expansion: nonmetallic solids. IFI/Plenum, New York, p 1369