A. Selective laser treatments of alumina without absorption additive

A broad range of laser parameters was considered for sintering/melting of pellets top surfaces with the Nd:YVO₄ laser. Unfortunately, the irradiated surfaces did not reach a temperature high enough to allow material diffusion, grains bonding, and subsequent sintering or melting. As reported in the literature, the energy conversion of laser light into thermal energy is a function of the parameters reminded in Sec. II. A. However at a power of 20 W and considering the very poor coupling of alumina and laser at 1064 nm (alumina absorbs less than 10% of laser energy for wave length around 1000 nm as shown in Fig. 4 (Refs. Reference Hellrung, Yeh, Depiereux, Gillner and Poprawe12,Reference Pham, Dimov and Petkov13)), the energy density was clearly not enough to reach conditions of sintering/melting. Furthermore, the laser/matter interaction is not only a function of the laser wave length but also of the grain size and porosity of the pellets: although some fraction of the incident laser beam is reflected, a nonnegligible part of the remaining energy is absorbed by the material through multiple reflections on a certain depth of the pellets. ^{Reference Shao, Lin and Zhou14} Usually, this effect improves the effective amount of absorbed energy. ^{Reference Pham, Dimov and Petkov13–Reference Deckers, Shazad, Vleuels and Kruth15} Unfortunately, the low absorbance of alumina in the near-IR region was obviously not compensated by this mechanism in our preliminary experiments.

FIG. 4.

Comparison of wave length absorption of alumina adapted from Hellrung et al. ^{Reference Hellrung, Yeh, Depiereux, Gillner and Poprawe12}

Complementary tests were therefore performed at much higher laser powers on the SLM machine (up to 200 W) on single layers of powder. However, the results did not show any difference with those obtained on the marking machine at 20 W.

B. Selective laser treatments of alumina with absorption additive

As said previously, colloidal graphite was added within the ceramic granules to increase the optical coupling between the laser beam and alumina. Adding graphite presents two advantages: it exhibits a very high absorptivity at 1070 nm (>0.9) and it does not react with alumina, even at high temperature. Hence, there is no risk of forming new undesirable phases during the laser treatments. The effect of graphite addition on the alumina absorptivity was determined by spectrometry. The reflectance (R) of the material was measured on 5-mm-thick green pellets. Then the absorbance (A), defined as the ratio of absorbed radiation to the incident radiation, was calculated according the usual formula given in Eq. (3). ^{Reference Tolochko, Laoui, Khlopkov, Mozzharov, Titov and Ignatiev16}

(3) $$A = 1 - R - T$$

where T is the transmittance. In the present study, T was considered equal to zero considering the thickness of the pellets (whatever the amount of additive).

The absorbance for pure alumina is calculated in accordance with Hellrung et al. ^{Reference Hellrung, Yeh, Depiereux, Gillner and Poprawe12} Figure 5 shows the absorbance evolution for increasing amounts of absorption additive into alumina. At 1 vol%, the mixed powder exhibits an absorbance around 0.8.

FIG. 5.

Change of absorbance of alumina with absorption additive ratio.

Comparison of irradiated areas in pellets containing or not the additive reveals significant changes (Fig. 6). Consolidated microstructures with grains around 1 µm are visible for pellets containing 0.1 vol% of graphite. The grains are bonded to each other and form bigger particles with a spherical morphology and sintered necks. Unfortunately, the irradiated areas are significantly damaged with cracks and porosity. The presence of cracks can be explained by shrinkage occurring during sintering, thermal shocks, and/or thermal stresses induced by a heterogeneous sintering. Some pores are also noticed and are probably related to the graphite burning off during the laser treatment. Another possibility to explain these defects, according to the study of Deckers et al. ^{Reference Deckers, Shazad, Vleuels and Kruth15} is a local overheating of alumina leading to a subsequent boiling. Despite these defects, it is clear that the addition of graphite in alumina enhances the optical coupling and favors consolidation during the laser treatment.

FIG. 6.

SEM micrographs of (a) alumina green body without laser treatment, (b) laser treated area of top surface of alumina green body, and (c) laser treated area of alumina green body containing the absorption additive.

C. Density and microstructure

SLM cubes (10 × 10 × 10 mm³) were manufactured according to the design of experiment and starting from the alumina powder containing 0.1 vol% graphite. A total of ten cubes for each experiment were fabricated. Figure 7 shows an example of results.

FIG. 7.

Test specimens manufactured by SLM for density measurement.

All experiments from Table II, have not led to dense bulk cubes from the powder bed (Fig. 7). The powder bed was not densified for experiments E3 and E6. For experiment E9, the specimen exhibited a too weak mechanical resistance to be manipulated. Therefore, for experiments E3, E6, and E9, the densification levels of powder beds are considered equal to 0%.

The highest relative densities were obtained with experiments E4, E7, and E1 with 97.5, 97.3, and 92.9%, respectively (Table III). But slight shifts of geometry following the Z axis compared to the 3D-CAD model were observed for these experiments. In these cases, the temperature of the powder bed seems to have reached the alumina melting point, inducing menisci in irradiated areas. Distortions observed along the Z axis could be explained by the freezing of menisci, due to the rapid cooling. According to this explanation, the geometrical distortion shown in Fig. 7 (E1) would result from the addition of each slight layer deformation. Contrarily, a good geometric accuracy following X, Y, and Z axis, was observed with experiments E8, E5, and E2 leading to lower densities (76.9, 67.3, and 57.2%, respectively).

TABLE III.

Relative densities obtained by applying the experimental scheme of Table II.

Figure 8 shows SEM micrographs of polished cross sections of specimens obtained from experiments E4 and E8. SEM micrographs of experiment E8 [Figs. 8(c) and 8(d)] show porosity, cracks, and large periodic defects along the building direction. It is believed that some visible pores are the consequence of stacking defects in the powder bed. But large periodic defects likely results from a lack of cohesion between two consecutive layers. In this case, the energy density is probably no sufficient to weld consecutive layers, contrary to results obtained with experiment E4 [Figs. 8(a) and 8(b)] leading to a higher density and where cracks and pores are the major defects. The energy density applied for this experiment (16.6 J/mm²) seems to be suitable to avoid the largest defects, but not the smallest ones, like pores.

FIG. 8.

SEM micrographs of (a and b) experiment E4 and (c and d) experiment E8.

D. Influence of laser parameters on powder bed densification

The influence of each laser parameters on the density is reported in Fig. 9. Among the four factors investigated, point distance (pd) has the strongest influence, hatch space (HS) and exposure time (t) have a similar effect on the evolution of density. The change of power (P) leads to a surprisingly insignificant effect on densification of powder bed compared to point distance. Considering Fig. 9, point distance should be chosen as low as possible, i.e., around 50 µm according to the present study.

FIG. 9.

Average effect of laser parameters upon powder bed densification.

To understand how densification is affected by the laser treatment, it is useful to plot the relative density as a function of laser energy density (j). Figure 10 shows that globally, the densification of powder bed increases with laser energy density. But for experiments E3, E6, and E9 energy densities are similar to those of E2, E5, and E8 whereas densification rate is equal to zero. It is clear that the laser energy density is not sufficient to explain densification. Points overlapping [Eq. (4)] seems to be more relevant. If Ov > 0, there is an overlap of successive laser spots. And for Ov < 0, there is no contact between successive spots inside laser tracks.

(4) $${\rm{Point}}\;{\rm{overlapping:}}\;{\rm{Ov}} = (1 - ({{{\rm{HS}}} \mathord{\left/ {\vphantom {{{\rm{HS}}} {{\phi _z}}}} \right. \kern-\nulldelimiterspace} {{\phi _z}}}))$$

FIG. 10.

Change of parts relative density as a function of laser energy density.

As previously reported in the literature, the laser track overlapping exerts a significant influence on microstructures and geometrical feature of 3D parts. ^{Reference Su and Yang17}

For experiments E2, E5, and E8, the applied energy densities are 5.1, 6.3, and 7.6 J/mm², respectively, with Ov = −44% (a negative value means that no overlapping occurs). For experiments E3, E6, and E9, the energy densities are 5.3, 6.6, and 7.9 J/mm², respectively, with Ov = −477%. Finally, for experiments E1, E4, and E7, the energy densities are 13.3, 16.6, and 19.9 J/mm², respectively, with Ov = 28% (Table III). In the latest case only there is an overlap between consecutive laser spots contrary to the two other cases. The absence of powder bed densification can be explained by a too large distance between successive irradiated zones which do not lead to a single continuous track of densified material.

E. Influence of slice thickness on microstructure

In an attempt to increase further the density and quality of SLM parts, the influence of the layer thickness was studied. The initial layer thickness of 100 µm was reduced to 50 µm by sieving the alumina granules at 50 µm. Figure 11 shows the particle size distribution of the granules used to build items with a layer thickness of 50 µm. Figure 12 shows the change of microstructures obtained by applying parameters of experiment E8. By decreasing the layer thickness from 100 to 50 µm, an increase of density was observed from 76.9 to 85.1%, respectively. SEM micrograph of the parts cross section [Fig. 12(a)] shows a net improvement of the microstructure with smaller pores size [Fig. 12(b)]. This change may be explained by a more homogeneous and efficient melting of the powder bed allowing a better welding between two consecutive layers.

FIG. 11.

Particle size distribution of the spray-dried granules after sieving at 50 µm.

FIG. 12.

Influence of layer thickness upon SLM microstructures (a) 100 µm slice, (b) 50-µm slice, and (c) 50 µm slice with postthermal treatment at 1550 °C/2 h.

A suitable setting of laser parameters and a decrease of layer thickness clearly improve microstructures and the relative density of parts. Unfortunately, cracks and heterogeneities are still noticed. It is expected that the amount of defects may be decreased by considering some improvements of powder bed density. This could be achieved either through the use of more suitable powders and/or by considering specific devices to precompact the powder bed before laser irradiation. This clearly requires further technological developments. An alternative approach would be to use slurries instead of powder. This approach also known as the layer-wise slurry deposition leads to improved green density and is described by Tian et al. ^{Reference Tian, Günster, Melcher, Li and Heinrich11,Reference Tian, Mühler, Gomez, Günster and Heinrich18} However, this wet process required additional steps compared to our dry SLM process.

F. Influence of a thermal posttreatment

To increase the quality of parts, a thermal posttreatment was applied. Cubes manufactured by SLM with a 50 µm slice thickness were heat treated at 1550 °C for two hours. This temperature is usually considered to obtain fully dense objects with alumina P172 SB. Prior heat-treatment the density was 85.1%. Exactly the same value was obtained after postprocessing. Figure 12(c) shows the microstructure after posttreatment. Numerous cracks occurring perpendicular to building direction can now be observed. They may originate from a release of residual stresses in the structure during the thermal treatment, or an evacuation of gas resulting from the degradation of remnants of absorption additive. Dilatometric analysis of specimen before postthermal treatment shows no significant shrinkage (Fig. 13). SEM micrograph [Fig. 12(b)] exhibits clearly a molten structure suggesting that the temperature reached during SLM is over the melting temperature of alumina (T _m = 2070 °C). According to the study of Bertrand et al., ^{Reference Bertrand, Bayle, Combe, Goeuriot and Smurov19} the density cannot be increased by conventional postthermal treatment which is much lower than the shaping temperature. For a temperature lower than that achieved by SLM, internal rearrangement of matter is locked.

FIG. 13.

Dilatometric analysis of specimen manufactured by SLM with laser parameter of experiment E8 and slice thickness of 50 µm.

XRD analysis of SLM parts (Fig. 14) shows that the structures are well crystallized (despite melting) and that no change of the original crystalline structure of alumina occurs. Moreover, there is no glassy phase nor any secondary phase detected as a consequence of graphite addition.

FIG. 14.

Comparison of XRD patterns of raw alumina powder, SLM-treated alumina, and posttreated alumina parts at 1550 °C/2 h.

G. 3D complex shapes for demonstration purposes

By tuning laser parameters from the design of experiments, specimens with relative density of over 90% were manufactured without geometric deformation. Table IV shows that the level of density can be kept constant by decreasing the absorption additive ratio from 1.00 to 0.05 vol% for a same set of laser parameters.

TABLE IV.

Absence of effect of graphite amount addition (in the range 0.05–1 vol%) on the part density. (Laser parameters: P = 150 W; pd = 200 µm; HS = 50 µm; t = 700 µs).

For demonstrating purposes, 3D complex shapes have been manufactured from alumina powder containing 0.05 vol% absorption additive (Fig. 15). The relative density of these large objects was confirmed around 90% similar to that obtained on small cubes.

FIG. 15.

Pure alumina parts processed by SLM.