LZSA glass-ceramic laminates: Fabrication and mechanical properties

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 6 ( 2 0 0 8 ) 194–201

journal homepage: www.elsevier.com/locate/jmatprotec

LZSA glass-ceramic laminates: Fabrication and mechanical properties Cynthia M. Gomes a , Antonio P.N. Oliveira a , Dachamir Hotza a,∗ , Nahum Travitzky b , Peter Greil b a b

´ Group of Ceramic and Glass Materials (CERMAT), Federal University of Santa Catarina (UFSC), 88040-900 Florianopolis, SC, Brazil Department of Materials Science, Institute of Glass and Ceramics, University of Erlangen-Nuremberg, 91058 Erlangen, Germany

a r t i c l e

i n f o

a b s t r a c t

Article history:

The aim of this work was to fabricate LiO2 –ZrO2 –SiO2 –Al2 O3 (LZSA) glass-ceramics lam-

Received 21 June 2007

inates by laminated object manufacturing (LOM) and to characterise some properties of

Received in revised form

the laminates before and after sintering. Correlations between green tape properties, pro-

8 November 2007

duced by aqueous tape casting, and the green and sintered laminate properties were also

Accepted 4 December 2007

determined. Processing optimisation was based on a factorial design. The microstructure of the LZSA glass-ceramic laminates showed a homogeneous distribution of porosity and the main phases were identified as being ␤-spodumene and lithium metasilicate. Laminates

Keywords:

with a 0◦ /90◦ layer orientation attained a significantly higher bending strength of 120 MPa

Glass ceramics

compared to 70 MPa for the 90◦ /90◦ orientation.

Tape casting

© 2007 Elsevier B.V. All rights reserved.

Rapid prototyping Factorial design

1.

Introduction

LiO2 –ZrO2 –SiO2 –Al2 O3 (LZSA) glass ceramics present some interesting properties: e.g. good chemical and thermal shock resistances and a very low thermal expansion coefficient (4–6 × 10−6 ◦ C−1 ) (Giassi et al., 2006). Additionally, this system can be processed at low sintering temperatures (40 wt%) requires the use of some adhesive agents (such as double-side adhesive tapes or diluted binder solutions) essential to the LOM process. This procedure, often used for low pressure lamination (Piwonski and Roosen, 1999; Baud´ın et al., 2005), promotes the interconnection between the adjacent tapes since the boundary between the tapes should be undetectable after compression.

195

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 6 ( 2 0 0 8 ) 194–201

Table 1 – Experimental design for suspension compositions Run 1 2 3 4 5 6 7 8 9 10 11 a

Dispersant (wt%)

Bindera (wt%)

0.00 1.50 0.00 1.50 0.00 1.50 0.00 1.50 0.75 0.75 0.75

Plasticizer (wt%)

21.57 21.57 33.28 33.28 21.57 21.57 33.28 33.28 27.42 27.42 27.42

0.50 1.00 1.00 0.50 1.00 0.50 0.50 1.00 0.75 0.75 0.75

Total Solids (wt%) 50.74 50.32 47.98 47.81 50.08 49.88 47.61 47.28 48.87 48.87 48.87

Aqueous solution with 31.51 wt% solids.

The aim of this work is to investigate some properties of LZSA laminates (bending strength, density, porosity, microstructure and crystalline phases) as well as possible correlations between green and sintered laminate properties. In addition the influence of green tape properties on the final prototype produced by LOM was also investigated. A factorial design was used to find the suitable tape casting compositions useful for the LOM process.

2.

0.80 0.80 0.80 0.80 3.20 3.20 3.20 3.20 2.00 2.00 2.00

Antifoamer (wt%)

Experimental procedure

A powder of composition LZSA, d50 = 2.10 ␮m, bulk = 2.58 g/cm3 , surface area according to the Brunauer, Emmett and Teller method = 8.29 m2 /g (Brunauer et al., 1938) was used for aqueous suspension preparation, following a factorial experiment design 23 + 3 centre points (Montgomery, 1997). Nine different compositions are shown in Table 1. Ammonium polyacrylate, polyvinyl alcohol (molar mass = 30.000 g/mol), polyethylene glycol (molar mass = 400 g/mol) were used as dispersant, binder and plasticizer, respectively. As antifoaming agent, a blend of modified fatty and alkoxylated compound was used. The preparation of the suspensions was carried out in three stages: dispersion of the parent glass powder in distilled water with the dispersant for 24 h; addition and homogenisation of the binder solution for 12 h; and addition of the plasticizer and the antifoam followed by a mixing period of 12 h. In order to choose the carrier for the parent glass aqueous suspensions, the wetting behaviour was also analysed. The measurements of contact angle were carried out in a videobased semi-automatic contact angle meter (OCA 30, Data Physics, Germany), with silicone covered and non-covered polyethylene terephthalate (PET) films. Flow curves were measured to determine the apparent viscosity and shear-thinning index (n) of the suspensions. The measurements were carried out in a rotational rheometer with a cone and plate geometry (UDS 200, Paar Physica, Germany), at room temperature. Tape casting was performed in a laboratory tape caster with double blade device, at a constant speed of 450 mm/min. After casting, the tapes were dried for 48 h and then released from

the carrier. Releasing behaviour, homogeneity and thickness of the tapes were used as parameters to choose three suitable compositions for the lamination process by LOM. The LOM process was carried out in a CW-CO2 -laser equipment (1015, Helisys, USA). A retract from 0.10 mm was used, which is related to the distance in between the heated roller and the sample surface. Generally, the lower is the retract, the higher is the pressure applied. The laser power was 16.8 W, the cutting and roller speed were 50 mm/s and 25 mm/s, respectively, and the roller temperature was 80 ◦ C. The tapes were laminated using a thin layer of 5 wt% aqueous binder solution as adhesive agent applied with a painting brush on the richest organic side from each tape before stacking the subsequent tape (Baud´ın et al., 2005). Samples for mechanical tests and microstructure characterization consisted in rectangular bars with 27.0 cm × 3.5 cm × 2.5 cm, as result of 20 green-stacked tapes (Fig. 1a). Since the structure of the green samples is a result of two processes, tape casting and lamination, a more complex geometry (Fig. 1b) was also built up. The x- and y-directions were built up during the tape casting process. They present different shrinkage behaviour during heat treatment due to the polymer chain alignment in the casting. The LOM is responsible for the z-direction building. These differences are expected to bring also irregular shrinkages during the thermal treatment. The apparent and theoretical densities of the green and sintered laminates were determined geometrically and by Hepicnometry (AccuPyc 1330, Micrometrics, USA), respectively. Three-point bending strength tests were carried out in a universal testing machine (4204, Instron Corp., USA), at room temperature. The bending load was applied at a constant

Fig. 1 – LOM geometries: (a) rectangular bars for mechanical characterization (b) “stair-like” samples for dimensional analysis.

196

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 6 ( 2 0 0 8 ) 194–201

Fig. 2 – Response surfaces for the (a) viscosity and (b) shear-thinning index, as a function of the binder and dispersant content.

crosshead speed of 0.5 mm/min. The results were an average of ten measurements both for the green as for the sintered laminates. The heat treatment applied to the suitable compositions followed a constant weight loss program of 0.01%/min. The total cycle presented two main threshold temperatures for sintering and crystallisation. The binder was completely removed at 525 ◦ C. The first threshold temperature was at 750 ◦ C for sintering. At this temperature, three different holding times were applied: 30, 45 and 60 min. The crystallisation temperature followed at 850 ◦ C, for 30 min.

3.

Results and discussion

3.1. Characterization of the aqueous parent glass suspensions for tape production The viscosity (k) and shear-thinning index (n) of all suspensions were adjusted to the Herschel–Bulkley model, according

to Eq. (1) (Macosko, 1993): n

 = 0 + k()

(1)

where  is the shear stress,  0 , the apparent yield stress, and , the shear rate. The influence of each organic component on the rheological behaviour of the glass suspensions was verified through an analysis of variance (ANOVA). The parameters F and p on the ANOVA analysis represent this influence. The higher is the F factor and the lower is the p-value, the higher is the influence of the respective parameter analysed. The dispersant and binder concentration influences significantly the viscosity, according to the low p-values reached for these two variables on the ANOVA (p-values of 0.0009 and 0.03, respectively). A very high influence of the dispersant content was observed since it promotes a reduction of the internal particle attraction due to the deagglomeration of aggregates (Reed, 1988).

Fig. 3 – (a) Response surface adjusted for the contact angles of the aqueous parent glass suspensions as a function of the binder and dispersant contents in a PET carrier; (b) wetting behaviour of compositions 2 (high contact angle) and 8 (suitable contact angle).

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 6 ( 2 0 0 8 ) 194–201

197

Fig. 4 – (a) Response surface adjusted for the contact wetting angle of the aqueous parent glass suspensions as a function of the binder and dispersant contents in a silicone-coated PET carrier; (b) wetting behaviour of compositions 2 and 8, both presenting high contact angles.

Fig. 2a and b shows, respectively, the response surfaces adjusted for a linear model relating the viscosity and shearthinning index (n) and the binder and dispersant content on the suspensions. An effect between binder and dispersant content on the viscosity of the suspensions was noticed (loss of linearity on the response surface graphic). High values of viscosity (25 Pa s) were observed when high amounts of dispersant were used (Fig. 2a). This behaviour is undesired since it leads to lower values of green density, which increase firing shrinkage. Furthermore, values of viscosity higher than 20 Pa s characterise a paste system instead of a fluid system (Mistler and Twiname, 2000). The shear-thinning index was influenced mainly by the binder content (Fig. 2b), although no significant p-values were identified on the ANOVA for this property. As also observed in the literature (Chartier and Rouxel, 1997), all these rheological parameters influence directly the tape characteristics, such as presence of cracks after drying,

green density and microstructure homogeneity. In addition to the rheological properties, the easiness on removing the dried tape from the carrier should be evaluated, since some defects are introduced during this step on the tape production (Lutz and Roosen, 1998). This releasing ability was investigated through an ANOVA showing the influence of the organic composition on the contact angle of all suspensions cast on PET carriers. Only the dispersant amount was shown to present a significant influence on the wetting behaviour of the suspensions independently of the carrier used (p-values of 0.05 and 0.02, respectively for silicon coated an uncoated PET films). However the lowest contact angles (60◦ ) were observed when PET films were used as carrier. Figs. 3 and 4 present the response surfaces adjusted for the contact angles of the suspensions as a function of the dispersant and binder contents, when casting them on a PET and silicone-coated PET carrier, respectively. The contact angles of the suspensions cast on PET films were lower (75◦ ). However, only compositions 3, 4 and 8 were possible to be released from the PET carrier. The other tapes, although presenting good wetting behaviour, could not be easily removed from the PET film without cracking or breaking. According to Lutz and Roosen (1998), silicone-coated PET films can only be used as a tape carrier for aqueous-based slurries, if the contact angle of water is minimized to about 30◦ up to 45◦ .

3.2. Green properties of parent glass tapes and laminates

Fig. 5 – Green density and porosity from tapes as a function of the total organics content.

Green properties of tapes and laminates were studied to verify the influence of the tape properties on the final laminate. Fig. 5 shows the green density and porosity presented by tapes from compositions 3, 4 and 8 as a function of the total organic content.

198

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 6 ( 2 0 0 8 ) 194–201

Table 2 – Properties of green tapes and laminates Run

Properties Tensile strength (MPa)

Green density (g/cm3 )

Bending strength (MPa)

Porosity (%)

Tapes

3 4 8

1.45 1.31 1.41

5.6 ± 0.1 5.7 ± 0.2 4.4 ± 0.1

– – –

49.65 55.73 40.33

Laminates

3 4 8

2.39 2.35 2.34

– – –

17.0 ± 0.6 18 ± 4 15.8 ± 0.6

18.70 11.01 11.22

Fig. 6 – Tensile strength curves for tapes of compositions 3, 4 and 8.

Fig. 7 – Sintered bending strength of laminates with threshold times at the sintering temperature.

Tapes from composition 4 presented the lowest values of green density and highest porosity (1.31 g/cm3 and 55.73%, respectively). This composition presented also the highest viscosity (17.3 Pa s) comparing to compositions 3 (4.2 Pa s) and 8 (14.0 Pa s). According to the literature (Moreno, 1992), suspensions presenting high viscosity lead to tapes with low green densities, due to the reduction of the powder settling. This can be observed for composition 3, which presented the lowest value of viscosity and highest green density (4.2 Pa s and 1.45 g/cm3 , respectively). Fig. 6 shows the effect of the organics content on the tensile strength curves from tapes of compositions 3, 4 and 8. The highest tensile strength (5.7 MPa) was reached for tape 3,

which presents the lowest organic content (47.8 wt% solids) and the highest green density (1.45 g/cm3 ). Table 2 presents a resume of the main properties of the green tapes and laminates, with compositions 3, 4 and 8. Tapes with higher tensile strength led to laminates with higher bending strength. No interaction was observed concerning green density and bending strength of sintered laminates.

3.3.

Sintered laminates

Fig. 7 presents the bending strength of green and sintered laminates with different threshold times. It was not possi-

Table 3 – Properties of laminates after sintering Run

Properties Apparent density (g/cm3 )

Laminates

a b

3 4 8

60 min threshold at 700 ◦ C. Geometrically calculated.

2.09 2.29 2.08

Bending strength (MPa)a 106.4 ± 7 106 ± 12 90 ± 20

Shrinkage x directionb (%) 21.89 23.19 22.16

Shrinkage y directionb (%)

Shrinkage z directionb (%)

20.63 24.10 23.02

19.74 22.18 16.50

Porosity due to internal delamination (%) 87.70 53.05 85.74

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 6 ( 2 0 0 8 ) 194–201

199

Fig. 8 – Microstructures LZSA glass-ceramics sintered with different threshold times: (a) and (b) 30 min, (c) and (d) 45 min, (e) and (f) 60 min.

Table 3 presents some properties of the sintered laminates with compositions 3, 4 and 8, when applied a sintering threshold time of 60 min. As can be seen, the shrinkage values along the x- and y-directions were almost the same. In all compositions the shrinkage on z-direction was lower by a factor of 5–30% that in the other two directions. This behaviour could be expected since during lamination some pressure is applied to promote the interconnection between the layers. Compositions with the lowest sintered densities presented also the higher amounts of porosity due to internal delamination. The higher density presented for samples of composition 4 influenced also the higher shrinkage on zdirection.

Fig. 9 – Diffractogram of LZSA glass-ceramic showing the main crystalline phases: () ␤-LiAlSi2 O6 and () Li2 SiO3 .

ble to notice any influence of the organic content on the bending strength of the laminates. Moreover, a large standard deviation is observed in all samples. Since no obvious defect was detected on the samples submitted to the mechanical tests, this dispersion could be a result of some internal defects. Deviations on the flexural strength from square parts of hydroxyapatite/calcium phosphate glass produced by LOM was also observed and attributed to the incomplete layer fusion of the ceramic tapes (Chartoff et al., 2002).

Fig. 10 – Bending strength values presented for green and sintered laminates built up with different tapes orientation.

200

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 6 ( 2 0 0 8 ) 194–201

Fig. 11 – Stair-like geometry for the study of dimensional stabilisation: (a) green sample, (b) sintered sample.

3.4.

Laminates microstructure

Fig. 8 shows the microstructure of samples with composition 4, when submitted to different threshold times at the sintering temperature (750 ◦ C). It can be seen that a homogeneous porous distribution and the increase on the threshold time led to a decrease of the internal porosity. The white points were identified by EDX as being monoclinic ZrO2 , probably due to the devitrifying process during the melting process. This contamination helps the crystallizing process since this system mainly presents heterogeneous surface nucleation. Fine crystals, homogenously distributed in the glass matrix, were also observed. No influence from the threshold time could be observed on the size and distribution of the crystals. This behaviour was also observed in samples with compositions 3 and 8. The main crystalline phases were identified by X-ray diffraction (XRD) as being zirconium metasilicate (Li2 SiO3 ) and ␤-spodumene (␤ - LiAlSi2 O6 ), Fig. 9. The ␤-spodumene crystalline phase presents a low thermal expansion coefficient (TEC) in the order of 2 × 10−6 K−1 (Kingon et al., 1991) and a high temperature dimensional stability (Sakamoto and Yamamoto, 2006).

3.5.

Lamination optimisation

New rectangular samples from composition 4 were laminated, under the same LOM parameters used before, but with two different tapes orientation (related to the casting direction). The 90◦ /90◦ or 0◦ /0◦ orientation, already laminated, and a 0◦ /90◦ tape orientation in order to minimize defects originated due to anisotropic shrinkage. Fig. 10 presents the bending strength values reached for green and sintered laminates with different tape orientations. No significant differences were observed on the green bending strength values presented by laminates with 0◦ /0◦ and 0◦ /90◦ tape orientation. Both samples showed good handling during the de-cubing process. However, after sintering, the 0◦ /90◦ laminates presented higher values of sintered bending strength, compared to 0◦ /0◦ laminates. Comparing the bending strength values of composition 4 without (Table 3) and with (Fig. 10) tape orientation, an increase can be seen on the results presented for 0◦ /90◦ laminates. This is probably due to a decrease of the porosity as a consequence of internal delamination, since these samples presented 95% theoretical density. It can also be seen that a decrease of the standard deviation occurred.

Using the 0◦ /90◦ tapes orientation, which showed better mechanical properties, a stair-like geometry was laminated in order to test out the dimensional stability of the studied system to the LOM process (Fig. 11). The green sample presented good handling characteristics concerning the de-cubing process. After sintering, it retains its original profile, without delamination, although some warping was observed in z-direction.

4.

Conclusions

LZSA laminates were produced by Laminated Object Manufacturing using green tapes of parent glass produced by aqueous tape casting. The rheological properties of the suspensions determined three compositions that could be used to produce green tapes free of defects. The LOM process was successfully carried out using green tapes and a diluted binder solution. It was observed that tapes with higher tensile strength after lamination originated laminates with higher bending strength values; however no differences could be observed on the green density and porosity. The microstructure of the LZSA glass-ceramic laminates showed a homogeneous distribution of porosity and the main phases were identified as ␤-spodumene and lithium metasilicate. The LOM process is a new possibility to build up LZSA glassceramics substrates from glass precursor green tapes. Good dimensional stability was observed on the x, y-directions.

Acknowledgements The authors are thankful for the financial support from the CAPES/DAAD interchange program and for Colorminas (Brazil), Vanderbilt (USA) and KSE (Germany) for supplying samples.

references

´ Baud´ın, C., Sanchez-Herencia, A.J., Gurauskis, J., 2005. Joining green ceramic tapes made from water based slurries by applying low pressures at ambient temperature. J. Eur. Ceram. Soc. 25, 3403–3411. Brunauer, S., Emmett, P.H., Teller, E., 1938. J. Am. Chem. Soc. 60, 309.

j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 6 ( 2 0 0 8 ) 194–201

Chartier, T., Rouxel, T., 1997. Tape-cast alumina-zirconia laminates: processing and mechanical properties. J. Eur. Ceram. Soc. 17, 299–308. Chartoff, R.P., Steidle, C., Klosterman, D.A., Graves, G.A., Osborne, N.R., 2002. Automated fabrication of custom bone implants using rapid prototyping. In: Business Briefing: Medical Device Manufacturing & Technology. Dernovsek, O., Naeini, A., Preu, G., Wersing, W., Eberstein, M., Schiller, W.A., 2001. LTCC glass-ceramic composites for microwave application. J. Eur. Ceram. Soc. 21, 1693–1697. Giassi, L., Hotza, D., Alarcon, O.E., Fredel, M.C., Oliveira, A.P.N., 2006. Sintering and crystallization of LZSA glass powder compacts formed by injection molding. Am. Ceram. Soc. Bull. 84, 9301–9306. Gomes, C.M., Biscaia, F.N., Quinaud, J.T., Montedo, O.R.K., Oliveira, A.P.N., Hotza, D., 2006. Aqueous tape casting of LZSA glass ceramics. Ceram. Trans. 193, 9–16. Kingon, A.I., Davis, R.F., Thackery, M.M., 1991. Engineering properties of multicomponent and multiphase oxides. In: Engineered Materials Handbook, vol. 4: Ceramics and Glasses. ASM International, USA. Klosterman, K., Chartoff, R., Osborne, N., Graves, G., 1997. Automated fabrication of monolithic and ceramic matrix composites via laminated object manufacturing. In: Solid

201

Freeform Fabrication Symposium Proceedings, Austin, pp. 537–549. Lutz, C., Roosen, A., 1998. Wetting behavior of tape casting slurries on polymeric tape carriers. Ceram. Trans. 83, 163–170. Macosko, C.W., 1993. Rheology: Principles, Measurements and Applications. VCH Publishers, New York. Mistler, R.E., Twiname, E.R., 2000. Tape Casting—Theory and Practice. The American Ceramic Society, Westerville, OH, USA. Montedo, O.R.K., Reitz, G.M., Oliveira, A.P.N., Hotza, D., Siligardi, C., 2004. Extruded LZS glass-ceramics. Am. Ceram. Soc. Bull. 83, 9201–9206. Montgomery, D.C., 1997. Design and Analysis of Experiments. Wiley, New York. Moreno, R., 1992. The role of slip additives in tape casting technology: part II—binders and plasticizers. Am. Ceram. Soc. Bull. 71, 10. Piwonski, M.A., Roosen, A., 1999. Low pressure lamination of ceramic green tapes by gluing at room temperature. J. Eur. Ceram. Soc. 19, 263–270. Reed, J.S., 1988. Principles of Ceramics Processing, second ed. Wiley Interscience Publications, New York. Sakamoto, A., Yamamoto, S., 2006. Volume change of ␤-spodumene s.s. glass-ceramics due to structural relaxation. J. Ceram. Soc. Jpn. 114, 380–385.