Wetting and interfacial phenomena between a Ni-based superalloy and silica-based ceramic cores with ZrSiO4 additions

Wetting behavior and interfacial reactions between a Ni-based superalloy and silica-based ceramic cores containing 10 wt.%, 30 wt.% and 50 wt.% ZrSiO4 were studied using a sessile drop method. The reaction products were characterized by SEM/EDS, EPMA and XPS analysis. SiO2 was the main component in the ceramic cores taking part in the interfacial reactions and the reaction products were composed mainly of Al2O3 layer with some ZrOx scattered. A Cr2O3 halo formed outside the wetted ceramic surface. The wetting angle increased in the initial stage to a peak value and then decreased to a constant value. The increase of the wetting angle may be attributed to the gas evaporation at the interface and the decrease of the wetting angle was due to the formation of the reaction layer at the interface.


Introduction
Advanced superalloy hollow turbine blades are fabricated by investment casting, which allows dimensionally accurate components to be produced and is a cheaper alternative than forging or machining since the waste material is kept to a minimum.The internal air-cooling passages of the blades are provided by the use of ceramic cores, which form the internal geometry of the blades.With the increase of the turbine inlet temperature, the internal air-cooling passages of the blades become more complex in order to increase the cooling efficiency of the blades, i.e., the shape of the ceramic cores become more complicated and thus more excellent properties of the ceramic cores are needed [1][2][3][4] .
Generally, ceramic cores are comprised of silica-based and alumina-based materials and they are produced by the way of hot press molding.Silica-based ceramic cores are extensively used due to their good properties such as low coefficient of thermal expansion as well as fine removability from the interior of the thin-walled blades.Compared with silica-based ceramic materials, alumina-based materials exhibit more chemical stability with the alloy melt at temperatures higher than 1550 º C but poor removability from the blades [5][6][7] .The main component of silica-based ceramic materials is fused silica, which is in glass state.Fused silica may partly change to crystal-state materials named cristobalite in the temperature range of 1100 º C-1250 º C. The phase transformation from glass to crystals would induce cracks in the ceramic, which lower the strength of the ceramic cores.Zircon sand (ZrSiO4), with high strength and higher stability than fused silica, is often added into ceramic cores to improve the flexural strength and the creep resistance of the silica-based ceramic cores [8][9][10][11] .
Investment casting of columnar crystal blades and single crystal blades may take about one or two hours at high temperatures in excess of 1500 º C.During the long time and high temperature casting process, ceramic cores contact the alloy melt closely and serious interfacial reactions may occur at the alloy-ceramic interface.Removing the reaction products from the castings is costly and it would deteriorate the surface quality of the castings.Furthermore, the interfacial reactions would affect the wettability between the alloy melt and the ceramic cores.A smaller wetting angle caused by the interfacial reactions would facilitate the penetration of the alloy melt into the ceramic cores through the capillaries on the cores, which may bring about inclusions of the ceramic oxides on the surface of the solidified alloy [12] .Accordingly, it is of great importance to investigate the interfacial reactions and wettability between superalloys and the ceramic materials.
C, Hf, Cr and Al are reported to be the main active elements that cause interfacial reactions between superalloys and the ceramic materials.For instance, Zheng et al. [13] have reported that Cr in superalloys would react with silica-based ceramic cores at high temperatures and produce Cr2O3, which induce metallic nodular protrusions on the alloy surface.Li et al. [14] have found that Hf and Cr both take part in interfacial reactions between superalloys and ceramic cores and the reaction products are HfO2 and Cr2O3.As for the wettability, Valenza et al. [15] have studied the wettability between superalloys and different ceramic mold materials (sapphire, polycrystalline alumina, zirconia and mullite) and they have found that the composition of the ceramic mold would influence the wettability.However, to the best of our knowledge, only a few report has so far been systematically concerned with the interfacial reactions and wettability between superalloys and ceramic core materials.
Based on the above considerations, wetting and interfacial phenomena between superalloys and ceramic core materials are investigated in this study using a sessile drop method.The effect of ZrSiO4 contents on the wettability and the interfacial phenomena is also taken into account.

Materials and experiments
The alloy used in the present work was a Ni-based superalloy.The composition of the alloy (mass fraction, %) was Al 4.8, Ti 4.5, Cr 9.1, Co 10, Mo 3.1, C 0.2 and Ni in balance.Before experiment, the alloy was cut into 2 mm × 2 mm × 2 mm cubic samples, polished and ultrasonically cleaned in acetone to remove the oxides on the surface.
The ceramic cores were made through the way of injection moulding.The slurry was comprised of 88 wt.% ceramic powders and 12 wt.%wax-based plasticizer.The ceramic powders were commerciallyavailable fused silica and zircon sand with particle size of -300 mesh.The content of zircon sand in the ceramic powders was 10 wt.%, 30 wt.% and 50 wt.%,respectively.The purities of the ceramic powders were listed in table 1.The wax-based plasticizer was made up of 70 wt.%paraffin, 25 wt.% beeswax, 2 wt.% polyethylene and 3 wt.%stearic acid.The ceramic core slurry was injected into a mold at (120 ± 1) ℃ to obtain green samples of 120 mm length, 10 mm width and 4 mm height.The green ceramic cores were heated to 450 ℃ at a rate of 80 ℃/h and held 4 h for dewaxing.After dewaxing, the samples were heated to 1200 ℃ at a rate of 120 ℃/h and held 4 h for sintering.Ceramic substrates of 20 mm length, 10 mm width and 4 mm height were cut from the sintered samples.The surface roughness (Ra) of the ceramic substrates was measured by a DEKTAK 6 M surface profilometer over a distance of 2 mm at a speed of 100 μm/s and the average value of the surface roughness was about 2 μm.The substrates were ultrasonically cleaned in acetone before wetting experiment.
The ceramic substrate was placed on an alumina support inside the furnace and was adjusted to a horizontal position.An alloy sample was then placed on the ceramic substrate.Subsequently, the furnace was evacuated to 5 × 10 -4 Pa and then heated at a rate of 20 K min -1 to 1200 ℃.An Ar gas was then introduced to the chamber to about 0.12 Pa.To reduce water and oxygen levels, the Ar gas was purified by passing through a magnesium (99.9%) furnace at 673 K, a dehydrating column filled with molecular sieves and finally an oxygen-adsorption column filled with high effectiveness palladium-type agents.During the process of introducing Ar gas, the furnace was continuously heated to the experimental temperature of 1550 º C to melt the sample.High-resolution photographs of the melt profile were taken by a high resolution (1504 × 1000 pixels) charge-coupled device (CCD) camera.
After the experiment, the samples were furnace cooled at 20 K/min and the captured drop profiles were analyzed using drop-analysis software to calculate the wetting angles.The microstructures of the alloy drop and the ceramic substrate were observed by using a scanning electron microscope (SEM, JMS-6301F, Japan) equipped with energy dispersive spectroscopy (EDS) analysis.The distribution of elements in the reacted area was characterized by electron probe microanalysis (EPMA-1610, Japan).The composition and chemical state identification of the main products was performed by using X-ray photoelectron spectroscopy (XPS, ESCALAB250, Japan).

Results
(1)Wetting behavior Figure 1 presents the variation of the wetting angle with time for the superalloy on the ceramic substrate.10 wt.%, 30 wt.% and 50 wt.% in the figure represent the content of ZrSiO4 in the ceramic substrates.It can be seen that the initial wetting angle varies slightly from 150º to 155º with a ZrSiO4 content increasing from 10 wt.% to 50 wt.%.Such a result indicates that ZrSiO4 is less wetted by the superalloy melt compared with SiO2.The wetting angle for the alloy/ceramic couple of ceramic containing 10 wt.% ZrSiO4 increases gradually up to a peak value of 162º and then decreases exponentially to reach a constant value of 120º.The changes of the wetting angle with time are in agreement with the characteristics of reactive wetting [16]   .The similar trends are observed in the alloy/ceramic couples of ceramics containing 30 wt.% and 50 wt.%ZrSiO4.With increasing of ZrSiO4 content, the peak value of the wetting angle decreases slightly, while the constant value of the wetting angle increases greatly.The constant value for the couples containing 10 wt.%, 30 wt.% and 50 wt.%ZrSiO4 are 120º, 132º and 145º respectively.The time to tc tc tc Peak value reach the wetting equilibrium (tc) decreases from 730 s to 430 s when the content of ZrSiO4 in the ceramic increases from 10 wt.% to 50 wt.%.
(2)Microstructure analysis Figure 2 shows the typical macrograph of the ceramic substrate and the corresponding alloy drop after the wetting experiment.The wetted surface of the ceramic is circled by a black region.Due to physical and chemical interactions between the alloy melt and the ceramic, the wetted surface of the ceramic substrate is ruptured.Figure 3 (a)-3(c) represent the interface microstructures of the alloy drops removed from the ceramic with 10 wt.%, 30 wt.% and 50 wt.%ZrSiO4.For each alloy drop, a continuous grey layer is observed at the interface and some white lump products are found distributing in the grey layer.The thickness of the grey layer shows slight decrease and the amount of the white lump products exhibits obvious increases as the content of ZrSiO4 varies from 10 wt.% to 50 wt.%.   4 (a), 4(a1) -(a5) exhibit the microstructure and the distribution of Cr, Al, Zr, Si and O measured by EPMA analysis on the ceramic surface with 50 wt.%ZrSiO4.It is obvious that some irregular lump products circle the wetted surface of the ceramic and the lump products are rich in Cr, with some Al, seen from figure 4(a1) and figure 4(a2).Figure 4 (b) displays the detailed microstructure in the rectangle zone in figure 4 (a).There is a clear boundary between the wetted surface and the Crenriched region, as is marked by the white dot curve.XPS analysis is also used to confirm the phase of the Cr-enriched product.Figure 4 (c

Discussion
(1)Formation of the interfacial microstructures As mentioned above, the reaction products at the interface are Al2O3 layer with some ZrOx scattered in it.It has been reported that SiO2 is reduced by Al according to the reaction 4(Al)+3SiO2→ 3(Si)+2Al2O3 [17]   .Since some Si elements are observed in the alloy, we think that SiO2 has dissolved at the interface, as described in Eustathopoulos's research [18] .After its dissolution, O atoms are combined by Al to form Al2O3 at the interface and some Si atoms penetrate into the alloy.Considering theΔG Θ -T diagram (figure 5) using thermodynamic date from Liang [19] , ΔG Θ for the oxides such as TiO2, Cr2O3 and CoO, which could be produced by redox reactions between the alloy atoms and the ceramic materials, are higher than that of Al2O3.Accordingly, the layer of Al2O3 forms preferentially and accordingly almost no TiO2, Cr2O3 and CoO form at the interface.The remained Si produced by the redox reaction between Al and SiO2 may combine O atoms and form SiOx/Si at the interface.
As for the formation of ZrOX, we should consider the decomposition of ZrSiO4.It has been reported that pure ZrSiO4 decomposes to ZrO2 and SiO2 at temperatures higher than 1600 o C.However, a small amount of impurities may change the decomposition temperature to about 1500 o C [20] .In the present study, ZrSiO4 powders consist of a small amount of Fe2O3 and TiO2.The impurities would promote the decomposition of ZrSiO4, producing ZrO2 and SiO2.ZrO2 is then reduced by Al, which produce ZrOx at the interface.The role of Cr has been discussed by several papers and Cr is reported segregating at the alloyceramic interface [21][22] .In the present work, the enrichment of Cr is observed mainly outside the wetted surface (the black region in figure 2).Early well known theories such as surface diffusion and evaporation-condensation can not account for the formation of the Cr-enriched region that circles the wetted surface of the ceramic.On one hand, surface diffusion is hardly to produce such a wide halo whose width is about 2 mm, as indicated in figure 2. On the other hand, if the Cr-enriched region is caused by the evaporation of Cr, the vapor pressure of Cr would be higher than the vacuum degree in the furnace.The calculation for vapor pressure of Cr is based on the formula 0 e i i i P a P  where e i P is the vapor pressure of component i in the alloy, ai is the activity of i and 0 i P is the vapor pressure of pure material i. 0 i P for Cr is calculated according to the equation 0 31 lg 10 lg 10 where A, B, C and D are constants for pure materials which are available from Liang [19] .The activity of Cr in the alloy is calculated by using Thermo-Cal software.The vapor pressure of Cr at the texting temperature is calculated to be about 2 × 10 -5 Pa and the evaporation of Cr is therefore negligible.Therefore, there should be other mechanisms that account for the formation of the Cr-enriched region on the ceramic surface.Xian [23] has suggested a model of rapid adsorption and then film overflow.In the system with active elements in the alloy, the active elements may quickly move to the alloy-substrate interface under the attractive force.Then a thin layer of the liquid, which have enriched the active elements, overflows to the substrate surface and the wide halo is therefore produced.A distinguishing character of such flow lies in that the moving speed of active elements to the substrate surface is more quickly than the moving speed in surface diffusion.In our case, we guess that Cr elements follow the film overflow mechanism and move quickly to the surface of the ceramic.As Cr has a high affinity to O, Cr2O3 is thereby formed by Cr combing O on the ceramic surface.
(2)Effect of interfacial reactions on wetting Based on figure1, the wetting process for the superalloy melt on the ceramic substrate can be roughly divided into two stages.In the first stage, the initial wetting angle increases gradually to a peak value.This stage continues less than 200 s and the peak value varies for each couple.Eustathopoulos et al [18] have reported that the increase of the wetting angle is resulted by gas evaporation from the interface.There are two possible reactions that could produce gases in our experiment.One is that Si from the decomposition of SiO2 may react with the undecomposed SiO2 and form SiO by the reaction of Si+SiO2→2SiO (gas) [24] and the other is that C in the alloy may react with SiO2 following equation of C+SiO2→SiO (gas)+CO (gas) [25] , which has been reported in our former work.Once CO and SiO are produced, the evaporation of these gases will give rise to the increase of the wetting angle in the first stage.Obviously, the amount of gas produced by the interfacial reactions is corrected with the content of SiO2.The ceramic with 10 wt.% ZrSiO4 has a maximum SiO2 content than the ceramic with 30 wt.% and 50 wt.%ZrSiO4.More SiO2 may take part in the reactions, producing more gases compared with the other two couples.Accordingly, the couple of ceramic containing 10 wt.% ZrSiO4 shows the highest peak value of wetting angle and the couple with 50 wt.%ZrSiO4 exhibits the lowest peak value.
In the second stage, the wetting angle decreases exponentially to a constant value, showing the characteristic of reactive wetting.It is obvious that in the couple of ceramic containing 10 wt.% ZrSiO4, the spreading rate of the superalloy melt is more rapid and the time to reach the wetting equilibrium (tc) is smaller than in the other two couples.The reason may also lies in that there are more SiO2 that would take part in the interfacial reactions.As there are three reactions that consume SiO2 and the reaction products contain not only Al2O3 but some gases, there is no obvious linear dependence that the couple of ceramic containing 10 wt.% ZrSiO4 shows the maximum thickness of Al2O3 layer.

Conclusions
Wetting and interfacial phenomena between superalloy melt and silica-based ceramic cores are investigated taking into account the effect of ZrSiO4 contents.The main conclusions are as follows:(1) SiO2 is the main component in the ceramic core that takes part in interfacial reactions.The reaction products at the interface compose Al2O3 layer, a minor SiOx/Si and some ZrOx.Moreover, an annular zone of Cr2O3 forms near the wetted surface of the ceramic and its formation shows little or no influence on the wettability between the superalloy and the ceramic core materials.(2) The wetting angle increases in the initial 200 s to a peak value and then decrease to a constant value for the couples of ceramic containing 10 wt.% , 30 wt.% and 50 wt.%ZrSiO4.The increase of the wetting angle may be attributed to the evaporation of CO and SiO at the interface and the decrease of the wetting angle is due to the formation of the reaction layer at the interface.

Figure 1 .
Figure 1.Variations of the wetting angle with time for the alloy on the ceramic substrates.

Figure 2 .
Figure 2. Typical macrographs of the ceramic substrate and the corresponding alloy drop after wetting experiment.

Figure 3 .
Figure 3. (a)-(c) Interfacial microstructures of the alloy drop after wetting on ceramics with 10 wt.%, 30 wt.% and 50 wt.%ZrSiO4.(d)-(e) Al 2p spectrum and Si 2p spectrum for the grey product on the alloy bottom.By EDS analysis (table 2) at Points 1 and 2, the grey products are mainly composed of Al, O and a little of Si, Cr, Ni and the atom ratio of Al and O is about 2:3.To confirm the presence of Si in the grey product, the composition and elemental chemical state of the grey product are examined by XPS analysis.The full spectral scan analysis also indicates that the grey product comprises Al, Si and O. Using more precise scanning, the binding energies of Al 2p peak and Si 2p peak are determined to be 74.7 eV and 101.6 eV respectively (figure 3 (d) and (e)).Such values are well in accordance to the standard binding ) shows the Cr 2p spectrum examined on the lump products.The binding energies of Cr 2p (Cr 2p1/2 and Cr 2p3/2) peaks are 576.6 eV and 586.3 eV and they are well in accordance with the standard binding energies of Cr 2p peaks of Cr2O3, i.e., the lump phases are mainly composed of Cr2O3.

Figure 4 .
Figure 4. (a) Elemental mapping graphs on the ceramic surface.(b) Microstructure in the rectangle zone in a. (c) Cr 2p spectrum examined on the Cr-enriched phases near the wetted surface of the ceramic.

Figure 5 .
Figure 5. ΔG Θ -T diagram for the oxides of the alloy elements.

Table 1 .
Chemical composition of the ceramic powders.