Compositional Dependence of Phase Formation Mechanisms at the
Interface Between Titanium and Calcia-Stabilized Zirconia at 15501C
Yao-Wen Chang and Chien-Cheng Lin*
,wDepartment of Materials Science and Engineering, National Chiao Tung University, Hsinchu 30050, Taiwan ZrO2samples with various CaO contents were fabricated by hot
pressing, whereby CaO was dissolved by and/or reacted with ZrO2 to form a solid solution and/or CaZr4O9, respectively. After a reaction with Ti at 15501C for 6 h in argon, the inter-facial microstructures were characterized using X-ray diffrac-tion and analytical electron microscopy. Experimental results were very different from those found previously in the Y2O3– ZrO2 system. The 5 mol% CaO–ZrO2sample was relatively stable due to the formation of a thin TiO layer acting as a diffusion barrier phase. However,a-Ti(O), b0-Ti (Zr, O), and/or Ti2ZrO were found in 9 or 17 mol% CaO–ZrO2due to exten-sive interdiffusion of Ti, O, and Zr with a much thinner (b0-Ti1a-Ti) layer in 17 mol% CaO–ZrO2 than in 9 mol% CaO–ZrO2. Because CaO was hardly dissolved into Ti, it fully remained in the residual ZrO2, leading to the formation of spherical CaZrO3 in 9 mol% CaO–ZrO2 and columnar Ca-ZrO3in 17 mol% CaO–ZrO2. In the region far from the orig-inal interface, abundant intergranulara-Zr was formed in 5 or 9 mol% CaO–ZrO2. Scattereda-Zr and CaZrO3were found in 17 mol% CaO–ZrO2because a high concentration of extrinsic oxygen vacancies, which were created by the substitution of Ca12for Zr14, effectively retarded the reduction of zirconia.
I. Introduction
B
ECAUSEtitanium alloys have high tensile strength and tough-ness, light weight, and extraordinary corrosion resistance, they are widely applied to some parts of aircraft, medical devices, golf club heads, consumer electronics, etc. Titanium alloys are usually melted in a water-cooled copper crucible by the consumable electrode vacuum arc melting (VAR),1because the ceramic crucible used in the vacuum induction melting sig-nificantly reacts with titanium.2However, high cost of the equip-ment, scrape recycle, and long cycle time are some of drawbacks for the VAR process. Moreover, the reaction between the ceramic mold and titanium parts during invest casting inevita-bly results in a-casing and the resultant deterioration of me-chanical properties. Therefore, how to control the interfacial reactions between titanium and some ceramic materials is of great concerns.In the past several decades, extensive studies have been per-formed on the reactions between titanium and zirconia.3–10 Economos and Kingery3found that titanium penetrated along the grain boundaries of ZrO2 to form black oxygen-deficient
zirconia. Ruh6 indicated that up to approximately 10 mol% zirconia could be dissolved in titanium, while the residual zirconia became oxygen-deficient zirconia. Saha et al.10 also
revealed that the oxygen of zirconia was readily extracted and dissolved into the titanium to form a-Ti(O).
Recently, Lin et al.11–15have thoroughly investigated the phase formation mechanisms and microstructural evolution at the in-terface between titanium (or titanium alloys) and 3Y–ZrO2(or
various ratios of Y2O3/ZrO2) using analytical electron
micros-copy. a-Ti(O), b0-Ti (Zr, O), and/or Ti
2ZrO were formed near the
original interface due to the dissolution of ZrO2into Ti and vice
versa. Both lamellar orthorhombic Ti2ZrO and spherical
hexa-gonal Ti2ZrO were found in a-Ti(Zr, O) after reaction at 15501C. 11
The orientation relations of the acicular a-Ti and the b0-Ti were ½2110aTi==½001b0Ti and (0001)aTi//(100)b0Ti in
combi-nations of½2110aTi==½021b0Tiandð0001ÞaTi==ð112Þb0Ti,
re-spectively.13Lin and Lin12also found intergranular a-Zr, twinned t0-ZrO
2x, lenticular t-ZrO2x, and/or ordered c-ZrO2x in the
zirconia side, far from the interface between Ti and 3Y–ZrO2after
reaction at 15501C. Concerning the reaction of the Ti melt with various Y2O3/ZrO2samples at 17001C,15the incorporation of 430
vol% Y2O3in ZrO2could effectively suppress the reactions on the
Ti side, where only a very small amount of a-Ti and b0-Ti was found. Furthermore, Y2O3reprecipitated in the samples containing
30–70 vol% Y2O3because the solubility of Y2O3in Ti was very
low.
The yttria partially stabilized zirconia (Y-PSZ) has been con-sidered as one of the most popular industrial ceramic materials because of its good fracture toughness. Because Y2O3is much
more expensive than CaO or MgO, the zirconia crucibles used in casting industry are frequently made from CaO- or MgO-stabi-lized zirconia instead. In this study, powder mixtures with var-ious CaO/ZrO2 ratios were hot pressed and then allowed to
react with titanium at 15501C for 6 h in argon. The reaction layers at the interface were characterized using analytical scan-ning electron microscopy and analytical transmission electron microscopy. The effect of CaO on the interfacial reactions be-tween Ti and CaO/ZrO2samples is elucidated.
II. Experimental Procedures
Starting powders were zirconia (499.14 wt% ZrO21HfO2, with
HfO2accounting for approximately 1.84% of this total,o0.5
wt% SiO2,o0.11 wt% Y2O3,o0.08 wt% Na2O,o0.05 wt%
Al2O3,o0.05 wt% Fe2O3,o0.03 wt% CaO, o0.01 wt% MgO,
o0.005 wt% TiO2,o0.018 wt% U, o0.007 wt% Th; 8.54 mm
average particle size; Z-Tech LLC, Bow, NH), and calcia (499.9
wt% CaO, o0.034 wt% Sr, o0.02 wt% Mg, o0.02 wt% Na,
o0.01 wt% K, o0.005 wt% Ba, o0.005 wt% Pb, o0.003 wt% Fe, o0.002 wt% Cd, o0.001 wt% As; 10 mm average particle size; Ube
Material Industries Ltd., Chiba, Japan).
The CaO/ZrO2samples contained 5, 9, or 17 mol% CaO,
with the balance being ZrO2. The sample consisting of 5 mol%
CaO and 95 mol% ZrO2 was designated as 5C95Z, with the
same notation used for the other samples. Powder mixtures were dispersed in ethanol. The pH of the suspension was adjusted to 11 by adding NH4OH. The suspension was ultrasonically
vibrated for 10 min (Model XL-2020, Sonicator, Heat Systems Inc., Farmingdale, NY), dried in an oven at 1501C, ground with H.-J. Kleebe—contributing editor
Research supported by National Science Council of Taiwan under Contract No. NSC 96-2221-E-009-100.
*Member, The American Ceramic Society. w
Author to whom correspondence should be addressed. e-mail: chienlin@faculty. nctu.edu.tw
Manuscript No. 27356. Received January 8, 2010; approved May 30, 2010.
Journal
DOI: 10.1111/j.1551-2916.2010.03946.x r2010 The American Ceramic Society
an agate mortar and pestle, and then screened through an 80 mesh. Bulk specimens were fabricated by hot pressing in a graphite furnace at 1 atm argon (Model HP50-MTG-7010, Thermal Technology Inc., Santa Rosa, CA). As hot-pressed samples were blackened and were regarded as oxygen-deficient zirconia. They become stoichiometric after isothermal annealing at 13001C for 1 h.
The apparent densities of CaO/ZrO2powder mixtures were
measured using a gas pycnometer (Model MultiVolume Pycno-meter 1305, Micromeritics, Norcross, GA) with 99.99% pure helium. Bulk densities of hot-pressed CaO/ZrO2samples were
determined by the Archimedes method using water as an im-mersion medium. The relative densities of the hot-pressed sam-ples were calculated as follows: Relative density 5 (bulk density/ true density) 100%. For a nonporous powder, the apparent density approximates the true density and can be used as the reference point in calculating the relative density. The hot press conditions, compositions, relative densities, and designations of CaO/ZrO2samples are listed in Table I.
Commercially pure titanium plates (99.7% purity, Alfa Ae-sar, Ward Hill, MA) were brought to react with hot-pressed CaO/ZrO2samples at 15501C for 6 h in argon. First, bulk CaO/
ZrO2 samples and titanium plates were cut and machined to
dimensions of 10 mm 10 mm 5 mm. Their surfaces were ground and polished with a diamond paste and subsequently ultrasonically cleaned in acetone. One titanium specimen was inserted between two pieces of a CaO/ZrO2sample to produce a
sandwiched type, and the sandwich was then placed into the aforementioned graphite furnace. The furnace was initially pre-pared by first being pressurized to 5 MPa, then evacuated to 2 104torr, and filled with argon to 1 atmospheric pressure. This cycle of evacuation and purging was repeated at least three times. After the sample insertion, the furnace temperature was raised to 10001C at a heating rate of 301C/min, to 15501C at 251C/min, and then held at 15501C for 6 h. Thereafter, the tem-perature was lowered to 10001C at a cooling rate of 251C/min, and finally, the furnace was cooled down to room temperature. The phase identification of the as hot-pressed CaO/ZrO2
samples was performed using an X-ray diffractometer (XRD, Model MXP18, Mac Science, Yokohama, Japan). The operat-ing conditions of the XRD were CuKa radiation at 50 kV and 150 mA and a scanning rate of 21/min.
A scanning electron microscope (SEM, Model JSM 6500F, JEOL Ltd., Tokyo, Japan), which was equipped with an energy-dispersive X-ray spectrometer (EDS, Model ISIS 300, Oxford Instrument Inc., London, U.K.), was used for the microstruc-tural observation of the interfaces between Ti and various CaO/ ZrO2 samples. Cross-sectional SEM specimens were cut and
ground using standard procedures and finally polished using diamond pastes of 6, 3, and 1 mm in sequence.
The cross-sectional specimens of the interfaces between Ti and various CaO/ZrO2 samples for transmission electron
microscopy (TEM) were prepared by two different methods. Firstly, they were cut perpendicular to the interface and then polished, dimpled, and subsequently ion-beam-thinned using a precision ion-polishing system (PIPS, Model 691, Gatan, San Francisco, CA). The details of this traditional technique for preparing cross-sectional TEM specimens were described in a previous study.15Secondly, the TEM samples were acquired by an innovative technique. A specific location on a metallographic sample was ion-bombarded using a focused ion beam (FIB, Model Nova 200, FEI Co., Hillsboro, OR). The FIB operating parameters were adjusted so that the electron beam was 5 kV
from 98 pA to 1.6 nA, and the ion beam was 30 kV from 10 pA to 7 nA. A TEM specimen with a thickness of less than 100 nm was electron transparent. The final TEM specimen was approx-imately 12 mm 5 mm 0.05 mm in size.
The interfacial microstructures were then characterized using a TEM (Model JEM 2100, JEOL Ltd.) equipped with an EDS (Model ISIS 300, Oxford Instrument Inc.). Analyses of atomic configurations in various phases were performed using com-puter simulation software for crystallography (CaRIne Crystal-lography 3.1, Divergent S.A., Compiegne, France). Chemical quantitative analyses for various phases were conducted by the Cliff–Lorimer standardless technique.16 A conventional ZAF correction was operated using the LINK ISIS software.
III. Results and Discussion
(1) XRD Analyses
Figure 1 shows the XRD spectra of various hot-pressed CaO/ ZrO2 samples. These spectra are arranged in the sequence of
17C83Z, 9C91Z, and 5C95Z, from top to bottom. Monoclinic CaZr4O91719 and cubic ZrO2 were found in the hot-pressed
17C83Z, indicating that cubic ZrO2 was fully stabilized by
17 mol% CaO. While c-ZrO2, t-ZrO2, and m-ZrO2were found
in 9C91Z, the 5C95Z consisted of t-ZrO2and m-ZrO2. This
in-dicates that CaO was dissolved by or reacted with ZrO2to form
a solid solution and/or CaZr4O9, respectively, in hot-pressed
CaO/ZrO2samples.
Although Hellmann and Stubican19found Ca6Zr19O44after
sintering at 20001C for 4 h and subsequent prolonged annealing, Ca6Zr19O44was not found at all in this study. It was noted that
few CaZr4O9were found in 5C95Z and 9C91Z. Furthermore,
only a very small amount of CaZr4O9was distributed along the
grain boundaries of cubic zirconia in 17C83Z, as illustrated in Fig. 2. It was thus inferred that the equilibrium was not estab-lished after hot pressing. X-ray phases of these hot-pressed CaO/ ZrO2samples are summarized in Table I.
At the eutectoid temperature of CaO–ZrO2(11401C),19the
solid solution of ZrO2with more than 17 mol% CaO could
de-compose into cubic solid solution and two ordered phases (F1:
CaZr4O9,17–19and F2:Ca6Zr19O44).19,20 However, the rates of
decomposition to ordered phases are very slow at relatively low temperatures (o14001C).19
Table I. Designations, Compositions, Hot-Pressing Conditions, Relative Densities, and XRD Phases of Hot-Pressed CaO/ZrO2Samples
Specimens Composition (mol%) Hot-pressing conditions Relative densities (%) XRD phases
5C95Z 5% CaO195% ZrO2 16001C/30 min/1 atm Ar 98.4 t-ZrO2, m-ZrO2
9C91Z 9% CaO191% ZrO2 16001C/30 min/1 atm Ar 98.9 c-ZrO2, t-ZrO2, m-ZrO2
17C83Z 17% CaO183% ZrO2 16001C/30 min/1 atm Ar 98.0 c-ZrO2, CaZr4O9
Fig. 1. X-ray diffraction spectra of as hot-pressed CaO/ZrO2samples.
(2) SEM and TEM Analyses
Figures 2(a)–(c) display the backscattered electron images of the cross sections normal to the interfaces of Ti and various CaO/ ZrO2samples. Titanium is shown on the left of the micrograph,
while zirconia is on the right-hand side. The vertical arrows on the upper sides of individual figures indicate the original inter-faces of Ti and individual CaO/ZrO2samples. The original
in-terfaces were deliberately located according to the characteristic Ka X-ray maps of calcium (not shown), which was relatively
immobile compared with Zr, O, and Ti. The existence of pores in the ceramic side was attributed to the Kirkendall effect, as zir-conium and oxygen diffused to the titanium side much more rapidly than titanium diffused toward the zirconia side.
Figure 2(a) indicates that only a limited reaction occurred on the titanium side at the Ti/5C95Z interface, signifying that in-terfacial reactions were effectively suppressed. In the engineering respect of Ti castings, a well-controlled interfacial reaction be-tween Ti and 5C95Z can result in a thinner a-casing and thus in better mechanical properties. However, extensive reactions oc-curred at the Ti/9C91Z and Ti/17C83Z interfaces, as shown in Figs. 2(b) and (c). It was reported previously that needle-like a-Ti and some lamellar phases were usually found in the titanium side because of the interfacial reactions between Ti and
3Y–ZrO2.11,13,14 Even though the system became more stable
with the decreasing CaO, several reaction layers were found on the zirconia side after the interfacial reactions between Ti and various CaO/ZrO2samples. Microstructures of the reaction
lay-ers at the interface between Ti and various CaO/ZrO2samples
were characterized using SEM/EDS and TEM/EDS, with the results listed in Table II. Four dissimilarities of the interfacial microstructures were recognized after various CaO/ZrO2
samples reacted with Ti. They are described below.
(A) Two TiO and t-ZrO2Layers at the Ti/5C95Z Inter-face versus Four Complex Layers ofa-Ti, b0-Ti, and Ti2ZrO at the Ti/9C91Z and Ti/17C83Z Interfaces: The two outermost layers were TiO and t-ZrO2xlayers at the Ti/5C95Z interface.
In contrast, the three distinct outermost layers were composed of a-Ti, b0-Ti, and Ti
2ZrO at the Ti/9C91Z or Ti/17C83Z
inter-face. Figure 3(a) shows a bright-field image of a thin TiO reac-tion layer (about 2 mm thick) at the Ti/5C95Z interface. The arrow indicates the original interface between titanium (a-Ti) and zirconia (t-ZrO2x). Figures 3(b) and (c) show the selected
area diffraction patterns (SADPs) of TiO with [001] and [011] zone axes, respectively. It can be seen that TiO has a B1(NaCl) structure. The EDS spectrum in Fig. 3(d) shows that TiO dis-solved a small amount of Zr and was composed of 49.73 at.%
Fig. 2. The backscattered electron images of the interfaces between (a) Ti and 5C95Z, (b) Ti and 9C91Z, and (c) Ti and 17C83Z after reaction at 15501C for 6 h. The arrows indicate the original interfaces between Ti and CaO/ZrO2samples before reaction.
Ti, 49.26 at.% O, and 1.01 at.% Zr. This indicates that titanium could not dissolve Ca in the solid solution. Figure 3(e) shows the SADP of continuous t-ZrO2xphase with a zone axis of [111].
The EDS results indicate that t-ZrO2xwas composed of 4.96
at.% Ti, 2.72 at.% Ca, 32.62 at.% Zr, and 59.70 at.% O. Figure 4 shows the backscattered electron image of the reac-tion layers I and II at the Ti/17C83Z interface. Reacreac-tion layer I consisted of a-Ti (dark) and Ti2ZrO (bright). Reaction layer II
consisted of Ti2ZrO (bright), a-Ti (dark), and b0-Ti (bright). Lin
and Lin11reported that the Ti2ZrO lamellae were precipitated
from plate-like a-Ti by a eutectoid reaction during cooling. At high temperatures, the primary a-Ti dissolved a significant amount of oxygen and a relatively small amount of zirconium, forming metastable a-Ti, which could result in the precipitation of Ti2ZrO during cooling. As more Zr was dissolved in a-Ti,
b-Ti was formed, some of which was transformed into ortho-rhombic b0-Ti solid solution during cooling.
(B) A Much Thinner Reaction Layer III (b0-Ti1a-Ti) at the Ti/17C83Z Interface than at the Ti/9C91Z Inter-face: Figures 2(b) and (c) show a much thinner reaction layer III (b0-Ti1a-Ti) in 17C83Z than in 9C91Z, while their reaction layers (I1II) had approximately the same thickness. Reaction layer III, which dissolved a significant amount of zirconium (b stabilizer) and oxygen (a stabilizer), was composed of a-Ti1b-Ti with Zr and O in solid solution. Lin and Lin13indicated that the
acicular a-Ti was precipitated from the b0-Ti matrix.
As described in the following section, a stable 1:1 CaO ZrO2
compound (or CaZrO3) was formed in the ceramic side as Ti
and CaO/ZrO2samples were brought into contact at 15501C for
6 h. Assuming that all of CaO was consumed to form CaZrO3,
9C91Z, and 17C83Z had 82 and 66 mol% of excess ZrO2,
re-spectively. Even though this assumption is not fully correct (i.e., some CaO went into the solid solution of ZrO2), it was obvious
that more excess ZrO2existed in 9C91Z than in 17C83Z. As a
result, the thinner reaction layer III at the Ti/17C83Z interface was attributed to the fact that a smaller amount of excess ZrO2
and a relatively large amount of stable CaZrO3were formed in
the outermost ceramic region.
(C) Distinct Morphologies of CaZrO3in Reaction Layer
IV at the Ti/9C91Z and Ti/17C83Z Interfaces: Figure 5
demonstrates the microstructural variations of reaction layers IV at the Ti/9C91Z and Ti/17C83Z interfaces. Both reaction layers IV consisted of CaZrO3(dark) and b0-Ti (bright). The
CaZrO3 phase was sparse and isolated at the Ti/9C91Z
inter-face, while the CaZrO3phase was dense and interconnected at
the 17C83Z interface. As shown in Fig. 5(a), b0-Ti and a rela-tively small amount of spherical or worm-like CaZrO3existed in
reaction layers IV at the Ti/9C91Z interface. While the excess ZrO2was dissolved into Ti at the Ti/9C91Z interface, ZrO2
re-acted with CaO and formed the compound CaZrO3. Figure 5(b)
displays the b0-Ti and a relatively large amount of columnar or worm-like CaZrO3 coexisting in reaction layer IV at the
Table II. Thickness and Phases in Various Reaction Layers at the Interfaces of Ti and CaO/ZrO2Samples after Reaction at 15501C for 6 h
CaO content (mol%)
Reaction layers in the titanium side Reaction layers in the zirconia side
Layer Thickness (mm) Phases Layer Thickness (mm) Phases 5 I 2 TiO(1)w IV 8 t-ZrOð1Þ2x
V Very large t-ZrO2x, m-ZrO2x, a-Zr(1,4)
9 I 8 a-Ti1Ti2ZrO(1,2) IV 90 CaZrO3, b0-Ti(1,2,3)
II 56 a-Ti1Ti2ZrO1b0-Ti(1,2) V Very large c-ZrO2x, a-Zr(1,4)
III 38 b0-Ti1acicular a-Ti(1,2)
17 I 8 a-Ti1Ti2ZrO(1,2) IV 86 CaZrO3, b0-Ti(1,2,3)
II 56 a-Ti1Ti2ZrO1b0-Ti(1,2) V Very large c-ZrO2x, CaZrO3, a-Zr(1,5)
III 14 b0-Ti1acicular a-Ti(1,2)
w
Note that the formation mechanisms of various reaction layers are indicated by the superscript numbers in parenthesis:1
outward diffusion of O;2
outward diffusion of Zr;
3
inward diffusion of Ti;4
excluded from ZrO2x;5decomposition of CaZr4O9.
Fig. 3. (a) The bright-field image of reaction layers I and IV at the Ti/5C95Z interface after reaction at 15501C for 6 h. The arrow indicates the original interface between Ti and 5C95Z before reaction; (b) and (c) selected area diffraction patterns (SADPs) of TiO along the [001] and [011] zone axes, respectively; (d) an energy-dispersive spectrum of TiO; (e) an SADP of t-ZrO2xalong the zone axis [111].
Ti/17C83Z interface. The columnar CaZrO3was formed due to
the outward diffusion of O and Zr away from metastable CaO fully stabilized c-ZrO2x. This was usually called as a diffusion
zone, consisting of b0-Ti and columnar CaZrO
3. This result
in-dicates that CaZrO3was a stable phase and was not significantly
dissolved in Ti.
Figure 6(a) shows a bright-field image of reaction layer IV, consisting of b0-Ti and CaZrO
3, at the Ti/17C83Z interface. The
arrow in the upper right corner indicates the interface between reaction layer IV and reaction layer V. Columnar b0-Ti and Ca-ZrO3were aligned nearly perpendicular to the interface of
re-action layers IV and V. The crystal structures of both CaZrO3
and b0-Ti were identified to be orthorhombic from the superim-posed SADPs, as shown in Fig. 6(b). With the diffraction spots being indexed in Fig. 6(b), the orientation relationships of CaZrO3 and b0-Ti were thus recognized as follows:
½101CaZrO
3==½001b0Ti andð101ÞCaZrO3==ð100Þb0Ti. Figure 6(c)
shows the EDS spectrum of the CaZrO3, revealing that it
com-prised 19.93 at.% Ca, 20.61 at.% Zr, and 59.46 at.% O. Figure 6(d) displays the½101CaZrO3 or [001]b0Tistandard stereographic
projection corresponding to the SADPs illustrated in Fig. 6(b). It indicates that theð101Þ plane of CaZrO3 is parallel to the
(100) plane of b0-Ti. Lin and Lin21
reported that the reaction at 17501C/7 min between zirconia and titanium melt caused the formation of several CaZrO3 ovals embedded in a-Zr on the
zirconia side. No specific orientation relationship was identified in previous studies.
(D) Distinct a-Zr and/or CaZrO3 in Reaction Layer
V: Figure 7 displays the backscattered electron images of
reaction layer V on the zirconia side far away from the original interfaces. This layer could be termed as a reaction-affected zone. Figure 7(a) displays that dense a-Zr grains existed along the grain boundaries of c-ZrO2x in reaction layer V of Ti/
9C91Z. Dissolution did not play a significant role because the titanium was not detected by EDS in reaction layer V. A much denser a-Zr phase was found at the Ti/5C95Z interface, as shown in Fig. 2(a). In the other respect, very little a-Zr was found in reaction layer V of Ti/17C83Z, as shown in Fig. 7(b). In general, the amount of intergranular a-Zr decreases with the increasing CaO content.
Figures 8(a)–(b) show two bright-field images of reaction layer V on the zirconia side far away from the original interfaces. Figure 8(a) shows the bright-field image of m-ZrO2x,
(t1m)-ZrO2x, and a-Zr in the reaction layer V of Ti/5C95Z. As
5C95Z reacted with Ti, the metastable oxygen-deficient ZrO2x
was transformed into m-ZrO2x and a-Zr. Several martensite
lathes grew completely across the grain and became twinned m-ZrO2x(labeled as ‘‘m’’ in Fig. 8(a)) due to the stress
con-centrations at such a region.22Figure 8(b) shows a bright-field image of c-ZrO2x, CaZrO3, and a-Zr in reaction layer V of Ti/
17C83Z. The crystal structure of a-Zr was identified to be hex-agonal based upon the superimposed SADP, as shown in the upper right corner. In 17C83Z, ZrO2xwas fully stabilized as a
cubic phase because 17 mol% CaO was dissolved into ZrO2as a
solid solution. This c-ZrO2xwas so stable that no oxidation–
reduction was able to occur between zirconia and titanium in reaction layer V. In order to maintain charge neutrality, an ox-ygen vacancy was created in the crystal lattice of zirconia for every substitution of Ca21 for Zr41. As a result, there was 17 mol% of oxygen vacancies in 17C83Z if all of the CaO was consumed to form the solid solution of ZrO2. It was also
believed that the intergranular CaZrO3was found due to the
decomposition of intergranular CaZr4O9 at the Ti/17C83Z
interface.
Table II summarizes the formation of various reaction layers at the interface between titanium and various CaO/ZrO2
sam-ples after reaction at 15501C for 6 h. The formation mechanisms of a-Ti, b0-Ti, Ti2ZrO, ZrO2x, and a-Zr were described in detail
previously. However, the following phases were not found in previous studies: (1) TiO and continuous t-ZrO2xin reaction
layers I and IV of Ti/5C95Z; (2) CaZrO3in reaction layers IV
and/or V of Ti/9C91Z and Ti/17C83Z. As also indicated in Ta-ble II, the formation mechanisms of various reaction layers are described as follows: (1) reaction layers I, II, and III were formed because of the outward diffusion of O and/or Zr away from zirconia; (2) reaction layer IV was formed because of the inward diffusion of Ti into zirconia together with the outward diffusion of O and Zr away from zirconia; (3) reaction layer V was formed because of the decomposition of metastable oxygen-deficient ZrO2xand/or CaZr4O9as well as the outward
diffu-sion of O away from ZrO2.
Fig. 4. The backscattered electron image of reaction layers I and II at the Ti/17C83Z interface after reaction at 15501C for 6 h.
Fig. 5. The backscattered electron images of reaction layer IV at the interface between (a) Ti and 9C91Z, and (b) Ti and 17C83Z after reaction at 15501C for 6 h.
(3) Formation Mechanisms of TiO and Continuous t-ZrO2x at the Ti/5C95Z Interface
As mentioned above, TiO and continuous t-ZrO2xwere found
in reaction layers I and IV, respectively, at the Ti/5C95Z inter-face. The TiO reaction layer was formed due to the oxidation– reduction reaction between titanium and zirconia. Previous studies13,14reported that a-Ti(O) rather than a TiO layer was observed at the Ti/3Y–ZrO2interface after reaction at 15501C
for 6 h. It was inferred that 5 mol% CaO–ZrO2released oxygen
atoms much more easily than 3Y–ZrO2did because the Ca–O
bond is weaker than the Y–O bond. This thin TiO reaction layer functioned as a diffusion barrier phase23 because Ti and Zr
diffuse very slowly across the TiO reaction layer, resulting in a suppressed interfacial reaction between Ti and 5C95Z.
While oxygen-deficient zirconia was formed because of the oxidation–reduction between titanium and zirconia, the meta-stable zirconia with supersaturated oxygen vacancies had the tendency to decompose, leading to the formation of a-Zr along the grain boundaries. However, the dissolution of Ti12could stabilize the oxygen-deficient zirconia (or ZrO2x) due to the
creation of extra oxygen vacancies. Therefore, continuous t-ZrO2xrather than a-Zr was found in the ceramic outer layer
in 5C95Z.
It was noticeable that the TiO layer did not exist at the Ti/ 9C91Z and Ti/17C83Z interfaces. As mentioned above, almost all CaO and ZrO2went into a solid solution in as hot-pressed
9C91Z or 17C83Z. When these metastable solid solutions re-acted with Ti at 15501C, the excess Zr and O had a great ten-dency to be excluded from the solid solution, with CaZrO3being
left, and subsequently dissolved into Ti. Because TiO had a very limited solubility of Zr as indicated in the ternary phase diagram of Ti–O–Zr,13,14the extended dissolution of both Zr and O into Ti (as a-Ti, b0-Ti, and/or Ti2ZrO) obviously excluded
the possibility of the formation of the TiO layer at the Ti/9C91Z and Ti/17C83Z interfaces.
(4) Formation Mechanisms of CaZrO3and a-Zr at the Ti/9C91Z and Ti/17C83Z Interfaces
Figure 9 shows schematically the formation mechanisms of CaZrO3 and a-Zr at the Ti/9C91Z and Ti/17C83Z interfaces,
respectively. Upon heating at 15501C, the interfacial reactions resulted in the formation of a two-phase (b-Ti1CaZrO3) layer.
During cooling, cubic b-Ti was transformed into orthorhombic b0-Ti, where spherical or worm-like CaZrO3was formed.
Fig. 7. The backscattered electron images of reaction layer V in the zirconia side, far away from the original interface after reaction at 15501C for 6 h between (a) Ti and 9C91Z and (b) Ti and 17C83Z.
Fig. 6. (a) The bright-field image of reaction layers IV and V at the Ti/17C83Z interface after reaction at 15501C for 6 h; (b) selected area diffraction patterns of the CaZrO3and b0-Ti, Z 5 [101]CaZrO3//[001]b0Ti; (c) an energy-dispersive spectrum of CaZrO3; (d) the standard stereographic projection with [101]CaZrO3//[001]b0Ti.
Microstructural evolution in reaction layer IV at the Ti/ 9C91Z interface is schematically displayed in Fig. 9(a). As 9C91Z reacted with Ti at 15501C for 6 h, increasing amounts of O and Zr from the CaO–ZrO2solid solution were gradually
dissolved in titanium. Because CaO remained in the solid solu-tion due to the very limited solubility of CaO in Ti, the increase in the ratio of CaO to ZrO2gave rise to the formation of
Ca-ZrO3. The formation mechanisms of b0-Ti and CaZrO3in the
case of Ti/9C91Z in the reaction layer IV can be expressed as follows:
0:09CaOþ 0:91ZrO2! ðCa0:09Zr0:91ÞO1:91during hot pressing
(1)
x Ti þ ðCa0:09Zr0:91ÞO1:91! ðx Ti þ 0:82Zr þ 1:64OÞ
þ Ca0:09Zr0:09O0:27
! b0 TiðZr; OÞ þ 0:09CaZrO3
(2) Figure 9(b) displays a proposed model of microstructural evolution in reaction layer IV at the Ti/17C83Z interface. As 17C83Z reacted with Ti at 15501C for 6 h, titanium diffused into this region and dissolved a relatively small amount of O and Zr, which diffused out of the metastable ZrO2phase with
supersat-urated 17 mol% CaO in solid solution, leading to an appearance of a so-called diffusion zone. As the solubility of Ca in Ti was quite limited, Ca was retained in the residual ZrO2, causing the
formation of columnar CaZrO3. This diffusion zone consisted of
a two-phase (b-Ti1CaZrO3) layer and was featured by the
col-umnar CaZrO3parallel to the diffusion direction. The formation
mechanisms of b0-Ti and CaZrO3in the case of Ti/17C83Z in the
reaction layer IV can be expressed as follows:
0:17CaOþ 0:83ZrO2! ðCa0:17Zr0:83ÞO1:83during hot-pressing
(3)
x Ti þ ðCa0:17Zr0:83ÞO1:83! ðx Ti þ 0:66Zr þ 1:32OÞ
þ Ca0:17Zr0:17O0:51
! b0 TiðZr; OÞ þ 0:17CaZrO3
(4)
Moreover, as O and Zr diffused out of CaZr4O9 formed
previously during hot pressing, the ratio of CaO/ZrO2changed
from 1:4 to 1:1. Consequently, CaZr4O9 was transformed
into CaZrO3. This reaction mechanism can be expressed as
follows:
x Ti þ CaZr4O9! ðx Ti þ 3Zr þ 6OÞ þ CaZrO3 ! b0 TiðZr; OÞ þ CaZrO
3
(5)
It is worth mentioning that a columnar CaZrO3phase was
precipitated in 17C83Z from the CaO–ZrO2solid solution. As
Zr and O of the CaO–ZrO2solid solutions in 17C83Z diffused
outwards or were selectively dissolved into Ti, the matrix be-came enriched in CaO. There were two possibilities that CaZrO3
were formed. For example, it might have entered into the two-phase (CSS1CaZrO3) region18and then would decompose into
c-ZrO2 and CaZrO3 at 15501C through the nucleation and
growth mechanism. However, a diffusion zone was generally featured by columnar precipitates, which grew along the direc-tion parallel to that of diffusion. For instance, Goward and Boone24 observed a diffusion zone in the aluminized nickel-based superalloys. The morphology of CaZrO3 at the Ti/
17C83Z interface implied that it was a diffusion zone and thus excluded the possibility that CaZrO3 was formed through a
decomposition process. In other words, the diffusion of Zr and O atoms out of the c-ZrO2with 17 mol% CaO in solid solution
was the primary formation mechanism of the columnar CaZrO3
in the diffusion zone of 17C83Z.
Figure 9(c) illustrates the formation of intergranular a-Zr in reaction layer V of Ti/9C91Z. The oxygdeficient zirconia en-tered into a two-phase (a-Zr1ZrO2x) region25because of the
oxidation–reduction between Ti and zirconia at such a high temperature as 15501C. It was obvious that the oxygen-deficient zirconia was metastable because of the supersaturation of oxy-gen vacancies. The exsolution of Zr from ZrO2xto the grain
boundaries gave rise to the formation of intergranular a-Zr with oxygen in the solid solution.
Figure 9(d) illustrates that less intergranular a-Zr(O) was formed in 17C83Z than in 9C91Z or 5C95Z. The CaZr4O9in
reaction layer V of Ti/17C83Z could be decomposed into Ca-ZrO3and a-Zr(O), with O partially dissolved into b0-Ti in
reac-tion layer IV. The decomposireac-tion of CaZr4O9could be expressed
in terms of the following equation:
CaZr4O9! CaZrO3þ 3a Zrðwith O in solid solutionÞ þ 6Oðpartially dissolved in b0TiÞ
(6)
It is also believed that the enthalpy change DH for an extra oxygen vacancy significantly increased with the increasing va-cancy concentration. While the vava-cancy concentration increased with the increasing concentration of CaO in the solid solution, the tendency of extraction of oxygen from ZrO2by Ti also
di-minished with the increasing CaO content. It was thus possible that ZrO2xdid not become metastable or oxygen-vacancy
su-persaturated, so that very little a-Zr was found at the Ti/17C83Z interface.
(5) Thermodynamic Calculation for the Formation of CaZrO3at the Ti/17C83Z Interface
The metastable solid solution will be decomposed when it is in contact with Ti. The thermodynamic values for the decomposi-tion of 17C83Z are calculated as an example. In addidecomposi-tion to four end solid solution phases, there are four intermediate stoichio-metric compounds, i.e., o-CaZrO3, c-CaZrO3, CaZr4O9(F1),
and Ca6Zr19O44(F2), in the CaO–ZrO2 binary system. 26
The thermodynamic values acquired were mainly focused on the in-termediate compounds CaZrO3in previous studies,27–31while
those for the other two intermediate phases (F1 and F2) are relatively insufficient. It was reported that the intermediate com-Fig. 8. Bright-field images of reaction layer V in the zirconia side far
away from the original interface between (a) Ti and 5C95Z and (b) Ti and 17C83Z after reaction at 15501C for 6 h. Inset in the upper right hand corner of Fig. 9(b) is a selected area diffraction pattern of a-Zr along the [101] zone axis.
pounds F1 and F2 are stable at the temperatures 1408.5–1507.1 K and 1418.9–1625.6 K, respectively, and c-CaZrO3 was
stable above 2273 K32according to the CaO–ZrO2phase
dia-gram proposed by Wang et al.26Based on these arguments and experimental results, only o-CaZrO3 is considered among the
four intermediate compounds.
Because of lack of the experimental information for solid so-lution phases with a wide range of compositions, it is assumed that the decomposition is governed by the following reaction for simplicity. In other words, the 17C83Z solid solution was first decomposed into CaZrO3and ZrO2, and then the reduction and
dissolution occurred between Ti and ZrO2
0:66Tiþ Ca0:17Zr0:83O1:83
! 0:66Ti þ ð0:66ZrO2þ Ca0:17Zr0:17O0:51Þ ! 0:66a TiOxþ 0:66ZrO2xþ 0:17CaZrO3DG1
(7)
During hot pressing, ZrO2and CaO are mutually dissolved as
a homogeneous solid solution phase for the sake of kinetics, which can be expressed as follows:
0:17CaOþ 0:83ZrO2! Ca0:17Zr0:83O1:83DG2 (8)
Using solid electrolyte galvanic cells, the standard Gibbs free energies of formation of CaZrO3 from CaO and ZrO2 at a
different range of temperature were determined. The formation of CaZrO3can be expressed by the following equation:
CaOþ ZrO2! CaZrO3DG3 (9)
where DG3525.2(70.15)17.58(70.085) 103T (kJ/mol)
and T is the absolute temperature.28 When T 5 15501C
(1823 K), DG3557.25 (kJ/mol), the reduction and dissolution
between Ti and ZrO2 can be expressed by the following
equation:
Tiþ ZrO2! a TiOxþ ZrO2xDG4 (10)
Therefore, DG15DG210.17DG310.66DG4. The Gibbs
free-energy DG4can be estimated by taking into consideration the
following equivalent defect reaction: OXOðin zirconiaÞ! V
Oðin zirconiaÞþ 2e0ðin zirconiaÞ þ O00i ðin titaniumÞþ 2hðin titaniumÞDG5
(11) Fig. 9. Schematic diagrams showing the formation mechanisms of (a) CaZrO3in reaction layer IV at the Ti/9C91Z interface; (b) CaZrO3in reaction
layer IV at the Ti/17C83Z interface; (c) a-Zr in reaction layer V in 9C91Z; (d) a-Zr in reaction layer V in 17C83Z at various intervals (t0ot1ot2ot3). The
arrows indicate the diffusion directions for the individual atoms of Ti, Zr, and O assuming that Ti and ZrO2are on the left- and right-hand sides,
respectively.
This equivalent reaction is the combination of dissolution and reduction. Based upon the calculation by Lin and Lin,12 DG55DGred1DGdiss50.54379.30 5 378.76 (kJ/mol) at
15501C. Because DG45 xDG5, DG15DG210.17DG310.66
xDG5. Substituting DG3557.25 kJ/mol and DG55378.76
kJ/mol, DG15DG29.73249.98 x (kJ/mol). Because DG2is a
positive value,33DG
1must be negative.
Based upon the discussion mentioned above, the formation of CaZrO3and ZrO2xinduced by the interfacial reaction between
Ti and 17C83Z is thermodynamically favorable. This calcula-tion is in good agreement with the present experimental results, indicating the formation of CaZrO3, a-Ti(O), and
oxygen-deficient ZrO2x in layer V. It is believed that extensive
diffusion between Ti, Zr, and O is more thermodynamically favorable and results in b0-Ti (O, Zr) and CaZrO
3in layer IV at
the Ti/17C83Z interface.
IV. Conclusions
1. The phase formation mechanisms at the interface be-tween Ti and ZrO2strongly depended upon the types of
stabi-lizers as well as their amounts. This study shows the promising application of CaO-stabilized-ZrO2as molding ceramics for Ti
and its alloys while being compared with previous studies con-ducted on the Ti/Y2O3–ZrO2interfaces.
2. The 5C95Z shows the best performance due to the for-mation of a thin layer of TiO, which behaves as a diffusion bar-rier between Ti and ZrO2. In other words, the incorporation of 5
mol% CaO into ZrO2could effectively suppress the interfacial
reactions between Ti and 5C95Z. However, more complex layers consisting of a-Ti, b0-Ti, and Ti
2ZrO were found at interfaces
such as Ti/9C91Z and Ti/17C83Z. This indicates that the zir-conia with 5 mol% CaO in solid solution is one of the most potential candidates for crucible and mold materials in the Ti-casting industry.
3. Both b0-Ti and spherical CaZrO3were found at the Ti/
9C91Z interface after reaction at 15501C for 6 h because of ex-tensive dissolution of O and Zr together with a very limited sol-ubility of Ca in Ti.
4. The outward diffusion of Zr and O, which were subse-quently dissolved into Ti, gave rise to a diffusion zone featuring columnar CaZrO3in the matrix of b0-Ti after the reaction
be-tween Ti and 17C83Z. This implies that CaZrO3 is a stable
phase and thus as a potential refractory when it is taken into contact with titanium alloys at high temperatures.
5. For 5C95Z and 9C91Z, a large amount of a-Zr grains were excluded from metastable oxygen-deficient ZrO2xon the
zirconia side far away from the original interface. In this reac-tion-affected zone, oxygen-deficient zirconia was partially stabi-lized as tetragonal in 5C95Z and fully stabistabi-lized as cubic in 9C91Z.
6. A very small amount of a-Zr and CaZrO3was found on
grain boundaries of zirconia on the ceramic side far away from the original interface between Ti and 17C83Z. The amount of intergranular a-Zr decreased with the increasing CaO content.
References
1
G. Broihanne and J. Bannister, ‘‘Cold Crucible Melting Gets A New Spin,’’ Mater. World, 8 [8] 21–3 (2000).
2
K. Sakamoto, K. Yashikawa, T. Fujisawa, and T. Onaye, ‘‘Changes In Oxygen Contents of Titanium Aluminides by Vacuum Induction, Cold Crucible Induction and Electron Beam Melting,’’ Iron Steel Inst. Jpn. Int., 32 [5] 616–24 (1992).
3
G. Economos and W. D. Kingery, ‘‘Metal–Ceramic Interactions: II, Metal Oxide Interfacial Reactions at Elevated Temperatures,’’ J. Am. Ceram. Soc., 36 [12] 403–9 (1953).
4
B. C. Weber, H. J. Garrett, F. A. Mauer, and M. A. Schwartz, ‘‘Observations on the Stabilization of Zirconia,’’ J. Am. Ceram. Soc., 39 [6] 197–07 (1956).
5
B. C. Weber, W. M. Thompson, H. O. Bielstein, and M. A. Schwartz, ‘‘Ce-ramic Crucible for Melting Titanium,’’ J. Am. Ceram. Soc., 40 [11] 363–73 (1957).
6
R. Ruh, ‘‘Reaction of Zirconia and Titanium at Elevated Temperatures,’’ J. Am. Ceram. Soc., 46 [7] 301–6 (1963).
7
A. I. Kahveci and G. E. Welsch, ‘‘Effect of Oxygen on the Hardness and Al-pha/Beta Phase Ration of Ti–6Al–4V Alloy,’’ Scr. Metall., 20 [9] 1287–90 (1986).
8
R. L. Saha and K. T. Jacob, ‘‘Casting of Titanium and It’s Alloy,’’ Def. Sci., 36 [2] 121–41 (1986).
9
K. I. Suzuki, S. Watakabe, and K. Nishikawa, ‘‘Stability of Refractory Oxides for Mold Material of Ti–6Al–4V Alloy Precision Casting,’’ J. Jpn. Inst. Met., 60 [8] 734–43 (1996).
10
R. L. Saha, T. K. Nandy, R. D. K. Misra, and K. T. Jacob, ‘‘On the Eval-uation of Stability of Rare Earth Oxides Face Coats for Investment Casting of Titanium,’’ Metall. Trans., 21B [6] 559–66 (1990).
11
K. L. Lin and C. C. Lin, ‘‘Ti2ZrO Phases Formed in the Titanium and
Zir-conia Interface after Reaction at 15501C,’’ J. Am. Ceram. Soc., 88 [5] 1268–72 (2005).
12
K. L. Lin and C. C. Lin, ‘‘Zirconia-Related Phases in the Zirconia/Titanium Diffusion Couple after Annealing at 11001 to 15501C,’’ J. Am. Ceram. Soc., 88 [10] 2928–34 (2005).
13
K. L. Lin and C. C. Lin, ‘‘Microstructural Evolution and Formation Mech-anism of the Interface Between Titanium and Zirconia Annealed at 15501C,’’ J. Am. Ceram. Soc., 89 [4] 1400–8 (2006).
14
K. L. Lin and C. C. Lin, ‘‘Effects of Annealing Temperature on Microstruc-tural Development at the Interface Between Zirconia and Titanium,’’ J. Am. Ceram. Soc., 90 [3] 893–9 (2007).
15
C. C. Lin, Y. W. Chang, and K. L. Lin, ‘‘Effect of Yttria on Interfacial Re-actions Between Titanium Melt and Hot-Pressed Yttria/Zirconia Composites at 17001C,’’ J. Am. Ceram. Soc., 91 [7] 2321–7 (2008).
16
G. Cliff and G. W. Lorimer, ‘‘The Quantitative Analysis of Thin Specimens,’’ J. Microsc., 130 [3] 203–7 (1975).
17
R. C. Garvie, ‘‘The Cubic Field in the System ZrO2–CaO,’’ J. Am. Ceram.
Soc., 51 [10] 553–6 (1968).
18
V. S. Stubican and S. P. Ray, ‘‘Phase Equilibria and Ordering in the System ZrO2–CaO,’’ J. Am. Ceram. Soc., 60 [11–12] 534–7 (1977).
19
J. R. Hellmann and V. S. Stubican, ‘‘Stable and Metastable Phase Relations in the System ZrO2–CaO,’’ J. Am. Ceram. Soc., 66 [4] 260–4 (1983).
20
J. R. Hellmann and V. S. Stubican, ‘‘The Existence and Stability of Ca6Zr19O44Compound in the System ZrO2–CaO,’’ Mater. Res. Bull., 17 [4]
459–65 (1982).
21
K. F. Lin and C. C. Lin, ‘‘Interfacial Reactions between Zirconia and Tita-nium,’’ Scr. Metall., 39 [10] 1333–8 (1998).
22
A. H. Heuer and M. Ruhle, ‘‘On the Nucleation of the Martensitic Transfor-mation in Zirconia (ZrO2),’’ Acta Metall., 33 [12] 2101–12 (1985).
23
H. Hao, Y. Wang, Z. Jin, and X. Wang, ‘‘Joining of Zirconia Ceramic to Stainless Steel and to Itself Using Ag57Cu38Ti5Filler Metal,’’ J. Am. Ceram. Soc.,
78 [8] 2157–60 (1995).
24
G. W. Goward and D. H. Boone, ‘‘Mechanisms of Formation of Diffusion Aluminide Coatings on Nickel-Base Superalloys,’’ Oxid. Met., 3 [5] 475–95 (1971).
25
R. F. Domagala, S. R. Lyon, and R. Ruh, ‘‘The Pseudobinary Ti–ZrO2,’’
J. Am. Ceram. Soc., 56 [11] 584–7 (1973).
26
K. Wang, C. H. Li, Y. H. Gao, X. G. Lu, and W. Z. Ding, ‘‘Thermodynamic Reassessment of ZrO2–CaO System,’’ J. Am. Ceram. Soc., 92 [5] 1098–104 (2009).
27
G. Roga and M. Potoczek-Dudek, ‘‘Determination of the Standard Molar Gibbs Energy of Formation of Calcium Zirconate by a Galvanic Cell Involving Calcium-Ion Conducting Solid Electrolyte,’’ J. Chem. Thermodyn., 33, 77–82 (2001).
28
J. Tanabe and K. Nagata, ‘‘Use of Solid-Electrolyte Galvanic Cells to Deter-mine the Activity of CaO in the CaO–ZrO2System and the Standard Gibbs Free
Energies of Formation of CaZrO3from CaO and ZrO2,’’ Metall. Mater. Trans. B,
27B [8] 658–62 (1996).
29
K. T. Jacob and Y. Waseda, ‘‘Gibbs Energy of Formation of Orthorhombic CaZrO3,’’ Thermochim. Acta, 239, 233–41 (1994).
30
B. R. Brown and K. O. Bennington, ‘‘Thermodynamic Properties of Calcium Zirconate,’’ Thermochim. Acta, 106, 183–90 (1986).
31
V. A. Levitski, P. B. Narchuk, Ju. Hekimov, and J. I. Gerassimov, ‘‘Thermo-dynamic Study of Some Solid Solutions in CaO–ZrO2System by EMF Method,’’
Solid State Chem., 20, 119–25 (1977).
32
E. R. Andrievskaya, A. V. Shevchenko, L. M. Lopato, and Z. A. Zaitseva, ‘‘Interactions in the System CaZrO3–Y2O3,’’ Izv. Akad. Nauk SSSR Neorg.
Mater., 25 [9] 1338–40 (1989).
33
K. S. Mohandas and D. J. Fray, ‘‘Electrochemical Deoxidation of Solid Zir-conium Dioxide in Molten Calcium Chloride,’’ Metall. Mater. Trans. B, 40B [10]
685–99 (2009). &
