**3. Mechanical properties of Ag–20Pd–14.5Cu–12Au alloy**

#### **3.1. Unique hardening behavior**

**RQ**

**15 20 25 Cu (at%)**

**Figure 2.** Effect of Cu concentration (at%) on hardness (Hv) of Ag-30.32Pd-9.66Cu-5.04Au (A), Ag-25.4Pd-12.82Cu-9.96Au (B), Ag-28Pd-9.12Cu-12.04Au (C), and Ag-25.2Pd-9.88Cu-20Au (D) alloys. SQ and RQ indi‐

**Figure 3.** (a) Diffraction pattern, (b) TEM image, and (c) HRTEM image of Ag-Pd-Cu-Au alloy subjected to ST at 1023 K

for 3.6 ks followed by aging treatment at 623 K for 1.8 ks.

**CA D B**

**473 K 523 K 548 K 573 K SQ**

**300**

518 Advances in Biomaterials Science and Biomedical Applications

**200**

**Hv**

**100**

cate specimens subjected to slow quenching and rapid quenching after ST, respectively.

#### *3.1.1. Precipitation of β′ phase after high temperature solution treatment*

Commercial Ag–20Pd–14.5Cu–12Au dental alloy (as-received) is fabricated by a rolling process. The commercial Ag–20Pd–14.5Cu–12Au alloy has a multiphase microstructure (α1, α2, β). It is well known that the Ag–Pd–Cu–Au alloy exhibits age hardening behavior as described in the section 2, but the drastic increase in the hardness of Ag–20Pd–14.5Cu–12Au alloy through ST at temperatures over 1073 K subjected to water quenching has been newly reported [10]. This unique hardening behavior has been explained in terms of two hardening mechanisms: (1) solid solution hardening mechanism in which the alloying elements are dissolved into the matrix (α phase) during ST, and (2) precipitation hardening mechanism, in which L10-type ordered phases are precipitated during the quenching process after ST. Recent studies on the hardening mechanism of Ag–20Pd–14.5Cu–12Au alloy have revealed that the precipitation hardening mechanism is the probable mechanism for the unique hardening behavior exhibited [11, 26].

Conventionally, dentists have employed AT for the hardening of Ag–20Pd–14.5Cu–12Au alloy as mentioned above. Figure 4 [10] shows the effects of heat treatment (AT and ST) temperatures on the mechanical properties (tensile strength, elongation, and hardness (HRA)) of this alloy. The tensile strength and hardness increase until the temperature reaches 673 K, and then decrease for up to 923 K due to the removal of strain from the alloy. At temperatures higher than 923 K, this alloy exhibits a unique hardening behavior. Under AT, the tensile strength and hardness of this alloy drastically increase and the elongation decreases after treatment at 673 K. On the other hand, under ST, the tensile strength and hardness still increase but the elongation does not decrease after treatment at 1073 K. Since high temperature ST is very useful for the hardening of this alloy, this treatment will be widely adopted in the future.

The relationship between the microstructural changes in the L10-type ordered β′ phase and the hardening behavior in the solutionized alloys was investigated by changing the cooling rate. Figure 5 shows a schematic drawing of various cooling rates employed after ST [11]. The Vickers hardness of the as-received alloy and of the alloys subjected to ST followed by water quenching (WQ), air cooling (AC), and cooling in a furnace (FC) are shown in Fig. 6 [11]. ST subjected to WQ and AC leads to a significant and slight increase in the hardness of the alloy,

**Figure 4.** Effect of heat treatment temperature on mechanical properties of Ag–20Pd–14.5Cu–12Au alloy.

respectively, while ST subjected to FC decreases the hardness of the alloy. Thus, the hardness tends to decrease with a decrease in the cooling rate after ST.

Figures 7 and 8 show the microstructures of the as-received and solutionized alloy, respec‐ tively, as measured by backscattered electron (BSE) and energy dispersive X-ray spectroscopy (EDX) analysis [13]. The as-received alloy composed of a Cu-rich α<sup>1</sup> phase, an Ag-rich α2 phase, and Cu–Pd intermetallic β phase, while the solutionized alloy composed of α2 and β phases. After ST, the α1 phase dissolved into the Ag-rich α2 phase and the β phase remained in the matrix. The β′ phase precipitated in the matrix could not be observed by BSE, but could be observed by TEM.

**Figure 5.** Schematic drawing of heat treatments with various cooling rates after ST. ST1123K-WQ, ST1123K-AC, and ST1123K-FC indicate specimens subjected to ST at 1123 K followed by water, air, and furnace cooling, respectively.

**Figure 6.** Vickers hardness of as-received, ST1123K-WQ, ST1123K-AC, and ST1123K-FC.

respectively, while ST subjected to FC decreases the hardness of the alloy. Thus, the hardness

**Figure 4.** Effect of heat treatment temperature on mechanical properties of Ag–20Pd–14.5Cu–12Au alloy.

Figures 7 and 8 show the microstructures of the as-received and solutionized alloy, respec‐ tively, as measured by backscattered electron (BSE) and energy dispersive X-ray spectroscopy (EDX) analysis [13]. The as-received alloy composed of a Cu-rich α<sup>1</sup> phase, an Ag-rich α2 phase, and Cu–Pd intermetallic β phase, while the solutionized alloy composed of α2 and β phases. After ST, the α1 phase dissolved into the Ag-rich α2 phase and the β phase remained in the matrix. The β′ phase precipitated in the matrix could not be observed by BSE, but could be

**Figure 5.** Schematic drawing of heat treatments with various cooling rates after ST. ST1123K-WQ, ST1123K-AC, and ST1123K-

FC indicate specimens subjected to ST at 1123 K followed by water, air, and furnace cooling, respectively.

tends to decrease with a decrease in the cooling rate after ST.

520 Advances in Biomaterials Science and Biomedical Applications

Specimen

observed by TEM.

**Figure 7.** BSE image and elemental mappings obtained by EDX of as-received Ag–20Pd–14.5Cu–12Au alloy.

**Figure 8.** BSE image and elemental mappings obtained by EDX of Ag–20Pd–14.5Cu–12Au alloy subjected to ST at 1123 K for 3.6 ks.

The selected area diffraction patterns (SADP) and key diagrams obtained from TEM for the alloy subjected to ST and WQ, given in Fig. 9, show that three variants of the L10-type ordered β′ phase are present when the beam direction is parallel to [100] crystal direction: one variant (P) has a c-axis parallel to the electron beam; the other two variants (N1, N2) have c-axes normal to the electron beam [11]. The SADPs indicate that an L10-type ordered β′ phase is densely precipitated in the matrix. The dark field images using a (001)N1 and (001)N2 reflections (Fig. 9 (d) and (e)) show that the β′ phase is 2–6 nm wide and 20–60 nm long. In the alloy subjected to ST and AC, the β′ phase is 3–13 nm wide and 50–400 nm long, and in the alloy subjected to ST and FC the β′ phase is 3–25 nm wide and 70–700 nm long, as confirmed by the dark field images using a (001)N1 reflection. In the alloys subjected to ST followed by AC and FC, the β ′ phase is less densely precipitated than the alloy subjected to ST and WQ. These dark field images indicate that the size and number of the β′ phase vary with the cooling rates after ST. In the solutionized alloy subjected to WQ, the size of the β′ phase is small and the amount of the β′ phase is large. The size of the β′ phase decreased and the amount of the β′ phase increased with an increase in the cooling rate after ST. The driving force behind the nucleation of the β′ phase in the solutionized alloy subjected to WQ is stronger than that in the solution‐ ized alloy subjected to AC and FC, because the degree of undercooling in the solutionized alloy subjected to WQ is larger than that in the solutionized alloy subjected to AC and FC. The extent of nucleation of the β′ phase in the solutionized alloy subjected to WQ is larger than that in the solutionized alloy subjected to AC and FC. The β′ phase in the solutionized alloys subjected to AC and FC grew more coarsely than that in the solutionized alloy subjected to WQ, because the diffusion occurred more easily owing to the slow cooling process in the solutionized alloys subjected to AC and FC. Therefore, it is likely that the β′ phase is formed during the cooling process and grows by diffusion.

**Figure 9.** TEM micrographs of ST1123K-WQ: (a) a selected area diffraction pattern, (b) a key diagram, (c) a bright-field image, and (d) and (e) dark-field images using (001)N1 and (001)N2 reflections, respectively. The beam direction is par‐ allel to [100] crystal direction.

**Figure 8.** BSE image and elemental mappings obtained by EDX of Ag–20Pd–14.5Cu–12Au alloy subjected to ST at

The selected area diffraction patterns (SADP) and key diagrams obtained from TEM for the alloy subjected to ST and WQ, given in Fig. 9, show that three variants of the L10-type ordered β′ phase are present when the beam direction is parallel to [100] crystal direction: one variant (P) has a c-axis parallel to the electron beam; the other two variants (N1, N2) have c-axes normal to the electron beam [11]. The SADPs indicate that an L10-type ordered β′ phase is densely precipitated in the matrix. The dark field images using a (001)N1 and (001)N2 reflections (Fig. 9 (d) and (e)) show that the β′ phase is 2–6 nm wide and 20–60 nm long. In the alloy subjected to ST and AC, the β′ phase is 3–13 nm wide and 50–400 nm long, and in the alloy subjected to ST and FC the β′ phase is 3–25 nm wide and 70–700 nm long, as confirmed by the dark field images using a (001)N1 reflection. In the alloys subjected to ST followed by AC and FC, the β ′ phase is less densely precipitated than the alloy subjected to ST and WQ. These dark field images indicate that the size and number of the β′ phase vary with the cooling rates after ST. In the solutionized alloy subjected to WQ, the size of the β′ phase is small and the amount of the β′ phase is large. The size of the β′ phase decreased and the amount of the β′ phase increased with an increase in the cooling rate after ST. The driving force behind the nucleation of the β′ phase in the solutionized alloy subjected to WQ is stronger than that in the solution‐ ized alloy subjected to AC and FC, because the degree of undercooling in the solutionized alloy subjected to WQ is larger than that in the solutionized alloy subjected to AC and FC. The extent of nucleation of the β′ phase in the solutionized alloy subjected to WQ is larger than that in the solutionized alloy subjected to AC and FC. The β′ phase in the solutionized alloys subjected to AC and FC grew more coarsely than that in the solutionized alloy subjected to WQ, because the diffusion occurred more easily owing to the slow cooling process in the solutionized alloys subjected to AC and FC. Therefore, it is likely that the β′ phase is formed during the cooling

1123 K for 3.6 ks.

522 Advances in Biomaterials Science and Biomedical Applications

process and grows by diffusion.

**Figure 10.** HRTEM micrographs of ST1123K-WQ: (a) a selected area diffraction pattern, (b) a key diagram, and (c) a bright-field image. The beam direction is parallel to [100].

Figure 10 shows the SADP, key diagram, and HRTEM bright field image of ST1123K-WQ [11]. The SADP and the key diagram in Fig. 10 (a) and (b) indicate that the FCC α phase and the three variants of the L10-type ordered β′ phase are superimposed. The L10-type ordered β′ phase is precipitated in the FCC matrix after ST. The streaks on the reflection spots of the L10-type ordered β′ phase indicate that the shape of the β′ phase is similar to a thin plate. The HRTEM bright field image in Fig. 10 (c) shows that several nanometer-sized thick plateshaped β′ phase with two variants (N1, N2), whose c-axes are normal to the electron beam, are precipitated in the matrix.

It is well known that it is difficult to make a single phase by quenching after high temperature ST; the supersaturated vacancies help the solutes to diffuse more easily and also help to form more clusters, G–P zones, and metastable phases during quenching after high temperature ST. The precipitated β′ phase of ST1123K-WQ shown in Fig.9 (d), (e) and Fig. 10 (c) is like a thin plate with a nanometer-scale thickness. These images also suggest that the formation of the β′ phase is diffusion controlled. Generally, the formation of precipitates can be considered to be order– disorder transition, diffusionless transformation (martensitic transformation), or diffusional transformation. In the as-solutionized Ag–Pd–Cu–Au alloy used in this case, the dependence of microstructural changes in the precipitated β′ phase on both the cooling rate after ST and the ST temperature show that the precipitated L10-type ordered β′ phase is formed during the cooling process and that the growth of the β′ phase is influenced by the diffusion process. The hardness increases with an increase in the cooling rate after ST, and consequently, the hard‐ ness of ST1123K-WQ increases significantly by quenching after ST (Fig. 6). The fine β′ phase in ST1123K-WQ is densely precipitated in the matrix. The hardness of ST1123K-AC increased only slightly, while the hardness of ST1123K-FC decreases, as only coarse β′ phases are precipitated in the matrix of ST1123K-AC and ST1123K-FC. Thus, the increase in hardness may be strongly affected by the presence of finely precipitated β′ phase. The coherent precipitation of β′ phases with long andshort axes of around100 nmand10 nm,respectively, also occuredduring ST, although the amount of β′ phase decreases with an increase in the ST time. The effect of solid solution hardening in the α, α1, and α2 phases is lower than that exerted by the precipitation harden‐ ing due to β′ phases.

#### *3.1.2. Hardening behavior of Ag–20Pd–12Au–14.5Cu alloy fabricated by liquid rapid solidification*

An Ag–20Pd–14.5Cu–12Au alloy with a single α phase can be fabricated using a liquid rapid solidification (LRS) method that employs a melting mechanism, as shown schematically in Fig. 11 [12]. The critical temperature for the order–disorder transformation in the Cu–Pd binary phase diagram is below 1023 K, and hence, at 1023 K, the Cu-rich phase α1 and Ag-rich phase α2 decompose, as shown in the Ag–Cu binary phase diagram. Figure 12 shows TEM micro‐ graphs of an Ag–20Pd–14.5Cu–12Au alloy fabricated by the LRS method [13]. No precipita‐ tion is observed in the matrix.

As shown in Fig. 12 (b) and (c), the LRS alloy consists of a single α phase with face centered cubic structure (FCC). The tensile properties of the as-received Ag–20Pd–14.5Cu–12Au alloy (AS), AS subjected to ST at 1123 K for 3.6 ks in vacuum (STAS/3.6 ks), LRS alloy (LRS), and LRS alloy subjected to ST at 1123 K for 3.6 ks in vacuum (STLRS/3.6 ks) are shown in Fig. 13 [13]. The

**Figure 11.** Schematic drawing of LRS method.

Figure 10 shows the SADP, key diagram, and HRTEM bright field image of ST1123K-WQ [11]. The SADP and the key diagram in Fig. 10 (a) and (b) indicate that the FCC α phase and the three variants of the L10-type ordered β′ phase are superimposed. The L10-type ordered β′ phase is precipitated in the FCC matrix after ST. The streaks on the reflection spots of the L10-type ordered β′ phase indicate that the shape of the β′ phase is similar to a thin plate. The HRTEM bright field image in Fig. 10 (c) shows that several nanometer-sized thick plateshaped β′ phase with two variants (N1, N2), whose c-axes are normal to the electron beam,

It is well known that it is difficult to make a single phase by quenching after high temperature ST; the supersaturated vacancies help the solutes to diffuse more easily and also help to form more clusters, G–P zones, and metastable phases during quenching after high temperature ST. The precipitated β′ phase of ST1123K-WQ shown in Fig.9 (d), (e) and Fig. 10 (c) is like a thin plate with a nanometer-scale thickness. These images also suggest that the formation of the β′ phase is diffusion controlled. Generally, the formation of precipitates can be considered to be order– disorder transition, diffusionless transformation (martensitic transformation), or diffusional transformation. In the as-solutionized Ag–Pd–Cu–Au alloy used in this case, the dependence of microstructural changes in the precipitated β′ phase on both the cooling rate after ST and the ST temperature show that the precipitated L10-type ordered β′ phase is formed during the cooling process and that the growth of the β′ phase is influenced by the diffusion process. The hardness increases with an increase in the cooling rate after ST, and consequently, the hard‐ ness of ST1123K-WQ increases significantly by quenching after ST (Fig. 6). The fine β′ phase in ST1123K-WQ is densely precipitated in the matrix. The hardness of ST1123K-AC increased only slightly, while the hardness of ST1123K-FC decreases, as only coarse β′ phases are precipitated in the matrix of ST1123K-AC and ST1123K-FC. Thus, the increase in hardness may be strongly affected by the presence of finely precipitated β′ phase. The coherent precipitation of β′ phases with long andshort axes of around100 nmand10 nm,respectively, also occuredduring ST, although the amount of β′ phase decreases with an increase in the ST time. The effect of solid solution hardening in the α, α1, and α2 phases is lower than that exerted by the precipitation harden‐

*3.1.2. Hardening behavior of Ag–20Pd–12Au–14.5Cu alloy fabricated by liquid rapid solidification*

An Ag–20Pd–14.5Cu–12Au alloy with a single α phase can be fabricated using a liquid rapid solidification (LRS) method that employs a melting mechanism, as shown schematically in Fig. 11 [12]. The critical temperature for the order–disorder transformation in the Cu–Pd binary phase diagram is below 1023 K, and hence, at 1023 K, the Cu-rich phase α1 and Ag-rich phase α2 decompose, as shown in the Ag–Cu binary phase diagram. Figure 12 shows TEM micro‐ graphs of an Ag–20Pd–14.5Cu–12Au alloy fabricated by the LRS method [13]. No precipita‐

As shown in Fig. 12 (b) and (c), the LRS alloy consists of a single α phase with face centered cubic structure (FCC). The tensile properties of the as-received Ag–20Pd–14.5Cu–12Au alloy (AS), AS subjected to ST at 1123 K for 3.6 ks in vacuum (STAS/3.6 ks), LRS alloy (LRS), and LRS alloy subjected to ST at 1123 K for 3.6 ks in vacuum (STLRS/3.6 ks) are shown in Fig. 13 [13]. The

are precipitated in the matrix.

524 Advances in Biomaterials Science and Biomedical Applications

ing due to β′ phases.

tion is observed in the matrix.

tensile strength and 0.2% proof stress of the AS alloy drastically increased after ST. The elongation of AS subjected to ST is smaller than that of the AS alloy. On the other hand, the tensile strength and 0.2% proof stress of the LRS alloy decrease after ST. The reduction in strain and the coarsening of the α phase during ST result in the decrease in the tensile strength and 0.2% proof stress and the increase in elongation. The tensile strength of the LRS alloy and solutionized LRS are relatively smaller than those of the AS alloy and the AS alloy subjected to the ST.

**Figure 12.** TEM micrographs of alloy fabricated by LRS: (a) bright field image, (b) diffraction pattern and (c) key dia‐ gram. Beam direction is parallel to [100].

Figure 14 shows the XRD profiles of the LRS alloy solutionized at 1173 K for 3.6 ks (1173WQLRS/3.6 ks) and the LRS alloy subjected to AT at 673 K for 1.8–28.8 ks following the ST (673WQLRS/1.8 ks-28.8 ks) [12]. A single α phase is identified in the solutionized LRS alloy, whereas, in the LRS alloy subjected to AT after the ST, α2 and β phases are observed. The Vickers hardness of the solutionized LRS alloy and the LRS alloy subjected to AT are shown in Fig. 15. The hardness of the LRS alloy subjected to AT increases greatly as compared to that of the solutionized LRS alloy with a single α phase owing to the precipitation of β phase.

**Figure 13.** Tensile properties of AS, STAS/3.6ks, LRS, and STLRS/3.6ks.

**Figure 14.** XRD profiles of (a) 1173WQLRS/3.6 ks, (b) 673WQLRS/1.8 ks, (c) 673WQLRS/3.6 ks, (d) 673WQLRS/7.2 ks, (e) 673WQLRS/14.4 ks, and (f) 673WQLRS/28.8 ks.

Figure 14

**Figure 15.** Vickers hardness of 1173WQLRS/3.6 ks, 673WQLRS/1.8 ks, 673WQLRS/3.6 ks, 673WQLRS/7.2 ks, 673WQLRS/14.4 ks, and 673WQLRS/28.8 ks.

#### **3.2. Fatigue properties**

Figure 14 shows the XRD profiles of the LRS alloy solutionized at 1173 K for 3.6 ks (1173WQLRS/3.6 ks) and the LRS alloy subjected to AT at 673 K for 1.8–28.8 ks following the ST (673WQLRS/1.8 ks-28.8 ks) [12]. A single α phase is identified in the solutionized LRS alloy, whereas, in the LRS alloy subjected to AT after the ST, α2 and β phases are observed. The Vickers hardness of the solutionized LRS alloy and the LRS alloy subjected to AT are shown in Fig. 15. The hardness of the LRS alloy subjected to AT increases greatly as compared to that of the solutionized LRS alloy with a single α phase owing to the precipitation of β phase.

Figure 14

**Figure 14.** XRD profiles of (a) 1173WQLRS/3.6 ks, (b) 673WQLRS/1.8 ks, (c) 673WQLRS/3.6 ks, (d) 673WQLRS/7.2 ks, (e)

**Figure 13.** Tensile properties of AS, STAS/3.6ks, LRS, and STLRS/3.6ks.

526 Advances in Biomaterials Science and Biomedical Applications

673WQLRS/14.4 ks, and (f) 673WQLRS/28.8 ks.

Dental prosthetic products produced by dental casting method are subjected to cyclic stress, i.e., fatigue, because of mastication. Dental castings contain a number of casting defects such as microshrinkages, pores, and surface roughness. The effects of these casting defects on the fatigue properties of cast Ag–20Pd–14.5Cu–12Au alloy were investigated in comparison with the fatigue properties of a drawn Ag–20Pd–14.5Cu–12Au alloy. The tensile properties of the cast specimens and drawn specimens are shown in Fig. 16 [14]. The cast specimens were prepared using a lost wax method. The drawn alloy bars were solutionized at 1073K for 3.6 ks in vacuum and then cooled in air (D-1073AC (drawn)). The heat treatment produced a microstructure with a matrix similar to that in an as-cast alloy. For the mechanical testing, the following specimen dimension were used: gauge diameter 3 mm and gauge length 20 mm for cast specimens, gauge diameter 2 mm and gauge length 20 mm for drawn specimens. The surfaces of the cast specimens for tensile tests were sand blasted (non-polished). The tensile strength and 0.2% proof stress of the cast specimens were higher than those of the drawn specimens. The distribution of the elongation of the cast specimens was larger than that of the elongation of the drawn specimens. The amount of intermetallic β phase that leads to the high strength and low ductility is greater in the cast specimens than in the drawn specimens. The microstructure of the cast specimens may be coarser than that of the drawn specimens. The relationship between the volume fraction and the number of the microshrinkage as measured on the fracture surface of the non-polished cast specimen and the elongation is shown in Fig. 17 [14]. Although no correlation was obtained between the volume fractions of the micro‐ shrinkage and elongation, the elongation decreases with an increase in the number of the microshrinkage. The surfaces of some fatigue test specimens were finished by buff-polishing (polished). Drawn specimens with a gauge diameter of 5 mm and a gauge length of 20 mm were used for the fatigue tests. Fatigue tests were carried out in order to obtain an S–N curve for each specimen at a stress ratio, R, of 0.1 and a frequency of 10 Hz with a sine waveform in air at room temperature (295 K). As can be seen in Fig.18 [14], the fatigue strength of the nonpolished specimen is similar to that of the polished specimen in both the low-cycle fatigue life region and the high-cycle fatigue life region. The distribution of the fatigue strength of both the cast specimens is greater than that of the fatigue strength of the drawn specimens. The fatigue strength of the non-polished cast specimens was 352–492 MPa (Δσmax (range of σmax) = 140 MPa) at 105 cycles and 209–284 MPa (Δσmax = 75 MPa) at 106 cycles. The fatigue strength of the polished cast specimens was 347–565 MPa (Δσmax = 218 MPa) at 105 cycles and 233–251 MPa (Δσmax = 18 MPa) at 106 cycles. The fatigue strength of both cast specimens was lower than that of the drawn specimens, and is extremely low in the high-cycle fatigue life region, where Nf exceeds 105 cycles. The fatigue limits, which represent the fatigue strength for which the number of cycles to failure is over 107 cycles, are around 210 MPa. On the other hand, although the fatigue limit of the drawn specimen was not obtained, it is expected to be near 400 MPa. Figure 19 [14] shows SEM fractographs taken near the site at which fatigue crack initiated in the cast specimens and the drawn specimen, which then broke in the high-cycle fatigue life region. In the case of the drawn specimen, the fatigue crack initiates at the slip band on the specimen surface, but in the case of the cast specimens, the fatigue crack initiates from the microshrinkage near the specimen surface. In general, slip damage accumulates on the specimen surface, after which extrusion and intrusion take place. Next, the stress concentration occurs and a fatigue crack initiates along the slip plane. If there are polishing scars, defects, etc. on the specimen surface, stress concentration occurs, and the fatigue strength becomes lower. The sizes of the microshrinkage areas and pores whose size exceeds 10 μm were measured, because the microshrinkage areas and pores whose size is less than 10 μm were difficult to distinguish from dimples on the fracture surface in the measurement on the fractographs. The number and size of the microshrinkage, whose size exceeds 10 μm as measured on the fatigue fracture surface, are greater than those measured on the cross- section near the fatigue fracture surface. Therefore, the fatigue crack propagates by preferentially linking the areas of microshrinkage. The number and size of the pores measured on the fatigue fracture surface is nearly equal to those measured on the cross-section near the fatigue fracture surface. Therefore, the effect of the pores on the fatigue properties is much smaller than that of the microshrinkage.

In general, dental prosthetic materials sustain a stress of 20–230MPa during mastication. Moreover, they must be able to sustain such cyclic stress over 10,000,000 times (107 cycles), which is the calculated number of cycles that is equivalent to the number of times food will be chewed in a span of ten years. Therefore, the target value of the fatigue limit of the cast specimen is considered to be 230MPa, which is the greatest mastication stress. Since the fatigue strength of this cast specimen is strongly dependent on the size of the microshrinkage that acts as the fatigue crack initiation site, it is prudent to estimate the size of the microshrinkage. It is also beneficial to know the size that can be tolerated, in order to achieve reliability in casting that is subjected to fatigue fracture.

**Figure 16.** Tensile properties of non-polished cast specimens (Cast) and drawn specimens (Drawn: D-1073AC (drawn)).

**Figure 17.** Relationships between volume fraction or number of shrinkage on fractograph and elongation of non-pol‐ ished cast specimen.

#### **3.3. Fretting–fatigue properties**

microshrinkage. The surfaces of some fatigue test specimens were finished by buff-polishing (polished). Drawn specimens with a gauge diameter of 5 mm and a gauge length of 20 mm were used for the fatigue tests. Fatigue tests were carried out in order to obtain an S–N curve for each specimen at a stress ratio, R, of 0.1 and a frequency of 10 Hz with a sine waveform in air at room temperature (295 K). As can be seen in Fig.18 [14], the fatigue strength of the nonpolished specimen is similar to that of the polished specimen in both the low-cycle fatigue life region and the high-cycle fatigue life region. The distribution of the fatigue strength of both the cast specimens is greater than that of the fatigue strength of the drawn specimens. The fatigue strength of the non-polished cast specimens was 352–492 MPa (Δσmax (range of σmax) = 140 MPa) at 105 cycles and 209–284 MPa (Δσmax = 75 MPa) at 106 cycles. The fatigue strength of the polished cast specimens was 347–565 MPa (Δσmax = 218 MPa) at 105 cycles and 233–251 MPa

of the drawn specimens, and is extremely low in the high-cycle fatigue life region, where Nf

the fatigue limit of the drawn specimen was not obtained, it is expected to be near 400 MPa. Figure 19 [14] shows SEM fractographs taken near the site at which fatigue crack initiated in the cast specimens and the drawn specimen, which then broke in the high-cycle fatigue life region. In the case of the drawn specimen, the fatigue crack initiates at the slip band on the specimen surface, but in the case of the cast specimens, the fatigue crack initiates from the microshrinkage near the specimen surface. In general, slip damage accumulates on the specimen surface, after which extrusion and intrusion take place. Next, the stress concentration occurs and a fatigue crack initiates along the slip plane. If there are polishing scars, defects, etc. on the specimen surface, stress concentration occurs, and the fatigue strength becomes lower. The sizes of the microshrinkage areas and pores whose size exceeds 10 μm were measured, because the microshrinkage areas and pores whose size is less than 10 μm were difficult to distinguish from dimples on the fracture surface in the measurement on the fractographs. The number and size of the microshrinkage, whose size exceeds 10 μm as measured on the fatigue fracture surface, are greater than those measured on the cross- section near the fatigue fracture surface. Therefore, the fatigue crack propagates by preferentially linking the areas of microshrinkage. The number and size of the pores measured on the fatigue fracture surface is nearly equal to those measured on the cross-section near the fatigue fracture surface. Therefore, the effect of the pores on the fatigue properties is much smaller than that

In general, dental prosthetic materials sustain a stress of 20–230MPa during mastication. Moreover, they must be able to sustain such cyclic stress over 10,000,000 times (107 cycles), which is the calculated number of cycles that is equivalent to the number of times food will be chewed in a span of ten years. Therefore, the target value of the fatigue limit of the cast specimen is considered to be 230MPa, which is the greatest mastication stress. Since the fatigue strength of this cast specimen is strongly dependent on the size of the microshrinkage that acts as the fatigue crack initiation site, it is prudent to estimate the size of the microshrinkage. It is also beneficial to know the size that can be tolerated, in order to achieve reliability in casting

cycles. The fatigue limits, which represent the fatigue strength for which the

cycles. The fatigue strength of both cast specimens was lower than that

cycles, are around 210 MPa. On the other hand, although

(Δσmax = 18 MPa) at 106

of the microshrinkage.

that is subjected to fatigue fracture.

number of cycles to failure is over 107

528 Advances in Biomaterials Science and Biomedical Applications

exceeds 105

Fretting–fatigue properties are also important for alloys that are used for dental applications, because during mastication, fretting occurs between the alloys and the teeth opposite them. Fig. 20 [15] shows the S–N curves of AS alloys subjected to ST at 1123 K for 3.6 ks followed by WQ and AT at 673 K for 1.8 ks followed by WQ that were obtained from plain fatigue and fretting–fatigue tests. The fretting–fatigue strength of the Ag–20Pd–14.5Cu–12Au alloy subjected to ST and AT decreases significantly as compared to the fatigue strength without fretting (plain–fatigue strength). The fretting–fatigue strength after the ST decreases by

**Figure 18.** S-N*<sup>f</sup>* curves of non-polished cast specimen (Non-polished (cast)), polished cast specimen (Polished (cast)), and drawn specimen (D-1073AC(drawn)).

**Figure 19.** SEM fractographs in high-cycle fatigue life region: (a) non-polished cast specimen, (b) polished cast speci‐ men, and (c) drawn specimen. Arrows indicate crack initiation sites. σmax and N*<sup>f</sup>* are the maximum cyclic stress and number of cycles to failure, respectively.

approximately 13% in the low-cycle fatigue life region and by approximately 40% in the highcycle fatigue life region, as compared to the fatigue strength of the solutionized alloy. More‐ over, the fretting–fatigue strength after the AT decreases by approximately 60% as compared to that after the ST, especially in the high-cycle fatigue life region. A schematic drawing of crack initiation from fretting damage region is shown in Fig.21 [15]. Although a slip between the specimen and a fretting pat does not occur in the stick region, a microslip between the specimen and the fretting pat occurs in the slip region during the deformation of the specimen. A significant stress concentration is generated by the damage to the specimen surface caused by the microslip. Fracture morphologies caused by crack initiation and propagation then appear. It can be observed that several traces of fretting wear are distributed in the slip region of both materials. These wear traces are generated by the accumulation of wear debris on the fretting pad or on the fretting–fatigue specimen. These traces of fretting wear are distributed more closely in the slip region of the material that was subjected to AT. Therefore, the fatigue life decreases significantly because the fretting–fatigue crack initiation life and the propagation life decrease in the material that was subjected to AT.

**Figure 20.** S–N curves of Ag–20Pd–14.5Cu–12Au alloy subjected to ST and AT obtained from plain fatigue and fret‐ ting fatigue tests.

#### **3.4. Friction wear properties of Ag–20Pd–12Au–14.5Cu alloy in corrosive environments**

approximately 13% in the low-cycle fatigue life region and by approximately 40% in the highcycle fatigue life region, as compared to the fatigue strength of the solutionized alloy. More‐ over, the fretting–fatigue strength after the AT decreases by approximately 60% as compared to that after the ST, especially in the high-cycle fatigue life region. A schematic drawing of crack initiation from fretting damage region is shown in Fig.21 [15]. Although a slip between the specimen and a fretting pat does not occur in the stick region, a microslip between the specimen and the fretting pat occurs in the slip region during the deformation of the specimen. A significant stress concentration is generated by the damage to the specimen surface caused

**Figure 19.** SEM fractographs in high-cycle fatigue life region: (a) non-polished cast specimen, (b) polished cast speci‐

are the maximum cyclic stress and

men, and (c) drawn specimen. Arrows indicate crack initiation sites. σmax and N*<sup>f</sup>*

curves of non-polished cast specimen (Non-polished (cast)), polished cast specimen (Polished (cast)),

**Figure 18.** S-N*<sup>f</sup>*

and drawn specimen (D-1073AC(drawn)).

530 Advances in Biomaterials Science and Biomedical Applications

number of cycles to failure, respectively.

Mastication also leads to friction wear in dental alloys and in teeth. As the friction wear progresses, this eventually causes problems with mastication. Therefore, the evaluation of the friction wear properties of dental alloys is quite important to the health of the teeth and oral cavity. The friction wear property of Ag–20Pd–14.5Cu–12Au alloy that was subjected to various heat treatments was evaluated in three corrosive environments: distilled water, 0.9% NaCl solution, and 3% NaCl solution. In the friction wear testing, the alloys were subjected to

**Figure 21.** Schematic drawing of crack initiation from fretting damage region.

STs at 1073 and 1123 K subjected to WQ, AT at 673 K, and ST at 1073 K subjected to AC. It was found that the friction wear properties are influenced by the microstructures of this alloy as well as by the corrosive environments. The friction wear tests were performed using a pin-ondisk-type friction wear tester. The applied load, sliding diameter, sliding velocity, sliding distance, and test duration were 9.8 N, 3 mm, 31.4 mm/s (100 rpm), 4.71×105 mm, and 115 ks, respectively. The temperature of all the solutions was 310 K. The weight loss was calculated by subtracting the combined weight of a specimen and the mating material it after the friction wear test, from their combined weight before the test. Figure 22 [16] shows the total weight loss and wear surface roughness of specimens that were subjected to each heat treatment in each corrosive solution. The total weight loss (the sum of the weight losses of the specimen and the material opposite it) of the specimens was largest in distilled water, 0.9% NaCl and then 3% NaCl solution. A pin with a diameter of 1 mm, which was made from the as-received alloy, was used as the mating material. The weight losses of the specimens and the mating materials increased with a decrease in the kinematic viscosity of the solutions, due to the increase in the average friction coefficient during the friction wear tests. In every environment, the total weight loss of the specimen subjected to AT at 673 K, which gives a large amount of precipitated β phase, was small compared with that of the other heat-treated specimens, due to the higher degree of hardness near the contact surface, as also shown in Fig.4 [10]. The total weight loss increased with the surface roughness of the contact surface. The hardness near the contact surface changed the surface roughness of the contact surface. The mode of wear of these alloys in a corrosive environment is adhesive wear. No tarnish, which is corrosion caused by chloride ion and the production of silver chloride coatings were found on the wear marks or wear particles after the friction wear tests in each environment.

**Figure 22.** Total weight loss and wear surface roughness of specimen subjected to each heat treatment in each corro‐ sive solution. 1073AC, 1073WQ, 1123WQ, and AT673 indicate specimens subjected to ST at 1073 K for 3.6 ks fol‐ lowed by AC or WQ, ST at 1123 K for 3.6 ks followed by WQ and AT at 673 K for 1.8 ks.

#### **3.5. Corrosion properties**

STs at 1073 and 1123 K subjected to WQ, AT at 673 K, and ST at 1073 K subjected to AC. It was found that the friction wear properties are influenced by the microstructures of this alloy as well as by the corrosive environments. The friction wear tests were performed using a pin-ondisk-type friction wear tester. The applied load, sliding diameter, sliding velocity, sliding

respectively. The temperature of all the solutions was 310 K. The weight loss was calculated by subtracting the combined weight of a specimen and the mating material it after the friction wear test, from their combined weight before the test. Figure 22 [16] shows the total weight loss and wear surface roughness of specimens that were subjected to each heat treatment in each corrosive solution. The total weight loss (the sum of the weight losses of the specimen and the material opposite it) of the specimens was largest in distilled water, 0.9% NaCl and then 3% NaCl solution. A pin with a diameter of 1 mm, which was made from the as-received alloy, was used as the mating material. The weight losses of the specimens and the mating materials increased with a decrease in the kinematic viscosity of the solutions, due to the increase in the average friction coefficient during the friction wear tests. In every environment, the total weight loss of the specimen subjected to AT at 673 K, which gives a large amount of precipitated β phase, was small compared with that of the other heat-treated specimens, due to the higher degree of hardness near the contact surface, as also shown in Fig.4 [10]. The total weight loss increased with the surface roughness of the contact surface. The hardness near the contact surface changed the surface roughness of the contact surface. The mode of wear of these alloys in a corrosive environment is adhesive wear. No tarnish, which is corrosion caused

mm, and 115 ks,

distance, and test duration were 9.8 N, 3 mm, 31.4 mm/s (100 rpm), 4.71×105

**Figure 21.** Schematic drawing of crack initiation from fretting damage region.

532 Advances in Biomaterials Science and Biomedical Applications

The study of the corrosion properties of dental alloys is very important due to their usage in severe oral environments. The formation of corrosion compounds on the surface causes the alloy to tarnish. The corrosion behaviors of commercial Ag-22.41Pd-15.64Cu-12.1Au alloys in various solutions were investigated by Ichinose [17]. The alloy was casted into a plate with 2 mm in thickness by heating at 973 K for 30 min. Table 2 [17] shows the rest-potential of the ascast alloy in 1% NaCl solution, artificial saliva, human saliva, and 1% C3H6O3 solution. The rest-potentials in 1% NaCl solution and 1% C3H6O3 solution show obviously higher values than those for artificial and human saliva. The organic compounds in the artificial and human saliva inhibit the redox reaction by adsorbing on the surface of the alloy or by the formation of complexes with the metal. AgCl was formed on the surface of the alloy after anode polari‐ zation sweeping up to 1800 mV in 1% NaCl solution, artificial saliva, and human saliva, which contain Cl ion. 90% of the released ions of the as-cast alloy immersed in human saliva for 24 h were Cu ions (Fig.23 [17]). Cu contributes to the enhancement the hardness of the alloy. However, a large amount of Cu in the alloy decreases the corrosion resistance. The as-cast alloy was subjected to a softening treatment at 1073 K for 3 min followed by WQ (softened). An ingot of the commercial Ag-22.41Pd-15.64Cu-12.1Au alloy was subjected to ST at 1123 K for 2 h and then heated at 723 K for 1 h or 20 h (723 K 1h aged and 723 K 20 h aged, respectively). As shown in Fig. 24 [17], the amounts of released Cu and Ag ions depend on the heat treatment that the alloy is subjected to, because the corrosion behavior is influenced by the microstructure of the alloy. The amount of Cu ions released from the as-cast alloy is higher than that released from heat-treated alloys. Cavities formed by shrinkage during casting in the as-cast alloy are the reason for this phenomenon. Table 2


**Table 2.** Rest-potential of as-cast Ag-22.41Pd-15.64Cu-12.10Au alloy in various corrosive solutions.

Figure 23 **Figure 23.** Amounts of ions released from as-cast Ag-22.41Pd-15.64Cu-12.10Au alloy immersed in human saliva for 24 h.

#### **4. Summary**

Figure 23 This chapter details the microstructure, the mechanical properties (hardness, fatigue, frettingfatigue, and friction-wear), and the corrosion properties of the Ag–Pd–Cu–Au alloys, espe‐ cially the Au-20Pd-14.5Cu-12Au alloy. Most studies on these alloys have been carried out in Japan and Korea. Although other dental materials such as high carat gold alloys, amalgam,

rest-potentials in 1% NaCl solution and 1% C3H6O3 solution show obviously higher values than those for artificial and human saliva. The organic compounds in the artificial and human saliva inhibit the redox reaction by adsorbing on the surface of the alloy or by the formation of complexes with the metal. AgCl was formed on the surface of the alloy after anode polari‐ zation sweeping up to 1800 mV in 1% NaCl solution, artificial saliva, and human saliva, which contain Cl ion. 90% of the released ions of the as-cast alloy immersed in human saliva for 24 h were Cu ions (Fig.23 [17]). Cu contributes to the enhancement the hardness of the alloy. However, a large amount of Cu in the alloy decreases the corrosion resistance. The as-cast alloy was subjected to a softening treatment at 1073 K for 3 min followed by WQ (softened). An ingot of the commercial Ag-22.41Pd-15.64Cu-12.1Au alloy was subjected to ST at 1123 K for 2 h and then heated at 723 K for 1 h or 20 h (723 K 1h aged and 723 K 20 h aged, respectively). As shown in Fig. 24 [17], the amounts of released Cu and Ag ions depend on the heat treatment that the alloy is subjected to, because the corrosion behavior is influenced by the microstructure of the alloy. The amount of Cu ions released from the as-cast alloy is higher than that released from heat-treated alloys. Cavities formed by shrinkage during casting in the as-cast alloy are the

> Solution Rest–potential 1% NaCl solution 112.7 ± 22.4 mV Artificial saliva 24.7 ± 17.2 mV Human saliva 28.0 ± 20.4 V 1% C3H6O3 solution 113.6 ± 37.7 mV

**Table 2.** Rest-potential of as-cast Ag-22.41Pd-15.64Cu-12.10Au alloy in various corrosive solutions.

Table 2

Solution Rest–potential 1% NaCl solution 112.7 ± 22.4 mV Artificial saliva 24.7 ± 17.2 mV Human saliva 28.0 ± 20.4 V 1% C3H6O3 solution 113.6 ± 37.7 mV

Table 2

Figure 23

This chapter details the microstructure, the mechanical properties (hardness, fatigue, frettingfatigue, and friction-wear), and the corrosion properties of the Ag–Pd–Cu–Au alloys, espe‐ cially the Au-20Pd-14.5Cu-12Au alloy. Most studies on these alloys have been carried out in Japan and Korea. Although other dental materials such as high carat gold alloys, amalgam,

Figure 23 **Figure 23.** Amounts of ions released from as-cast Ag-22.41Pd-15.64Cu-12.10Au alloy immersed in human saliva for

reason for this phenomenon.

534 Advances in Biomaterials Science and Biomedical Applications

24 h.

**4. Summary**

**Figure 24.** Cu and Ag release amounts from as-cast, softened, WQ, 723 K 1h aged, and 723 K 20 h aged Ag-22.41Pd-15.64Cu-12.10Au alloys immersed in human saliva for 24 h.

and non-precious alloys such as titanium and its alloys, cobalt-chromium alloys, and nickel– chromium alloys are also important, the authors intended to introduce present results of studies on the Ag–Pd–Cu–Au alloys to researchers who study dental materials all over the world through this chapter.

A variety of test results showed that the hardness of the alloy can be drastically increased through ST at a temperature over 1073 K subjected to WQ. The L10-type ordered phase (β′ phase) precipitated during the quenching process leads to this unique hardening behavior. The hardness of the alloy increased with an increase in the cooling rate following ST, because the size of the β′ phase decreased and the number of the β′ phase increased with an increase in the cooling rate following ST.

It was also found that the fatigue strength of a cast alloy was considerably less than that of a drawn alloy. The fatigue crack of a cast alloy initiated preferentially at the shrinkage near the specimen surface. The deviation of the fatigue strength of the cast alloy became small by relating the fatigue life to the maximum stress intensity factor that was calculated, assuming that the shrinkage that begins as a fatigue crack initiation site becomes an initial crack. This means that the size of the shrinkage strongly affects the fatigue strength of this cast alloy. A tolerable shrinkage size that satisfies the target value of the fatigue limit (230 MPa) of this cast alloy was calculated to be below 80 μm, using a derived equation that describes the relationship between the maximum stress intensity factor and the number of the cycles to failure.

The fretting–fatigue strength of the Ag–20Pd–12Au–14.5Cu alloy subjected to ST and aging treatment decreased significantly as compared to the fatigue strength without fretting. The fretting–fatigue strength after the aging treatment decreased by approximately 60% as compared to that following the ST, especially in the high-cycle fatigue life region.

The total weight losses (the sum of the weight losses of a specimen and of the mating material) of specimens were largest in distilled water, subjected to 0.9% NaCl and then 3% NaCl solution. In every environment, the total weight loss of a specimen subjected to aging treatment at 673 K, which gives a large amount of precipitated β phase, was small compared with that of the other heat treated specimens, due to the higher degree of hardness near the contact surface.

The Ag-Pd-Cu-Au alloys have excellent mechanical and corrosion properties, but the relatively high cost and supplied amounts of these alloys are issues to be used widely. The authors hope that the Ag-Pd-Cu-Au alloys is further investigated by many researchers for its wide usage.

### **Acknowledgements**

This study was supported in part by ISHIFUKU Metal Industry Co., Ltd. for their support during this study. This study was financially supported in part by a Grant-in-aid for Scientific Research from the Japan Society for the Promotion of Science (Grant Number 21360332), the Global COE program "Materials Integration International Center of Education and Research, Tohoku University," Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan, and Tohoku Leading Women's Jump Up Project, Tohoku University. The authors are very grateful to Dr. YH. Kim (Institute for Materials Research, Tohoku University) and Prof. H. Fukui (Aichi-Gakuin University) for their experimental work and fruitful discussion.
