Dent Mater 14:1120, January, 1998

Cyclic contact fatigue of a silver alternative to amalgam

Hockin H. K. Xu, Frederick C. Eichmiller, Anthony A. Giuseppetti, Christian E. Johnson

Paffenbarger Research Center, American Dental Association Health Foundation, National Institute of Standards and Technology, Gaithersburg, MD, USA

ABSTRACT
Objective. The purpose of this study was to compare the cyclic contact fatigue resistance of a novel mercury-free silver direct lling material to that of a dental amalgam (Dispersalloy). Methods. The silver specimens were made by pressing a precipitated powder at room temperature with a pressure of 150 MPa, which can be achieved in clinical hand-consolidation. To simulate clinical contact of restorations against enamel cusps, a cyclic contact fatigue methodology was employed. A spherical indenter was used to repeatedly indent the specimen, while the accumulation of deformation and damage was examined as a function of the number of cycles up to 5 10 5. Students t test, analysis of variance (ANOVA) and Duncans multiple range test were used to compare the specimen groups for signicant differences in exural strength, indentation impression diameter, and hardness. A type I error of a = 0.05 was considered as signicant. Subsurface damage was examined by using a bonded-interface technique. Results. As a result of cyclic indentation, microcracks were produced in the amalgam, but no cracks were found in the silver lling material. At fewer numbers of cycles, indentation produced larger impressions in silver (e.g., diameter = [450 31] mm at 10 2 cycles) than in amalgam ([145 20] mm) due to a lower hardness of the former. However, with increasing number of cycles, damage accumulated more rapidly in the amalgam, while the silver benecially workhardened in repeated indentations. At 5 10 5 cycles, the difference in impression diameter between silver and amalgam ([582 20] mm vs. [568 42] mm) became insignicant (p 0.1, Students t test). Signicance. The mercury-free silver direct lling material is more resistant to microcracking and to cyclic contact fatigue than amalgam, and the indentation impression sizes in the consolidated silver and dental amalgam are not statistically different at large numbers of cycles. Published by Elsevier Science Ltd on behalf of the Academy of Dental Materials

INTRODUCTION
Dental amalgam has been widely used as a direct lling
PII: S0109-5641(98)00004-9

material due to its favorable mechanical properties as well as low cost and ease of placement. The mercury that dental amalgams contain, however, raises concerns about its biological toxicity and environmental hazards (e.g., Macker, 1991; Mandel, 1991). Therefore, increasing efforts have been devoted to developing mercury-free alternatives to dental amalgam. Polymeric composites, although a promising aesthetic alternative, have drawbacks such as low durability, especially in posterior stress-bearing applications (Ferracane, 1992; Leinfelder, 1994; Bayne et al., 1994; Ausavice, 1996). Gallium-based metallic alloys are more prone to corrosion (Oshida and Moore, 1993). Recent studies have reported a silver-based material free of mercury for dental direct lling applications (Lashmore et al., 1994; Dariel et al., 1995a,b). The fabrication of this material relies on the room-temperature welding of a loose powder into a cohesive solid by fusing silversilver interfaces that are relatively uncontaminated on an atomic level. Flexural and compressive strength values, equivalent to those of amalgam, were obtained (Dariel et al., 1995a,b). In vitro corrosion studies using articial saliva showed that the consolidated silver possessed a corrosion resistance comparable to that of amalgam (Mueller, 1998). The previous strength results (Dariel et al., 1995a,b) were obtained by conventional single-load fast fracture tests. To evaluate in vitro the long-term clinical durability of this material, fatigue studies are needed. Dental restorations in vivo are not subjected to a single load, but to many cycles of load. Clinical studies have observed a gradual accumulation of damage and degradation for dental amalgam restorations (Marshall et al., 1980; Osborne et al., 1991; Lian and Meletis, 1996). Therefore, information on cyclic fatigue is essential to the evaluation of dental materials. Conventional fatigue tests involve the measurement of
Dental Materials/January 1998 11

fracture strength of specimens subjected to dynamic or cyclic loading in biaxial and uniaxial exure, and in compression (Zardiackas and Bayne, 1985; Sutow et al., 1985; Morena et al., 1986; McCabe and Carrick, 1987; Twiggs et al., 1995; Watkins et al., 1995). These tests provide important information for applications involving tensile stresses where failure is manifested as the macroscopic fracture of the bulk specimen. These tests, however, may not reect certain clinical situations relative to dental restorations where fatigue is governed by cyclic contacts and failure results from localized damage accumulation. Recently, a cyclic indentation technique was developed to characterize the contact fatigue behavior of ceramics (Guiberteau et al., 1993; Cai et al., 1994; Xu et al., 1995). Originally used by Hertz (1896) to analyze cone fracture in brittle solids and hence named Hertzian indentation, this test is as simple as using a sphere to indent the at surface of a solid. When the Hertzian indentation test is used as a fatigue test, the same spot in the surface of a specimen is repeatedly indented, and damage evolution is examined as a function of the number of indentation cycles. This methodology has been applied to the study of deformation and fatigue of dental glass ceramics (Cai et al., 1994; Fisher-Cripps and Lawn, 1996) and feldspathic porcelain (White et al., 1995). Metallic dental restorative materials may undergo embrittlement in cyclic contact fatigue due to dislocation accumulation and stress build-up at grain boundaries. Indeed, dental amalgam restorations have been observed clinically to exhibit microfracture along the margins, rather than macroscopic bulk fracture of the entire restoration (Marshall et al., 1980; Osborne et al., 1991). In the present study, the cyclic Hertzian indentation test was employed to study the contact fatigue behavior of the silver-based direct lling material in comparison to a dental amalgam (Dispersalloy). The silver-based specimens were made by pressing a precipitated powder at room temperature with a degree of pressure that resulted in a sample density equivalent to clinical hand-consolidation. In order to examine the contact fatigue response, a spherical indenter was used to repeatedly indent the specimens at numbers of indentation cycles ranging from 1 to 5 10 5. Subsurface deformation and damage were examined by using a bonded-interface technique.

MATERIALS AND METHODS

Materials. Three materials were prepared in the present study: a dental amalgam, a silver-based material with the surface polished (designated as polished silver), and the same silver-based material with the surface burnished (burnished silver). Fourteen exure specimens of dental amalgam (Dispersalloy, Dentsply, Milford, DE) were fabricated. The amalgam was mixed in an amalgamator (VARIMIX III, Dentsply, Milford, DE) for 13 s, then condensed into a steel mold cavity of approximately 32 3 2 mm using a condenser of 2 mm diameter at tip under a load of approximately 13 N. After demolding, the specimens were incubated at 37C for 24 h before polishing and testing.
12 Xu et al./Fatigue of a silver alternative to amalgam

The early versions of the silver-based material involved compositions of silver and tin and silvercoated alloy particles (Lashmore et al., 1994; Dariel et al., 1995a,b). These powders required high compaction pressures to consolidate. In the present study, a different powder was used that can be consolidated at a lower pressure more relevant to clinical handconsolidation. A chemically precipitated pure silver powder with particle size ranging from 0.2 mm to 2.0 mm were admixed with a volume fraction of 2% g-alumina of particle size 0.05 mm. The powder was passed through a 200 mesh sieve to eliminate large agglomerates. The sieved powder was then annealed for 2 h at 450C to soften the silver particles and promote deformability during subsequent consolidation. To remove any possible oxide lms from the particle surfaces and to promote inter-particle welding during subsequent consolidation, the powder was submerged in an aqueous solution with a weight fraction of 10% of uoboric acid for 4 min, then rinsed in an aqueous solution with a weight fraction of 2% uoboric acid for 1 min. The wet powder was then used to make a specimen within 15 min after the rinsing. The clinical effects of uoboric acid treatment of silver powder are being investigated with consolidated silver llings in monkey teeth. Flexure specimens approximately 32 6 2.3 mm (length width depth) were made by pressing the wet silver powder in a thick-wall steel mold with slidable steel punches. Approximately 3.5 g of powder was used to make each specimen by pressing at 150 MPa pressure (or a load of 28 800 N) at room temperature. This pressure was chosen because it consolidated specimens to a density equal to what can be achieved by hand-consolidation with clinically realistic loads. The dimensions of the silver specimens were different from those of the amalgam specimens because the small mold used to make the amalgam specimens did not have slidable punches and would not stand the load used to press the silver specimens. Fourteen silver specimens were thus made with a density of approximately 7.9 g/mm 3 (72% of theoretical density of 10.5 g/mm 3). The amalgam and silver specimens were polished consecutively with 15, 6, 3, 1, and 0.25 mm diamond paste to remove a surface layer approximately 300 mm thick. The 14 silver specimens were randomly divided into two groups: seven were kept as-polished; the other seven were burnished on the polished surfaces. The burnishing was performed by rst scratching the specimen surface with a stainless steel dental burnisher at a normal load of approximately 13 N. The tip radius of the burnisher was 0.5 mm. Each burnishing stroke was approximately 10 mm long; approximately 3 min were required to burnish a specimen surface of 192 mm 2 area. Burnishing was followed by nishing or smoothing with a plastic dental burnisher. The amalgam specimens were not burnished because this procedure induced damage, especially in areas near the specimen edges. Flexural strength. An articulating four-point bending xture with an inner/outer span of 10 mm/20 mm was used in a screw-driven universal testing machine

(Model 1122, Instron Corp., Canton, MA) to fracture the specimens. Five measurements were made for each of the three materials: amalgam, polished silver, and burnished silver, with the polished or burnished surfaces in tension. A displacement rate of 0.5 mm/ min was used. The broken halves were repolished on the previously polished surfaces to remove a layer approximately 300 mm thick and used for indentation testing. Vickers hardness and fracture toughness measurements. A Vickers diamond pyramid was used in a standard microhardness indenter (Zwick of America, Inc., East Windsor, CT) to indent the surfaces of specimens. Seven different loads of 2, 5, 10, 20, 50, 100, and 200 N were used, with four separate indentations per load for each material. A distance of at least ve times the indentation size was kept between neighbouring indents to minimize interference between indentations. The indentation size was measured using an optical microscope with Nomarski interference contrast (Nikon Diaphot, Mager Scientic, Inc., MI) coupled with a video micrometer (Model 305, Colorado Video, Inc., Boulder, CO). The hardness was then calculated by using a standard equation (ASTM E 384, 1991): H = 1X854Pa(2a)2 (1)

where P is the load and a is half the diagonal length of the indentation. The lengths of the cracks produced at the corners of the indentations were also measured; c is dened as the crack length measured from the center of the indentation to the crack tip. The fracture toughness T was then calculated by using the following expression (Anstis et al., 1981; Lawn, 1993): T = y(EaH )1a2 Pac3a2 (2)

Fig. 1. a Hertzian indentation test where the surface of a specimen is indented with a spherical indenter. b Bonded-interface sectioning technique to reveal indentation-induced subsurface damage.

where the indentation coefcient y = 0.016 (Anstis et al., 1981; Lawn, 1993), E is Youngs modulus, and H is hardness. Mostly used for ceramics, this technique has also been applied to steels (Fields et al., 1984) and composite resins (Ferracane, 1989) that formed cracks during indentation. The toughness value of a steel obtained by using the indentation technique was within 10% of that determined by a conventional technique with chevron-notched specimens (Fields et al., 1984). Hertzian indentation and cyclic contact fatigue. As shown schematically in Fig. 1a, the Hertzian indentation test involved indenting the surface of a specimen with a spherical indenter in a servo-hydraulic machine (Model 8502, Instron Corp., Canton, MA). The indenter was a tungsten carbide sphere with a radius of 3.2 mm. Single- and multiplecycle contacts were made at a peak load of 50 N and at a frequency of 10 Hz under sinusoidal load control. The sphere radius and the load were chosen to be clinically realistic. Five different numbers of cycles were used: 1, 10 2, 10 4, 10 5, and 5 10 5. With the load and number of cycles used, the largest depth of Hertzian impressions produced in these materials was observed to be less

than 20 mm, which is two orders of magnitude smaller than the specimen depth. After indentation, the impression size in the specimen surface was measured with the optical microscope using Nomarski contrast. The enlargement of the impression with increasing number of cycles was monitored. The impression size, together with the nature of damage within and beneath the impression, was used as an indication of the contact fatigue response of the material. In this article, the terms cyclic Hertzian indentation and cyclic contact are used interchangeably. Subsurface damage. In order to examine the mechanism and the extent of subsurface deformation and damage induced by cyclic contacts, a bonded-interface sectioning technique, detailed elsewhere (Guiberteau et al., 1994; Cai et al., 1994; Xu et al., 1995), was used. In this technique (Fig. 1b), the polished surfaces of two specimens of the same material were bonded together by adhesive (Superglue, Loctite Corp., Newington, CT). Clamping pressure was applied during bonding to make the interface narrow (less than 1 mm thick) in order to avoid articial damage associated with the interface during subsequent

Dental Materials/January 1998 13

Fig. 2. Polished and burnished surfaces. SEM micrographs show a polished surface of amalgam, b polished surface of silver, and c burnished surface of silver. Note AgCu eutectic particles (indicated by A) imbedded in the AgHg matrix (M). Arrows indicate pores. Burnishing has reduced the surface porosity in silver.

Fig. 3. Vickers indentations at a load of 50 N in a amalgam, b polished silver, and c burnished silver. Well-developed cracks (arrow) were produced in amalgam. Circumferential cracks (arrow) were visible around the impressions in polished silver. Cracks were absent in burnished silver. Note the smaller magnication for silver than for amalgam.

indentations. The top surface to be indented (Fig. 1b) was polished consecutively with 15, 6, 3, 1, and 0.25 mm diamond paste to remove a thickness of approximately 0.5 mm. This polished top surface was then used for cyclic Hertzian indentation testing. During indentation, the sphere was loaded symmetrically on the top surface along the trace of the bonded interface, in the manner of Fig. 1b. After indentation, the bonded specimens were separated by dissolving the adhesive in acetone, and the surfaces were cleaned and viewed in the optical microscope with Nomarski contrast and with a scanning electron microscope (SEM, model JSM-5300, JEOL, Inc., Peabody, MA) to reveal subsurface damage produced during cyclic contacts. Four specimens of each material were used for bonded-interface experiments. Statistical analysis of data. Students t test, analysis of variance (ANOVA) and Duncans multiple range test were used to compare the specimen groups for signicant differences in exural strength, indentation impression diameter, and hardness. A
14 Xu et al./Fatigue of a silver alternative to amalgam

type I error of a = 0.05 was considered as signicant.

RESULTS
Microstructure. SEM micrographs in Fig. 2 show a the polished surface of amalgam, b the polished surface of silver, and c the burnished surface of silver. The amalgam microstructure consisted of AgCu eutectic particles (indicated by A in Fig. 2a), embedded in a matrix of AgHg grains (M). Occasional pores are present as indicated by the arrow. Clusters of pores are visible in the polished surface of the silver lling (arrows in Fig. 2b) together with polishing scratches. Burnishing appears to have reduced the surface porosity of the silver specimen, producing a denser surface with burnishing scratches (Fig. 2c). Vickers indentation. Representative examples of Vickers indentations at a load of 50 N are shown in the optical micrographs in Fig. 3ac for amalgam, polished silver, and burnished silver, respectively. Well-developed cracks were produced at the corners of the indentation in amalgam, as indicated by the arrow

TABLE 1: FLEXURAL STRENGTH OF AMALGAM, POLISHED SILVER, AND BURNISHED SILVER*

Materials Strength (MPa) Amalgam 83 8A Polished silver 97 13A Burnished silver 147 30B

*Each datum is the average valueone standard deviation of ve measurements. Values with similar letters are not statistically different (p>0.1, Duncans multiple range test). Values with dissimilar letters are statistically different (p<0.05, Duncans multiple range test).

Fig. 5. Optical micrographs of Hertzian cyclic indentation fatigue in a amalgam, b polished silver, and c burnished silver. Two different numbers of contact cycles, 1 and 5 10 5, were used to produce the impressions as indicated in each micrograph.

Fig. 4. Hardness vs. indentation load. Each datum is the average value of four measurements with the error bar showing one standard deviation. Note that amalgam is the hardest, followed by the burnished silver, then the polished silver. Also note that the near-surface hardness (measured at small loads) of silver was increased by a factor of almost ve due to burnishing.

in Fig. 3a. These cracks are usually categorized as either radial or median cracks (Lawn, 1993). No such cracks were found in the polished silver, but circumferential cracks were visible around the indentations (arrow in Fig. 3b). Both types of cracks were absent in the burnished silver (Fig. 3c). Note the smaller magnication for silver than for amalgam due to the larger indentation for the polished silver. The burnished surface of silver had a smaller indentation than that of the polished silver (compare Fig. 3b with c). These observations suggest that the burnished surface of silver is tougher (i.e., free of indentation cracks) as well as harder than the polished silver. The crack lengths were used in Eq. (2) to calculate the fracture toughness of amalgam at indentation loads of 50, 100, and 200 N with four indentations at each

load. This yielded an average value with one standard deviation of 0.98 0.09 MPam 1/2. Fracture toughness at smaller loads was not calculated since the crack size was too small, as indentation fracture mechanics requires c 2a (Anstis et al., 1981; Lawn, 1993). Fracture toughness of polished and burnished silver was not obtained due to the absence of radial or median cracks in these specimens at indentation loads up to 200 N. Hardness at different loads was calculated by using Eq. (1), and the data are plotted in Fig. 4. Amalgam was the hardest, followed by burnished silver, then polished silver. The hardness values of amalgam and polished silver were insensitive to the indentation load. However, the hardness value of the burnished silver decreased with increasing load; this decrease was most likely a result of the nite thickness of the burnished layer being penetrated by the Vickers indenter at large loads. Note that the near-surface hardness of silver (measured at small loads) was increased almost by a factor of ve due to burnishing. Flexural strength. The strength in the four-point bending test was calculated by using the following formula (ASTM, 1984): S = 3F (L 1 L 2)/2bw 2, where S is strength, F is the maximum applied load when the specimen fractured, L 1 is the outer span and L 2 is the inner span of the four-point xture, b is the specimen
Dental Materials/January 1998 15

Fig. 6. Diameter of Hertzian impressions vs. number of indentation cycles. Each datum is the average value of four measurements with the error bar showing one standard deviation. At 5 10 5 cycles, the impression diameters in amalgam and in burnished silver are not statistically different ( p 0.1; Students t ).

width, and w is the specimen depth. During exural testing, little plastic deection occurred before the specimens failed in a brittle manner. Amalgam is a relatively brittle material; the nearly 30% porosity in the consolidated silver is likely responsible for its reduction in ductility. The strength results thus obtained are listed in Table 1 for amalgam, polished silver, and burnished silver. The difference between the strength of amalgam and of polished silver is not signicant ( p 0.1, Duncans multiple range test), but that between the polished silver and burnished silver is signicant ( p 0.05, Duncans multiple range test). Burnishing produced a relative increase of 52% in the strength of silver, or 50 MPa. Hertzian indentation. Optical micrographs of representative Hertzian impressions in a. amalgam, b. polished silver and c. burnished silver are shown in Fig. 5. A single indentation cycle was used to produce the impressions in the left column of Fig. 5, and 5 10 5 cycles in the right column, as indicated in each micrograph. Little permanent deformation occurred in the amalgam specimens after indenting at a single cycle. However, the impression size increased rapidly when the number of cycles was increased. A large impression was produced in the polished silver at the rst cycle, and the impression size was increased at 5 10 5 cycles. In the burnished silver, the impression produced by the rst cycle was substantially smaller than that in the polished silver, and the impression
16 Xu et al./Fatigue of a silver alternative to amalgam

size increased only slightly with increasing number of cycles. At 5 10 5 cycles, the impression size in the burnished silver and in amalgam became similar. This cyclic indentation test was repeated in four different specimens for each material. The average impression diameters are plotted in Fig. 6 against the number of indentation cycles. At fewer numbers of cycles, indentation produced bigger impressions in burnished silver (e.g., diameter = [450 31] mm at 10 2 cycles) than in amalgam ([145 20] mm) due to the lower hardness of silver. However, with increasing numbers of cycles, the indentation impression in amalgam enlarged at a faster rate than that in burnished silver. At 5 10 5 contact cycles, the difference in impression diameter between burnished silver and amalgam ([582 20] mm vs. [568 42] mm) became insignicant ( p 0.1, Students t). The impressions were examined by SEM to elucidate the mechanisms of deformation and damage. An area of approximately 10 5 mm 2 of two impressions was examined for each material. Fig. 7ac show typical micrographs within the indentation impressions for amalgam, polished silver, and burnished silver, respectively, at 5 10 5 cycles. Microcracks were observed within the impressions in amalgam, an example of which is indicated by the arrow in Fig. 7a. The surface appeared to be smeared by repeated indentations (compare with the polished surface in Fig. 2a). Pores were still present in silver after indentations (Fig. 7b), but their sizes were decreased (compare Fig. 7b with Fig. 2b). Similarly, in burnished silver, repeated indentation attened and smoothed the burnishing scratches (Fig. 7c). Subsurface deformation and damage. Fig. 8ac show representative cross-section views of subsurface damage obtained from the bonded-interface technique after 5 10 5 indentation cycles in a. amalgam, b. polished silver, and c. burnished silver. The two solid arrows in each micrograph indicate the approximate position of the indentation. The vertical bar in Fig. 8c indicates the depth of the burnished layer outside the indentation area. The indentation-induced surface depression is visible, especially for the polished silver in Fig. 8b. The nature of subsurface damage and deformation is not visible at this magnication. Therefore, two methods were used to characterize the indentation-induced subsurface deformation and damage: SEM examination at higher magnications and hardness measurements of these cross-sections in areas immediately beneath the indentations. SEM views of the areas in the proximity of the open arrows in Fig. 8ac are shown in Fig. 9ac, respectively. Subsurface microcracks and damaged areas (arrows in Fig. 9a) were found in amalgam. Pores are evident in the polished silver in Fig. 9b as are a few polishing scratches, but no cracks nor damaged areas were observed. Subsurface cracks were also absent in the burnished silver (Fig. 9c). These cross-section views show that burnishing reduced the subsurface porosity in silver. The hardness results on the bonded-interface cross sections are listed in Table 2. The hardness values of

Fig. 7. Indented surfaces. SEM micrographs of areas within the Hertzian impressions after 5 10 5 cycles for a amalgam, b polished silver, and c burnished silver. Arrows in a indicates a microcrack in amalgam. The surface of amalgam appears smeared due to indentation (compare with the polished surface in Fig. 2a). Arrows in b indicate pores in silver.

Fig. 8. Indented cross-sections. Subsurface views obtained from the bonded-interface technique after 5 10 5 indentation cycles for a amalgam, b polished silver, and c burnished silver. The two solid arrows in each micrograph indicate the sphere indentation position. The vertical bar in c outside the indentation area indicates the depth of the burnished layer. The indentation-induced surface depression is visible especially for polished silver in b. The open arrow in each micrograph indicates the area examined in SEM in Fig. 9.

Fig. 9. Indented cross-sections. High magnication SEM views in the proximity of the open arrows in Fig. 8ac, respectively. Subsurface microcracks and damaged areas in amalgam are indicated by arrows in a. No cracks were observed in silver b and c. Burnishing has reduced the subsurface porosity in silver.

amalgam inside and outside the impression zones are not signicantly different ( p 0.1, Students t), indicating little work-hardening due to Hertzian indentation, even after 5 10 5 cycles. In contrast, there is a signicant difference ( p 0.05, Students t) between the two hardness values for polished silver, suggesting a work-hardening effect from cyclic

contacts. The hardness values for burnished silver are not signicantly different ( p 0.1, Students t), implying that the work-hardening in silver was nearly complete during burnishing; hence, subsequent cyclic indentations resulted in negligible additional work-hardening.

Dental Materials/January 1998 17

TABLE 2: VICKERS HARDNESS (GPA) AT A LOAD OF 2N IN THE BONDED-INTERFACE CROSS SECTIONS OF AMALGAM, POLISHED SILVER, AND BURNISHED SILVER*
Amalgam Un-indented areas on 1.57 0.25A cross-sections Under the Hertzian impressions** 1.63 0.14A Polished silver 0.27 0.04B 0.38 0.10C Burnished silver 0.80 0.18D 0.85 0.18D

*Each datum is the average valueone standard deviation of six measurements. For each material, values with similar letters are not statistically different (p>0.1, Students t). **Hardness measurements under the Hertzian impressions were made in areas near those indicated by the open arrows in 8.

DISCUSSION
Cracking. The present study demonstrates that, compared to dental amalgam, the mercury-free silver material is more resistant to cracking during sharp diamond indentation and cyclic Hertzian contact. Sharp indentation in amalgam produced cracks (Fig. 3a) similar to those observed in brittle materials such as glasses and ceramics (Lawn, 1993). This is consistent with other studies (Lian and Meletis, 1996) and with amalgam being classied as a brittle material (e.g., Phillips, 1991; Sturdevant et al., 1995). Although sharp indentation produced circumferential cracks in polished silver, the cracks were absent in burnished silver (Fig. 3c). In cyclic Hertzian indentation, microcracks were produced in amalgam both on the surface and beneath the surface. These microcracks were likely formed due to cyclic shear stresses during cyclic loading and unloading (Guiberteau et al., 1993, 1994; Cai et al., 1994; Xu et al., 1995). These microcracks in amalgam would be anticipated as the multiple phases present pin dislocations creating stress concentrations at interphase boundaries. The cracks produced in cyclic fatigue are consistent with cracks observed in the Vickers indentation toughness measurement of amalgam. In contrast, no cracks could be found in polished or burnished silver at numbers of cycles up to 5 10 5. This suggests that the silver experimental material possesses a greater resistance to cracking than amalgam, both in sharp indentation fracture toughness measurement and in cyclic Hertzian contact fatigue. Cyclic contact response. The contact response of a material can be characterized by two parameters: the contact impression size, and the nature of deformation and damage within and beneath the impression (e.g., cracking, work-hardening, etc.). The contact impression size produced by cyclic indentation is determined by two factors: the hardness of the material, and the fatigue resistance of the material. While the hardness determines the impression size after the rst cycle, the fatigue resistance of the material determines the rate of
18 Xu et al./Fatigue of a silver alternative to amalgam

increase in the impression size with increasing number of indentation cycles. The results illustrated in Figs. 5 and 6 show that with increasing number of cycles, the impression size in amalgam specimens increased more rapidly than in silver. This indicates that the silver material possesses a higher resistance to contact fatigue than amalgam. It should be noted that the fatigue response of materials may show a frequency dependence, and further studies should address whether the deformation behavior will be different if a slower frequency (e.g., 2 Hz) is used. Other characteristics of the cyclic contact response of a material are its deformation and damage behavior. Microcracks were observed in the contact impressions in amalgam. These microcracks could lead to microfracture and dislodgement of material (Guiberteau et al., 1994; Xu et al., 1995, 1996). This may explain the clinical observations of ditching and chipping in amalgam restorations, especially at margins (Phillips, 1991; Sturdevant et al., 1995). Similar to many brittle materials, amalgam deteriorates during prolonged use due to damage accumulation. In contrast, the properties of silver are improved by use as a result of work-hardening. After 5 10 5 contact cycles, the average hardness of the polished silver beneath the contact area was increased by more than 40%. This, together with the absence of microcracks associated with cyclic contact in silver, suggests that the clinical ditching and chipping problems associated with amalgam could be less severe in the silver material. Long-term clinical studies are needed to verify this point. Burnishing. The benet of burnishing the surface of silver is twofold. First, it substantially hardens the surface layer of silver, rendering it more resistant to plastic deformation as well as to crack formation by sharp indentation. In cyclic Hertzian indentation, burnishing resulted in much smaller impression sizes in silver. This again is a result of its enhanced resistance to plastic deformation. Second, burnishing was shown in this study to signicantly improve the exural strength of silver when the burnished surface is in tension. This is likely a result of the increased density (i.e., reduced porosity in surface, Fig. 2c, and subsurface, Fig. 9c) as well as the compressive residual stresses due to dislocation accumulation induced by burnishing. Burnishing can be performed in a short time (approximately 3 min to burnish an area of 192 mm 2, or 9 s for a square of 3 3 mm). Therefore, the results of this study would seem to recommend burnishing as a nishing step in the clinical applications of this silver material. Finally, it should be noted that the cyclic Hertzian indentation technique was found to be a simple and useful fatigue test. It provides information that is expected to be clinically more relevant than conventional fatigue tests involving tensile macroscopic fracture of the entire specimen. Improvements, however, can still be made. Cyclic sliding or scratching experiments (Xu and Jahanmir, 1995a,b) may better simulate the clinical activities of dental restorations compared to cyclic indentation. Separate studies are

underway that include a cyclic sliding test to investigate the sliding fatigue response of this silver material in comparison with amalgam, as well as a three-body wear test to evaluate the wear resistance of the consolidated silver as a function of wear cycles.

DISCLAIMER
Certain commercial materials and equipment are identied in this paper to specify the experimental procedure. In no instance does such identication imply recommendation or endorsement by the National Institute of Standards and Technology or the ADA Health Foundation or that the material or equipment identied is necessarily the best available for the purpose.

ACKNOWLEDGMENTS
The authors are grateful to Hongyan Liao for help in statistical analysis, to Brian R. Lawn for discussions and lab facilities, and to Irene M. Peterson and Douglas T. Smith for experimental assistance. This study was supported by NIDR IA 1Y01 DE30045, NIDR grant P 50 DE09322, NIST and the ADAHF.
Received January 6, 1997 Address for correspondence and reprint requests to: H.H.K. Xu American Dental Association National Institute of Standards & Technology Paffenbarger Research Center Building 224, Room A-153 Gaithersburg, MD 20899, USA. Tel: 001-301-975-6804; Fax: 001-301-963-9143; E-mail: hockin.xu@nist.gov

REFERENCES
Anstis, GR, Chantikul, P, Lawn, BR, Marshall, DB (1981). A critical evaluation of indentation techniques for measuring fracture toughness: I, direct crack measurements. J Am Ceram Soc 64 (9):533 538. ASTM Designation F417-78 (1984). Standard test method for exural strength of electrical grade ceramics. Philadelphia, PA: American Society for Testing and Materials. ASTM Designation E 384 (1991). Standard test method for microhardness of materials, part 3 in book of ASTM standards. Philadelphia, PA: American society for testing and materials. Ausavice, KJ (1996). Challenges to the development of esthetic alternatives to dental amalgam in an academic research center. Pro Academy Dent Mater 9:2550. Bayne, SC, Heymann, HO, Swift, EJ (1994). Update on dental composite restorations. J Am Dent Assoc 125:687701.

Cai, H, Kalceff, MAS, Hooks, BM, Lawn, BR (1994). Cyclic fatigue of a mica-containing glass-ceramic at Hertzian contacts. J Mater Res 9 (10):26542661. Dariel, MP, Admon, U, Lashmore, DS, Ratzker, M, Giuseppetti, A, Eichmiller, FC (1995). A silver alternative to dental amalgam. J Mater Res 10:505511. Dariel, MP, Lashmore, DS, Ratzker, M (1995). A new technology for direct restorative alloys. Dent Mater 11:208217. Ferracane, JL (1989). Indentation fracture toughness testing of and crack propagation mode in dental composites. Mat Res Soc Symp Proc 110:619624. Ferracane, JL (1992). Using posterior composites appropriately. J Am Dent Assoc 123:5358. Fields RJ, Harne DE, Fields BA (1984). Fracture toughness of a steel matrix, titanium carbide composite. In: Early JG, Shives TR, Smith JH, editors. Proceedings of the 39th meeting of the mechanical failures prevention group. London: Cambridge University Press. Fisher-Cripps, AC, Lawn, BR (1996). Indentation stressstrain curves for quasi-ductile ceramics. Acta Metall Mater 44:519527. Guiberteau, F, Padture, NP, Cai, H, Lawn, BR (1993). Indentation fatigue: a simple cyclic Hertzian test for measuring damage accumulation in polycrystalline ceramics. Phil Mag A 68 (5):10031016. Guiberteau, F, Padture, NP, Lawn, BR (1994). Effect of grain size on Hertzian contact damage in alumina. J Am Ceram Soc 77:18251831. Hertz H (1896). Hertzs miscellaneous papers, chapters 5 and 6. London: Macmillan. Lashmore DS, Dariel MP, Johnson CE, Ratzker M, Giuseppetti A, Eichmiller FC, Beane GL, Kelley DR (1994). Acid assisted cold welding and intermetallic formation and dental application thereof. US Patent application 08/317,719. Lawn BR (1993). Fracture of brittle solids, 2nd ed. London: Cambridge University Press. Leinfelder, KF (1994). After amalgam, what? Other materials fall short. J Am Dent Assoc 125:586589. Lian, K, Meletis, EI (1996). Corrosion of amalgams under sliding wear. Dent Mater 12:146153. Macker, JR Jr. (1991). Dental amalgam and mercury. J Am Dent Assoc 122:5461. Mandel, ID (1991). Amalgam hazards. J Am Dent Assoc 122:6265. Marshall, GW, Jackson, BL, Marshall, SJ (1980). Copper-rich and conventional amalgam restorations after clinical use. J Am Dent Assoc 100:4347. McCabe, JF, Carrick, TE (1987). Dynamic creep of dental amalgam as a function of stress and number of applied stress cycles. J Dent Res 66:13461349. Morena, R, Beaudreau, GM, Lockwood, PE, Evans, AL, Fairhurst, CW (1986). Fatigue of dental ceramics in a simulated oral environment. J Dent Res 65 (7):993997. Mueller HJ (1998). Tarnish, corrosion, and electrochemical considerations of a consolidated
Dental Materials/January 1998 19

silver particle material. J Dent Res, AADR abstract # 108. Osborne, JW, Norman, RD, Gale, EN (1991). A 14-year clinical assessment of 12 amalgam alloys. Quintessence Int 22:857863. Oshida, Y, Moore, BK (1993). Anodic polarization behavior and microstructure of a gallium-based alloy. Dent Mater 9:234241. Phillips RW (1991). Science of dental materials, 9th ed., pp. 318319. Philadelphia: WB Saunders Company. Sturdevant CM, Roberson TM, Heymann HO, Sturdevant JR (1995). Operative dentistry, 3rd ed., p. 227. St Louis, MO: Mosby. Sutow, EJ, Jones, DW, Hall, GC, Milne, EL (1985). The response of dental amalgam to dynamic loading. J Dent Res 64:6266. Twiggs, SW, Fairhurst, CW, Lockwood, PE, Ringle, RD (1995). Cyclic fatigue of a model feldspathic porcelain. Dent Mater 11:273276. Watkins, JH, Nakajima, H, Hanaoka, K, Zhao, L, Iwamoto, T Okabe, T, (1995). Effect of zinc on strength and fatigue resistance of amalgam. Dent Mater 11:2433.

White, SN, Zhao, XY, Li, ZC, Kipnis, V (1995). Cyclic mechanical fatigue of a feldspathic dental porcelain. Int J Prosthodont 8 (5):413420. Xu, HHK, Jahanmir, S (1995). Microfracture and material removal in scratching of alumina. J Mater Sci 30:22352247. Xu, HHK, Jahanmir, S (1995). Scratching and grinding of a machinable glass-ceramic with weak interfaces and rising T-curve. J Am Ceram Soc 78:497500. Xu, HHK, Smith, DT, Jahanmir, S (1996). Inuence of microstructure on indentation and machining of dental glass-ceramics. J Mater Res 11 (9):2325 2337. Xu, HHK, Wei, L, Padture, NP, Lawn, BR, Yeckley, RL (1995). Effect of microstructural coarsening on Hertzian contact damage in silicon nitride. J Mater Sci 30:869878. Zardiackas, LD, Bayne, SC (1985). Fatigue characterization of nine dental amalgams. Biomaterials 6:49 54

20 Xu et al./Fatigue of a silver alternative to amalgam