d e n t a l m a t e r i a l s 3 0 ( 2 0 1 4 ) 381–391

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.intl.elsevierhealth.com/journals/dema

Are linear elastic material properties relevant predictors of the cyclic fatigue resistance of dental resin composites? Renan Belli ∗ , Anselm Petschelt, Ulrich Lohbauer Laboratory for Biomaterials Research, Dental Clinic 1, Operative Dentistry and Periodontology, Friedrich-Alexander-University of Erlangen-Nürnberg, Glueckstrasse 11, D-91054 Erlangen, Germany

a r t i c l e

i n f o

a b s t r a c t

Article history:

Objective. The aim of this study was to measure the linear elastic material properties of direct

Received 26 March 2013

dental resin composites and correlate them with their fatigue strength under cyclic loading.

Received in revised form

Methods. Bar specimens of twelve resin composites were produced according to ISO 4049 and

30 August 2013

tested for elastic modulus (Emod) in 3-point bending (n = 10), flexural strength (FS) (n = 15)

Accepted 14 January 2014

and single-edge-notch-beam fracture toughness (FT) (n = 15), both in 4-point bending. Using the same specimen geometry, the flexural fatigue strength (FFS) was determined using the staircase approach after 104 cycles at 0.5 Hz in 4-point bending (n = 25). The observation of the

Keywords:

fracture surface and fracture profiles was conducted using a scanning electron microscope in

Resin composite

order to evaluate the respective fracture mechanisms according to the two different loading

Fracture toughness

conditions.

Flexural strength

Results. Materials were ranked differently according to the tested parameters. Only weak

Fatigue

correlations were found between any of the initial properties and FFS or strength loss. The

Staircase approach

best correlation to FFS was found to be the Emod (r2 = 0.679), although only slightly. Crack path in both loading conditions was mainly interparticle, with the crack propagating mainly within the matrix phase for fatigued specimens and eventually through the filler/matrix interface for statically loaded specimens. Fracture of large particles or prepolymerized fillers was only observed in specimens of FS and FT. Initial properties were better associated with microstructural features, whereas the fatigue resistance showed to be more dependent on aspects relating to the matrix phase. Significance. Our results show that linear elastic properties such as elastic modulus, flexural strength and fracture toughness are not good descriptors of the fatigue resistance of dental resin composite under cyclic bending, and may therefore have limited clinical relevance. © 2014 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved.

1.

Introduction

Predictions of material performance based on mechanical properties are anchored in the theory of correlation, a statistical tool that describes the degree to which two sets of data

∗

show any association. Depending on the data distribution, high correlation coefficients may characterize a high functional dependency, express susceptibility or causality between independent variables. Mechanical properties such as fracture toughness, surface hardness and elastic modulus were found to have a defined relationship to the wear resistance of

Corresponding author. Tel.: +49 9131 854 3741; fax: +49 9131 853 4207. E-mail address: [email protected] (R. Belli). 0109-5641/$ – see front matter © 2014 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.dental.2014.01.009

382

d e n t a l m a t e r i a l s 3 0 ( 2 0 1 4 ) 381–391

resin-based composites [1]. This theoretically enables the prediction of clinical wear rates of composites based on experimentally obtained parameters. By increasing the predicability of in vitro data, costly and time-consuming clinical studies for material performance loose urgency and material development processes become more dynamic. Traditionally, the characterization of the mechanical behavior of resin composites starts with the measurement of intrinsic properties under a condition of uniform increase in load (static loading). The most relevant of these properties concern the stiffness (elastic modulus) and those involving the resistance to fracture (i.e., strength and fracture toughness). For the measurement of both strength and toughness a high amount of energy is applied to cause fast fracture, resulting in parameters that define the state of a material in a point in time. Under these conditions, material behaviors liable to dynamic or cyclic loading are poorly described. That is especially true for materials that present a high susceptibility to fatigue, where the mechanical properties are ever-changing by influence of environmental and stress-related phenomena. For fatigue-resistant materials, measurements of strength and fracture toughness serve to characterize their mechanical state at different periods even after long loading challenges. Relationships between clinical performance and mechanical parameters in dental resin composites are complex, since both derive from scenarios of very different loading conditions. Under function resin composite restorations are loaded periodically, in which the dynamic mechanical behavior of the polymer matrix and water-/stress-degradation processes play a significant role. The viscoelastic effects occurring in the resin during in vivo cyclic fatigue are absent in specimens tested under high strain rates, where the release of high elastic energy levels results in brittle-like fracture patterns [2]. Also in vitro, factors related to test design and specimen geometry add to the apparent discrepancy to in vivo data [3–5]. Strength measurements for example, naturally amplify the role of surface flaws on fracture initiation, whereas clinical wear tends to reshape the surface and introduce new flaw populations to the loaded restoration. Such complexities are in part disregarded in the literature, and although no concrete correlations between mechanical properties measured under static loading and the fatigue resistance of resin composites have been ever established [6], inferences on the latter based on the former have become commonplace. For other restoratives such as glass–ceramics, the in vivo clinical performance has shown to be better characterized by in vitro fatigue parameters that consider time-dependent processes involved in their failure [7]. Conversely to the measurement of strength and fracture toughness, where failure results from the unstable propagation of a crack of critical length, the mechanical degradation, or “fatigue”, of a material involves the growth of subcritical defects at subcritical loads. The fatigue susceptibility can be assessed using parameters defined by fracture mechanics, such as the stress intensity factor, by measuring its development during crack extension at consecutive cycle intervals. Outside the fracture mechanics realm, phenomenological approaches base their evaluation on the residual strength after fatigue loading, where endurance ranges can be extracted and compared to clinically relevant stress amplitudes.

In the present study we selected for testing twelve commercial direct resin composites that represent the current state of the art regarding matrix composition and filler technology. These materials were tested for inherent mechanical properties (e.g., elastic modulus, fracture toughness and flexural strength) and fatigue resistance after 104 cycles using the staircase approach. The tested null-hypotheses were that (i) there is no correlation between the inherent properties and the fatigue resistance or strength loss (in %) in resin composites and (ii) there is no correlation between elastic modulus, fracture toughness and strength in these materials.

2.

Materials and methods

2.1.

Materials

Twelve currently commercially available resin composite materials for direct restorations were selected for testing. In Table 1 details about their filler content, filler configuration and matrix composition can be found. This selection of composites included those with novel monomer formulations (Filtek Silorane, Venus Diamond and Kalore), those containing prepolymerized fillers (Kalore, Tetric EvoCeram, Miris 2, Clearfil Majesty Esthetic) and a mixture of glass particles and long glass fibers for the use as a base material (Xenius Base). Apart from the nanofilled Filtek Supreme XT, most are nanohybrid materials, with an inorganic filler content ranging from 54 vol% (Tetriv EvoCeram) to highly filled systems, such a Grandio SO (73 vol%) and Clearfil Majesty Posterior (82 vol%).

2.2.

Specimen preparation

Sixty-five bending bars (2 mm × 2 mm × 25 mm) were produced for each material using a tungsten carbide/glass mold under calibrated conditions of 23 ◦ C and 50% humidity. The materials were inserted in the mold in one previously weighted increment, and light-cured with five overlapping spots of ∅ = 8 mm on both upper and lower sides with a commercial light-curing unit (Elipar® Trilight, 3M ESPE, Germany) with an output intensity of 800 mW/cm2 for the time recommended by the manufacturers (20 s per spot for all materials for the curing intensity range used). For Xenius Base, the material was inserted in the mold out of the cannula so that the parallel glass fibers were oriented in the long axis of the specimen. The specimen fabrication procedures followed the manufacturers’ recommendations and ISO 4049 standard [8]. After removal from the mold, the specimens were ground in the lower side with silicon carbide paper of down to 4000 grit to remove surface flaws that could influence strength values [9]. The specimens were stored in distilled water at 37 ◦ C for 24 h before the fracture toughness and elastic modulus testing and for 14 days before testing the initial flexural strength and the flexural fatigue resistance.

2.3.

Elastic modulus (Emod) testing

To determine the elastic modulus (Emod) (n = 10), the stress–strain curves were obtained between 10 and 40 MPa for each specimen by loading them in a 3-point bending test

Table 1 – Name of the composites tested in this study, their manufacturer, filler content by volume, filler size, matrix composition and batch number. Composite

Manufacturer

Filler content (vol%)a

Filler configuration

Kuraray, Tokyo, Japan

82

0.5–10 ␮m particles; 20 nm particles

Grandio SO (GR)

Voco, Cuxhaven, Germany

73

Filtek Silorane (SI)

3M ESPE, St. Paul, USA

55

0.5–3 ␮m particles; 20–40 nm nanoparticles 0.04–1.7 ␮m particles

Filtek Supreme XT (SU)

3M ESPE, St. Paul, USA

59.5

Tetric EvoCeram (EV)

Ivoclar-Vivadent, Schaan, Liechtenstein

55

Miris II (M2)

Coltène, Altstätten, Switzerland

65

Venus Diamond (VD)

Heraeus, Hanau, Germany

64

Kalore (KA)

GC, Tokyo, Japan

69

Xenius Base (XE)

Stick Tech, Turku, Finland

53.6

EsthetX (ES)

Dentsply, Konstanz, Germany Kuraray, Tokyo, Japan

60

Ivoclar-Vivadent, Schaan, Liechtenstein

58

Clearfil Majesty Esthetic (CLE)

Tetric Ceram (TC)

66

Nanoparticles 20 – 70 nm; nanoclusters of 0.01–3.5 ␮m with 20 nm particles 0.5 ␮m mean particle size; 5–50 ␮m prepolymerized fillers containing 0.4–0.7 ␮m fillers 0.2–2 ␮m particles; 20 nm nanoparticles; 10–50 ␮m prepolymerized fillers 0.05–1 ␮m particles; 5 – 20 ␮m irregular particles; 20 nm nanoparticles 0.7 ␮m particles, 16 nm nanoparticles; 17 ␮m prepolymerized fillers of 0.1–0.4 ␮m particles 0.1–2.2 ␮m particles; 1–2 mm glass fibers 0.02–2.5 ␮m particles; 20 nm nanoparticles Particles with mean size of 1.5 ␮m; 20 nm nanoparticles; 5–40 ␮m prepolymerized fillers containing nanoparticles 0.04–1 ␮m particles

Matrix composition

Batch

Bis-GMA, TEGDMA, hydrophobic aromatic dimethacrylates Bis-GMA, Bis-EMA, TEGDMA 3,4-Epoxycyclohexylethylcyclopolymethylsiloxane, bis-3,4-epoxycyclohexylethylphenylmethylsilane Bis-GMA, Bis-EMA, UDMA, TEGDMA

134468

Dimethacrylates

P67534

Bis-GMA-based methacrylates

D56212

Tricycledecane-urethane dimethacrylate, UDMA

010029

UDMA, DX-511 monomer, dimethacrylates

906021

PMMA, Bis-GMA, TEGDMA

XB1003

UDMA, Bis-EMA, TEGDMA

011227

Bis-GMA, TEGDMA, hydrophobic aromatic dimethacrylates

00057A

Bis-GMA, UDMA, TEGDMA

G01410

00111B

1207326

040028 d e n t a l m a t e r i a l s 3 0 ( 2 0 1 4 ) 381–391

Clearfil Majesty Posterior (CLP)

All composites tested were from shade A3 (S3 for Miris II and A3B for Filtek Supreme XT). Filler contents were obtained from the manufacturers.

a

383

384

d e n t a l m a t e r i a l s 3 0 ( 2 0 1 4 ) 381–391

Fig. 1 – (a) Diagram of the test set-up for the fracture toughness and (b) a random sequence of 4 specimens under cyclic loading according to the staircase approach, where the first specimen survives and the subsequent two fail. In (b) “d” represents the stress increment.

set-up at a loading rate of 0.75 mm/min (Zwick Z2.5, Zwick, Germany) according to the relation Emod = d/dε = const, where  is the stress and ε is the strain. A precision extensometer (Zwick, Germany) was used on the lower side of the specimens to record the displacement.

2.4.

Fracture toughness (FT) testing

After the storage period, a notch at the midspan of the lower side of the specimens was fabricated with a low-speed diamond saw (150 ␮m thick) set-up (Buehler, USA) and sharpened with a razor blade and 3 ␮m diamond suspension for testing according to the so-called single-edge-notched beam method (SENB) (n = 15). The notch depth:specimen height ratio was controlled within 0.2 and 0.3, according to DIN CEN/TS 14425-5. The specimens were fixed in the test rig (fin distance: 10/20 mm, d = 2 mm) in a 4-point bending configuration (Zwick Z2.5, Zwick, Ulm, Germany) and loaded until fracture with a crosshead speed of 0.75 mm/min (Fig. 1a). The notch was placed in the center of the test rig and set under tensile load. The notch depth of each specimen was measured post-fracture using an optical light microscope (SV6, Zeiss, Germany) using the average of the values obtained in three locations for both fragments. The toughness was calculated according to the equation below [10]:

KIc =

Fc √ Y B W

(1)

where Fc is the critical load, B is the specimen width, W is the specimen height and Y is the stress intensity shape factor, defined by:

Y =

√ L−l 3 ˛ W 2(1 − ˛)3/2



× 1.9887 − 1.1326˛ −

(3.49 − 0.68˛ + 1.35˛2 )˛(1 − ˛) (1 − ˛)

2

2.5. Initial flexural strength (FS) and flexural fatigue strength (FFS) testing For the initial flexural strength (FS) measurements, 15 specimens per material were loaded in 4-point bending at 0.75 mm/min in an universal testing machine (Z2.5, Zwick). The initial flexural strength () was calculated using the relation [11]: =

3Pd wb2

(3)

where P is the maximum load at failure, d is the distance between the midpoint of the lower supports and the upper supports (5 mm), w is the width and b is the height of the specimen. The initial strength data was further analysed according to a two-parameter Weibull distribution in terms of a shape (m) and scale parameters  0 and  −2.97 . The m modulus describes the scatter in stress to failure, and parameters  0 and  −2.97 indicate their magnitude and are given for a failure probability of 63.2% and 5%, respectively. The Weibull modulus m, the characteristic strength  0 and their respective 95% confidence intervals were corrected by a factor corresponding to the number of specimens (n = 15), according to the European standard EN 843-5 [12]. The flexural fatigue strength (FFS) of the composite materials was determined for 104 cycles under equivalent test conditions at a frequency of 0.5 Hz (n = 25). The “staircase” approach method [13,14] was used for the fatigue resistance evaluation. For each cycle, the stress alternated between 1 MPa and maximum stress. Tests were conduced sequentially, with the maximum applied stress in each succeeding specimen being increased or decreased by a fixed increment of stress, according to whether the previous 104 cycles run resulted in failure or survival. The first specimen was tested at 50% level of the initial mean flexural strength value. All tests were carried out in distilled water at 37 ◦ C. The FFS and standard deviation (SD) were determined using Eqs. (3) and (4), respectively;

 (2)

FFS = X0 + d

  in  i ± 0.5

  where L is the distance of the support fins, l is the distance of the loading fins and ˛ = a/W, being a is the notch depth.

SD = 1.62d

(4)

ni

ni

i2 ni −



 2 ni

ini

2

+ 0.029

(5)

385

d e n t a l m a t e r i a l s 3 0 ( 2 0 1 4 ) 381–391

Table 2 – Mean and standard deviations () of fracture toughness, elastic modulus, flexural strength, flexural fatigue limit (FFS) after 104 cycles and strength loss (in %). Composite

Fracture toughness (MPa m0.5 )

E-modulus (GPa)

Clearfil Majesty Posterior Grandio SO Filtek Silorane Filtek Supreme XT Tetric EvoCeram Miris II Venus Diamond Kalore Xenius Base EsthetX Clearfil Majesty Esthetic Tetric Ceram

2.21 (0.19)b 1.85 (0.16)c 1.96 (0.17)c 2.13 (0.14)b 1.49 (0.13)d 1.40 (0.15)de 2.59 (0.31)a 1.29 (0.08)e 2.17 (0.21)b 2.02 (0.22)bc 1.24 (0.08)e 2.18 (0.17)b

19.18 (1.3)a 14.64 (1.7)b 10.18 (0.26)d 8.99 (0.35)e 8.26 (0.24)f 8.35 (0.35)ef 8.73 (0.95)ef 6.02 (0.4)h 12.19 (0.9)c 10.11 (0.37)d 6.99 (0.31)g 8.67 (0.25)ef

Flexural strength  (MPa) 119.78 (24.2)b 139.20 (13.6)a 112.37 (11.6)b 85.12 (9.1)cd 77.34 (6.0)d 117.36 (9.2)b 130.12 (11.0)a 93.82 (13.4)c 110.68 (9.6)b 96.39 (14.67)cd 70.43 (7.59)e 101.46 (13.16)b

FFS 104

81.4 (8.01)a 60.5 (3.17)b 62.6 (6.71)b 35.85 (3.72)d 42.30 (2.45)c 44.75 (5.16)c 46.30 (6.34)c 44.12 (5.83)c 41.18 (4.39)c 45.37 (3.64)c 41.01 (3.66)c 44.33 (4.45)c

Strength loss %

32.04 56.54 44.29 57.90 45.31 61.87 64.42 52.97 62.79 51.02 41.77 57.13

The same letters in columns indicate same statistical subsets at ˛ = 0.05.

where X0 is the lowest stress level considered in the analysis and d is the fixed stress increment. To determine FFS, the analysis of the data is based on the least frequent event (failures versus survivals). In Eq. (4), the negative sign is used when the analysis is based on failures, otherwise the positive sign is used. The lowest stress level considered is designated i = 0, the next i = 1, and so on, and ni is the number of failures or survivals at the given stress level. A cross-section of the cyclic loading profile for the fatigue experiment used in the staircase approach for a given hypothetical sequence of 4 specimens is illustrated in Fig. 1b.

After fracture, two specimens for each group from FS, FT and FFS tests were sputter-coated with gold and observed at the fracture surfaces with a scanning electron microscope at 20 kV (SEM) (Leitz ISI SR50, Akashi, Japan) for representative fractographic markings. Two additional specimens for FS and FFS testing were produced for each material and polished at the tensile side until 4000 grit SiC paper for observation in the SEM (back-scattered mode) at the site of fracture to determine crack propagation paths (interparticle or intraparticle).

2.6.

3.

Statistical analysis

The Emod, FT, and FFS were analysed using a one-way ANOVA test and modified post hoc LSD test (p < 0.05). The initial FS was analysed using a two-parameter Weibull distribution, in which the 95% confidence intervals for the groups were computed and differences were considered significant when the 95% confidence intervals did not overlap. Correlation analysis among the tested parameters, strength loss and filler loading were also computed.

2.7.

Fractographic analysis

Results

In Table 2 the values and standard deviations for FT, Emod, FS, FFS and strength loss are summarized. Strength loss in percentage was calculated in relation to the FFS after 104 cycles and the average value of FS calculated using Eq. (3). Table 3 depicts the Weibull distribution parameters for the specimens tested in flexure in 4-point bending. The rank of materials changed according to the measured property, as can be observed in Table 2. Usually, materials that displayed a

Table 3 – Weibull scale ( 0 and  −2.97 ) and shape (m) parameters for initial strength and 95% confidence intervals (Cl /Cu ) and (Dl /Du ) for  0 and m, respectively. Composite Clearfil Majesty Posterior Grandio SO Filtek Silorane Filtek Supreme XT Tetric EvoCeram Miris II Venus Diamond Kalore Xenius Base EsthetX Clearfil Majesty Esthetic Tetric Ceram

Weibull  −2.97 (MPa) 76.64 113.71 90.37 89.80 82.61 99.98 108.38 68.96 92.33 69.74 55.34 76.33

Weibull  0 (MPa) 129.67ab 145.11a 117.45bc 109.94cd 92.5e 121.39b 135.19a 99.53de 114.94bc 102.57cde 74.13f 109.27bcd

The same letters in columns indicate same statistical subsets at ˛ = 0.05.

Cl /Cu  0 115.49–145.54 137.54–153.08 110.87–124.40 105.15–114.93 90.20–94.85 116.52–126.44 128.78–141.90 91.12–108.78 109.72–120.38 91.76–115.03 69.52–79.04 100.98–118.21

Weibull m 5.12c 11.05abc 10.28abc 13.32ab 23.58a 13.98ab 12.2abc 7.39bc 12.33abc 6.56bc 9.22bc 7.51bc

Dl /Du m 3.2–7.8 7.0–17.0 6.5–15.8 8.4–20.4 15.0–36.2 9.0–21.1 7.7–18.7 4.4–11.5 7.9–18.6 3.6–11.3 5.8–14.1 4.7–11.4

386

d e n t a l m a t e r i a l s 3 0 ( 2 0 1 4 ) 381–391

Fig. 2 – SEM images of the fractures surface of specimens under static loading. (a and b) The microstructural fracture pattern of Clearfil Majesty Posterior (1000× and 3000×, respectively); white arrows indicate particles covered with resin, (c and f) the polished side of a fractured specimen of Tetric EvoCeram under back-scattered mode (both 1000×). In (c) the crack is deflected by a prepolymerized filler (ppf) (dcp, direction of crack propagation), whereas in (f) the white arrow points to a fractured ppf, and (d and e) the microstructural fracture pattern of Grandio SO (1000× and 3000×, respectively).

high fracture toughness did not necessarily showed high flexural strength, with exception to Venus Diamond and Clearfil Majesty Posterior, which showed high values for both properties. The highest FT value was obtained by Venus Diamond, which was statistically different than all other materials. The materials containing prepolymerized fillers (Clearfil Majesty Esthetic, Kalore, Tetric EvoCeram and Miris 2) showed the lowest values for FT. Despite the high characteristic strength shown by Clearfil Majesty Posterior, it also resulted in the highest scatter (the lowest m-value). The most homogeneous material was Tetric EvoCeram, with a m-value of 23.5. The stiffest materials were Clearfil Majesty Posterior and Grandio SO, which contain filler contents of 76 vol% and 82 vol%, respectively. After a fatigue challenge of 104 cycles all materials showed a significant drop in strength, represented by the FFS value. This value of residual strength indicates the range of stress below which a specimen could endure 104 cycles without fracture. From the values of FFS in Table 2, four distinct ranges of residual strengths can be distinguished. The highest residual strength (81.4 MPa) after cyclic loading was obtained for Clearfil Majesty Posterior, which was significantly higher than all other materials. The second highest interval of residual strength was found between 60 and 63 MPa, for Filtek Silorane and Grandio SO. Filtek Supreme XT showed the significantly lowest residual strength (35.8 MPa), while the other materials remained in the range between 41 and 47 MPa. Since the reference is the initial strength, the fatigue resistance is better represented by the strength loss rather than by the FFS value. A look at the percentage strength loss identifies also four distinct ranges. Isolated as the most fatigue resistant material

was Clearfil Majesty Posterior, showing a strength loss of only 32%. Clearfil Majesty Esthetic, Filtek Silorane and Tetric EvoCeram form the group that shows a strength loss around 42–45% after cyclic loading. Grandio SO, which showed the second to third highest FFS values lost 56.5% of flexural strength after the fatigue challenge, a similar value obtained by Kalore (53%), EsthetX (51%) and Filtek Supreme XT (57.9%). The nanofilled material, which showed the lowest FFS value, ranked 8th in fatigue resistance. Miris 2, Xenius Base and Venus Diamond were the least fatigue resistant materials, showing a strength loss between 61.8% and 64.4%. Venus Diamond, the material that showed the highest fracture toughness and the second highest flexural strength was the least fatigue resistant material. Fractographic findings could establish a markable difference in macroscopic fracture patterns and microscopic crack mechanisms between loading conditions and among materials. Specimens tested under static loading showed clear markings of fast fracture, with evident surface roughness and multiple crack planes. Fracture origin, mist and hackle region could be identified in the specimens that tested the initial flexural strength. Specimens that fractured during the cyclic loading experiment produced a smooth fracture surface where none of the typical brittle failure marking could be seen. At the microstructure level, fractures under static loading showed a crack path involving the matrix, occasionally through the filler/matrix interface as well as fracture of prepolymerized fillers (Fig. 2). Interparticle fracture path (cracking deflection by particle) was observed as the dominant crack mechanism for all composites under static loading as well as under cyclic loading. Particle fracture was observed only on specimens of

d e n t a l m a t e r i a l s 3 0 ( 2 0 1 4 ) 381–391

387

Fig. 3 – SEM images of fractured specimens of Venus Diamond under static loading. (a) The fracture surface, where the black arrow points to a particles which surface is covered with a thin layer of resin, and the white arrow indicates a particle not covered with resin with signs of brittle fracture (1000×) and (b) back-scattered image of the polished size of a fractured specimen where a crack fractures a 30 ␮m-sized particle (500×). The magnification of (b) shows in detail a crack bridging toughening mechanism in play formed by secondary crack branches and the uncracked composite ligament (2000×).

Venus Diamond in particles bigger than 10 ␮m (Fig. 3a). Interfacial fracture were observed sporadically in specimens tested under static loading, whereas patterns of cyclic loaded specimens showed the crack path involving mainly the resin matrix phase (Fig. 4). Larger particles and particle clusters have shown to deflect the crack at a zone near the interface, while keeping the particle covered with a thin layer of resin (Fig. 4b and e). Cracks growing under cyclic fatigue tended to deflect around prepolymerized fillers instead of causing fracture (Fig. 4c and f), as occurred under static loading. Toughening mechanisms such as crack bridging, difficult to observe on the fractures surface of specimens, were only detected at the polished tensile side of fractured FS specimens of Venus Diamond (Fig. 3b).

The fractured profile of other composites presented, as a rule, crack deflection around particles for FS and FFS specimens. Correlations between inherent mechanical properties and FFS or strength loss were all poor. Fig. 5 shows some of these correlations, where the best correlation to FFS being the Emod (r2 = 0.679). Correlations to the filler vol% content were also weak (r2 < 0.45).

4.

Discussion

An approach for the assessment of the fatigue resistance is to measure the inherent properties of a material after cyclic

Fig. 4 – SEM images of the fracture surfaces of specimens fractured under cyclic loading. (a and b) The microstructural fracture pattern of Filtek Supreme XT (1000× and 4000×, respectively); the white arrow indicates a intact nanofiller cluster, and the black arrow the concavity of a cluster, (c) intact prepolymerized fillers (ppf) of Miris 2; the white arrow point to the interface containing rests of resin (2000×), (d and e) the microstructural fracture patterns of Clearfil Majesty Posterior; the white arrow point to a filler covered with a thick resin layer (1000× and 2000×, respectively) and (f) the fracture surface of Tetric EvoCeram in which the crack deflects around ppfs instead of causing their fracture; the white arrow points to the negative of a detached ppf (1000×).

388

d e n t a l m a t e r i a l s 3 0 ( 2 0 1 4 ) 381–391

Fig. 5 – (a) Correlation between flexural fatigue strength and strength loss against the elastic modulus and (b) correlation between flexural fatigue strength and characteristic strength against fracture toughness.

loading, describing the mechanical degradation in terms of property loss. The staircase approach used herein follows that strategy and provides the residual strength after a defined number of cycles. This test is known to provide a good estimation of the mean fatigue strength, although an underestimation of the strength scatter can be expected [15]. A simple assessment of the strength loss (here presented in percentage) results from the difference between the mean initial strength and mean residual strength after cyclic loading. Regardless of recent advances in monomer and filler technologies, the tested resin composites showed a high susceptibility to fatigue (substantial strength loss), confirming previous results [6,16–18]. The measurement of flexural strength and fracture toughness showed that fatigue is poorly correlated to these initial properties. Among themselves, inherent properties of dental resin composites also showed no correlation, following the trend of natural and engineered materials [19]. Both null-hypotheses are therefore confirmed. The plotting of elastic modulus, flexural strength and fracture toughness against volume filler content also yielded a poor correlation. The range of filler volume in the materials tested here (55–82 vol%) represents the higher end of the packing spectrum, within which the effects of filler content already reach a peak or also begin to be detrimental [4,20,21]. In model and commercial composites, tests with a wider range of filler volume content showed an exponential [22–24] and a linear [25] relationship for elastic modulus and flexural strength, respectively. The increase in elastic modulus and strength with filler content measured experimentally indicates some agreement with empirical and semi-empirical relations for composites with well bonded particles that assume stress transfer between matrix and particles [26]. To a certain level that relationship holds true, but filler contents higher than 61.7 vol% were reported by Ikejima et al. [20] to cause a slight decrease in strength for a micro-hybrid system. For fracture toughness, a threshold has been found at 55–57 vol% [21,24] and a slight decrease after 65 vol% [21]. Due to the high amplitude of filler size distributions used in commercial composites, the effect of filler size apparently shows a minor role in flexural strength, with nanofilled, microhybrid and nanohybrid performing similarly [27–30]. The effect of filler size on elastic modulus, in turn, seems to be dependent on the range

of filler size distributions employed [31]. The values for filler loading used herein were taken from manufactures’ data sheets, rather than measured by the authors. Uncertainties regarding the reliability of these data should therefore be accounted before definite conclusions regarding the relations to the tested mechanical properties are drawn. Those observations are introductory to our fractographic analyses, which revealed markings that seem to relate particle features to fracture mechanisms and loading conditions. In specimens loaded statically, crack deflection appeared as the dominant toughening mechanism, with particles protruding from the fracture surface for microstructures in the range of nanometer to a few microns. An atomic force microscopy evaluation of crack mechanisms in dental composites has confirmed the preferential crack path through the matrix and matrix/filler interface for composites composed of 40–120 nm spherical silica particles or ∼0.4 ␮m irregular glass particles [32]. In that range of particle size, composite toughness was shown to be governed by matrix and interface toughnesses, while particles seem to survive composite fracture. The microscopic evidence of resin covering the particles at the fractured surfaces (Fig. 2) is indicative of a strong – and probably nondegraded – interface. In storage times longer than the ones used in the present study, interfacial debonding during fracture is expected to have a greater effect on degradation of properties [33]. An exception to the rule of interparticle fracture in static tests occurred for Venus Diamond, where particles measuring >10 ␮m showed markings of brittle fracture and the absence of resin on the surface (Fig. 3), confirming particle fracture as a common event for this composite. At smaller scales, crack behavior in Venus Diamond followed those of the other composites (interparticle fracture). Microstructural dimensions and particle size become important factors as the stress concentration is defined by the crack tip scale. At high-energy fractures, particles that are small in comparison to the plastic zone surrounding the crack tip are more prone to induce crack deflection and interface decohesion. The same does not hold for microstructures that scale up to the size of the plastic zones present around the crack tip. Following the encounter with a large particle, the region ahead of the crack tip is deprived of deformable regions (i.e., resin matrix) to dissipate some of the

d e n t a l m a t e r i a l s 3 0 ( 2 0 1 4 ) 381–391

crack-opening stress. The plastic zone length is momentarily reduced and the stress redistribution concentrates the stress at a small region right ahead of the crack tip, which becomes high enough to induce particle fracture. The relation for the plastic zone radius r proposed by Irwin r = (1/6)(KI / Y )2 valid for small scale yieldings under plain strain conditions [34], implies an amplification of particle-size effects for composites with more flexible matrices. Within this zone, a higher amount of energy can be dissipated within the same area in comparison to highly filled, stiffer composites. The flexible TCD-urethane polymer and the presence of particles exceeding a certain size may then be accredited for the high FT values measured for Venus Diamond. That is not because of particle fracture events (which does not increase the overall toughness of a composite), but to local interactions at the crack tip that induce efficient toughening mechanisms, such as crack bridging and microcracking ahead of the crack tip (Fig. 3). Large-scale crack bridging mechanisms lead to more efficient toughening development during crack extension, and have been recently linked to particle size in composites with similar organic compositions [35,36]. These effects are clearly dependent on the crack trajectory and orientation in relation to the stationary particle, but our observations seem to indicate that a threshold value of particle size for particle fracture exists around 10–20 ␮m for a particle being hit by a crack at its center. Interface fracture of particles in such dimensions then becomes energetically unfavorable due to the increased interfacial area. In the case of irregular particles, the increased tilting angles for crack deflection plays further against interface fracture, as it tends to direct the crack tip away from the principal stress field orientation. On the other hand, the high strength loss shown by Venus Diamond and the interparticle fracture patterns observed for all composites loaded cyclically point to a negligible effect of particle size in the fatigue resistance of resin composites. Other authors found an opposite trend. Composites with a broader filler size distribution and higher maximum filler size have shown lower susceptibility to slow crack growth in a study by Ornaghi et al. [37]. Shah et al. [38] measured a higher stress intensity threshold for crack initiation under cyclic loading for a microhybrid in comparison to a nanofilled composite having the same filler vol% and similar matrix composition. Takeshige et al. [39] found a similar result when comparing a composite with particles reaching 50 ␮m in size to another with particles in the range of 1–3 ␮m. The two last studies, however, dealt with pre-cracked compact tension specimens, in which the pre-cracks propagates into a nonfatigued material, where the effects of high cyclic fatigue phenomena cannot be appreciated. Rather than microstructure-related, the fatigue resistance of the tested composites seem to be better linked to factors associated to the resin phase, such as matrix content, composition, network rigidity and susceptibility to stress and environmental degradation. The first indication of that is the extremely low strength loss shown by Clearfil Majesty Posterior, the composite with the lowest amount of resin phase. With fewer matrix, fewer are the loci for flaws and cracks to develop. Also, composites that incorporate prepolymerized resin fillers in their composition seem to benefit from the lower amount of in situ polymerizable resin. The

389

polymerization under ideal conditions of energy, temperature and pressure, used for the fabrication of prepolymerized fillers, has proven to provide superior mechanical stability under fatigue [40]. The deflection of cracks around prepolymerized fillers observed in the FFS test specimens (see Fig. 4c and f) in contrast to fracture occurring under static loading (see Fig. 2f), indicates that cracks under fatigue tend to preferentially propagate through the weaker matrix phase where stress levels are not capable of inducing the fracture of prepolymerized fillers. Apart from Miris II, the composites Kalore, Clearfil Majesty Esthetic and Tetric EvoCeram presented strength losses between 41 and 53%, lower than many composites not containing prepolymerized fillers. Under high-energy fractures the same effect is not observed, and composites having prepolymerized fillers have shown, in general, low values of FT and FS. The role of matrix composition and network mechanical behavior on fatigue resistance is also assumed based on the large differences in strength loss shown by composites that present different matrix compositions (not Bis-GMA-based), such as Venus Diamond, Filtek Silorane and Kalore. The high backbone flexibility of the TCD-urethane monomer in Venus Diamond apparently makes it more susceptible to fatigue crack growth, which can be aggravated by lower degrees of conversion [41], which has been reported for this composite by Boaro et al. [42]. More flexible polymeric networks have also been shown to present higher creep deformation [43], a viscoelastic response to loading that may undermine the structural integrity of the network under repetitive strain challenges. The silorane-based material, on the other hand, has shown comparable mechanical properties to conventional composites [44,45] and an apparent higher fatigue resistance (shown herein) when compared to Bis-GMA-based composites at similar filler levels. The fatigue resistance of Bis-GMA-based composites, in fact, when compared to FFS values measured in the same laboratory under the same conditions and methods, has not shown any significant improvements when compared to composites available 7 years ago (at similar filler contents) [46]. The hydrolyzing and plasticizing effects of water on the cured polymer are not secondary [47] and further aggravate the stress-related degradation of mechanical properties in composites. Indeed, Takeshige et al. [39] showed that the presence of water do not necessarily reduces the fatigue threshold for crack initiation, but it leads to crack acceleration once a crack has initiated (steepening in the Paris-law linear regime). Fracture toughness has also shown to be minimally affected by cyclic loading when specimens are unaged or only aged in air [48]. Combined with the effects of water and/or solvents, cyclic loading leads to a significant decrease in fracture toughness. The effect of mechanical fatigue is highlighted by the progressive deterioration of FT with the number of loading cycles in the presence of water. Due to the limited water storage period in the present study, filler/matrix interface degradation was not readily evident, and the fatigue degradation induced by cyclic loading is assumed to have been limited to the resin phase. Testing resin composites under cyclic conditions emerges as an important method in that it characterizes failure mechanisms developing in the bulk of the material. Looking at

390

d e n t a l m a t e r i a l s 3 0 ( 2 0 1 4 ) 381–391

the clinical loading conditions, where defects are introduced on the surface of restorations and crack initiation may arise from wear and Hertzian contact cracks, strength and fracture toughness, unlike fatigue testing, better reflect surface related properties. These properties seem to have a complex relationship to the fatigue behavior of resin composites and their use for the prediction of clinical performance of composite restorations is not appropriate.

5.

Conclusions

Based on our results, the linear elastic mechanical properties such as elastic modulus, fracture toughness and flexural strength are inappropriate to characterize or predict the fatigue resistance of dental resin composites. While features related to the microstructure seem to affect inherent mechanical properties more directly, the susceptibility to fatigue degradation seem to be rather linked to aspects related to the matrix phase. More precise statements on the effects of matrix and microstructure related factors on fatigue mechanisms of dental composites could be achieved by testing model compositions rather than commercial products.

Acknowledgements The authors declare no conflict of interest. The present work was performed in partial fulfillment of the requirements for obtaining the degree “Dr. med. dent.”

references

[1] Heintze SD, Zellweger G, Zappini G. The relationship between physical parameters and wear of dental composites. Wear 2007;263:1138–46. [2] Quinn JB, Quinn GD. Material properties and fractography of an indirect dental resin composite. Dent Mater 2010;26:589–99. [3] Garcia-Godoy F, Frankenberger R, Lohbauer U, Feilzer AJ, Kramer N. Fatigue behavior of dental resin composites: flexural fatigue in vitro versus 6 years in vivo. J Biomed Mater Res B Appl Biomater 2012;100:903–10. [4] Pick B, Meira JB, Driemeier L, Braga RR. A critical view on biaxial and short-beam uniaxial flexural strength tests applied to resin composites using Weibull, fractographic and finite element analyses. Dent Mater 2010;26:83–90. [5] Rodrigues Jr SA, Ferracane JL, Della Bona A. Flexural strength and Weibull analysis of a microhybrid and a nanofill composite evaluated by 3- and 4-point bending tests. Dent Mater 2008;24:426–31. [6] Lohbauer U, von der Horst T, Frankenberger R, Kramer N, Petschelt A. Flexural fatigue behavior of resin composite dental restoratives. Dent Mater 2003;19:435–40. [7] Lohbauer U, Krämer N, Petschelt A, Frankenberger R. Correlation of in vitro fatigue data and in vivo clinical performance of a glass ceramic material. Dent Mater 2008;24:39–44. [8] International Standards Organization ISO 4049. Dentistry – polymer-based filling, restorative and luting materials; 2000. [9] Lohbauer U, Müller FA, Petschelt A. Influence of surface roughness on mechanical strength of resin composite versus glass ceramic materials. Dent Mater 2008;24:250–6.

[10] Munz D, Fett T. Mechanical properties, failure behavior, materials selection. Berlin: Springer; 1999. [11] European Standard EN 843-1. Monolithische Keramik. Mechanische Eigenschaften bei Rraumtemperatur. Teil 1: Bestimmung der Biegfestigkeit; 1995. [12] European Standard EN 843-5. Monolithische Keramik. Mechanische Prüfungen bei Raumtemperatur. Teil 5: Statistische Auswertung; 1997. [13] Dixon WJ, Mood AM. A method for obtaining and analysing sensitivity data. J Am Stat Assoc 1948;109:109–26. [14] Draughn RA. Compressive fatigue limits of composite restorative materials. J Dent Res 1979;58:1093–6. [15] Wallin KRW. Statistical uncertainty in the fatigue threshold staircase test method. Int J Fatigue 2011;33: 354–62. [16] Abe Y, Braem MJ, Lambrechts P, Inoue S, Takeuchi M, Van Meerbeek B. Fatigue behavior of packable composites. Biomaterials 2005;26:3405–9. [17] Braem M, Davidson CL, Lambrechts P, Vanherle G. In vitro flexural fatigue limits of dental composites. J Biomed Mater Res 1995;28:1397–402. [18] McCabe JF, Wang Y, Braem M. Surface contact fatigue and flexural fatigue of dental restorative materials. J Biomed Mater Res 2000;50:375–80. [19] Ritchie RO. The conflicts between strength and toughness. Nat Mater 2011;10:817–22. [20] Ikejima I, Nomoto R, McCabe JF. Shear punch strength and flexural strength of model composites with varying filler volume fraction, particle size and silanization. Dent Mater 2003;19:206–11. [21] Ilie N, Hickel R, Valceanu AS, Huth KC. Fracture toughness of dental restorative materials. Clin Oral Investig 2012;16:489–98. [22] Braem M, Lambrechts P, Van Doren V, Vanherle G. The impact of composite structure on its elastic response. J Dent Res 1986;65:648–53. [23] Goncalves F, Kawano Y, Braga RR. Contraction stress related to composite inorganic content. Dent Mater 2010;26:704–9. [24] Kim KH, Ong JL, Okuno O. The effect of filler loading and morphology on the mechanical properties of contemporary composites. J Prosthet Dent 2002;87:642–9. [25] Kim KH, Park JH, Imai Y, Kishi T. Microfracture mechanisms of dental resin composites containing spherically-shaped filler particles. J Dent Res 1994;73:499–504. [26] Fu S-Y, Feng X-Q, Lauke B, Mai Y-W. Effects of particle size, particle/matrix interface adhesion and particle loading on mechanical properties of particulate–polymer composites. Composites B 2008;39:933–61. [27] Beun S, Glorieux T, Devaux J, Vreven J, Leloup G. Characterization of nanofilled compared to universal and microfilled composites. Dent Mater 2007;23:51–9. [28] Ferracane JL. Resin composite – state of the art. Dent Mater 2011;27:29–38. [29] Ilie N, Rencz A, Hickel R. Investigations towards nano-hybrid resin-based composites. Clin Oral Investig 2013;17:185–93. [30] Leprince J, Palin WM, Mullier T, Devaux J, Vreven J, Leloup G. Investigating filler morphology and mechanical properties of new low-shrinkage resin composite types. J Oral Rehabil 2010;37:364–76. [31] Masouras K, Akhtar R, Watts DC, Silikas N. Effect of filler size and shape on local nanoindentation modulus of resin-composites. J Mater Sci Mater Med 2008;19:3561–6. [32] Chan KS, Lee Y-D, Nicolella D-P, Furman BR, Wellinghoff S, Rawls HR. Improving fracture toughness of dental nanocomposites by interface engineering and micromechanics. Eng Fract Mech 2007;74:1857–71. [33] Ferracane JL, Berge HX, Condon JR. In vitro aging of dental composites in water – effect of degree of conversion, filler

d e n t a l m a t e r i a l s 3 0 ( 2 0 1 4 ) 381–391

[34] [35]

[36]

[37]

[38]

[39]

[40] [41]

volume, and filler/matrix coupling. J Biomed Mater Res 1998;42:465–72. Irwin GR. Fracture. In: Handbuch der Physik. Berlin: Springer; 1958. p. 551–90. De Souza JA, Goutianos S, Skovgaard M, Sorensen BF. Fracture resistance curves and toughening mechanisms in polymer based dental composites. J Mech Behav Biomed Mater 2011;4:558–71. Shah MB, Ferracane JL, Kruzic JJ. R-curve behavior and micromechanisms of fracture in resin based dental restorative composites. J Mech Behav Biomed Mater 2009;2:502–11. Ornaghi BP, Meier MM, Rosa V, Cesar PF, Lohbauer U, Braga RR. Subcritical crack growth and in vitro lifetime prediction of resin composites with different filler distributions. Dent Mater 2012;28:985–95. Shah MB, Ferracane JL, Kruzic JJ. Mechanistic aspects of fatigue crack growth behavior in resin based dental restorative composites. Dent Mater 2009;25:909–16. Takeshige F, Kawakami Y, Hayashi M, Ebisu S. Fatigue behavior of resin composites in aqueous environments. Dent Mater 2007;23:893–9. Lin L, Drummond JL. Cyclic loading of notched dental composite specimens. Dent Mater 2010;26:207–14. Lohbauer U, Rahiotis C, Krämer N, Petschelt A, Eliades G. The effect of different light-curing units on fatigue behavior

[42]

[43]

[44]

[45]

[46]

[47] [48]

391

and degree of conversion of a resin composite. Dent Mater 2005;21:608–15. Boaro LC, Goncalves F, Guimaraes TC, Ferracane JL, Versluis A, Braga RR. Polymerization stress, shrinkage and elastic modulus of current low-shrinkage restorative composites. Dent Mater 2010;26:1144–50. El-Safty S, Silikas N, Watts DC. Creep deformation of restorative resin composites intended for bulk-fill placement. Dent Mater 2012;28:928–35. Ilie N, Hickel R. Macro-, micro- and nano-mechanical investigations on silorane and methacrylate-based composites. Dent Mater 2009;25:810–9. Lien W, Vandewalle KS. Physical properties of a new silorane-based restorative system. Dent Mater 2010;26:337–44. Lohbauer U, Frankenberger R, Krämer N, Petschelt A. Strength and fatigue performance versus filler fraction of different types of direct dental restoratives. J Biomed Mater Res B Appl Biomater 2006;76B:114–20. Ferracane JL. Hygroscopic and hydrolytic effects in dental polymer networks. Dent Mater 2006;22:211–22. Ravindranath V, Gosz M, De Santiago E, Drummond JL, Mostovoy S. Effect of cyclic loading and environmental aging on the fracture toughness of dental resin composite. J Biomed Mater Res B Appl Biomater 2007;80B: 226–35.