A study on the material properties and fatigue life of natural
rubber with different carbon blacks
J.-H. Kim a , H.-Y. Jeong b, * a Korea Railroad Research Institute, 360-1, Woulam-Dong, Uiwang-City, Kyounggi-Do 437-050, South Korea b Department of Mechanical Engineering, Sogang University, 1 Shinsoo-Dong, Mapo-Gu, Seoul 121-742, South Korea Received 14 November 2003; received in revised form 6 April 2004; accepted 9 July 2004 Abstract Natural rubber compounds lled with N330, N650 or N990 were experimentally investigated in order to examine the effects of carbon black on the fatigue life, the hysteresis, the fracture surface morphology and the critical J-value. The fatigue life was obtained by conducting a displacement-controlled fatigue test on an hourglass-shaped specimen, and the hysteresis was calculated from the loading and unloading curves obtained during the fatigue test. The fracture surface obtained after the fracture test was examined by a microscope, and its roughness was calculated for the surface prole obtained by using a stylus-type roughness tester. In addition, the critical J-value was obtained by conducting a tensile test for pre-cracked dumbbell specimens. Based on the test data, it was noticed that the fatigue life, the hysteresis, the small-scale roughness and the critical J-value were ranked in the following order for the NR compound lled with N330, N990 or N650. It was also noticed that the logarithmic value of the fatigue life was linearly proportional to the square root of the product of the critical J-value and the hysteresis. q 2004 Elsevier Ltd. All rights reserved. Keywords: Carbon black; Natural rubber; Fatigue life; Critical J-value; Hysteresis; Fracture surface morphology; Roughness 1. Introduction Carbon black is used as a reinforcement to increase the structural and thermal properties of NR (natural rubber). When a rubber component lled with carbon black is under loading, the rubber matrix around carbon black is strained more than the rest. If a large or repeated force is applied, carbon black is separated from the matrix nucleating microcracks. There are other sources of microcracks in rubber including contaminants or voids in the matrix, imperfectly dispersed compounding ingredients, mold lubricants and surface aws [1,2]. Upon further loading, the microcracks coalesce resulting in a noticeable crack that may lead to failure [3]. On the torn surface of stronger elastomers vertical steps are more frequent and larger, and each step is associated with a carbon black particle [4]. Therefore, it is important to gure out the characteristics and effects of carbon black in order to understand the material behavior of NR. For this purpose, many researchers have studied the characteristics and effects of carbon black using diverse methods. Greensmith [5] investigated the effect of carbon black on the tearing behavior of NR and SBR (styrene butadiene rubber, then called GR-S), and he discovered that the reinforcement of carbon black was effective only in a limited range of temperature and rate of crack propagation. Hess and Ford [6] reviewed microscopy of lled rubbers, and they showed evidence that a micro- crack nucleated from a carbon black, and a crack followed a path through reagglomeration areas. Lake and Lindley [1] investigated the effects of deformation, testing environment, protective agents and llers on the fatigue life and the crack growth rate, and they noticed that the crack growth rate decreased by addition of carbon black especially small carbon black particles such as HAF (high abrasion furnace), and the fatigue life of NR decreased 0142-1123/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijfatigue.2004.07.002 International Journal of Fatigue 27 (2005) 263272 www.elsevier.com/locate/ijfatigue * Corresponding author. Fax: C82 2 712 0799. E-mail address: jeonghy@sogang.ac.kr (H.-Y. Jeong). less than that of SBR as temperature increased. Goldberg et al. [7] investigated the effect of carbon black loading on stretching, tearing and failure of NR and SBR, and they noted that knotty tearing was more frequently observed in NR than in SBR, and the tearing energy increased as the loading increased from 0 to 35 phr. Chung et al. [8] measured the tearing energies of SBR, HNBR (hydrogen- ated nitrile rubber) and NR with different loading of carbon black, and they noticed that the tearing energy became maximized around the loading of 40 phr in SBR and NR, but the effect of carbon black on the tearing energy was not considerable in NR because of crystal- lization occurring in front of a crack. Wang [9] thoroughly reviewed the effect of carbon black on dynamic properties of lled vulcanizates such as the elastic modulus, the viscous modulus and the loss tangent, and he noted that the loss tangent was dependent on the temperature, the strain amplitude and the loading of carbon black. Legorju-jago and Bathias [10] investigated the effects of the load ratio, the temperature, the thickness and the stress state on the fatigue life and the crack growth rate of NR, CR (polychloroprene rubber) and SBR, and they noticed that for the high load ratio or thickness, the fatigue life of NR increased because of crystallization. However, no research was conducted on the change of the material property and the failure behavior of lled NR due to different types of carbon black. Thus, in this paper the material properties (the critical J-value and the hysteretic loss), the fatigue life and the fracture mor- phology of NR compounds lled with three types of carbon black, N330, N650 and N990, were experimentally obtained. The fatigue life was measured for an hourglass- type specimen under a displacement control with three different amplitudes. The fracture morphology was examined not only by photos but also by fracture surface roughness. In addition, the roughness was classied into a large-scale roughness and a small-scale roughness. By reviewing the experimental data, several interesting points could be noted. First, the material properties, the fatigue life and the fracture morphology became different due to variant types of carbon black. Secondly, the trend of the large-scale roughness was opposite to that of the small- scale roughness, and the critical J-value increased as the small-scale roughness increased. Thirdly, the logarithmic value of the fatigue life for each of the displacement amplitude was linearly proportional to the square root of the product of the critical J-value and the hysteretic loss. Thus, the logarithmic value of the fatigue life could be represented as a function of the displacement amplitude, the critical J-value and the hysteretic loss. This implies that the fatigue life of a rubber component can be predicted from the displacement amplitude and the material properties such as the critical J-value and the hysteretic loss. 2. Experimental results 2.1. NR lled with three types of carbon black In this research, NR compounds lled with three kinds of carbon black, N330, N650 and N990, were investigated. Table 1 shows the characteristics of the NR compounds along with carbon black. Carbon black can be classied into the small, medium and large size according to ASTM D1765 [11], and a representative carbon black for each size was selected. When a rubber component is under develop- ment in industry, the static stiffness of a rubber component is usually specied. Since there is a relationship between Youngs modulus and the hardness of rubber, the hardness can be determined from the required static stiffness [12,13]. For this study, the amount of carbon black in each compound was determined in a way that the hardness of the rubber compound remained the same at 62. In addition, the hardness was measured using durometer type A on a compression specimen of diameter of 28.8 mm and height of 12.5 mm [14,15]. 2.2. Fatigue life The hourglass-shaped specimen designed by Takeychi et al. [16] was used to measure the fatigue life of the lled NR compounds, and it is shown in Fig. 1. It has an elliptical cross section, and the seam line is located on the surface of the minor axis. Since the stress on the surface of the major axis is higher than that on the surface of the minor axis, a crack usually starts at the major axis, and the seam line hardly affects the fatigue life of the specimen. The fatigue test was conducted at room temperature, 23 8C, and it was displacement-controlled using MTS810. The displacement was prescribed as a sinusoidal pulse at the frequency of 1 Hz with the maximum displacement of 16, 19 or 22 mm and the minimum displacement of 0 mm. Thus, the maximum displacement was equal to the double displacement amplitude, and the load ratio, which is dened to be the ratio of the minimum load (or displacement) to the maximum load (or displacement), was equal to 0. During the test, the specimen was air-cooled to keep the temperature around 23 8C. The fatigue life was dened to Table 1 Characteristics of NR compounds lled with N330, N650 and N990 N330 N650 N990 CB particle diameter (nm) 30 61 285 Specic weight 1.1245 1.1485 1.2270 CB phr 46 54 90 Volume fraction (%) 16.8 18.4 28.0 Total phr 171.26 183.00 219.26 Curing time (s) 420 420 420 Curing temperature (8C) 173 173 173 Hardness 62 62 62 J.-H. Kim, H.-Y. Jeong / International Journal of Fatigue 27 (2005) 263272 264 be the cycle at which the load became half the load measured at 10,000 cycles. Three fatigue tests were conducted at three different double displacement amplitudes for each type of carbon black. Table 2 shows the average and the standard deviation of the fatigue life of the NR compounds. As expected, the fatigue life reduced as the amplitude increased. Regardless of the amplitude, the NR lled with N650 resulted in the shortest fatigue life, but the NR lled with N330 resulted in the longest fatigue life. However, there were comparatively small deviations among tests showing the repeatability of the test. Previous research showed that the fatigue life of NR is dependent on testing environment [1], the load ratio [10] and additives [17]. In addition, this study showed that the fatigue life is also dependent on the type of carbon black. 2.3. Measurement of the hysteresis Rubber has a characteristic of viscoelasticity resulting in an energy loss during a cyclic loading. The energy loss is called the hysteresis (or hysteretic loss), and it eventually is converted to heat causing temperature increase [18]. Since the temperature increase gradually degrades rubber, it is important to understand the mechanism and effects of the hysteretic loss. The specimen shown in Fig. 1 was also used to measure the amount of hysteresis, and the conditions were the same as that for the fatigue test. It is well known that the material properties of NR change during the rst set of loadings due to so-called Mullins effect [19]. Thus, the hysteresis was measured after several thousands of loadings. For example, when the double displacement amplitude was 22 mm, the specimen was loaded up to 10,000, 15,000, and 20,000 cycles, and the force vs. displacement curve was measured for the next 100 cycles. Then, the hysteresis was obtained from the average of the energy loss occurring for the 100 cycles. The number of cycles for the rst measurement, 10,000 cycles, was chosen at about a quarter of the minimum fatigue life which occurred for the rubber compound with N650 at the fatigue test with the amplitude of 22 mm. When the amplitude was 16 or 19 mm, the cycles for which the hysteresis was measured were higher than those for the amplitude of 22 mm. Three sets of tests were conducted for each test condition, and the averages and the standard deviations of the hysteresis for three rubber compounds are shown in Table 3. As expected, the hysteresis increased as the amplitude increased. Note that the hysteresis was ranked in the following order for the rubber compound lled with N330, N990 or N650, and there were negligible deviations showing the repeatability of the test. Carbon black particles are rmly fused together in rubber, and the smallest discrete entity existing in rubber is always the aggregate [20]. By virtue of their irregular morphology, the aggregates are bulky, and occupy an effective volume considerably larger than that of carbon black itself. The bulkiness of the aggregate is the property commonly called structure. It is adequate to describe carbon black morphology by two parametersone related to the structure and the other to the mean particle size. The most commonly used techniques for measuring carbon black structure have been based on internal void volume using absorptive measurements or volumetric measurements under pressure [20]. Dibutylphthalate (DBP) is continuously added using a constant rate buret to dry carbon black in an internal mixer tted with a device measuring the torque on the mixer blades. When the interstices between carbon black particles are lled with DBP, an appreciable torque level is developed, and the absorptometer and buret will shut off automatically. The volume of DBP per unit mass of carbon black is the DBP Fig. 1. Hourglass-shaped fatigue test specimen (all dimensions in mm). Table 2 Fatigue life test data (cycle) N330 N650 N990 Double amplitude of 16 mm Ave. 625,800 213,940 457,310 Std. 7860 34,150 6720 Double amplitude of 19 mm Ave. 480,650 98,930 323,860 Std. 10,530 7200 5550 Double amplitude of 22 mm Ave. 382,210 41,110 214,340 Std. 6140 1040 14,000 Table 3 Hysteresis test data (N mm) N330 N650 N990 Double amplitude of 16 mm Ave. 355.3 44.5 205.3 Std. 7.8 1.6 3.3 Double amplitude of 19 mm Ave. 465.5 47.4 276.3 Std. 0.9 0.4 0.8 Double amplitude of 22 mm Ave. 487.1 49.3 359.3 Std. 5.4 1.3 3.3 J.-H. Kim, H.-Y. Jeong / International Journal of Fatigue 27 (2005) 263272 265 absorption (DBPA) number. The test method for DBPA number determination is described in ASTM2414-00 [21]. The traditional particle size count of electron micro- graphs is difcult and inherently imprecise. Thus, the specic surface area is determined instead. The absorption of cetyltrimethyl ammonium bromide (CTAB) by carbon black in an aqueous solution is a method for surface area determination. This method covers the measurement of the specic surface area of carbon black exclusive of area contained in micropores too small to admit CTAB molecules, and it is described in ASTM 3765-99 [22]. The amount of hysteresis depends on both the structure and the surface area per unit volume of carbon black aggregate [23]. In Table 4, the DBPA number and the CTAB surface area are shown for the three types of carbon black. If the structure of carbon black aggregate develops well, the carbon black aggregate is more likely to be cut off by a blade in the mixing process, and new surface is more likely to be created. This causes stronger cohesion and smaller friction between rubber and carbon black, sub- sequently resulting in less hysteresis. When the specic surface area of carbon black is high, there occurs more friction between rubber and carbon black, subsequently resulting in more hysteresis. Therefore, the hysteresis increases as the DBPA number decreases or the CTAB surface area increases. Data in Table 4 indicate that the structure of N650 was well developed, and the surface area of the carbon black was comparatively small. Therefore, the hysteresis of the NR lled with N650 turned out to be the smallest. The opposite situation happened to N330, and the hysteresis of the NR lled with N330 turned out to be the largest. Hysteresis of an elastomer can be approximately obtained from the loss tangent multiplied by input energy [24]. In addition, the loss tangent is a function of the strain amplitude, temperature as well as the type of carbon black. As an example, the loss tangent of lled SBR has a critical temperature around 0 8C below which it decreases but over which it increases as the CTAB surface area increases [9]. Therefore, the ranking of hysteresis of the three lled NR compounds could change at different temperatures, and the hysteresis obtained from this study should be accepted with caution. 2.4. Measurement of the critical J-value Since the fatigue life depends on the growth of a crack, which is related to the critical J-value, it is important to investigate the effect of carbon black on the critical J-value. Thus, the critical J-value of the NR compounds was measured. Based on the method of Begley and Landes [25], the critical J-value was measured using the dumbbell specimen shown in Fig. 2 [26]. For each type of the NR compound, a set of ve specimens were pre-cracked, the crack length ranging from 1 to 3 mm, every 0.5 mm apart. The specimen was stretched under the standard speed of 500 mm/min using UTM (Universal Test Machine). The J-value can be calculated from the following equation: J ZK 1 B vU va
D (1) Here, B is the thickness of the specimen, U is the strain energy, a is the crack length, and D is the displacement. The J-value is the amount of decrease of the strain energy per unit area of new crack surface with the displacement xed. While the specimen was being loaded, it was video- recorded by a digital motion camera. The moment when the geometry around the crack became asymmetric with respect to the crack line was regarded as the moment of the crack growth [7], and the J-value at the moment was assumed to be the critical J-value. Thus, the critical J-value is the same physical quantity as the tearing energy [27]. The critical J-value remained almost the same regardless of the pre-crack length, and the test showed a good repeatability. Thus, only the average and the standard deviation of the critical J-values for the NR compounds obtained for three sets of ve specimens are shown in Table 5. Note that the critical J-value is ranked in the following order for the compound lled with N330, N990 or N650, which is in the same order for the fatigue life and the hysteretic loss. The critical J-value of the NR lled with N330 is smaller by a factor of about 3 than that of gum NR and lled NR measured by Chung et al. [8]. This difference stemmed from different denitions on Table 4 DBPA number and CTAB surface area of N330, N650 and N990 N330 N650 N990 DBPA a (cm 3 /100 g) 100 129 35 CTAB a (m 2 /g) 84 38 9 a Carbon black science and expanded. Marcel Dekker; 1993. p. 113. Fig. 2. Dumbbell specimen used to measure the critical J-value (all dimensions in mm). Table 5 Critical J-value of NR compounds lled with N330, N650 or N990 (N/mm) N330 N650 N990 Ave. 1.416 0.379 1.167 Std. 0.012 0.019 0.030 J.-H. Kim, H.-Y. Jeong / International Journal of Fatigue 27 (2005) 263272 266 the moment of the crack growth. The process of crack growth can be described by two steps [8]. First, the crack starts at one or two spots on the inside surface of the tip, then it propagates across the thickness of the specimen to form a full initial crack before it propagates along the width of the specimen. In this study, the moment of the crack growth was dened to be the time when the crack surface lost its symmetry, but Chung et al. dened the moment to be the time when a full crack through the thickness started to grow. Thus, the crack growth dened in this study happened earlier, and the critical J-values measured turned out to be smaller. 2.5. Measurement of the fracture surface morphology In general, fracture surface morphologies give infor- mation about failure process and mechanisms [3]. As shown in Fig. 3, the (large-scale) fracture surface morphologies of the three types of the NR compounds look quite rough. Thus, the stereo optical microscope that had a poor depth of focus could not be used for the analysis of the fracture surfaces. Instead, LG DSP color camera, a microscope for the inspection of semiconductor, which had a good depth of focus, was used and its magnication was from !100 to ! 1000. The large-scale and the small-scale fracture surface morphologies of the NR compounds are shown in Figs. 3 and 4, respectively. Figs. 3 and 4 show that the NR compound lled with N330 or N990 has many small humps, but the NR compound lled with N650 has a small number of large humps. In other words, the NR compound lled with N330 or N990 has a smoother surface in the large-scale but a rougher surface in the small-scale than the NR compound lled with N650. To quantify the roughness of the fracture surfaces, the prole of the fracture surfaces was obtained by using a stylus type roughness tester, and the root-mean-square (RMS) roughness was employed. Moreover, the RMS roughness was calculated in two different scales. The prole along the major axis of 14 mm long was obtained, and the RMS roughness was calculated for the prole. This roughness was named the large-scale roughness in this paper. The prole of four 1 mm long segments along the major axis was obtained, and the average of the roughness of the four segments was assumed to be the RMS roughness for the prole. This roughness was named the small-scale roughness. The large and small-scale RMS roughness data are shown in Tables 6 and 7, respectively. It is striking to notice that the order of magnitude of roughness is opposite to each other; the order of large-scale roughness is N650O N990ON330, but the order of small-scale roughness is N330ON990ON650. Fukahori and Andrews [28] found out that the hyteresis of highly deformable polymers decreased as the fracture surface roughness increased. As shown in Fig. 5, the hysteresis of the NR compounds studied in this paper decreased as the large-scale roughness increased. However, as shown in Fig. 6, the hysteresis increased as the small- scale roughness increased. Therefore, when the roughness of a NR compound is examined, the scale should be taken into account, otherwise the conclusion would be opposite. Fig. 3. Large-scale fracture surface. Fig. 4. Small-scale fracture surface. J.-H. Kim, H.-Y. Jeong / International Journal of Fatigue 27 (2005) 263272 267 Note also that the test data are located near the regression line in Fig. 5, but they are located off the regression line in Fig. 6. The poor linearity of the test data shown in Fig. 6 stemmed from the fact that the small-scale roughness was obtained for four short proles, and it in itself had a high variance relative to the average value (refer to Table 7). In other words, the small-scale roughness is quantitatively less reliable than the large-scale roughness. However, the trend of the small-scale roughness is correct in that the small-scale roughness increases as the critical J-value increases (refer to Fig. 8). As Gent and Pulford [4] noted, vertical steps are more frequent, higher and less spaced on the torn surfaces of tougher elastomers, and the spacing is in the order of 10 mm. Therefore, the small-scale roughness should be higher for tougher elastomers. 2.6. Analysis of the carbon black distribution The fracture surface was sliced to a thin sheet using a glass knife in liquid nitrogen. Three locations on the thin sheet were randomly selected to examine the carbon black distribution using an optical microscope. Each location was a square of 1 mm!1 mm, and it was divided into 10,000 cells. The dispersion of carbon black agglomerates was examined, and Table 8 shows the percentage of cells in which carbon black agglomerates were observed, proving that they were well distributed in all three NR compounds. Then, the number of cells in which an agglomerate of carbon black larger than 50 mm 2 existed was counted, and the diameters of those large agglomerates were also measured. In Fig. 7, the number of cells is shown as a function of the diameter of the agglomerates. In the rubber compound lled with N330, an agglomerate with diameter of 9 or 10 mm was observed in one or two cells. In the rubber compound lled with N990, an agglomerate with diameter of 9 or 10 mm was observed in seven or three cells, Table 6 Large-scale RMS roughness data (mm) N330 N650 N990 Ave. 0.311 0.771 0.425 Std. 0.012 0.027 0.066 Table 7 Small-scale RMS roughness data (mm) N330 N650 N990 Ave. 0.095 0.028 0.045 Std. 0.040 0.008 0.017 Fig. 5. Hysteresis as a function of the large-scale roughness. Fig. 6. Hysteresis as a function of the small-scale roughness. Table 8 Carbon black dispersion data N330 (%) N650 (%) N990 (%) 1 99.90 98.10 98.40 2 99.70 96.90 97.80 3 99.70 97.20 97.70 Avg. 99.77 97.40 97.40 Fig. 7. Diameter and frequency of carbon black agglomerates. J.-H. Kim, H.-Y. Jeong / International Journal of Fatigue 27 (2005) 263272 268 and an agglomerate with diameter of 12 mm at highest was observed. However, in the rubber compound lled with N650, an agglomerate with diameter of 9 mm was observed in ten cells, and even an agglomerate with diameter of about 15 mm was observed. This observation on the distribution is compatible with the DBPA numbers shown in Table 4. In the case of the NR compound lled with N650, for example, the DBPA number is high. This means that the carbon black is well developed, and the agglomerate of the carbon black is big. Knotty tearing is an important aspect of the reinforcing mechanism of carbon black, and it results in a rough torn surface and a high critical J-value [5]. For the NR compounds studied in this paper, the critical J-value was also high when the small-scale roughness was high, as shown in Fig. 8. The rubber compound lled with N330 had a rough surface in the small-scale, and its critical J-value turned out to be high. This means that the carbon black, N330, formed small agglomerates, which acted as a local stress raiser. The small agglomerates deected the direction of a crack growth causing the knotty tearing [29]. However, the rubber compound lled with N650 had a clean surface in the small-scale, and its critical J-value turned out to be low. This means that the carbon black, N650, formed large agglomerates, which were easily separated from the rubber matrix. Thus, a crack propagated rapidly creating a cleaner surface in the small-scale. 3. Function of the fatigue life NR has good mechanical characteristics of large elastic deformation and damping. In addition, it can be readily manufactured at a comparatively low cost. Because of these characteristics, rubber is widely used in numerous products such as automobiles, locomotives and commod- ities [30]. However, a rubber component is susceptible to fatigue failure, and its durability is one of the major concerns. Thus, when a rubber component is under development using various compositions, the fatigue life of the component is to be measured from a fatigue test, which usually takes a long period of time. If the fatigue life of a rubber component can be predicted from material properties which can be obtained from simple exper- iments, it can save time and efforts to select a composition which results in the best durability. By reviewing the experimental data obtained from this study, it was easily noticed that the fatigue life increased as the critical J-value or the hysteresis increased. After trying many different functional forms, it was nally noticed that the logarithmic value of the fatigue life was linearly proportional to the square root of the product of the critical J-value and the hysteresis. In Fig. 9, the logarithmic value of the fatigue life is shown with respect to the square root of the product of the critical J-value and the hysteresis. For the case of N650 for which the value of the square root was low, the fatigue life was also low. In contrast, for the case of N330 for which the value of the square root was high, the fatigue life was also high. As mentioned previously, the fatigue life increased as the displacement amplitude decreased. The regression lines for the test data are also shown in Fig. 9, and it is noteworthy that they t the test data well for all three displacement amplitudes. Thus, the regression lines can be expressed as the following Eq. (2) logN f ZA J C EL
Hysteresis EV 1 2 CB (2) Here, N f is the number of cycles at failure, J C is the critical J-value, and A and B are proportionality constants which are different for a different displacement amplitude. Since the fatigue life is given in cycles, i.e. it is dimensionless, the right-hand side of Eq. (2) should be made dimensionless by dividing the critical J-value and the hysteresis by some representative values, E, L and V, Fig. 8. Critical J-value as a function of the small-scale roughness. Fig. 9. Logarithmic value of the fatigue life as a function of [(J C /EL)(hysteresis/EV)] 1/2 . J.-H. Kim, H.-Y. Jeong / International Journal of Fatigue 27 (2005) 263272 269 of the rubber specimen. E is the representative Youngs modulus of the rubber compounds, and it is given as a function of IRHD (International Rubber Hardness Degree) as long as the hardness of the rubber is between 30 and 85 IRHD [12,13]. In addition, hardness measured by durometer type A is compatible with that in IRHD [12]. Since the NR compounds studied in this paper had the same hardness, 62, the representative Youngs modulus turned out to be 3.87 MPa. L was selected to be the radius of the minor axis, i.e. LZ5 mm. In addition, V was assumed to be the volume of the most deformable part of the specimen, i.e. between K2.5 and C2.5 mm from the middle plane of the specimen, and it turned out to be 549.78 mm 3 . However, note that as long as the same specimen is used to evaluate the fatigue life, any representative value can be used. Even if a different representative value is used, the constant A will be adjusted accordingly. The constants A and B are shown with respect to the double displacement amplitude D in Figs. 10 and 11, respectively. Note that there is a linear relationship of A or B to the double displacement amplitude. Thus, A and B can be expressed as a function of D as shown in the following Eqs. (3) and (4) A Za CD b L (3) B Zc CD d L (4) Plugging Eqs. (3) and (4) into (2), the logarithmic value of the fatigue life can be expressed as a function of the critical J-value, the hysteresis, and the double displacement amplitude as in the following Eq. (5) logN f Z a CD b L
J C EL
hysteresis EV 1 2 C a CD d L
Z0:623 J C EL
hysteresis EV 1 2 d L
K0:135 J C EL
hysteresis EV 1 2 K4:878 D L
C7:4 (5) In order to evaluate the adequacy of the nal equation, the equation is shown graphically along with the test data for the amplitude of 16 mm (shown in Fig. 12), 19 mm (shown in Fig. 13) and 22 mm (shown in Fig. 14). Note that the equation is in good agreement with the test data for all three amplitudes. In other words, the fatigue life can be Fig. 10. Coefcient A as a function of the double displacement amplitude (R 2 Z0.9777). Fig. 12. Comparison of the fatigue life equation with test data for the double displacement amplitude of 16 mm (R 2 Z0.9621). Fig. 11. Coefcient B as a function of the double displacement amplitude (R 2 Z0.9956). J.-H. Kim, H.-Y. Jeong / International Journal of Fatigue 27 (2005) 263272 270 predicted from the displacement amplitude and the material properties, the critical J-value and the hysteresis, which can be obtained from simple tests. Therefore, it may be said that when a rubber component is under development using various compositions, a composition for which the product of the critical J-value and the hysteresis is the highest will result in the best durability. In this study, the hysteresis was measured from the same specimen and at the same deformation mode as the fatigue life was evaluated. If hysteresis had been measured from a different specimen or at a different deformation mode, the fatigue life prediction equation (5) would not be in good agreement with the test data mainly because hysteresis is dependent on the strain amplitude and the load ratio [9,10]. Therefore, in order to predict the fatigue life based on the critical J-value and the hysteresis, more reliable prediction will be obtained if the hysteresis is measured at the same deformation mode and for the same rubber component for which the fatigue life is to be predicted. 4. Conclusion In this paper, NR compounds lled with three types of carbon black, N330, N650 and N990 were experimentally investigated in order to evaluate the effects of carbon black on the fatigue life, the hysteresis, the critical J-value and the fracture morphology. By reviewing all the test data, the following conclusions could be reached. (1) The fatigue life, the hysteresis and the critical J-value were ranked in the following order for the NR compound lled with N330, N990 or N650. (2) The small-scale roughness showed a trend opposite to that of the large-scale roughness. The hysteresis or the critical J-value increased as the small-scale roughness increased. (3) In the NR compound lled with N650, there existed large carbon black agglomerates, which separated from the rubber matrix comparatively easily. Thus, the fatigue life of the NR compound lled with N650 was shorter, and its fracture surface was cleaner in the small- scale than that of the NR compound lled with N330 or N990. (4) The logarithmic value of the fatigue life of the NR compounds was linearly proportional to the square root of the product of the critical J-value and the hysteresis. In addition, the proportionality constants were linearly proportional to the double displacement amplitude. Thus, the logarithmic value of the fatigue life could be expressed as a function of the double displacement amplitude, the critical J-value and the hysteresis. Acknowledgements The specimens were provided by Pyoung Hwa Industrial Company, and some experiments were conducted at Hankook Tire R&D Center. All of these supports are highly appreciated. References [1] Lake GJ, Lindley PB. Ozone cracking, ex cracking and fatigue of rubber. Rubber J 1964;146:309. [2] Mars WV, Fatemi A. A literature survey on fatigue analysis approaches for rubber. Int J Fatigue 2002;24:94961. [3] Sawyer LC, Grubb DT. Polymer microscopy, 2nd ed. London: Chapman & Hall; 1996. [4] Gent AN, Pulford CTR. Micromechanics of fracture in elastomers. J Mater Sci 1984;19:36129. [5] Greensmith HW. Tear properties of vulcanizates containing carbon black. J Polym Sci 1956;21:17587. [6] Hess FM, Ford FP. Microcopy of pigmentelastomer systems. Rubber Chem Technol 1963;36:1175228. [7] Goldberg A, Lesuer DR, Patt J. Observations made during stretching, tearing, and failure of NR and SBR loaded with various amounts of carbon black. Rubber Chem Technol 1988;62:288304. Fig. 14. Comparison of the fatigue life equation with test data for the double displacement amplitude of 22 mm (R 2 Z0.9983). Fig. 13. Comparison of the fatigue life equation with test data for the double displacement amplitude of 19 mm (R 2 Z0.9825). J.-H. Kim, H.-Y. Jeong / International Journal of Fatigue 27 (2005) 263272 271 [8] Chung B, Funt JM, Ouyang GB. Effect of carbon black on elastomer ultimate propertiesIR compounds. Rubber World 1991;204:4651. [9] Wang MJ. Effect of polymerller and llerller interactions on dynamic properties of lled vulcanizates. Rubber Chem Technol 1998;71:52089. [10] Legorju-jago K, Bathias C. Fatigue initiation and propagation in natural and synthetic rubbers. Int J Fatigue 2002;24:8592. [11] ASTM D1765. Standard classication system for carbon blacks used in rubber products. [12] ASTM D1415-88. Standard test method for rubber property international hardness. [13] Hawley SW. Anomalies in ISO 48, hardness of rubber. Polym Test 1997;16:32733. [14] ASTM D2240-97. Standard test method for rubber property durometer hardness. [15] ASTM D575-91. Standard test methods for rubber properties in compression. [16] Takeychi K, Nakagawa M, Yamaguchi H, Okumoto T. Fatigue test technique of rubber materials for vibration insulators and their evaluation. Int Polym Sci Technol 1993;20(10). [17] Mathew NM, Bhowmick AK, De SK. Chemical and scanning electron microscopy studies on fatigue failure of natural rubber. Rubber Chem Technol 1982;55:5161. [18] Young RJ, Lovell PA. Introduction to polymers. London: Chapman & Hall; 1991. [19] Mullins L. Softening of rubber by deformation. Rubber Chem Technol 1969;42(1):33962. [20] Mark JE, Erman B, Eirich FR. Science and technology of rubber. New York: Academic Press; 1994. p. 387418. [21] ASTM D2414-00. Standard test method for carbon black-n-dibutyl phthalate absorption number. [22] ASTM D3765-99. Standard test method for carbon black-CTAB (cetyltrimethylammonium bromide) surface area. [23] Lee BC. Effect of carbon black on the reinforcement and vibration damping of rubber vulcanization. Korea: Chungbuk National University; 1997 p. 703. [24] Wei YT, Tian ZH, Du XW. A nite element model for the rolling loss prediction and fracture analysis of radial tires. Tire Sci Technol, TSTCA 1999;27(4):25076. [25] Begley JA, Landes JD. The J-integral as a fracture criterion. ASTM STP 514. Philadelphia: American Chemical Society for Testing and Materials; 1972 p. 120. [26] ASTM D412-98a. Standard test method for vulcanized rubber and thermoplastic elastomers-tension. [27] Rivlin RS, Thomas AG. Characteristic energy for tearing. J Polym Sci 1953;10:291318. [28] Fukahori Y, Andrews EH. Fracture surface roughness in highly deformable polymers. J Mater Sci 1978;13:77785. [29] Lee DJ. Fracture, aging and microstructural changes in carbon black lled natural rubber.: University of Massachusetts; 1987. p. 10820. [30] Treloar LRG. The physics of rubber elasticity. Oxford: Clarendon Press; 1975. J.-H. Kim, H.-Y. Jeong / International Journal of Fatigue 27 (2005) 263272 272