魔 Availableonlineatwww.sciencedirect.col . ScienceDirect WEAR ELSEVIER Wear265(2008)612-619 www.elseviercom/locate/wea Grain size dependence of scratch-induced damage in alumina ceramics Michael Klecka a Ghatu Subhash b,* a Materials Science and Engineering, Michigan Technological University, Houghton, MI 49931, USA b Mechanical and Aerospace Engineering, University of Florida, 129 NEB, PO Box 116250, Gainesville fl 3261l usa Received 6 June 2007: received in revised form 9 November 2007; accepted 11 December 2007 Available online 19 February 2008 Variable depth single and double scratch experiments were conducted on alumina ceramics of three distinct grain sizes. Utilizing laser profilometry, SEM, and force transducers, an in-depth analysis of volume of material removed, microstructural damage features, and applied force, respectively, was performed as a function of grain size. In single scratches, for a given depth of cut the extent of lateral cracking and material removal volume increased with grain size. In interacting double scratches, a critical separation distance was observed where the extent of damage reached a maximum. For all three grain sizes, the volume of material removed at this critical separation distance reached four times the volume removed by a single scratch. This critical separation distance decreased with decreasing grain size. It was also noticed that the force required to create the second scratch in the vicinity of the first scratch reduced considerably depending on the separation distance. These results are rationalized on the basis of material properties and inherent flaw size Intergranular fracture and grain dislodgement were found to be the dominant modes of lateral C 2007 Elsevier B V. All rights reserved Keywords: Alumina; Scratch-induced damage; Interacting scratches; Separation distance 1. Introduction standing residual surface and subsurface damage imposed by grinding is necessary to minimize post-machining strength In order to gain a fundamental understanding of abrasive degradation in ceramic components (26, 27). To better repre material removal mechanisms and damage characteristics in sent the evolved damage due to either grinding or abrasive brittle materials, numerous studies using both static indenta- machining, interacting scratches are often utilized. Xu et al. tion and scratch processes have been conducted [l-11]. While [28] have studied the effect of scratch spacing and grain indentation tests impart only normal loads, scratch tests, on the size on resulting damage in low speed scratches (0.5 mm/s) other hand, impart both tangential and normal loads, and thus They found a substantial increase in material removal vol- can provide better insight into abrasion resistance of ceramic ume when scratches become adequately spaced. Studies which components and residual damage during grinding [ 12-18]. Alu- include microstructural effects in similar scratch experiment mina, being a widely used structural ceramic, has been the focus [19, 20, 25, 28] and grinding experiments [29-33] have provided of a number of such studies [ 19-25] additional understanding in terms of surface and subsurface Grinding accounts for a substantial cost in the damage tion of ceramic components [23, 26-28. Reducing ti The current study is aimed at understanding damage evo- on this step is critical for ceramics to be furthe lution in alumina under abrasive conditions as a function of into technical and structural applications. Furthermore, under- grain size. Damage induced by single and double scratches pro- duced by a rotating diamond tool was investigated. In-depth analysis of microstructural damage features, volume of material Corresponding author. Tel. +1 352 392 7005: fax: +1 352 392 7303 removed, and applied force was performed in order to investigate E-mail address. subhash@ufl.edu(G. Subhash). the influence of grain size on damage evolution. 0043-1648 front matter 2007 Elsevier B V. All rights reserved doi:10.1016wea2007.1202
Available online at www.sciencedirect.com Wear 265 (2008) 612–619 Grain size dependence of scratch-induced damage in alumina ceramics Michael Klecka a, Ghatu Subhash b,∗ a Materials Science and Engineering, Michigan Technological University, Houghton, MI 49931, USA b Mechanical and Aerospace Engineering, University of Florida, 129 NEB, P.O. Box 116250, Gainesville, FL 32611, USA Received 6 June 2007; received in revised form 9 November 2007; accepted 11 December 2007 Available online 19 February 2008 Abstract Variable depth single and double scratch experiments were conducted on alumina ceramics of three distinct grain sizes. Utilizing laser profilometry, SEM, and force transducers, an in-depth analysis of volume of material removed, microstructural damage features, and applied force, respectively, was performed as a function of grain size. In single scratches, for a given depth of cut the extent of lateral cracking and material removal volume increased with grain size. In interacting double scratches, a critical separation distance was observed where the extent of damage reached a maximum. For all three grain sizes, the volume of material removed at this critical separation distance reached four times the volume removed by a single scratch. This critical separation distance decreased with decreasing grain size. It was also noticed that the force required to create the second scratch in the vicinity of the first scratch reduced considerably depending on the separation distance. These results are rationalized on the basis of material properties and inherent flaw size. Intergranular fracture and grain dislodgement were found to be the dominant modes of lateral crack propagation. © 2007 Elsevier B.V. All rights reserved. Keywords: Alumina; Scratch-induced damage; Interacting scratches; Separation distance 1. Introduction In order to gain a fundamental understanding of abrasive material removal mechanisms and damage characteristics in brittle materials, numerous studies using both static indentation and scratch processes have been conducted [1–11]. While indentation tests impart only normal loads, scratch tests, on the other hand, impart both tangential and normal loads, and thus can provide better insight into abrasion resistance of ceramic components and residual damage during grinding [12–18]. Alumina, being a widely used structural ceramic, has been the focus of a number of such studies [19–25]. Grinding accounts for a substantial cost in the production of ceramic components [23,26–28]. Reducing time spent on this step is critical for ceramics to be further adopted into technical and structural applications. Furthermore, under- ∗ Corresponding author. Tel.: +1 352 392 7005; fax: +1 352 392 7303. E-mail address: subhash@ufl.edu (G. Subhash). standing residual surface and subsurface damage imposed by grinding is necessary to minimize post-machining strength degradation in ceramic components [26,27]. To better represent the evolved damage due to either grinding or abrasive machining, interacting scratches are often utilized. Xu et al. [28] have studied the effect of scratch spacing and grain size on resulting damage in low speed scratches (0.5 mm/s). They found a substantial increase in material removal volume when scratches become adequately spaced. Studies which include microstructural effects in similar scratch experiments [19,20,25,28] and grinding experiments [29–33] have provided additional understanding in terms of surface and subsurface damage. The current study is aimed at understanding damage evolution in alumina under abrasive conditions as a function of grain size. Damage induced by single and double scratches produced by a rotating diamond tool was investigated. In-depth analysis of microstructural damage features, volume of material removed, and applied force was performed in order to investigate the influence of grain size on damage evolution. 0043-1648/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.wear.2007.12.012
M. Klecka, G. Subhash/ Wear 265(2008)612-619 613 2. Experimental procedure ter of the transducer. The signal from the transducer is amplified and output to a digital oscilloscope which records the normal 2.1. Material preparation force during the scratch. Initially, scratches were spaced far enough apart to avoid any interactions. A range of scratch lengths Three cylindrical rods of fully densified 99.9% purity between a maximum of 2 mm and a minimum of 0.5 mm were alumina, each with different grain sizes(fabricated by produced to investigate the relationships among scratch depth dyne, Inc. Costa Mesa, CA), were obtained from Oak and damage zone size National Laboratories(ORNL), Oak Ridge, TN Specified grai Interacting scratches of equal length were produced sequen- sizes for the three rods were 2 um, 15 um, and 25 um(later tially with varying separation distances. The distance between confirmed by observing fracture surfaces). The rods were sec- these double-grit scratches was gradually reduced such that the tioned using a low speed saw to produce test samples sized residual damage left by the first scratch was allowed to influence Omm x 10 mm x 3 mm. The square surfaces were ground suc- the induced damage due to the second scratch. Scratch lengths cessively using 240, 320, and 400 grit SiC paper for no less within a pair of adjacent scratches were kept constant; however, than 20 min each. They were then polished successively using some minor variation in length was inherent in the test setup 35 um, 15 um, 6 um, 3 um, and finally 1 um diamond paste. This issue will be addressed in a later section Each step was continued for approximately 30-45 min until After each scratch test. the volume removed was mea evidence of previous polishing marks and grain dislodgement sured using an Optical Gauging Products Cobra 3D Laser was minimized as verified under an optical microscope. Vick- Profilometer. Single grit scratches were scanned separately ers macro-hardness measurements were also performed for while interacting scratches were scanned together in order each sample at a load of l kg. The average hardness for six calculate the combined removal volume from the 3D scratch indentations plus other relevant material properties are given in profiles. Optical microscopy and scanning electron microscopy Table 1 (SEM) were utilized to evaluate the extent of scratch damage and determine the fracture characteristics as a function of grain 2. 2. Scratch tests Scratch testing was performed using an instrumented scratch 3. Results and test fixture described in previous publications [13, 14, 34-36] In brief, the fixture contained a pendulum supported by two 3.. Single scratch high-precision bearings housed in a rigid steel frame designed to minimize extraneous vibrations. A 90 included angle phono- 3.1.1. Microstructural feature point diamond-dressing tool was inserted into the pendulum and An SEMimage, a laser profilometer scan, and the correspond used as the scratch tool. The reach of this tool tip can be adjusted ing force profile from a typical scratch of 1.5 mm length on the with a micrometer in the pendulum housing, allowing control 2 um grain size alumina are shown in Fig. 1. Due to the nature over scratch depth and length. A fully adjustable pneumatic pis- of the rotating grit scratch process, the groove starts at very low ton drives the pendulum, creating a scratch in the form of a depth, increases until reaching a maximum depth at the center of circular arc in a plane perpendicular to the specimen surface. the scratch path, and then gradually decreases in-depth for the Scratch velocities were controlled by regulating the nitrogen remainder of the scratch. The damage is minimal at the entry pressure feeding the piston. Typical scratch velocity was on the and exit of the scratches, while more intense damage is seen in order of 1.5 m/s. After each scratch, tool tips were observed the central region of the scratch. By varying the scratch length, under an optical microscope for signs of damage or excessive a large volume of data on induced damage for a range of scratch wear and replaced when necessary. Test samples were loaded depths was generated onto a moving platform with dial indicators to assist in select Throughout the analysis it was observed and confirmed that ing the designated scratch location. The specimen holder also the grain size distributions in the 15 um and 25 um samples houses a Kistler miniature force transducer(200kHz), posi- contained considerable overlap. This was noted in both micro- tioned to measure normal force during the scratch event The graphs and measured data during experiments which showed pecimen was positioned directly on top of the transducer, such similarities between these two grain sizes. While separate data that the scratch is conducted at a location directly above the cen- for the two grain sizes is maintained throughout this work, remain concise the 15 um and 25 um grain size samples are Table 1 often lumped together aslarge versus the 2 um grain Material properties for alumina samples size when considering qualitative analysis Grain size(um) 15 Short scratches below 1 mm length reveal a smooth scratch 25 Hv(GPa), I kg load groove interior and minimal brittle damage along the groove ural strength, ball on ring(MPa) 691 edges(see Fig. 2). Only at very low depth-of-cut does the scratch Modulus(GPa)- 399 a completely ductile appearance(e.g, at the entry Poisson’ s ratIo2 0. 237 0.237 0.237 and exit of the scratch 37)). The mode of material removal 3.95 3.96 3.97 in the damaged regions of short scratches was commonly dis- Supplied by ORNL. lodgement of individual grains along the groove edges. With
M. Klecka, G. Subhash / Wear 265 (2008) 612–619 613 2. Experimental procedure 2.1. Material preparation Three cylindrical rods of fully densified 99.9% purity alumina, each with different grain sizes (fabricated by Ceradyne, Inc. Costa Mesa, CA), were obtained from Oak Ridge National Laboratories (ORNL), Oak Ridge, TN. Specified grain sizes for the three rods were 2 m, 15m, and 25m (later confirmed by observing fracture surfaces). The rods were sectioned using a low speed saw to produce test samples sized 10 mm × 10 mm × 3 mm. The square surfaces were ground successively using 240, 320, and 400 grit SiC paper for no less than 20 min each. They were then polished successively using 35m, 15m, 6m, 3m, and finally 1m diamond paste. Each step was continued for approximately 30–45 min until evidence of previous polishing marks and grain dislodgement was minimized as verified under an optical microscope. Vickers macro-hardness measurements were also performed for each sample at a load of 1 kg. The average hardness for six indentations plus other relevant material properties are given in Table 1. 2.2. Scratch tests Scratch testing was performed using an instrumented scratch test fixture described in previous publications [13,14,34–36]. In brief, the fixture contained a pendulum supported by two high-precision bearings housed in a rigid steel frame designed to minimize extraneous vibrations. A 90◦ included angle phonopoint diamond-dressing tool was inserted into the pendulum and used as the scratch tool. The reach of this tool tip can be adjusted with a micrometer in the pendulum housing, allowing control over scratch depth and length. A fully adjustable pneumatic piston drives the pendulum, creating a scratch in the form of a circular arc in a plane perpendicular to the specimen surface. Scratch velocities were controlled by regulating the nitrogen pressure feeding the piston. Typical scratch velocity was on the order of 1.5 m/s. After each scratch, tool tips were observed under an optical microscope for signs of damage or excessive wear and replaced when necessary. Test samples were loaded onto a moving platform with dial indicators to assist in selecting the designated scratch location. The specimen holder also houses a Kistler miniature force transducer (200 kHz), positioned to measure normal force during the scratch event. The specimen was positioned directly on top of the transducer, such that the scratch is conducted at a location directly above the cenTable 1 Material properties for alumina samples Grain size (m) 2 15 25 Hv (GPa), 1 kg load 19.8 18.3 17.8 Flexural strength, ball on ring (MPa)a 691 459 408 Modulus (GPa)a 394 399 399 Poisson’s ratioa 0.237 0.237 0.237 Density (g/cc)a 3.95 3.96 3.97 a Supplied by ORNL. ter of the transducer. The signal from the transducer is amplified and output to a digital oscilloscope which records the normal force during the scratch. Initially, scratches were spaced far enough apart to avoid any interactions. A range of scratch lengths between a maximum of 2 mm and a minimum of 0.5 mm were produced to investigate the relationships among scratch depth and damage zone size. Interacting scratches of equal length were produced sequentially with varying separation distances. The distance between these double-grit scratches was gradually reduced such that the residual damage left by the first scratch was allowed to influence the induced damage due to the second scratch. Scratch lengths within a pair of adjacent scratches were kept constant; however, some minor variation in length was inherent in the test setup. This issue will be addressed in a later section. After each scratch test, the volume removed was measured using an Optical Gauging Products Cobra 3D Laser Profilometer. Single grit scratches were scanned separately; while interacting scratches were scanned together in order to calculate the combined removal volume from the 3D scratch profiles. Optical microscopy and scanning electron microscopy (SEM) were utilized to evaluate the extent of scratch damage and determine the fracture characteristics as a function of grain size. 3. Results and discussion 3.1. Single scratch 3.1.1. Microstructural features An SEM image, a laser profilometer scan, and the corresponding force profile from a typical scratch of 1.5 mm length on the 2m grain size alumina are shown in Fig. 1. Due to the nature of the rotating grit scratch process, the groove starts at very low depth, increases until reaching a maximum depth at the center of the scratch path, and then gradually decreases in-depth for the remainder of the scratch. The damage is minimal at the entry and exit of the scratches, while more intense damage is seen in the central region of the scratch. By varying the scratch length, a large volume of data on induced damage for a range of scratch depths was generated. Throughout the analysis it was observed and confirmed that the grain size distributions in the 15 m and 25 m samples contained considerable overlap. This was noted in both micrographs and measured data during experiments which showed similarities between these two grain sizes. While separate data for the two grain sizes is maintained throughout this work, to remain concise the 15m and 25m grain size samples are often lumped together as ‘large’ versus the ‘small’ 2 m grain size when considering qualitative analysis. Short scratches below 1 mm length reveal a smooth scratch groove interior and minimal brittle damage along the groove edges (see Fig. 2). Only at very low depth-of-cut does the scratch groove assume a completely ductile appearance (e.g., at the entry and exit of the scratch [37]). The mode of material removal in the damaged regions of short scratches was commonly dislodgement of individual grains along the groove edges. With
M. Klecka, G. Subhash/Wear 265(2008)612-619 八M Time(ms) Fig 1. Micrograph(a) of a medium length(1.5 mm) scratch, laser profilometer scan(b)and the corresponding force profile (c)on a 2 um grain size sample increasing scratch length (i.e, at higher depths of cut), however, lateral cracking starts to evolve as seen in the following sections Scratches in the range of 1.0-2.0 mm length generally contain a scratch groove with damage features similar in appearance for all three grain sizes. The scratch groove appears smooth throughout the length of the scratch, indicating the evidence of ductile mode deformation [34]. Depending on the depth of the scratch, considerable brittle damage surrounds the scratch groove in the form of lateral cracking resulting in large scale material removal. Fig 3 provides images of the central sections plastic groove from scratches on 2 um and 15 um grain size samples revealing the above damage features with further increase in scratch length, the damage is more scratch length =0. 83 mm severe and the central ductile region of the groove also becomes dislodged as shown in Fig. 4. It is clearly seen that intergranu lar fracture is the dominant mode of material removal for both grain dislodgement grain sizes. Grain dislodgement in the scratch path(beneath the groove)is more common in the larger grain size samples; while only in regions size samples display this type of damage As a supplement to the qualitative information on the nature of material removal and scratch features gathered from the above SEM analysis, additional data was collected from the scans pro- duced by the laser profilometer. Utilizing the scan data, a 3D wireframe model of each scratch was constructed [13, 14], see scratch length =0. 84 mm Fig. 1(b). The measured volume of material removed per scratch Fig. 2. Micrographs of central portions of short(<l mm) scratches revealing for each grain size as a function of overall scratch length was inimal brittle damage surrounding the groove and smooth groove edges in (a) determined from the profilometer scans and is summarized in um and(b)25 um grain size alumina. Fig. 5. Although the depth of groove can be measured from
614 M. Klecka, G. Subhash / Wear 265 (2008) 612–619 Fig. 1. Micrograph (a) of a medium length (1.5 mm) scratch, laser profilometer scan (b) and the corresponding force profile (c) on a 2m grain size sample. Fig. 2. Micrographs of central portions of short (<1 mm) scratches revealing minimal brittle damage surrounding the groove and smooth groove edges in (a) 2m and (b) 25m grain size alumina. increasing scratch length (i.e., at higher depths of cut), however, lateral cracking starts to evolve as seen in the following sections. Scratches in the range of 1.0–2.0 mm length generally contain a scratch groove with damage features similar in appearance for all three grain sizes. The scratch groove appears smooth throughout the length of the scratch, indicating the evidence of ductile mode deformation [34]. Depending on the depth of the scratch, considerable brittle damage surrounds the scratch groove in the form of lateral cracking resulting in large scale material removal. Fig. 3 provides images of the central sections from scratches on 2 m and 15m grain size samples revealing the above damage features. With further increase in scratch length, the damage is more severe and the central ductile region of the groove also becomes dislodged as shown in Fig. 4. It is clearly seen that intergranular fracture is the dominant mode of material removal for both grain sizes. Grain dislodgement in the scratch path (beneath the groove) is more common in the larger grain size samples; while only in regions with very high depth of cut do the 2m grain size samples display this type of damage. 3.1.2. Material removal volume As a supplement to the qualitative information on the nature of material removal and scratch features gathered from the above SEM analysis, additional data was collected from the scans produced by the laser profilometer. Utilizing the scan data, a 3D wireframe model of each scratch was constructed [13,14], see Fig. 1(b). The measured volume of material removed per scratch for each grain size as a function of overall scratch length was determined from the profilometer scans and is summarized in Fig. 5. Although the depth of groove can be measured from
M. Klecka, G. Subhash/Wear 265(2008)612-619 615 plastic groove scratch length= 1. 26 mm scratch length=1.96 mm lateral cracking scratch length=1.22 mm Fig. 3. Central sections of medium length(1-2 mm)scratches revealing brittle scratch length= 1.94 mm cracking surrounding the groove in(a)2 um and(b)15 um grain size alumina. Fig. 4. Central sections of large scratches revealing damage that extends beneath the central scratch groove in(a)2 um and(b)15 um grain size alumina. Note that the damage mode is predominantly intergranul the laser scans, cracks also extend beneath the scratch groove single grain sized 15 um or 25 um as c into the bulk material. In some places where large fragments ared to a single 2 um are removed, the groove depth can be significantly greater than the depth of cut(e.g. Fig. 4). For this reason, data is plotted The second noticeable trend is that at low depth of cut (i.e as a function of scratch length rather than scratch depth, as the below a scratch length of 1 mm)the volume removed for a sin- depth of cut can no longer be measured accurately with the pro- gle scratch increases almost linearly with depth as shown in nilometer. The plot includes trend lines following a power-law the inset of Fig. 5. This is due to the absence of large lateral onset of lateral crack propagation ume removal comes primarily from the ductile mode of material From Fig. 5, a number of observations are noted. First, the vol- ume of material removed for scratches of similar length increases 2mR2a096 with larger grain size; i.e., scratches on the 2 um samples remove 06 considerably less volume of material than in the larger grai sizes. The data on 15 um and 25 um alumina are more similar. due to considerable overlap in grain sizes and microstructural features. The data for scratches on the 2 um samples appear to more closely follow the trend line, while the larger grain sizes 20.000400 50Q60070080090100 exhibit considerable scatter in the data. As will be seen in the 20.000300 following sections, this is a recurring issue when dealing with scratches on larger grain size samples due to a number of fac- tors. In general, larger grain size ceramics contain inherently larger flaws. Being the controlling factor in strength behavior of ceramics,larger flaws increase the relative volume of material Scratch Length(mm) removed for each scratch. Another consequence is that consid- Fig. 5. Volume of material removed for all three grain sizes as a function of erably more volume is removed during a scratch by plucking a scratch length
M. Klecka, G. Subhash / Wear 265 (2008) 612–619 615 Fig. 3. Central sections of medium length (1–2 mm) scratches revealing brittle cracking surrounding the groove in (a) 2m and (b) 15m grain size alumina. the laser scans, cracks also extend beneath the scratch groove into the bulk material. In some places where large fragments are removed, the groove depth can be significantly greater than the depth of cut (e.g. Fig. 4). For this reason, data is plotted as a function of scratch length rather than scratch depth, as the depth of cut can no longer be measured accurately with the pro- filometer. The plot includes trend lines following a power-law function to illustrate the increasing material removal after the onset of lateral crack propagation. FromFig. 5, a number of observations are noted. First, the volume of material removed for scratches of similar length increases with larger grain size; i.e., scratches on the 2 m samples remove considerably less volume of material than in the larger grain sizes. The data on 15m and 25m alumina are more similar, due to considerable overlap in grain sizes and microstructural features. The data for scratches on the 2 m samples appear to more closely follow the trend line, while the larger grain sizes exhibit considerable scatter in the data. As will be seen in the following sections, this is a recurring issue when dealing with scratches on larger grain size samples due to a number of factors. In general, larger grain size ceramics contain inherently larger flaws. Being the controlling factor in strength behavior of ceramics, larger flaws increase the relative volume of material removed for each scratch. Another consequence is that considerably more volume is removed during a scratch by plucking a Fig. 4. Central sections of large scratches revealing damage that extends beneath the central scratch groove in (a) 2m and (b) 15m grain size alumina. Note that the damage mode is predominantly intergranular. single grain sized 15m or 25m as compared to a single 2m grain. The second noticeable trend is that at low depth of cut (i.e., below a scratch length of 1 mm) the volume removed for a single scratch increases almost linearly with depth as shown in the inset of Fig. 5. This is due to the absence of large lateral cracking in these short scratches (see Fig. 2). In this case, the volume removal comes primarily from the ductile mode of material Fig. 5. Volume of material removed for all three grain sizes as a function of scratch length.
616 M. Klecka, G. Subhash/Wear 265(2008)612-619 removal within the scratch groove, with a smaller contribution reduced, interaction between scratches intensified, resulting in from slight chipping along the scratch edges. A detailed descrip- an increased material removal volume due to extended damage tion of low depth scratches is presented in a separate manuscript in the region between the two scratches. Cracking induced dur- [37]. In these short scratches, material removal volume was still ing the first scratch may also reactivate and propagate farther higher in the larger grain samples: another consequence of the due to the stress state imposed by a neighboring scratch volume associated with the dislodgement of isolated large grains Figs. 7 and 8 depict resulting damage in paired scratches with along the scratch edges. With increasing scratch length, large varied spacing(measured from scratch center to center) for all grain samples also show a faster increase in removal volume, three grain size samples. For each pair of scratches shown in due to the earlier onset of lateral cracking the figures, the top scratch was produced first while the bottom scratch of approximately equal length was produced second. As 3.1.3. Scratch force profiles lateral cracking extends in a similar manner in all three grain From the force transducer mounted under the specimen, data sizes, images with a variety of scratch separation distances were regarding the normal force required to produce a scratch was selected in order to provide an overview of crack interaction gathered for individual scratches[13, 14]. One such force profile features. Note that the intensity of damage is more severe in the was shown in Fig. 1(c). Large fluctuations in normal force are central region of the pairs of scratches. When the second scratch generally indicative of removal of large chips from the sample is in the vicinity of the first scratch, the overall damage produced ] These oscillations in the force profiles reflect the intermit- surrounding the second scratch is also significantly greater tha tent nature of material removal during scratching. Fig. 6 displays the first scratch. Fig 8 contains two pairs of interacting scratches the average force required as a function of scratch length for the on the 25 um grain size sample where a small change in the sep- three grain sizes. In general, the average force for a given scratch aration distance results in nearly two times the volume removed, length decreases with increasing grain size. This result is in gen- reflecting a greater sensitivity to the scratch separation distance ral agreement with the trends observed in both the hardness and in the larger grain size samples. the characteristic strength measurements summarized in Table 1 The measured volume of material removed for each pair of Although there is some scatter, the overall trends indicate that the interacting scratches is summarized in Fig. 9. Due to the varia- smaller grain size ceramic required considerably higher force for tions in scratch length between all pairs of scratches considered damage when conducting tests on the smaller grained samples. normalized with respect to the volume removed by a single non- The data for both the 15 um and 25 um grain size samples are interacting scratch of equal length on the same grain size sample closer together and show a similar trend. As mentioned before, The resulting normalized data for all three grain sizes is plotted this is due to the similarities in microstructure and closeness in as a function of scratch separation distance. A fourth order poly- graIn size nomial trend line has been added to the figure in an effort to better illustrate the trends in the data and to identify the scratch sep- 3.2. Double scratch features aration spacing at maximum removal volume. The plot shows that for non-interacting scratches (i.e, large scratch spacing) Interacting pairs of scratches were produced in order to fur- the volume removed for two scratches is equivalent to twice ther investigate the effect of grain size on the evolved damage. the volume removed by a single scratch of equal length. As the All scratches were produced sequentially at varying separation spacing between the pairs of scratches is reduced, interaction distances in order to determine how the residual damage from between the scratches begins and an increase in the volume of the first scratch interacts with and amplifies damage due to material removed for each pair of scratches is observed. The a neighboring scratch. As the scratch separation distance was data indicates that scratches on the 25 um grain size samples begin to interact at a greater spacing(200 um) than scratches on 15 um samples(180 um), followed by scratches on the 2 um samples(150 pm). This corresponds well with the previous trend on extent of damage(or the volume removed) in single scratches with increasing grain size(e.g, Fig. 5). SEM analysis revealed that damage among all three grain sizes occurred primarily in the form of intergranular cracking; however, some amount of transgranular cracking was also present in the larger grain size samples. Large lateral cracking still remains as the principa mechanism for material removal. It is likely that some subsur 叹 face cracks which may not produce material removal during the first scratch alone are allowed to propagate further during th 0.50 0.75 1.00 1.25 1.50 1.75 2.00 second scratch and successfully remove additional material cratch Length(mm) It appears from Fig 9 that a critical separation distance exists Fig. 6. Plot of average normal force as a function of length for all three grain for each grain size at which the maximum volume of material is removed for a pair of interacting sequential scratches. Following
616 M. Klecka, G. Subhash / Wear 265 (2008) 612–619 removal within the scratch groove, with a smaller contribution from slight chipping along the scratch edges. A detailed description of low depth scratches is presented in a separate manuscript [37]. In these short scratches, material removal volume was still higher in the larger grain samples: another consequence of the volume associated with the dislodgement of isolated large grains along the scratch edges. With increasing scratch length, large grain samples also show a faster increase in removal volume, due to the earlier onset of lateral cracking. 3.1.3. Scratch force profiles From the force transducer mounted under the specimen, data regarding the normal force required to produce a scratch was gathered for individual scratches [13,14]. One such force profile was shown in Fig. 1(c). Large fluctuations in normal force are generally indicative of removal of large chips from the sample [34]. These oscillations in the force profiles reflect the intermittent nature of material removal during scratching. Fig. 6 displays the average force required as a function of scratch length for the three grain sizes. In general, the average force for a given scratch length decreases with increasing grain size. This result is in general agreement with the trends observed in both the hardness and the characteristic strength measurements summarized in Table 1. Although there is some scatter, the overall trends indicate that the smaller grain size ceramic required considerably higher force for a given scratch length than the larger grain ceramics. This result was also confirmed by the increased tendency for scratch tip damage when conducting tests on the smaller grained samples. The data for both the 15m and 25m grain size samples are closer together and show a similar trend. As mentioned before, this is due to the similarities in microstructure and closeness in grain size. 3.2. Double scratch features Interacting pairs of scratches were produced in order to further investigate the effect of grain size on the evolved damage. All scratches were produced sequentially at varying separation distances in order to determine how the residual damage from the first scratch interacts with and amplifies damage due to a neighboring scratch. As the scratch separation distance was Fig. 6. Plot of average normal force as a function of length for all three grain sizes. reduced, interaction between scratches intensified, resulting in an increased material removal volume due to extended damage in the region between the two scratches. Cracking induced during the first scratch may also reactivate and propagate farther due to the stress state imposed by a neighboring scratch. Figs. 7 and 8 depict resulting damage in paired scratches with varied spacing (measured from scratch center to center) for all three grain size samples. For each pair of scratches shown in the figures, the top scratch was produced first while the bottom scratch of approximately equal length was produced second. As lateral cracking extends in a similar manner in all three grain sizes, images with a variety of scratch separation distances were selected in order to provide an overview of crack interaction features. Note that the intensity of damage is more severe in the central region of the pairs of scratches. When the second scratch is in the vicinity of the first scratch, the overall damage produced surrounding the second scratch is also significantly greater than the first scratch. Fig. 8 contains two pairs of interacting scratches on the 25m grain size sample where a small change in the separation distance results in nearly two times the volume removed, reflecting a greater sensitivity to the scratch separation distance in the larger grain size samples. The measured volume of material removed for each pair of interacting scratches is summarized in Fig. 9. Due to the variations in scratch length between all pairs of scratches considered, the raw data is difficult to interpret. Instead, the measured volume of material removed for each pair of double scratches was normalized with respect to the volume removed by a single noninteracting scratch of equal length on the same grain size sample. The resulting normalized data for all three grain sizes is plotted as a function of scratch separation distance. A fourth order polynomial trend line has been added to the figure in an effort to better illustrate the trends in the data and to identify the scratch separation spacing at maximum removal volume. The plot shows that for non-interacting scratches (i.e., large scratch spacing), the volume removed for two scratches is equivalent to twice the volume removed by a single scratch of equal length. As the spacing between the pairs of scratches is reduced, interaction between the scratches begins and an increase in the volume of material removed for each pair of scratches is observed. The data indicates that scratches on the 25 m grain size samples begin to interact at a greater spacing (200m) than scratches on 15m samples (180 m), followed by scratches on the 2 m samples (150m). This corresponds well with the previous trend on extent of damage (or the volume removed) in single scratches with increasing grain size (e.g., Fig. 5). SEM analysis revealed that damage among all three grain sizes occurred primarily in the form of intergranular cracking; however, some amount of transgranular cracking was also present in the larger grain size samples. Large lateral cracking still remains as the principal mechanism for material removal. It is likely that some subsurface cracks which may not produce material removal during the first scratch alone are allowed to propagate further during the second scratch and successfully remove additional material. It appears from Fig. 9 that a critical separation distance exists for each grain size at which the maximum volume of material is removed for a pair of interacting sequential scratches. Following