A new test methodology for evaluating scratch resistance of polymers

Wear 256 (2004) 1214–1227

A new test methodology for evaluating scratch resistance of polymers M. Wong, G.T. Lim, A. Moyse, J.N. Reddy, H.-J. Sue∗ Department of Mechanical Engineering, Polymer Technology Center, Texas A&M University, College Station, TX 77843-3123, USA Received 30 May 2003; received in revised form 3 October 2003; accepted 8 October 2003

Abstract A new approach to conduct scratch tests on polymers is proposed. A brief review of the different definitions of scratch resistance and surface damages, currently available scratch test devices and the need for such a new approach is given. The new test methodology is developed using the concepts from materials science and solid mechanics, which include the consideration of material parameters, use of microscopy for image analysis and the finite element method (FEM). The consistency and reproducibility of test results are shown using a new scratch test device on two sets of neat and talc-filled polypropylene (PP) systems. Three different test conditions, i.e. linear load increase under constant speed, constant load under constant speed, and linear speed increase under constant load, have been conducted to determine the most effective, informative test conditions for evaluation of scratch resistance of polymers. The effect of loading and scratch speeds on the scratch-induced deformation of PP is also investigated using FEM simulations. Experimental observations and FEM results show a good qualitative correlation. The unique advantages of the new scratch test method for evaluating scratch resistance of polymers are discussed. © 2003 Elsevier B.V. All rights reserved. Keywords: Mar; Scratch; Surface damage; Test method; Finite element modeling; Polypropylene

1. Introduction and review of test methods Scratch deformation of polymeric surfaces has become an important area of research in the field of materials science and mechanics. The surge of interest in the subject of scratch resistance stems from the increasing use of polymers in applications. It is generally recognized that there are two types of surface damage—mar and scratch. A mar is a mark caused by a sliding body that is too shallow to be perceived by the casual human eyes alone but nevertheless does become visible when present in large quantities. Good examples are the typical damage found on paint coats and dashboard surfaces damaged by small pointed objects such as rough stones, sticks, keys, etc. A scratch is a mark that forms visible grooves and/or surface damage; this is the typical damage mode for surfaces that withstand heavy moving loads by swivels, ball bearings, etc. The complexity of the subject is underlined by the numerous other factors that influence the material response of polymers to scratches; these include scratch loads and speeds [1–4], coefficients of friction [1–4], geometry [1,4] and number of scratch tips, amount and types of fillers or additives [5–7].

∗ Corresponding author. Tel.: +1-979-845-5024; fax: +1-979-862-3989. E-mail address: [email protected] (H.-J. Sue).

0043-1648/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.wear.2003.10.027

The effect of fillers on scratch resistance is a major concern since almost any commercial plastic contains some amount of fillers. Fillers often have a deleterious effect on the surface appearance of the polymer due to the poor scratch resistance it imparts. The reason(s) for such an effect is still poorly understood [8] and it will be shown in this work the usefulness of the current method in investigating this effect. Until now, there is no generally recognized method for the quantification of the scratch resistance or surface damage of polymers. Analogous to the indentation hardness, Williams [9] proposed the definition of scratch hardness and ploughing hardness to characterize the scratch resistance of metals. Other parameters such as tangential hardness, dynamic hardness and specific grooving energy were considered by various authors [10–12]. Amongst these parameters, scratch hardness has gained popularity with an ASTM test method [13] published in recent months. Severe shortcomings for obtaining consistent and reproducible data using these definitions have yet to be overcome. Firstly, to apply these definitions, scratch widths and the corresponding contact areas need to be determined from experiments and are not easily quantified, especially when the width values scatter during scratch; hence, their readings can be subjective and difficult to reproduce. Secondly, as most materials will undergo elastic and viscoelastic recovery upon unloading, scratch deformation recovery can reach as much as 70–95%

M. Wong et al. / Wear 256 (2004) 1214–1227

of the penetration depth for polymers. The issue of when these contact areas should be measured will arise, which might give very different results at different times. As for the quantification of surface damage, Gauthier and Schirrer [14], Krupiˆcka and Johansson [15] and Sue and coworkers [16,17] have attempted to define scratch damage based on the measurements from the geometry of the cross-sectional scratch profiles. One of the objectives of this paper is to provide a test method that can address this issue by showing that highly reproducible data can be obtained with ease. It is essential to perform tests using a reliable scratch test device. A reliable scratch test device is defined as one that can produce consistent and reproducible data, and has reasonable capability and scalability in changing test conditions to capture the essential scratch characteristics of the test specimen. Over the years, numerous scratch test devices have been built commercially or custom-built by researchers to study scratch responses of polymers at various length scales. At the macroscopic scale of testing, there are simplistic test methods like the pencil hardness test [18,19] and others that employ more sophisticated devices like the scratching machine [1,2,20,21], Taber test and pin-on-disc machine [5,22], Ford five-finger test [6,7,23,24], single-pass pendulum sclerometer [10–12,25], scratch apparatus [14], Revetest scratch tester [15], needle test [26], scratch test rig [27] and in-house scratch test apparatus [28]. To perform scratch tests at the micro- and nano-meter scales, one can turn to several commercially available machines, or customized test machines built by individual researchers [29–33]. Scanning probe microscopy instruments, such as atomic force microscope [34–36], are also adopted and modified by researchers to perform scratch tests at such small scales. The review of current scratch test devices readily reveals that the ranges of normal loads and scratch speeds for most devices are rather limited while some of them may only be good for the evaluation of mar and thus insufficient for scratch studies. All of the current test devices reviewed are unable to specify a critical quantity, such as load or speed, that may be useful in comparing scratch resistance. The fundamental study of the scratch behavior of polymers if based solely on material science and experimental efforts is naturally inadequate. Understanding of the mechanical process of scratch is essential if we are to make any quantitative predictions in the scratch behaviors of polymers. However, the complexity of the subject easily renders analytical solutions unattainable and requires the use of computational techniques like the finite element method (FEM) to solve the problem [37,38]. Very often, there is a lack of synergy between materials science and mechanics approaches. A careful survey of the documentation on the existing test devices may also reveal that there exists little or no correlation of experimental test results with solutions computed analytically or numerically (e.g. from the FEM) [2,5,20,21,27,28], thereby restricting the interpretation of the experimental results solely via imprecise materials science reasoning.

1215

The need to implement a new scratch methodology that can characterize scratch properties of polymers at the macroscale is emphasized. There is no widely accepted method in the quantification of scratch properties of commercial polymers. One of the main objectives of this work is to propose such a methodology that is based on the current state of scratch research and valuable inputs from the automotive industry. The methodology primarily comprises of a new scratch test device and is accompanied by a detailed description of the test procedures and explicit definitions to quantify surface damages. The proposed test device has several attributes including the ability to carry out multi-pass and load-controlled scratch tests using linearly increasing speed- and load-modes as well as to provide corresponding normal and tangential load data during scratching. From the scratch tests on neat and talc-filled polypropylene (PP), it will be illustrated that the various testing capabilities of our custom-built test device allows one to capture the change in the scratch damage and the transition of damage modes, thereby providing a critical quantity that can be used in the characterization of scratch property. The consistency and reproducibility of experimental results from the new test device will be emphasized. Another objective is to implement a three-dimensional finite element analysis (FEA) for simulating the experimental test conditions on PP material. It will be shown that the numerical results indicate a good qualitative correlation with the experimental observation. Using the framework of mechanics and the numerical solutions, discussion on the explanation of the various experimental phenomena will be presented. Through the numerical results and the discussion presented, it will be demonstrated that the use of the finite element analysis can lead to a better understanding of the scratch behavior of polymers.

2. Experimental study 2.1. Custom-built scratch test device A new scratch device was developed for this research. Although the focus of the research is mainly on automotive applications, the custom-built scratch device schematically illustrated in Fig. 1 is designed with various functionalities to address macroscopic scratch issues for a wide range of applications. These various functionalities are discussed below. The scratch test device is built with the capability to execute multi-pass, multi-indenter, constant load, constant speed, increasing load and increasing speed test with the potential to operate under various ambient temperatures. The scratch test unit comprises of a servo gear-driven motor that drives the scratch tips or styli with constant or linearly increased speeds. For constant speeds, the stylus can move in a range from 0 to 400 mm/s. As for linearly increased speeds, the stylus can be set to move from a zero speed to a peak speed of 400 mm/s. A choice of up to five scratching styli

1216

M. Wong et al. / Wear 256 (2004) 1214–1227

Fig. 1. (a) Schematic of the spring-loaded scratch stylus, (b) photograph of spring-loaded scratch stylus and (c) photograph of dead-loaded scratch stylus.

can be used for the scratch test device to perform single- or multi-pass tests. The test device is also designed to conduct tests with dead weights or load-controlled spring loads. This allows the test device to have a wider load range for testing: 0–50 N for dead weights and 0–100 N for spring loads with a load cell sensitivity of 0.01 N. The reasons for incorporating spring loads are not only to allow for operation of increasing-load tests but also to prevent the occurrence of chattering of indenters as found in the dead weights loading case [39]. The test device is also equipped with sensing and data acquisition functions to record vital test data during testing, such as the tangential force acting on the stylus with an accuracy of 0.1 N for a load range up to 1000 N. The data acquired for depth, horizontal position and velocity of the stylus have accuracies of 5 ␮m, 5 ␮m and 10 ␮m/s, respectively. During tests, these test data will be sent to an external computer for data storage and processing. Test parameters, such as number of scratch passes, start and end positions and speeds of the stylus, are controlled through an on-board microprocessor housed in an instrumentation unit. An environmental chamber will be incorporated to house the test device (not shown in Fig. 1) so that scratch tests can be conducted under specified temperatures (−50 to 100 ◦ C).

Table 1 shows the comparison of the functionalities between our test device and the selected devices in the literature [1,2,15,20,21,27,28]. It is indicative from the table that the new test device presented herein, despite its relatively low construction cost, has the necessary capabilities to satisfy the needs of the polymer industry. More importantly, researchers can use the new scratch test device to design a variety of scratch tests on different polymeric bulk or coating systems through its various intended functionalities; some of these suggested tests are shown in Table 2. Several of the suggested tests are applied to the model polypropylene systems to illustrate their usefulness in scratch characterization. In the description of the scratch tests, emphasis will be placed on the test procedure and scratch damage quantification to help establishing a standard test method for scratch evaluation of polymers, which is among the most urgent needs of this field. 2.2. Model material system and test procedures In this study, four PP-based material systems are selected and their compositions are shown in Table 3. For these material systems, the PP resin and a dark gray coloring pigment was provided and blended by Solvay Engineered Polymers.

Functionality

Scratching machine by Briscoe and coworkers [1,2,20,21]

Scratch apparatus by Gauthier and Schirrer [14]

Scratch test rig by Wang et al. [27]

In-house scratch test apparatus by Ni and Faou [28]

Revetest scratch tester [15]

Current custom-built scratch device

Constant load test (range)

Yes—dead weights

Yes (0.05–5 N)

Yes (1–100 N)

Yes (0.1–10 N)

Yes (1–200 N)

Constant speed test (range) Increasing load test (range) Increasing speed test (range) Temperature control (range)

Yes (0.001–40 mm/s) No No Yes

Yes (0.01–100 mm/s) No Yes (0.01–100 mm/s) Yes (−70 to 120 ◦ C)

Yes (1–200 mm/s) Yes (1–100 N) No Yes

Yes (0.011–0.46 mm/s) No No No

Yes (0.003–6.67 mm/s) Yes (0.01–30 N) No No

Multi-indenter test Data acquisition Optical observatory device

No Yes No

No Yes No

No Yes No

No Yes No

No Yes Yes

Yes (0–50 N : dead weight) (5–100 N : spring load) Yes (0–400 mm/s) Yes (5–100 N) Yes (0–400 mm/s) Module to be added soon (−50 to 100 ◦ C) Yes Yes Provision provided for upgrading

M. Wong et al. / Wear 256 (2004) 1214–1227

Table 1 Comparison of functionalities of different scratch devices

1217

1218

M. Wong et al. / Wear 256 (2004) 1214–1227

Table 2 Suggested tests for scratch characterization Important scratch characterization

Suggested tests

Effect of scratch speed Effect of scratch load Effect of temperatures

Increasing speed tests Increasing load tests Scratch tests with environmental chamber Multiple-indenter tests

Influence of multiple scratches

Table 3 Composition of PP systems Material system

PP type

Filler

Coloring compound

1 2 3 4

Homopolymer Homopolymer Copolymer Copolymer

– Talc (20 wt.%) – Talc (20 wt.%)

2NCA 2NCA 2NCA 2NCA

(2 wt.%) (2 wt.%) (2 wt.%) (2 wt.%)

Talc additive was provided by Luzenac. Injection molding of the plaques, having dimensions of 340 mm × 180 mm × 3 mm, was performed by Advanced Composites Inc. For testing, the plaques were cut and machined into dimensions of 140 mm × 10 mm × 3 mm. All test specimens were prepared according to ASTM D 618-99 Procedure A [40]. Three sets of scratch tests (Tests A–C) were conducted. In Test A, a constant stylus speed of 100 mm/s with a linear increasing normal load of 0–50 N was performed. While in Test B, a 30 N dead load was utilized with a constant stylus speed of 100 mm/s, which is consistent with the Ford five-finger test. Finally for Test C, a dead weight of 30 N was used with a linearly accelerated stylus speed of 0–140 mm/s. The scratch lengths of all tests were set to be 100 mm and tests were conducted at room temperature. Stainless steel ball with a diameter of 1 mm was used as the scratch stylus tip.

into 2-cm long rectangular blocks, and mounted in an epoxy resin. The mounted polymer block was glued onto a microslide and further cut down to a 2-mm thick section by an ISOMET® 1000 diamond saw. The thick sections were then polished to a thickness of 100–150 ␮m, using polishing papers stepwise with roughness from grit 800 to grit 4000 (grain size 5 ␮m) to achieve the final polish. Scanning electron microscopy (SEM) was also performed to study the microscale surface damage features using a JEOL JSM-6400 system. A flatbed scanner with a resolution of 1200 dpi was used to scan the test specimens and generate digital images for the quantification of scratch damage. To quantify the scratch damage, measurements were taken from the TOM, SEM and scanned images using the definitions of scratch widths and depths by Kotaki et al. [16], as shown in Fig. 2. SW1 represents the inner width of the scratch groove. SW2 represents the outer width of the scratch groove, i.e. the distance between the points where the slopes of the hills meet the unscratched plane. SD1 represents the depth of the scratch groove calculated from the unscratched plane. SD2 is the height of the peak to the trough of the scratch groove. For spherical indenters, the scratch grooves generally show a symmetric cross-sectional profile. In cases where asymmetry occurs, i.e. one side of the pile-up is higher than the other, the higher point was taken to obtain scratch depths.

3. Finite element analysis The finite element method [41] is used as the numerical tool to help elucidate the phenomena observed in the experiments. A well-established commercial package ABAQUS/Explicit® [42] has been adopted to perform the finite element analysis of the concerned problem.

2.3. Scratch damage evaluation and quantification

3.1. Computational model and its material properties

Transmission optical microscopy (TOM) observation, using an Olympus® BX60 microscope, of thin sections of PP systems was performed to study the scratch damage of selected cross-sections along and across the scratch groove. The thin sections were prepared by cutting the polymer strips

The modeling work is primarily set out to model the scratch problem as closely and realistically as possible to the actual testing conditions. As highlighted in [17,43], the simulation of the scratch process by a spherical indenter on a polymeric surface ought to be carried out using a

Fig. 2. (a) Definitions of scratch widths and scratch depths and (b) actual cross-section of a scratch groove.

M. Wong et al. / Wear 256 (2004) 1214–1227

1219

Fig. 3. Computational model with a plane of symmetry.

50

True Stress s (MPa)

40

30 Rate = 1/sec 20

Rate = 0.1/sec Rate = 0.01/sec

10

0 0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

True Strain e

Fig. 4. True compressive stress–strain curves of PP under various strain rates at 23 ◦ C [44].

three-dimensional FE modeling coupled with a realistic material model [37,38]. A computational model of 50 mm × 10 mm × 3 mm was first considered. By exploiting the plane of symmetry, the computational model was reduced to the dimensions of 50 mm × 5 mm × 3 mm, as illustrated in Fig. 3. Not only will it save computational resources, the results of the reduced computational model can be extended to those of the original model. For the indenter, its tip was modeled to be spherical in shape, with a diameter of 1 mm. The material of the computational model is taken to be PP with a density of 0.905 g/cm3 , Young modulus of 1.65 GPa and Poisson’s ratio of 0.4. To account for the viscoelastic behavior of PP at various strain rates, the present study simply adopts the true stress–strain curves of PP at various strain rates [44] as the material input for FEA (Fig. 4). The indenter was assumed to be rigid. A three-dimensional dynamic and plastic stress analysis was executed. For the plastic analysis, von Mises yielding criterion with isotropic hardening rule was implemented. Coulomb frictional model was included and the coefficient of friction between the contact surfaces was taken to be 0.3. For a more detailed discussion of the setup, boundary and

loading conditions of the computational model and various considerations of the FEA, one can refer to [17,43,45].

4. Results and discussion 4.1. Experimental results The scratch damage cross-sectional profile is reported based on an average of five specimens for each test condition. For Test A, the cross-section was taken at a location where the normal load is equivalent to 30 N load. While for Test C, the cross-section was taken at a location where the scratch speed corresponds to 100 mm/s. In this way, the three tests could be compared under the same loads and speeds of 30 N and 100 mm/s. Following the definition specified in Fig. 2, the trend suggests that the scratch width is the greatest for Test C, followed by Tests B and A (Fig. 5a). This trend has also been observed in FEA modeling (Fig. 5b), which will be discussed in Section 4.2. For Test C, the accelerating scratch tip will induce both horizontal (in the direction of scratch) and vertical inertias (acting downwards). The vertical inertia

1220

M. Wong et al. / Wear 256 (2004) 1214–1227

Fig. 5. Comparison of (a) experimental and (b) FEA results.

induced is due to the frictional effect. Both of the inertia will increase the normal and tangential forces acting on the substrate, thereby increasing the scratch width and depth. While the increasing load imposed in Test A also induced additional vertical inertia, the magnitude of the induced inertia is much smaller than that for Test C. With the presence of induced inertia, it is however contrary to the engineering intuition that the scratch width for Test A is smaller than that for Test B, where there should not be any additional inertia induced. This anomaly will be subjected to our further study. One possible reason for such an anomaly is because of the pre-existing high penetration depth due to the high initial dead load for Test B, which leads to a much higher resistance against horizontal sliding. This, in turn, induces a higher ‘scratching force’ required to drive the scratch tip to maintain a constant speed of 100 mm/s, when compared to Test A. Comparing the scanned images of the scratch morphology of a talc-filled PP copolymer under the three test conditions, the scratch width remains constant along the scratch path for Tests B and C conditions; while there is a gradual increase in scratch width along the scratch path for Test A (Fig. 6). The

damage induced in the scratch groove undergoes a transition as the scratch progresses in Test A. Minimal surface features are observed in the beginning while severe damage with prominent ripple marks is present toward the end of the scratch. It is found that the ripple marks are actually curved fracture lines that appear periodically. The same phenomena are also observed in other model PP systems. It should be noted that the existing initial scratch width of 0.33–0.45 mm found in specimens is caused by the pre-existing small mass of the scratch tip and the load control unit, which measures at 5.47 N. Future improvement to the test device will be made to minimize such a pre-existing dead load prior to testing. It is apparent that the linear load increase test, i.e. Test A, is a more sensible test method in characterizing scratch damage resistance in polymers. Subsequent tests done on different material systems will demonstrate the usefulness and effectiveness of this test. The test has shown that copolymer systems suffer more damage than homopolymer systems (Fig. 7). This is to be expected as the Young’s modulus and yield strength of copolymer PP are lower than those of

Fig. 6. Talc-filled copolymers scratched under different conditions: (a) linear load increase and constant speed; (b) constant speed and load; (c) linear speed increase and constant load.

M. Wong et al. / Wear 256 (2004) 1214–1227

1221

50 45

Homopolymer Homopolymer + Talc

40

Normal Load (N)

35

Copolymer Copolymer + Talc

30 25 20 15

Fig. 7. Scratch widths and depths from linear load increase test condition on four different model PP systems.

10 5

the homopolymer PP [46]. Interestingly, the addition of talc does not cause significant changes in the size of scratch damage as quantified by the scratch depths and scratch widths. The test also found that all scratch depths and scratch widths show the same general trend between the copolymer and homopolymer PP, and between neat and talc-filled PP systems. Figs. 8 and 9 illustrate a typical complex surface feature and its sub-surface damage profile after a scratch is performed on a polymer. It is evident that complex surface damage mechanisms, such as plastic ironing, brittle fracture, fibril drawing, filler debonding, stick–slip, etc., can evolve, causing the scratch depths to vary within the same scratch

Fig. 8. SEM of talc-filled homopolymer scratched under Test A conditions.

Fig. 9. Variation of scratch depth along scratch groove in talc-filled copolymer.

0 0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

Scratch Width m m

Fig. 10. Variation of scratch width with normal load.

pass. Thus, it is recommended that scratch widths, as opposed to scratch depths, be considered as a more reliable and consistent measure to quantify scratch damage. Adopting the scratch widths as a measure of severity of surface damage can be quite practical since flatbed scanners can be used for the measurement. However, it should be highlighted that the scanned images generally do not produce a reliable measurement of SW2 when compared to TOM or SEM images. Nevertheless, scanned images allow us to have a reliable measurement of SW1 and thereby producing a quick assessment of the scratch damage. More sophisticated imaging tools can always be used for a more detailed study, if needed. To establish a relationship between scratch widths and normal loads, the linear load increase test as in Test A can be used. Fig. 10 shows the plot of the nominal normal loads applied by the spring load against the scratch widths measured by the scanner for various PP systems. For all PP systems, the scratch width follows a reasonable linear relationship with normal load, with the copolymer PP exhibiting larger scratch widths than homopolymer PP. Fig. 10 is a useful plot for revealing the load needed to form a given scratch width for a given polymer. Since it has been shown that scratch width correlates well with scratch visibility as well as the severity of surface damage if the surface damage characteristics stay the same [7], it is therefore possible to easily determine the critical load needed to cause such a surface damage based on the scratch widths data shown in Fig. 10. Most significantly, this plot will also allow material designers to quantitatively formulate a workable system to achieve specified surface damage resistance for a given polymeric system under a given testing condition.Furthermore, the Test A method permits a mar–scratch transition to be identified. This will help determine the critical normal load for such

1222

M. Wong et al. / Wear 256 (2004) 1214–1227

Fig. 11. Mar–scratch damage transition of (a) homopolymer and (b) talc-filled homopolymer in Test A observed under SEM.

a transition. For illustration, scratched specimens from Test A were scrutinized for the exact load and location along the scratch path where the scratch groove becomes highly visible. SEM images that show the mar–scratch transition for homopolymer and talc-filled homopolymer are given in Fig. 11. The distance, which was measured visually, and normal load for the mar–scratch transition are also listed in Table 4. The damage modes for homopolymer and talc-filled homopolymer are observed to be distinctly different. For homopolymer PP, the surface is smooth with no prominent features except for the faintly discernable edges before the line of transition (see Fig. 11a). After the line of transition, curved fracture lines appear and are closely spaced together, indicating an increase in the severity of surface damage. In addition, a change in damage mode from plastic ironing to plastic drawing and cracking is found as the load increases. For talc-filled homopolymer PP, before the line of transition, the surface damage is barely observable where a very shallow depression is formed due to the sliding of the scratch tip (Fig. 11b). The scratch groove is so shallow that it is more consistent with mar damage. After the line of transition, surface drawing and large-scale plastic deformation occur, creating the damage features that scatter light more significantly from the scratch groove. Note that there will Table 4 Mar–scratch transition values

Mar–scratch transition distance (cm) Mar–scratch transition load (N)

Homopolymer

Homopolymer + talc

2.65

1.90

18

15

be room for ambiguity as to the exact position where the transition begins in the micrographs. However, due to the contrasting features observed under SEM, the approximate region where transition begins can still be easily discerned. The resulting error is expected to be less than 100 ␮m, which is less than the precision, 0.1 mm, quoted in Table 4. This precision is sufficient for the quantification of macroscopic scratch behavior. The two SEM micrographs contrast the differences in surface damage mechanisms that occur during the mar–scratch transition. Homopolymer PP exhibits a more brittle damage mode, which is evidenced by the regular plastic drawing and crack lines; whereas talc-filled homopolymer shows more plastic drawing. This finding suggests that the addition of talc will alter the mode of scratch damage. This study suggests that the addition of talc into PP will lower the normal load required to cause mar–scratch transition by about 3 N (Table 4). It should be noted that the critical load for the stress-whitening transition, which does not necessarily coincide with mar–scratch transition described above, can also be determined using the linear load increase test. The findings have been demonstrated by using a commercial image analysis tool, VIEEW® . The details of the results and their significance will be discussed in a separate paper. The scratches performed under Test B and Test C do not exhibit such a transition. This is mainly due to the pre-existing severe initial indentation caused by the 30 N dead weight (Fig. 6). From the width measurements and the gray level analysis via a scanner, it can be shown that the scratch width does not vary significantly along the scratch grooves for Test B and Test C. In contrast, the linear load increase test (Test A) does not introduce such a severe

M. Wong et al. / Wear 256 (2004) 1214–1227

1223

Fig. 12. Scanned image showing scratch damage transition in a talc-filled homopolymer under Test A conditions: (a) entire scratch length; (b)–(d) enlarged details showing transition in scratch damage.

initial indentation because of its minimal starting load. The transition in damage can thus be observed. A scanned image shown in Fig. 12 clearly illustrates the transitions as the scratch groove progresses. Both the mar–scratch and the stress-whitening transitions can be observed. Another advantage of using the linear load increase test is the prevention of “chattering” of the scratch tip. In the work done by Kita et al. [39], it has been found that at constant speed and constant dead weight test, the scratch tip has a tendency of skipping or jumping during scratching, depending on the polymer type and the testing conditions applied. The same effect has also been observed in scratch tests done under similar conditions in our study. This effect probably comes about when the tip ploughs too deep. When the ploughing resistance becomes higher than scratching force, skipping occurs as the tip can only continue the forward motion by climbing up vertically. The linear load increase test reduces this effect because the scratch depth is shallower and the frictional force that entails will not overcome the scratching force. Fig. 13 shows the normal load of the scratch stylus as it traverses a neat PP specimen. Notice that the load plot is linear and well-behaved without any large spikes, proving that severe chattering did not occur during the linear load increase scratch test.

4.2. Repeatability The scratch tests performed above show that our custom-built scratching machine, if executed with care, can generate results that are highly repeatable. Fig. 14 demonstrates the consistency of the scratches produced by the linear load increase test performed on talc-filled copolymer system, the scratch features and transition points are shown to be very similar among the scratches. To show the repeatability of test results, standard deviation of the scratch widths and depths from Test A are calculated and plotted in Fig. 15. From Fig. 15, the percentage standard deviation is found to be lowest for SW1 and SW2, while it can go as high as 33% for SD1. This further suggests that the scratch widths give a more reliable measure of scratch damage. Apart from scanned images, the repeatability of test results in terms of scratch widths and depths has also been evaluated using a commercial image analysis system, VIEEW® and the findings are very similar to the analysis given above. Scratch width is also an important parameter that can be used in the determination of the contact area of the tip on the material. An accurate measurement of scratch width will allow the accurate determination of scratch hardness. Scratch hardness serves as a quantity in characterizing

60

Normal Load (N)

50 40 30 20 10 0 0

20

40

60

80

100

120

Position (mm)

Fig. 13. Wavy line represents actual normal load profile of neat PP under linear load increase test during scratch. Straight line represents ideal linear load increase.

Fig. 14. Scratch tests repeated under linear load increase conditions, demonstrating consistency of scratches obtained. Scratch direction is left to right.

1224

M. Wong et al. / Wear 256 (2004) 1214–1227

Fig. 15. Percentage standard deviation for scratch widths and depths in the linear load increase test.

scratch resistance. The derivation of scratch hardness using the current test method is discussed elsewhere [47]. 4.3. Numerical analysis findings To evaluate the effect of loading conditions and scratch speeds on the stress field of the computational model, three different load cases that are similar to the three tests performed in the experimental section (Tests A–C) were considered for the present FEA work. The three load cases1 are: (a) linearly increasing load (10–30 N) under constant scratch speed (10 m/s), (b) dead load (30 N) under constant scratch speed (10 m/s), and (c) linearly increasing scratch speed (0–20 m/s) under dead load (30 N). Using the same scratch damage quantification in Section 2.3, the scratch widths (SW1 and SW2) predicted by FEA at sections where the normal load and the scratch speed are the same for the three load cases are shown in Fig. 5. Good qualitative correlation can be noted. Similar to the earlier experimental findings, the scratch widths are found to be the smallest for load case (a), i.e. Test A, followed by load case (b), i.e. Test B, and load case (c), i.e. Test C. There is, however, a noted quantitative difference between both sets of results and may be due to the material model and scratch speeds adopted in the FEA, which may require further refinement. Based on the elasto-plastic FEA as discussed earlier, the maximum von Mises octahedral stress [48] is expected to exist underneath the tip of the indenter as it traverses across the scratch path. The variation of the maximum von Mises stress along the scratch path will indicate the amount of the plastic flow taken place during the scratch and this will indirectly help to evaluate the extent of damage along the 1 Note that due to the symmetry of the problem, the reduced computational model was used in FEA (see Fig. 3) that requires the normal loads specified to be reduced by half [41]. The computed FEA results remain valid for the normal loads as stated.

scratch path. The amount of plastic flow can also be gauged by the magnitude of the equivalent plastic strain, which is used to describe the change of the yield stress for isotropic hardening rule in plasticity. To study the plastic flow and thereby evaluating the extent of material damage across the scratch path, the maximum envelopes2 of the von Mises stress and equivalent plastic strain as computed by FEA for the three load cases have been presented in Figs. 16 and 17, respectively; in both figures, a plot of the ultimate von Mises stress and equivalent plastic strain has been included accordingly. From Fig. 16, the maximum envelope of the von Mises stress for load case (b) reaches the ultimate strength value the most rapidly, followed by load case (c) and (a). The reason for such a trend is that the load case (b) has the most severe loading conditions right from the beginning of the scratch process. While the inertia due to the increasing scratch speed of load case (c) will result in a more severe response than load case (a). Also, the increasing normal load in load case (a) expectedly produces an increasing variation of the von Mises stress before the ultimate strength value is reached. As prescribed in the model, the material has no additional load-carrying capacity beyond the ultimate strength value, which will explain why the maximum envelopes do not exceed the ultimate strength value. For the equivalent plastic strain, similar conclusion can be reached such that the maximum envelope reaches the ultimate strain value first for load case (b), (c) and finally (a). Beyond the ultimate strain value, the equivalent plastic strain continues to rise, indictating that more plastic flow is taking place despite that there is no further hardening of the material. From the two figures, it can be concluded that changing the normal load and/or scratch speed can influence the stress and strain state significantly and affect the surface damage of the material. Another worthwhile observation to note from Figs. 16 and 17 is that by using the increasing normal load condition in load case (a), there will be a critical region along the scratch path, beyond which fully plastic deformation will take place. Such a critical region can in turn be possibly related to the critical normal load that causes the onset of scratch damage and scratch visibility, which is one of the main objectives set out to achieve in this and our future study. Though as highlighted earlier that the material model adopted and the scratch speed may require refinement, the FEA results still qualitatively match reasonably well with the experimental data. From the above discussion, it is clear that the FEA can be used to gain a better understanding of the scratch damage under different load and scratch conditions and thereby providing the crucial link between the material science and mechanics. Since the von Mises yielding criterion only addresses the shear yielding of the material, 2 Due to resource constraint, only a limited number of time intervals were used for producing the plots and numerical fluctuation is hence observed.

M. Wong et al. / Wear 256 (2004) 1214–1227

Fig. 16. Maximum envelope of the von Mises stress for different load cases along the scratch path.

Fig. 17. Maximum envelope of the equivalent plastic strain for different load cases along the scratch path.

1225

1226

M. Wong et al. / Wear 256 (2004) 1214–1227

the current computational effort can be extended to consider crazing deformation during scratch. The current FEA can also be extended readily to include different scratch tip geometries such as conical and Berkovich to address different needs. It is proposed that by employing the current method, a standardized test can be utilized in scratch studies of polymers and establish a common basis for comparison and reduce the number of tests required to characterize scratch behavior of polymers.

5. Concluding remarks In this paper, a new scratch test method has been introduced to evaluate polymer scratch resistance. The proposed scratch test method is used to investigate four sets of model PP systems. By employing the linear load increase method, the chattering phenomena commonly seen in dead weight methods are eliminated, and the scratch damage resistance of different PP systems can be quantified. It is found that copolymer PP suffers greater scratch damage than homopolymer PP. Addition of talc does not change scratch widths and depths of both homopolymer and copolymer significantly. Good repeatability in all three test conditions is also found using our custom-built scratcher. The proposed linear load increase test enables the observation of mar–scratch and stress-whitening transitions during scratch. From the three-dimensional FEA, a better understanding of several influencing factors, such as the change in the loading and scratching speeds and stress distribution around the indenter, is gained. Through the correlation between the FEA and experimental results, it is indicative that the FEA is able to qualitatively capture the important characteristics of the scratch process, and hence warrants further utilization of FEA for fundamental understanding of scratch behavior of polymers.

Acknowledgements The authors would like to thank the financial support provided by the Texas A&M Scratch Behavior Consortium (Advanced Composites—Brian Coleman, BP Chemical— Kathryn Shuler, Luzenac—Richard Clark, Solvay Engineered Polymers—Edmund Lau, Visteon—Beth Wichterman and Rose Ryntz) in this research endeavor. The authors would like to acknowledge the generous loan of equipment from Atlas Materials Testing Technology—Fred Lee. The authors would also like to acknowledge the support from the State of Texas (ARP #32191-73130) and Defense Logistic Agency (SP0103-02-D-0003). The second author gratefully acknowledges the support by Applied Materials® Inc. through a graduate fellowship provided. The fourth author acknowledges the support of the research by the Oscar S. Wyatt Endowed Chair. Special thanks are also

due to the Society of Plastics Engineers—South Texas Section, for their generous donation of equipment for this research.

References [1] B.J. Briscoe, P.D. Evans, E. Pelillo, S.K. Sinha, Scratching maps for polymers, Wear 200 (1996) 137–147. [2] B.J. Briscoe, E. Pelillo, S.K. Sinha, Scratch hardness and deformation maps for polycarbonate and polyethylene, Polym. Eng. Sci. 36 (24) (1996) 2996–3005. [3] B.J. Briscoe, P.D. Evans, S.K. Biswas, S.K. Sinha, The hardness of poly(methylmethacrylate), Tribol. Int. 29 (2) (1996) 93–104. [4] C. Xiang, H.-J. Sue, J. Chu, B. Coleman, Scratch behavior and material property relationship in polymers, J. Polym. Sci. Polym. Phys. 39 (2001) 47–59. [5] R.S. Kody, D.C. Martin, Quantitative characterization of surface deformation in polymer composites using digital image analysis, Polym. Eng. Sci. 36 (2) (1996) 298–304. [6] J. Chu, L. Rumao, B. Coleman, Scratch and mar resistance of filled polypropylene materials, Polym. Eng. Sci. 38 (11) (1998) 1906– 1914. [7] J. Chu, C. Xiang, H.-J. Sue, R.D. Hollis, Scratch resistance of mineral-filled polypropylene materials, Polym. Eng. Sci. 40 (4) (2000) 944–955. [8] R.N. Rothon, Mineral fillers in thermoplastics: filler manufacture and characterization, Mineral fillers in thermoplastics; I. Adv. Polym. Sci. 139 (1999) 67–107. [9] J.A. Williams, Analytical models of scratch hardness, Tribol. Int. 29 (8) (1996) 675–694. [10] B.J. Briscoe, A. Delfino, E. Pelillo, Single-pass pendulum scratching of poly(styrene) and poly(methylmethacrylate), Wear 225–229(1) (1999) 319–328. [11] O. Vingsbo, S. Hogmark, Single-pass pendulum grooving-a technique for abrasive testing, Wear 100 (1984) 489–502. [12] Y.N. Liang, S.Z. Li, D.F. Li, S. Li, Some developments for single-pass pendulum scratching, Wear 199 (1996) 66–73. [13] ASTM G171-03. [14] G. Gauthier, R. Schirrer, Time and temperature dependence of the scratch properties of poly(methylmethacrylate) surfaces, J. Mater. Sci. 35 (2000) 2121–2130. [15] A. Krupiˆcka, M. Johansson, Use and interpretation of scratch tests on ductile polymer coatings, A. Hult, Prog. Org. Coat. 46 (2003) 32–48. [16] M. Kotaki, M. Wong, C. Xiang, H.-J. Sue, Scratch behavior of polypropylene-based blends, ANTEC 2 (2002) 1535–1539. [17] G.T. Lim, M. Wong, A. Moyse, J.N. Reddy, H.-J. Sue, Mechanical modeling and experimental observation of surface damage phenomena of polymers, in: Procedings of the SPE Int. Conf. Polyolefins, 2003, 577–584. [18] P.R.J. Guevin, State-of-the-art instruments to measure coating hardness, J. Coat. Technol. 67 (840) (1995) 61–65. [19] T. Triplett, Two-component: the magic’s in the mix, Ind. Paint Powder 72 (4) (1996) 34–37. [20] B.J. Briscoe, E. Pelillo, F. Ragazzi, S.K. Sinha, Scratch deformation of methanol plasticized poly(methylmethacrylate) surfaces, Polymer 39 (11) (1998) 2161–2168. [21] B.H. Stuart, B.J. Briscoe, Scratch hardness studies of poly(ether ether ketone), Polymer 37 (17) (1996) 3819–3824. [22] A. Chanda, D. Basu, A. Dasgupta, S. Chattopadhyay, A.K. Mukhopadhyay, A new parameter for measuring wear of materials, J. Mater. Sci. Lett. 16 (1997) 1647–1651. [23] J. Chu, Scratch resistance of PP composites, in: Proceedings of the 43rd International SAMPE Symposium, 1998, 1149–1157.

M. Wong et al. / Wear 256 (2004) 1214–1227 [24] J.R. Grasmeder, Scratch-resistant polypropylene compounds, in: Proceedings of the International Conference on Polypropylene Institute of Polymer, 1994, 98–108. [25] B. Lamy, Effect of brittleness index and sliding speed on the morphology of surface scratching in abrasive or erosive processes, Tribol. Int. 17 (1) (1984) 35–38. [26] F. Ramsteiner, T. Jaworek, M. Weber, S. Forster, Scratch resistance and embrittlement of coated polymers, Polym. Test. 22 (2003) 439– 451. [27] P.Z. Wang, I.M. Hutchings, S.J. Duncan, L. Jenkins, Quantitative characterization of scratch damage in polypropylene (TPO) for automotive interior applications, SAE Trans. 1999-01-0243 (1999) 134–150. [28] B.Y. Ni, A.L. Faou, Scratching behavior of polymer films using blunt spherical styli, J. Mater. Sci. 31 (1996) 3955–3963. [29] V. Jardret, H. Zahouani, J.L. Loubet, T.G. Mathia, Understanding and quantification of elastic and plastic deformation during a scratch test, Wear 218 (1998) 8–14. [30] S.R. Kim, J.S. Song, Y.J. Choi, J.H. Kim, Preparation of hard coatings on polycarbonate substrate by high frequency ion beam deposition using CH4 /H2 gases, Mater. Res. Soc. Symp. Proc. 504 (1997) 265– 270. [31] L. Lin, G.S. Blackman, R.R. Matheson, Micro-mechanical characterization of mar behavior of automotive topcoats: micro- and nanowear of polymeric materials, American Chemical Society, Polymer Preprints, Div. Polym. Chem. 39 (2) (1998) 1224–1225. [32] A.C.-M. Yang, T.W. Wu, Abrasive wear and craze breakdown in polystyrene, J. Mater. Sci. 28 (1993) 955–962. [33] P. Leroux, A. Raveh, J.E. Klemberg-Sapieha, L. Martinu, Mechanical properties of plasma deposited functional coatings determined by microscratch measurements, in: Proceedings of the Annual Technical Conference, Society of Vacuum Coaters,1993, 472–477. [34] B.Y. Du, M.R. Vanlandingham, Q.L. Zhang, T.B. He, Direct measurement of plowing friction and wear of a polymer thin film using the atomic force microscope, J. Mater. Res. 16 (5) (2001) 1487–1492.

1227

[35] E. Hamada, R. Kaneko, Micro-tribological evaluations of a polymer surface by atomic force microscopes, Ultramicroscopy (Part A), 42–44 (1992) 184–190. [36] Y.C. Han, S. Schmitt, K. Friedrich, Nanoscale indentation and scratch of short carbon fiber reinforced PEEK/PTFE composite blend by atomic force microscope lithography, Appl. Compos. Mater. 6 (1) (1999) 1–18. [37] J.L. Bucaille, E. Felder, G. Hochstetter, Mechanical analysis of the scratch test on elastic and perfectly plastic materials with the three-dimensional finite element modeling, Wear 249 (2001) 422– 432. [38] J.H. Lee, G.H. Xu, H. Liang, Experimental and numerical analysis of friction and wear behavior of polycarbonate, Wear 251 (2001) 1541–1556. [39] H. Kita, M. Ishiki, M. Maki, T. Kitamura, T. Kuriyama, Scratch behaviors of moldings, ANTEC 3 (2003) 2992–2996. [40] ASTM D 618-99. [41] J.N. Reddy, An Introduction to the Finite Element Method, 2nd ed., McGraw-Hill, New York, 1993. [42] ABAQUS Inc., ABAQUS/Explicit, Version 6.3, 1–3, 2002. [43] M. Wong, G.T. Lim, A. Moyse, M. Kotaki, J.N. Reddy, H.-J. Sue, Surface damage phenomena of polyolefin materials, TPOs in Automotive, June 2003. [44] E.M. Arruda, S. Ahzi, Y. Li, A. Ganesan, Rate dependent deformation and semi-crystalline polypropylene near room temperature, J. Eng. Mater.—Trans. ASME 119 (1997) 216–222. [45] G.T. Lim, J.N. Reddy, H.-J. Sue, Computational modeling of scratch behaviour of polymers, Surf. Coat. Technol., in review. [46] A. Dasari, J. Rohrmann, R.D.K. Misra, Micro- and nanoscale evaluation of scratch damage in poly(propylene)s, Macromol. Mater. Eng. 287 (2002) 889–903. [47] M. Wong, A. Moyse, F. Lee and H.-J. Sue, Study of surface damage of polypropylene under progressive loading, J. Mater. Sci., in review. [48] A.S. Khan, S. Huang, Continuum Theory of Plasticity, Wiley, New York, 1995.