Advances in NanoScratch Testing of Automotive Clearcoats

Abstract

The primary functions of automotive clearcoats are to protect the underlying layers and maintain a glossy showroom appearance for as long as possible. One of the most common sources of damage is from polymeric carwash brushes and small embedded dirt and grit particles that create small scratches in the clearcoat surface. Complex multiasperity contact, unknown geometries, and stress states make it difficult to quantify this damage. Since the mid-1990s, researchers have developed sophisticated single indenter nano- and microscratch techniques to quantify the surface mechanical properties and to understand the connection to scratch performance. Indenter size and shape were selected so that the scratches produced would be similar in size and appearance to the damage produced under real-world conditions. In this chapter, four automotive clearcoats with different crosslink densities, including a 2K urethane, a 1K melamine, and two experimental clearcoats, are evaluated and compared. New statistical methods for creating and analyzing scratch damage based on NanoScratch testing are discussed. Recovery of scratches in both short timescale (minutes) and long timescale (days) is analyzed by the NanoScratch testing and Atomic Force Microscope (AFM) analysis. The connection between the scratch morphology and the visual appearance is explored using dark field imaging as an objective surrogate for appearance. Two components to scratch resistance, damage resistance and scratch visibility, are analyzed. The damage resistance can be characterized by residual depth and fracture threshold, and the scratch visibility can be characterized by the contrast and size of a scratch image. Crosslink density appears to affect the residual depth and scratch visibility. A coating with good damage resistance does not automatically lead to low scratch visibility. The methods presented offer new ways to further understand scratch performance of coatings.

Appendix MATLAB ® and Kornucopia ® ML™ Analysis of Scratch Data

The raw data from the NanoScratch test (NST) is exported and processed so that powerful statistical tools such as ANOVA (analysis of variation) and two-sample t-tests can be applied and even subtle variations between samples quantified. The amount of raw data to be processed from the NST is extremely large. In addition, the data need to be sliced into subsections for each load step that is applied to a step scratch. Performing this type of data processing using manual cut and paste operations in Excel or similar tools is possible, but it quickly becomes an inefficient process, especially when you need to process tens to hundreds of scratch profiles. The authors applied a few specialized scripts created using Kornucopia^® ML (Bodie Technology, Inc.) and MATLAB^® (The Mathworks, Inc.) to address the raw data processing, slicing, filtering, decimation, statistics, and plotting used for much of this work.

Figure 15.11 is one data set from a six-step scratch on a 2K urethane clearcoat. The load transitions are located, and roughly 50 μm at the beginning and end of each 0.5 mm step is removed. This is the region where finite probe size effects, creep, and feedback anomalies most affect the data. The raw data have over 5600 data points in each load step, and after slicing the data and removing the transition regions, there are still over 4500 data points at each load indicated by the different color regions in Fig. 15.11.

Fig. 15.11

Raw data exported from the NanoScratch tester and imported into Matlab. Kornucopia routines are used to slice the data as indicated by the colored steps in the figure. The top panel contains the normal and tangential force data and the lower panel contains the surface profile and residual depth. The data are for the step scratch for the 2K urethane coating aged for 1 day after spraying

The penetration and residual depth measurements are relative to the initial profile of the surface; however, there is often deformation associated with the pre-scan load. In the NanoScratch test method, the first load step is 0.2 mN (the pre-scan or profiling load). The displacement is corrected by subtracting the mean value in this first load step from each of the other load steps. Figure 15.12 shows the residual depth for a region around the 5 mN load step with the processing steps indicated.

Fig. 15.12

Residual depth data for 5 mN load step for the 2K urethane coating illustrating the various steps in the treatment of the data

The instrument is capable of collecting data at 100 points per second or higher; however, due to the finite size of the probe, the data is highly correlated. The correlation arises because the spacing between data points is 0.1 μm, and the indenter has a radius of 2 μm. Most statistical analyses require independent measurements; therefore the exported data must be downsampled or decimated prior to the ANOVA calculations (Fig. 15.8). The correlation is significant and is most easily detected in an autocorrelation calculation such as in Fig. 15.13a. The decimated data have very little correlation as shown in Fig. 15.13b.

Fig. 15.13

Autocorrelation function performed on residual depth data from 5 mN load step on 2K urethane clearcoat in Fig. 15.12. The red lines represent the 5% confidence limits

In the final step, any slope due to creep during the scratch is subtracted from each of the steps. The slope in this region represents a short timescale relaxation of the surface during the scratch. No further analysis was done on the short timescale creep measurements.