Anodic coatings used on aluminum and magnesium alloy light metals can provide significant protection from corrosion or wear; thiner for corrosion and thicker for wear. However, anodizing aluminum alloys causes fatigue strength loss. This is also observed in magnesium light metals. This post is primarily concerned with the the fatigue reduction in the aluminum-copper alloys, i.e., 2000 series. The focus is on AA2219 and AA2024 Prudent and safe design must consider the influence of anodizing on fatigue based on the product form used in a product.
In the AA2219 alloy, depending on the cross section of the wrought product form, the fatigue debit can vary widely from less than 5 % to greater than 45 % (Figure 2, A, B, C). Here, fatigue debit is the percent difference at a given life from the unanodized to the anodized state. This can have marked product integrity ramifications if not addressed with proper characterization, materials selection, and design. While the extent of the debit is not the same for all aluminium-copper alloys, similar behavior is observed in AA2024 and, by inference to the microstructure and literature, in AA2014. The results shown here are from rotary bending testing at the implied stress ratio of R = -1, i.e., fully reversed bending. Elsewhere on this web site, we discuss the Ashby method of materials selection. Data such as these can readily be incorporated in to that graphical means of material selection.
My coauthors and I spent many years investigating fatigue strength loss from anodization in a wide range of light metal alloys. AA6061-T6, when hard anodized for wear, can loose 60 % of its fatigue strength. This was attributed to stress risers from coating cracks. In the same paper it was shown that cast alloy C355-T6 showed no significant fatigue strength loss. This was attributed to the intrinsic presence of silicides in the microstructure which were of the same scale as anodize cracks. Magnesium alloy WE43A-T6  lost about 10 % of its fatigue strength after application of Magoxide anodize coating. More recent work by others  has shown similar losses in magnesium alloy AM60 with an unspecified anodize process. These authors investigated coating thickness and found fatigue strength loss increased with thickness.
What makes AA2219 so interesting, in comparison to these other alloys, is the direct link between the Al2Cu constituent particle size and the resulting fatigue strength. Both the cooling rate of the ingot and the extent of cold work during ingot breakdown can influence this particle size. Figure 1 shows a cross section of anodized 150 mm plate. The anodic coating is at the bottom between the arrows. Note how the pits are anodized. Thus, the content of the pit prior to anodization was very quickly dissolved away at the start of anodization. Some of the remaining unconnected Al2Cu is evident as the bright phase in the back scatter SEM image of Figure 1. Chromic acid anodizing was also observed to cause even deeper pitting in AA2219 and it makes sense that this might explain the even greater fatigue strength loss with chromic acid anodizing.
The constituent Al2Cu is an artifact of casting the ingot. Quantitative metallography of 150 mm plate vs. 25 mm bar revealed the same volume fraction of the Al2Cu constituent. However, the extent of ingot breakdown during wrought processing, i.e., mechanical working, or cooling rate was much greater for the small cross section material, which resulted in significantly reduced particle size (see figure 6 in the cited paper). The smaller particle size gave rise to smaller pits and concomitantly smaller fatigue reducing flaws. This explained the difference in fatigue debit of 45 % in thick cross section material vs. <5 % in thin cross section material.
Given what we found for AA2219, we decided to investigate AA2024, an alloy leaner in copper ( 4.4 % vs 6.3 %). As there is less Cu to form the Al2Cu constituent, one would expect less influence of anodization on fatigue, and this was indeed the case as shown in Figure 3. Large diameter bar had 30 % fatigue strength loss after sulfuric acid anodizing vs. 13 % for the small diameter bar.
AA2014 was not investigated in my work. However, there is some data in the literature.  Again, the copper content is lower than in AA2219 (4.5 % vs. 6.3 %). Shiozawa showed some interesting results where at a stress ratio of R = -1, there was perhaps an improvement in fatigue life, while at R = 0.01, the debit was about 26 %. This material was 25 mm dia. extruded rod.
Finally, there is another interesting but not unexplained phenomenon. Note how in many cases shown in fatigue 2, the fatigue strength loss is greatest in the vicinity of one million cycles. We speculated that under high cyclic stresses, low cycle fatigue nucleated the strength controlling flaws independent of the anodic coating, where as at high cycles, the influence of the stress riser was less significant as most of the life was spent actually nucleating the strength controlling flaw. This same situation was observed for the case of WE43A-T6 
In conclusion, anodizing aluminum causes fatigue strength loss and its extent is alloy dependent. In Aluminum-Copper, it is related to the size of casting artifacts, chiefly Al2Cu, during ingot casting and wrought processing. Larger casting artifacts give rise to greater fatigue strength losses for a given alloy in the aluminum=copper system. Design data must account for the real microstructure of the actual aluminum product form and anodizing process to be valid.
if you have a fatigue sensitive product made from an aluminum or magnesium alloy, contact us to discuss how we can help you develop a test and data analysis plan or to perform design trade studies to prevent failures.
 R. G. Rateick, R. J. Griffith, D. A. Hall, and K. A. Thompson, “Relationship of microstructure to fatigue strength loss in anodised aluminium-copper alloys,” Materials Science & Technology, vol. 21, no. 10, pp. 1227–1235, 2005. Copyright © 2005 Institute of Materials, Minerals and Mining, figures reprinted by permission of Taylor & Francis Ltd, http://www.tandfonline.com on behalf of Institute of Materials, Minerals and Mining.