Study of Corrosion and Corrosion Inhibition of. Chromate and Chromate-Free Primers

Transcription

1 Study of Corrosion and Corrosion Inhibition of Chromate and Chromate-Free Primers Thesis Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in the Graduate School of The Ohio State University By Longfei Li, B.S. Graduate Program in Material Science and Engineering The Ohio State University 2012 Master s Examination Committee: Dr. Rudolph G. Buchheit, Adviser Dr. Gerald S. Frankel

2

3 ABSTRACT Chromate and chromate-free primers were characterized in this study. For chromate primer examined, PPG EWEDA 144A (Pittsburgh Plate Glass Company), SrCrO 4 was the active inhibitor. The three chromate-free primers examined were Deft 02GN084, Hentzen TEP and Sicopoxy In the Deft product, the active inhibitor was PrO/ Pr(OH) 3. CaSiO 3 was the active inhibitor for Hentzen TEP and BaSO 4 for Sicopoxy How each pigment reacted within its coating environment during exposure was an object of study and primers coatings were characterized by several techniques. Focused ion beam (FIB) milling was used to prepare cross sections for energy dispersive x-ray analysis (EDX) to identify pigments and inhibitors derived from them. A Matlab TM program was developed to analyze the resulting EDX maps to characterize the dispersion of elements in inhibitor ions produced by pigment dissolution during exposure. Pigment characteristics and pigment dissolution varied widely from primer to primer. Chromate pigments were present in high volume fractions in chromated primers. As a result pigment particles were in close proximity to one other and touching in 2D sections. In the chromate-free primers, inhibitor pigment particles were present in much lower volume fractions and were far away from each other. To quantify the protectiveness of chromate and chromate-free coatings systems in standard exposure testing on mixed metal structures, optical profilometry was used on ii

4 painted and primed 7075-T6 coupons outfitted with AISI 316 or Ti-6Al-4V bolts. The presence of the bolts provided galvanic stimulation to corrosion attack occurring under the coating during ASTM B117 exposure. In these experiments, coatings were stripped and corrosion damage was quantified by optical profilometry, which allowed measurement of the total corroded volume that accumulated around the bolts that were placed in the coated samples. The volume was converted to corrosion rate. In general, 316 CRES fasteners caused more serious corrosion to AA2024-T3 panel than Ti-6Al-4V. And samples with chromium-containing pretreatments, Surtec 650 or Alodine 1200S, had smaller corrosion volume that those with adhesion promoter, Prekote or BoeGel. In the aspect of primers, ANAC P2100P003 and Deft 44-GN-098 gave the substrate best protection, while PPG EWDE 144A and Henzten KEP had the worst performance on corrosion prevention among the six types of primers examined. In addition, the application of topcoat did not guarantee a better protection on the scribed AA2024-T3 panel. And the rate of pigment dissolution was affected by the application of topcoat. iii

5 DEDICATION For all my loving family who support me back at my home country, especially my parents and my sister. iv

6 ACKNOWLEDGMENTS I would like to thank my advisor, Dr. Rudy Buchheit, who accepted me as graduate research assistant one day before my undergraduate commencement. His guidance and encouragement made me confident and energetic, especially when I was struggling. I also would like to acknowledge those who have sponsored my work. This project is funded by Strategic Environmental Research Development Program (SERDP) under contract W912HQ-08-C Finally, I wish to appreciate the past and present members of the FCC group who assisted me in finishing this work (Anusha Chilukury, Zhicao Feng, Dr. Yang Guo, Dr. Belinda Hurley, Guangyi Liu and Meng Tong). In addition, I am so grateful for Zhao Wang, Wenjia Luo and Yunlu Pan helping develop the Matlab TM program. v

7 VITA March 8, Born Daqing, China Sep 2005 Dec Attended, Beihang University, China Jan 2008 May B.S. Material Science and Technology, Iowa State University Aug 2012 present...graduate Research Assistant, The Ohio State University FIELDS OF STUDY Major Field:... Material Science and Engineering vi

8 Table of Contents ABSTRACT... ii DEDICATION... iv ACKNOWLEDGMENTS...v VITA... vi LIST OF TABLES...x LIST OF FIGURES... xiii CHAPTER 1: REVIEW OF LITERATURE Introduction Techniques for Characterization of Primers Introduction Techniques for Characterization of Pigment Morphology and Distribution within Primers Techniques for Characterization of Pigment Composition Techniques for Characterization of Pigment Reactions Non-electrochemical Methods of Examining Effectiveness of Inhibitors Introduction Common Non-electrochemical Methods of Evaluating Inhibitor Efficacy...6 vii

9 Other Non-electrochemical Methods of Evaluating Inhibitor Efficacy Weakness of Non-electrochemical Methods...12 CHAPTER 2: CHARACTERIZATION OF PIGMENT DISSOLUTION in coatings Introduction Experimental Procedures Sample Preparation Non-electrochemical Characterization Results and Discussion Cross Section by FIB Milling EDX Mapping Image Analysis by Matlab TM Program Conclusion...39 CHAPTER 3: EVALUATION OF CORROSION PREVENTION Introduction Experimental Procedure Sample Preparation Cleaning Volume Measurements of Material Lost Measurements of Corrosion Area Calculation of Corrosion Rate Results and Discussion Types of Corrosion Corrosion of Samples Exposed in ASTM B117 and ASTM G viii

10 3.3. Conclusion Types of Corrosion Effect of Fastener Material Effect of Surface Treatment Effect of Primers Effect of Topcoat Conclusion Future Work REFERENCES APPENDIX A: Average Elemental X-ray Intensity within Fixed Area of Primers APPENDIX B: Corrosion Volume, Area and Rate of Coated AA2024-T3 Panels Exposed in ASTM B117 and G ix

11 LIST OF TABLES Table Page Table 1. 1: Results of adhesion test [36] Table 2. 1: Phases within tested primers Table 2. 2: Average percentage residual of calcium X-ray intensity within fixed area of topcoated Hentzen TEP with, (a) unexposed, (b) 30-day exposure, (c) 90-day exposure Table 2. 3: Average percentage residual of calcium X-ray intensity within fixed area of Deft 02G084, (a) unexposed, (b) 30-day exposure, (c) 90-day exposure Table 2. 4: Average percentage residual of praseodymium X-ray intensity within fixed area of untopcoated Deft 02N084, (a) unexposed, (b) 30-day exposure, (c) 90-day exposure Table 2. 5: Average percentage residual of barium X-ray intensity within fixed area of topcoated Sicopoxy , (a) unexposed, (b) 30-day exposure, (c) 90-day exposure 56 Table 3. 1: Collection of corrosion type Table 3. 2: Number of counts in each rank range by fastener in ASTM B117 for 500 hrs Table 3. 3: Number of counts in each rank range by surface treatment in ASTM B117 for 500 hrs Table 3. 4: Number of counts in each rank range by primer in ASTM B117 for 500 hrs x

12 Table 3. 5: Number of counts in each rank range by existence of topcoat in ASTM B117 for 500 hrs Table 3. 6: Number of counts in each rank range by fastener type in ASTM G85 for 360 hrs Table 3. 7: Number of counts in each rank range by surface treatment in ASTM G85 for 360 hrs Table 3. 8: Number of counts in each rank range by primer in ASTM G85 for 360 hrs Table 3. 9: Number of counts in each rank range by existence of topcoat in ASTM G85 for 360 hrs Table A. 1: Calcium X-ray intensity within fixed area of top-coated Hentzen TEP, (a) unexposed, (b) 30-day exposure, (c) 90-day exposure Table A. 2: Calcium X-ray intensity within fixed area of untopcoated Deft 02-GN-084, (a) unexposed, (b) 30-day exposure, (c) 90-day exposure Table A. 3: Praseodymium X-ray intensity within fixed area of untopcoated Deft 02-GN- 084, (a) unexposed, (b) 30-day exposure, (c) 90-day exposure Table A. 4: Barium X-ray intensity within fixed area of topcoated Sicopoxy , (a) unexposed, (b) 30-day exposure, (c) 90-day exposure Table B. 1: Corrosion volume of samples in ASTM B117 for 500 hrs Table B. 2: Corrosion area of samples in ASTM B117 for 500 hrs Table B. 3: Corrosion rate of samples in ASTM B117 for 500 hrs Table B. 4: Corrosion volume of samples in ASTM G85 for 360 hrs Table B. 5: Corrosion area of samples in ASTM G85 for 360 hrs xi

13 Table B. 6: Corrosion rate of samples in ASTM G85 for 360 hrs xii

14 LIST OF FIGURES Figure Page Figure 1. 1: SEM cross-sectional observation of Mg-rich pigmented primer coating on AA 2024-T3: unexposed (a) and after 5090 hrs (b) [11] Figure 1. 2: Pigment distribution within chromate primer by Raman spectroscopy [13].14 Figure 1. 3: Phase identification within chromate primer by Raman spectroscopy [13]. 15 Figure 1. 4: Spectra pf mixture of ZnFe 2 O 4 and polyaniline (PANI) within coating by FTIR Figure 1. 5: XPS spectrum on bare aluminum sample (a) and silicate-passivated aluminum coupon [17] Figure 1. 6: Optical and Raman images of a section close to a scribe slot after being exposed to neutral salt spray for 5 days (b-d) and before salt spray (e). The maps are for TiO 2 (anatase) at 638 cm -1 (b), BaSO 4 at 988 cm -1 (c) and SrCrO 4 at 866 cm -1 (d-e) [18].18 Figure 1. 7: Cr distribution along a 0.6 mm slot for different exposure times to neutral salt fog. (orange means depletion and blue represents chromate particles) [18] Figure 1. 8: Morphology of primer coating after 53 hrs cathodic polarization. The scribe was made at the center of the sample [25] xiii

15 Figure 1. 9: Total area of blisters (a) and maximum distance of blister from scratch (b) versus time during cathodic polarization at -1.1V SSE at 5% NaCl solution [25] Figure 1. 10: RBS profile shift of Xe sputtered on titanium versus time immersed into 1M H 2 SO 4 [39] Figure 1. 11: Structure of Bragg grating fiber [43] Figure 2. 1: Position of samples when trenched by FIB Figure 2. 2: Cross-sectional view of calcium intensity profile of primer (Hentzen TEP) after 30-day exposure. Yellow circles indicate increasingly larger areas with radii of 5, 10 and 15 pixels Figure 2. 3: SEM and EDX mapping images of unexposed PPG EDWE 144A with conversion coating and phase identification Figure 2. 4: SEM and EDX mapping images of unexposed Deft 02GN084 and phase identification Figure 2. 5: SEM and EDX mapping images of unexposed Hentzen TEP and phase identification Figure 2. 6: SEM and EDX mapping images of unexposed Sicopoxy and phase identification Figure 2. 7: Cr elemental X-ray intensity map on PPG EWDE 144A by Matlab TM program Figure 2. 8: EDX maps of primer cross sections by FIB (PPG EDWE 144A) Figure 2. 9: EDX maps of primer cross sections by FIB (Deft 02GN084) Figure 2. 10: EDX maps of primer cross sections by FIB (Hentzen TEP) Figure 2. 11: EDX maps of primer cross sections by FIB (Sicopoxy ) xiv

16 Figure 3. 1: Positions of mounted fasteners Figure 3. 2: Coating systems in detail Figure 3. 3: Morphology of samples with corrosion induced in a limited area Figure 3. 4: Morphology of samples with uniform corrosion induced in a wide area Figure 3. 5: Sample before cleaning (a) and after cleaning (b) (The red squares were drawn later for OP measurements) Figure 3. 6: Weight loss measurements during cleaning Figure 3. 7: An example of a situation when the surface close to a hole and scratches are higher than that of surfaces far away Figure 3. 8: Example of corrosion area calculation by Clemex Vision Figure 3. 9: Collection of corrosion volume in ASTM B117 and G Figure 3. 10: Collection of corrosion rate in ASTM B117 and G Figure 3. 11: Collection of corrosion area in ASTM B117 and G Figure 3. 12: Corrosion volume listed by fastener type in ASTM B117 for 500 hrs Figure 3. 13: Corrosion rate listed by fastener type in ASTM B117 for 500 hrs Figure 3. 14: Corrosion area listed by fastener type in ASTM B117 for 500 hrs Figure 3. 15: Corrosion volume listed by fastener type in ASTM G85 for 360 hrs Figure 3. 16: Corrosion rate listed by fastener type in ASTM G85 for 360 hrs Figure 3. 17: Corrosion area listed by fastener type in ASTM G85 for 360 hrs Figure 3. 18: Corrosion volume listed by surface treatment in ASTM B117 for 500 hrs Figure 3. 19: Corrosion rate listed by surface treatment in ASTM B117 for 500 hrs Figure 3. 20: Corrosion area listed by surface treatment in ASTM B117 for 500 hrs xv

17 Figure 3. 21: Corrosion volume listed by surface treatment in ASTM G85 for 360 hrs. 97 Figure 3. 22: Corrosion rate listed by surface treatment in ASTM G85 for 360 hrs Figure 3. 23: Corrosion area listed by surface treatment in ASTM G85 for 360 hrs Figure 3. 24: Corrosion volume listed by primer in ASTM B117 for 500 hrs Figure 3. 25: Corrosion rate listed by primer in ASTM B117 for 500 hrs Figure 3. 26: Corrosion area listed by primer in ASTM B117 for 500 hrs Figure 3. 27: Corrosion volume listed by primer in ASTM G85 for 360 hrs Figure 3. 28: Corrosion rate listed by primer in ASTM G85 for 360 hrs Figure 3. 29: Corrosion area listed by primer in ASTM G85 for 360 hrs Figure 3. 30: Corrosion volume listed by topcoat in ASTM B117 for 500 hrs Figure 3. 31: Corrosion rate listed by topcoat in ASTM B117 for 500 hrs Figure 3. 32: Corrosion area listed by topcoat in ASTM B117 for 500 hrs Figure 3. 33: Corrosion volume listed by topcoat in ASTM G85 for 360 hrs Figure 3. 34: Corrosion rate listed by topcoat in ASTM G85 for 360 hrs Figure 3. 35: Corrosion area listed by topcoat in ASTM G85 for 360 hrs xvi

18 CHAPTER 1: REVIEW OF LITERATURE 1.1 Introduction Protective coatings are used widely to protection metallic engineering structures of all types [1]. In many cases, corrosion protection derives from dissolution of sparingly soluble pigment particles that release ionic corrosion inhibitors into moisture that has permeated the coating system. More recently pigments based on ion exchanging materials have been explored to deliver novel inhibitors such as vanadates [2]. A range of chromate-free coating systems including chromate-free primers has been developed to reduce the usage of chromate in coating manufacture and application. Chromate is objectionable due to its carcinogenicity [3]. Unfortunately, chromate is an extraordinarily powerful corrosion inhibitor, and chromate-free primer formulations do not result in coatings that approach to protectiveness or chromated coating systems. To fully account for the performance differentia it is necessary to characterize differences in performances among each class of primers, and in turn develop and deploy the best possible methods to assess primer and coating structure, properties and performance. This chapter presents several techniques that are used for characterizing chemical composition and morphology of pigments within primers. Meanwhile, 1

19 some other techniques that are capable of resolving pigment reactions due to exposure are also explained. Non-electrochemical methods for characterization are also discussed. The strengths and weaknesses of these various approaches are presented and discussed. 1.2 Techniques for Characterization of Primers Introduction A range of inhibiting pigments has been explored for use in protective coatings [4]. It is necessary to know the physical-chemical properties of these pigments, such as composition, shape, size range, and volume fraction [5-8]. The distribution of pigments through the thickness of the coating may also affect protection [9]. Exposure itself may change pigments [10]. Characterizing the basic information and changes in these aspects of pigment particles may aid in understanding coating protection. This section describes approaches that have been used to characterize the pigments particles within primers Techniques for Characterization of Pigment Morphology and Distribution within Primers Scanning electron microscopy (SEM) has been used to qualitatively assess the morphology change of pigments due to exposure. From Figure 1.1 (a) and (b )[11], illustrate obvious changes of Mg-containing pigments due to extended 2

20 cabinet exposure testing. In Figure 1.1 (a), the pigment particles are intact while small portion of each pigment has disappeared due to dissolution after exposure (Figure 1.1 (b)). These two images indicate reaction of the pigment within Mg-rich primer[11]. These images show that SEM can resolve pigment particles in coatings, reveal their shape, size and volume fraction. Raman spectroscopy has been used for pigment characterization. It is based on illumination of sample with a laser source and measurement of scattered radiation by an appropriate spectrometer [12]. It is capable of measuring the depth profile of pigments in the primers. Figure 1.2 shows the distribution of primary pigments within chromate primers which was examined by this method [13] Techniques for Characterization of Pigment Composition Raman spectroscopy can also help to identify specific phases within the primers. In Figure 1.3[13], primary pigments within chromate-inhibited primer were identified by Raman spectroscopy over a 100 X 25 μm area of a coated surface. In a similar fashion Fourier transform infrared spectroscopy (FTIR) can aid identification of pigment components as well. It based on the absorption of infrared radiation by the sample and transmission of infrared radiation through the sample [14]. Light scattering and absorption leads to characteristic spectra that can be diagnosed to identify the chemical compounds present in the sample. In Scholes work [13], FTIR was used to identify the mixture of ZnFe 2 O 4 ferrite and polyaniline (PANI). The spectra appeared in the middle was between that of ZnFe 2 O 4 ferrite and 3

21 polyaniline (PANI). Its appearance demonstrated the successful blending of ZnFe 2 O 4 and polyaniline in the coating. X-ray photoelectron spectroscopy (XPS) is a technique that yields information on elemental composition, and electronic state of the elements in an analysis volume [15]. Its working mechanism is to measure the energy of electrons escaping from sample surface irradiated by a low-power x-ray source. It is well suited for analysis of the near-surface region of samples as the electron escape depths are extremely shallow. Calcium silicate has been tested as an inhibitor of aluminum and steels. As an example, a passivation layer having thickness of about 10 nm was found on the AA 3003 surface and identified by XPS when calcium silicate was added into the system [16]. From XPS spectra in Figure 1.5 [17], Al 2 O SiO 4 was confirmed as the major part within the passivation layer Techniques for Characterization of Pigment Reactions Raman spectroscopy can also be used to assess the change of pigment chemistry due to exposure. Figure 1.6 [18] shows Raman intensity maps that characterize the movement of multiple species along a scratch. The maps show the existence of TiO 2 and BaSO 4. They also show the movement of SrCrO 4 before and after 5-day neutral salt spray exposure. Another technique called proton induced X-ray emission (PIXE) has been used to measure the dissolution of inhibitors from a cut edge slot [18]. In Figure 1.7 4

22 [18], the leaching out of chromium was measured after different lengths of exposure to neutral salt spray exposure. The results indicated that the depletion zone increased in extent as exposure time increased. Electron microprobe analysis (EMPA) is also used for inhibitor depletion measurements. Its working principle is similar to SEM/EDX by scanning the sample with electron beam and analyzing the characteristic X-rays produced [18]. 1.3 Non-electrochemical Methods of Examining Effectiveness of Inhibitors Introduction For the purposes of this discussion, inhibitors are the ions dissolved or released from the embedded pigment particles in the primer coating that dissolve when moisture is taken up [19]. Numerous inhibitors have been explored for corrosion prevention, and their efficacy varies. Therefore, a range of methods for evaluating inhibitors is needed. However, many evaluations depend on personal judgments which are subjective. As an example, ASTM D1654 is a standard for evaluating painted or coated specimens subjected to corrosive environments. It involves determination of rust creepage by recording the largest and smallest creepage zone from the scribe visually [20]. Similar amounts of creepage may be evaluated differently by different examiners. Thus, quantitative and objective assessments on inhibitor efficacy are needed in order to eliminate subjectivity in visual assessments. Within this section, the focus is on existing methods of how to characterize inhibitor efficacy objectively. Non-electrochemical methods are 5

23 discussed. With the help of objective assessments, the corrosion rates can be estimated and the relative efficacy of inhibitors more appropriately assessed Common Non-electrochemical Methods of Evaluating Inhibitor Efficacy Introduction Common methods of measurements for evaluating inhibitor efficacy nonelectrochemically are addressed here [21]. First among these is weight loss. The second is to measure the concentration of corroded ions within contacting solutions. Third is to collect the gas evolved from the supporting cathodic reaction in the corrosion cell process. Finally there is ph change that can reveal locations of anodes and cathodes that comprise the corrosion cell. Most experiments in this regard use ph indicator to detect and control corrosion, such as ph responsive or fluorescing compounds [22, 23]. Due to the limitation of the ph method, first three of those four methods will be the focus of discussion. In addition, there are some methods that have been developed, but not used extensively. They are also explained so as to give a comprehensive portrayal of corrosion assessments Weight Loss Measurements The method of weight loss is commonly used for determining corrosion damage and average corrosion rate. By knowing the weight loss, the area and the length of exposure, corrosion rate can be calculated easily [24]. The equation below shows how corrosion rate is calculated from weight loss data. In this expression, Δm 6

24 is weight loss, S is the area and t is the length of exposure time, (2) The unit of corrosion rate depends on units used with the variables in the formula. Common units are mg cm -2 h -1 [25]. Weight loss can be converted into other expressions, such as volume lost per unit time or penetration depth per unit time [26]. Weight loss measurements can be used to estimate inhibitor efficiency. The inhibitor efficiency, E (%), is calculated based on following equation, where W and W o are the corrosion rate having inhibitors and no inhibitors [27]. E (%) = ( ) * 100 (3) Concentration Measurements of Corroded Material in Solution The corrosion rate also can be estimated from the concentration of ions produced by dissolution. This method is limited to situations where the corrosion product is soluble or can be made soluble by known additions of acid or base. Colorimetry and Inductively coupled plasma atomic emission spectroscopy (ICP- AES) are used extensively for measuring the concentration of ions dissolved within the solution due to corrosion [28]. Colorimetry is a method for determining the concentration of certain ions based upon the color of the solution compared to standard solutions[29]. Designated additives are required for specific ions [30]. For identifying aluminum, 9- Hydroxyquinoline is needed. Thiocyanate is good for iron measurements. However, the color identification depends on observer s judgment which could cause 7

25 subjective error [29]. A spectrophotometer is designed and applied in order to get accurate results from colorimetry. The device is able to quantitatively measure the solution composition by detecting transmission or adsorption of light with certain bands of wavelength [12]. Inductively coupled plasma atomic emission spectroscopy (ICP-AES) is another popular analytical instrument for determination of trace elements in solution [31]. It is capable of examining over 70 elements. The total time consumed is approximately two minutes [32]. The liquid sample can be nebulized when going through the spectrometer with a stream of plasma argon [31]. The atoms within the solution are excited by the plasma. The excited electrons return to ground state by emission of light. The light from each element has its own specific energy and wavelength. Usually there are multiple elements in the solution, so a spectrum of is collected. A grating separates the light and directs each light to a designated photomultiplier tube detector. The light received by the detector is proportional to the concentration of the relative element. Therefore, ICP-AES is a fast technique with high accuracy Measurements of Gas Evolution from Corrosion Reaction Gasmetric methods are another approach based on collecting gas evolving from the cathodic reaction within sealed system [33, 34]. The evolution of gas is proportional to the corrosion rate [33]. The gas evolution rate is expressed as V. It equals to the total volume of gas (υ) divided by the reaction time (t). 8

26 V= (4) Similar to weight loss, inhibition efficiency (E %) can be calculated through the gas evolution rate, E%= (5) where V and V o are the gas evolution rate with and without inhibitors Other Non-electrochemical Methods of Evaluating Inhibitor Efficacy Other than the methods described above, there are still some less common ways to characterize corrosion and the performance of inhibitors. As we know, adhesion of the coating to the substrate is a key factor in coating efficacy. The bonding between coating and substrate is widely influenced by the additives into the coating. Polysiloxane is a well known adhesion promoter [35]. From the results of pull-off adhesion test under dry conditions in Table 1.1 [36], it is clear that the adhesive strength of coating A on low carbon steels is promoted by one and half times due to the addition of polysiloxane-containing inhibitors. In comparison with results of coating A, coating B shows even better adhesion, which has commercial zinc-aluminum phosphate pigments. The adhesive strength rises about twice comparing to that of blank coating. Therefore, the adhesion could be seen as one of the aspects when considering the effectiveness of pigments/inhibitors. Other than coating adhesion, suppression of blister formation is also a demonstration of inhibitor efficacy. In work by Galliano et al. [36], the time of first blister formed was recorded after an artificial scribe was made with holding at 9

27 cathodic voltage. Scanning acoustic microscopy (SAM) was used to observe the process of blister formation and growth. Figure 1.8 [36] shows the image of one coating after 53h cathodic polarization. After analyzing the final image from SAM, the total area of blister and the maximum distance of blisters away from the scratch were plotted versus time for three coating systems studied in Figure 1.9 (a) and (b) [36]. From the given Figure 1.9 (a), a linear relationship between blister area and cathodic polarization time could be seen indicating that the blister formation follows the diffusion mechanism of water and ions transportation at coating-metal interface [37]. Besides being used to analyzing blister under cathodic polarization, this method could be applied for salt spray chamber testing as well. The inert marker method is a time-saving manner for directly measuring corrosion rate [38]. It involves the use of an ion-implanted Xe layer into the corroding surface and subsequent analysis by Rutherford backscattering analysis (RBS) [39]. RBS is a technique of measuring material composition by detecting the backscattering of helium atoms from the sample [40]. In an experiment conducted by MaCafferty [21], titanium loss in boiling 1M H 2 SO 4 was examined by determining the energy shift in the Xe profile which can be converted into loss of thickness. This worked on the premise that the energy shift only results from Xe removal caused by corrosion. Fig 1.10 [39] presents the shape change of Xe profile versus the depth of Xe moving inward. The depth, in other words, was the thickness lost of titanium layer. The whole test was accomplished within minutes and was more time-efficient when considering the duration of a traditional electrochemical 10

28 method and colorimetric analysis. Its results were consistent with the results from the methods mentioned above [21]. Optical method has also been developed for studying corrosion. A singlepitch Bragg grating fiber sensor designed for this intention was based on Lo s previous work [41]. The principle of the sensor is based upon Bragg s law [42]. λ = 2n eff Λ (6) where λ is the wavelength meets reflection situation, n eff and Λ is the effective refractive index of optical fiber and the single pitch of the optical fiber. The basic design is shown in Figure 1.11 [43]. In Figure 1.11 (a), optical fibers were embedded into the fiber core. The fibers reflect the light with certain wavelength. Then a preload was forced on the core fiber and a metal coating was sprayed on the core. After the preload was removed, a uniaxial residual strain was left. Being exposed to aggressive solution with and without inhibitors, the metal would be corroded away and caused change of the metal thickness and fiber core s strain. The strain difference would affect the light reflected by the optical fiber. From the difference of the wavelengths before and after corrosion, metal coating thickness could be obtained and then converted into corrosion rate. By knowing the corrosion rate and thickness change, the metal protection under types of inhibitors can be assessed. A free section was recommended in Figure 1.9 (d) in order to eliminate the impact of temperature fluctuation. 11

29 Weakness of Non-electrochemical Methods All the methods above regarding measurements of corrosion damage are reasonable and feasible to various degrees, but they have disadvantages [21]. Shortterm corrosion rate measurements many not be representative of long-term behavior. Initial high rates typically decrease as time goes by. Predictions made using short term tests could be overly conservative for long-term service. In addition, the total corrosion loss usually will be converted into mass loss per unit time per unit area during data analysis. However the premise for this conversion is the corrosion happens uniformly across the exposure surface. Many engineering alloys experience localized corrosion damage where the total amount of material loss is very small to begin with. In many of these cases, computation of inhibitor efficiency does not give a clear characterization of corrosion protection. Therefore, it is critical to clarify corrosion type when utilizing and comparing the extent of corrosion. The length of the corrosion test should be provided as well. 12

30 Figure 1. 1: SEM cross-sectional observation of Mg-rich pigmented primer coating on AA 2024-T3: unexposed (a) and after 5090 hrs (b) [11] 13

31 Figure 1. 2: Pigment distribution within chromate primer by Raman spectroscopy [13]. 14

32 Figure 1. 3: Phase identification within chromate primer by Raman spectroscopy [13]. 15

33 Figure 1. 4: Spectra pf mixture of ZnFe 2 O 4 and polyaniline (PANI) within coating by FTIR 16

34 Figure 1. 5: XPS spectrum on bare aluminum sample (a) and silicate-passivated aluminum coupon [17]. 17

35 Figure 1. 6: Optical and Raman images of a section close to a scribe slot after being exposed to neutral salt spray for 5 days (b-d) and before salt spray (e). The maps are for TiO 2 (anatase) at 638 cm -1 (b), BaSO 4 at 988 cm -1 (c) and SrCrO 4 at 866 cm -1 (d-e) [18]. 18

36 Figure 1. 7: Cr distribution along a 0.6 mm slot for different exposure times to neutral salt fog. (orange means depletion and blue represents chromate particles) [18]. 19

37 Figure 1. 8: Morphology of primer coating after 53 hrs cathodic polarization. The scribe was made at the center of the sample [25]. 20

38 Figure 1. 9: Total area of blisters (a) and maximum distance of blister from scratch (b) versus time during cathodic polarization at -1.1V SSE at 5% NaCl solution [25]. 21

39 Figure 1. 10: RBS profile shift of Xe sputtered on titanium versus time immersed into 1M H 2 SO 4 [39]. 22

40 Figure 1. 11: Structure of Bragg grating fiber [43]. 23

41 Table 1. 1: Results of adhesion test [36]. 24

42 CHAPTER 2: CHARACTERIZATION OF PIGMENT DISSOLUTION IN COATINGS 2.1 Introduction Primers coatings are applied to metals and alloys to provide corrosion prevention [44]. Primers improve adhesion of subsequently applied paint and can actively inhibit corrosion due to embedded inhibitive pigments. Corrosion protection is due to the release of ionic species by dissolution of embedded pigment particles that work as corrosion inhibitors. Chromate is considered to be one of the most effective inhibitors for aluminum alloys [45]. It provides both anodic and cathodic inhibition [46]. However, chromates cause DNA change and cancer [47]. Thus, the use of chromate pigments is restricted. Chromate-free pigments have and continue to be developed. Currently, there are some commercial available chromatefree primers, but the fundamental understanding of how they perform is still absent. This study focuses on characterizing and analyzing the diffusion and dissolution of pigments within organic primers subject to aggressive exposure. Focused ion beam (FIB) sectioning was used to create samples for analysis. Energy dispersive X-ray (EDX) mapping was used to identify phases within pigments. Then a Matlab TM program was developed to analyze the dispersion of 25

43 inhibitive ions from images collected from exposed samples. A SrCrO 4 -pigmented epoxy primer was examined as were several chromate-free primers. Pigments within chromate-free primers have elements such as barium, calcium and silicon, which are thought to be dissolved and diffused from pigments (Table 2.1). The presence of a sparingly soluble pigment within coatings does not assure that it will constitute to corrosion protection. It must dissolve to such an extent that it results to accomplish inhibition. In this chapter, results of experiments are presented to help assess the behavior of pigments in Cr-free primers. 2.2 Experimental Procedures Sample Preparation Coating Systems A chromated primer and several Cr-free primers were examined in this study as shown in Table 2.1. The three chromate-free epoxy based primers used were Deft 02GN084, Hentzen TEP, and Sicopoxy With the help of EDX mapping and other information collected, the details of the primer formulations were developed [48]. Deft 02GN084 is a high-solids epoxy-polyamide nonchromate primer. It is a solvent-based paint and mainly containing PrO x, CaSO4, TiO2 pigment carriers. Among these three compounds, PrO x represents Pr 2 O 3 or Pr 6 O 11, which are sparingly soluble oxides believed to providing active corrosion protection [49, 50]. For aluminum alloys, Pr cations decrease the corrosion current 26

44 by at least one magnitude at ph 5 to 7 range. And they also increase the range of passivity and shift open circuit potential (OCP) to more negative potential by dissolution and precipitation of praseodymium hydroxide or hydroxide carbonate. CaSiO 3 is the primary inhibiting pigment phase within the Hentzen TEP primer which is epoxy-based. BaSO 4 and SiO 2 are also present in this coating. CaSiO 3 and BaSO 4 are found in Sicopoxy All primers were mixed and applied following manufacturer's instructions [48]. They were then sprayed and kept in lab fume hood for 24 hrs to air dry. Afterwards, primers were cured at 65ºC for 48 hours. The thickness of these chromate-free primers as applied ranged from 20 to 30 μm, similar to that of chromate epoxy primers. A polyurethane topcoat was applied after the primers were applied. Samples with primers and topcoat were air-dried for 24 hours at room temperature, and then cured at 65 o C for 48 hours. Average thickness of topcoat was about 40 μm. Half of the primed samples was top-coated, the other half was not [48]. Within the present experiment, samples coated with Hentzen and Sicopoxy primers had polyurethane topcoat. The samples with the Deft 02GN084 primer had no topcoat. The chromated coating systems were prepared at the Patuxent Naval Air Station Patuxent River Maryland. The coating system was composed of chromate conversion coating, chromated primer and topcoat. The chromate conversion coating was Alodine 1200S (MIL-DTL Type I, chromate control). The chromate primer was a solvent-based PPG CA7233 (MIL-PRF Type I, Class C control). The inhibiting pigment in PPG CA7233 was soluble strontium chromate. 27

45 The topcoat was a two-component polyurethane (PPG CA 8211/F36375 (MIL-PRF Type IV). The primer and topcoat were spray applied according to standard procedures. The primed panels were allowed to cure for 24 hours at room temperature before applying topcoat, and then cured for 14 days at room temperature before further use Salt Spray Exposure The size of AA2024-T3 samples for salt spray was 125 x 75 mm. These samples were exposed in a salt spray chamber (Q-FOG, Inc.) during which pigments within primers dissolved to various extents. Before salt spray exposure, black electrical tape was used to wrap edges of the samples so as to prevent corrosion from the edges of samples. Coupons with primers were put on fixtures to orient them at an angle slightly off vertical during exposure. The exposure followed conditions described in the ASTM B117 standard [51], which directs the use of a solution of 5% sodium chloride at 35 C. The exposure duration for chromate-free primer samples was 90 days. Two sets of samples with a chromated primer and topcoat were also exposed. One set was sat in salt fog for 30 days and the other one for 90 days. Only the set of samples exposed for 30 days was tested within this experiment. Chromate-free primers exposed in ASTM B117 for 30 days as part of a previous study [48]. Unexposed samples of each coating system was obtained and examined as controls. 28

46 Sample Mounting and Polishing Exposed samples were cut into 12 x 6 mm pieces and mounted in epoxy. Samples were polished using successively finer SiC grit papers starting at 240 and finishing with 1200 grit (Buehler Inc.). DP-Lubricant Blue (Struers A/S), a nonaqueous fluid was sprayed for lubrication and preventing sample attack by water. Diamond paste 1μm and 3 μm (Buehler Inc.) was used with micro diamond compound extender (LECO Corporation) on a microcloth for final polishing. Between each polishing step, the samples were rinsed with ethanol and dried with air. Surfaces were examined under optical microscopy to assure readiness for further preparation. After final polishing, samples were cleaned ultrasonically to remove polishing residue. To facilitate SEM imaging, a thin gold coating was applied by a Pelco Model 3 sputter coater (Schneider Electric). Colloidal graphite paint (Ted Pella Inc.) was brushed all over the mount surface to increase conductivity and prevent charging. With the help of sputtering gold and carbon paint, the mounted samples were examined under SEM without charging Focused Ion Beam (FIB) Trenching Focused ion beam (FIB) milling was used to create cross sections for analysis. Samples were prepared using FEI Helios 600. When trenched, samples were placed horizontally (Figure 2.1). Titled cross sections were prepared by ion beam milling craters. The size of the observed area was approximately 30 x 25 µm. The beam voltage and current was 30 kv and 21nA for trenching, and 16kV, 4.7nA 29

47 for surface cleaning. Two cross sections were prepared for each coating system at each exposure time analyzed Non-electrochemical Characterization Scanning Electron Microscopy (SEM) and Energy Dispersive X-ray (EDX) Mapping A FEI/Philips XL-30 environmental scanning electron microscope (ESEM) equipped with energy dispersive spectroscopy was used to image FIBprepared cross sections. Energy dispersive X-ray (EDX) elemental maps were collected on samples using a 12 kev accelerating voltage and a spot size of 3-5. Pigment particle types within primers were identified by collocation of element maps with SEM images, along with product specification information. There was no gold coating and carbon paint on the surface of prepared cross sections as they were milled away by focused ion beam when trenching Image Analysis by Matlab TM Program In order to analyze EDX mapping data to assess pigment dissolution, a Matlab TM program was developed based on elemental X-ray signal intensity. In this approach, the absolute value of elemental X-ray signal intensity could not be used as concentration directly. The X-ray signal intensity is an accumulation of counts at each image point at characteristic energy range. The higher the concentration, the 30

48 greater the number of counts. However, the counts could not be converted to concentration, so the signal intensity indicated only the relative magnitude of elemental concentration at each pixel. In this analysis there was no correction for absorption atomic number or fluorescence. A technique was developed to make comparative assessments of particle dissolution. The average (Ave) of the elemental X-ray signal intensity in a fixed area on a particle was used to estimate particle dissolution. Increasingly larger areas with radii of 5, 10 and 15 pixels were analyzed depending on the size of the particle. The center of the circle in which analysis was conducted was set as that with the highest average signal intensity across the selected pigment within radius of 5 pixels. The center selected did not only rely on its own intensity value, but the intensity values of the pixels around it. This method eliminated the possibility of selecting single pixel with unusually high intensity, which might not have seen the real center of the particle. When selecting a single particle for analysis, it was found that sizes and shapes of the particles varied, even though the particles had similar elemental types. In order to limit the differences caused by shape and size, the pigments selected were larger than 30 pixels in any direction and the shape was close spherical. However, the size of pigments could not be too large in case that the center of the pigments did not dissolve during exposure due to its large size. As an example, the fixed area for calculating average of calcium intensity within Hentzen TEP was shown in Figure 2.2. A similar evaluation was done for calcium and praseodymium within Deft 02N084, and barium within Sicopoxy For the deft primer, the size of praseodymium-enriched pigments was small, so the radius of 31

49 fixed area was limited to 5 and 10 pixels respectively. The percentage decrease of elemental signal intensity within fixed area was collected in order to see whether particle dissolved and soluble products spread out through the surrounding polymer. The calculation for estimating decrease in elemental signal intensity was: where I 5 is the average signal intensity of the fixed area within radius of 5 pixels and I b is the average signal intensity of within radius of 10 or 15 pixels depending on the area selected. The smaller the percentage residual, the higher the dissolution rate during the same amount of time. Calculations were carried out for each element on multiple particles. Only the values of average percentage residue (APR) are shown in Tables The standard deviation of APR for each fixed area is shown as well. Average intensity of the fixed area of particles within each primer was shown in Appendix A. 2.3 Results and Discussion Cross Section by FIB Milling A 25 x 30 μm cross-sectional area was made by FIB trenching on a chromate primer. Cross sections of other primer coatings were prepared by FIB milling as well. In Figures 2.3 through 2.6, it can be seen that the size of pigment particles varied in the range from 0.5 to 10 μm, and the shapes of particles were quite irregular. 32

50 EDX Mapping EDX mapping was done on cross sections to determine particle identification. SEM images and EDX maps for primers examined are shown in Figure The phases were identified by analyzing SEM images and collocation of EDX maps. The results of phase identification process for each primer are shown in Table 2.1. A series of EDX maps of each primer are listed in Figure Only selected elemental maps were shown. Carbon and oxygen X-ray maps were not shown as they were not the primary inhibiting species. In Figure 2.3, Cr and Si were identified in the unexposed chromate coating. Cr and Si both appeared in the mapping images. This was consistent with the information provided by Naval Air Station Patuxent River and confirmed the existences of chromate. Since K series x-ray emission energy of Sr was much larger than 12 kev, the energy used in EDX, so analysis of Sr was not done [52]. By comparing the EDX maps and secondary electron (SE) images, the location of SrCrO 4 pigments could be determined by the location of Cr in the EDX maps. The size of chromate pigments ranged from 1-5 μm. The distance between two nearby individual chromate particles was quite small, about 1 to 2 μm. In the mapping images, chromium covers majority of area of the cross section and revealed the high volume concentration of chromate pigments. The loading level was so high that adjacent SrCrO4 pigment particles often appeared to touch. In Figure 2.7, the general intensity of chromium x-ray signal was quite high. This revealed the overall high concentration of chromium distributed within the primer. In addition, there was an inherent Cr-rich layer that appeared at the coating-metal interface. This was due 33

51 to the chromate conversion coating since the coating was not exposed to the salt spray chamber at all. Oxygen was found to be collocated with Si and the particles were identified as SiO 2 (Figure 2.3). SiO 2 flake is widely used as pigment within primers [53]. The size of SiO 2 particles was in the range of 1 to 5 μm. TiO 2 particles were identified from the collocation of Ti and O. They spread widely across the cross section, but had low volume fraction. The size of TiO 2 particles was very small and hard to detect within the SE image. In unexposed Deft 02GN084 (Figure 2.4), there was calcium, silicon, titanium and praseodymium, which was consistent with manufacturer s product information [48]. The overlap of Ca and S indicated the existence of CaSO 4. This phase appears as dark grey particles in the BSE image. CaSO 4 pigment particles were much larger than other types of particles within Deft primer. TiO 2 was determined from EDX mapping of elemental Ti. It was distributed through the cross section as small clusters. In the BSE image, Pr(OH) 3 appears as white spots due to atomic number contrast. Compared to chromium in chromate primer, the volume fraction of praseodymium was much smaller than chromium. Praseodymium was not detected at the coating-metal interface, even after ninety-day s exposure (Figure 2.9). This indicated that little if any praseodymium appeared to reach the coating metal interface. This lack of interaction may be due to a lack of inhibitive ion release or poor transport through the resin phase. Particles in Hentzen TPE and Sicopoxy were identified as described above. In the BSE image of Hentzen (Figure 2.5), the brightest 34

52 particles were identified as BaSO 4 due to the overlap of Ba and S. The size of BaSO 4 particles was about 1 to 3 μm. Overlapping of Ca and Si confirmed the existence of CaSiO 3. CaSiO 3 pigments were bright gray. CaSiO 3 is thought to be the inhibitor, but most of Ca element was far away from the interface, and particles were isolated from each other. Apparently, no calcium layer formed at the interface at the end of ninety-day test (Figure 2.10). In terms of the amount of calcium on the cross sectional area, it was greater than that of praseodymium, but still less than chromium. The area of Si was much larger than that of Ca. This could be explained by existence of SiO 2 particles, which were dark gray. Ca and Si were detected in Sicopoxy and were collocated in the mapping images (Figure 2.7). This demonstrated the existence of CaSiO 3. However, only several CaSiO 3 particles were seen on the cross section and none of them were larger than 1 μm. Ba and S overlapped and were prevalent in the primer. This indicated that there was BaSO 4 in this primer and it appeared to be a primary inhibitor in view of its prevalence. The volume fraction of barium sulfate was the largest among the three particles types. However, the amount of barium was still not as great as chromium in the chromate primers. And the intensity of barium was smaller than chromium, which meant lower concentration when comparing (Figure 2.3 and 2.6). And no clear barium layer was observed at the interface after long exposure times. EDX maps of primer cross sections from exposed samples showed no obvious evidence of dissolution and dispersion of the pigment particles (Figure ). 35

53 Image Analysis by Matlab TM Program Chromium x-ray intensity within chromate primer did not change significantly with increasing exposure time as the pigments were connected and mixed together and the distribution of chromium intensity was quite even across the cross sectional area. Therefore, PR could not be used to indicate the dissolution of chromate as there was almost no change of PR due to exposure. However, in terms of isolated pigment particles, analyses were carried out to determine if the dissolution of pigment particles could be detected. The APR of calcium X-ray intensity within Hentzen TEP in a fixed area was calculated and is collected in Table 2.2. With increasing exposure time, the APR for calcium within fixed area of a single pigment particle decreased. As seen in Table 2.2, it fell from 84 to 71 percent after 90-day exposure indicating that the intensity of calcium dropped dramatically within same fixed area (15 pixels) after exposure. However, the relative large standard deviation made the conclusion weak. Similarly in Table 2.3, PR of calcium X-ray intensity decreased from 75 to 62 percent after salt spray test. All those above demonstrated calcium ion dissolved away from the center of the pigment particle during exposure regardless of the presence of a topcoat. The results of praseodymium from analysis of the Deft coating followed the same trend, (Table 2.4) reducing from 79 to 66 percent when comparing data from unexposed and 90-day exposed sample, even though the results were affected by the standard deviation of APR. Barium within Sicopoxy also dissolved as its APR went down from 83 to 72 percent (Table 2.5). 36

54 The possibility of barium dissolution was higher than other tested elements as it had relative small APR standard deviation. From the analysis above, barium, calcium and praseodymium all dissolved during exposure since the APR of each element decreased with increasing exposure time. However, the rate of pigment dissolution was different and was affected by existence of topcoat. With topcoat applied, the APR in the circle with radius of 15 pixels hardly changed after 30-day exposure for Hentzen. In Table 2.2, with the existence of topcoat, APR of calcium X-ray intensity only decreased by 4% after 30-day exposure. Due to the effect of standard deviation, the decrease could be even smaller. Meanwhile, the APR of calcium intensity dropped from 75% to 64% without application of topcoat during the same period of time (Table 2.3). The average decrease rate of APR in the absence of topcoat was almost 2.5 times as that when there was topcoat during the first 30 days. However, there was only 2% drop from 30 to 90 day exposure for calcium with no topcoat in the next sixty days. Big value of standard deviation may lead to larger drop. Generally, the decrease of APR with no topcoated primer was smaller than that with topcoat in the later sixty-day exposure. Regarding the non-topcoated samples, calcium intensity decrease rate in the first 30 days was ten times larger than that in 30 to 90 day period, while the change of APR was very stable during the exposure when there was topcoat. This result suggests that calcium dissolves quickly at the outset of exposure, but slows down if there is no topcoat. When a topcoat is present the dissolution of calcium is restricted. However, due to the effect of the standard deviation, the conclusion above needs further demonstration. 37

55 Within only primed coating, APR of praseodymium decreased by 8% after 30-day exposure, but by just 5% from 30 to 90 day exposure. Dissolution of praseodymium performed similarly to calcium. It followed the trend that dissolution was strong initially, but diminished gradually. Similarly, the rate of dissolution was influenced when considering the standard deviation. The APR of intensity of Barium dropped only 2% in the first 30 days when there was no topcoat present. Then it fell by 9% for the next 60 days. The dissolution rate was become faster, but just increased by 2.5%. Therefore, we can say that under aggressive environment barium rich pigments dissolved very fast and then slow down. Small standard deviation for barium made the conclusion more persuasive than the conclusion drawn from other inhibiting element. The change of dissolution rate may be because water could easily penetrate through the primer and dissolve the pigments. Thus, the dissolution becomes fairly obvious in a short time and then slows down. This also aligns with the driving force of dissolution which is proportional to concentration gradient. Dissolution of inhibitive ions was quick as concentration gradient was large at the beginning and then become very slow as the gradient decrease. However, the pigments with topcoated primer dissolve very slowly and constantly. The water permeation required for pigment particle dissolution was limited by the barrier properties of topcoat [54, 55]. The exception to these general trends was barium sulfate whose dissolution rate increased with time. 38

56 Conclusion Chromated primers are characterized by a high volume fraction of SrCrO4 particles. Their volume fraction appears to be greater than the volume fraction on inhibitor-bearing pigment particles in any of the other Cr-free primers examined in this study. By visual assessment, the volume concentration of praseodymium was smallest within Deft 04GN098. Barium particles in Sicopoxy had the highest volume fraction in its primer and its volume fraction was close to, but not as great as that of chromium within PPG CA7233. The barium concentration was lower than chromium as seen in EDX maps. The amount of pigments particles with primers is a critical factor for inhibitor efficacy. In addition, none of these three non-chromate primers released inhibitors to s sufficient extent that a concentrated protective layer at the interface was formed. In contrast, chromate pigment particles were close to each other. The distance between two nearby individual chromate particles was quite small, about 1 to 2 μm. In the mapping image, chromate pigment particles often appeared to touch one another and Cr element blended together through the outlay of the coating to the coating-metal interface, so they were able to form a channel providing inhibitive ions to coating-metal interface. Although the Cr-free primers did not perform as well as the chromated primer, they still dissolved during the exposure. As the primers exposed to aggressive environment, the APR of elemental x-ray intensity in a fixed area decreased as exposure time increased. This indicated the dissolution of tested element. However, the rate of dissolution was influenced by the existence of topcoat. When there was no topcoat, the dissolution of pigments was fast at the beginning, 39

57 but slowed down. For the instance of the presence of only primer, water moisture could easily penetrate into the primer and help dissolve the pigment. Then the dissolution slowed down as the concentration gradient became smaller as pigments dissolved. However, the dissolution rate of pigments was very similar along the whole exposure if there was topcoat. This appears to be a result from the barrier properties of topcoat, which constrains water penetration [54]. Extra attention needs to be drawn as the standard deviation has to be considered. In terms of the effect of topcoat on pigment dissolution, the general conclusion requires further tests. 40

58 Figure 2. 1: Position of samples when trenched by FIB. 41

59 Figure 2. 2: Cross-sectional view of calcium intensity profile of primer (Hentzen TEP) after 30-day exposure. Yellow circles indicate increasingly larger areas with radii of 5, 10 and 15 pixels. 42

60 Figure 2. 3: SEM and EDX mapping images of unexposed PPG EDWE 144A with conversion coating and phase identification. 43

61 Figure 2. 4: SEM and EDX mapping images of unexposed Deft 02GN084 and phase identification. 44

62 Figure 2. 5: SEM and EDX mapping images of unexposed Hentzen TEP and phase identification. 45

63 Figure 2. 6: SEM and EDX mapping images of unexposed Sicopoxy and phase identification. 46

64 Figure 2. 7: Cr elemental X-ray intensity map on PPG EWDE 144A by Matlab TM program. 47

65 Figure 2. 8: EDX maps of primer cross sections by FIB (PPG EDWE 144A). 48

66 Figure 2. 9: EDX maps of primer cross sections by FIB (Deft 02GN084). 49

67 Figure 2. 10: EDX maps of primer cross sections by FIB (Hentzen TEP). 50

68 Figure 2. 11: EDX maps of primer cross sections by FIB (Sicopoxy ). 51

69 Primer Possible Inhibiting Pigments Deft 02GN084 Pr(OH) 3, CaSO 4 and TiO 2 Hentzen TEP CaSiO 3, BaSO 4 and SiO 2 Sicopoxy CaSiO 3 and BaSO 4 Chromate Primer SrCrO 4, SiO 2 and TiO 2 Table 2. 1: Phases within tested primers. 52

70 Average Percentage Residual (%) No. of Pixel Exposure Time Unexposed 30 days 90 days ± ±0.4 87± ±0.3 80±8 71±11 Table 2. 2: Average percentage residual of calcium X-ray intensity within fixed area of topcoated Hentzen TEP with, (a) unexposed, (b) 30-day exposure, (c) 90-day exposure. 53

71 Average Percentage Residual (%) No. of Pixel Exposure Time Unexposed 30 days 90 days ±1 82±3 83± ±2 64.4±0.6 62±18 Table 2. 3: Average percentage residual of calcium X-ray intensity within fixed area of Deft 02G084, (a) unexposed, (b) 30-day exposure, (c) 90-day exposure 54

72 Average Percentage Residual (%) No. of Pixel Exposure Time Unexposed 30 days 90 days ±0 71±3 66±9 Table 2. 4: Average percentage residual of praseodymium X-ray intensity within fixed area of untopcoated Deft 02N084, (a) unexposed, (b) 30-day exposure, (c) 90- day exposure. 55

73 Average Percentage Decrease (%) No. of Pixel Exposure Time Unexposed 30days 90 days ±2 91±2 85± ±0.4 81±3 71.6±0.8 Table 2. 5: Average percentage residual of barium X-ray intensity within fixed area of topcoated Sicopoxy , (a) unexposed, (b) 30-day exposure, (c) 90-day exposure 56

74 CHAPTER 3: EVALUATION OF CORROSION PREVENTION 3.1 Introduction Weight loss is the most extensively used method for measuring corrosion. In this approach, a sample with known exposure area and mass are exposed in an environment for a certain amount of time [26]. The mass loss is measured after withdrawing the sample and removing the corrosion product [38]. Weight loss is more commonly used for uniform corrosion, but not for localized corrosion [56]. The mass loss of localized corrosion is minimal and hard to measure, even though the corrosion is serious. The corrosion rate can be estimated based on the weight lost, exposure time and area. However, weight loss of substrate material may not be able to be obtained accurately if there are coatings on the substrate as the coatings must be removed carefully. The corrosion rate will be larger than its actual value if the weight of coating is not eliminated from the total weight loss. Therefore, it is necessary to develop a method measuring material loss directly after exposure without considering the impact of coatings. Measurement of volume loss could be an alternative that eliminates some of the complications posed by the presence of a coating. 57

75 Optical profilometry (OP) is a technique that is able to measure volume loss due to corrosion. Previous work showed OP could measure wear volume loss of bare AA2024 in different solutions [57]. This chapter discusses efforts to use the OP to estimate volume and mass loss. In this way, it is possible to compare the efficacy of different coating combinations, even though there is no information prior to exposure available but only the samples after exposure. 3.2 Experimental Procedure Sample Preparation The samples were prepared by Naval Air Station Patuxent River. They were x 79.2 x 6.5 mm and made of AA2024-T3. Eight holes were drilled on each panel at first (Figure 3.1). Then four kinds of surface treatment and six types of primer combinations were applied. Half of each sample had topcoat above primer coating, but the other half only had primer and surface treatment. After the panels were painted, scribes were applied manually using a carbide-tipped stylus in an "X" pattern across the holes before fasteners were installed. Two types of fasteners were mounted on the samples through the holes. The positions of fasteners mounted were shown in Figure 3.1. Overall, there were 96 types of coating and fastener combinations shown in Figure 3.2. Then scratches were made across bottom four holes. Each coating system was applied on two panels. One group of samples was exposed by following ASTM B117 for 500 hrs, and the other one by following ASTM G85 for 360 hrs [58]. Scribes were applied manually using a carbide-tipped 58

76 stylus in an "X" pattern across the fastener holes before fasteners were installed. Before the panels were processed and painted, the holes were put in the panels. The four surface treatments used were Prekote, BoeGel, Surtec 650 and Alodine 1200S. PreKote is a chromate-free inorganic inhibitor and also silane adhesion promoter. BoeGel also is an adhesion promoter. Surtec 650 is a liquid concentrate based on trivalent chromium that could passivate aluminum. Alodine 1200S is a chromate-containing pretreatment. All test panels were surface treated following regular process. The six primers used are Deft 02-GN-084, Deft-44-GN-098, ANAC s P2100P003, PPG CA 7233, PPG EWDE 144A and Hentzen KEP. All primers were epoxy-amides based paints, which means there is a part A (epoxy) and a part B (amide) that is mixed prior to painting. Deft 02GN084 was introduced in Chapter 2. It is a high-solids epoxypolyamide non-chromate primer. It is a solvent-based paint. Praseodymium oxide is considered to be the active inhibitor compound. The primer s thickness was about 25 μm. Deft 44-GN-098 is also a non-chromate primer qualified to MIL-PRF Class N Type I. Sparingly soluble praseodymium oxide is thought to be the inhibitor source in this coating system. It was 22μm thick and was slightly thinner than the Deft 02-GN-084 application. The difference between this primer and Deft 02-GN- 084 is that Deft 44-GN-098 is a water-based paint, but Deft 02-GN-084 is a solventbased. 59

77 ANAC P2100P003 refers to Akzo Nobel Aerodur MgRP003 (P003). It is reported to be a solvent-based paint with a Mg-pigment phase. Magnesium within the primer will be sacrificially corroded to protect the aluminum alloy substrate. This is a non-chromate paint as well. The average thickness was about 25-30μm. Hentzen KEP is a solvent-based paint. It is a chromate-free paint with thickness of 22μm. PPG CA7233 (MIL-PRF Type I, Class C control) and PPG EWEDA144A (MIL-PRF Type I, Class C1 control) are both chromate primers. PPG CA 7233 is solvent based while PPG EWEDA is water based. Their thicknesses were approximately 23μm. Also, half of each sample was with topcoat. The topcoat a two-component polyurethane (PPG CA 8211/F36375 (MIL-PRF Type IV). The average dry thickness of the topcoat was 50μm. Panels were primed and topcoated according to standard procedures. The primed panels were allowed to cure for 24 hours at room temperature before applying topcoat, and then cured for 14 days at room temperature before further use. The two types of fasteners were made of Ti-6Al-4V and 316 CRES. Ti-6Al- 4V is composed of 6% aluminum, 4% vanadium, 0.25% iron, 0.2% oxygen, and titanium for the rest. It is used extensively in aerospace, medical and marine due to its high strength and good corrosion resistance. Stainless steel 316 CRES has 16 to 18% chromium and 2 to 3% molybdenum. The addition of molybdenum increases its corrosion resistance, particularly to pitting and crevice corrosion in chloride 60

78 environment. Both of the fastener induced galvanic corrosion of aluminum alloys in these tests. When the samples were received, the fasteners were already taken out. The paint was stripped off by immersion in paint stripper PR-3505 MIL-R (Eldorado Chemical Company, Inc) for an hour. However, some coating remnants were remained on the substrate and needed to be cleaned carefully by hand. Generally, there were two types of corrosion existing. These are shown in Figure 3.3 and 3.4. One type is that corrosion induced within area tightly close to scribes (Figure 3.3). The other one is corrosion induced within a relative large area (Figure 3.4) Cleaning Due to the size of the samples, they were cut into half for ease of cleaning. Original weights were recorded before cleaning. One half of a sample was immersed in concentrated nitric acid to remove corrosion product. The amount of acid for each cleaning was about 100 ml to assure that the substrate was totally immersed. The acid was replaced if the cleaning process lasted more than an hour since nitric acid would start to attack the substrates as it decomposed [59]. Decomposition was identified as the solution changed from colorless to yellow. During immersion, the coatings lost adhesion with the substrates and were suspended within the acid. The only exceptions were coatings applied on anodized AA2024-T3 samples, which could be removed by the concentrated nitric acid. 61

79 Therefore, the two groups of anodized samples were not able to be measured with OP. With the loss of corrosion product and coating, the substrate showed its original uncorroded surface, which had a metallic luster (Figure 3.5). Weight change was measured every minutes during cleaning. The cleaning process was terminated when the weight loss was less than 0.01 gram between two consecutive weight measurements. Before each weight measurement, rinsing and drying were necessary. Immersed samples were rinsed in continuous streams of deionized water. Samples were then dried with a paper towel and compressed air. Samples were checked visually to make sure there were no white solid corrosion products within the scratches. In Figure 3.6, a typical example of the weight loss versus time was plotted. Generally, there were three steps. Large weight losses occurred in the first twenty minutes due to the loss of coatings and corrosion products. Then from 20 to 40 minutes, the curve had smaller slope mainly because of the removal of corrosion product. After 40 minutes, the corrosion product was removed completely and the curve became nearly flat since only limited aluminum alloy substrate was washed out. This plot demonstrated that the effectiveness of this cleaning procedure Volume Measurements of Material Lost An optical profilometer was used to measure the volume of material lost due to corrosion. After cleaning, the volume of material lost around the scribed area was measured to characterize the extent of corrosion. As the weight of the sample before 62

80 exposure was not known, weight loss was not known. Optical profilometery (OP) was used to measure the depth of corrosion and test the surface morphology profile. A stitching function was used as the area measured was much larger than that of a single OP measurement. Multiple partially overlapping measurements were matched together to form a single dataset. The area measured by stitching function was in the range from 20 x 20 mm to 40 x 40 mm depending on the actual condition of corrosion. The volume was calculated after the morphology profile of corroded area around each hole at the bottom row. There were situations that uniform corrosion spread out which caused corroded area around nearby holes intersect. In order to deal with the intersection, the corroded area was bisected at the midpoint between holes to delineate and attribute corrosion to each hole. When morphology measurements were completed and stitched, a reference surface height was established against which corrosion attack depth was measured. The reference height was manually adjusted to be the height of the unattacked surface. After setting the reference height, the corrosion depth was measured and total corrosion volume was calculated by an automated routine in the software.. Standard deviation was calculated based on multiple measurements of similar surface. It was around 0.2 mm 3, so the significant figures were left to the first decimal place. There were some unique situations. Some measurements showed the surface close to the holes and scratches was higher than surfaces far away (Figure 3.7). This was caused by mechanical deformation during drilling of the samples processing. 63

81 Abnormally high surfaces were masked to assure corrosion volume was calculated correctly. Each corrosion volume of scribed area was measured twice and then the average value was collected. The overall results of samples with different coating and fastener systems were gathered in Table B.1 and B.4 for both exposure conditions. In addition, cumulative distribution plots of corroded volumes were plotted for each of the main attribute variables in the study such as material of fastener, surface treatment, primer and existence of topcoat. This enabled assessment of the impact of each factor on efficacy of corrosion prevention. In thee plots the y-axis was corrosion volume. The x-axis was calculated as: (eq. 3.1) where n was number of individual observation within each category and N was total number of observations. It was easy to see the percentage of total numbers of samples related to corrosion volume Measurements of Corrosion Area In order to measure the corrosion area of each sample, corroded surfaces of all cleaned samples were scanned. The scanned images were analyzed by the image analyzing software Clemex Vision (Clemex Technologies Inc.). The scale bar was calibrated appropriately and then lines were drawn at the boundary of corrosion area (Figure 3.8). The boundary was determined as the juncture of the corroded surface and uncorroded bright metal surface. Then the software was able to discern the corroded area based on previous calibration. The area of screw holes was subtracted 64

82 when reporting the corroded area. The area of the screw hole was about 24 mm 2. The results of corrosion area measurements were shown in Table B.3 and B.6 for both exposure environments. Average standard deviation was calculated and ended up with 3 mm 2, so that all significant figures after the decimal were dropped Calculation of Corrosion Rate Corrosion rate, R, in mm/yr was estimated from the measured corroded area and volume [60], (eq. 3.2) where V is volume of material loss in unit mm 3, t the exposure time in unit hour and A corrosion area in unit mm 2. And 8760 is obtained on the numbers of hours in one year (24 365). The unit of R is mm/yr. The results of corrosion rate were listed in Table B.2 and B.4 for ASTM B117 and G85. The corrosion rate was plotted as the way of plotting corrosion volume by the category of material of fastener, surface treatment, primer and existence of topcoat respectively to see the impact of each factor on protecting substrate AA2024-T Results and Discussion In general, samples exposed in ASTM B117 for 500 hrs had larger material losses than those exposed in ASTM G85 for 360 hrs as the line labeled as ASTM 65

83 B117 is above that labeled as ASTM G85 (Figure 3.9). The corrosion volume was in the range from 0.9 to 29.4 mm 3. In terms of the corrosion rate, about 40% had similar corrosion rate regardless of exposure condition (Figure 3.10). In the other 60% cases, the corrosion rate of samples exposed to ASTM B117 was bigger than that exposed to ASTM G85. The corrosion rate ranged from 0.48 to 4.8 mm/yr. The corrosion area of half samples was similar to one another no matter which exposure condition the samples were immersed in (Figure 3.11). However, for the other half cases, samples within ASTM G85 has obvious larger corrosion area that those within ASTM B117. The values of corrosion area were between 11 and 1200 mm Types of Corrosion In both ASTM B117 and G85, the samples with Prekote or Boegel as surface treatment ended up with corrosion in a small area close to the scribes (Table 3.1), while the samples with Surtec 650 or Alodine 1200S all resulted in being corroded in a relative large area. PreKote and Boegel were chromate-free pretreatments. Surtec 650 and Alodine 1200S are two types of pretreatments involving with chromium. The results in Table 3.1 proved the types of corrosion did not depend on experiment environment, the existence of topcoat, choices of primer and fastener material. The explanation was that chromium-based pretreatments became a integral part of the metal surface and provided good adhesion and bonding properties for paints and coatings[61]. Therefore, the excellent adhesion and bonding between 66

84 pretreated surface and coating prevented the infiltration of water and oxygen, which resulted in corrosion in a very limited area. In contrast, the two adhesion promoters, Prekote and BoeGel, were not able to provide as great adhesion properties as chromium-based surface pretreatment Corrosion of Samples Exposed in ASTM B117 and ASTM G Effect of fasteners In Figure 3.12, the corrosion volume of samples exposed in ASTM B117 is listed by fasteners material. The corrosion volume around Ti-6Al-V4 fasteners was smaller than that around 316 CRES fasteners regardless of coatings applied. There were thirty-one out of top ranked thirty-two samples that were mounted with Ti- 6Al-4V fasteners (Table 3.2). However, the corrosion rate did not follow the same trend as corrosion volume. Figure 3.13 showed the corrosion rate caused by both fasteners was very close. The corrosion volume caused by 316 CRES was bigger than that by Ti-6Al-4V, while the corrosion rate was very close, so the corroded area affected by 316 CRES was larger than that by Ti-6Al-4V. This also was confirmed the corrosion area measurements in Figure The line referred as 316 CRES was always above that as Ti-6Al-4V. All in all, the reason why galvanic corrosion between 316 CRES and AA2024-T3 were stronger than that between Ti- 6Al-4V and AA2024-T3 was due to the larger corrosion area but not corrosion rate. For samples exposed in ASTM G85 for 360 hrs, the corrosion volume was plotted by type of fastener. And the overall results followed the same trend as that in 67

85 ASTM B117. The corrosion volume around Ti-6Al-V4 fasteners was smaller than that around 316 CRES fasteners regardless of coatings applied (Figure 3.15). There were twenty-four out of first thirty-two top ranked samples that were mounted with Ti-6Al-4V fasteners (Table 3.3). The area affected by 316 CRES fasteners was larger than that by Ti-6Al-4V fasteners (Figure 3.16). However, the corrosion rate was very close (Figure 3.17). This confirmed the conclusion above that galvanic corrosion between 316 CRES and AA2024-T3 were stronger than that between Ti- 6Al-4V and AA2024-T3 due to the corrosion area, but not corrosion rate Efficacy of Surface Treatment In Figure 3.18, Surtec 650 and Alodine 1200S, two types of surface treatments involving with chromium, provided better protection than the other chromium-free surface treatment in ASTM B117, as the samples with those two surface treatments had smaller volume of material lost. Alodine 1200S performed particularly well as none of the last thirty-two ranked corrosion volume was from samples applied with it, whereas the samples with both Prekote or BoeGel made nearly half of the worst thirty-two cases of material lost (Table 3.4). However, the results of corrosion rate were totally opposite to what the corrosion volume expressed (Figure 3.19). It was mainly due to the corrosion area. As stated in Section 3.3.1, samples with Alodine 1200S or Surtect 650 were corroded in the area close to scratches while Prekote or BoeGel coated samples were corroded in an extensive area. The corrosion area of samples applied with Prekote or BeoGel was 68

86 always larger than that with Surtec 650 or Alodine 1200S (Figure 3.20). Although samples coated with Surtec 650 or Alodine 1200S had high corrosion rate, the corrosion area was small resulting in small corrosion area in comparison with Prekote and BoeGel. These findings indicate that the corrosion area differentiated the performance of surface treatment, but not corrosion rate. In general, samples with Alodine 1200S or Surtec 650 were small in corrosion area, but high in corrosion rate. As the corrosion area was the dominative factor when considering surface treatment, samples treated with Alodine 1200S ans Surtec 650 still have smaller corrosion volume than those with Prekote or BoeGel. For samples exposed to ASTM G85 for 360 hrs, Alodine 1200S, the surface treatments involving with chromate, provided the best protection than the other surface treatment in terms of corrosion volume (Figure 3.21). Seventeen out of the top thirty-two ranked samples were coated with Alodine 1200S (Table 3.5). And none of the bottom ranked thirty-two samples had it. Surtec 650 was not as good as Alodine 1200S, but still performed quite well as ten out of the top thirty-two ranked samples were coated with it and only one of the bottom ranked thirty-two samples had it. Samples with Prekote or BoeGel made nearly half of the worst thirty-two cases of material lost while only four samples with PreKote and one of BoeGel sample were within the top ranked thirty-two. The corrosion area of samples applied with Prekote or BeoGel was much larger than that with Surtec 650 or Alodine 1200S (Figure 3.23). Although samples coated with Surtec 650 or Alodine 1200S had high corrosion rate (Figure 3.22), they limited the corrosion area and resulted in small corrosion area in comparison with Prekote and BoeGel. Thus, similar 69

87 conclusion could be drawn that the corrosion area differentiated the performance of surface treatment, but not corrosion rate. This was consistent with the conclusion described in the condition of ASTM B117. In general, samples with Alodine 1200S or Surtec 650 were small in corrosion area, but high in corrosion rate. As the corrosion area was the dominant factor when considering surface treatment, samples treated with Alodine 1200S and Surtec 650 still have smaller corrosion volume than those with Prekote or BoeGel Efficacy of Primers In Figure 3.24, ANAC s P2100P003 and Deft 44-GN-084 were more effective as the corrosion volume of samples with these two was generally smaller than that with the other four primers. About half of samples applied with those two primers were ranked in the best thirty-two (Table 3.6). The performances of these two primers were better than the other primers, but not too much as the lines in Figure 3.24 were not widely separately from each other as in Figure 3.12 and PPG EWDE 144A and Henzten KEP were the worst among these six primers as the lines standing for them in Figure 3.24 were always above other lines. Aside from that, about half of samples applied with those two primers were ranked in the worst thirty-two, while only two samples applied with PPG EWDE 144A and four with Henzten KEP were in the top thirty-two. In addition, results of corrosion rate in Figure 3.25 did not fully comply with results in Figure Corrosion rate of samples coated with PPG CA7233 was overall the lowest. However, PPG CA 70

88 7233 did not perform best. Corrosion area of samples coated with PPG CA 7233 was larger than that with ANAN s P2100P003 (Figure 3.26). The corrosion volume was determined by both corrosion rate and corrosion area. Thus, Deft 44-GN-098 and ANAC s P2100P003 performed best when considering corrosion volume. PPG CA7233 coated samples had the lowest corrosion rate and samples with the other five types generally had similar corrosion rate. ANAC s P2100 and PPG 7233 had better impact in terms of limiting corrosion area. In addition, the effect of applying different kinds of primers on corrosion was not as strong as different types of surface treatments or fasteners. For samples exposed to ASTM G85 for 360 hrs, material loss of samples with each primer varied in a large range (Figure 3.27). All primers were able to protect samples fairly well and keep the corrosion volume lower than 5 mm 3 within the first four tenths of samples. However, PPG EWDE 144A and Hentzen KEP did not always perform that well. As seen in Figure 3.27, PPG EWDE 144A and Hentzen KEP protected the substrate poorly in some cases. In Figure 3.27, largest corrosion volume for samples painted with PPG EWDE 144A was 29.4 mm 3 and with Hentzen KEP was 16.4 mm 3, while Deft 44-GN-098 and ANAC s P2100P003 provided excellent anticorrosive effect and kept the corrosion volume smaller 0.5 mm 3 in every scenario. Eight of bottom thirty-two samples were painted with Hentzen KEP primer, while eight samples with Deft 44-GN-098 were ranked in top thirty-two (Table 3.7). Deft 02-GN-084 and PPG CA7233 was not as great as the best two, but they were still able to remain corrosion volume not over 1 mm 3. Corrosion rates were not distinct from one another. In terms of 71

89 corrosion area, samples with Deft 44-GN-098 and PPG EDWE 144A had generally the smallest and largest corrosion area (Figure 3.29). The effects of the other four primers on corrosion area were not very different. However, smaller corrosion area did not always mean smaller corrosion volume. Neither corrosion rate or corrosion area was quite prevalent on determination of the extent of corrosion when considering effects of primer. Generally speaking, ANAC s P2100P003 and Deft 44-GN-098 provided the best overall protection, while PPG EWDE 144A and Hentzen KEP had the poorest influence on AA 2024-T3. PPG EDWE 144A coated samples had the highest corrosion rate, while samples with the other five types generally had similar corrosion rate. Deft 04-GN-098 performed best on limiting corrosion area. In addition, all of the six primers performed quite well in their best 40% conditions Efficacy of Topcoat Figure 3.30 reflected that the corrosion volume of samples with topcoat was not necessarily smaller than that with primer only as lines was close to each other and intersected several times. Within the 96 coating and fastener systems, the number of samples with topcoat in each rank was nearly equivalent to that with no topcoat (Table 3.8). This reason for this was that values of corrosion rate of samples with and without topcoat were quite similar (Figure 3.31). So were the values of corrosion area (Figure 3.32). Therefore, applying topcoat does not guarantee a better protection against corrosion in ASTM B

90 For samples exposed in ASTM G85 for 360 hrs, the corrosion volume of samples with topcoat was very close to that with primer only (Figure 3.33). What happened upon corrosion rate and area was identical to corrosion volume (Figure 3.34 and 3.35). Within the 96 coating and fastener systems samples with and without topcoat distributed evenly in each rank (Table 3.9). Therefore, applying topcoat does not guarantee a better protection against corrosion in ASTM G85 Generally, topcoat is expected to retard the corrosion rate due to barrier properties [54]. However, a topcoat s superior barrier property limits water permeation, which results into limited dissolution of pigments. And most primers require water penetration to dissolve the pigments in order to release inhibitors. Then the inhibitors reach metal-coating interface to prevent corrosion. In this experiment, all the coatings were scribed. Since the coatings were scribed which means the pigments within primers could reach water moisture, the barrier property of topcoat would be lost. Then the anticorrosive ability would fully reply on the surface pretreatment and primer. Therefore, there is no obvious increase in corrosion prevention with topcoat Conclusion Types of Corrosion Within this study, corrosion type was mainly controlled by the surface treatment, but not fastener, primer and topcoat. Corrosion was always limited in the 73

91 area close to scribes on the samples with Alodine 1200S and Surtec 650, while corrosion in a wide area occurred on the samples with Prekote and BoeGel. This showed that chromium-containing surface treatments were able to restrict the corrosion in a small area by forming great adhesion and bonding with the coating above, so that the infiltration of water and oxygen through metal-coating interface was diminished which resulted in limitation on corrosion expansion. However, the non-chromate pretreatments were not capable of forming as strong adhesion as chromate pretreatments with paints above. Therefore, they cannot limit the expansion of corrosion Effect of Fastener Material Under both exposure conditions, corrosion volume caused by 316 CRES was larger than that by Ti-6Al-4V. However, the corrosion rate of samples mounted with these two fasteners was similar in each environment. The corrosion area influenced by 316 CRES was much larger than that by Ti-6Al-4V. Therefore, galvanic corrosion between 316 CRES and AA2024-T3 were stronger than that between Ti- 6Al-4V and AA2024-T Effect of Surface Treatment In terms of corrosion volume, Alodine 1200S and Surtec 650 provided better protection than PreKote and BoeGel. However, the corrosion rate of samples with chromium-containing coatings was much larger since the corrosion area was 74

92 restrained. In contrast, samples with PreKote and BoeGel has more material lost, but at a slow rate. The restriction of corrosion area outperformed the effect of differences of corrosion rate. Thus, the corrosion volume of samples with Alodine 1200S and Surtec 650 was smaller Effect of Primers In ASTM B117, Deft 44-GN-098 and ANAC s P2100P003 performed best when considering corrosion volume, while samples with Hentzen KEP and PPG EWDE 144A had very large volume of material lost. However, the corrosion volume did not align neatly with the results of either corrosion rate and corrosion area. Therefore, none of those two factors was ruling the corrosion volume. It was necessary to consider both corrosion rate and relative corrosion area together in order to characterize the effects of primers on corrosion. In regards to corrosion rate, PPG CA7233 coated samples had the lowest corrosion rate and samples with the other five types generally had similar corrosion rates. ANAC s P2100 and PPG 7233 had better impact in terms of limiting corrosion area. In addition, the effect of applying different kinds of primers on corrosion was not as strong as different types of surface treatments or fasteners. In ASTM G85 exposures, Deft 44-GN-098 and ANAC s P2100P003 were the best two since they were able to keep both corrosion volume and corrosion rate smaller than that of the other four. The other four could restrict the corrosion volume, but not for every coating combination. Deft 02-GN-084 and PPG CA

93 was not as great as the first two, but they were still able to suppress corrosion volume to less than 10 mm 3. Both Hentzen KEP and PPG EWDE 144A could not limit the corrosion since over half of cases ended up with quite large corrosion volumes. In addition, all of the six primers performed quite well in the best 40% of cases. However, the corrosion volume did not align with the results of either corrosion rate and corrosion area; similar to the outcome in ASTM B117. Therefore, none of those two factors dominated the corrosion volume attribute. It was necessary to consider both corrosion rate and relative corrosion area together in order to characterize the effects of primers. In terms of corrosion rate, PPG EDWE 144A coated samples had the highest corrosion rate and samples with the other five types generally had similar corrosion rate. Deft 04-GN-098 performed best on limiting corrosion area Effect of Topcoat Topcoat did not provide assurance of better protection in both exposure conditions. All in all, the general performance of sample with topcoat was close to that without topcoat. This may be due to scratches, which breached the topcoat. Then the corrosion prevention fully depended on surface treatment and primer coating. That is probably why no obvious increase in protection was seen with application of a topcoat. 76

94 Figure 3. 1: Positions of mounted fasteners. 77

95 Figure 3.2: Coating systems in detail. Figure 3. 2: Coating systems in detail 78

96 Figure 3. 3: Morphology of samples with corrosion induced in a limited area. 79

97 Figure 3. 4: Morphology of samples with uniform corrosion induced in a wide area. 80

98 Figure 3. 5: Sample before cleaning (a) and after cleaning (b) (The red squares were drawn later for OP measurements). 81

99 Weight ( gram) Time after being immersed into concentrated HNO 3, (min) Figure 3. 6: Weight loss measurements during cleaning. 82

100 Figure 3. 7: An example of a situation when the surface close to a hole and scratches are higher than that of surfaces far away. 83

101 Figure 3. 8: Example of corrosion area calculation by Clemex Vision. 84