F. Deflorian, S. Rossi, L. Fedrizzi and P.L. Bonora
Keywords:
Abstract:
Because you are not logged-in to the journal, it is now our policy to display a 'text-only' version of the paper. This version is obtained by extracting the text from the PDF or HTML file, and it is not guaranteed that the text will be a true image of the text of the paper. The text-only version is intended to act as a reference for search engines when they index the site, and it is not designed to be read by humans!
If you wish to view the human-readable version of the paper, then please Register (if you have not already done so) and Login. Registration is completely free.
JCSE Volume 2 Paper 11 ifnotmode(3,'Submitted 13th September 1999, published for public review 9th November 1999') Statistical analysis of salt spray and EIS testing data concerning organic coated metals F. Deflorian*, S. Rossi*, L. Fedrizzi°, P.L. Bonora* *Dip. di Ingegneria dei Materiali, Università di Trento, Via Mesiano 77, Trento (Italy) °Dip. ICMMPM, Università di Roma la Sapienza, Via Eudossiana 18, Roma (Italy) E-Mail:mailto2(' Flavio.Deflorian','ing.unitn.it'); §1 Abstract Electrochemical Impedance Spectroscopy (EIS) and salt spray chamber are very popular techniques for testing the protective properties of organic coatings. It is a common experience however that the results obtained by the two methods often show big difference and it is difficult to compare the performance. §2 In this paper are discussed some experimental results obtained on polyester coated galvanised steel, analysing the data dispersion of 36 identical samples tested both with EIS and salt spray. The morphology of degradation is very different because some blisters are developed in salt spray chamber, but not in immersion. In order to quantify the degradation in salt spray test we proposed a blister index (BI); and the data dispersion of BI and the data dispersion of the parameters obtained by EIS are compared and discussed. A mechanism is proposed explaining the reason because only the double layer capacitance can be compared with the development of blisters in salt spray chamber while the barrier properties of the coating measured by EIS follow a completely different behaviour. §3 Keywords: Electrochemical Impedance Spectroscopy, salt spray test, statistics, blistering. Introduction §4 In many practical cases two, very different, testing methods are used for characterising organic coated metals: electrochemical measurements, mainly electrochemical impedance spectroscopy (EIS) and salt spray chamber test. Between these two testing methods there are many other approaches as natural exposure, cyclic weathering, other electrochemical methods like cathodic delamination, noise, etc [1,2]. §5 In the literature it is very frequent to find strong criticisms about the use of salt spray for testing organic coatings because this test does not simulate actual application conditions and the results are not completely independent from the operator [3]. However electrochemical testing methods, as electrochemical impedance spectroscopy are very popular in research laboratories, but they are not very diffused in the industry, also because sometimes it is not easy to interpret correctly the EIS results and it is possible to find a high dispersion of data on identical samples [4]. §6 This work is part of a larger project and it is devoted to the study of the difference in the statistic properties of the results obtained by EIS measurements and salt spray tests on organic coated galvanised steel, in order to better understand the degradation mechanism and better choose the testing methods and the experimental set-up. Materials and Experimental §7 Commercial protective coatings on phosphatised zinc electroplated steel, based on epoxy resin (thickness 30 µm) were formed by coil coating process. A set of 72 square identical samples (7x7 cm2) were cut from the same coil sheet and they were observed by a stereo-microscope in order to see if some macro-defects (scratch, pin holes, etc.) were present in the coating. A first set of 36 samples was used for the EIS characterisation and 36 were exposed in the salt spray chamber following the ASTM B117 standard. §8 The samples characterised with electrochemical impedance spectroscopy were immersed in 3.5% NaCl in aerated water. The three electrode electrochemical cell was obtained by sticking a plastic cylinder on the sample sheet and filling it with the test solution. The exposed surface area was about 10 cm2. A platinum counter electrode and a saturated calomel electrode (SCE +242 mV SHE) as reference were used. §9 Impedance data were obtained on coated samples at the open circuit potential using a PAR 283 potentiostat and a Solartron 1255 Frequency Response Analyser. The impedance measurements were carried out over a frequency range of 100 kHz to 1 mHz (5 frequencies per decade) using a 10 mV amplitude of sinusoidal voltage, in a Faraday cage in order to minimise external interference on the system. The impedance measurements were made for all the samples after 1 hour, 7, 21, 35 57 and 100 days of immersion, with particular attention to measure the impedance of all the samples exactly after the same time and in the same experimental conditions. The impedance spectra were analysed using the fitting software Equivcrt [5] elaborated by Boukamp. Results Salt spray data. §10 The samples after salt spray exposure were observed following the standard procedure (ASTM D714) recording at different times for every sample, the dimensions and numbers of blisters. Actually the standard requires only the dimension and the frequency of the blisters and not the exact number. The presence of blisters is the only visible degradation of the coating (no presence of zinc or iron rusts, cracks, etc.). §11 The blister dimension is defined as a number starting from 8 (small blisters) reaching 1 (biggest ones). In order to have a quantitative definition of the blisters amount, we proposed a blister index (BI) which is defined as a number which is the result of the sum of the number of blister with a specific dimension multiplied by a coefficient of dimension. BI = (1) §12 In equation 1 i is the blister dimension (from 8 to 1), n is the number of blisters and a the coefficient of the blister of dimension i. The definition of the coefficient of blister can be done in different ways. In table 1 are reported two different scale of coefficients. The first one is linear, it means that the coefficient grows linearly with the blister dimension, the second one is quadratic, it grows with the square of the blister dimension. Passing from a blister of dimension n to a blister bigger with dimension n-2 (from example from 8 to 6) the radius becomes about double, it means that the area of blister is 4 times higher. For this reason we choose the quadratic scale, because in this case the coefficient of blister is proportional to the area of blister. The total blister index (BI) is therefore in direct proportion with the blister area. Blister dimension (ASTM) Linear coefficient Quadratic coefficient 8 1 1 7 2 2 6 3 4 5 4 8 4 5 16 3 6 32 2 7 64 1 8 128 §13 Table 1: Blister coefficients §14 Figure 1: Blister index (BI)ifscreen(' (click for a larger image)'). §15 Figure 2: Coefficient of variation of the blister indexifscreen(' (click for a larger image)'). §16 In figure 1 is shown the trend with the immersion time of the average of the blister index (BI) of the 36 samples; before 21 days no blisters are visible and after 57 days the degradation was so large to suggest to stop the testing. The trend is almost linear using the quadratic coefficients. Interesting is also the trend of the coefficient of variation as a function of the time of testing (the coefficient of variation is the standard deviation divided by the mean value). The trend is in figure 2 and it is possible to note that the value is quite high (in the order of 100%), but more or less constant, in fact it is impossible to note a clear trend increasing the testing time. EIS data. §17 The experimental results obtained by EIS have been analysed using the equivalent electrical circuit shown in figure 3 in order to obtain information concerning the barrier and adhesion properties of the coating and corrosion reaction of substrate. §18 The elements composing the circuit are the coating capacitance Qc (expressed by a constant phase element CPE), the coating resistance Rp, the charge transfer resistance Rct, the double layer capacitance Qdl and Zdiff that is a generic element modelling the diffusion contribution to impedance. §19 Figure 3: Equivalent electrical circuit. §20 All the parameters have been obtained from fitting results with a relative deviation (χ2) between the EIS measurements and the fitting that is in general lower than 10-4 . §21 The mean values of each considered parameter are shown in figure 4 a,b,c and d. Despite the few number of points in the plots, in particular in the case of Qdl and Rct, which it is possible to measure regularly only after 35 days of immersion, it is possible to note a trend with the time of immersion for all the parameters. §22 The Qc values increase during the initial period of immersion because of the water uptake, after they remain almost constant (saturation) and after 20 days they start to increase further. The decrease of Rp is indicative of the reduction of the barrier properties of the coating due to the ions penetration. The values of Qdl which is related to the area of the substrate in contact with the electrolyte because of loss of adhesion, increase with the immersion time, while the Rct values, related in the inverse proportion with the corrosion rate, decrease. It is interesting to note that the Qdl variation is larger than the Rct one in the same time; Qdl increase of about 30 times, while Rct decreases by only a factor 6. §23 Figure 4a: Coating capacitance Qcifscreen(' (click for a larger image)'). §24 Figure 4b: Pore resistance Rpifscreen(' (click for a larger image)'). §25 Figure 4c: Double layer capacitance Qdlifscreen(' (click for a larger image)'). §26 Figure 4d: Charge transfer resistance Rctifscreen(' (click for a larger image)'). §27 As far as the data dispersion is concerned, we have calculated also in this case the coefficient of variation. The results are shown in figure 5a, b and c. §28 Figure 5a: Coefficient of variation of the coating capacitance Qcifscreen(' (click for a larger image)'). §29 Figure 5b: Coefficient of variation of the pore resistance Rpifscreen(' (click for a larger image)'). §30 Figure 5c: Coefficients of variation of Rct and Qdlifscreen(' (click for a larger image)'). §31 All the parameters, except the double layer capacitance, show a clear increase of the coefficient of variation, which means that the data dispersion increases with the time of testing. The parameters describing the barrier properties: the coating capacitance Qc and the pore resistance Rp show lower values of the coefficient of variation, which means lower data dispersion, than the parameters related to the metal coating interface: the charge transfer resistance Rct and the double layer capacitance Qdl. It is clear however that the only coefficient of variation which is stable, in similar way of blister index in the case of salt spray results, is the double layer capacitance. Discussion §32 A first very clear difference, comparing the testing results obtained by EIS and salt spray is that in the case of testing in immersion the samples are apparently almost intact after more than 3 months of immersion by visual observations. In the case of salt spray exposure, on the contrary, after only few weeks many blister are visible on the coating surface. This is the result of different testing conditions, involving for example different temperatures, oxygen concentration on the coating surface, etc. §33 A second very important aspect is the different trend of the coefficient of variation in the two testing methods. Increasing the time of exposure in the salt spray chamber the data dispersion is almost constant. The dispersion is quite high (the standard deviation of the blister index is in the order of the mean value) but it does not change. On the contrary, in the case of the EIS parameters (Rp and Qc describing the barrier properties and Rct related to the corrosion reaction) the coefficient of variation increase with the testing time. It means that the samples become more and more different from each other increasing the testing time. This result is not due to problem in the testing procedure because, reducing the total impedance with the testing time, the reliability and the accuracy of the EIS data increases. There is only one exception which is the double layer capacitance Qdl which shows a coefficient of variation which is constant and in the order of the mean value, exactly as the blister index. §34 In general, when a system of data, obtained theoretically in identical way, shows a statistical variation which increases with time, this system is influenced by conditions which are not controlled, for example initial conditions. In this case it is very difficult to predict the final behaviour of the single sample. §35 On the contrary if the data dispersion is independent of time it is always possible that the initial condition influence the behaviour (there is a data dispersion) but these conditions are not critical for the evolution of the system. §36 In other words the first system is dominated by defects which determine the total behaviour, the second system can be influenced by defects only during nucleation of the phenomenon (in our case the coating degradation), but not for the further growth. §37 In the case of salt spray test the main degradation morphology, and the only one visible, is blisters formation. It is therefore possible to explain the behaviour of the coefficient of variation of the blister index: the number of blisters is influenced by the presence of defects at the beginning, but the growth of blisters is dominated by the properties of the metal-polymer interface, which means adhesion. During exposure also new defects through the coatings are developed, reducing the barrier properties, but this phenomenon is not visible by simple sample observation. §38 Completely different is the behaviour in the case of EIS characterisation obtained in immersion. The data dispersion of the barrier properties increase, despite the fact that blisters are not present. The degradation of the coating, the reduction of the barrier properties, is mainly due to the development of new pores and defects through the coating. This process is completely dominated by defects which are randomly present on the samples, and not by the interface properties which have deterministic nature. §39 The different behaviour of the coefficient of variation of double layer capacitance Qdl is consistent with this data interpretation. The Qdl values increase because of loss of adhesion at the coating interface due to electrolyte penetration. The process is controlled by the stability of the metal-coating interface, like in the case of blistering, and not by the presence of defect. Also in this case, therefore, the data dispersion does no increase with the time of testing but it remains constant. §40 The only one element of the equivalent electrical circuit which can be related to the blister index is therefor the double layer capacitance Qdl, not only because is the parameter related to loss of adhesion process like the blister growth, but also because the data dispersion is similar. §41 In the case of EIS measurements moreover it is also important to remember that with the measurement of Rp and Qc it is possible to monitor different phenomena. For example considering Qc, in the initial stage the growth of Qc is due to homogeneous water uptake in the coating, while the further uptake, after plateaux is due to water accumulation in specific sites in the coating (metal-polymer or pigment-polymer interfaces) [6]. The overlapping of different mechanisms could influence the data dispersion. Also in the case of Rp it is possible that different mechanism influence the evolutions of the values. For example the increase of the D areas in the coating, in comparison with I areas, following the classical theory proposed by Mayne [7]. §42 Also the corrosion reaction (and therefore the specific charge transfer resistance Rct) can change during testing, for example because of the presence of corrosion products, affecting the data dispersion. The only parameters which seems more directly correlates to a single physical phenomenon is the double layer capacitance Qdl which is a measure of the loss of contact between coating and metal surface because of an aqueous interlayer. Conclusions §43 By testing organic coatings with salt spray exposure and EIS measurements it is possible to obtain in general different results. This well known conclusion is evident not only by looking the different degradation appearance of the coatings (presence of blisters), but also analysing statistically the experimental data using the coefficient of variation which in the case of EIS become larger increasing the time of testing proving an increase of data dispersion, but the data dispersion remains constant in the case of blisters in salt spray chamber. §44 The degradation phenomena which it is possible to observe with EIS are different: barrier coating degradation (coating capacitance Qc and pore resistance Rp) and metal-coating interface degradation (double layer capacitance Qdl and charge transfer resistance Rct). The only parameter which can be correlated with the blistering is the double layer capacitance Qdl. References comment(45) B.R. Appelman, Journal of Coating Technology, 62, (1990), 57; F.D Timmins, JOCCA, 62 (1979) 131; B.S. Skerry, C.H. Simpson, Corrosion, 49, (1993), 663; F. Deflorian, L. Fedrizzi, S. Rossi, P.L. Bonora, Materials Science Forum, 289-292, (1998), 337; B. Boukamp, Solid State Ionics, 20, (1986), 31; F. Deflorian, L. Fedrizzi, S. Rossi, P.L. Bonora, Electrochimica Acta, 44 (1999) 4243; C.C Maitland, J.E.O. Mayne, Official Dig., 34 (1962) 972.