Hi
I have done potentiodynamic (Tofel) tests on 9 samples of Aluminium 1XXX in 3.5% NaCl. The results of icorr were in the range of 0.337 to 0.877 microamperes (about 2.5 times difference). Is this amount of variation is acceptable or usual in corrosion results?
In many research papers, the amount of icorr of an alloy is reported by just one test. So how much deviation is usual in the results?
Hello Milad Mojtahedi,
The results obtained are interesting and the variations are in a wide range for the same material. This can often happen if there are differences in roughness. Another reason for this may be due to samples taken from different places in the rental. In it we can have different size of the structure. This causes varying degrees of corrosion. Another reason is due to the use of different concentrations of the solution. This often occurs when using the same solution or not the same concentration. Reasons for the difference may be the temperature in the cell and the environment. Atmospheric pressure also has the effect of accelerating the process. There are other reasons, but more data from the experiment are needed. How was the roll obtained and how were the samples taken from it?
Wishing you success in your research.
Hello Milad Mojtahedi,
The results obtained are interesting and the variations are in a wide range for the same material. This can often happen if there are differences in roughness. Another reason for this may be due to samples taken from different places in the rental. In it we can have different size of the structure. This causes varying degrees of corrosion. Another reason is due to the use of different concentrations of the solution. This often occurs when using the same solution or not the same concentration. Reasons for the difference may be the temperature in the cell and the environment. Atmospheric pressure also has the effect of accelerating the process. There are other reasons, but more data from the experiment are needed. How was the roll obtained and how were the samples taken from it?
Wishing you success in your research.
Assuming all the 9 measurements are in carried on in the same conditions and on "virtually" the same sample, the data you have obtained should be normally (from a statistic viewpoint) distributed. The sample size (9 tests) is small, however you may extract the average value and a standard deviation. Only from those parameters you may build an idea of the statistical quality of your experiment.
Dear Dr. Milad Mojtahedi,
I don't think it is possible to give "a priori" an exact "acceptable deviation" for the estimation of Tafel slopes, icorr, polarization resistance etc.. The precision with which the curve is converted from the "true" behaviour into a graph or data table -in this case the answer is essentially a function of the system being used to make the measurement- should be be of the order of 1%. The precision related to a dynamically-measured curve relative to the ideal of a steady-state response -in this case it will depend on the dynamic response of the system being studied- the sweep rate and probably several other factors. The repeatability from one measurement to another you can (and should) measure it and apply the statistic. Similarly there are the parameters linked to lab reproducibility (you might look at the ASTM regulations). I suggest that you "standardize" the parameters of your measurements to the maximum and evaluating "statistically" the repeatability and try to do all validation checks you can.
For more details, you can see at:
https://www.palmsenscorrosion.com/knowledgebase/polarization-curves/#processing
Good work and best regards, Pierluigi Traverso
I'm not an expert of corrosion. However, I think that the observed variability can be classified into two types:
- aleatory or intrinsic: does the variability persist even if you would perform one million measurements under virtually identical conditions? If yes, you cannot avoid it;
- epistemic or lack of knowledge: does the variability reduce if you increase the amount of information (i.e. the number of measurements)? If yes, the sampling variability can be decreased by performing as much measurements as possible.
Another point, equally important, is: are your measurement conditions really identical under repeated measures? If not, you have a measurement error that leads to an additional variability in the outcomes
Acceptance of results and applicable tolerance is agreement between supplier and user. Sometimes, as per governing code requirement.
Dr. Milad,
The number of tested samples should be enough. Corrosion test is very sensitive to environmental parameters. Any tiny distraction from surrounding environment can remarkably affect your results. I would recommend to review your results carefully and look for potential outliers. The other thing that might help is reducing the scanning rate. Also, if you are creating a constrained area on your samples by a tape, make sure that the solution does get under the tape since it would affect your measurement.
Good luck!
Some times oxygen entrapment will lead to wide variations leaving aside surface preparation etc.
In my opinion, the deviations you get just show that the measuring method is not adequate. The corrosion potential of Al alloys in NaCl solution usually approaches the pitting potential thus the anodic branch of your polarization will induce local depassivations that are not reproducible. In other words, the conditions for Tafel behaviour are not fulfilled in this case. Perhaps you can estimate your corrosion rate just by extrapolating the cathodic branch up to the corrosion potential value.
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers