Research in the recent article has further highlighted the importance of proper deaeration after each test. Without deaeration the reduction efficiency is significantly lower, which negatively influences both accuracy and repeatability.
Adding a deaerator to a system is relatively inexpensive, adding only about 300 USD to a system. Furthermore, the amount of inert gas released by a properly designed is low and has a negligible cost per test.
Proper Electrolyte Refreshing:
quality has a substantial impact on the test results. As the electrolyte is
used, its pH typically lowers due to creation of ions, which inhibit current
flow through the solution. Also increased contamination of the electrolyte tends
to reduce the reduction efficiency.
The amount of tests that can be made in one batch of electrolyte before repeatability begins to suffer depends upon the cell volume, the cleaning practice for the test samples, the constituents in the surface films (non-oxides tax the electrolyte more) and the surface areas under test.
The electrolyte we use is sodium carbonate which is both very inexpensive and easily recycled (in the US). Furthermore, providing distilled or deionized water is also a low expense.
Equation #3 solves for each film constituent in this test:
Thickness for each constituent = ε * Kc * timec * CD 
The proper method of finding the timec to use in this Equation is to find the inflection point for each constituent, which in turn is found by selecting either the appropriate peak in the first derivative or the appropriate zero-point in the second derivative. Due to the fast rate required in QC labs, solving these curves can be rather complex to computerize, just like character recognition still baffles computers. Thus, when there are more than two clear peaks, it becomes harder to solve by computer. However, the user can find the correct inflection points rapidly with a small investment in training. Thus, the controller for each Surface Oxide Tester, requires, at a minimum, an interface allowing the operator to select the appropriate inflection points. Once the user has selected these points, the computer will easily calculate the result.
So this only requires a simple computer algorithm to allow the user to chose the correct inflection points and about an hour training time to allow each operator to learn how to chose these points, which is discussed here: Finding inflection points .
We provide such a system that allows the operator to select the proper inflection points from both the color touch screen and from software run on an external computer. We have also attempted to provide a system which can correctly chose the right inflection points, if the user does not have the time to perform this task. Our artificial intelligence algorithms are not perfect, but they are improving.
Avoid Non-Peak Finding Methods:
On the other hand, some systems used in copper rod plants
use a much simpler but poorer method, by only allowing the
voltage levels to trigger the times. There are numerous problems with this
method. First, systems which use this method do not apply a cell design that
maintains a constant voltage levels over time, so we have heard from previous
users of these systems that their results tend to drift over the years.
Furthermore, even with proper cell design the two
features discussed above (deaeration and electrolyte refreshing) must be applied
even more judiciously with the voltage trigger method. Finally, after all
those measures are added, there are issues that are not controllable by the
instrument-makers, such as:
Surface roughness variation cause actual current
density change. (See
Figure for the influence of current density on voltage. )
Porosity of the constituents. (These
constituent are grown far lower than the theoretical densities in continuous casting and
hot rolling. Growth in annealing oven tend to produce films closer to the
theoretical densities.) For example, the voltage levels for
shaved rod will be different than for hot rolled rod.
For example, the voltage levels for shaved rod will be different than for hot rolled rod.
� Mixing of the constituents. (Again in continuous casting and hot rolling the constituents do not ultimately arrange in isolated layers, but as a mixture.) Again this is seen in the difference between shaved rod and hot rolled rod voltages.
� Presence of non-oxide constituents that are both reducible and not reducible in this test.
These methods that use voltage only have never been recommended by an article or specification, they are just a short-cut for the instrument maker and should be avoided.