We have hosted a live webinar about best practice for the control of the melting process in foundries, together with the Foundry Planet. If you have missed it, you can watch the recording here.
This is the second post where we answer questions received from attendees of webinar. The first post answered questions around calibration and you can read it here. In this second post, our Business and Product manager Wilhelm Sanders gives his expert answers to questions around measurement, accuracy and precision, and maintaining your instrument.
Q: For Hitachi High-Tech, what is the most important: precision or accuracy?
Both accuracy and precision are important, however accuracy is almost impossible to give values for. This is why none of the spark OES manufacturers give values for accuracy in application notes or give any kind of guarantee. This is because accuracy is hugely dependent on a large range of parameters, including, but not limited to:
On the other hand, precision can be easily found by measuring homogeneous samples for all elements across all concentration ranges. This is why precision values and guarantees are typically given in application notes.
Q: Is carbon measurement accurate with OES? What are the pitfalls with measuring carbon?
For the vast majority of grades and material, carbon measurements can be obtained with spark optical emission spectrometers with good accuracy and precision for low detection limits. There is, however, one significant exception – cast iron. This is because the carbon within cast iron is treated to form free carbon in nodules, laminar or vermicular structures. Unfortunately, this free carbon can be burnt off during the pre-burn stage of making a measurement with OES, leading to inaccurate results. This is also the reason why it’s inadvisable to test grey casted final products for carbon with OES.
However, you can use OES for melt control within cast iron foundries under certain controlled conditions:
For further details on this, please see our cast iron guide.
Q: For %C analysis, why is it more common to use combustion analysis than OES?
My personal view is that is not the case for most foundries because combustion analysis instruments are relatively expensive when compared with OES. However, for large steel plants and bigger foundries, it becomes more cost effective and it’s here that we see combustion analysis used for carbon. The reason this method delivers excellent results is that the effective sample is bigger, which makes it more representative of the material. Disadvantages are high purchase price, relatively difficult sample preparation, high cost of ownership due to consumables and the need for specialist knowledge to use the instrument. If you are careful in your sample preparation (as outlined in our cast iron guide) then OES is a feasible low-cost alternative.
Q: Is sulfur measurement accurate with OES?
The issue with sulfur is due to its solubility. Sulfur will dissolve in liquid iron at any concentration. But, the solubility of sulfur in solid iron is limited: 0.002% in α-iron at room temperature and 0.013% in γ-iron at 1832°F (1000°C).
When the liquid steel cools down and solidifies, the solubility of the sulfur falls and the sulfur is liberated from the solution as iron sulfide (FeS). The FeS forms a eutectic with the surrounding iron and segregates at the grain boundaries. The eutectic temperature is quite low, about 1810°F (988°C).
The Fe-FeS eutectic weakens the blinding between the grains and increases brittleness at hot deformation temperatures (e.g. rolling, forging).
Because of this, the sulfur is not homogeneous in the mix, and this segregation at the grain boundaries increases as the amount of sulfur increases. This is why the results for sulfur tend to show poor precision than for completely dissolved elements in the iron matrix.
Q: How important is it to clean the OES and how often?
During the sparking sequence, a small amount of the sample material is vaporized. On cooling, this material forms a deposit. The amount of deposit generated depends on several factors, including:
For our OE750, we recommend cleaning around 2000 measurements for steel or similar materials, such as Ni, Co and Ti. For lower melting point matrices, such as Al and Cu, we recommend cleaning after every 1000 burns. The spark stand in the OE750 is very easy to clean. You don’t need any special tools as the spark stand has been designed with quick release fasteners, meaning the whole procedure takes less than three minutes.
Q: What is the effectie life of the OES equipment?
Again, this depends on several factors. The main ones are:
When I first started working with OES equipment, more than 30 years ago, most electronic components were stand-alone which made it relatively easy to replace individual components on a PCB. This means that, with a little work being, you could make an instrument last around 30 years. On the other hand, those instruments were much larger than today’s equivalents and 3-4 times more expensive. The small footprint of today’s spectrometers means that the PCBs are highly integrated, using ICs that are often discontinued at relatively short intervals. However, my estimation of today’s spectrometer lifetime is 15-20 years max.
Q: Do you recommend 3 sparks for the same sample for optimum process control?
If this brings the precision (reproducibility) of the measurements within acceptable limits, then yes, standard practice is 2-3 burns.
Q: What types of detector do your stationary spectrometers have?
Most of our instruments use CCDs (charge coupled devices), with the exception of our latest model OE750, which uses CMOS detectors. Both of these types are semiconductor-based and can be tailored to cover the whole spectrum within optical systems.
For the OE750, the CMOS detector covers 119nm up to 766nm, which means it can analyse all relevant elements within metals, from hydrogen upwards. This is because CMOS detectors have a better resolution and dynamic range, and they are more linear. They can also be used for TRS (time resolved spectroscopy).
Photomultipliers can only be used for a specific wavelength selection. However, they are very sensitive and their fast response makes them good for TRS. A useful summary of the differences between different detector technologies can be found from Frauenhofer Instiute.Watch the recorded webinar