Why do we need high purity metals?
The effect of impurities on alloy properties is well established (1) but is largely forgotten in discussions about bulk materials. The bottom line is this: without high purity metals, researchers may be missing, or not seeing, the true physical properties (or physics of the material) which results in missed opportunities to refine theories, discover new materials and phemomena, or develop devices. Let the image below serve as an example why purity is a critical factor in fundamental research:
The obvious difference in the two entropy curves for the two Gd5Si2Ge2 samples is due to purity. The large peak in the high purity (KAG) sample is derived by both a magnetic and structural transition. This structural transition is not exhibited in samples made from low purity metal.
Another example of purity driving properties in superconducting MgB2 is found in reference (2).
How pure are Ames Laboratory's rare earth metals?
Please examine the image below:
The above chart is based on a comparison of Ames Laboratory lanthanum vs. a 4N (99.99%) commercial lanthanum product. In terms of atomic percent, both are very good with respect to intra-rare earth impurities (orange highlighted elements), with the Ames material being slightly improved, as shown in the orange highlighted values. If we add in alkaline, alkali, transition metals, metalloids and some of the non-metals (yellow highlighted elements), the Ames lanthanum is still a 4N metal while the commercial is a 3N metal. Now add the gasses (the green highlighted elements) and the Ames metal is 3N while the commercial material is just breaking a 2N rating. This type of relation holds for Ames and commercial materials starting at 5N rare earths purity.
What to you mean by 5N or 3N?
It is important to understand how purity is defined when acquiring alloying materials. When a material is declared as 99.99% pure, or 4-nines, or 4N, the declared value and the true value of the concentration of impurities most likely are not the same. This is due to many manufactures not testing for all possible elemental impuirties.
Let us first lay out what the % or Nines scale means in terms of parts-per-million (ppm)
|Nines||Purity %||Total Parts||Matrix ppm||Impurity ppm|
So, for example, in a 4N pure tin (Sn) ingot, for every million atoms of matter, 999,900 of the atoms are Sn atoms with 100 atoms of other elements.
What is the basis?
The issue of purity is often confusing, as there is the absolute purity and the common metals basis purity. On an absolute scale, 99.999% is a very high bar to achieve, since this allows for only 10 ppm impurity with respect the all elements on the periodic table. Metals basis, as is commonly indicated in catalogues, does not include many of the metallurgically important interstitial elements or other nonmetallic impurities. For example, a certificate of analysis that declared Mn is 99.95% purity (min. metals basis) also reported that the Mn has 0.3% oxygen. The Mn is not 99.95% on an absolute basis. If you include the oxygen content, at best the Mn is 99.65% pure on an absolute basis. A metals basis purity may not include: H, B, C, N, O, F, Si, P, S, Cl, As, Se, Br, Te, I, At. Some certificate of analysis forms may only list a handful of elements that were tested; the remainder are untested or unreported. The true value of the impurity concentration is therefore not known.
Why are such limited assays reported? Because vendors are pragmatic. For example, the semiconductor industry demands extremely tight purity specifications. As they are the biggest players/buyers in the game, the Si vendors do extensive testing and reporting on the purity of their silicon. To the contrary, nickel is heavily used in metal castings were purity is not as critical except for some specific contaminants - sulfur for example. Hence vendors will limit their assays to the elements of concern. Also, testing can be expensive so vendors will limit the assays to avoid paying for tests that are not required by the customer.
Atomic versus weight based reporting.
Most reports use ppm by weight; however, some reports are in ppm atomic. The difference is shown in the following example:
|Matrix||Impurity||Reported as Weight ppm||Reported as Atomic ppm|
Reporting the impurity concentration in weight percent masks the true atom-to-atom ratio of the impurity. If we were only concerned with the oxygen in the tungsten, we could state that the tungsten is 99.98% pure with respect to oxygen on a weight basis, or the tungsten is 99.8 % pure with respect to oxygen on an atomic basis. One must recognize this effect in order to keep purity in perspective, especially when the atomic weights are very different: impurity Z << matrix Z.
What is the error? (3)
Each analytical measurement is an experiment attempting to arrive at the true value of the measured quantity. The error in the measurement is:
Error = measured value - true value
As we don't know the true value we don't really know the error. What we do know is the magnitude of the uncertainty in our measured value is based on our understanding of the methods employed. Error/uncertainty is typically not reported in certificates of analysis. This (±) error range will vary for each analysis technique; for example, the sum of the three constituent elements in a ternary alloy may add up to 102% in an ICP measurement where the relative error is ± 3%.
It has become the norm that analytical assay numbers are reported without error ranges. It is up to the individual to have an understanding that each method has its particular accuracy and precision.
(1) Metals, alloys and compounds-high purities do make a difference!
Journal of Alloys and Compounds, Volume 193, Issues 1-2, 15 March 1993, Pages 1-6
K. A. Gschneidner Jr.
(2) Effect of Boron powder purity on superconducting properties of bulk MgB2
Physica C: Superconductivity, Volumes 460-462, Part 1, 1 September 2007, Pages 602-603
Xun Xu, Dayse I. dos Santos, J. H. Kim, W. K. Yeoh, M. J. Qin, K. Konstantinov, S. X. Dou.
(3) Analytical Techniques in the Sciences, Chapter 1: Analytical Measurements, Edited by Graham Currell, (2000) John Wiley & Sons, Ltd.