The efficiency of in-plant spectrometer for their task-sorting, identification, approximate analysis, analysis-is assessed according to criteria similar to those for laboratory spectrometers：

a) Measurement value precision, long-term stability

b）Discernibility of concentrations( detection sensitivity)

c) Inter-element effects from third elements (accuracy).

Measurement value precision

  To improve the RSD, intensity ratios are determined, which are obtained from the intensities of the analytical lines divided by the intensity of the reference line.

  This is state of the art and is used in all simultaneous AE spectrometers. For a good approximate analysis or identification, a good RSD for the measurement values is essential. When comparing two spectrometers a number of times in succession. Care must be taken to ensure that all measurement values are used, and that the results are not “massaged” by leaving out outliers. Leaving-out outliers is not permissible since, for identification work, as a rule only one measurement is used and thus outliers cannot be recognized as such.

   Long-term stability can be determined by measuring one and the same sample repeatedly over a lengthy period and determining the RSD, or examining the highest and lowest value.

Discernibility of concentrations

   AE spectrometers measure the radiation from spectral lines. Even if an element is completely absent, a single appears which is the background. To be able to assess this, it must be converted into a concentration value which is known as the BEC.

    The lower the BEC, the better the optics since only a little radiation from the back-ground, stray light or other lines passes through the exit slit to fall on the CCD. The BEC can most simply be determined from sets of low-alloy samples.

Logically, this also applies at higher concentrations. In this case, it is not the quality of the optics which is so critical but the use of suitable analysis lines which allow differences to be detected even at high concentrations and are not within the saturation range.

Inter-element effects from third elements

     For identification testing, inter-element effects from third elements are not important since, with all elements being measured, only too many elements may be indicated as different and thus detection of mix-ups is not reduced.

     In the case of approximate analysis, interference from third elements is compensated by corrections.

     The quality of correction depends on the quantity and quality of the calibration samples. It can be ascertained whether many corrections are necessary if measurement values for a large number of samples from one alloy group are plotted against concentration values. If a “starry  sky “ is obtained, numerous correction are necessary: with a “string of pearls” no correction are required.