The concentration of an analyte is usually determined by means of a line pair ie.  From the ratio of the intensity of an analytical line to that of a reference line. With photograghic detection of line intensities, because of the wavelength dependence of film and photographic plate sensitivity, it was necessary to use, in each case, a reference line close to analytical line. With photoelectric detection this drawback is eliminated; often one reference line is sufficient.

  A reference line undergoes changes in radiation produced as for the analytical lines, to the same or similar extent. Changes in radiation production are of various kinds and the individual processes, as well as their effect on intensity, are largely unknown. It is mainly a question of processes in the discharge gap, or faults in the radiation offtake.

  A line is suitable for use as a reference line if the RSD of the line pair becomes less than that of the analytical line alone. This simple criterion is jointly determined by the following requirements:

  a) The intensity ratio of the line pair should not be affected by interference from other lines.

  b) The ratio should only be dependent on the concentration and not on discharge parameters, so that the lines are associated with the same ionization state and should have similar excitation energies. Analyte and reference should have similar ionization energy.

  c) Vaporisation of the reference element should take place in a manner similar to that of the analytes. This applies in particular for methods with total vaporization, or if a steady state does not occur. This holds for ADA eg. for the analysis of minerals and similar substances with the cup electrode or the globular arc.

  With simultaneous spectrometers the analytical and reference lines are preset by the manufacturer according to the analytical task to be performed by the user and the experience acquired by the manufacturer.

  With monochromators, an almost inexhaustible field of activity opens up when looking for suitable line pairs. There will seldom be time for unconstrained development, however, and it is therefore usually necessary to refer to data from equipment manufacturers, the literature or information from other users regarding similar analytical tasks.

  The reference line is often a line of the base element eg. in steel, a Fe line, in brass a Cu line. If no element can be found which is suitable as a reference because of variations in the sample composition, a constant quantity of an element not present in the raw sample is added during sample preparation eg. 2%SrO in the case of slag analysis. It is essential to ensure uniform distribution which, in most cases, is not achieved by mixing and grinding, so that fusion becomes necessary. In so doing, isoforming of form of bond is simultaneously carried out, so that the SR of the calibration curves are improved. In the case of liquids, the addition of an external reference is simple eg. in the form of ≤0.1% of a soluble salt of the reference element.

  In the case of SDAR it has been found that with elements mainly or entirely precipitated from the base the RSD is not improved by relating the analytical line to a reference line. In addition, in the case of these elements, as a result of reference the SR of the calibration curves is generally slightly impaired for non-alloy and low-alloy qualities, while being distinctly impaired for medium-and high-alloy qualities. Intensities are then plotted against concentrations in the calibration curves. This is understandable if we take into account that the introduction of precipitated elements into the plasma proceeds in a different way from that analytes dissolved in the base, including the base. Thus for precipitated elements it is only necessary to heat the sample surface sufficiently for it to be vaporized. In the case of dissolved elements and the base metal the crystal must first be destroyed and then vaporized.

  Occasionally the intensity of the background at a line-free location in the spectrum, the total discharge radiation or zero order radiation is used.

  Since these variable take intensity change in the spectral background into account they are suitable as a reference for low concentrations if the proportion of the spectral background in the total measurement significant.

The main purpose of the reference is to improve the RSD and SR. It can also impair the SR, however, if:

  a) the dependence of the intensity of the reference on its concentration in the calibration curve has not been taken into account ie. Intensity ratios have been plotted against the concentration of the analyte.

  b) the intensity of the reference has been affected by changes in the concentration of other elements ie. Inter-element effects are present. It is obvious that such effects exist as the reference is an element like any other and the reference line is a line like any other. The aim in making these comments is to help to do away with the widespread opinion that the reference is a “universal remedy” with which all calibration problems can be cleared up.

  For the analysis of non-alloy and low-alloy metal samples by SDAR the reference line leads to satisfactory RSD. For the elements dissolved in the base and the base the mechanism of removal by the discharge is identical and leads to RDS≤0.5%. For precipitated elements, the reference only works in the steady state, and then only to a very limited extent. The RSD of 2-5% are mostly sufficient because of the small concentrations. The calibration curves are good and have low SR, since the reference element at 95%constitutes the main part of the sample and itself largely determines processes occurring in the discharge gap.

  With high-alloy metal samples the relationships are more complex. In addition to processes occurring in the plasma, the quantities removed per discharge and which depend on the type of alloy, are responsible for inter-element effects. The quantities removed depend on the type of alloy, are responsible for inter-element effects. The quantities removed  depend not only on the discharge parameters but also on physical properties of the metal which are determined, almost unequivocally, by the chemical composition of the sample-which is to be determined.

  Indications that these ideas are correct. Our own measurements of quantities removed in the case of Fe +30%Ni and Fe +30% Cr under identical conditions gave 6.0mg and 3.3mg. The dependence of concentration on intensity in the case of the reference lines Fe 271.44 nm; Fe 187.74nm;Al 305.99nm; Al266.04nm; Al256.79nm;Cu 310.86nm and Cu 296.11 nm in the case of binaries and alloys, and the fact that they are not dependent on the type of discharge indicate that the assumption is correct. Changes in the calibration curves for the reference line Cu 296.11nm were only found with arc-like discharges. It has yet to be clarified whether this is due to self-absorption.

  A regression calculation applied to all elements should allow correction of randomly composed systems with reference to base calibration curves. It can be gathered from this work that most alloy constituents reduce the intensities of the other analytes. The quantity of material removed does not alone determine intensity since it should increase due to alloy constituents in keeping with the reduction in vaporization nugy and thermal conductivity. Added to this are the changes in transition probability due to temperature changes, so that selection of the correct reference line is extremely important for an improvement in the RSD and SR, and this unfortunately receives too little attention. It is understandable, because this means “endless” painstaking work, which nobody wants to do.

  Since the fundamental work done by Gerlach and Schweitzer about 1930 it has become the practice, in the case of AD and SD, to work with references although a misconception still often persists regarding the mode of operation, leading to incorrect applications. Thus the reference is primarily intended to compensate for changes in the vaporization behaviour and probability of transition, and only secondarily for instability due to absorption or in the geometry of the radiation source with resultant fluctuations in the signals from the CCD because of inadequate radiation offtake.