General Aspects of the Graphite Furnace Solid Sampling Method

  • Ulrich Kurfürst


The dosing of small amounts of solid test samples is totally different to that of liquid or slurry test samples in every step of the handling procedure which are separation, quantification and transportation. Using a manual or automatic pipette, all three steps are integrated with the dosing of liquid samples, while for solids these have to be carried out separately: For separation mostly a microspatula is used, for quantification a (semi- or full-) microbalance is required, and for transportation a sample carrier is highly advantageous.


Solid Sampling Certified Reference Material Reference Solution Graphite Powder Combine Uncertainty 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1).
    A precondition was the development of a very stable top loading microbalance M2P (Sartorius, D-Göttingen) whose working head is integrated with the automaton (see Fig. 2.2)Google Scholar
  2. 2).
    The data is taken, with permission, from the Promochem catalogue “Reference Materials for Micro, Macro and Trace Element Analysis” [64]. The PROCHEM group of companies, from their offices in Europe, India and North America, distribute reference materials, including those from BCR, NIST, LGC(UK), NRC, CANMET, NWRA (Canada), IAEA (Austria) and other CRM producers. Their catalogue gives valuable information about matrix, contents, use and handling of these materials.Google Scholar
  3. 3).
    The regression of response Rc on the subsample mass m s is recommended only if the content of the unknown sample is calculated by comparison of the slopes of the regression lines for the calibrant and the sample material [95]. With this technique the uncertainty for the result may be enlarged by the error of two regressions proceduresGoogle Scholar
  4. 4).
    Analyte addition method (AAM) is be a better designation of this method [99] than the commonly used term standard addition method (SAM) Google Scholar
  5. 5).
    On the mass-level of field or bulk samples, the segregation effect can be the dominant factor of the total sampling errorGoogle Scholar
  6. 6).
    For analysts who were against the solid sampling technique at that time, these results proved a presumed serious drawback for its application. Of course, the advocates of the solid sampling technique would have preferred that these results had been found to be irregular (e. g. due to a contamination source)Google Scholar
  7. 7).
    The existing ISO 5725 (2nd edn., 1986) gives a different definition of accuracy which corresponds to the difinition of the term “trueness” given below. However, the proposed definition enables better differentiation between the results of random and systematic effects. Furthermore, the term trueness is used in other guides from ISO (e.g. [146] ) that are relevant for this topic. For this reasons, the above defined terms are used here.Google Scholar
  8. 8).
    Additional demands of AQC such as representativity (field or bulk sampling procedure) and tracebility (e.g. documentation) are not discussed here since the respective principles are identical to these applied with conventional methods.Google Scholar
  9. 9).
    This is particularily true if the certified confidence range is misinterpreted as the interval in which the determined value cq of the CRM must lie.Google Scholar
  10. 10).
    From the author’s experience, such a procedure is more or less common in laboratories which carry out routine work (for all kinds of analytical methods), although these techniques of “internal AQC” are usually not revealed.Google Scholar
  11. 11).
    Often only the precision of the analytical measurements, mostly expressed as the RSD, has been documented in the literature. However, without information about the number of test measurements, the influence of random effects to the uncertainty in the analytical result cannot be estimated.Google Scholar
  12. 12).
    Even if uncertainties of type B have to be considered, the coverage factor can be orientated by the values in the “t-table” as long as the convoluted distribution can be assumed to deviate not to much from normal. The expanded uncertainty then provides a range having an approximate level of confidence that may differ from the level given by the t-distribution; consequently, the term “confidence interval” should by no means be used.Google Scholar
  13. 13).
    For the analyis of solid materials after prior decomposition of the sample as well [660].Google Scholar
  14. 14).
    A recently published software tool gives a practical guidance for the raw data processing and facilitates the uncertainty evaluation considerably [709].Google Scholar
  15. 15).
    In the literature other terms are used for this process, e.g. ashing, thermal decomposition, charring. Because of the furnace shielding gas (argon or nitrogen), not enough oxygen is available this process should be termed pyrolysis; ashing should be used only if oxygen is added with the pretreatment step.Google Scholar
  16. 16).
    There is a discrepancy in the literature about the necessity of matrix modifying. Some of workers, specially those who are familier with the classical approaches, have mentioned the use of matrix modifiers without showing or giving the reason for its need.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1998

Authors and Affiliations

  • Ulrich Kurfürst

There are no affiliations available

Personalised recommendations