Variation in apatite fission-track length measurement: implications for thermal history modelling

Predictive thermal history modelling using apatite fission-track (FT) data is dependent on an algorithm to describe the time and temperature dependency of FT annealing which, in turn, relies on the empirical determination of FT length as a measure of the annealing process. Assessment of variation in FT length measurement is poorly described, with few comparisons between analysts and little interlaboratory standardisation. Using apatites of various compositions containing induced tracks annealed to differing degrees, this study has assessed variation in horizontal confined track-length measurement for a variety of procedural conditions. Replicate analysis by a single observer is typically within 3% but increases inversely with track length. Comparison between observers on the same samples shows significant, generally nonsystematic variation between observers; for a complex length distribution variation is similar to 12%. Sources of variation are identified as: (a) variation from track revelation, including etching, track-in-track (TINT) vs. track-in-cleavage (TINCLE) measurement and use of Cf-252 irradiation to produce additional etching channels; (b) bias in measurement, including equipment, analytical procedures, and sample size; and (c) observer bias, principally differences in and consistency of personal technique. 5 M HNO3 of is preferred to weaker etchants: although more anisotropic, tracks are better defined, permitting more rigorous measurement, while c-axis parallel sections (where 2 pi/4 pi geometry is better defined) are more easily identified. For all but the longest length distributions, TINCLEs are significantly longer than TINTS, with few short TINCLEs at high angles; measurement of TINCLEs effectively masks the anisotropy of annealing. Cf-252 irradiation is effective in increasing the number of TINTs sampled and measured. Variation between values measured for unirradiated and Cf-irradiated aliquots does not exceed that found for a single analyst, although a slight systematic shift to longer lengths for Cf-irradiated samples is seen. As reported by other workers, track-length distributions are anisotropic, anisotropy increasing with annealing level. Track angle exerts a major influence on measured length, summing affects from annealing and etching anisotropies with observer bias. Track angle should be accommodated within the annealing algorithm. It is recommended that similar track revelation, observation and measurement conditions are used for the analysis of field samples as are used in annealing experiments, and subsequently employed in numerical models to predict thermal history. A parallel argument can be advanced for using samples of similar composition. Further, we recommend that the FT community should seek as a matter of some urgency a programme of interlaboratory comparison of track-length measurement using standard apatite samples containing artificial length distributions typifying various levels of complexity. Such comparisons would provide a more rigorous baseline for thermal history prediction in geological case studies. (C) 2003 Elsevier Science B.V. All rights reserved.