Quantitative Reproducibility

Most pyrolysis mass spectrometry applications also require a definite degree of quantitative reproducibility, that is the continuous presence of features with similar relative intensities in analysis patterns obtained from the same sample. Obviously, short-term quantitative reproducibility, better termed "repeatability", is much more readily achieved than long-term quantitative reproducibility. Typical figures for short-term (less than 1 day) and long-term (more than 1 month) reproducibility as determined in Curie-point Py-MS studies on selected biopolymers (ref. 100) are in the 1-3% and 8-11% ranges (average difference observed in the relative intensities of the 40 most intense peaks), respectively. Figure 22 shows the level of long-term reproducibility observed for glycogen. The above-mentioned levels of quantitative reproducibility can be achieved only after some basic considerations regarding ion signal statistics and dynamic range of the amplifier are carefully observed. For example, an average reproducibility level of ±5% cannot be obtained if ' the average peak signal in the mass spectra is produced by less than 400 ions (for

400 detected ions the uncertainty in the signal magnitude measured is roughly 1/2

± 400 ' = + 20, or ± 5%). On the other hand, if the signals become too large, both pulse and analogue amplifiers may suffer signal loss, leading to a non-linear response. This may adversely affect the quantitative reproducibility, especially if the sample sizes vary considerably between duplicate analyses. For a more comprehensive discussion of dynamic range problems in pulse amplifiers, the reader is referred to Section 4.6.

In view of the inherent variability of many samples of biological origin, a 5 - • 10% level of instrument ireproducibility is satisfactory for most Py-MS applications. Further, some applications do not require a high degree of long-term reproducibility since reference samples are co-analysed with the unknown samples, as discussed in Section 2.1 ("operational fingerprinting"). Typical examples of operational fingerprinting, where the need for library spectra is obviated, are provided by studies on the subspecies identification of mycobacteria reported by Wieten et aZ-.(refs. 123 -125), as well as by the analyses of tissue and body fluid samples described in Chapters 6 and 7, respectively. However, even if not required in some applications, a high level of long-term quantitative reproducibility is an indispensable ingredient for achieving any degree of inter-laboratory reproducibility.

Figure 22. Long-term reproducibility of glycogen pyrolysis mass spectra. Conditions: sample 5 yg; Tc 610°C; tj 0.1 s; tj 0.9 s; inlet temperature 150°C; Eel 14 eV. (a) Glycogen, suspended in methanol; (b) the same suspension, analysed after 26 days of storage at -20°C; (c) fresh suspension of glycogen prepared from the same batch, analysed 34 days after the analysis of (a).

Figure 22. Long-term reproducibility of glycogen pyrolysis mass spectra. Conditions: sample 5 yg; Tc 610°C; tj 0.1 s; tj 0.9 s; inlet temperature 150°C; Eel 14 eV. (a) Glycogen, suspended in methanol; (b) the same suspension, analysed after 26 days of storage at -20°C; (c) fresh suspension of glycogen prepared from the same batch, analysed 34 days after the analysis of (a).

0 0

Post a comment