Peak-width errors

Geometrical errors associated with manual measurements include height error, peak width error, half-height error, and baseline position error. 

It is seen that the molecular weight of a solute having a density between 0.85 and 1.25 can be estimated experimentally from peak width measurements for 90% of the compounds within an error of 13% (80% of the samples gave an error of less than 10%). Again, depending on the field of application, it should be noted that such data for a given substance can be obtained in addition to its separation from... 

In addition, as mentioned before, the visual identification of the spot width is also open to considerable error but, if the plate is scanned, and the elution curve obtained, then both the peak width (w) or the peak width at the base (wb) can be obtained fairly accurately (assuming appropriate software is available). 

Having chosen the test mixture and mobile diase composition, the chromatogram is run, usually at a fairly fast chart speed to reduce errors associated with the measurement of peak widths, etc.. Figure 4.10. The parameters calculated from the chromatogram are the retention volume and capacity factor of each component, the plate count for the unretained peak and at least one of the retained peaks, the peak asymmetry factor for each component, and the separation factor for at least one pair of solutes. The pressure drop for the column at the optimum test flow rate should also be noted. This data is then used to determine two types of performance criteria. These are kinetic parameters, which indicate how well the column is physically packed, and thermodynamic parameters, which indicate whether the column packing material meets the manufacturer s specifications. Examples of such thermodynamic parameters are whether the percentage oi bonded... 

Prior knowledge allows to include fixed relations between some of the four parameters (amplitude, phase, frequency position, peak width) describing a symmetrical well-shaped resonance. Signal ratios, chemical shift difierences, linewidth relations and zero-order phase relations can be included. The reduction of the number of unknown parameters leads to a reduced calculation time, better convergence behaviour and improved results. However, the assumptions made to include the prior knowledge must be validated for each experiment. Differences between the parameter values set by the prior knowledge and the actual parameters could lead to systematic errors. 

The shape ofthe elution curve for a pulse injection can be approximated by the Gaussian error curve for AT > 100, which is almost the case for column chromatography [2]. The value of N can be calculated from the elution volume Vg (m ) and the peak width W (m ), which is obtained by extending tangents from the sides of the elution curve to the baseline and is equal to four times the standard deviation (Ty (m ) = as shown in Figure 11.7. 

The ELS detector was previously also referred to as a mass detector, pointing to the fact that the response is (mainly) determined by the mass of the sample rather than by its chemical structure. Van der Meeren et al., though, demonstrated that the ELSD calibration curves of phospholipid classes were also dependent on the fatty acid composition (52). The dependence on the fatty acid composition is, however, completely different in nature and much less pronounced than for UV detection. The reason for this behavior is to be found in the partial resolution of molecular species, even during normal-phase chromatography. Thus, the peak shape depends not only on the chromatographic system but also on the fatty acid composition and molecular species distribution of the PL sample (47). Because it was shown before, based on both theoretical considerations and practical experiments, that the ELS detector response is generally inversely proportional to peak width (62,104), it follows that the molecular species distribution of the PL standards used should be similar to the sample components to be quantified. It was shown that up to 20% error may be induced if an inappropriate standard is used (52). 

C4 peak on the shot reaction chromatogram, making measurement difficult. The estimated probable error has been indicated by bars in Figure 3. (A typical peak is shown in Figure 5, d the dashed line shows the peak width for a shot passed through the analytical column alone.) There was also a loss of material in the shot reactions—i.e., the difference between the initial and final quantities of thiophene (upper curve, Figure 3) was greater than the C4 produced (middle curve). Presumably the loss was due to irreversible adsorption of either butene or thiophene, possibly as a polymer. 

The perpendicular drop method produces accurate peak areas for symmetrical overlapped peaks of similar height and width. In this case, the portion of the second peak attributed to the first peak is offset by the portion of the first peak attributed to the second peak and vice versa. For overlapped peaks of significantly different size, the perpendicular drop method always overestimates the peak area of the smaller peak as the smaller peak gains more area from the larger peak than it loses to the larger peak. Quantitation errors are further exacerbated in the case of a smaller peak imposed on the tail of a much larger tailing peak. This method tends to show little injection-to-injection variability because of its simplicity. 

The peak area errors for the two most studied de-convolution methods (i.e., perpendicular drop and linear tangential skim) are dependent on a complex combination of resolution, relative peak width, relative peak height, and asymmetry ratio [1]. Exponential skimming assumes that the tailing of the first peak can be described by an exponential decay and that the peaks are sufficiently resolved to determine the decay parameters. Nonetheless, some broad generalizations can be made ... 

Resolution is also used to compare the theoretical (calculated or predicted) mass value to the experimentally observed value. Resolution defines the peak width, and a parts per million (ppm) error value can be calculated with Eq. 15.8 this allows estimation of the precision of an experimental mass assignment ... 

The assumption for the measurement of peak widths in Eqs. (8)—(10) is that the peak is Gaussian. Unfortunately, few peaks are truly Gaussian and, in general, for asymmetrical peaks, N, calculated by a Gaussian-based equation, increases the higher up on the peak the width is measured. Figure 3 shows the error in plate-count determinations for asymmetrical peaks as a function of the peak height at which the width is measured. A more accurate approach to efficiency measurement has been presented by Foley and Dorsey (7), which takes into account the peak asymmetry. 

---