Chromatographic Quantitation Using Fractions of the Peak Areas Antonio L. Pires Valente*, Fabio Augusto, and Eduardo Carasek da Rocha zyxwvu Instituto de Quimica, Universidade Estadual de Campinas, 13083-970 Campinas, SP, Brazil Key Words: Chromatography Quantitative Analysis Computer calculation of peak areas Summary Quantitative chromatographic analysis is liable to errors due to peak asymmetry because the uncertainty in the detected position of the end of the peak tail decreases the reliability of the computed peak area. This dependence may be a severe drawback whenever peaks of different areas must be compared, asin the case of calibra- tion curves. A new approach to overcome the uncertainties of area calculation due to peak asymmetry is reported in this paper. The approachconsists of calculatingonly the area included between the start and the maximum of the chromatographic peak. Simulated and experimental chromatographicdata were used in this study. Both the peak start-to-peak maximum area (SMA) and the start- to-end or total area (TA) were calculated and the quantitative results were compared.Within the scope of this work it is concluded that the SMA yields calibration curves that are more linear and have intercepts closer to zero than the calibration curves obtained using the TA. 1 Introduction In the computer analysis of digitized chromatograms, the end of asymmetric peaks is anticipated and thus the corresponding peak areas are underestimated [l]; noise and noise reduction algo- rithms may contribute to augment this undesired experimental limitation. Procedures to solve or minimize errors in area assign- ment, such as the estimation of the area of the peak tail by fitting the data to an exponential function which is then algebraically integrated, have been suggested in the literature [2]. The use of the Exponentially Modified Gaussian (EMG) for the calculation of the peak area zyxwvutsrq - and other peak parameters - has also been proposed [3-51. However these approaches have not come into common usage. The approach proposed herein consists of simply not computing this troublesome part of the peak area. It may be implemented as a sub-routinein any chromatographic data analy- sis software that is user-modifiable. It is partially based on the fact that the allocation of the peak start and of the peak maximum is much less subject to computational errors than the allocation of the peak end [6]. The discussion of the peak start-to-peak maximum area (SMA) as compared to the total area (TA) is done with both simulated and experimental chromatographic data. The simulated tailed peaks are used to ascertain the potentiality of the SMA as substitute for the TA, without the interference of chro- matographic side effects such as noise and peak distortions due to the geometry of the detector cell. The simulation also allows the comparison of the software calculated SMA and TA with the known real area of the peak. Thus, by varying the size of the simulated peaks the accuracy of the SMA and of the TA may be evaluated from the parameters of calibration curves emulated in the form of SMA zyxwvuts vs Real Area and TA vs Real Area. For experi- mentally obtained data, the comparison of the SMA and TA calibration curves is also used. If a calibration curve of known behavior (the vast majority reported in the literature are linear) shows that behavior when obtained from SMA data then it may be concluded that, within the studied analyte concentration range, the relation of the SMA to the varying detected amount of analyte is consistent. Comparison of the intercepts of calibration curves obtained with SMA and TA is useful to ascertain if either of these parameters contribute to systematic quantitative errors. 2 Materials and Methods zyxw 2.1 Data Analysis Sojiiwaue The SMA were calculated from digitized chromatograms by a procedure incoiporated into the ANACROM chromatographic software package [6]. In ANACROM, the peak start, the peak maximum and the peak end are assigned after analysis of the derivative of the chromatographic signal. The peak area (SMA and TA) is then calculated as the sum of the sequentially digitized signals - the slices - that belong to the pre-defined boundary (peak start-to-peak end or peak start-to-peak maximum). The software was written in Pascal with theTurbo Pascal4.0 compiler (Borland Corp., Scotts Valley, CA) and runs on IBM-PC com- patible microcomputers. 2.2 Simulated Chromatographic Data Simulated chromatograms were constructed using the EMG function to generate asymmetric peaks [7]. The software-gener- ated chromatograms emulate real chromatograms collected with a 12bits A/D converter at an acquisition rate of 18 signal elements s . The peaks of these chromatograms differ in their Aslo (asym- metry at 10 % of height) and w0.1(widths at 10 % of height) and have theoretical areas ranging from 25,000 area units to 250,000 area units. -1 2.3 Experimental Data Chromatograms were obtained for l-octanol and l-nonanol dis- solved in isooctane. The range of injected masses were of 238 g to 1430 g for l-nonanol and of 287 g to 1724 g for l-octanol. Chromatograph: PU-104 (Pye-Unicam Ltd.) with FID. Column: 3 % SE-30 on 100/120 mesh Diatomite-C; borosilicate glass tubing, 1.5 m length, dl = 4 mm. Temperatures: Injector = 200 "C, Column Oven = zyxw 1 10 "C ( 1 -0ctanol) or 120 "C (1-nonanol) and Detector Oven = 200 "C. On-column injected volume: 1.0 L. Carrier gas (N2) = 25 mL min-' (l-octanol) or 30 mL min-' (l-nonanol). These chromatographic conditions were ad'usted to generate tailed peaks. Detector gases = 40 mL min-'Hz and 600 mL min-' air. Signal conditioning and collection: from a model Wide Range Amplifier (Pye-Unicam Ltd.) interfaced to an XT-2002 microcomputer (Microtec Sistemas, S.Paulo, Brazil) through a DACA 12 bits zyx A/D converter (IBM Corp.). The acqui- sition rate was of 18 signal e1ements.s-l. The digitized chroma- J. High Resol. Chromatogr. VOL. 18, MAY 1995 315