Improved optimization of polycyclic aromatic hydrocarbons (PAHs) mixtures resolution in reversed-phase high-performance liquid chromatography by using factorial design and response surface methodology
Résumé
Optimization of reversed-high performance liquid chromatography for separation of polycyclic aromatic hydrocarbons (PAHs) mixtures is frequently based on resolution and efficiency criteria. However, when very complex mixtures are analysed (such as diesel or kerosene soot) optimization must be made up based on resolution criterion rather than on efficiency. In fact, low efficiency is recommended in these cases because unexpected compounds not included in synthetic mixtures can appear in real samples.
Based on resolution criterion, optimization process for a 16 EPA PAHs mixture was performed on three sets of difficult-to-separate PAHs pairs: acenaphthene–fluorene, benzo[g,h,i]perylene–dibenzo(ah)anthracene and benzo(ghi)perylene-indeno(123cd)pyrene. Resolution of acenaphthene-fluorene pair was used for evaluation of the first part of the chromatogram, and evaluation of the second one was carried out by using the resolution of the pairs dibenzo(ah)anthracene-benzo(ghi)perylene and benzo(ghi)perylene-indeno(123cd)pyrene.
Two-level full factorial designs were applied to detect interactions among variables to be optimized: speed flow, temperature of column oven and ramp in both parts of the chromatogram. Experimental data were fitted by multivariate nonlinear regression models. Optimum values of speed flow and temperature were obtained through mathematical analysis of the constructed models.
For good resolution of acenaphthene-fluorene 10°C (column oven temperature) and 1.0 mL/min (mobile phase flow rate) are recommended but 40°C and 0.5 mL/min provides us with the best resolution for the pairs dibenzo(ah)anthracene-benzo(ghi)perylene and benzo(ghi)perylene-indeno(123-cd)pyrene. We have found that resolution in the second part of RPLC chromatogram does not depend on conditions used in the first part of the chromatogram. Thus, good resolution was achieved when operating at 1.0 mL/min in the first part of the chromatogram and 0.5 mL/min in the second one. Programmed temperature gradient (10°C until 30 minutes and progressively increasing temperature until reaching 40°C at 45 minutes) was used.
Due to the difficulty of achieving 10°C in the column oven and high pressures developed inside the column when so low temperatures are selected, combination of 15°C (in the first part of the chromatogram) with 40°C (the last 45 min of the chromatogram) were finally selected. Speed flow increases from 0.5 mL/min in the first part of the chromatogram to 1 mL/min after 45 minute. Good sensitivities were also achieved.