Universal Response in Liquid Chromatography Using Charged

The detector is characterized by a nearly universal response at a given, constant mobile-phase composition for sufficiently nonvolatile analytes. A se...
0 downloads 0 Views 128KB Size
Anal. Chem. 2006, 78, 3186-3192

Universal Response in Liquid Chromatography Using Charged Aerosol Detection Tadeusz Go´recki,†,‡ Frederic Lynen,† Roman Szucs,§ and Pat Sandra*,†

Pfizer Analytical Research Centre, Ghent University, Krijgslaan 281 S4-bis, B-9000 Ghent, Belgium, and Analytical R&D, Pfizer Limited, Ramsgate Road, Sandwich, Kent, CT13 9NJ, United Kingdom

A new, empirical approach is introduced to correct for the varying response of aerosol-based detectors with the varying composition of the mobile phase during gradient elution in HPLC. A Corona charged aerosol detector was used in the experiments. The detector is characterized by a nearly universal response at a given, constant mobilephase composition for sufficiently nonvolatile analytes. A second pump was used to deliver an exactly inverse gradient compared to the analytical HPLC system, and both flows were mixed in a tee piece before introduction to the Corona detector. The approach proposed made it possible to extend the universal response from isocratic to gradient elution conditions in HPLC, vastly improving the usefulness of this detection technique. The constant response of the detector obtained in this way was first demonstrated in flow injection analysis. Very similar calibration curves were obtained for six sulfonamide drugs after mobile-phase compensation. The approach was also applied to gradient elution with excellent results. The data were characterized by good precision ranging from 4% RSD at 10 mg/L to 1.6% RSD at 780 mg/L. The average limit of detection with a 2-µL injection was 0.5 mg/L, corresponding to 1 ng injected on the column. The approach proposed allows quantification of unknown compounds, e.g., in pharmaceutical mixtures. Measurement of analytes at a relative concentration of 0.05% versus the main component is demonstrated. The need for ever faster analysis in the pharmaceutical and other industries leads to less time available for method development. This results in an increasing interest in the use of generic techniques. One of the reasons why the development of generally applicable methods for HPLC proves challenging is the lack of sensitive and universal detection techniques suitable for both qualitative and quantitative analysis under a wide range of conditions. Several detectors partially fulfilling these needs are currently employed. They include refractive index (RI), lowwavelength UV,1 chemiluminescent nitrogen,2 mass spectrometry (MS), flame ionization,3 and evaporative light scattering (ELSD)4 * To whom correspondence should be addressed. E-mail: pat.sandra@ richrom.com. † Ghent University. ‡ On sabbatical leave from: Department of Chemistry, University of Waterloo, Waterloo, ON, N2L 3G1 Canada. § Pfizer Limited. (1) Collins, A. J. J. Chromatogr. 1986, 354, 459-462.

3186 Analytical Chemistry, Vol. 78, No. 9, May 1, 2006

detectors. In contrast to specific detectors such as UV and fluorescence, the above detectors respond to bulk or generic properties of the analytes. Mass spectrometry is considered a specific and universal detection method, but quantitative analysis with LC-MS is currently less robust as the response depends on the ionization process. RI detection is widely used as a universal technique but suffers from limited sensitivity and cannot be used in gradient elution analysis. Low-wavelength UV detection is sensitive for many analytes in HPLC, but its sensitivity is dependent on the molecular structures of the analytes; in addition, several mobile-phase constituents exhibit high UV cutoffs. The chemiluminescent nitrogen detector for HPLC is sensitive for most pharmaceutical compounds (as long as they contain nitrogen) but is limited in the choice of mobile phases (for example, acetonitrile modifier or amine additives cannot be used). The use of FID with HPLC has been described, but without mechanical interfaces is limited to the analysis of volatile compounds, and the results are often poorly reproducible. The evaporative light scattering detector is increasingly used as a universal detector in combination with UV. The detector is more sensitive than the RI detector and can be used in combination with gradient elution. However, the dayto-day reproducibility and the precision of the results obtained with ELSD are not very good, which leads to the need for regular recalibration. An advantage of this aerosol-based detector is that it has an amount (mass)-dependent response that is substantially more uniform than that in either UV or MS.5,6 Recently, a new type of detector, the Corona charged aerosol detector (CAD), has been introduced. In the CAD, aerosol particles are charged with an ionized gas (typically nitrogen). After the removal of high-mobility particles (mainly excess N2 ions), the aerosol particles are then electrically measured.7 The detector is claimed to be more sensitive than the ELSD, shows a dynamic range of up to 4 orders of magnitude, and requires little or no human intervention. Most importantly, it has been demonstrated to provide a uniform response to nonvolatile analytes independently of their nature.7-9 However, since the CAD is an aerosolbased detector, it is subject to the same limitations as the ELSD. (2) Fujinari, E. M.; Courthaudon, L. O. J. Chromatogr., A 1992, 592, 209214. (3) Hooijschuur, E. W. J.; Kientz, C. E.; Brinkman, U. A. J. High Resolut. Chromatogr. 2000, 23, 309-316. (4) Ford, D. L.; Kennard, W. J. Oil Colour Chem. Assoc. 1966, 49, 299-313. (5) Charlesworth, J. M. Anal. Chem. 1978, 50, 1414-1420. (6) Mourney, T. H.; Oppenheimer, L. E. Anal. Chem. 1984, 56, 2427-2434. (7) Gamache, P. H.; McCarthy, R. S.; Freeto, S. M.; Asa, D. J.; Woodcock, M. J.; Laws, K.; Cole, R. O. LC-GC Eur. 2005, 18, 345-354. 10.1021/ac060078j CCC: $33.50

© 2006 American Chemical Society Published on Web 03/28/2006

The main one is the fact that the response of the detector varies as a function of the mobile-phase composition. An increase in the organic content of the mobile phase leads to an increase in the transport efficiency of the nebulizer.10 This results in a greater number of particles reaching the detector chamber and in a higher signal. Mathews et al.11 corrected for this phenomenon in ELSD by developing a generic calibration routine. A 3-D calibration surface was constructed including the varying response with the mobile-phase composition, as opposed to the 2-D calibration normally used. The main drawbacks of this approach are the large number of analyses that must be performed for calibration and the fact that the results are method-specific. In this contribution, an empirical and more universal approach is demonstrated to solve this problem. The principle is to provide the detector at all times with a constant composition of the mobile phase. The eluent from the column is therefore mixed before the detector with the eluent from a second pumping system that produces an exactly opposite gradient. The combination of this approach with the CAD provides a powerful tool for fast quantitative or semiquantitative analysis of analytes, for which, for example, no pure standards are available. In this way, calibration and analysis can be done with a limited number of analyses and a universal response is approached. EXPERIMENTAL SECTION Materials. Water, acetonitrile (ACN), and formic acid used to prepare the mobile phases were all LC-MS grade from Biosolve (HA Valkenswaard, The Netherlands). Sulfamethoxazole, sulfadmiethoxin, sulfamerazin, and sulfamethizol from RiedeldeHae¨n, were delivered by Sigma-Aldrich (Seelze, Germany). Sulfamethazine and sulfaguanidyne were from Sigma-Aldrich Chemie, GmbH. Stock solutions of the individual sulfonamides (5000 and 500 mg/L) were prepared in methanol (Biosolve) and stored in a refrigerator. New stock solutions were prepared weekly, with the exception of sulfadmiethoxin, for which they were prepared twice a week because of instability. Standard solutions were prepared fresh daily by appropriate dilutions of the stock solutions in water/acetonitrile (90/10). Nitrogen 4.0 (Messner, Mechelen, Belgium) was used as the makeup gas for the Corona detector. A new cylinder was connected whenever pressure in a cylinder dropped below 20 bar. Instrumentation. Two HPLC 1100 Series systems from Agilent Technologies (Waldbronn, Germany) were used in the research. Both systems were equipped with diode array detectors (DAD) and autosamplers. The first system was equipped with a single binary pump, while the second system was equipped with two binary pumps, both controlled from within the same software (Chemstation). In the flow injection experiments with mobile(8) Gallaghar, R. T.; Goodall, E. Evaluation of charged aerosol detection for use as a relative or absolute purity indicator. 29th International Symposium on High Performance Liquid-Phase Separations and Related Techniques, Stockholm, Sweden, June 26-30, 2005. (9) Christensen, J.; Goodal, E. HPLC with charged aerosol detection for the measurement of different lipid classes. 29th International Symposium on High Performance Liquid-Phase Separations and Related Techniques, Stockholm, Sweden, June 26-30, 2005. (10) Cobb, Z.; Barret, D.; Shaw, P.; Meehan, E.; Watkins, J.; O’Donohue, S.; Wrench, N. J. Microcolumn Sep. 2001, 13, 169-175. (11) Mathews, B. T.; Higginson, P. D.; Lyons, R.; Mitchell, J. C.; Sach, N. W.; Snowden, M. J.; Taylor, M. R.; Wright, A. G. Chromatographia 2004, 60, 625-633.

phase compensation, a Gilson pump model 307 (Villiers Le Bel, France) was used with premixed mobile phases. PEEK tubing of 0.127-mm i.d. was used for all connections after the DAD detector. The Corona CAD was from ESA Analytical (Aylsbury, Buckinghamshire, England). In all experiments, this detector was connected in series after the diode array UV detector of the 1100 Series HPLC. Data acquisition for the Corona detector was carried out using Peak Simple Chromatography Data System model 202 and Peak Simple software (both from SRI Instruments, Torrance, CA). Data acquisition was triggered automatically using the start signal from the LC autosampler and finished after a predetermined time. Methods. Flow Injection Analysis. Experiments were performed by injecting standard solutions of the analytes directly into the transfer line leading to the DAD/Corona detectors. Injection volume was 2 µL in each case. Each injection was carried out in triplicate. Experiments were performed for five different compositions of the mobile phase: 10, 30, 50, 70, and 90% ACN in water with 0.1% formic acid. Mobile-phase flow rate was 1 mL/min. UV detection was carried out at 254 nm. Analyte solutions at five different concentration levels were injected: 2, 10, 50, 500, and 1000 mg/L. The mobile-phase compensation procedure is described in the Results and Discussion section. HPLC Analysis. The 1100 Series HPLC equipped with two pumps was used in the experiments. Separation was carried out on a Zorbax SB-C18 column, 4.6 mm i.d. × 150 mm L, packed with 3.5-µm particles (Agilent Technologies). The calibration curves were determined by analyzing standard solutions of the analytes at six concentration levels (2, 10, 50, 100, 500, and 780 mg/L). Each analysis was carried out in triplicate, and average values were taken for the calibration. Injection volume was 2 µL. The following mobile-phase gradient was used: 10% B for 2 min, increased to 90% B between 2 and 15 min, held at 90% B for 3 min, and reduced back to 10% B at 19 min (where B is ACN, and A is 0.1% HCOOH in water). Mobile-phase flow rate was 1 mL/min. The column was equilibrated for 15 min after each run. The procedure for HPLC analysis with mobile-phase compensation is described in the Results and Discussion section. The limits of detection were determined according to the EPArecommended procedure.12 A standard solution containing each analyte at a 2 mg/L level (estimated S/N ) 10) was injected seven times, and the standard deviations of the results were calculated. The limits of detection were calculated by multiplying the standard deviations by 3. RESULTS AND DISCUSSION The Corona charged aerosol detector is characterized by a uniform response toward aerosol-forming compounds with low vapor pressures.7-9 However, the magnitude of the response depends on the organic content of the mobile phase, with higher response observed at higher organic concentrations. Initial experiments were performed to determine the response of the detector toward six sulfonamide drugs: sulfamethoxazole, sulfadimethoxine, sulfamerazine, sulfamethizole, sulfamethazine, and sulfaguanidine. The drugs were selected because they are characterized by very low vapor pressures at room temperature (on the order of 10-6 Pa) and they cover a wide range of polarities, owing to which they elute from C18 columns over a broad range of mobile-phase Analytical Chemistry, Vol. 78, No. 9, May 1, 2006

3187

Figure 1. Averaged calibration curves obtained for all six analytes at different compositions of the mobile phase in flow injection analysis. (A) linear coordinates; (B) logarithmic coordinates, including power regression lines. Error bars show one standard deviation.

compositions. The response factors of the analytes at mobile-phase compositions selected for the study (from 10 to 90% ACN in water with 0.1% HCOOH) were determined through flow injection analysis, since isocratic separations at those mobile-phase compositions were impractical or impossible. In a study of response factors of the ELSD, Matthews et al.11 repeatedly injected a nonretained compound (5-fluorocytosine) into the column during gradient elution, and the magnitude of the response recorded versus time (thus also mobile-phase composition) was later used to calculate the concentrations of unknown analytes during gradient elution. The main limitation of such an approach is the fact that the response factor is determined based on a single calibrant, which might lead to errors if the detector response to the analyte is significantly different from the response toward the calibrant. Flow injection used in this study allowed the determination of the response factors of six compounds at five different mobile-phase compositions. Consequently, the average response factor for those compounds could be used instead of any of the individual response factors. Flow Injection Analysis. To perform the study, each compound was injected three times at each concentration level and mobile-phase composition directly to the transfer line leading to the diode array detector. Figure 1 presents the averaged calibration curves for all six compounds at five different compositions 3188

Analytical Chemistry, Vol. 78, No. 9, May 1, 2006

of the mobile phase. Error bars ((1 standard deviation) were included in this figure to illustrate the range of uncertainty of the results. Very strong dependence of the detector response on the mobile-phase composition is evident in this figure, with peak areas at 90% ACN being nearly five times greater than at 10% ACN. The increase was not linearsgreater differences were observed when going from low to medium organic content in the mobile phase (10-50% ACN) than when going from medium to high concentration (50-90% ACN). As expected, the response of the Corona detector was not linear (Figure 1A), but reasonably good linearity was obtained when the curves were plotted in a logarithmic coordinate system (Figure 1B). The lines connecting the experimental data points in Figure 1A were added to help visualize the course of the curves. Lines in Figure 1B are the lines of best fit for power regression in the form y ) axb. The regression lines in Figure 1B fit the data well, with the correlation coefficients exceeding 0.99 in all cases except 90% ACN, for which the correlation coefficient was 0.954. The greatest relative uncertainty was observed for the 2 mg/L standard. This was related to significant signals observed for the blanks and the necessity of blank correction by subtraction of the blank peak areas from the analyte peak areas. The blank signal was most likely related to trace impurities in the solvents, even though they were of the highest grade available. Blank correction had the greatest implica-

Figure 2. Calibration curves obtained for all six analytes using a mobile phase consisting of 50% ACN in water with 0.1% HCOOH; flow injection analysis, logarithmic coordinates.

tions for the lowest standard (2 mg/L), which was close to the limits of detection of the method with 2-µL injection; for the 50 mg/L standard and above, the correction was insignificant. It should be emphasized that blank correction would not be required if chromatographic separation was involved. Figure 2 shows an example of a set of calibration curves obtained for all six analytes for a fixed composition of the mobile phase (50% ACN in water with 0.1% HCOOH). The figure illustrates that, for a given composition of a mobile phase, the response factors of the Corona detector were very close independently of the compound. Similar results were obtained for other mobile-phase compositions studied. The relatively large spread of results observed for the 2 mg/L standard was related to the low analyte levels and significant blank signals. Flow Injection Analysis with Mobile-Phase Compensation. The strong dependence of the response factors on the composition of the mobile phase is certainly an undesirable characteristic of the Corona detector. The organic content of the mobile phase affects the transport efficiency of the nebulizer and the particle size distribution of the droplets.11 While it is possible to correct the results for this variability using computational methods, as was done in ref 11 for ELSD, it would be more practical to eliminate the source of the variability in the first place. A straightforward way to do it is to make sure that the mobile-phase composition reaching the detector remains always constant. In flow injection experiments, this could be done by combining the flow of the mobile phase into which the injection was made with an additional stream of a mobile phase of reverse composition delivered at the same flow rate. For example, if the injection was made into a mobile phase containing 10% ACN, the secondary mobile-phase stream would contain 90% ACN. In this way, the mobile phase reaching the detector would always contain 50% ACN and would be delivered at a flow rate equal to double the flow rate of the primary stream (2 mL/min in this case). In the experiments, the two streams of the mobile phase were combined before the Corona detector, but after the DAD. The second stream was supplied by an additional pump using premixed mobile phases. Mixing took place in a PEEK tee. The lines from the two pumps were connected to the opposite arms of the tee, and the inlet line for the Corona detector was connected to the perpendicular arm.

Figure 3. Averaged calibration curves obtained for all six analytes at different compositions of the mobile phase using mobile-phase compensation in flow injection analysis. (A) linear coordinates; (B) logarithmic coordinates. Error bars show one standard deviation.

Figure 3 illustrates the results of this experiment. It shows the averaged calibration curves obtained for all six analytes at different compositions of the mobile phase with “mobile-phase compensation” using the technique described above. The method proposed worked very efficiently and assured uniform response factors independently of the mobile-phase composition. This is particularly evident when comparing the calibration curves in Figures 1A (without compensation) and Figure 3A (with compensation). The sensitivity of the method was very similar to the sensitivity obtained with 50% ACN in the mobile phase without compensation (see Figure 1A, 50% ACN line). While somewhat higher sensitivity could be achieved at higher organic content in the mobile phase, the difference was not very significant and certainly did not outweigh the benefits of having similar response factors for all analytes at all mobile-phase compositions. The sensitivity did not suffer even though the concentration of the analyte in the combined mobile-phase stream reaching the detector was cut by half compared to the original concentration because the analyte mass reaching the detector in unit time remained unchanged, and the Corona is a mass-sensitive detector. The higher spread of results observed for the lowest concentration standard (2 mg/L) was related to the blank signals, as described above. HPLC Analysis. The mobile-phase compensation technique used with flow injection analysis could in principle be applied without any modifications to isocratic HPLC separations. However, most HPLC separations are carried out under gradient elution conditions. This leads to problems with quantification of unknown analytes with the Corona detector, because the mobile-phase Analytical Chemistry, Vol. 78, No. 9, May 1, 2006

3189

Figure 4. Chromatograms with superimposed organic modifier composition obtained for the separation of the six sulfonamides. (A) UV signal in normal gradient run; (B) CAD signal without mobile-phase compensation in a gradient run; (C) CAD signal with mobile-phase gradient compensation. Peak identification: 1, sulfaguanidine; 2, sulfamerazin; 3, sulfamethazine; 4, sulfamethizole; 5, sulfamethoxazole; 6, sulfadimethoxin.

composition changes throughout the run and so do the response factors of the analytes. The problem is illustrated in Figure 4A and B. Figure 4A shows a chromatogram of a mixture of the six analytes (100 mg/L each) obtained during gradient elution with UV detection at 254 nm. Even though the concentrations of all the analytes were the same, peak areas were slightly different because of the different absorptivities of the analytes. Figure 4B shows the chromatogram of the same mixture obtained with the Corona detector. It is evident that the response factors increased with increasing concentration of ACN in the mobile phase, which gave rise to the characteristic sloped profile of peak apexes. This in itself is not an issue if the identity of the analytes is known and pure standards are available. However, it becomes a serious limitation when the identities of the peaks are unknown, as for example in the analysis of impurities in pharmaceutical preparations at the drug development stage. Combined experimental-computational approaches as described in ref 11 could be used to overcome this limitation; however, this approach requires new sets of experiments to be carried out every time the gradient is changed. Elimination of the root cause of the variability of the response factor, changes in mobile-phase composition reaching the Corona detector, seems to be a more practical approach. 3190

Analytical Chemistry, Vol. 78, No. 9, May 1, 2006

Mobile-phase compensation in gradient elution can be done by combining the effluent from the analytical column with a mobile phase whose composition is modified throughout the run according to a gradient that is the reverse of the gradient used for the analytical separation. For example, if the separation starts with 10% ACN in the mobile phase, and ACN concentration increases to 90% between 2 and 15 min, the mobile phase used for Corona signal compensation should contain 90% ACN for the first 2 min, and the ACN concentration should decrease from 90 to 10% between 2 and 15 min. In order for the approach described to be successful, it is critical that the composition of the compensation mobile phase is synchronized with the composition of the separation mobile phase at the point of mixing. This in itself might pose a challenge, because there is always a delay between the instant in which the instruction to change the mobile-phase composition is sent to the system and the time at which the new composition reaches the detector. The delay depends on many factors, including the method of mixing of the mobile-phase components (low- vs highpressure side), the volume of the mixer, the geometry of the column, the length and diameter of the connecting tubing, etc. To account for the delay, we decided to include similar components in both channels. Thus, the outlet of the second HPLC pump was fed to an identical HPLC column, and the length of the connecting tubing was adjusted to be as close as possible to that in the analytical channel (including the injector). The end of the outlet tubing was connected to the mixing tee as in the flow injection experiments. Figure 4C shows the chromatogram of the sulfonamide mixture obtained under conditions identical to those in Figure 4B except that mobile-phase gradient compensation was used. It is evident from this figure that the approach proposed produced the expected results. While the responses of the individual analytes in the conventional run (Figure 4B) increased with increasing ACN content in the mobile phase, they remained practically constant in the run with compensation (Figure 4C). Equally importantly, in this run, all responses were brought to the highest level recorded during the conventional run; therefore, increased sensitivity was obtained for all but the latest eluting analyte. Panels A and B of Figure 5 illustrate the calibration curves determined for the six analytes using the Corona detector in the conventional fashion. The effect of the mobile-phase composition on the response factors is clear in this figure. One should also keep in mind that the calibration would have to be repeated if the timing of the gradient was changed, which could result in the analytes reaching the detector at somewhat different mobilephase compositions. Panels C and D of Figure 5 illustrate the calibration curves obtained under identical conditions with mobilephase gradient compensation. The responses of all analytes were very similar in this case and would not change if the timing of the gradient was changed simultaneously in both channels. With mobile-phase gradient compensation, the calibration would have to be redone only if the average concentration of the organic component of the mobile phase in the gradient had changed (i.e., if the final ACN concentration was reduced/ increased or the initial concentration was increased/reduced). An additional advantage of this approach is that a single calibrant

Figure 5. Calibration curves obtained for all six analytes by HPLC in gradient analysis. (A) Without compensation, linear coordinates; (B) without compensation, logarithmic coordinates; (C) with mobile-phase gradient compensation; linear coordinates; (D) with mobile-phase gradient compensation, logarithmic coordinates.

(or an average of a few calibrants) can be used to quantify all analytes that show up during the run, including the unidentified ones. This feature is extremely important in some areas, including the pharmaceutical industry. The reproducibility of the results was examined based on triplicate injections of the standard mixtures at the six concentration levels selected. Overall, the reproducibility was excellent, with RSDs ranging from 4.0% at the10 mg/L level to 1.6% at 780 mg/L (higher concentrations could not be examined because of limited solubility of sulfamerazine). The RSD was higher at 2 mg/L (∼14%), but that was expected considering that 2 mg/L was close to the limit of detection of the method with a 2-µL injection. The limits of detection (LODs) with mobile-phase gradient compensation were estimated according to the EPA-recommended procedure.12 Peak heights were considered rather than peak areas in the determination, as they determine the signal-to-noise ratio. The LODs were equal to 0.5, 0.7, 0.4, 0.6, 0.3, and 0.7 mg/L for sulfaguanidyne, sulfamerazin, sulfamethazine, sulfamethizol, sulfamethoxazole, and sulfadmiethoxin, respectively. The average LOD for the six compounds was 0.5 mg/L. Considering that the injection volume was 2 µL, this is equivalent to the detection of 1 ng of a compound on column, which is an excellent result for a quasi-universal detector. In the pharmaceutical industry, it is critical that impurities can be detected at a level corresponding to 0.05% of the concentration of the main component. In this application, it is especially important that the response factors of the detector toward all compounds are similar, otherwise it is impossible to properly estimate the concentration levels of unidentified impurities. No

single detector thus far fulfills the requirements of this application; therefore, various combinations of detectors are used instead. For example, Lane et al.13 compared the performance of ELSD, nitrogen chemiluminescence, and proton NMR for universal detection. The Corona detector with mobile-phase gradient compensation seems to be an excellent candidate for such a detector provided that the analytes are nonvolatile. Figure 6 presents two chromatograms of standard mixtures of sulfadimethoxine in which the remaining analytes were present at 0.05% level. Figure 6A shows a chromatogram of a standard solution in which sulfadimethoxine was present at 10 000 mg/L, and the remaining analytes were present at 5 mg/L level. The injection volume was 2 µL. The unmarked peaks were most likely unidentified sulfadimethoxine impurities. All analytes could be easily quantified at this level (the limit of quantitation was 1.5 mg/L in this case). Figure 6B presents a chromatogram of a 2000 mg/L standard solution of sulfadimethoxine, in which the remaining analytes were present at 1 mg/L level. The injection volume in this case was 5 µL. The result was very similar to that obtained for a 2-µL injection of a more concentrated solution. Also in this case the impurities could be quantified at the required level. While the results presented in the paper are extremely promising, one should keep in mind that the Corona detector does not respond to volatile compounds, and the response to compounds of intermediate volatility can be inconsistent. For example, in preliminary flow injection experiments, butyl paraben showed inconsistent responses that were generally lower than expected and varied from experiment to experiment, even though its vapor pressure at room temperature is on the order of 0.01 Pa. Thus,

(12) Longbottom, J. E., Lichtenberg, J. J., Eds. Methods for Organic Chemical Analysis in Industrial Wastewater, EPA-600/4-82-057; U.S. EPA, Environmental Monitoring and Support Laboratory: Cincinnati, OH, 1982.

(13) Lane, S.; Boughtflower, B.; Mutton, I.; Paterson, C.; Farrant, D.; Taylor, N.; Blaxill, Z.; Carmody, C.; Borman, P. Anal. Chem. 2005, 77, 43544365.

Analytical Chemistry, Vol. 78, No. 9, May 1, 2006

3191

Figure 6. Chromatograms obtained for the 0.05% level analysis. (A) A 2-µL injection of 10 000 mg/L sulfadimethoxine standard solution containing 5 mg/L (each) of the remaining five sulfonamides; (B) 5-µL injection of 2000 mg/L sulfadimethoxine standard solution containing 1 mg/L (each) of the remaining five sulfonamides; peak identifications see Figure 5.

the Corona detector should probably be used in addition to other detectors rather than instead of them. CONCLUSIONS The quest for universal detection with uniform response factors for all analytes is an ongoing challenge. Thus far, no ideal solution has been found, and detector combinations have to be used when assessing unidentified impurities. The recently introduced Corona charged aerosol detector responds to all nonvolatile species, offers good sensitivity and a quasi-linear response, but shares a common drawback with ELSD in that its response depends on the organic content of the mobile phase. The method presented in the paper,

3192

Analytical Chemistry, Vol. 78, No. 9, May 1, 2006

mobile-phase gradient compensation, eliminates this drawback through normalization of the mobile-phase composition before the detector. This is accomplished by mixing the mobile phase leaving the column with a secondary mobile-phase stream of the opposite composition. This extremely simple approach is equally effective in flow injection experiments (and by extension in isocratic separations) and in separations involving gradient elution. As long as the compounds are nonvolatile, the response factors obtained on the Corona detector are known to be uniform independently of their nature,7-9 which opens the door to quantitation of unidentified species or single-compound calibration. Precision of the results and limits of detection are very good. The same approach would most likely be applicable to evaporative light scattering detectors, because the same mechanism is responsible for varying responses in both detectors. Evaporative detectors seem to be particularly well suited to this technique, as they do not require perfect mixing ahead of the detectorsthe mobile phase evaporates before the detection takes place anyway; therefore, only its average composition is important. The technique could also make it possible to use refractive index detectors in gradient elution, but in this case, perfect mixing would have to be accomplished before the mixed mobile phase enters the detector. In general, it seems that the Corona charged aerosol detector with mobile-phase gradient compensation could be an excellent addition to the toolbox of analytical chemists requiring universal HPLC detectors with uniform response for all analytes. The detector should be used in addition to rather then instead of other detectors, however, as it does not respond to all species. ACKNOWLEDGMENT We thank ESA Analytical (Aylsbury, Buckinghamshire, England) for a loan of the Corona charged aerosol detector. Received for review January 11, 2006. Accepted March 6, 2006. AC060078J