Charged Aerosol Detection: Factors for consideration in its use as a generic quantitative detector Ian Sinclair and Richard Gallagher - Global Compound Sciences/DMPK, AstraZeneca, Mereside, Alderley Park, Macclesfield, Cheshire, SK10 4TG. Charged Aerosol Detection (CAD) is now established as a universal detector for High Performance Liquid Chromatography (HPLC) applications, and a complimentary technique or alternative to Evaporative Light Scattering or Ultraviolet detection. CAD has already been used to quantify structurally diverse samples, providing that the sensitivity variation with changes in eluent composition was taken into consideration. Routine use of CAD in measuring solubilised drug standards resulted in a discrepancy in expected results and led to investigations into how mobile phase buffer could significantly affect quantification results. This study has shown the distinctive property of CAD in detecting the solution based ion pair of a drug substance and its adduct in a single analysis. Introduction Monitoring the concentration and quality of compounds in solution intended for physical or biological screening is of paramount importance within the pharmaceutical industry. Instrumentation that can produce quantifiable information, irrespective of the compound s physical or chemical properties, and can be readily incorporated into on-line systems is the ideal solution. Several techniques have appeared over the last few years that showed promise as the elusive universal detector. The need for universal detection methods for liquid chromatography led to the development of evaporative light scattering detection (ELSD) in the late 1970 and 1980s 1,2. This detection method is useful both for detection of compounds without a chromaphore and detection of compounds separated with gradient elution, yielding advantages over ultraviolet absorption detection and refractive index detection respectively 3. Within similar compound classes a constant molar response can be measured. This makes it ideal for the determination of polymer concentrations and latterly as a tool within small molecule library production as a means of measuring yield and purity 4. However, there are three main problems with using ELSD as a universal detector: as it is a particle technique, the changes in solvent composition across the gradient change the system s sensitivity and later running compounds will exhibit a disproportional response to similar concentration analytes eluted early in a gradient. Mobile phase compensation is one way round this see ref 6. Secondly, as the name suggests, this is an evaporative technique, hence volatile compounds cannot be fully detected and finally, although it gives an equimolar response for compounds within a class there can be an order of magnitude difference between compound classes. One of the more recently available universal detectors is the Charged Aerosol Detector (CAD) that has been shown to produce direct quantitation of compounds relative to a known standard 5-7. Charged Aerosol Detection involves first nebulizing the HPLC column eluent with nitrogen and then drying in a drift tube to produce analyte particles. A secondary flow of nitrogen becomes positively charged as it passes through a corona generated by high voltage. This charged nitrogen is then mixed with the stream of analyte particles where the charge migrates to the analyte. Charged analyte particles proceed to a collector region where a highly sensitive electrometer produces a signal that is proportional to the weight of sample present, independent of chemical structure 8. This results in an equal response for equal mass amounts, not equal molar amounts. A weakness of all detectors employing a nebulisation process is that the response is dependant on the composition of the mobile phase and increases with the organic solvent content 7 Isocratic HPLC analyses are unaffected by this phenomena
Figure 1 LC-MS system used for screening a liquid sample, including measuring concentration by Charged Aerosol detector. Figure 3: Traces collected from the Charge Aerosol Detector from the analysis of: (a) Standard drug in dimethyl sulphoxide (DMSO) solvent (b) Drug with a Hydrogen Bromide adduct. but gradient analyses have a non-uniform response. To overcome this shortcoming modifications to the flow conditions can be made 6,7,9. Gradient Effects To allow quantification to be performed on liquid samples using the CAD, the variation in sensitivity of the detector with mobile phase gradient has to be addressed. This variation is removed by adding an inverse gradient mobile phase to the eluent so that all samples entering the detector do so under the same mobile phase composition and hence no correction of the peak area is required 6,9. A photograph and schematic of the experimental layout is shown in figures 1 and 2. Figure 2: Schematic of HPLC system used. In figure 2, the pumps initiate a gradient simultaneously, with pump 2 supplying a gradient that is the exact inverse of the analytical gradient from pump 1. Using the inverse gradient also stabilises the baseline when high ph buffers are used, whereas in a traditional single gradient system the background level can increase to unacceptably high levels, masking the peaks of interest. Whilst more eluents are accessible when the inverse gradient method is used it must be pointed out that the supplies for the two gradient systems should be drawn from the same source, as slight differences in eluent make-up can disturb the baseline and may affect sample sensitivity across the gradient. When the sample injection solvent reaches the CAD detector a peak is observed. This is due to either a change in the baseline sensitivity as the slug of different solvent composition passes into the detector or because there is a non-volatile entity in the solvent front. The traces in figure 3 reflect these two scenarios. In the first chromatogram (a), the elemental analysis shows that only the drug is present in the sample, whereas in the second (b), elemental analysis shows the presence of hydrogen bromide as an adduct to the drug Assessing the Accuracy of the Detector In order to assess the accuracy of CAD as a detector some primary standards were analysed. Weighed standards may introduce errors so 1 H-Nuclear Magnetic Resonance (NMR) was used to determine the strength of these standards. NMR has been used for many years in the pharmaceutical industry to measure the concentration of compounds and is regarded as the Gold Standard for this application 4,10-12. A randomly selected sample of 29 drug like compounds (sample set 1) were submitted to NMR and the quantified solutions then analysed by CAD. These samples were dissolved in perdeuterated DMSO containing maleic acid (0.122mg/ml), as an internal standard. The CAD detector produced an area that was proportional to the absolute weight of sample loaded. A calibration factor was
Figure 4: Correlation between weights determined by NMR vs. calculated sample weight determined by CAD for sample set 1. applied to allow sample weight to be expressed. By assuming the first sample is correct a factor (K) was calculated and applied to all the other samples: CAD Area = CAD sample weight = NMR K weight for the first sample The predicted weights produced by the peaks of interest were plotted against the NMR values and the results can be seen in figure 4. Table 1 Assuming NMR strength is accurate then the CAD results show a standard deviation of 15% from the true weight. Additional Features of CAD detector In the work we have presented so far we have shown the generic abilities of this detector, but there are other factors to consider. We showed earlier that the CAD was not only able to detect and quantify the drug molecule but also the inorganic counter ions Sample Adduct ClogP Protein Charge state of Number Binding parent compound 1 HCl High High Cation 2 HCl High High Neutral 3 Citric acid High High Cation 4 HCl High High Zwitterion 5 None Low High Cation 6 HCl High High Cation 7 None Low High Neutral 8 None Low Low Cation 9 HCl Low High Zwitterion 10 Na Low Low Anion 11 HCl High Low Cation 12 HCl Low Low Cation 13 None Low Low Zwitterion 14 None High High Neutral 15 HBr Low Low Cation 16 None Low High Neutral 17 None High Low Anion 18 None n/a n/a Anion 19 None n/a n/a Neutral 20 None n/a n/a Neutral 21 None n/a n/a Cation 22 None n/a n/a Neutral Sample set 2 characteristics Figure 6: Variation in measured peak area for the sodium ion in three different mobile phases included in the sample make up. If we quantify the area at the solvent front, and subtract the small area produced by the DMSO, we should be able measure the amount of adduct present. A set of 22 compounds (sample set 2) was chosen for its coverage of physicochemical space with representatives of acids, bases, neutrals and zwitterions that contained a variety of characterised adducts with known stoichiometry, see list below: The peak areas obtained for sample set 2 were converted to analyte weight and the results showed an overestimation in the amount of adduct present (data not shown). When we substituted the ammonium acetate in the eluent system with either ammonium hydroxide or trifluoroacetic acid the drug areas remained constant with just a few notable exceptions (discussed later in this section). However, the adduct areas varied enormously or even disappeared. The difference in quantified adduct weight appears to be specific to not only the adduct involved but also the eluent make up. An example of this is when the drugs have a sodium counter ion, (figure 6). When measuring the area/weight ratio for the sodium ion a five-fold difference in the measured amount in the three systems was observed. However, if it is assumed that the ion exists in the form of sodium acetate, sodium trifluoroacetate and sodium hydroxide respectively in the three different systems, then a variation <2% is seen (figure 6). If the CAD system is an ion pair detector for inorganic molecules, then perhaps the same is true for the organic ones? The data in figure 7 represents seven of the molecules from the same set of compounds and shows the measured area/weight ratios in the three solvent systems. It also provides area:weight ratios of the drug:salt entities.
Daltons (Da). The CAD measures this mass change but this is not observed in the Mass Spectrometer. Sample 3 has two basic centres and likewise coordinate with two TFA anions under acid conditions. After taking this mass shift into account there is a good correlation with the ratio obtained for the basic and neutral eluents. Figure 7: CAD peak area/weight measurements for seven drugs taken from sample set 2 and measured in various mobile phase modifiers This data shows that unless the mobile phase composition and compound character are accounted for then an overestimation of the amount of compound present can be made. Limitations of CAD Figure 8: Measured concentration by CAD for a variety of drug compounds nominally at 10mM vs. their molecular weight Figure 9: Measured concentrations of a sample set separated by molecular weight above and below 300 Da Figure 10: CAD area versus Sample loading for four serially diluted standards As with the adduct molecules we have to consider the charge state of the drug molecules in each buffer system. Once the charge state is understood then the weight based on the expected ion pairing produced in solution can be calculated. From figure 7 we see that the neutral samples 7 and 19 are unaffected by the presence of the modifiers and the same area/weight ratios were obtained in all three buffer systems. Samples 2, 4, 9 and 21 show a large variation when analysed in the TFA system relative to the neutral and basic mobile phase. These compounds each have one basic centre and will therefore ion-pair with a single TFA anion in the acid mobile phase effectively increasing their detected mass by 114 Compound volatility plays an important part in the detector s abilities. As the volatility of every sample submitted to the system cannot be determined we have been able to show where the boundary lies between good generic detection and sample specific detection. The data in figure 8, for a set of nominal 10mM solutions, shows a significant difference in measured concentrations above and below 300 Da. Separation of this data into samples above and below the 300 Da boundary produced the following distributions (fig 9). The wide variation in the measured concentration for samples below 300 Da shows a loss of generic ability and precludes the data from being used to determine concentration. The detector is still useful for relative purity measurements in this region and the detection of compounds with little or no UV absorbance. Above 300 Da the data exhibits a much tighter distribution. The linearity of the detector was tested using serial dilutions of four randomly selected samples from sample set 1, and as can be seen in figure 10, over this range the detector is not linear. However, over shorter regions the output can be considered linear with sufficient accuracy for this application. Conclusion Charged Aerosol Detection is a powerful tool in the analysis of diverse compound collections. It has been proved to be simple to integrate and operate within a high throughput analytical environment and can
provide evidence of the contents of the original solid material and indicate compound concentrations. The role envisaged for the CAD detector is in concentration measurement for samples from the liquid screening collection. These samples typically have a MW distribution between 300-600 Da and are nominally at a concentration of 10 mm. Providing a pass/fail status on each sample will not require a large linear range on the detector output. When using CAD as a generic detector it is advisable to choose a buffer that contains the smallest possible counter ions (i.e. NH 4 OH or formic acid). Using this method to analyse typical drug-like molecules in the MW range of 350-450 Da the largest errors expected from unpredicted ion pairing would be 5% from NH 4 OH and 13% from formic acid buffers. Ion pairing allows CAD detection to be extended to volatile acids and bases, but an understanding of each compound s state in solution is required. Acknowledgements With thanks to Isabel Charles for reading and encouragement on production of this document and to Dave Temesi for compiling the test samples. References 1. JM Charleworth, Anal Chem 50 (1978) 1414-1420 2. K Petritis, I Gillaizeau, C Elfakir, M Dreux, A Petit, N Brongibault, W Luitjen, J Sep Sci, 25 (2002) 593-600 3. K Petritis, C Elfakir, M Dreux, J Chrom A, 961 (2002) 9-21 4. S Lane, B Boughtflower, I Mutton, C Paterson, D Farrant, N Taylor, Z Blaxill, C Carmody, P Borman, Analytical Chemistry, 77-14, (2005) 4354-4365 5. RW Dixon, DS Peterson, Analytical Chemistry, 74 (2002) 2930-2937 6. T Gorecki, F Lynen, R Szucs, P Sandra, Analytical Chemistry, 78-9 (2006) 3186-3192 7. R Gallagher, E Goodall, 29th Symposium on High Performance Liquid Phase Separations and Related Techniques, (2005) Stockholm, Sweden 8. P Gamache, R McCarthy, S Freeto, D Asa, M Woodcock, K Laws, R Cole, LC-GC Europe, 18-6 (2005) 345-346 352-354. 9. A de Villiers, T Gorecki, F Lynen, R Szucs, P Sandra, Journal of Chromatography, A 1161-1-2 (2007) 183-191 10. R Wells, J Hook, T. Al-Deen, D Hibbert, J. Agric. Food Chem. 50 (2002) 3366 3374. 11. G Maniara, K Rajamoorthi, S Rajan, G Stokton, Anal. Chem. 70 (1998) 4921 4928. 12. F Malz, H Jancke, J.Pharm and Biomedical Analysis, 38 (2005) 813 I started using the Corona CAD for ion measurements then turned off my Ion Chromatography system for good! Simultaneous Measurement of Anions and Cations Use your standard HPLC system and Corona CAD Measure counter ions (anions, cations, and even the API) at the same time without the need for an ion chromatography system Cost effective: no special equipment, no suppressor to replace, no special operator qualifications Robust and easy to use, the Corona CAD is ideally suited for both the methods-development lab and the factory floor O N E I N J E C T I O N O N E S Y S T E M O N E R U N To learn more, visit www.esainc.com/ion_analysis ESA Biosciences, Inc 22 Alpha Road Chelmsford, MA 01824 USA +1 978.250.7000 Ph +1 978.250.7090 Fax +1 800.959.5095 U.S. Toll free ESA Analytical, Ltd. Dorton, Aylesbury Buckinghamshire HP18 9NH England +44 (0)1844 239381 Ph +44 (0)1844 239382 Fax