Simultaneous dual capillary column headspace GC with flame ionization confirmation and quantification according to USP <467> Application Note Joseph M. Levy Michael Kraft Abstract Agilent Equipment 7890A GC system G1888 headspace sampler Application Area Pharmaceutical quality control Pharmaceutical development Residual solvents analysis The Application Note describes the implications for laboratories of the latest revision of the United States Pharmacopoeia (USP) residual solvents analysis method that will take effect on July 1, 2008. Laboratories will eventually face the task of performing confirmation analysis with their gas chromatography (GC) systems for the various target compounds. In this note a GC method based on headspace sampling is presented, which meets the requirements of the latest USP revision and provides the high precision and sensitivity necessary for quantitative confirmation analysis of residual solvents in pharmaceutical products. The method uses an Agilent 7890A GC system with the following components: Agilent headspace sampler (G1888) Dual matched capillary columns (DB-WAX and DB-624) of different polarities in a single split/splitless inlet Dual flame ionization detector
Introduction In 1988, the United States Pharmacopoeia (USP) provided control limits and testing criteria for several organic volatile impurities (OVIs) under the official General Chapter <467>. The compounds were chosen based on relative toxicity and only applied to drug substances and some excipients. In an effort to harmonize with the International Conference for Harmonization (ICH), the USP has proposed the adoption of a slightly modified version of Quality-3C (Q3C) methodology, which has been scheduled for implementation on July 1, 2008. The methodology provides an approach to residual solvent analysis that considers a patient s exposure to a solvent residue in the drug product. Solvents have been classified based on their potential health risks into three main classes: Class 1: Solvents should not be used because of the unacceptable toxicities or deleterious environmental effects. Class 2: Solvents should be limited because of inherent toxicities. Class 3: Solvents that may be regarded as less toxic and of lower risk to human health. Testing is only required for those solvents used in the manufacturing or purification process of drug substances, excipients, or products. pharmaceutically acceptable intake level of a residual solvent. When the solvent level in drug substances, excipients, and drug product are below the PDE limit for a given solvent, testing is not required when the daily dose is less than 10 grams. When the level of solvent is expected to be above the PDE limit, testing would be required to determine whether the solvent was removed during the formulation process. The USP has provided a method for the identification, control, and quantification of Class 1 and 2 residual solvents for either water soluable or unsoluable compounds. The method calls for a gas chromatographic analysis with flame ionization detection (FID) and headspace sampling from either water or organic diluents. The monograph has suggested two procedures: Procedure A G43 (DB- 624) phase and Procedure B G16 (DB-WAX) phase. Procedure A is used first. If a compound is determined to be above the specified concentration limit, then Procedure B be should be used to confirm its identity. Since there are known co-elutions on both phases, the orthogonal selectivity ensures that co-elutions on one phase will be resolved on the A other. Neither procedure is quantitative, so to determine the concentration the monograph specifies Procedure C, which utilizes whichever phase will give the fewest co-elutions. In this study all of the three procedures were combined. This was achieved by the use of simultaneous dual capillary column GC. Instead of injecting onto two independent capillary columns of differing polarities, which is required for component confirmation, a single injection was made by headspace sample introduction onto two columns simultaneously. Each column was connected to its own FID. This simultaneous GC injection can be accomplished in several ways: Splitter (analogous to a glass tee or a capillary flow splitter) Press-fit connector (where the capillary column is seated by pressing into a y-shaped glass connector). Two-holed ferrule with a single two-holed column nut (this is the approach highlighted in this application note). Experimental A two-holed ferrule (figure 1) and a single two-holed column nut B The new <467> General Chapter provides an optional method to determine when residual solvent testing is required for Class 2 solvents. Each Class 2 solvent is assigned a permitted daily exposure (PDE) limit, which is the 2 Figure 1 Two-holed ferrule/column nut for the installation of simultaneous dual capillary column into a single inlet with two FIDs.
were used for both capillary columns, resulting in just as many connections as there are when using a single capillary column. The traditional approaches that utilize splitters and press-fit connectors are plagued with multiple connections and therefore multiple leak possibilities. There is also the ability to use the new multidimensional capillary flow device technology (such as Deans switching) that can be configured into an Agilent 7890A GC. However, the 7890A GC used for this study was not configured with that capability. Using the two-holed ferrule approach was very convenient because it did not require the purchase of any extensive parts. Any split/splitless inlet in a 7890A GC could be easily adapted to this approach. The only requirement to ensure proper precision and quantification for simultaneous twoholed dual column GC is to use two matched columns (each with identical lengths and internal diameters). After careful installation of both columns into the single inlet, each capillary column was connected to two identical flame ionization detectors. The signals from both detectors were simultaneously acquired using Agilent ChemStation software (version B.03.01). Table 1 lists the optimized GC and headspace operational variables. Figure 2 represents a schematic block diagram of the 7890A and headspace configuration for the experiments. Standard solutions (appropriate standards were purchased from Sigma-Aldrich) were all prepared quantitatively in dimethyl sulfoxide (DMSO). Multilevel standards were prepared in 100 ml volumetric flasks GC inlet mode Split 7890A GC GC inlet mode Split Inlet temperature 175 C Inlet pressure 30.0 psi Split ratio 15:1 Split flow 154.78 ml/min Carrier gas Helium Column 1 carrier flow 10.3 ml/min Column 2 carrier flow 9.92 ml/min Initial oven temperature 35.0 C Initial time 20.0 min Temperature ramp rate 30.0 C/min Final temperature 240 C Final hold time 0.500 min Column mode Constant flow FID temperature 250 C FID hydrogen air flow 40.0 ml/min FID air flow 400 ml/min FID air flow constant column + makeup Sµm of 30.0 ml/min Total run time 27.3 min Headspace sampler Loop size 1.00 ml Vial pressure 15.4 psi Headspace oven 85.0 C Loop temperature 100 C Transfer line temperature 150 C Equilibration time 30.0 min (high shake) GC cycle time 39.0 min Pressurization 0.200 min Vent (loop fill) 0.100 min Loop equilibration time 0.0500 min Inject time 1.00 min GC columns Column 1: DB-WAX, 30 m, 0.32 mm (Front FID) ID, 0.25 µm (1237032) Column 2: DB-624, 30 m, 0.32 mm ID, 1.80 µm (123-1334) (Back FID) Table 1 Optimized GC and headspace operational variables. Headspace Vent 30 m X 0.32 m m id X 1.8 um DB 624 Figure 2 Instrument schematic. Electronic Backpressure Control G1888 Headspace Sampler Vial Pressure Headspace Transfer Line Two-holed ferrule/column nut Forward flow Aux Module Capillary Inlet Dual Mode Pressure Control Module 7890 GC Restrictor FID FID 30 m X 0.32 mm id X 0.25 um DB WAX 3
using organic-free water. These aqueous dilutions were then transferred with electronic pipettes as 1 ml and 5 ml aliquots into 10 ml and 20 ml screw-cap headspace vials. The screw caps included Teflon-faced septa. All of the standard solutions were analyzed as multiple replicates using the following sequence of sample types: DMSO blank Water blank Calibration standard mixes (between 3 to 12 replicates) Blanks were prepared with 1 ml of the water or DMSO diluents, respectively. Results and discussion Table 2 lists the elution patterns for the various targets on the highly polar DB-WAX column and the intermediately polar DB-624 column. Figures 3 and 4 show representative chromatograms for the various calibration standard analyses that were obtained. These results represent culminations of several experiments where both the HS and GC parameters were optimized. Retention time (minutes) Signal Target analyte 0.605 1 Hexane 0.693 1 Cyclohexane 0.699 1 1,1-dichloroethane 0.758 1 Methyl cylohexane 0.935 2 Methanol 0.939 2 Hexane 0.978 1 trans 1, 2 dichloroethene/tertrahydrofuran* 1.062 1 1,1,1-trichloroethane/carbon tetrachlride* 1.156 1 Methanol 1.278 1 1,2-dimethoxyethane 1.311 1 Methylene chloride 1.364 1 Benzene 1.501 2 1,1-dichloroethane 1.734 2 Acetonitrile 1.788 1 cis 1,2-dichloroethene 1.803 1 Trichloroethylene 1.852 2 Methylene chloride 1.874 1 Acetonitrile 2.074 2 trans 1,2-dichloroethene 2.139 1 Chloroform 2.316 1 Toluene 2.339 2 Nitromethane 2.653 1 1,4-dioxane 2.817 1 1,2-dichloroethane 3.046 1 2-hexanone 3.069 2 Chloroform 3.124 2 cis 1,2-dichloroethene 3.487 2 Tertrahydrofuran 3.601 2 1,2-dimethoxyethane 3.821 2 1,1,1-trichloroethane 3.871 2 Cyclohexane 4.031 1 Ethyl benzene 4.072 2 Carbontetrachloride 4.227 1 p-xylene 4.454 1 m-xylene 4.455 2 Benzene/1,2-dichloroethane** 4.611 2 Trichloroethylene 4.994 1 Nitromethane 5.870 1 o-xylene 5.918 2 Pyridine 5.966 1 Pyridine 6.343 2 Methyl cyclohexane 7.070 2 1,4-dioxane 7.272 1 Chlorobenzene 10.193 2 2-hexanone 10.803 2 Toluene 16.530 1 DMF 16.611 2 Tetralin 20.699 2 DMF 21.044 2 Chlorbenzene 21.541 2 Ethyl benzene 21.810 2 m-xylene/p-xylene** 21.874 1 DMA 22.363 2 o-xylene 22.924 1 Tetralin 22.939 2 DMA 23.266 2 DMSO 23.638 1 DMSO Table 2 Dual column headspace-gc/fid target analyte elution order (signal 1: FID1, DB-WAX; signal 2: FID2, DB-624). *coelute on the DB-WAX column only **coelute on the DB-624 column only 4
A B 1: Methanol 2: Acetronitrile 3: Methylene chloride 4: Trans 1,2-dichloroethene 5: cis 1,2-dichloroethene 6: THF 7: Cyclohexane 8: Methyl cyclohexane 9: 1,4-dioxane 10: Toluene 11: Chlorobenzene 12: Ethyl benzene 13: m-xylene/p-xylene 14: o-xylene 15: Dimethylacetamide C D 1: 1.1-dichloroethene, 2: 1.1.1-trichloroethane/carbon tetrachloride, 3: Benzene, 4: Dichloroethane 1: 1.1-dichloroethene, 2: 1.1.1-trichloroethane, 3: Carbon tetrachloride, 4: Benzene/dichloroethane Figure 3 Selected dual column headspace-gc/fid calibration standard analysis. A) Dual Column Headspace GC/FID, Matched columns (30 M x 0.32 mm ID), Target Mix 1 B) Dual Column Headspace GC/FID, Matched Columns (30 M x 0.32 mm ID), Target Mix 2 C) Dual Column Headspace GC/FID, DB-WAX (30 M x 0.32 mm ID, 0.25 µm), Target Mix 3 D) Dual Column Headspace GC/FID, DB-WAX (30 M x 0.32 mm ID, 1.8 µm), Target Mix 3.1 5
Figure 4 Representative low level standard calibrations and headspace-gc system blank (using 1 ml of diluent water in a 10 ml headspace vial). A) Dual Column Headspace GC/FID, DB-WAX (30 M x 0.32 mm ID, 0.25 µm), Zoomed in snapshot at low concentration levels B) Dual Column Headspace GC/FID, DB-624 (30 M x 0.32 mm ID, 1.8 µm), Zoomed in snapshot at low concentration levels C) Dual Column Headspace GC/FID, Water Blank, Zoomed in snapshot Table 3 summarizes the compositions of the various standard mixtures. These mixtures were all custom made. Several method modifications were made to ensure the required sensitivities and precision of the respective target analytes, while at the same time trying to minimize total analysis times without compromising peak separations. Headspace sampling precision, sensitivity and turnaround were all improved by the following: Setting the vial-shake setting to high Pressurizing the headspace vials to 15 psi Pressurizing the headspace sampling loop to 9 psi (with back pressure regulation) Setting the carrier gas pressure to 35 psi Setting the headspace venting time (which essentially fills the sample loop) to 0.1 minutes or less Target mix 1 DB-624 DB-WAX Methanol Cyclohexane Acetonitrile Methyl cyclohexane Methylene chloride trans 1,2-dichlorethene/Tetrahydrofuran trans 1,2-dichloroethene Methanol cis 1,2-dichloroethene Methylene chloride Tetrahydrofuran cis 1,2-dichloroethene Cyclohexane Acetonitrile Methyl cyclohexane Toluene 1,4--dioxane 1,4-diosane Toluene Ethyl benzene Chlorobenzene p-xylene Ethyl benzene m-xylene m-xylene/p-xylene o-xylene o-xylene Chlorobenzene DMA DMA Target mix 2 DB-624 DB-WAX Hexane Hexane Nitromethane 1,2-dimethoxyethane Chloroform Trichloroethylene 1,2-dimethoxyethane Chloroform Trichloroethylene 2-hexanone Pyridine Nitromethane 2-hexanone Pyridine Tetralin Tetralin DMSO DMSO Target mix 3 DB-624 DB-WAX 1,1-dichloroethene 1,1-dichloroethene 1,1,1-trichloroethane 1,1,1-trichloroethane/carbon tetrachloride Carbon tetrachloride Benzene Benzene 1,2-dichloroethane Table 3 Composition and elution orders of target analyte standards. With the evolution of the electronic pneumatics to the newest generation in the 7890A GC, the gas 6
sampling loop in the headspace sampler can now be controlled to 0.001 psi. It can also be efficiently pressurized throughout the duration of the timed headspace events cycle with back pressure control. The elevation of the column pressures for the GC runs allowed column flows near 10 ml/min for both of the 0.32 mm id columns and not only enhanced precision and sensitivity but also the resolution of the peaks. In short, the capillary columns functioned better at elevated flow (pressure). Elevating the column flows too high could result in loss of separation. Target analyte ICH Excipient limit Retention time Area Excipient class concentration repeatability repeatability MDL# (µg/ml) (%RSD) (%RSD) (ppm) Benzene 1 2 0.011 1.43 0.1 1,2-dichloroethane 1 5 0.015 2.47 0.3 1,1-dichloroethene 1 8 0.010 2.24 0.4 Carbon tetrachloride 1 5 0.010 2.11 0.4 Methylene chloride 2 600 0.010 2.15 40.1 Hexane 2 290 0.014 3.18 12.9 Cyclohexane 2 3880 0.040 2.59 46.1 Trichloroethylene 2 80 0.010 1.49 2.1 Toluene 2 890 0.015 2.02 11.8 Ethylbenzene 2 369 0.004 2.11 20.2 o-xylene 2 195 0.001 1.33 7.1 1,4-dioxane 2 80 0.010 1.7 9.6 Table 4 Retention time and peak area precision, and calculated method detection limits of representative target analytes for the dual column method (note that benzene and 1,2-dichloroethane are completely resolved by almost 1.5 minutes on the DB-WAX column this is a dramatic improvement over the traditional co-elution on the DB-624 column for the low level detection of benzene and 1,2-dichloroethane). This is why it would even be more beneficial to reduce column diameters further to 0.18 mm. Column pressure can be increased even more with lower flows that would help maintain the same linear velocities. The resulting narrow peaks in this study (peak widths as small as 0.001 seconds) provided improved separations which worked well with the control of column pressures with the 7890A GC. Table 4 lists the reproducibility of the various selected targets for the dual GC column separations. A C B D In term of instrumentation, the results were also significantly impacted by the following capillary inlet parts: Inlet liner, low pressure drop (Agilent part number 5183-4647) Agilent exclusive NEW molded gold-plated seal (Agilent part number 5188-5367) Figure 5 Calibration curves for some of the target analytes on the DB-WAX column. A) Chloroform, Linearity: 1.0000 B) Benzene, Linearity: 0.99929 C) 1,4-Dioxane, Linearity: 1.0000 D) 1,2-Dichloroethane, Linearity: 0.99983 To maintain excellent precision and peak shapes, these inlet parts needed to be replaced on a regular basis and were to be treated as ultra-clean parts. Typically, the 7
two-holed column nuts (Agilent part number 05921-21170) were used to coincide with the short style graphite/vespel two-holed ferrules (0.5 mm for 0.32 mm ID columns, Agilent order number 5062-3581; 0.4 mm for 0.25 mm ID and smaller, Agilent order number 5062-3580). Five dilutions of standard solutions were prepared ranging from one tenth to two times the limit concentration to determine the linearity of the calibration curves. These linearity plots of the calibration curves are shown in figures 5 and 6 for the DB-WAX and DB-624 capillary columns. The linearity results for methylene chloride, 1,2 dichloroethane, 1,4-dioxane, chloroform and trichloroethylene are shown for these target analytes for each of the columns. Conclusion A C Figure 6 DB-624 calibration curves for selected target analytes on the DB-624 column. A) Chloroform, Linearity: 0.99992 B) Trichloroethylene, Linearity: 0.99544 C) 1,4-Dioxane, Linearity: 1.0000 D) Methylene Chloride, Linearity: 0.99646 B D Using the on-line combination of a sample preparation device (Agilent headspace sampler) and the most advanced GC in the world (Agilent 7890A), a viable, precise and quantitative method was developed for the revised USP <467> regulations for residual solvents with the following benefits: Simultaneous dual capillary column confirmation of targets using DB-WAX and DB-624 columns with dual flame sionization detection. Optimized GC and headspace sampler operational variables. Direct dual column inlet connection without any instrumental modifications. Use of electronic pneumatic control to enhance the operational characteristics of the headspace sampler (G1888) and the GC 7890A. High precision and sensitivity coupled with an automated headspace sampler and a GC cycle time in less than one hour per sample. Joseph M. Levy is a freelance application chemist. Michael Kraft is Industry Marketing Manager for Pharma/Biopharma Solutions and Process Development & Manufacturing QA/QC. www.agilent.com/chem/pharmaqaqc 2008 Agilent Technologies Inc. Published May 1, 2008 Publication Number 5989-8085EN