Fast and Accurate EPA 8270 Quantitation Using Deconvolution

Transcription

1 Fast and Accurate EPA 8270 Quantitation Using Deconvolution Application Note Environmental Authors Mike Szelewski and Chinkai Meng Agilent Technologies, Inc Centerville Road Wilmington, DE USA Jin Yang, Dawei Zhang, and James Xue SGS Shanghai Environmental Services 3rd Building, 889 Yishan Rd. Shanghai China Abstract Environmental GC/MSD semivolatiles samples can be complex due to matrix interferences. A semivolatiles-specific library and deconvolution database are used in conjuction with AMDIS to clean spectra and target ions. Full spectra and quantitative results from the deconvoluted target ion can be compared in the familiar QEdit view. Identification and quantitation of samples are compared using the Agilent MSD ChemStation software and Deconvolution Reporting Software.

2 Introduction EPA 8270 semivolatiles analysis is one of the most widely used GC/MS methods in the world. The method is developed to analyze acids, bases, and neutrals in soils and water. The list of target compounds typically ranges from 160 to 250, including internal standards and surrogates. SGS Shanghai analyzes more than 5,000 semivolatile samples a year. Due to the complexity of sample matrices (background ion interferences), it is always time-consuming and challenging to confirm found target compounds and get accurate quantitation results. This study is designed to improve productivity by simplifying the data review process and improving the ability to ensure positive peak identification using the Deconvolution Reporting Software (DRS) [1, 2]. Agilent G1716AA DRS is an application for target compound analysis that processes results from the Agilent GC/MSD ChemStation using the NIST Automated Mass Spectral Deconvolution and Identification System (AMDIS) and companion Mass Spectral Search Program into one easy-to-read report. This "deconvolution" process reduces the manual spectral interpretation time significantly, whenever a total ion chromatogram (TIC) from a complex matrix is being analyzed. In addition, after removing interfering and background ions in the deconvolution process, quantitation on the "cleaned" ions can provide more reliable results. Experimental Sample Preparation Soil and sediment samples (mechanical shaker) Decant and discard any water layer on a sediment sample. Discard any foreign objects, such as sticks, leaves, and rocks. Weigh wet subsample 0.5 to 20 g (as-received condition) and place in a 250-mL Schott bottle. Nonporous or wet sample must be mixed with 20 g of anhydrous sodium sulfate, using a spatula. If required, more sodium sulfate may be added. After addition of sodium sulfate, the sample should be free flowing. Add 250 µl of surrogate standard solution (100 ng/µl) to all samples, spiked samples, QC samples, and blanks. For the sample in each batch selected for spiking, add 250 µl of the matrix spiking solution (100 ng/µl). Immediately add 100 ml of dichloromethane (DCM)/acetone (1:1 v/v), Place on the mechanical shaker for 2 hours at 200 rpm. Concentrate the extract to 5 ml using a Kuderna-Danish (KD) concentrator. Quantitatively transfer a 200-µL sample into a 2-mL GC autosampler vial. Add 5 µl of the internal standard (200 µg/ml) to give 5 µg/ml in solution. Seal the vial with a Teflon-faced cap and mix briefly for GC/MS. Water samples (liquid/liquid extraction) Take 1 L water sample to 2 L funnel, check the ph, and adjust, if necessary, to ph 6~8 using 1:1 (v/v) sulfuric acid or 10N sodium hydroxide. Add 50 µl stock surrogate standard solution (100 µg/ml) to each sample and mix well. For the sample in each batch selected for uses as a matrix spike sample, add 50 µl of the matrix spiking solution standard. Add 50 g of NaCl, extract with 60 ml DCM, adjust ph < 2, and extract with 60 ml DCM twice. Combine and reduce the extracts in volume to 1 ml using KD concentrator. Quantitatively transfer 200-µL sample into 2-mL GC autosampler vial. Add 5 µl of the SVOC internal standard (200 µg/ml) to give 5 µg/ml in solution. Seal the vial with a Teflon-faced cap and mix briefly for GC/MS. Instrument Parameters The GC/MS operating parameters are shown in Table 1. Deconvolution In GC/MS, deconvolution is a mathematical technique that separates overlapping mass spectra into "cleaned" spectra of the individual components. Figure 1 is a simplified illustration of spectral deconvolution. Here, the TIC and apex spectrum are shown on the left. In a complex matrix, a peak may be composed of multiple overlapping components and matrix background ions; therefore, the apex spectrum is actually a composite of these constituents. A mass spectral library search would give a poor match, at best, and certainly would not identify all the individual components that make up the composite spectrum. 2

3 Table 1. GC and MSD Operating Parameters GC Agilent Technologies 6890 Inlet EPC split/splitless Mode Splitless, 1.0 µl injected (10-µL syringe, p/n ) Inlet temperature 300 C Pressure 48 kpa Purge flow ml/min Purge time 0.75 min Total flow 56.5 ml/min Septum purge 3 ml/min Gas saver On Saver flow 20 ml/min Saver time 2 min Gas type Helium Liner 4 mm, p/n Liner O-ring Nonstick, p/n Septum Advanced green, 11 mm, p/n Column ferrule 0.4 mm id Vespel/Graphite 85%/15%, p/n Column Agilent J&W DB-5ms, p/n Length 30.0 m Diameter 0.25 mm Film thickness 0.25 µm Mode Constant flow Nominal initial flow 1.0 ml/min Outlet MSD Oven Oven ramp C /min Final ( C) Hold (min Initial 40 4 Ramp Ramp Ramp Runtime 40 min Oven equilibration time 0.5 min Thermal AUX 2 MSD transfer line, 280 C MSD Agilent Technologies 5975 MSD Tune file dftpp.u Mode SIM/Scan Solvent delay 3.80 min EM voltage offset 200 V Low mass 35 amu High mass 550 amu Threshold 150 Sampling 2 Quad temperature 150 C Source temperature 230 C Software GC/MSD ChemStation G1701EA rev E Deconvolution Reporting Software G1716AA rev A.04 Semivolatiles DRS Database G1677AA rev A.01 (273 compounds) NIST08 Search Engine and Library, including AMDIS 2.66, build AMDIS Settings There is no single group of settings that will guarantee finding all compounds in all matrices. The following settings produced the best results for the data set studied. Component width 24 Adjacent peak subtraction 1 Resolution Medium Sensitivity High Peak shape Low Minimum match factor 30 RI window 10 Level Infinite Maximum penalty 100 The deconvolution process first corrects for the spectral skew that is inherent in quadrupole mass spectra and determines a more accurate apex retention time (RT) of each chromatographic peak. Then, the deconvolution process groups ions whose individual abundances rise and fall together and have the same RT. As illustrated in Figure 1, deconvolution can produce a "cleaned" spectrum for each overlapping component. These "cleaned" spectra ("components" as they are called in AMDIS) are compared against a library. This comparison uses full spectra (not just three ion ratios) and is RT independent. However, Agilent GC/MSD systems can be retention time locked (RTL), and AMDIS can, after spectral identification, accept or reject components based upon their proximity to the locked RTs. Identified targets are saved in a table for later use and are also sent to NIST for further confirmation. TIC and spectrum TIC Deconvolution Deconvoluted peaks and spectra Component 1 extracted spectrum Component 2 extracted spectrum Component 3 extracted spectrum Library search each component to identify Figure 1. Illustration of spectral deconvolution in AMDIS. 3

4 DRS Database/Library The G1677AA Environmental Semivolatiles RTL Database/ Library (DBL) is a set of mass spectral libraries in the Agilent and NIST/AMDIS formats. There are three separate sets of libraries and methods. An 8270 set includes the mass spectra and locked retention times for 243 single-component semivolatiles compounds and internal standards specified by USEPA Method 8270 plus 30 additional compounds of environmental interest a total of 273 compounds. Additional information can be found in reference 3. The system used in this study was not RTL to the DRS DBL. The DRS DBL RTs were updated with the SGS quant database RTs via a menu function that is supplied with the DRS A.04 revision. This is a useful feature for labs that have established methods that they do not want to change. Results and Discussion A typical TIC from one sample can be seen in Figure 2, and is quite complex. DRS can be used as a tool to aid in both identification and quantitation of complex samples. As part of the process, deconvoluted data from AMDIS are imported into the MSD ChemStation QEdit View (see Figure 3). 1) The panel in the middle of the display shows a list of all target compounds. Target compounds quantified by ChemStation are labeled by x and target compounds quantified by DRS-AMDIS are labeled by A. 2) The top left panel shows a multi-ion overlay where MSD target (quant) ion, up to three qualifier ions, and the target (quant) ion from AMDIS (if a compound with an AMDIS hit A is selected), are displayed for quick confirmation. If the AMDIS ion does not overlay with the MSD ions, it implies that two different compounds were quantified. 3) The bottom left panel shows three spectra the raw "dirty" spectrum, the AMDIS "clean" deconvoluted spectrum, and the AMDIS library spectrum are displayed if a compound with an AMDIS hit A is selected or if the AMDIS target ion is manually integrated. By visually comparing the bottom two spectra (deconvoluted spectrum and the library spectrum), a good or poor match can be determined easily. 4) At the top right of the screen, the MSD target ion used for quantitation and 5) the AMDIS deconvoluted target ion are shown in two separate panels. Redrawing the baseline (manual integration) may be performed in each panel to improve the accuracy of the quantitation results. The panel of the AMDIS target (quant) ion has a very flat and clean baseline due to the deconvolution process. This helps to get a more reliable quant value. 6) The bottom right panel shows both the integrated area and calculated amounts for MSD and AMDIS target ions. Qualifier ion ratios are also listed for evaluation purposes. A # flags the out-of-range ion ratio. All these panels make it easy to confirm the deconvoluted compounds and compare quantitation results (refer to the DRS A.04 Help file for more information) Figure 2. Total ion chromatogram of a semivolatiles sample. 4

5 Figure 3. QEdit screen displaying MSD and AMDIS information, Acenaphthene selected. A detailed analysis of one sample is in Table 2, information combined from two separate DRS reports for presentation here. The first three columns, blue highlighted, list the retention time at which the compound was found, the CAS#, and the name. The four columns to the right, green highlighted, starting with the AMDIS Match, list the match factor of the deconvoluted component compared to the DRS Semivolatiles DBL and its RT difference from the expected. A low minimum match factor of 30 was used in AMDIS. This may result in a few more false positives but will minimize the number of false negatives. The NIST Match factor is a further check on identification, comparing the component to the NIST library. The last column lists the hit number in the list of the top 100 hits from NIST. The three remaining columns in Table 2 show amounts calculated against the MSD ChemStation quant database. The Auto column lists the amounts that were originally found by the ChemStation with automatic integration. If a compound showed qualifier ion ratios that were out of range, this is indicated by Qualifier MisMatch (QMM). The first three peaks with QMM were successfully integrated automatically, but the amounts were wrong. Either the wrong peak was detected or the integration was incorrect due to matrix. False positives and false negatives are indicated by FP and FN, respectively. The Man column lists amounts that were based on manual integration. The RT used for integration was based on the AMDIS component time. Both of the columns labeled MSD used raw data as normally seen without deconvolution. The two compounds labeled "Can't int" showed no distinct peak that could even be manually integrated. The AMDIS column shows the amount based on the deconvoluted "clean" extracted ion. 5

6 Table 2. Details of MSD ChemStation and DRS Results for a Single File Auto Man RT CAS# Name MSD MSD AMDIS AMDIS AMDIS NIST NIST ng ng ng Match RT diff Match Hit # Aniline Phenol QMM NF Bis(2-chloroethyl) ether QMM-320 Can't int NF o-toluidine QMM-410 Can't int N-Nitroso-di-n-propylamine FP ,2,4-Trichlorobenzene Naphthalene Acenaphthene ,4-Dinitrophenol QMM Dibenzofuran Chlorophenyl phenyl ether Diphenylamine Azobenzene Phenanthrene Anthracene FP Di-n-butylphthalate FN NF Benzo(a)anthracene Bis(2-ethylhexyl)phthalate Di-n-octylphthalate FP An example illustrating the value of quantitation of a deconvoluted target ion can be found in Figure 4. The black trace is the raw extracted target ion 93, and the two qualifiers are shown in blue 95, and purple 63. Without deconvolution, the software or an analyst may choose any of the peaks in the region for quantitation. All of these would result in incorrect amounts and misidentifications. The deconvoluted target ion 93 is shown in red and is easy to integrate. Eight soil and water samples with high interference were processed with the MSD ChemStation including DRS. The results are summarized in Table 3. Careful manual review shows that 183 compounds are actually present. Using only the MSD ChemStation automated quant results, there were a large number of false positives (FPs) and false negatives (FNs). This indicates that all compounds in the quant database would have to be individually reviewed. It should be noted that the results presented in Table 3 are based on automatic integration and data processing before manual review by an experienced analyst. AMDIS showed fewer FPs and FNs, but these are also not predictable. Using the ChemStation and AMDIS together, through DRS, showed the fewest number of FPs and FNs. There were no FNs that were the same in both programs, which saves significant time for the analyst, as only the positives need to be reviewed. Ion 95 Figure 4. Ion 63 MSD Target ion 93 Deconvoluted ion Expected RT Bis(2-chloroethyl) ether target and qualifier ions. 6

7 Table 3. Summary Results for Eight Samples References Total for 8 samples Total real positives based on manual review 183 ChemStation false positives at 20% absolute 35 qualifier ion agreement ChemStation false negatives at 20% absolute 56 qualifier ion agreement AMDIS false positives 11 AMDIS false negatives 11 Both AMDIS and ChemStation same false negatives 0 Both AMDIS and ChemStation same false positives 2 Conclusions Environmental GC/MSD semivolatiles samples can be complex due to matrix interferences. Data review can be simplified, data quality improved, and productivity increased through the use of integrated deconvolution software. Using the familiar QEdit view, operating training is minimized. The fewest number of false positives and false negatives will be found in the shortest amount of time using the MSD ChemStation in combination with DRS. Among the eight complex samples analyzed, no false negatives were found. 1. Philip L. Wylie, Michael J. Szelewski, Chin-Kai Meng, and Christopher P. Sandy, "Comprehensive Pesticide Screening by GC/MSD Using Deconvolution Reporting Software," Agilent Technologies, publication , May Bruce Quimby and Mike Szelewski, "Screening for Hazardous Chemicals in Homeland Security and Environmental Samples Using a GC/MS/ECD/FPD with a 731 Compound DRS Database," Agilent Technologies, publication , February Michael J. Szelewski, "Semivolatiles Retention Time Locked (RTL) Deconvolution Databases for Agilent GC/MSD Systems," Agilent technologies, publication , February 2008 For More Information For more information on our products and services, visit our Web site at 7

8 Agilent shall not be liable for errors contained herein or for incidental or consequential damages in connection with the furnishing, performance, or use of this material. Information, descriptions, and specifications in this publication are subject to change without notice. Agilent Technologies, Inc., 2008 Published in the USA December 9, EN