• Author / Uploaded
• pajovicm

• Categories
• Cromatografía de gases – Espectrometría de masas
• Cromatografía de gases
• Espectrometría de masas
• Cromatografía
• Ciencias fisicas

Citation preview

Solutions that meet your demands for

food safety testing

Excellent choices for food applications

Table of Contents The table of contents on the right has been linked to the individual sections in this compendium. Click on the text to jump to a specific section. If you prefer to search for words or phrases used in any of the enclosed applications notes, click on the Search button below to open Acrobat’s Search window.

Search

Solution Segments Pesticides and Residues Pesticides and Herbicides Antibacterial Drug Residues

Contaminants Acrylamides Food Packaging PCBs Toxins Trace Metals

Natural Compounds and Additives Additives Amino Acids Carbohydrates Fats and Oils Flavors and Fragrances Miscellaneous Proteins Vitamins

Bioanalysis Applications

Productivity Tools Applications

Index

Pesticides & Residues Pesticides and Herbicides

> Return to Table of Contents > Search entire document

Replacing Multiple 50-Minute GC and GC-MS/SIM Analyses with One 15-Minute Full-Scan GC-MS Analysis for Nontargeted Pesticides Screening and >10x Productivity Gain Application Food Safety

Author

Introduction

Chin-Kai Meng and Mike Szelewski Agilent Technologies 2850 Centerville Road Wilmington, DE 19808

To have a plentiful food supply, most fruits and vegetables are treated with pesticides (insecticides, fungicides, herbicides, etc.) to protect primarily against insects, molds, and weeds. Therefore, in order to ensure food safety, the food supply is frequently monitored for pesticide residues. Nowadays, the pesticide monitoring is expanding beyond food, for example, to botanical dietary supplements.

Abstract Pesticide analysis of fruits and vegetables requires finding trace-level residues in complex matrices. Up to now, the typical trade-off is between sensitivity and confirmation. Therefore, multiple injections are needed for screening and confirmation using gas chromatography with mass spectrometry (GC-MS) or GC-MS in combination with GC with element-selective detection. With the recent introduction of hardware and software tools, for example, capillary flow three-way splitter, trace ion detection, and deconvolution, a 15-minute fast analysis can match the results obtained from three injections of approximately 50 minutes each. A table comparing the results from the Food and Drug Administration/Center for Food Safety and Applied Nutrition (FDA/CFSAN) procedure using the traditional multi-instrument approach and the new Agilent single injection approach shows that Agilent's fast analysis is capable of finding all the target analytes in less than one-tenth of the current FDA/CFSAN total analysis time.

The analytical challenge to monitor (identify and quantify) trace multiresidues requires an effective and universal extraction and analysis method for maximum productivity and efficiency. Up to now, the trade-off in analysis has been between sensitivity and confirmation. Element-selective gas chromatograph (GC) detectors, such as the flame photometric (FPD), electron capture (ECD), electrolytic conductivity (ELCD), and halogen selective (XSD) detectors, provide excellent selectivity and sensitivity; however, they lack the capability to identify. On the other hand, mass spectrometry (MS) is capable of identifying an analyte by fullscan library match or multiple target and qualifier ion ratios from selected ion monitoring (SIM). However, MS sometimes lacks the selectivity to

find target analytes in a complex matrix full of interferences and chemical background. Analyte spectra are sometimes overwhelmed by similiar ions contributed from the coextractives in the matrix that prevent the analyte of interest from being identified or confirmed. The compromise and the typical approach are to use selective GC detector(s) to flag potential target analytes and use MS SIM for confirmation. For instance, many laboratories screen food samples for semivolatile pesticides using the ECD or ELCD (or XSD) for organohalogen, FPD or pulsed FPD (PFPD) for organophosphorus, and NPD for nitrogen-containing targets [1 – 5]. Any found targets are further confirmed by GC-MS/SIM. In addition, other procedures have used GC-MS/SIM entirely for the screening of pesticides in foods [6 – 8]. In most of these procedures, multiple injections are needed to identify hundreds of compounds at the detection limit in the low parts-per-billion (ppb) levels. To improve the efficiency and increase the productivity of screening for all of these pesticides, the challenge is to reduce the GC-MS or the combination of GC and GC-MS analysis times.

results from the current FDA/CFSAN multi-instrument approach and the new Agilent single-injection approach shows that not only is Agilent’s fast analysis capable of finding all the target analytes, but it is also accomplished in just one-tenth of the current FDA/CFSAN total analysis time.

Experimental Sample Preparation Sample extracts of fresh produce were prepared by FDA based on modifications of the QuEChERS protocols [9, 10]: • Homogenize 1 to 2 kg of sample • 15 g sample + 15 mL 1% AcOH/ACN, homogenized • Add 6 g MgSO4 and 2.5 g NaOAc, shake vigorously for 1 minute, and centrifuge • Transfer ~15 mL + 0.5 g C-18 + 1.2 g MgSO4, shake, and centrifuge for 5 min at 3,000 rpm • Transfer ~12 mL + 0.4 g PSA + 0.2 g GCB + 1.2 g MgSO4, vortex

There are several hundred pesticides typically used in the world, and each country has its own pesticide tolerance levels for different agricultural commodities. This presents another analytical challenge in multiresidue monitoring: to develop a nontargeted procedure to identify pesticides at trace levels in different food matrices.

• Add 4 mL toluene, shake, and centrifuge

These challenges are met by the recent introduction of hardware and software tools, including GCMS, capillary flow three-way splitter, trace ion detection, and deconvolution reporting software (DRS). The splitter allows multiple GC as well as MS signals to be acquired from a single injection for productivity gains (from three injections down to one). Trace ion detection minimizes noise on the signal and DRS separates target analyte ions from matrix background ions.

Sample preparation of dried ginseng powder is similar to that used for fresh produce, but smaller sample sizes (2 g) were used [11].

Several sample extracts were analyzed by the current Food and Drug Administration/Center for Food Safety and Applied Nutrition (FDA/CFSAN ) multiple injection process and this new Agilent pesticide system. With DRS, the demand for chromatographic resolution is minimum; therefore, the Agilent system was running the analysis at a 3x faster speed (one-third the analysis time) to further increase productivity. A table comparing the

2

• Transfer 6 to 8 mL, evaporate and bring to volume with toluene, add I.S. • Add MgSO4, vortex and centrifuge, transfer to ALS vials • GC and GC-MS analysis

Capillary Flow Three-Way Splitter One of the capillary flow devices is a three-way splitter, which consists of two half plates bonded together (diffusion bonding) to form a plate with the etched flow channels inside. The splitter is only 6.5 cm tall and 3 cm wide and is mounted on the side of the oven wall (see Figure 1). The low thermal mass minimizes cold spots and peak broadening. All capillary flow devices use metal column ferrules, have extremely low dead volumes, are inert, and do not leak, even after many oven cycles.

To MSD

To FPD To µECD

Aux EPC in

Column in

Etched flow channel inside two diffusion bonded plates Capillary tube connection via metal ferrule

Figure 1.

Flow diagram of the three-way splitter. The picture shows the splitter mounted on the oven wall.

The three-way splitter enhances productivity by splitting column effluent proportionally to multiple detectors: MSD, dual flame photometric detector (DFPD) and micro-electron capture detector (µECD). Therefore, two GC detector signals can be acquired together with the MS data (both SIM and scan signals if desired) from one injection [12]. The exit end of the analytical column is installed into one of the four ports on the splitter using a metal ferrule. The other three ports are connected to three detectors via restrictors (deactivated capillary tubing) of varying diameter and length to set the split ratio among the three detectors. Restrictors are sized for 1:1:0.1 split ratio in favor of MSD and DFPD (µECD has 1/10 of the flow to MSD), with similar hold-up times. The splitter uses auxiliary (Aux) electronic pneumatics control (EPC) for constant pressure makeup flow. The makeup gas (Aux pressure 6) at the splitter is fixed at 3.8 psi to maintain the split ratio throughout the run. This multisignal configuration provides full-scan data for library searching, SIM data for trace analysis, DFPD (phosphorus or sulfur mode), and µECD data for excellent selectivity and sensitivity from complex matrices. The trade-off is the decrease of analyte concentration in any detector due to the flow splitting and the additional makeup gas from the splitter. An analyte would have similar retention times in all three detectors. Therefore, the GC data can be used in two ways: first, to confirm the presence of target analytes found by the MSD deconvolution reporting software (DRS), and second, to highlight potential target compounds to be further confirmed by MSD. With the new 7890A GC software, up to six columns/ restrictors can be configured/assigned to

different inlets and outlets. Aux pressure can be either an inlet (for the splitter flow restrictors connected to different detectors) or an outlet (for the analytical column). A graphical user interface makes the configuration easy to set up. Once all the columns and restrictors are configured, the backflush can be executed easily. Backflush Traditional bakeout step for removing late eluters could be very time consuming, or even as long as the analysis time depending on the matrix. Backflush is a simple technique to remove high boilers from the column faster and at a lower column temperature to cut down analysis time and increase column lifetime. Capillary flow devices (in this case, a three-way splitter) also provide backflush [13, 14] capability. “Backflush” is a term used for the reversal of flow through a column such that sample components in the column are forced back out the inlet end of the column. By reversing column flow immediately after the last compound of interest has eluted, the long bake-out time for highly retained components can be eliminated. Therefore, the column bleed and ghost peaks are minimized, the column will last longer, and the MS ion source will require less frequent cleaning. The split vent trap may require replacement more frequently than usual. Figures 2 and 3 are two screen shots from the MSD ChemStation software, providing a summary of the backflush operation. In Figure 2, the column and three restrictor dimensions and respective detectors are shown (the setup came from the column configuration section). For MSD, the user can choose the vacuum pump installed on the system. This information will be used to calculate if the backflush is within the system flow limits. By clicking on the “Evaluate…” button, the screen shown in Figure 3 appears, listing the maximum flow for each detector and the void volumes for a certain backflush time. In this example, Aux pressure is at 60 psi, inlet is at 1 psi, and oven is at 280 °C. The backflushing flow is shown to be 8.66 mL/min, and the void time is shown to be 0.16 min. Therefore, backflushing for 2.5 minutes will send 15.6 void volumes through the column. This is useful for developing the backflush method. Figures 2 and 3 simplify the setup and development of a backflush method.

3

Figure 2.

Backflush setup in ChemStation.

Figure 3.

Automated backflush calculations in ChemStation.

4

Deconvoluted peaks and spectra

TIC and spectrum

Component 1 extracted spectrum

TIC

Component 2 extracted spectrum Deconvolution

Another useful feature in Figure 3 is the “warning,” shown as highlighted yellow cells. In this example, setting the backflush pressure to 60 psi sends more than the allowable flow (60 mL/min) to the FPD. Therefore, the backflush pressure setting and the actual flow value to FPD are shown in yellow as “warnings.” Although the system will accept the setup, the high flow may cause consequences in the analysis, for example, flameout.

Component 3 extracted spectrum

Trace Ion Detection Trace ion detection [15] is a filtering algorithm to smooth peaks. This filtering is an advanced form of averaging used to remove the noise riding on the signal. The implications from TID are typically a slight loss in peak height and some peak broadening. The default setting in ChemStation for TID is off. It should be turned on for any analysis that uses deconvolution and has more than six sampling points across a peak. TID provides better signal-to-noise ratios and helps deconvolution to confirm target compounds as shown in the Results section. Deconvolution In GC/MS, deconvolution is a mathematical technique that separates overlapping mass spectra into deconvoluted spectra of the individual components. Figure 4 is a simplified illustration of this process. Here, the total ion chromatogram (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 of the individual components that make up the composite spectrum. The deconvolution process groups ions whose individual abundances rise and fall together within the spectrum. The deconvolution process first corrects for the spectral skew that is inherent in quadrupole mass spectra and determines a more accurate apex retention time of each chromatographic peak. As illustrated in Figure 4, deconvolution produces a “cleaned” spectrum for each overlapping component. These individual spectra can be library searched with a high expectation for a good match. Deconvolution significantly reduces chromatographic resolution requirements, allowing much shorter analysis times.

Library search each component to identify

Figure 4.

Deconvolution process of three overlapped peaks.

Agilent Deconvolution Reporting Software (DRS) utilizes the AMDIS deconvolution program from the National Institute of Standards and Technology (NIST), originally developed for trace chemical weapons detection in complex samples [16]. DRS presents the analyst with three distinct levels of compound identification: (1) ChemStation, based on retention time and four ion agreement; (2) AMDIS, based on “cleaned spectra” full spectral matching and expected retention time window as a qualifier; and (3) NIST05 search using a >163,000compound library [17, 18]. In this application, both the ChemStation quantitation database and the AMDIS library have the same 926 entries. These entries include pesticides, numerous metabolites, endocrine disruptors, important PCBs and PAHs, certain dyes, synthetic musk compounds, and several organophosphorus fire retardants [18]. The AMDIS software, shipped with the NIST05 Library CD-ROM, is also capable of deconvoluting selected ion monitoring (SIM) data [19], while previous AMDIS revisions were not. Testing has shown that proper compound identification requires four ions per compound. All Agilent DRS databases are retention time locked and have both full-scan and SIM libraries for AMDIS. Instrument Method The system used for this study consists of an Agilent 7890A GC with split/splitless inlet, a threeway splitter, µECD, DFPD, and 5975 MSD. For a detailed description of SIM/scan and the splitter system configuration, please refer to the experimental section of reference [12]. See Table 1 for hardware detail and settings.

5

Table 1. Gas Chromatograph, Mass Spectrometer, and Three-Way Splitter Operating Parameters GC

Agilent Technologies 7890A with 240V fast oven option

Injector Syringe size Injection volume Solvent A wash Solvent B wash Sample wash Sample pump Plunger speed

Agilent Technologies 7683 10 µL 1 µL 1 (pre), 3 (post) 1 (pre), 3 (post) 0 4 Fast

Inlet Mode Inlet temperature Pressure

EPC split/splitless Splitless 250 °C ~24.4 psi (chlorpyrifos methyl RT locked to 5.531 min, 3x speed) constant pressure mode 50.0 mL/min 2 min 3 mL/min Switched Off Helium Helix double taper liner, deactivated, p/n 5188-5398

Purge flow Purge time Septum purge flow Septum purge mode Gas saver Gas type Liner Oven Oven ramp Initial Ramp 1 Ramp 2 Ramp 3

75 9 24

Runtime Oven equilib time Post-run time Post-run temperature

13.96 min 1.0 min 2.5 min 280 °C

Column

Agilent Technologies HP-5MS, p/n 19091S-431 15.0 m 0.25 mm 0.25 µm Constant pressure RT locked to chlorpyrifos methyl at 5.531 min, 3x analysis speed 3.5 mL/min Aux pressure 6 3.8 psi (Aux EPC pressure to three-way splitter), helium gas

Length Diameter Film thickness Mode

Nominal initial flow Outlet Outlet pressure

°C /min

Final (°C) 70 150 200 280

Backflush (post-run) Oven Time Inlet Aux pressure 6

280 °C 2.5 min 1 psi 60 psi (column outlet)

Front detector Temperature Const col + makeup Make gas type Data rate

µECD 300 °C 60.0 mL/min Nitrogen 20 Hz

Back detector Temperature

Dual FPD 250 °C

6

Hold (min) 0.67 0 0 3.33

Hydrogen flow Air flow Const Col + Makeup Make gas type Lit offset Data rate Transfer line

75.0 mL/min 100.0 mL/min 60.0 mL/min Nitrogen 2.00 20 Hz 250 °C

AUX Thermal 1 AUX Pressure 6 Gas type Initial pressure Backflush pressure

MSD transfer line, 280 °C Three-way splitter Helium 3.8 psi 60 psi

MSD Tune file Mode Solvent delay EM voltage Low mass High mass Threshold Samples Scans/sec Quad temp Source temp

Agilent Technologies 5975C MSD Atune.u Scan 1.50 min Atune voltage 50 amu 550 amu 0 2 2.91 150 °C 230 °C

Three-way splitter

Agilent 7890A Option 890, installed during factory assembly 3.8 psi (Aux pressure 6 setting) 1:1:0.1 MSD:DFPD:µECD 1.444 m × 0.18-mm id deactivated fused silica tubing, p/n 160-2615-10 0.532 m × 0.18-mm id deactivated fused silica tubing, p/n 160-2615-10 0.507 m × 0.10-mm id deactivated fused silica tubing, p/n 160-2635-1 3.43 mL/min (at 70 °C), 1.53 mL/min (at 280 °C) 3.43 mL/min (at 70 °C), 1.53 mL/min (at 280 °C) 0.343 mL/min (at 70 °C), 0.153 mL/min (at 280 °C) 3.19 mL/min (at 70 °C), 1.52 mL/min (at 280 °C)

Pressure Split ratio MSD restrictor DPFD restrictor µECD restrictor Flow to MSD Flow to DFPD Flow to µECD Makeup (Aux 6) Software GC/MSD ChemStation MS Libraries

Deconvolution software Library searching software

Deconvolution reporting software

Agilent part number G1701EA (version E.01.00 or higher) NIST05a mass spectral library (Agilent part number G1033A) Agilent RTL Pesticide and Endocrine Disruptor Libraries (926 entries) in Agilent and AMDIS formats (part number G1672AA) Automated Mass Spectral Deconvolution and Identification Software (AMDIS_32 version 2.65 Build 116.66) NIST MS Search (version 2.0d or greater) (comes with NIST'05a mass spectral library – Agilent part number G1033A) Agilent part number G1716AA (version A.03.00 or higher)

Results and Discussion

Trace Ion Detection

Backflush Example

Figure 6 compares the signals when TID is on and off. Visually, it is obvious that TID smoothes the noise riding on top of the signal. When TID was on, Atrazine was successfully identified by AMDIS. When TID was off, Atrazine was not found by AMDIS and resulted in a false negative. Figure 7 compares TID on and off for two different analytical conditions of the same ginseng extract. On the right, the fast (3x) analysis was a 1-µL splitless injection with TID on. The analyte Diazinon was found by AMDIS with a peak width less than 5 seconds. On the left side, the normal (1x)

Blank runs, made after separate milk analyses with different backflush (BF) times, are shown in Figure 5. The top TIC is a blank run after a milk extract analysis stopped at 42 minutes and the system backflushed for 1 minute. The next TIC is a blank run after another milk extract analysis stopped at 42 minutes and backflushed for 2 minutes and so on for the other five TICs. It is interesting to confirm graphically that the latest eluters disappeared from the TIC earliest in backflushing. 100000 50000 0 100000 50000 0 100000 50000 0 100000 50000 0 100000 50000 0 100000 50000 0 100000 50000 0

BF for 1 min

BF for 2 min

BF for 3 min

BF for 4 min

BF for 5 min

BF for 6 min

BF for 7 min 10.00

Figure 5.

Note: late eluters were backflushed out first

Column is clean 20.00

30.00

40.00

50.00

60.00

70.00

80.00

Differences in blank runs as the result of seven different backflush times. Atrazine found by AMDIS

TID on

TID off

Atrazine not found by AMDIS Figure 6.

The power of deconvolution with TID for atrazine, from AMDIS. 7

analysis was a 5-µL cold splitless injection using a PTV with TID off. The 9-second-wide Diazinon was not found by AMDIS, also a false negative. Both examples show that TID is a very useful feature for trace target analysis. Benefits of TID: • Improves the signal-to-noise ratio • AMDIS is more thorough in identifying components, resulting in fewer false positives • Improves library match quality • Improves area repeatability, resulting in more reliable quantitation DRS and Splitter

Deconvolution Figures 12 through 15 show the results from AMDIS. There are three spectra for each target compound found by AMDIS. The top window shows the spectrum (scan) from the TIC. This is the only spectrum that would be available for library searching without deconvolution – obviously quite useless. The middle window shows the deconvoluted spectrum and the bottom window is the target compound’s spectrum in the library. The compound confirmation can be done easily and with confidence by visually comparing the bottom two spectra. The power of deconvolution is appreciated while comparing the top two spectra (the raw scan and the spectrum hidden in the raw scan).

Figures 8 and 9 are DRS reports for ginseng and peach extracts with pesticides highlighted (for a detailed explanation of the report, please refer to references [17] and [20]). Figures 10 and 11 show simultaneously collected GC and MS signals (RT locked) for the corresponding ginseng and peach extracts from a three-way splitter. The presence of the GC peaks from the µECD and FPD (P) helps confirm the targets reported by DRS. Each run is finished at 15 minutes using the 3x speed and a 240V oven. With deconvolution, less peak resolution is required for compound identification. A 4minute backflush is added after the run to make sure that the column is clean to maintain the next run’s locked RTs for all peaks.

It is easy to further confirm the hits found by deconvolution. In Figure 9, four pesticides found by AMDIS in the peach extract have a match factor of about 80 or lower. The four pesticides are Cabaryl, Captan, Propiconazole, and Fenbuconazole. A SIM method of these compounds was set up to analyze the peach extract. By selecting the proper AMDIS library (full-scan or SIM), DRS can process full-scan as well as SIM data files [19]. Figure 16 is the DRS report of the peach SIM analysis. The high match factor (99 or higher) and the small RT difference of all targets found by AMDIS confirm the presence of all compounds.

PTV, 1x speed, 5 µL injection No TID, Diazinon not found (false negative)

Splitless, 3x speed, 1 µL injection with TID, Diazinon found

Figure 7.

8

The power of deconvolution with TID for diazinon.

Figure 8.

DRS report for ginseng with pesticides highlighted.

Figure 9.

DRS report for peach with pesticides highlighted.

9

Ginseng 1e+07 8000000

TIC

6000000 4000000 2000000

9e+07 8e+07 7e+07 6e+07 5e+07 4e+07 3e+07 2e+07 1e+07

6000000

2.00

3.00

4.00 3.524

5.00

6.00

7.00

8.00

Tetrachloro-m-xylene (ISTD) Chlorthaldimethyl

3.00

5000000

4.00 3.665

8.290

7.783

4.501

2.00 1.660

9.00

8.699

5.00

6.248

7.005

6.00

7.00

8.00

10.00 9.413

11.00

8.744 8.992 9.093 9.059 9.265 9.528 9.646 9.784

9.00

12.00

13.00

µECD Azoxystrobin

10.00

11.00

12.00

13.00

Tributylphosphate (ISTD) 9.413

4000000

FPD (P)

Diazinon

3000000

4.831

2000000

9.313 7.641

1000000 2.00

3.00

4.00

5.00

6.00

7.00

8.00

9.00

10.00

11.00

12.00

13.00

Figure 10. Simultaneous display of MSD and GC selective detector signals for ginseng.

Peach

2200000 1800000

TIC

1400000 1000000 600000 200000 2.00 3.5e+07

3.00

4.00

5.00

6.00

8.00

9.00

10.00

11.00

3.519

2.5e+07 2e+07 1.5e+07

4.696

2.703

3.932

5000000 2.00 5000000

3.00 4.00 3.641

13.00

Carbaryl 5.640

5.00

7.093 6.680

6.00

µECD

Endosulfan(alpha) 7.556 Phosmet 9.538 Captan

7.00

9.081 8.00

9.00

10.559 10.00

12.382 11.00

12.00

13.00

9.523

Phosmet

Tributylphosphate (ISTD)

4000000

FPD (P)

3000000 2000000 9.358

1000000

9.826 9.910

4.773 2.00

3.00

4.00

5.00

11.066 6.00

7.00

8.00

9.00

10.00

11.00

Figure 11. Simultaneous display of MSD and GC selective detector signals for peach. 10

12.00

Tetrachloro-m-xylene (ISTD)

3e+07

1e+07

7.00

1.838

12.00

13.00

Scan at 12.299 min

Deconvoluted/extracted spectrum

Library spectrum Azoxystrobin

Figure 12. Raw (dirty) spectrum, deconvoluted (clean) spectrum, and library spectrum of azoxystrobin found in ginseng, from AMDIS.

11

Scan at 5.615 min

Deconvoluted/extracted spectrum

Library spectrum Carbaryl

Figure 13. Raw (dirty) spectrum, deconvoluted (clean) spectrum, and library spectrum of cabaryl found in peach, from AMDIS.

12

Scan at 10.776 min

Deconvoluted/extracted spectrum

Library spectrum Fenbuconazole

Figure 14 Raw (dirty) spectrum, deconvoluted (clean) spectrum, and library spectrum of fenbuconazole found in peach, from AMDIS.

13

Scan at 8.934 min

Deconvoluted/extracted spectrum

Library spectrum Endosulfan sulfate

Figure 15 Raw (dirty) spectrum, deconvoluted (clean) spectrum, and library spectrum of endosulfan sulfate found in tomato, from AMDIS.

14

Figure 16. DRS report from the SIM analysis of peach. Please refer to reference 18 for the explanation of the fictitious CAS number assigned to Propiconazole-II (999048032).

Comparison of Incurred Samples The current approach at FDA/CFSAN is to find a wide suite of organohalogen and organophosphorus pesticide residues. This requires four injections (GC-MS/SIM and GC-ELCD for organohalogen and GC-MS/SIM and GC-FPD for organophosphorus screening) of approximately 50-minutes runtime each (total runtime = 200 minutes). Table 2 shows that FDA found several target compounds in three extracts as well as quantitation Table 2.

results from both GC and MS. In comparison, using the new tools (splitter, TID, and deconvolution) found as many target compounds and a few more in just one short (15-minute) full-scan analysis. The three-way splitter was used to get selective GC signals (µECD and FPD) for confirmation purposes. Due to column effluent splitting to three detectors (1:1:0.1), the MSD is getting less than half of the amount injected. FDA/CFSAN GC and GC/MS/SIM analyses for organohalogen monitor-

Comparison of the Agilent Pesticide System Results with the FDA Results Agilent DRS (full scan/TID)

FDA (FPD, ELCD, SIM)

GC-FPD or ELCD

GC-MS/SIM

Ginseng

Diazinon Chlorthal-dimethyl Azoxystrobin

Diazinon (FPD, SIM)

25 ± 3 ppb

25 ± 2 ppb

Peach

Carbaryl Captan Endosulfan (alpha) Phosmet Propiconazole I and II Fenbuconazole

Phosmet (FPD, SIM)

320 ± 37

230 ± 23

Chlorothalonil (ELCD, SIM) Endosulfan (alpha) (ELCD, SIM) Endosulfan (beta) (ELCD, SIM) Endosulfan sulfate (ELCD, SIM)

205 ± 10 16 ± 2 34 ± 4 14 ± 2

153 ± 47 26 ± 4 47 ± 5 21 ± 6

Tomato

Chlorothalonil Endosulfan (alpha) Endosulfan (beta) Endosulfan sulfate 1 15-min injection (splitter) found these

2 50-min injections found these

FDA quant results

15

ing found endosulfan sulfate at 14/21 ppb (pg/µL) in tomato. Agilent MSD/DRS also found this compound in full-scan mode with less than half the amount reported by FDA/CFSAN available at the MSD due to the split. Several other target compounds that did not contain any halogens or the organophosphorus moeity at the low ppb concentrations were also identified by DRS, such as carbaryl (C12H11NO2) in peach and azoxystrobin (C22H17N3O5) in ginseng. The two FDA/CFSAN procedures for organohalogen and organophosphorus pesticides never would have been able to detect these additional nitrogen-containing pesticides. This shows that deconvolution of data acquired with TID is capable of identifying compounds below 10 pg on column in full-scan mode.

Conclusions The trade-off in trace-level pesticide residue analysis is sensitivity versus confirmation. Therefore, the common practice is to use element-selective GC detectors to screen the extracts and use MS/SIM to confirm hits found by GCs. This can take as many as four injections to have a complete residue analysis from a sample extract. Recent introduction of hardware and software tools, which include the capillary flow three-way splitter, trace ion detection, and deconvolution reporting software, can increase productivity dramatically. With deconvolution the demand for chromatographic resolution is lowered; therefore, the Agilent system can run the analysis at a 3x faster speed to further increase productivity. A single-injection approach even at the 3x fast speed can replace the three-injection approach.

Department of Food and Agriculture,” Fresenius J. Anal. Chem, 1991, 339, 376–383 2. J. Cook, M. P. Beckett, B. Reliford, W. Hammack, and M. Engel, “Multiresidue Analysis of Pesticides in Fresh Fruits and Vegetables Using Procedures Developed by the Florida Department of Agriculture and Consumer Services,” J. AOAC Int, 1999, 82, 1419–1435 3. G. E. Mercer, “Determination of 112 Halogenated Pesticides Using Gas Chromatography/ Mass Spectrometry with Selected Ion Monitoring,” J. AOAC Int, 2005, 88, 1452–1462 4. G. E. Mercer and J. A. Hurlbut, “A Multiresidue Pesticide Monitoring Procedure Using Gas Chromatography/Mass Spectrometry and Selected Ion Monitoring for the Determination of Pesticides Containing Nitrogen, Sulfur, and/or Oxygen in Fruits and Vegetables,” J. AOAC Int, 2004, 87, 1224–1236 5. USDA Pesticide Data Program Analytical Methods: http://www.ams.usda.gov/science/pdp/ Methods.htm 6. J. Fillion, R. Hindle, M. Lacroix, and J. Selwyn, “Multiresidue Determination of Pesticides in Fruit and Vegetables by Gas ChromatographyMass-Selective Detection and Liquid Chromatography with Fluorescence Detection,” J AOAC Int, 1995, 78, 1252–1266 7. S. Nemoto, K. Sasaki, S. Eto, L. Saito, H. Sakai, T. Takahashi, Y. Tonogai, T. Nagayama, S. Hori, Y. Maekawa, and M. Toyoda, “Multiresidue Determination of 110 Pesticides in Agricultural Products by GC/MS(SIM),” J. Food Hyg. Soc, Japan, 2000, 41, 233–241

A table comparing the results from the current FDA/CFSAN multi-instrument approach and the new Agilent single-injection approach shows that not only is Agilent’s fast analysis capable of finding all the target analytes, but it can also do it in just one-tenth of the current FDA/CFSAN total analysis time.

8. G. F. Pang, C. L. Fan, Y. M. Liu, Y. Z. Cao, J. J. Zhang, X. M. Li, Z. Y. Li, Y. P. Wu, and T. T. Guo, “Determination of Residues of 446 Pesticides in Fruits and Vegetables by Three-Cartridge SolidPhase Extraction Gas Chromatography-Mass Spectrometry and Liquid ChromatographyTandem Mass Spectrometry,” J. AOAC Int, 2006, 89, 740–771

References

9. M. Anastassiades, S. J. Lehotay, D. Stajnbaher, and F. J. Schenck, “Fast and Easy Multiresidue Method Employing Acetonitrile Extraction/Partitioning and ‘Dispersive Solid-Phase Extrac-

1. S. M. Lee, M. L. Papathakis, H. C. Feng, G. F. Hunter, and J. E. Carr, “Multipesticide Residue Method for Fruits and Vegetables: California

16

tion’ for the Determination of Pesticide Residues in Produce,” 2003, J. AOAC Int, 86:412–431 10.S. J. Lehotay, K. Maštovská, and A.R. Lightfield, “Use of Buffering and Other Means to Improve Results of Problematic Pesticides in a Fast and Easy Method for Residue Analysis of Fruits and Vegetables,” 2005, J. AOAC Int, 88:615–629 11.J. W. Wong, M. K. Hennessy, D. G. Hayward, A. J. Krynitsky, I. Cassias, and F. J. Schenck, “Analysis of Organophosphorus Pesticides in Dried Ground Ginseng Root by Capillary Gas Chromatography-Mass Spectrometry and -Flame Photometric Detection,” J. Agric. Food Chem, 2007, 55, 1117–1128 12.Chin-Kai Meng and Bruce Quimby, “Identifying Pesticides with Full Scan, SIM, µECD, and FPD from a Single Injection,” Agilent Technologies publication, 5989-3299, July 2005 13.Chin-Kai Meng, “Improving Productivity and Extending Column Life with Backflush,” Agilent Technologies publication, 5989-6018EN, December 2006

17. 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, 5989-1157EN, May 2004 18.Philip L. Wylie, “Screening for 926 Pesticides and Endocrine Disruptors by GC/MS with Deconvolution Reporting Software and a New Pesticide Library,” Agilent Technologies publication, 5989-5076EN, April 2006 19.Mike Szelewski and Chin-Kai Meng, “New Features of Deconvolution Reporting Software Revision A.02,” Agilent Technologies publication, 5989-4159EN, November 2005 20.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, 5989-4834EN, February 2006

Acknowledgements

14.Matthew Klee, “Simplified Backflush Using Agilent 6890 GC Post Run Command,” Agilent Technologies publication, 5989-5111EN, June 2006

The authors would like to thank Jon Wong, FDA/CFSAN, for the sample extracts and results used in this study as well as valuable feedback on this application.

15.Randy Roushall and Harry Prest, “The 5975C Series MSDs: Method Optimization and Trace Ion Detection,” Agilent Technologies publication, 5989-6425EN, March 2007

For More Information

16.http://chemdata.nist.gov/mass-spc/amdis/ explain.html

For more information on our products and services, visit our Web site at www.agilent.com/chem.

17

www.agilent.com/chem

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. 2007 Printed in the USA December 18, 2007 5989-7670EN

Reducing Analysis Time Using GC/MSD and Deconvolution Reporting Software

Application Food and Flavors

Authors Mike Grady, Steve Morrison, and Bob Deets Campbell Soup Company Campbell Place Camden, NJ 08103 USA Mike Szelewski Agilent Technologies, Inc. 2850 Centerville Road Wilmington, DE 19808 USA

Abstract Analyzing a complex matrix can be accomplished using multiple specific detectors, but a significant time savings is realized using GC/MSD. Using an available Retention Time Locked database with Deconvolution Reporting Software (DRS) adds a second “expert” opinion. GC/MSD with DRS provides the fastest methodology to the fewest number of false positives/negatives with the greatest confidence in the results.

Introduction Analyzing a complex matrix can be accomplished using multiple specific detectors such as electron capture (ECD), nitrogen-phosphorus (NPD), and

dual flame photometric for phosphorus and sulfur (DFPD). Most of the analytes, depending on their molecular formula, could be run on two of these specific detectors for confirmation. However, that would take twice the time. A second choice is to run each sample on unlike columns of differing polarities to the same type detector. Both of these approaches, using ECD, NPD, and/or DFPD, involve more sample handling/tracking, usually twice the number of runs and subsequent data interpretation, and still only have retention time as an identifier. Some methodologies require final confirmation by GC/MSD. Also, it is very difficult to analyze for hundreds of compounds using a nonMS method due to peak overlaps. Campbell Soup Company has been moving from GC-specific detector analyses to GC/MSD analyses for the above-mentioned reasons. A significant time savings can be realized using GC/MSD. It is imperative, however, that data quality be maintained while increasing productivity. Agilent Technologies recently introduced Deconvolution Reporting Software (DRS) for use with a GC/MSD system [1]. DRS automatically combines the results from the MSD ChemStation Quantitation with AMDIS Deconvolution and NIST Search into an easy-to-read report. Using an available Retention Time Locked database reduces methods development for DRS and speeds data comparison among labs. Positives found by normal quantitation are confirmed, and the analyst is directed to further target compound hits to verify. The analyst

need not spend the time to review hundreds of possible compounds, which could take hours per sample. GC/MSD with DRS provides the fastest methodology to the fewest number of false positives/negatives with the greatest confidence in the results. It typically takes less than 2 minutes to process a DRS Report. The purpose of this application note is to show the use of GC/MSD with DRS on complex samples in Campbell’s Research Lab. It is also intended to show a time savings while maintaining data quality.

Experimental Instrument Operating Parameters The instrument operating parameters are listed in Table 1. These conditions may have to be optimized for use in another laboratory. A Programmable Temperature Vaporizing inlet (PTV) is used in the solvent vent mode (SV). The sample is injected at or below the boiling point of the solvent, in this case 50 °C. Solvent is evaporated through the split vent line with helium at 200 mL/min for 0.3 min. At 0.5 min, the PTV is rapidly heated to 320 °C, transferring the analytes, with minimum solvent, onto the column. The column is held at the initial temperature of 70 °C during this process. The PTV-SV allows larger volumes of sample to be injected, 10 µL for this study, versus the typical 1 µL for this column. The PTV inlet liner, 5183-2037, is multi-baffled and deactivated. It does not contain glass wool, which could contribute to active compound degradation. This liner has sufficient capacity to accommodate the 10 µL injection volume

2

The HP-5MS column was used by Agilent to develop the original method and is run in constant pressure mode. Constant pressure methods can be precisely scaled, or sped up, for faster analyses. The retention times of 927 analytes have been recorded on this column. The system is Retention Time Locked to methyl chlorpyrifos at 16.596 min. The primary benefit of RTL for a food laboratory is the ability to maintain retention times after clipping or changing the column. Other benefits include: 1) constant quantitation database and integration events times; 2) switching group times remain constant for laboratories performing SIM analyses; 3) multi-site laboratories can easily compare data; 4) commercially available RTL databases can be used. Additional information is available at www.agilent.com/chem, with application notes detailing the numerous benefits of RTL. The injector parameters are modified from the default “Fast Injection” to “Variable”. This allows matching the injector parameters to the PTV-SV requirements. Most importantly, the injection speed is slowed to accommodate the evaporation of the solvent during injection. The 5973N MSD had been upgraded to an inert source and Performance Electronics. The inert source has shown improved response and less degradation for active compounds. Performance electronics minimize noise at faster Scan sampling rates and allow shorter dwell times and more ions/group for SIM acquisitions. A sampling rate of 2 and a scan range of 35-500 yields 3.12 scans/sec. This results in at least 10 data points across the earliest narrow peaks that are 0.055 min wide. This is a good number of points for both quantitation and deconvolution while minimizing noise.

Table1.

Gas Chromatograph and Mass Spectrometer Conditions

GC

Agilent Technologies 6890N

Inlet

EPC PTV

Mode

Solvent vent

Temperature ramp Initial Ramp 1 Ramp 2

°C/min

Cryo

On

Cryo use temperature

50 °C

Cryo timeout

10.00 min (On)

Cryo fault

On

Pressure

17.98 psi (On)

Vent time

0.30 min

Vent flow

RTL

System Retention Time Locked to methyl chlorpyrifos at 16.596 min

Front injector

200.0 mL/min

Sample washes Sample pumps Injection volume Syringe size PreInj Solv A washes PreInj Solv B washes PostInj Solv A washes PostInj Solv B washes Viscosity delay Plunger speed Injection speed Draw speed Dispense speed PreInjection dwell PostInjection dwell

0 3 10.00 µL 25.0 µL 0 1 2 2 1s Variable 50.00 µL/min 600.00 µL/min 1000.00 µL/min 0.00 min 0.00 min

Vent pressure

0.0 psi

MSD

Agilent Technologies 5973N

Purge flow

400.0 mL/min

Upgrades

Purge time

1.00 min

Total flow

403.9 mL/min

Gas saver

On

Saver flow

20.0 mL/min

Saver time

2.00 min

Gas type

Helium

PTV liner

Agilent PTV Liner part# 5183-2037

Oven

120 V

Oven ramp Initial Ramp 1 Ramp 2 Ramp 3

°C/min

Inert source and Performance Electronics 4.00 min 35 amu 500 amu 50 2 3.12 150 °C 230 °C 280 °C Autotune

Total run time

41.87 min

Equilibration time

0.5 min

Oven max temperature

325 °C

Column

Agilent Technologies HP 5 MS, part# 19091S-433

Length Diameter Film thickness Mode Pressure Nominal initial flow Inlet Outlet Outlet pressure

30.0 m 0.25 mm 0.25 µm Constant Pressure 17.98 psi 1.9 mL/min Front MSD Vacuum

600 50

25 3 8

Next °C 50 320 200

Next °C 70 150 200 280

Hold min 0.50 3.00 0.00

Hold min 2.00 0.00 0.00 10.00

Solvent delay Low mass High mass Threshold Sampling Scans/sec Quad temperature Source temperature Transfer line temp Tune Type Calibration Standards

Prepared from certified reference standards available from ChemServe and Crescent Chemical Company. All standards were corrected for purity.

3

Extraction Procedure An appropriate amount of commodity is weighed, typically 10-15 grams. Surrogates and, if necessary, fortification standards (spike) are added. The commodity is extracted with 1% acetic acid in acetonitrile, centrifuged [2], and passed through an SPE cartridge [3]. Analytes are eluted from the cartridge using acetonitrile/toluene. A 1-gram volume equivalent is taken from the eluant and internal standard(s) is added. The extract is brought to near dryness and solvent exchanged into ethyl acetate for GC/MSD analysis.

Results Six commodities - apples, lettuce, carrots, celery, green peppers, strawberries, and tomatoes - were purchased at local supermarkets. They were extracted and analyzed using the GC/MSD conditions described earlier. Aldrin was added to each during the extraction process and acts as both a surrogate standard (SS) and an internal standard (IS). The datafiles were processed using Agilent’s Deconvolution Reporting Software. DRS automatically combines the results from the MSD ChemStation Quantitation with AMDIS Deconvolution and NIST Search into an easy-to-read report. The results are shown in Table 2. Each of the samples showed at least one residue, ranging from Table 2. Market Basket Commodity Results Commodity Apple Lettuce Compound Table R.T. Methamidopho Acephate Diphenylamine Chlorothalonil Carbaryl Metalaxyl Malathion Isodrin Thiabendazole Captan Phosmet Permethrin Cyfluthrin Cypermethrin Fenvalerate

5.655 7.690 10.516 14.784 16.806 17.337 18.800 20.031 20.939 21.227 28.504 31.369* 32.218* 32.690* 34.271*

Addional DRSonly compounds * First R.T. of multiple isomers.

4

Carrot

trace quantities (< 0.02 ppm) to 0.48 ppm. In most cases, the trace quantities were not found by the GC/MSD standard quantitation using 3 qualifier ion identification. As a verification of the methodology, a second sample of strawberries and tomatoes were each fortified with six analytes at the 0.1 ppm level, together with the Aldrin SS/IS. The analyses’ results for these spiked samples are shown in Table 3. There are excellent recoveries for most analytes. Responses for two analytes in tomato, atrazine, and permethrin were higher than expected and could be due to matrix enhancement during injection. Time constraints did not allow for further investigation. Duplicate results are also shown in Table 3. The chlorothalonil in tomato at 0.08 ppm compares favorably with the 0.09 ppm found in the first sample. The same is true for captan at 0.47 ppm in the duplicate versus 0.48 ppm in the original sample. The GC/MSD system was calibrated for more than 50 compounds. Using DRS, the analyst can get a verification of the presence of those 50 compounds together with an automated “expert second opinion” of other compounds that may be present. This second opinion is in two distinct parts. First, the deconvolution of the complex TIC with subsequent matching of clean spectra to a database is provided. Second, the matching of these clean spectra to an independent database, in this case the NIST05a library of > 163,000 compounds. Celery

Gr pepper

Strawberry

Tomato

0.04 0.23

t t 0.02

t

0.09

0.02 t t t 0.02 0.48 0.03 t t t t 8

t = Trace quantity

0.03

0

1

3

t

13

10

1

Table 3.

Spiked and Duplicate Commodity Results, ppm Spikes

Compound Table R.T. Atrazine 13.159 Lindane 13.461 Carbaryl 16.806 Linuron 18.187 Parathion 19.275 Permethrin I 31.369 Permethrin II 31.550 Chlorthalonil Captan

Strawberry 0.15 0.11 0.11 0.13 0.09 0.14 0.15

14.784 21.227

Tomato 0.25 0.11 0.15 0.19 0.13 0.26 0.25 Duplicates 0.08

0.47

A portion of the DRS Report for celery is shown in Figure 1. Chlorothalonil was found at 14.823 minutes at 0.02 ppm. AMDIS verified chlorothalonil with a 97 match eluting only 2.7 seconds from its expected RTL time. NIST search further verified chlorothalonil with a 93 match, as the third hit out of the top 100 hits. Aldrin and permethrin are similarly verified with permethrin below the normal reporting limit. Malathion was found by AMDIS and verified by NIST. It was not found by ChemStation because one of three qualifier ions was out of range. AMDIS mitigates this problem because deconvolved spectra are cleaned of interferences and full spectrum matching is used. It would be nearly impossible to identify the analytes of interest in the presence of > 650 individual components in the celery without deconvolution. The TIC for celery is shown in Figure 2.

For all of the commodities tested, DRS verified the presence of all the calibrated peaks found by the ChemStation. Most laboratories only calibrate a fixed number of compounds, say 50-100, as it is not practical to calibrate for all 927. At the same time, these laboratories are interested in identifying other compounds that may be present in samples. Numerous uncalibrated compounds were identified using the 927 compound DRS database. The number of these additional compounds is shown on the last line in Table 2, excluding phthalates, cresols, and sulfur. When these are important to the laboratory, the GC/MSD system can, of course, be calibrated using additional standards. If an estimate of the amount is needed, or if a standard is unavailable, an average response factor can be used. The DRS database provides an average response factor for all 927 compounds. In contrast to the above methodology is the use of multiple element specific detectors such as ECD, NPD, and DFPD. Most of the analytes could be run on two of these specific detectors, but that would take twice the time. Also, it is very difficult to analyze for 927 compounds using a non-MS method due to peak overlaps. A second choice is to run each sample on unlike columns of differing polarities to the same type detector. Both of these approaches, using ECD, NPD, and/or DFPD, involve more sample handling/tracking, usually twice the number of runs, and still only have retention time as an identifier.

MSD Deconvolution Report Sample Name: celery Data File: C:\msdchem\1\Data\sjm03.D Date/Time: 02:30 PM Friday, Jan 19 2007 The NIST library was searched for the components that were found in the AMDIS target library. R.T.

CAS #

14.823 18.5992 18.7799 31.3134

1897456 309002 121755 52645531

Figure 1.

Compound Name Chlorothalonil Aldrin Malathion Permethrin I

Agilent ChemStation Amount (ppm) 0.02 1 0.03

Match

AMDIS R.T. Diff Sec

97 99 59 71

2.7 4.3 -1.2 -3.3

NIST Reverse Match Hit Num. 91 93 47 53

3 1 1 5

DRS report for celery.

5

www.agilent.com/chem

Aldrin

Chlorothalonil Permethrin Malathion

4

6

Figure 2.

8

10

12

14

16

18

20

22

24

26

28

30

32

Market basket celery total ion chromatogram.

Conclusions Analyzing a complex matrix can be accomplished using multiple specific detectors, but a significant time savings is realized using GC/MSD. Many more analytes can be determined simultaneously using mass spectra for confirmation. Using an available Retention Time Locked database reduces methods development and speeds data comparison among labs. Deconvolution Reporting Software adds a second “expert” opinion. Positives found by normal quantitation are confirmed, and the analyst is directed to further target compound hits to verify. GC/MSD with DRS provides the fastest methodology to the fewest number of false positives/negatives with the greatest confidence in the results.

References 1. “Comprehensive Pesticide Screening by GC/MSD using Deconvolution Reporting Software,” Philip L. Wylie, Michael J. Szelewski, and Chin-Kai Meng, Agilent Technologies Pub # 5989-1157EN.

2. “Fast and Easy Multiresidue Method Employing Acetonitrile Extraction/Partitioning and ‘Dispersive Solid-Phase Extraction’ for the Determination of Pesticide Residues in Produce,” M. Anastassiades, S. J. Lehotay, D. Stajnbaher, F. J. Schenck. (2003) J.AOAC Int. 86, 412-431. 3. “Analytical Methods for Residual Compositional Substances of Agricultural Chemicals, Feed Additives, and Veterinary Drugs in Food,” Japan Department of Food Safety, Ministry of Health, Labour and Welfare.

For More Information For more information on our products and services, visit our Web site at www.agilent.com/chem. 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. 2007 Printed in the USA May 14, 2007 5989-6677EN

Rapid Analysis of Herbicides by Rapid Resolution LC with Online Trace Enrichment Application

Environmental

Author Michael Woodman Agilent Technologies, Inc. 2850 Centerville Road Wilmington, DE 19808-1610 USA

Abstract Environmental and Food Safety agencies are constantly updating methods to improve detection limits and to resolve interfering compounds. One particular method, EPA 555, is used for the analysis of chlorinated phenoxy acid herbicides in drinking water. A mandated trace enrichment step significantly impacts the ease of use and reliability of the method. The method uses 5-µm analysis columns and online trace enrichment. The variation here uses small ZORBAX 3.5- and 1.8-µm RRHT columns and an autoSPE (Solid Phase Extraction) cartridge with an automated switching valve mounted in the column compartment. Combined with sample introduction via direct injection to the autoSPE cartridge, instead of the loading pump specified in the EPA method, we dramatically reduce the overall analysis time and virtually eliminate the potential of sample cross-contamination.

Introduction Trace analyte detection in relatively clean matrices is an excellent application for online SPE procedures. Compared to manually loading samples with

disposable SPE cartridges, which require an elution step into a vial prior to analysis, online SPE assures 100% sample transfer to the analysis column and dramatically increases sensitivity by increasing the analyte mass delivered to the column. In EPA Method 555, 20 mL of drinking water is loaded through a pump to an SPE cartridge mounted on a high-pressure switching valve on the HPLC system. Because few, if any, autosamplers can inject this large volume, the sample must be pumped onto the cartridge. Contamination of the loading pump with prior samples is always a concern, and adequate flushing and blank runs become an important part of the overall method procedure. To reduce the sampling volume sufficient for available automatic preparative samplers, without losing sensitivity in the method, it is necessary to reduce the analysis column size while preserving resolving power. Ancillary benefits of using smaller columns generally include reduced analysis time and solvent consumption, and greater compatibility with ionization sources in mass spectrometers. If the ratios of their column length to particle size are equal, columns are considered to have equal resolving power. Significant reductions in column volume can be made by reducing the length and/or internal diameter of the column. In the latter case, the flow rate would normally be reduced as in Equation 1.

Flowcol. 1 ×

Diam.column2 Diam.column1

2

= Flowcol. 2

(eq. 1)

The combined effect of reduced length and diameter contributes to a reduction in solvent consumption. We normally scale the injection mass to the size of the column and a proportional injection volume would be calculated from the ratio of the void volumes of the two columns, multiplied by the injection volume on the original column, as in Equation 2 below.

Inj. vol.col. 1 ×

Volumecolumn2 Volumecolumn1

= Inj. vol.col. 2

(eq. 2)

Short columns packed with small particle sizes are typically operated at high linear velocities. The increase in elution speed will decrease absolute peak width and may require the user to adjust data acquisition rates and reduce signal filtering parameters. This will ensure that the chromatographic separation is accurately recorded in the acquisition data file. For gradient elution separations, where the mobile phase composition increases through the initial part of the analysis until the analytes of interest have been eluted from the column, successful method conversion to smaller columns requires that the gradient slope be preserved. We can express the gradient slope as in Equation 3.

% Gradient slope =

(End% – Start%) #Column volumes

Experimental Conditions See figure 1 for configuration. System Agilent 1200 Series Rapid Resolution LC consisting of: G1379B micro degasser G1312B binary pump SL G1312A binary pump with solvent selection valve option, or G1354A quaternary pump G1367C HiP ALS autosampler SL, and G2258A Dual Loop Prep autosampler 5 ml G1316B Thermostatted column compartment SL with 6- or 10-port 2-position switching valve G1315C UV/VIS diode array detector (DAD) SL, flow cell as indicated in individual chromatograms ChemStation 32-bit version B.02.01 Columns Agilent ZORBAX SB-C18, 4.6 × 250 mm, 5 µm Agilent ZORBAX SB-C18, 3.0 × 150 mm, 3.5 µm Agilent ZORBAX SB-C18, 2.1 × 80 mm, 1.8 µm Agilent ZORBAX SB-Aq, 4.6 × 12.5 mm, 5 µm Mobile phase conditions Organic solvent: Aqueous solvent:

Acetonitrile 25 mm phosphoric acid in Milli-Q water

Gradient conditions Gradient slope:

7.8 or 2.3% per column volume, as indicated. See individual chromatograms for flow rate and time

Sample (eq. 3)

Note that the use of % change per column volume rather than % change per minute frees the user to control gradient slope by altering gradient time and/or gradient flow rate. A large value for gradient slope yields very fast gradients with minimal resolution, while lower gradient slopes produce higher resolution at the expense of increased solvent consumption and somewhat reduced sensitivity. Longer analysis time may also result unless the gradient slope is reduced by increasing the flow rate, within acceptable operating pressure ranges, rather than by increasing the gradient time. Resolution increases with shallow gradients because the effective capacity factor, k*, is increased. Much like in isocratic separations, where the capacity term is called k', a higher value directly increases resolution. The effect is quite dramatic up to a k value of about 5–10, after which 2

little improvement is observed. In the subsequent examples, we will see the results associated with the calculations discussed above.

EPA 555 Group A chlorinated phenoxy acid herbicides (picloram, chloramben, dicamba, bentazon, 2,4-D, dichlorprop, 2,4,5-TP, acifluorfen), 100 µg/mL in methanol or diluted to 20 ng/L (20 ppb) in reagent water acidified with 25 mm phosphoric acid.

Results The separation was initially performed via direct injection of concentrated standard on a 4.6 × 250 mm, 5-µm ZORBAX SB-C18 column, thermostatted to 25 °C, using conditions referenced in US EPA method 555 (Figure 2). The described trace enrichment procedure using pump A as the loading pump was performed (Figure 3). The method was then scaled in flow and time for exact translation to a 3.0 × 150 mm 3.5-µm column (Figure 4) using 5-mL trace enrichment injection. Finally, a 2.1 × 80 mm 1.8-µm configuration (50-mm plus 30-mm columns in series) is used to demonstrate the feasibility of this separation under conditions for trace enrichment requiring less than 1.5-mL injection. (Figure 5)

Load/Wash position

Elute/Analyze position

2 Position/6 Port valve

Trace enrichment autoSPE scheme.

Figure 1 shows the schematic placement of modules and columns in the system. The A pump is the loading pump in case of volumes exceeding the 5-mL capacity of the G2258A Dual Loop Autosampler, thus pump A uses one line for sample and a second line for the aqueous eluent, 25 mm phosphoric acid. If direct injection from the autosampler is used, pump A is delivering 25 mm phosphoric acid. If the A pump is fitted with a degasser, the sampling line should bypass the degasser module to minimize contamination with sample solutions. To conduct sampling through the A pump, the valve should be in position B while the new sample is flushed through the A pump. Then switch the valve to the A position and load the required 20 mL sample volume. The analysis

29.609

17.653

13.303

20

24.678 - Acifluorfen

40

20.238

17.845 - Dicamba

60

23.095 - 2,4,5-TP

80

19.480 - 2,4-D

100

Figures 2 and 3 show the standard separation by direct injection and pumped trace enrichment, respectively. With column regeneration steps, this results in a total analysis time of 60 minutes. Translation of the gradient to the 3.0- × 150-mm column requires a reduction in flow rate, due to the smaller diameter, and a reduction in gradient time because of the shorter column length. The resulting analysis is reduced from 60 to 36 minutes and solvent consumption is proportionately reduced from 60 mL to 15.5 mL.

Conditions EPA Method 555 with ZORBAX SB-C18 columns and fast DAD detector ZORBAX SB-C18 4.6 mm × 250 mm, 5 µm Column temp: 25 °C Gradient: 25 mM H3PO4, ACN, 10% to 90% ACN in 30 min Gradient slope: 7.8% ACN/column volume Analysis flow rate: 1 mL/min Group A compounds: 1 µL of 100 µg/mL Total analysis time: 60 min Detection: UV 230 nm, 10-mm 13-µL flow cell, filter 2 seconds (default) 18.895 - Bentazon

12.679 - Picloram

14.436 - Chloramben

mAU 120

begins when the valve is returned to the B position, at which time the sampling line on the A pump would be flushed with reagent water or the next sample, as appropriate.

21.104 - Dichlorprop

Figure 1.

0 10

Figure 2.

12

14

16

18

20

22

24

26

28

min

Gradient separation of herbicides on 4.6 mm × 250 mm, 5 µm ZORBAX SB-C18.

3

150

21.720

18.439

17.607

13.194

50

27.714

100

29.595 29.786

24.667 - Acifluorfen

200

23.050 - 2,4,5-TP

250

21.063 - Dichlorprop

18.871 - Bentazon

300

19.414 - 2,4-D

350

17.779 - Dicamba

14.380 - Chloramben

12.557 - Picloram

mAU

0 10

12

14

16

18

20

22

24

26

28

min

Conditions EPA Method 555 with ZORBAX SB-C18 columns and fast DAD detector ZORBAX SB-C18 4.6 mm × 250 mm, 5 µm Column temp: 25 °C Gradient: 25-mM H3PO4, ACN, 10% to 90% ACN in 30 min Gradient slope: 7.8% ACN/column volume Analysis flow rate: 1 mL/min Group A compounds: 20 mL of 20 µg/L trace enrichment Total analysis time: 60 min Detection: UV 230 nm, 10-mm 13-µL flow cell, filter 2 seconds (default)

10.352

80

9.211

7.073

100

7.870

120

Conditions: EPA Method 555 with ZORBAX SB-C18 columns and fast DAD detector ZORBAX SB-C18 3 mm × 150 mm, 3.5 µm Column temp: 25 °C Gradient: 25 mm H3PO4, ACN, 10% to 90% ACN in 18 min Gradient slope: 7.8% ACN/column volume Analysis flow rate: 0.43 mL/min Group A compounds: 5 mL of 20 µg/L (20 ppb) Detection: UV 230 nm, 3-mm 2-µL flow cell, filter 0.2 seconds Total analysis time: 36 min

5.118

4.094

mAU

Trace enrichment (20 mL) of 20-ppb solution on 4.6 × 250 mm 5-µm ZORBAX SB-C18.

8.221

Figure 3.

60

11.555

6.642

4.462

20

14.041

40

0 2

Figure 4.

4

6

8

12

14

min

Trace enrichment (5 mL)of 20-ppb solution on 3.0 × 150 mm, 3.5-µm ZORBAX SB-C18.

The last peak in Figure 4 is missing due to a valve timing error that was not detected until sometime after the lab work was completed. Peak 8 was not eluted from the trace enrichment column before 4

10

the valve switched offline for regeneration and equilibration. Note the baseline shift that occurs after peak 7, not seen in other autoSPE examples.

300

2.723

400

2.577

2.048

500

2.385

600

2.161 2.206

700

1.672

1.532

mAU

Conditions EPA Method 555 with ZORBAX SB-C18 columns and fast DAD detector ZORBAX SB-C18, 2.1 mm × 80 mm (50mm + 30mm in series), 1.8 µm Column temp: 50 °C Gradient: 25-mM H3PO4/ACN, 10% to 90% ACN in 2.7 min 7.8% ACN/column volume Analysis flow rate: 0.72 mL/min Detection: UV 230 nm, 3-mm 2-µL flow cell, filter 0.2 seconds Sample: Aged 1 µL 100 µg/mL Total analysis time: 6 min

3.025

2.830

2.626

2.310

1.121

100

1.643

200

0 1.25

Figure 5.

1.5

1.75

2

2.25

2.5

2.75

3

3.25

min

High-speed gradient separation of herbicides on 2.1 × 80 mm, 1.8-µm ZORBAX SB-C18.

In Figure 5 we see the combination of highest speed and resolution, using the full capability of the 1200 Rapid Resolution LC. Operating pressure was, at the maximum point, about 520 bar. We maintain comparable resolution to the original 4.6 × 250 mm, 5-µm method, a 60-minute run time, with an analysis time of only 6 minutes.

Conclusion As is the case for many existing methods, it is both possible and practical to modernize this method to improve throughput and overall performance. Here we see the potential for a 10-fold increase in analysis speed and elimination of the loading pump scheme found in the original method. Approximately 1.3 mL of sample solution, injected to the autoSPE setup using the 2.1 × 80 mm configuration, is all that is needed to replace the 20-mL injection previously loaded through the pump. This approach can greatly improve productivity and ensure minimal sample cross-contamination.

For More Information For more information on our products and services, visit our Web site at www.agilent.com/chem.

5

www.agilent.com/chem

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. 2007 Printed in the USA March 30, 2007 5989-5176EN

Improving Productivity and Extending Column Life with Backflush

Application Brief Chin-Kai Meng

All Industries

A previous application note [1] has shown that multiple GC signals and MS signals can be acquired from a single sample injection. When a 3-way splitter is connected to the end of a column, column effluent can be directed proportionally to two GC detectors as well as the MSD. This multi-signal configuration provides full-scan data for library searching, SIM data for quantitation, and element selective detector data for excellent selectivity and sensitivity from complex matrices. The system used in this study consists of a 7683ALS, a 7890A GC with split/splitless inlet, 3-way splitter, µECD, dual flame photometric detector (DFPD), and a 5975C MSD. Figure 1 shows four chromatograms from a single injection of a milk extract. The synchronous SIM/scan feature of the 5975C MSD provides data useful for both screening (full scan data) and quantitation (SIM data). DFPD provides both P and S signals without the need to switch light filters. Noticeably in the full scan TIC in Figure 1, a significant number of matrix peaks were observed after 32 minutes. It is not uncommon to add a “bake-out” oven ramp to clean the column after analyzing complex samples. The bake-out period is used to quickly push the late eluters out of the column to be ready for the next injection. Therefore, it is common to use a higher oven temperature than required for the analysis and an extended bake-out period at the end of a normal Full scan TIC

SIM

µECD

DFPD(P)

5

Figure 1.

10

15

20

25

30

35

40

Four chromatograms collected simultaneously from a single injection of a milk extract.

Highlights •

Backflush – a simple technique to remove high boilers from the column faster and at a lower column temperature to cut down analysis time and increase column lifetime.

•

The milk extract example shows that a 7-minute 280 °C backflush cleaned the column as well as a 33-minute 320 °C bake-out. The cycle time was reduced by more than 30%.

•

Using backflush, excess column bleed and heavy residues will not be introduced into the MSD, thus reducing ion source contamination.

www.agilent.com/chem over program to clean out the column, which adds to the cycle time and shortens the column lifetime. Adding the bake-out period to the milk extract analysis, additional matrix peaks were observed even up to 72 minutes, while target compounds already eluted before 42 minutes. This means that 30 minutes were lost in productivity for each injection. Backflush [2] is a simple technique to drastically decrease the cycle time by reversing the column flow to push the late eluters out of the inlet end of the column. Late eluters stay near the front of the column until the oven temperature is high enough to move them through the column. When the column flow is reversed before the late eluters start to move down the column, these late eluters will take less time and at a lower oven temperature to exit the inlet end of the column. There are many benefits in using backflush: •

Cycle time is reduced (no bake-out period, cooling down from a lower oven temperature)

•

Column bleed is reduced (no high-temperature bake-out needed), resulting longer column life

•

Ghost peaks are eliminated (no high boilers carryover into subsequent runs)

•

Contamination that goes into the detector is minimized, which is especially valuable for the MSD (less ion source cleaning)

Figure 2 shows three total ion chromatograms from the Agilent 7890A GC/ 5975C MSD. The top chromatogram is a milk extract analysis with all the target compounds eluted before 42 minutes (over program goes to 280 °C). However, an additional 33-minute bake-out period at 320 °C was needed to move the high boilers out of the column. This bake-out period was almost as long as the required time to elute all target compounds. The middle chromatogram is the same milk extract analysis stopped at 42 minutes with a 7-minute backflush post-run at 280 °C added to the analysis. The bottom chromatogram is a blank run after the backflushing was completed. The blank run shows that the column was very clean after backflushing. The example shows that a 7-minute backflush cleaned the column as well as a 33-minute bake-out.

of 320 °C. A column effluent splitter or QuickSwap is required to do the backflush.

References 1. Chin-Kai Meng and Bruce Quimby, “Identifying Pesticides with Full Scan, SIM, µECD, and FPD from a Single Injection,” Agilent Application Note, 5989-3299EN, July 2005. 2. Matthew Klee, “Simplified Backflush Using Agilent 6890 GC Post Run Command,” Agilent Application Note, 5989-5111EN, June 2006.

Acknowledgement Milk extract is courtesy of Dr. Steven Lehotay from USDA Agricultural Research Service in Wyndmoor, Pennsylvania, USA.

For More Information For more information on our products and services, visit our Web site at www.agilent.com/chem.

The milk extract example in Figure 2 illustrates the backflush technique in reducing cycle time and column bleed. The cycle time was reduced by more than 30% and the column was kept at 280 °C, without going to the bake-out temperature

It took an additional 33 min and heating the column to 320 °C to remove these high boilers

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. 2006 Printed in the USA December 26, 2006 5989-6018EN

Run stopped at 42 min and backflushed at 280 °C for 7 mins Blank run after backflushing showing the column was clean 5

10

Figure 2.

15

20

25

30

35

40

45

50

55

60

65

70 min

Three total ion chromatograms comparing the results with and without backflush.

Using RTL and 3-Way Splitter to Identify Unknown in Strawberry Extract Application Brief Chin-Kai Meng

Food Safety and Environmental

Fruit and vegetable extracts are usually very complex to analyze. It is common to use the very selective GC detectors, for example NPD, µECD, and FPD, to look for trace pesticide residues in the extracts. Mass spectrometry is most often used to confirm the hits from GC detectors. A previous application note [1] describes a GC/MS system with a three-way splitter added to the end of the column. The column effluent could be split three ways to two GC detectors and the MSD. The splitter system is therefore capable of providing up to four signals (two GC signals, SIM, and full-scan chromatograms) from a single injection. The combination of element selective detectors, SIM/scan, and Deconvolution Reporting Software (DRS) makes a very powerful pesticide analysis system [2]. The trade-off is the decrease of analyte concentration in any detector due to the flow splitting at the end of the column. The system used for this study consists of an Agilent 7890A GC with split/splitless inlet, a three-way splitter, µECD, dual flame photometric detector (DFPD), and 5975C MSD. Figure 1 shows chromatograms from 2 separate injections (each injection provides two GC signals) of the same strawberry extract without any hardware or filter changes. All of the target compounds were found and confirmed by DRS, GC, and MS signals except the unknown peak at about 41 minutes. The peak shows responses from µECD, DFPD(S) and DFPD(P). However, no peak was observed in the MS full-scan signal. This makes it difficult to confirm the unknown peak using the full-scan TIC. Since the analysis was retention time locked, it is therefore possible to find potential matches by examining the RTL pesticide database (part number G1672AA). The unknown compound, containing electron-capturing atoms (for example, Cl or O), P, and S atoms, would have a target retention time inside the

Highlights Splitter+an inert, easy-to-use capillary flow technology that splits column effluent to multiple detectors (for example, MSD, DFPD, and µECD). The splitter configuration provides a comprehensive screening and quantitative system. By combing RTL, element-selective detector chromatograms, and the RTL pesticide database, a trace level pesticide residue was identified without the full-scan mass spectrum.

TIC

µECD

DFPD (S)

DFPD (P)

5.00

Figure 1.

Table 1.

10.00

15.00

20.00

25.00

30.00

35.00

Unknown compound detected by GC signals not found in strawberry extract TIC.

Compound List Extracted from the RTLPest3.tab File

Name Fluthiacet-methyl Benzo[g,h,i]perylene Temephos PBB 169 hexabrombiphenyl Rotenone

CAS 117337196 191242 3383968 60044260 83794

Mol form C15H15CIFN3O3S2 C22H12 C16H20O6P2S3 C12H4Br6 C23H22O6

Mol wt 403.9 276.3 466.5 627.6 394.4

R.T. 39.10 39.13 40.74 40.93 41.70

41 \ 0.5-minute window (that is, 40.5 to 41.5 min) in the database, if it is included in the database. Table 1 is a portion of the RTLPest3.tab file1 opened in Microsoft Excel. The compound temephos at locked retention time 40.74 min meets all the criteria for the unknown peak. To further confirm peak identity, extracted ion chromatograms (EICs) of the four major ions of temephos were plotted. Figure 2 shows EICs of target ion and three qualifiers (ions 466, 125, 93, and 109 from Table 1) of temephos. Although the ion intensities were weak, the noticeable presence of all four ions at 40.9 min helped to confirm that the unknown peak was temephos.

1. The RTLPest3.tab file is created in the C:\Database directory while executing the Tools\List Screen Database… command (in MSD Enhanced Data Analysis software) and selecting the RTLPest3.scd from the C:\Database directory.

2

40.00

Target Ion 403 276 466 308 192

Q1 56 277 125 468 191

Q2 405 138 93 148 394

Q3 232 275 109 154 177

200 100 0

Ion 466

200 100 0

Ion 125

Ion 93

200 100 0

200 100 0

Ion 109

36

Figure 2.

37

38

39

40

41

42

43

44

EICs of target ion 466 (temephos) and three qualifier ions.

References 1. Chin-Kai Meng and Bruce Quimby, “Identifying Pesticides with Full Scan, SIM, µECD, and FPD from a Single Injection,” Application Note, 5989-3299, July 2005. 2. Mike Szelewski and Bruce Quimby, “New Tools for Rapid Pesticide Analysis in High Matrix Samples,” Application Note, 5989-1716, October 2004.

Acknowledgement Strawberry extract is courtesy of Dr. Steven Lehotay from USDA Agricultural Research Service in Wyndmoor, Pennsylvania, USA.

For More Information For more information on our products and services, visit our Web site at www.agilent.com/chem.

3

www.agilent.com/chem

Copyright © 2006 Agilent Technologies All Rights Reserved. Reproduction, adaptation, or translation without prior written permission is prohibited, except as allowed under the copyright laws. Printed in the USA December 13, 2006 5989-6007EN

Automated Screening of 600 Pesticides in Food by LC/TOF MS Using a Molecular-Feature Database Search Application

Food Safety

Authors E. Michael Thurman and Imma Ferrer Pesticide Residue Research Group University of Almería 04120 Almería, Spain Jerry A. Zweigenbaum Agilent Technologies, Inc. 2850 Centerville Road Wilmington, DE 19808-1610 USA

Abstract Searching a database using a molecular feature (MF) algorithm was developed for the screening of 600 pesticides and degradates in extracts of food by liquid chromatography time-of-flight mass spectrometry in positive ion mode with full-scan accurate mass spectra. The database search works by compiling the accurate mass of the ions detected and identified as real compound chromatographic peaks without ion extraction and compares them with the monoisotopic exact masses of the compounds in the database. The screening criteria consisted of ± 5 ppm accurate mass window, ± 0.2 minute retention time window, and a minimum area count of 1,000 counts (signal-to-noise ratio of ~10:1). The limit of detection and retention time was determined for 100 of the 600 compounds and varied from 147,000 compound library. The DRS results show good match quality at the locked retention times for seven target compounds. No single software package can produce this same confidence level in compound identification. An experienced analyst would take 1–4 hours to process this complex sample manually with each of the three software packages used by DRS. DRS produced this report is less than two minutes.

Backflushing Citrus oils contain significant amounts of material that elute after the last pesticide. This requires a 150-min hold at 320 °C to elute all of the heavy material with a 1× method. The total run time for the 1× method is therefore 195 min, as shown in Table 3. Table 3.

Method Run Time Comparison

Column Speed Run time Pesticides No backflush matrix With backflush matrix

30 m 1× min 42 195 n/a

15 m 3× min 14 65 20

10 m 4.8× min 8.75 40.6 12.5

The 3× method requires a 50-min hold at 320 °C, as shown at the top in Figure 6, resulting in a 65-min run time. With backflushing, all of this heavy material is removed in 5 min at 300 °C, as shown in the bottom of Figure 6. This is a 9-fold reduction in analysis time compared to the 1× method. 4.8x Method Using the 220V oven, SP1 2310-0236, and oven insert accessory, the method can be scaled to 4.8× faster, as shown in Table 3. There is a practical limit to the amount of matrix that can be tolerated with the reduced resolution using a 10-m column. However, for matrices less complex than a citrus oil, the 4.8× method can be successfully used to save even more time. The Performance Electronics allows running at faster scan speeds while maintaining signal/noise ratio. Sufficient points across a peak are maintained even with faster chromatography.

Normal bakeout: 65 min run

Backflush time range

Backflush: 19 min run

10

Figure 6.

20

30

40

50

60

Top - 3× lemon oil analysis with 50 min bakeout at 320 °C. Bottom - 3× lemon oil analysis with 5 min backflush at 300 °C.

11

www.agilent.com/chem

Conclusions New tools are available for the analysis of pesticides in complex matrices. These tools can be combined to significantly reduce analysis and data processing time. • Fast oven - programming rates needed for faster runs • Shorter column - 3× faster runs with sufficient resolution • Microfluidic splitter - confidence in results using selective detection simultaneous with MS data • Backflush - additional 3× reduction in run time with lower column temperatures for extended lifetime • Performance Electronics - maintain signal/noise at faster sampling rates • DRS - three levels of target analyte identification in less than two minutes Using the above tools, the run time for analysis of lemon oil was reduced from 195 minutes to 20 minutes (nine-fold). DRS reduced the data analysis from hours to minutes.

References 1. B. D. Quimby, L. M. Blumberg, M. S. Klee, and P. L. Wylie “Precise Time-Scaling of Gas Chromatographic Methods Using Method Translation and Retention Time Locking” Agilent Technologies publication 5967-5820E www.agilent.com/chem 2. P. Wylie, M. Szelewski, and C. K. Meng “Comprehensive Pesticide Screening by GC/MSD using Deconvolution Reporting Software” Agilent Technologies publication 5989-1157EN www.agilent.com/chem

For More Information For more information on our products and services, visit our Web site at www.agilent.com/chem. 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. 2004 Printed in the USA October 13, 2004 5989-1716EN

A Blind Study of Pesticide Residues in Spiked and Unspiked Fruit Extracts Using Deconvolution Reporting Software Application

Food

Author Christopher P. Sandy Agilent Technologies, Inc. Block A, CSC Eskdale Road Winnersh Triangle Estate Wokingham, Berk RG41 5DZ United Kingdom

Abstract The Agilent Technologies mass selective detector (MSD) coupled with deconvolution reporting software (DRS) provides additional powerful data processing capabilities to the MSD ChemStation software. Reviewing full scan gas chromatography/mass spectrometry data for the confirmation of pesticide residues can be a labor-intensive and time-consuming process requiring great skill and concentration by an experienced analyst. The DRS is able to process a complex food extract total ion chromatogram in about 1 minute, whereas an experienced analyst may take more than 30 minutes to achieve the same quality result. Extensive data shown in this report supports the high confidence level that an analyst can have in results rapidly produced by the DRS.

Introduction Typical mass spectral pesticide residue analysis requires finding target ions and meeting qualifier

ion ratios. It is sometimes very difficult to confirm target compounds from high matrix background because the matrix affects the ion ratios of the target compounds or complicates the spectrum with additional ions. To be certain of the results, background subtraction and manual integration are often practiced. It is, therefore, a timeconsuming process to confirm target compounds in a dirty matrix. It can take an experienced analyst 15 to 30 minutes to review/confirm one data file. Two powerful gas chromatography/mass spectrometry (GC/MS) techniques - Retention Time Locking (RTL) and deconvolution were combined to create a quantitation and screening tool that can identify 567 pesticides and endocrine disrupters from a single run in 1–2 minutes. The Agilent Technologies GC/MSD-DRS provides the additional functionality to the MSD ChemStation.

Experimental DRS Overview A detailed overview of the DRS is given in an application note 5989-1157EN [1], available for download at www.agilent.com/chem. The operating principles of the DRS appear in Figure 1.

2.5E7 2E7

Total ion chromatogram (TIC)

1.5E7 1E7 .5E7 5

10

15

Targets are identified by comparison to locked retention times (RTs) and three qualifying ion ratios, quantified using target ion area versus internal standard (ISTD) calibration table

20

25

AMDIS 32 deconvolutes component spectra and searches target MS database, locked RT used as a qualifier

30

35

Deconvoluted target spectra confirmed by AMDIS searched against NIST02 MS database

Confirmed AMDIS hits

Quant results

40

Confirmed NIST02 hits

Combined quantitative and qualitative HTML Summary report

Figure 1.

Schematic diagram summarizing the GC/MS DRS.

The quantitation capabilities of the MSD ChemStation are combined with the deconvolution power of the industry standard AMDIS program from NIST. AMDIS is able to separate spectra of interest from dirty matrix spectra present in samples analyzed for pesticides. A third level of confidence is obtained by sending the deconvoluted spectra for library searches of the NIST02 145,000 compound library. A comprehensive report is produced in about 1 minute.

Samples Six samples of fruit extracts, supplied in 90/10 iso-octane/toluene solvent were received for analysis by GC/MS. The samples were prepared by an accredited food pesticide laboratory based in Scandinavia. Three of the samples were spiked with a number of pesticides at varying concentration levels. Although the range of concentrations of the pesticides in each sample was given, neither the actual number of pesticides spiked into each control sample nor the identities were supplied. Details of the samples appear in Table 1. The other three samples were ‘real’, unspiked extracts.

Table 1.

Sample Details for Blind Study

Sample number 1

Matrix extracted Orange

Number of pesticides 20–40

Concn range (mg/Kg) 0.02–0.20

Comments Control sample - spiked

2

Lettuce

20–40

0.02–0.20

Control sample - spiked

3

Apple

20–40

0.01–0.20

Control sample - spiked

4

Grapes

2–4

0.1–1.0

Real sample

5

Orange

2–4

0.2–5.0

Real sample

6

Apple

2–4

0.05–2.0

Real sample

2

Instrumentation The samples were analyzed by full-scan GC/MS using the analytical conditions given in Table 2. Data processing and reporting were performed using the default settings provided with the DRS.

Table 2.

RTL GC/MS Analysis Conditions for Fruit Extract Samples

Gas chromatograph

Agilent 6890N

Column

30 m x 0.25 mm id x 0.25 µm HP-5MS (p/n 19091S-433)

Carrier gas

Helium

Flow rate

1.9 mL/min at 70 °C

Head pressure

18 psig, constant pressure mode Method RTLocked to methyl chlorpyrifos at 16.593 min

Injector type

PTV, septumless head

Injector temperature (°C), hold time (min), and ramp rate (°C/min)

90 °C (0.3 min) - 1720 °C/min - 250 °C

Vent time

0.2 min

Vent flow

30 mL/min

Vent pressure

0 psig

Purge flow

60 mL/min

Purge time

1.0 min

Syringe volume

50 µL

Injection volume

15 µL

Liner

Empty multibaffle

Oven program: temperature (°C), hold time (min), and ramp rate (°C/min)

70(2)-25-150(0)-3-200(0)-8-280(10)

MSD

Agilent 5973 inert

MS interface

280 °C

MS source

230 °C

MS quad

150 °C

Detection mode

EI, Scan 40–550 amu

EM voltage

ATUNE value

3

Results The results for the three spiked extracts appear in Table 3 - note that the details of which pesticides were added to the spiked samples were not supplied until after the results were shown to the customer. Those pesticides confirmed by the DRS, are shown lightly shaded. The analytes, shown darkly shaded, are not present in the Agilent RTL Pesticides database. Analyte entries left unshaded were not confirmed.

Table 3.

MSD-DRS Results for Three Spiked Fruit Extract Samples

Sample 1: Control-orange, spiked Pesticide Methamidofos* Dichlorvos* Acephate* Omethoate Propachlor Chlorprofam Monocrotophos Dimethoate Quintozene Parathion-methyl Dichlofluanid Fenpropimorph Triadimefon Thiabendazole Tolylfluanid Mecarbam Methidation Vamidothion Imazalil Myclobutanil Kresoxim methyl Tebuconazole Phosmet Fenpropathrin Tetradifon Azinphos-methyl Fenarimol Azinpfos-ethyl Prochloraz Flucythrinate Esfenvalerate Azoxystrobin

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 * See Discussion item 1. ** See Discussion item 2.

4

Sample 2: Control-lettuce, spiked Added mg/kg 0.10 0.10 0.10 0.10 0.20 0.10 0.10 0.04 0.02 0.10 0.10 0.10 0.04 0.10 0.04 0.10 0.10 0.10 0.10 0.10 0.10 0.10 0.10 0.04 0.04 0.10 0.10 0.10 0.10 0.10 0.04 0.04

Pesticide Diphenylamine HCB Lindane (HCH-gamma) Diazinon Chlortalonil Vinclozolin Carbaryl Metalaxyl Pirimiphos-methyl Malathion Chlorpyrifos Cyprodinil Penconazole Captan Folpet** Procymidone Endosulfan-a pp-DDE Bupirimate Endosulfan-b Aclonifen Ethion Triazophos Endosulfan-sulfate Iprodione Bromopropylate Methoxychlor Phosalone Lambda-Cyhalothrin Permethrin Cypermethrin Fenvalerate Deltamethrin

Sample 3: Control- apple, spiked Added mg/kg 0.10 0.02 0.04 0.04 0.04 0.04 0.20 0.10 0.10 0.10 0.10 0.04 0.04 0.10 0.10 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.10 0.10 0.10 0.04 0.10 0.10 0.04 0.10

Pesticide Mevinphos Trichlorfon Heptenophos Tecnazene HCH alpha HCH beta Dichloran Pyrimethanil Etrimphos Ethiofencarb Metribuzin Toclophos methyl Linuron Aldrin Diethofencarb Trichloronate Triadimenol Disulfoton sulfoxide Disulfoton sulfone Fluazinam Chlorbenzilate Oxadixyl Benalaxyl Dicofol Fenazaquin Pyrazophos Acrinathrin Bitertanol Cyfluthrin beta Alpha cypermethrin

Added mg/kg 0.05 0.05 0.02 0.01 0.01 0.02 0.05 0.02 0.02 0.10 0.05 0.01 0.05 0.02 0.02 0.02 0.05 0.20 0.02 0.05 0.05 0.05 0.05 0.05 0.02 0.05 0.02 0.05 0.05 0.05

The results for the three ‘real’ extracts appear in Table 4. Those pesticides confirmed by the DRS are shown lightly shaded. The darkly-shaded analytes are not present in the Agilent RTL Pesticides database. Analyte entries left unshaded were not confirmed. Analytes with an associated concentration were confirmed as present by the customer using NPD/ECD. Lightly-shaded analytes without a concentration label were detected and confirmed by the DRS, but not by the customer. Table 4. MSD-DRS Results for Three ‘Real’ Fruit Extract Samples Sample 4: Grapes 0.68 mg/Kg Captan 0.21 mg/Kg Cyprodinil 0.27 mg/Kg Fludioxinil Diphenylamine Sample 5: Orange 2.5 mg/Kg Imazalil 0.25 mg/Kg Medidathion 3.0 mg/Kg Thiabendazole Sample 6: Apple 0.86 mg/Kg Diphenylamine 0.05 mg/Kg Chlorpyrifos 0.79 mg/Kg Thiabendazole Dimethoate Ethoxyquin Methyl parathion Endosulfan sulfate Propargite

Discussion 1. Control - Orange spiked extract This control sample was spiked with 32 pesticides at levels ranging between 0.02 and 0.10 mg/kg. Twenty-six pesticides were detected and confirmed by the DRS software, two were not reported since they are not present in the Agilent RTL Pesticide database and four were not detected. The spiking was done to the raw matrix, not to a matrix extract. For the polar pesticides (methamidofos and acephate), the recovery was in the 20%–30% range as confirmed by NPD/ECD. Therefore, that explains why these pesticides were not detected by DRS.

2. Control - Lettuce spiked extract This control sample was spiked with 33 pesticides at levels ranging between 0.02 and 0.20 mg/kg. Twenty-nine pesticides were detected and confirmed by the DRS software, three were not reported since they are not present in the Agilent RTL Pesticide database and one was not detected. The one undetected analyte, (Folpet, marked with two asterisks in Table 3), was detected and confirmed if a higher sensitivity setting was used in the AMDIS deconvolution program. 3. Control - Apple spiked extract This control sample was spiked with 30 pesticides at levels ranging between 0.01 and 0.20 mg/kg. Twenty-two pesticides were detected and confirmed by the DRS software, six were not reported since they are not present in the Agilent RTL Pesticide database and two were not detected. Overall, of the 95 spiked analytes in the three control samples, 93% of the pesticides present in the Agilent RTL Pesticide database were detected and confirmed by full-scan library searching of the deconvoluted mass spectra. 4. ‘Real’ Grape extract The customer had detected and confirmed three pesticide residues in the Grape extract sample Captan, Cyprodinil, and Fludioxinil. Of these three analytes, Captan was confirmed by the DRS and Cyprodinil and Fludioxinil are not entries in the Agilent RTL Pesticide database. However, DRS also confirmed an additional pesticide residue Diphenylamine, which was not reported by the customer. 5. ‘Real’ Orange extract The customer had detected and confirmed three pesticide residues in the Orange extract sample Imazilil, Methidathion, and Thiabendazole. All three of these pesticides were confirmed by the DRS software and no other analytes were confirmed.

5

www.agilent.com/chem 6. ‘Real’ Apple extract The customer had detected and confirmed three pesticide residues in the Apple extract sample Diphenylamine, Chlorpyriphos, and Thiabendazole. All three of these pesticides were confirmed by the DRS. In addition, the DRS also confirmed the presence of five additional pesticide residues Dimethoate, Ethoxyquin, Methyl Parathion, Endosulfan Sulfate, and Progargite. These five pesticides had not been reported by the customer.

Conclusions The Agilent Technologies MSD-DRS provides additional powerful data processing capabilities to the MSD ChemStation software. Reviewing full scan GC/MS data for the confirmation of pesticide residues can be a labor-intensive and time consuming process requiring great skill and concentration by an experienced analyst. The DRS is able to process a complex food extract TIC in the order of 1 minute, whereas an experienced analyst may take more than 30 minutes to achieve the same quality result. The DRS software was proven to report the lowest number of false positives and false negatives in the shortest time period. In scan mode, the detection limit is not as low as in selected ion monitoring (SIM) mode; however, any prior knowledge of the target analytes (retention times or characteristic ions) is not required for the DRS.

The extensive data shown in this report, run under totally blind conditions, shows the high degree of confidence that an analyst can have in the results produced by the DRS in minutes.

Reference 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 5989-1157EN, www.agilent.com/chem

For More Information For more information on our products and services, visit our Web site at www.agilent.com/chem.

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. 2004 Printed in the USA October 5, 2004 5989-1654EN

Analysis of Organochlorine and Pyrethroid Pesticides with Agilent 6820 Gas Chromatograph/Micro-Electron Capture Detector Application

Environmental and Food Analysis

Author Chuanhong Tu Agilent Technologies Co., Ltd. (Shanghai) 412 YingLun Road Waigaoqiao Free Trade Zone Shanghai, 200131 P. R. C.

Abstract The Agilent 6820 gas chromatograph (GC) with microelectron capture detector (µECD) was used to analyze organochlorine and pyrethroid compounds. All compounds demonstrated good linearity among a wide concentration range. The sensitivity provided by µECD was much better than the requirements of the routine pesticide residue analysis.

Introduction The electron capture detector (ECD) is a type of detector with high sensitivity and selectivity for halogenated compounds. However, there are some drawbacks in ECD design. The ECD is inherently nonlinear, with a limited linear range. Due to the narrow linear range, sample concentration or dilution, and re-analysis have to be employed, resulting in lower productivity. In addition, in traditional ECD design, a large flow cell is necessary to be compatible with both packed and capillary columns, leading to lower detector sensitivity [1].

To address these problems, the µECD, developed by Agilent Technologies, uses a smaller flow cell. The µECD is optimized for capillary columns and designed for improved sensitivity. It was successfully used with an Agilent 6890 series GC with better detector sensitivity and a wider linear range. In this note, the Agilent 6820 GC with µECD was used to determine organochlorine and pyrethroid pesticides following the Chinese National Standard Method GB/T 5009.146-2003 [2]. System sensitivity and limits of detection (LOD) were examined for organochlorine and pyrethroid compounds.

Experimental All experiments were performed on an Agilent 6820 GC with split/splitless inlet and µECD. Single-tapered deactivated liner (p/n 5183-4696) and Agilent green septa (p/n 5183-4759) were used. Cerity Networked Data System (NDS) software was used for instrument control, signal acquisition, and data processing. Samples were manually introduced into the GC with a 10-µL micro-syringe (p/n 5182-3428). Experimental conditions are listed in Table 1. All organochlorine and pyrethroid compounds were diluted with hexane.

Table 1. Instrumental Parameters Instrument Agilent 6820 GC

Results

Software

Cerity NDS Chemical for QA/QC

µECD Sensitivity

Inlet

Split/Splitless; 250 ºC; splitless mode; purge time: 0.75 min

Injection volume

2 µL

Column

HP-1, 30 m × 0.32 mm × 0.25 µm (p/n 19091Z-413)

Carrier

Nitrogen, 6.0 psi, constant head pressure mode, 1.2 mL/min (60 ºC)

Oven

60 ºC (1 min), 30 ºC/min to 180 ºC, 5 ºC/min to 250 ºC (5 min), 3 ºC/min to 280 ºC (10 min)

Detector

µECD; 330 ºC; make-up: nitrogen, 60 mL/min

A chromatogram of organochlorine and pyrethroid pesticides on an HP-1 column is shown in Figure 1. The concentrations, for four benzene hydrochlorides (BHCs), heptachlor, aldrin, heptachlor epoxide, and six pyrethroids are 10 ppb; for p,p'-DDE, dieldrin, endrin, endosulfan I, and endosulfan II, 20 ppb; and for p,p'-DDD, endrin aldehyde, endosulfan sulfate, and p,p'-DDT, 60 ppb. Except for endrin and endosulfan II, the other 20 compounds were fully separated. Among pyrethroid pesticides, two permethrin, four cypermethrin and two fenvalerate isomers were separated.

Hz 1400

Peak identification 1. 2. 3. 4. 5. 6. 7. 8.

1200

1000

800

1

α-BHC β-BHC ϒ-BHC δ-BHC Heptachlor Aldrin Heptachlor epoxide Endosulfan I

9. 10. 11. 12. 13. 14. 15. 16.

p,p'-DDE Dieldrin Endrin Endosulfan II p,p’-DDD Endrin aldehyde Endosulan sulfate p,p'-DDT

15

13

16

9 8

3 600

10

4

12

6 5

14

11

7

2

400

200

10

12

Hz 240

14

16

18

20

min

18 17. 18. 19. 20. 21. 22.

220

200

Fenpropathrin Cyhalothrin Permethrin Cypermethrin Fenvalerate Deltamethrin

17 180

21

20

22

19 160

140

120 25

Figure 1. 2

27.5

30

32.5

Chromatogram of organochlorine and pyrethroid pesticides on HP-1 column.

35

min

The signal-to-noise ratios are larger than 20 for organochlorine pesticides at concentrations of 1–6 ppb. For 5 ppb pyrethroids, the signal-to noise ratios are larger than 10. They can be easily quantitated. Therefore, the µECD provides more than enough sensitivity to meet the requirements of quantitative analysis of pesticides residues. Linear Range and Response Factors The calibration curves of γ-BHC and permethrin, typical of organochlorine and pyrethroid pesticides, are shown in Figures 2 and 3, respectively. The linear range and response factors (RFs) are listed in Table 2. The RFs are the ratios of compound concentrations to peak areas. The relative standard deviations (RSDs) of RFs are less than 20%, better than the precision requirements for response factors in the contract laboratory program of the USEPA (the United States Environmental Protection Agency). The linear correlation coefficients for all compounds are better than 0.995.

pA 60000

γ-BHC y = 119.98x + 354.05 R2 = 0.9988

40000

20000

0 0

100

200

300

400

Concentration, ppb

Figure 2.

Calibration curve of γ-BHC, a typical organochlorine pesticide.

pA 4000

Permethrin y = 9.0737x + 97.257 R2 = 0.9986

3000

2000

1000

0 0

Figure 3.

100

200 Concentration, ppb

300

400

Calibration curve of permethrin, a typical pyrethroid.

3

www.agilent.com/chem Table 2.

Linearity Results for Organochlorine and Pyrethroid Pesticides

Compound α-BHC ß-BHC γ-BHC δ-BHC Heptachlor Aldrin Heptachlor epoxide Endosulfan I p,p'-DDE Dieldrin Endrin Endosulfan II p,p'-DDD p,p'-DDT Endrin aldehyde Endosulfan sulfate Fenpropathrin Cyhalothrin Permethrin Cypermethrin Fenvalerate Deltamethrin

Average RF

%RSD of RF

0.0064 0.0159 0.0078 0.0087 0.0097 0.0078 0.0100 0.0109 0.0088 0.0123 0.0152 0.0121 0.0379 0.0175 0.0139 0.0140 0.0500 0.0225 0.1007 0.0809 0.0728 0.0461

7.5 18.1 9.4 8.8 13.2 8.0 11.7 11.7 11.0 13.0 16.3 15.8 8.7 15.8 10.6 10.0 18.0 8.6 10.2 7.2 6.9 15.0

Linear range (ppb) 1–400 1–400 1–400 1–400 1–400 1–400 1–400 2–800 2–800 2–800 2–800 2–800 6–2400 6–2400 6–2400 6–2400 1–400 1–400 10–400 10–400 10–400 10–400

R2 0.9983 0.9984 0.9988 0.9986 0.9987 0.9982 0.9983 0.9983 0.9948 0.9981 0.9991 0.9987 0.9983 0.9972 0.9989 0.9988 0.9951 0.9982 0.9986 0.9991 0.9990 0.9996

Conclusion The Agilent 6820 GC/ ECD system shows good sensitivity and wide linear range for organochlorine and pyrethroid pesticides, and are much better than routine pesticide residue analysis requirements.

References 1. Channel, I., and Chang, I. L., “Analysis of Organochlorine Pesticides and PCB Congeners with the Agilent 6890 Micro-ECD”, Agilent Technologies, Publication (23) 5965-8556E www.agilent.com/chem 2. China National Standard Method GB/T 5009. 146-2003, Multiresidue analytical methods for organochlorine and pyrethroid pesticides for plant-originated food, August, 2003

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.

For More Information

© Agilent Technologies, Inc. 2004

For more information on our products and services, visit our Web site at www.agilent.com/chem.

Printed in the USA August 13, 2004 5989-1333EN

Analysis of Organophosphorus Pesticides with Agilent 6820 Gas Chromatograph/ Nitrogen Phosphorus Detector Application

Environmental and Food Analysis

Author Chuanhong Tu Agilent Technologies Co., Ltd. (Shanghai) 412 YingLun Road Waigaoqiao Free Trade Zone Shanghai, 200131 P. R. C.

Abstract Fifteen of the most common and important organophosphorus pesticides (OPs) were studied using the Agilent 6820 gas chromatograph (GC) equipped with the new nitrogen-phosphorus detector (NPD). The 6820 GC-NPD demonstrated good linearity for concentrations of pesticides range from 1 to 500 ng/mL (R2 >0.999) for most compounds. All OPs produced high signal-to-noise ratios for splitless injections at 10 ng/mL (ppb) concentrations with the NPD detector. Instrumental limits of detection for most OPs studied were at low or sub-ppb concentrations. This suggests the 6820 GC with an NPD is well suited for OP pesticide residue determinations in foods, water, or other samples.

Introduction Synthetic organic pesticides are widely used in modern agriculture to protect crops and improve production. The “green revolutions” of many countries are obtained through the application of these compounds. There are increasing concerns over

food safety and the identities and residual concentrations of pesticides due to their stability, inappropriate or illegal usage. In China, organophosphorus pesticides (OPs) account for 70% of the total amount of the pesticides used [1]. Maximum residue levels (MRLs) have been set up for 27 OPs in different kinds of food, and analytical methods have been developed for their analysis [2]. The China National standard method GB/T 5009.1452003 is a method for the determination of 16 organophosphorus and 4 carbamate pesticides by GC-NPDs [3]. In this application note, the Agilent 6820 GC equipped with an NPD is employed to determine 15 of the most common OPs of concern.

Experimental The experiments were carried out on an Agilent 6820 GC with split/splitless inlet and NPD. De-activated liners for splitless injection (p/n 5183-4696) were used to improve the inertness of the system; the septa were Agilent green septa (p/n 5183-4759). Cerity Networked Data System (NDS) for Chemical QA/QC software was used for instrument control, data collection, and data processing. The sample was introduced manually with a 10-µL syringe (p/n 5182-3428). All target compounds were dissolved in acetone. The experimental conditions are listed in Table 1. All flows were set using the Veriflow-500 digital flowmeter (p/n HVF-500-2).

Table 1.

by 1-µL injections of standards at 1, 10, 50, 100, and 500 ng/mL concentrations were linear with R2 >0.999 for most compounds. The calibration curve for dichlorvos, a typical OP, is shown in Figure 2.

Instrumental Parameters

Software

Cerity NDS for chemical QA/QC

Inlet

Split/Splitless inlet

Inlet temperature

250 ºC

Injection mode

Splitless

Injection volume

1 µL

Purge time

0.75 min

Column

HP-5ms, 30 m × 0.32 mm × 0.25 µm (p/n 19091S-413)

Carrier gas

He, head pressure: 12 psi, 2.5 mL/min at 60 ºC.

Oven temperature

60 ºC for 1 min, to 200 ºC at 10 ºC/min, to 250 ºC at 5 ºC/min, 5 min hold.

Detector

NPD at 325 ºC with white rubidium bead (p/n G1534-60570)

Detector gases

H2: 3 mL/min; Air: 60 mL/min; makeup N2: 10 mL/min

Table 2. Retention Times of Target Pesticides Peak number Compound Retention time 1 Methamidophos 8.26 2 Dichlorvos 8.63 3 Acephate 11.21 4 Monocrotophos 14.44 5 Phorate 14.60 6 Dimethoate 15.00 7 Parathion-methyl 17.03 8 Fenitrothion 17.75 9 Malathion 18.04 10 Fenthion 18.27 11 Parathion 18.34 12 Chlorpyrifos 18.34 13 Methidathion 19.97 14 Ethion 22.52 15 Triazophos 22.92

Results and Discussions The separation of 15 OPs is illustrated in Figure 1. The compounds, except for chlorpyrifos and parathion, were well separated with the HP-5ms column. Peak identifications are listed in Table 2. Calibration curves constructed from data obtained

pA 600 11, 12

6

400 2 1

3

14

7

5

10

4

8 9

13 15

200

0 10

Figure 1.

2

15

Chromatogram of 15 OPs at 1-ppm using the NPD.

20

min

pA 800

600

Dichlorvos y = 1.3849x + 3.3834 R2 = 0.9999

400

200

0 0

100

200

300

400

500

600

Concentration, ppb

Figure 2.

Calibration curve for dichlorvos, a typical OP.

Approximate Instrumental Limits of Detection The chromatogram for 10-ppb pesticides using the NPD is shown in Figure 3. All compounds are easily quantitated. Acephate, with the lowest response factor, provided around a 30 ratio of signal to noise. In fact, most compounds at 1 ppb show good peaks except methamidophos, acephate, and monocrotophos. The limits of detection (LODs) are much lower than the maximum residue levels (MRLs) for the OPs. 11, 12

pA

10

52

14

9 7

50

6

8

13 15

5 48

2

4

46

44

3 1 42

40

7.5

Figure 3.

10

12.5

15

17.5

20

22.5

min

Chromatogram of 10-ppb OPs using the Agilent 6820 GC/NPD.

3

www.agilent.com/chem Sample Solvent Precautions In some sample extraction procedures, organic solvents, containing elements with high electronegativity, (such as methylene chloride), may be used. However, these solvents may cause baseline shifts, change the detector sensitivity, shorten lifetime of the rubidium bead, or even make the detector unusable. This is true for all types and vendors of NPD [4]. If the final solution for injection consists of methylene chloride or chloroform, it is best to change the solvent to acetone or hexane.

Conclusion The Agilent 6820 GC with NPD can be used for the sensitive and selective determination of OPs. The NPD detector provides good linearity for most of these compounds in the 1–500-ppb range.

References 1. Xingui Sun et al, Survey of organophosphorus pesticide residues in vegetables and fruits in Beijing, (2003) Chinese Journal of Food Hygiene, 15, No. 6, 536-538. 2. Jiming Ye et al, Introduction of maximum residues limits in China, (2000) Pesticide Science and Management. 3. China National Standard Method GB/T 5009.145-2003, Determination of organophosphorus and carbamate pesticides multiresidues in vegetable foods. 4. Kai Meng, Yeugeny Kaplun, Rich White, “The New Features of the HP 6890 NitrogenPhosphorus Detector to Deal with Hostile Solvents,” Agilent Technologies, publication (23) 5963-6808E. www.agilent.com/chem

For More Information For more information on our products and services, 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. 2004 Printed in the USA August 13, 2004 5989-1335EN

visit our Web site at www.agilent.com/chem.

Analysis of Carbamate Pesticides Using Agilent 6820 Gas Chromatograph/Nitrogen Phosphorus Detector Application

Environmental and Food Analysis

Authors Weijun Yao, Chuanhong Tu Agilent Technologies Co., Ltd. (Shanghai) 412 YingLun Road Waigaoqiao Free Trade Zone Shanghai, 200131 P. R. C.

Abstract The Agilent 6820 gas chromatograph/nitrogen phosphorus detector (GC/NPD) was employed to determine carbamate pesticides, following the China National Standard Method GB/T 5009.104-2003. Seven common carbamates were fully separated using an HP-5ms column, 30 m × 0.32 mm × 0.25 µm. Each compound showed good linearity in the 10–1000-ppb range, with a detection limit lower than 10 ppb, which is 10–1000 times lower than the maximum allowable residue level.

Introduction The nitrogen phosphorus detector (NPD) is both sensitive and selective for organic compounds containing nitrogen and phosphorus, and is widely used in the trace analysis of organonitrogen or organophosphorus pesticides. The carbamates are types of organic synthetic pesticides, widely used to prevent plant diseases. The application of these pesticides may leave residues in agricultural products that may pose potential human health risks via the food chain. The Chinese government publishes maximum residue limits for the pesticides carbofuran, primicarb, and carbaryl in food. GB/T 5009.104-2003 (originally GB14877-1994) is a method used to determine six carbamate residues based on GC/NPD using a packed column [1]. In recent years, capillary-column methods have gradually replaced packed-column methods

because of the excellent separation capability of the capillary column. In this note, the Agilent 6820 GC/NPD is used to determine seven common carbamates including metolcarb, isoprocarb, baycarb, propoxur, carbofuran, primicarb, and carbaryl.

Experimental All experiments were performed on an Agilent 6820 GC with NPD, deactivated splitless inlet liner (p/n 5183-4696), and Agilent advanced green septa (p/n 5183-4759). Agilent Cerity Networked Data System (NDS) is used for instrument control and data acquisition. Instrumental parameters are listed in Table 1. The standard pesticides were dissolved in acetone solution. Samples were injected manually using a 10-µL syringe (p/n 5182-3428). Table 1. Instrumental Parameters Instrument Agilent 6820 GC Software Cerity NDS for chemical QA/QC Inlet Split/Splitless inlet Inlet temperature 250 ºC Injection mode Splitless Injection volume 1 µL Purge time 0.75 min Column HP-5ms, 30 m × 0.32 mm × 0.25 µm (p/n 19091S-413) Carrier gas Nitrogen, constant pressure of 5 psi, 1.0 mL/min (50 ºC) Oven temp 50 ºC (1 min) Ramp 1: 20 ºC/min to 100 ºC Ramp 2: 5 ºC/min to 150 ºC (5 min) Ramp 3: 10 ºC/min to 200 ºC (10 min) Detector NPD, 325 ºC, white bead (p/n G1534-60570) H2, 3 mL/min Air, 60 mL/min Detector gas Makeup, N2, 10 mL/min

Results and Discussions All compounds were separated with a HP-5ms column. The resolution of baycarb and propoxur is 1.0, although their structures are very similar. The chromatogram of these pesticides is shown in Figure 1. The peak areas are linear with concentrations for all carbamates in the range of 10–1000 ppb. Typical calibration curves for carbofuran and primicarb are shown in Figure 2. pA 90

Peak identification 5. Carbofuran 1. Metolcarb 6. Primicarb 2. Isoprocarb 7. Carbaryl 3. Baycarb 4. Propoxur

80

70

6

60

1 5

3 4

50

7

2 40

30

20 16

Figure 1.

18

20

22

24

26

Chromatogram of 1-ppm carbamates using NPD detector.

pA 140

120

Primicarb y = 0.1277x + 0.5557 R2 = 0.9992

100

80

60

Carbofuran y = 0.0706x + 0.7961 R2 = 0.9969

40

20

0 0

200

400

600

800

Concentration, ppb

Figure 2.

2

Calibration curves for primicarb and carbofuran.

1000

1200

min

The chromatogram for the 10-ppb carbamate standard is shown in Figure 3. The signal to noise ratios are between 6–12 for the 10-ppb carbamates. The limits of detection are 4–6 ppb; these are 10 to 1000 times lower than the maximum allowable residue limits stipulated in China National Standard (GB14928.2-94, GB14928.7-94, and GB14971-94), which is 50-ppb primicarb and 5000-ppb carbaryl in cereals [2, 3, 4]. The NPD can easily meet these requirements for the routine analysis of carbamate pesticide residues. pA

6 30.6

5

30.5

30.4

4

7

3

1 2

30.3

30.2

30.1

30

16

Figure 3.

18

20

22

24

26

min

Chromatogram of 10-ppb carbamates standard with 1-µL injection.

Conclusion The Agilent 6820 GC with NPD detector can be used for the sensitive and selective measurement of carbamate pesticide residues in food. The detection limit is much lower than the maximum residue limits. Good linearity was obtained in the range of 10–1000 ppb.

References 1. GB/T 5009.104-2003, Determination of carbamate pesticide residues in plant-originated foods. 2. GB14928.2-94 Maximun residue limits of primicarb in food. 3 GB14928.7-94 Maximun residue limits of carbofuran in rice grains. 4. GB14971-94 Maximun residue limits of carbaryl in food. 3

www.agilent.com/chem

For More Information For more information on our products and services, visit our Web site at www.agilent.com/chem.

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. 2004 Printed in the USA August 11, 2004 5989-1334EN

Comprehensive Pesticide Screening by GC/MSD using Deconvolution Reporting Software Application

Food Safety

Author Philip L. Wylie, Michael J. Szelewski, and Chin-Kai Meng Agilent Technologies, Inc. 2850 Centerville Road Wilmington, DE 19808-1610 USA Christopher P. Sandy Agilent Technologies, Inc. Block A, CSC Eskdale Road Winnersh Triangle Estate Wokingham, Berk RG41 5DZ United Kingdom

Introduction According to The Pesticide Manual, more than 700 pesticides are currently approved for use around the world [1]. About 600 more were used in the past, but are either banned or no longer marketed. In spite of their discontinuance, some of these still persist in the environment where they may bioaccumulate in the flora and fauna. Many pesticides or their degradation products can be found at trace levels in food and beverages; in soil, water, and air; in aquatic and terrestrial flora and fauna; and in human blood, adipose tissue, and breast milk. The World Health Organization has classified pesticides into five groups based upon their acute toxicity to humans [2]. The categories range from “Acutely Hazardous” to those that are “Unlikely to Present Acute Hazard in Normal Use.” Certain pesticides are classified as persistent organic pollutants (POPs), carcinogens, teratogens, or endocrine disrupters. It is now common to analyze for

pesticides in food and environmental samples to track their distribution in the environment and to ensure a safe food supply. Current analytical methods target only a subset of the possible compounds. Whether for food or environmental samples, analyses are often complicated by the presence of co-extracted natural products. Food or tissue extracts can be exceedingly complex matrices that require several stages of sample cleanup prior to analysis [3]. Even then, it can be difficult to detect trace levels of contaminants in the presence of the remaining matrix. For efficiency, multiresidue methods (MRMs) must be used to analyze for most pesticides. Traditionally, these methods have relied upon gas chromatography (GC) with a constellation of element-selective detectors to locate pesticides in the midst of a variable matrix [4, 5, 6]. GC with mass spectral detection (GC/MS) has been widely used for confirmation of hits. Liquid chromatography (LC) has been used for those compounds that are not amenable to GC [2]. Today, more and more pesticide laboratories are relying upon LC with mass spectral detection (LC/MS) and GC/MS as their primary analytical tools [7, 8]. Still, most MRMs are target compound methods that look for a small subset of the possible pesticides. Any compound not on the target list is likely to be missed by these MRMs. Using the techniques of retention time locking (RTL) [9, 10, 11] and spectral deconvolution [12], a method has been developed to screen for 567 pesticides and suspected endocrine disrupters in a single GC/MS analysis. Spectral deconvolution

helps to identify pesticides even when they are buried under co-eluting matrix compounds. RTL helps to eliminate false positives and gives greater confidence in the results. Users can easily add compounds to the method if they wish.

Experimental Table 1 lists the instrumentation, software, and analytical parameters used by Agilent for pesticide analysis. Depending upon the desired injection volume, a PTV inlet or split/splitless inlet can be used.

Table 1.

Samples Vegetable extracts were obtained from Dr. Mark Lee and Stephen Siegel at The California Department of Food and Agriculture (CDFA; Sacramento, CA USA) and from Dr. J.G.J. Mol at TNO Nutrition and Food Research (Zeist, The Netherlands). Seventeen data files from the GC/MS analysis of surface water samples were also contributed by CDFA and were processed in this laboratory using the Deconvolution Reporting Software (DRS). GC/MS data files (locked to the Agilent Pesticide Method) for 17 crop extracts were supplied by NRM Laboratories, Berkshire, UK.

Instrumentation and Conditions of Analysis

Gas chromatograph

Agilent 6890N

Automatic sampler

Agilent 7683

Inlet

Agilent PTV operated in the solvent vent mode

Column

Agilent 30 m × 0.25 mm × 0.25 µm HP-5MS (p/n 19091S-433)

Carrier gas

Helium in the constant pressure mode

RTL

Chlorpyrifos-methyl locked to 16.596 min (nominal column head pressure = 17.1 psi)

Oven temperature program

70 °C (2 min), 25 °C/min to 150 °C (0 min), 3 °C /min to 200 °C (0 min), 8 °C /min to 280 °C (10–15 min)

PTV inlet parameters

Temp program: 40 °C (0.25 min), 1600 °C/min to 250 °C (2 min); Vent time: 0.2 min; Vent flow: 200 mL/min; Vent pressure: 0.0 psi; Purge flow: 60.0 mL/min; Purge time: 2.00 min

Injection volume

15 µL (using a 50-µL syringe)

Mass Selective Detector (MSD)

Agilent 5973 inert

Scan range

50–550 amu

Source, quad, transfer line temperatures

230, 150, and 280 °C, respectively

Solvent delay

4.00 min

Multiplier voltage

Autotune voltage

Software GC/MSD ChemStation

Agilent p/n G1701DA (Version D01.00 sp1)

Deconvolution Reporting Software (DRS)

Agilent p/n G1716AA

Library searching software

NIST MS Search (version 2.0) (included with NIST '02 mass spectral library, Agilent p/n G1033A)

Deconvolution software

Automated Mass Spectral Deconvolution and Identification Software (AMDIS) (included with NIST '02 mass spectral library, Agilent p/n G1033A)

MS Libraries

NIST '02 mass spectral library (Agilent p/n G1033A); Agilent RTL Pesticide Library (p/n G1049A)

2

Results and Discussion

Basics of 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 this process. Here, the total ion chromatogram (TIC) and apex spectrum are shown. As is often the case, the peak is composed of multiple overlapping components and 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 of the individual components that make up the composite “spectrum.”

RTL and RTL Databases RTL is a technique developed by Agilent that allows users to match analyte retention times (RTs) on any Agilent 6890 GC, in any laboratory in the world, so long as the same nominal GC method and capillary column are used [13]. Using RTL, Agilent has developed several retention-timelocked databases for GC and GC/MS that include the locked retention time, compound name, CAS number, molecular formula, molecular weight, and mass spectrum (GC/MS databases only) for each entry [14]. The Agilent RTL Pesticide Library contains this information for almost all GC-amenable pesticides, as well as several endocrine disrupters - 567 compounds in all. For use with the DRS discussed below, this library was converted into the NIST format [15]. Separate Automated Mass Spectral Deconvolution and Identification Software (AMDIS) libraries for the RTs and compound information were created from the original RTL Pesticide Library. Users can easily augment these libraries with newer pesticides or other compounds of interest [15].

The deconvolution process finds ions whose individual abundances rise and fall together within the spectrum. In this case, it first corrects for the spectral skew that is inherent in quadrupole mass spectra and determines a more accurate apex RT of each chromatographic peak. As illustrated in Figure 1, deconvolution produces “clean” spectra for each overlapping component. These individual spectra can be library searched with a high expectation for a good match. The AMDIS that is incorporated into the Agilent DRS is supplied by the National Institute of Science and Technology (NIST) [12].

TIC and spectrum

Deconvoluted peaks and spectra

TIC Component 1 Component 2 Matrix

Component 3

Deconvolution Interference

Target

Figure 1.

An illustration of mass spectral deconvolution process. 3

DRS Agilent's DRS results from the combination of three different GC/MS software packages: 1) the Agilent GC/MS ChemStation, 2) the NIST Mass Spectral Search Program with the NIST '02 MS Library, and 3) the AMDIS software, also from NIST. Included in the DRS, are mass spectral and locked RT libraries for 567 pesticides and suspected endocrine disrupters. Three separate, but complimentary, data analysis steps are combined into the DRS. First, the GC/MS ChemStation software performs a normal quantitative analysis for target pesticides using a target ion and up to three qualifiers. An amount is reported for all calibrated compounds that are detected. For other compounds in the database, an estimate of their concentration can be reported based upon an average pesticide response factor

Figure 2.

4

(RF) that is supplied with the DRS software. The DRS then sends the data file to AMDIS, which deconvolutes the spectra and searches the Agilent RTL Pesticide Library (in AMDIS format) using the deconvoluted full spectra. A filter can be set in AMDIS, which requires the analyte's RT to fall within a user-specified time window. Because RTL is used to reproduce the RTL database RTs with high precision, this window can be quite small typically 20 seconds or less. Finally, the deconvoluted spectra for all of the targets found by AMDIS are searched against the 147,000-compound NIST mass spectral library for confirmation; for this step, there is no RT requirement. Once the appropriate method is loaded, the DRS report can be generated with a single mouse click as shown in Figure 2. The software can run automatically after each analysis or at a later time on a single file or a batch of files.

ChemStation pull down menu showing options for running the DRS on single or multiple files.

Pesticides in an Herbal Mix Figure 3 shows a TIC from the extract of an herbal mix. Figure 4 shows the MSD Deconvolution Report for this sample, which is produced in html format so it can easily be emailed or copied into a spreadsheet. This sample was chosen because herbs are among the most difficult vegetable products to analyze. Their extracts contain a large number of natural products that interfere with pesticide analysis.

450000

TIC: MOL_4A.D

400000

Abundance

350000 300000 250000 200000 150000 100000 50000 5.00

10.00

15.00

20.00

25.00 Time

30.00

35.00

Figure 3.

TIC of an herbal mix.

Figure 4.

MSD Deconvolution Report generated for the herbal mix extract shown in Figure 3.

40.00

45.00

5

The DRS report in Figure 4 lists the RT, CAS number, and compound name for each hit. Phenanthrene-d10, listed at the bottom of the report, is the internal standard (ISTD) used by the ChemStation to estimate the quantity of each compound that it found. Since an average pesticide response factor was used for all 567 target compounds, the amounts listed in column 4 are only estimates. Experience has shown that most estimates reported using an average pesticide response factor fall within a factor of 10 of their actual values. True quantitation requires calibration with pesticide standards in the normal way, but this is not practical for all of the pesticides in the database. A laboratory would normally generate calibration curves for their target set of pesticides and use the average RF for the remaining compounds in the database. In this way, when a new compound is detected, the lab can immediately get a rough estimate of its concentration and decide if it should be added to the calibration list. Column 5 in the report shows the match factor obtained through AMDIS deconvolution and RTL Pesticide Library searching using the deconvoluted full spectra. In this case, several more targets were identified by AMDIS than were found by the ChemStation software (for example, Prometon and p,p’-DDE), which is typical for complex samples. When locked RTs are available, it is a significant advantage to set a RT requirement in the AMDIS software. In this case, hits that did not fall within ±10 seconds of the database RT were eliminated. Column 6 shows the RT difference (in seconds) between the compound's library RT and its actual value in the chromatogram. Figure 4 shows that the software identified two phthalates (suspected endocrine disrupters) in addition to the pesticides. Phthalates are ubiquitous in the environment and are extremely difficult to remove from the background. In this case, no attempt was made to determine if the phthalates were actually extracted from the sample or were introduced in the laboratory. The last two columns in the DRS report show the results from searching all of the AMDIS hits against the NIST 147,000-compound mass spectral library. When the NIST library search finds a compound in the top 100 matches (a user-settable value) that agrees with the AMDIS results, its match factor is listed in column seven. The hit number is shown in the last column, with “1” being the best match (highest match factor) in the NIST database. Occasionally, the NIST library search does not find the AMDIS hit among the top

6

100 spectral matches. In this case, the next line in the report shows the best library match for that spectrum. This is evident for fluvalinate-tau-I (Figure 4), which eluted at 34.779 min. The next line shows the best NIST library match for that spectrum - fluvalinate. In this case, no compound with the same CAS number as fluvalinate-tau-I is contained in the NIST mass spectral library. In fact, fluvalinate-tau-I is the D isomer, while fluvalinate is the DL isomer mixture. Blind Comparison Between DRS and Traditional Data Review Many comparisons have shown that the DRS is much better than conventional methods at identifying target compounds in complex samples, such as food and environmental extracts. Two such studies are described here. In the first case, 17 unspiked crop samples were analyzed by NRM Laboratories in Berkshire, UK using Agilent's RT-locked pesticide method. The data files, but not their list of pesticide hits, were sent to Agilent for analysis using the new DRS. Table 2 shows a comparison of the results from the two laboratories. Using manual data review, NRM identified 28 pesticides in the 17 samples, four of which were below their lowest calibration level. Using the same data files, the DRS identified 33 pesticides. Agilent's automated method did not identify azoxystrobin in the spring onion sample because it is not included in the RTL pesticide library. While it can be found in the NIST library, it has a molecular ion at 403 amu and method used at NRM only scanned to 400 amu. The DRS method confirmed all four pesticides that were below the NRM calibration range and found five more (terbacil, pyrimethanil, methiocarb, pyridaben, and propamocarb) that were not included in their method. The agreement between the manual and automated methods was excellent. However, the DRS looks for many more pesticides and was able to find several that were missed by the manual method. In addition, manual data review took a chemist about 7 hours for the 17 samples while the DRS finished the task in 50 minutes of unattended computer time.

Table 2.

A Comparison of the Pesticides Found in 17 Unspiked Crop Samples Using Conventional Data Review and Agilent's DRS. Pesticides that Were Found by Only One Method Are Underlined

Sample

Agilent DRS results*

NRM Manual Analysis**

Coriander

Propyzamide Chlorthal-dimethyl p,p'-DDE

Propyzamide Chlorthal-dimethyl p,p'-DDE

Rosemary

Terbacil Pirimicarb Chlorthal-dimethyl

Not found*** Pirimicarb Chlorthal-dimethyl

Spring Onion

Propyzamide Pyrimethanil Pirimicarb Metalaxyl Iprodione Not in DRS library†

Propyzamide Not found*** Pirimicarb Metalaxyl Iprodione Azoxystrobin

Chives

Methiocarb Iprodione

Not found*** Iprodione

Cherry Tomato

Procymidone Pyridaben

Procymidone Not found***

Courgette

Propamocarb

Not found***

Aubergine

Procymidone Buprofezin Endosulfan sulfate Iprodione

Procymidone Buprofezin Endosulfan sulfate Iprodione

Flat Leaf Parsley

Chlorthal-dimethyl

Chlorthal-dimethyl

Lambs Lettuce

Iprodione

Iprodione†††

Cos Lettuce

Dimethoate Metalaxyl Procymidone Terbuconazole Omethoate††

Dimethoate Metalaxyl Procymidone Terbuconazole††† Omethoate

Fine Endive

Procymidone lamda-Cyhalothrin

Procymidone lamda-Cyhalothrin

Red Potato

Chloropropham Pirimicarb

Chloropropham Pirimicarb†††

Fine Endive

Pirimicarb

Pirimicarb†††

*

Pesticides found by re-analyzing NRM datafiles using Agilent's DRS software.

**

Pesticides found by NRM using target compound analysis and manual verification.

*** This compound was not in the NRM target compound list. †

This compound is not included in the Agilent RTL Pesticide Library or the DRS software.

††

Found by the Agilent ChemStation but not found by AMDIS or NIST library searching after deconvolution. After careful review of this hit, omethoate was judged not to be in the sample.

†††

Compound was detected but was below the calibration range.

7

Analysis of Surface Water Samples: In another study, the CDFA analyzed 17 surface water extracts for pesticides. TICs for two typical samples are shown in Figure 5. The CDFA used RTL and RTL database searching but without the benefit of spectral deconvolution. The same data files were then analyzed using the DRS for comparison. Table 3 shows the results from the CDFA manual analysis of the 17 samples compared to results using the DRS. The CDFA found 38 pesticide hits in the 17 samples, some of which were for the same pesticide in multiple samples. It took a skilled analyst about 8 hours to review the results, eliminate false positives, and verify all of the hits. The DRS found 37 of the compounds seen by the CDFA and identified one CDFA hit as a false positive. In addition, 34 more pesticides were

Table 3.

A Comparison of Results from the Analysis of 17 Surface Water Samples by GC/MS. The CDFA Used RTL and RTL Database Searching, but No Deconvolution. Agilent's DRS Was Used to Analyze the Same Data Files CDFA

DRS

Number of pesticide hits

37

Same 37 + 34 additional

Number of false positives

1

0

Time required for analysis

~ 8 hours

20 minutes

5.00

10.00

15.00

20.00

25.00

30.00

35.00

40.00

5.00

10.00

15.00

20.00

25.00

30.00

35.00

40.00

Figure 5.

8

found for a total of 71 hits in the 17 samples. The process was fully automated and took about 20 minutes of unattended computer time to process all of the data files.

TICs of typical surface water extracts provided by the CDFA.

Conclusions

Acknowledgements

Agilent's new DRS solution for pesticide analysis offers laboratories a number of real benefits.

The authors wish to thank Dr. Mark Lee and Stephen Siegel of the California Department of Food and Agriculture, Dr. J.G.J. Mol of TNO Research, The Netherlands, and the management and staff at NRM Laboratories, UK, for their contribution of samples and data.

• Ease of use: This software solution is very simple to use and takes no more skill than is needed to operate the 6890N/5973 inert GC/MS system. There is no need for the user to learn about the intricacies of deconvolution or to master a new software package.

References

• Automation: The deconvolution report can be generated automatically after each run or a batch of samples can be processed all at once.

1. C.D.S Tomlin, editor. The Pesticide Manual, 13th edition, British Crop Protection Council, Surry, UK (2003).

• Time savings: Data review is reduced from hours to minutes.

2. http://www.who.int/pcs/docs/ Classif_Pestic_2000-02.pdf

• Quality: It produces results with the fewest false positives and false negatives.

3. J. Cook, M.P. Beckett, B. Reliford, W. Hammock, and M. Engel (1999) J. AOAC Int., 82 (6), 14191435.

• Reproducibility: Results are not dependent upon the skill or experience of the operator. • Accuracy: Comparisons such as those discussed in this application note show that the DRS finds pesticides with greater accuracy than manual methods of data analysis. It is particularly useful for relatively complex samples where co-eluting matrix components might obscure traces of target pesticides. • Comprehensive: This method screens for almost all GC-amenable pesticides as well as several suspected endocrine disrupters in a single GC/MS run. With 567 compounds in the method, it is the most comprehensive pesticidescreening tool available. Users can add more compounds to the method as needed. • Produces quantitative, semi-quantitative, and qualitative results: All calibrated compounds can be quantified. The concentrations of any other compounds can be estimated using an average pesticide response factor provided with the software. Use of the DRS is not limited to pesticide analysis. Other target compound mass spectral libraries can be converted into the AMDIS format and used with this software. For example, one could use existing libraries for forensic drugs, flavors and fragrances, organic pollutants, etc. Users can even generate their own libraries and use them with the DRS. While not required, it is a big advantage to have an RTL library with locked RTs for each entry, as this will give the fewest false positives.

4. M.A. Luke, J.E. Froberg, and H.T. Masumoto, (1975) J. Assoc. Off. Anal. Chem. 58, 1020-1026. 5. M. Luke, J. Froberg, G. Doose, and H. Masumoto, (1981) J. Assoc. Off. Anal. Chem. 64, 1187-1195. 6. B. McMahon and N. Hardin (1994) Pesticide Analytical Manual, Vol. 1, 3rd Ed., U.S. Food and Drug Administration, Washington, DC. 7. J. Fillion, R. Hindle, M. Lacroux, and J. Selwyn (1995) J. AOAC Int. 78, 1252-1266. 8. J. Fillion, F. Sauvé, and J. Selwyn (2000) J AOAC Int., 83, 698-712. 9. P. Wylie and B. Quimby, “A Method Used to Screen for 567 Pesticides and Suspected Endocrine Disrupters,” Agilent Technologies, publication 5967-5860E www.agilent.com/chem. 10.H. Prest, P. Wylie, K. Weiner, and D. Agnew, “Efficient Screening for Pesticides and Endocrine Disrupters Using the 6890/5973 GC/MSD System,” Agilent Technologies, publication 5968-4884E www.agilent.com/chem. 11.K. Weiner and H. Prest, “Retention Time Locking: Creating Custom Retention Time Locked Screener Libraries,” Agilent Technologies, publication 5968-8657E www.agilent.com/chem. 12.National Institute of Standards and Technology, AMDIS Literature and Downloads website: http://www.amdis.net/What_is_AMDIS/ AMDIS_Literature_and_Downloads/ amdis_literature_and_downloads.html.

9

www.agilent.com/chem 13.V. Giarocco, B. Quimby and M. Klee, “Retention Time Locking: Concepts and Applications,” Agilent Technologies, publication 5966-2469E www.agilent.com/chem. 14.http://www.chem.agilent.com/cag/servsup/ usersoft/main.html#RTL. 15.M. Szelewski and C.K. Meng, “Building and Editing RTL Screener Databases and Libraries,” Agilent Technologies, publication 5989-0916EN www.agilent.com/chem.

For More Information For more information on our products and services, visit our Web site at www.agilent.com/chem.

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. 2004 Printed in the USA May 19, 2004 5989-1157EN

Improving the Analysis of Organotin Compounds Using Retention Time Locked Methods and Retention Time Databases Application

Environmental

Authors

Introduction

Frank David Research Institute for Chromatography Pres. Kennedypark 20, B-8500 Kortrijk, Belgium

For many years, organometal speciation has been an important topic in environmental analysis, primarily due to increasing awareness of the toxicological effects of many organometal compounds. Within the class of organometalics, organotin compounds are probably the most widely spread in the environment due to their use as additives in polymers and in antifouling paints. Organotin compounds degrade in the environment into more polar metabolites [1]. Tributyltin, one of the most frequently used organotin additives (as tributyltinchloride or tributyltinoxide), for instance, degrades into dibutyltin and monobutyltin species. Consequently, a large diversity of organotin compounds can be detected in various environmental samples [2]. More recently, organotin contamination of diapers and printed T-shirts was reported and numerous analyses were performed on different consumer products, including all types of absorbent hygiene products.

Pat Sandra University of Gent Krijgslaan 281 S4, B-9000 Gent, Belgium Philip L. Wylie Agilent Technologies 2850 Centerville Road Wilmington, DE 19808-1610 USA

Abstract The analysis of organotin compounds is becoming increasingly important in both environmental analysis and in food and consumer product analysis. This application note describes a retention time locked (RTL) gas chromatography/mass spectrometry (GC/MS) method for the analysis of derivatized organotin compounds. Three retention time locked libraries are made available, corresponding to three different derivatization methods. The retention time databases allow easy peak location and identification of the target solutes based on mass spectra and retention times.

Different methods were used for the extraction and analysis of organotin compounds in environmental, food, and consumer product matrices. Since the organotin compounds with less than four alkyl groups are very polar, they cannot be analyzed directly by GC and must be derivatized into tetraalkyltin compounds prior to analysis. Initially, most methods were based on extraction with

tropolone (a complexing agent) and n-hexane, followed by Grignard derivatization and determination with GC-flame photometric detection (FPD) [3–9]. Recently, in-situ ethylation with sodium tetraethylborate (NaBEt4) [10–13] has largely replaced Grignard derivatization. At the same time, mass selective detectors (MSD) and atomic emission detectors (AED) have replaced the FPD as the preferred GC detector for organotin compounds [11,13]. A few years ago, solid phase micro extraction (SPME) in combination with capillary gas chromatography-inductively coupled plasma mass spectrometry (CGC-ICP-MS) was used for the determination of volatile and semi-volatile organometal compounds, resulting in excellent sensitivity and selectivity [14,15]. SPME was performed in the headspace or directly in the aqueous sample using a 100 mm polydimethylsiloxane (PDMS) coated fiber. Using NaBEt4, organotin compounds could be derivatized in-situ and simultaneously extracted into the PDMS phase. More recently, stir bar sorptive extraction (SBSE) using a magnetic stir bar coated with a 0.5–1 mm PDMS layer was developed [16]. After extraction, the solutes were thermally desorbed online to GC/MS, GC-AED or GC-ICP-MS. SBSE in combination with CGC-ICP-MS was applied for the determination of organotins in environmental samples after in-situ derivatization with NaBEt4, resulting in unsurpassed sensitivity with detection limits reaching the ppq (pg/L) level [17]. For standard applications such as the determination of organotin compounds in sediments, or soils, and in extracts or leachates of consumer products, these extremely high sensitivities are not required. For these applications, sufficient sensitivity is obtained using mass spectrometric detection. In comparison to AED or ICP-MS, where specific tin-chromatograms are obtained, the chromatograms obtained by mass spectroscopy are far more complex, even when using the selected ion monitoring (SIM) mode. Several ions per solute need to be monitored, and the derivatized sample extracts often contain many co-extracted solutes

2

or by-products of the derivatization reaction. Therefore, data interpretation is more demanding requiring the use of extracted ion chromatograms, retention time matching, and calculation of the relative abundances of target and qualifier ions. In this respect, the use of retention time locked methods offers several advantages. If a selected ion method is used, the switching times between groups of monitored ions are fixed and do not need to be adjusted after column maintenance or column change, since the retention times of all solutes can be relocked. Moreover, quantification databases do not need to be updated for variations in retention times. Finally, a retention time locked database can be used, allowing easy peak allocation. Solute detection and confirmation are far more reliable using the results screener option [18,19], which combines the power of spectral matching with locked retention time matching. In this application note, a GC/MS method is described for the analysis of organotin compounds in environmental, food, or consumer product extracts. Since derivatization by Grignard reaction and derivatization using NaBEt4 are both easy and convenient, three types of derivatives are considered: methyl-derivatives using methylmagnesium bromide, pentyl- derivatives using pentylmagnesium bromide (both Grignard reagents), and ethylderivatives using NaBEt4. The most important organotin compounds are listed in Table 1 together with typical ions for the mass spectra of all three derivatives. Tin has several isotopes and the mass spectra are characterized by typical isotope clusters. The relative abundances of the tin isotopes are Sn-116 (14.24%), Sn-117 (7.57%), Sn-118 (24.01%), Sn-119 (8.59%), Sn-120 (32.97%), Sn-122 (4.71%), and Sn-124 (5.98%). For the organotin compounds listed in Table 1, mass spectral libraries and retention-time-locked screener libraries were created for all three types of derivatives. After selecting the appropriate derivitization method, a library and screener database can be selected, allowing fast data interpretation. Sample extraction and clean-up are beyond the scope of this application note.

Experimental Samples The organotin compounds listed in Table 1 were purchased from Dr Ehrenstorfer, Augsburg, Germany (http://www.analytical-standards.com). For analysis, the standards were dissolved in methanol at a 1000 ppm (1mg/mL) concentration. These solutions were further diluted, depending on the derivatization method used. For creation of the databases, approximately 10 µg of compound was derivatized, resulting in a final concentration of 10 ppm. Derivatization method 1: The sample extract is concentrated to 1 mL in an apolar solvent (typically hexane) in a reaction tube. To this solution, 0.5 mL methylmagnesiumbromide Grignard reagent (1.4 M in 75/25 toluene/THF, SigmaAldrich cat no 28,223-5) is added. The solution is vortexed for 10 s and allowed to stand at room temperature for 15 min. This procedure should be performed in a fume hood, since toxic vapors evolving from the reaction and the solvents are

Table 1:

flammable. The reaction is stopped and the excess reagent is removed by adding 2 mL of a saturated ammoniumchloride solution in water or 2 mL 0.25 mol/L aqueous sulphuric acid. The mixture is vortexed for 10 s and the two phases are allowed to separate. The clear upper layer (apolar hexane phase) is transferred to an autosampler vial for analysis. The resulting organotin compounds are the methyl-derivatives. Derivatization method 2: The sample extract is concentrated to 1 mL in an apolar solvent (typically hexane) in a reaction tube. To this solution, 0.5 mL pentylmagnesiumbromide Grignard reagent (2 M in diethylether, Sigma-Aldrich cat no 29,099-8) is added. The remaining steps in this procedure are identical to those used in derivitization method 1. The resulting organotin compounds are the pentyl-derivatives. Derivatization method 3: The sample extract is concentrated to 1 mL in a polar solvent (typically ethanol) in a reaction tube. To this solution, 1 mL acetate buffer (82 g/L sodium acetate in water, adjusted to pH 4.5 with acetic acid) and 50 µL

Organotin Compounds and Characteristic Ions for the Three Derivatization Products

Organotin solute Reagent

Abbreviation

Derivatives Triethyltin

TET

Derivatization 1 Methylmagnesium bromide Methyl193, 191, 165, 163

Derivatization 2 Pentylmagnesium bromide Pentyl179, 177, 249, 247

Derivatization 1 Sodium tetraethylborate Ethyl207, 205, 179, 177

Tetraethyltin

TeET

207, 205, 179, 177

207, 205, 179, 177

207, 205, 179, 177

Tripropyltin

TPT

179, 177, 221, 219

277, 275, 165, 163

235, 2331, 249, 247

Tetrapropyltin

TePT

249, 247, 207, 205

249, 247, 207, 205

249, 247, 207, 205

Monobutyltin

MBT

165, 163, 151, 149

319, 317, 193, 191

235, 233, 179, 177

Dibutyltin

DBT

151, 149, 207, 205

319, 317, 179, 177

263, 261, 207, 205

Tributyltin

TBT

193, 191, 249, 247

305, 303, 179, 177

291, 289, 207, 205

Tetrabutyltin

TeBT

291, 289, 179, 177

291, 289, 179, 177

291, 289, 179, 177

Monophenyltin

MPhT

227, 225, 223, 197

339, 337, 197, 195

255, 253, 197, 195

Diphenyltin

DPhT

289, 287, 285, 197

345, 343, 197, 195

303, 301, 197, 195

Triphenyltin

TPhT

351, 349, 347, 197

351, 349, 347, 197

351, 349, 347, 197

Tetraphenyltin

TePhT

351, 349, 347, 197

351, 349, 347, 197

351, 349, 347, 197

Tricyclohexyltin (Cyhexatin)

TCT

301, 299, 219, 217

357, 355, 205, 203

315, 313, 233, 231

Monooctyltin

MOT

165, 163, 263, 261

375, 373, 193, 191

291, 289, 179, 177

Dioctyltin

DOT

263, 261, 151, 149

417, 415, 375, 373

375, 373, 263, 261 3

derivatization reagent are added. The derivatization reagent is prepared by dissolving 2 g NaBEt4 (Sigma-Aldrich cat no 48,148-3) in 10 mL ethanol. This solution should be freshly prepared. The sample is shaken and allowed to react for 30 min. After addition of 5 mL water, the derivatized compounds are extracted in 1 mL hexane. The mixture is vortexed for 10 s and the two phases are allowed to separate. The clear upper layer (apolar hexane phase) is transferred to an autosampler vial for analysis. The resulting organotin compounds are the ethyl-derivatives. These derivatization methods can be adapted to the type of sample analyzed. For example, derivatization method 3 is often applied to aqueous samples directly, combining in-situ derivatization and simultaneous extraction. This method is also used for sediment samples. Typically 1 g sample (dry weight) is extracted with 10 mL acetate buffer, 7 mL methanol and 10 mL hexane. Four mL of a 5% NaBEt4 solution is added while stirring. The derivatized organotin compounds are simultaneously extracted into the hexane layer.

Table 2.

Analytical Conditions All analyses were performed on an Agilent 68905973N GC-MSD system. Automated splitless injection was performed using an Agilent 7683 automatic liquid sampler. The instrumental configuration and analytical conditions are summarized in Table 2. The retention time of tetrabutyltin (used as the locking standard) was locked at 16.000 min. To duplicate this method, the initial column head pressure can be set to the pressures indicated in Table 2 (nominal pressure). Then the retention time locking (RTL) calibration runs can be performed automatically (at –20%, –10%, +10% and +20% of the nominal pressure) [18]. The retention time versus head pressure curve is then calculated and stored in the method. Agilent’s RTL software uses this curve to set the column head pressure so that retention time of the locking standard (tetrabutyltin) is 16.000 min.

Instrumentation and Conditions of Analysis

Instrumentation Chromatographic system

Agilent 6890 GC

Inlet Detector Automatic sampler

Split/Splitless Agilent 5973 N MSD Agilent 7683

Liner

Splitless liner (part number 5062-3587)

Column

30 m × 0.25 mm id × 0.25 µm HP-5MS (Agilent part number 19091S-433)

Experimental conditions Inlet temperature Injection volume Injection mode Carrier gas Head pressure

280 °C 1 µL Splitless, purge time: 1 min, purge flow: 50 mL/min. Helium Tetrabutyltin is retention time locked at 16.000 min (pressure around 45 kPa at 50 °C, 34 cm/s at 50 °C)

Oven temperature Transfer line temperature Detector

50 °C, 1 min, 10 °C/min to 300 °C, 4 min. 300 °C Scan (40–550 amu), threshold 100, MS quad 150 °C, MS source 230 °C. Solvent delay: 4 min SIM mode: 50 ms dwell time per ion, ions listed in Table 3

4

Results and Discussion

derivatization. With this derivatization, the elution sequence of the butyltin compounds is MBT A typical chromatogram, for an organotin standard (10 C atoms) < DBT (12 C atoms) < TBT (14 C atoms) < TeBT (16 C atoms). The spectrum obtained for mixture, derivatized using method 3 (ethylderivatives with NaBEt4), is shown in Figure 1. The tributyltin (as tributylethyltin) is shown in Figure 2. The typical ion clusters, resulting from the different compounds elute according to their boiling point, tin isotopes, are clearly detected. and the elution sequence can be predicted by calculating the total number of carbon atoms after 3000000

TCT 2800000 2600000 2400000 2200000 2000000 Abundance

1800000

TeBT

TBT

1600000 1400000

DOT MOT

MBT TPT TePT DBT

1200000 1000000 800000 600000 400000 200000 0

10.00 11.00 12.00 13.00 14.00 15.00 16.00 17.00 18.00 19.00 20.00 21.00 22.00 23.00 Time

Figure 1.

GC/MS chromatogram for the analysis of an organotin standard mixture after derivatization with NaBEt4 (ethyl-derivatives). 207

130000 120000 110000 100000

Abundance

90000 80000 177

70000

263

151

60000

291 50000 121

40000

235

30000 20000 10000

41 57

0 40

60

81

103

80

100

313 120

140

160

180

200

220

240

260 280

300

389 406

334 320

340

360

380

400

m/z

Figure 2.

Mass spectrum of tributyltin after derivatization with NaBEt4 (ethyl-derivative). 5

The analysis of a coastal sediment sample is shown in Figure 3. In this case, derivatization method 2 (Grignard reaction with pentylmagnesium bromide) was applied and a complex chromatogram was obtained. Using the extracted ion chromatogram at m/e 179 the butyltin compounds were easily detected (Figure 4). Tetrabutyltin, eluting at 16.000 min, was added as internal standard. In this case, pentyl- derivatives are analyzed. Therefore the elution order is reversed since the derivatization adds a C5-group for every free valency. The elution sequence is now TeBT (16 C atoms = unchanged) < TBT (17 C atoms)