Analysis of Halogenated Organic Contaminants in Water Using GC-MS and Large Volume Injection

I recently attended the annual meeting of  AOAC International where I presented a poster on adapting EPA method 551.1 to GC-MS by using the CSR-LVSI technique our lab has had so much success with in the past.

EPA 551.1 is a dual column GC-ECD method used to measure a wide range of halogenated organic contaminants in drinking water including disinfection byproducts (DBP’s) and chlorinated solvents as well as herbicides and pesticides [1]. These compounds have previously been reported at low µg/L (ppb) levels in both source water and finished drinking water [2,3]. They are regulated under the Safe Drinking Water Act and are assigned the Maximum Contaminant Limits (MCL) shown in figure 1.

Fig. 1) Halogenated organic contaminants regulated by US EPA under the Safe Drinking Water Act and their associated Maximum Contaminant Limits in drinking water. All are target analytes of method 551.1

Fig. 1) Halogenated organic contaminants regulated by US EPA under the Safe Drinking Water Act and their associated Maximum Contaminant Limits in drinking water. All are target analytes of method 551.1

 

As is typical for drinking water contaminants, the MCL’s for most of these compounds are in the single digit ppb or even sub ppb range.  EPA 551.1 addresses the challenge of detecting these low levels by prescribing a liquid-liquid extraction where a 50 mL water sample is saturated with sodium sulfate and extracted with 3 mL of methyl t-butyl ether (MTBE). While this provides a preconcentration factor of more than 16 fold in the final extract, the chromatographic detection method must still be quite sensitive.

Dual column ECD methods like 551.1 provide very high sensitivity for halogenated compounds but they are not without drawback. Dual column analysis relies on differences in elution profile between the two analytical columns for positive peak identification. An analyst must consider two separate detector channels with different expected elution profiles along with possible ECD active interferences that can confound peak identification and quantification.

My colleague Dan Li has blogged on how common phthalate contaminants from plastic lab ware and packaging respond by ECD and can interfere with analysis.

The central goal of this work was to develop a single column GC-MS method for analysis of the EPA method 551.1 analytes in water. The CSR-LVSI large volume injection technique was used to achieve detection limits relevant to the contaminant MCL’s [4].

 

CSR-LVSI

Fig 2) The instrument configuration used for large volume injection.

 

For detailed information on CSR-LVSI see some of our previous blog posts and technical articles:

Combined Determination of 1,4-Dioxane and Nitrosamine Contaminants in Drinking Water

Large Volume Splitless Injection Using an Unmodified Split/Splitless Inlet and GC-TOFMS for Pesticides and Brominated Flame Retardants

Solvent Focusing – The Key to Successfully Implementing CSR-LVSI

Is there a limit to the volume you can inject into an Agilent split/splitless inlet?

Development of GC conditions involved modeling using Pro ezGC® as well taking into account the factors needed for optimal CSR-LVSI. In this case I was making 50 µL autosampler injections of MTBE using the GC conditions detailed below.

 

C-Grams

Fig 3) SIM-TIC chromatogram collected from 50 uL CSR-LVSI injection of an EPA 551.1 water extract. The water blank was spiked at 5 ug/L. Assuming 100% recovery this represents 4 ng of each component on column. Trace is the sum of all monitored ions.

Conditions

 

With the retention times set, I developed an MS-SIM method for the 36 analytes that spanned 13 groups!  Some of the SIM group switching was very tight, taking place in 10-20 sec between analyte retention times. Retention times were very stable and I had no problems running the complex SIM program.

When choosing SIM ions I often erred on the side of caution, sacrificing abundance for higher specificity. I also tried to avoid masses commonly associated with siloxanes, column bleed, and phthalates. It is important to analyze a solvent blank and reagent blank before samples to asses any background contamination.

In some cases the solvent dimethylformamide may be present in herbicide standards to aid solvation and stability. Additionally, commercial lots of MTBE are known to sometimes contain target analytes such as chloroform or carbon tetrachloride [1].

 

Pk list

Figure 4) Selected ions used for MS-SIM. Compound identification was confirmed by the presence of all 3 ions with coinciding retention time and intensity ratios +/- 20% of those collected for a standard in solvent.

 

At this point I was happy with how the GC method worked for high level standards so I ran a 5 point calibration across a 100 fold concentration range. The calibration samples were prepared by spiking 50 mL aliquots of reagent water at the levels

0.1 µg/L, 0.4 µg/L, 1.0 µg/L, 5.0 µg/L, 10 µg/L

and performing the salting out liquid-liquid extraction exactly as described in EPA 551.1. Internal standard is added to the resulting  3 mL MTBE extract at 100 µg/L before analysis. Linearity was good for most compounds with some deviation occurring for the polar herbicides and pesticides.

linearity

Figure 4) Linearity of calibration for the halogenated organics in water from 0.1 µg/L to 10 µg/L.

 

One pair of closely related herbicides, simazine and atrazine,  uniquely highlights the value of mass selective detection.  These triazine herbicides are structurally very similar and difficult to chromatographically separate,  requiring a slow oven ramp rate that results in an extended run time. Even in the best case baseline resolution is unlikely.

However, examining the EI mass spectra of the compounds reveals that despite their structural similarity each has unique fragmentation ions allowing for deconvolution of the TIC signal and individual quantitation.

 

sim vs. atra

Figure 5) Deconvolution of the simazine and atrazine signals from the TIC at 5 ug/L.

 

While sensitivity is sacrificed when moving from ECD to EI-MS the large volume injection technique used here (CSR-LVSI) can make up for much of the loss by introducing a larger amount of sample on column.  Choice of target ions for each compound can also have a dramatic effect on sensitivity. Smaller m/z are generally more abundant but prone to interference while larger m/z are more compound specific.

The GC-MS system was calibrated down to 0.1 µg/L  without any modification of the EPA 551.1 sample preparation procedure. Based on signal to noise ratio at the 0.1 µg/L level, many of the volatile compounds could be calibrated even lower.

Obtaining mass spectral information, even if only for 3 selected ions, gives higher confidence in compound identification, delineates target analytes from interferences, and results in fewer false positive detections.  The simazine and atrazine case shows how mass spectral detection can also remove some of the burden of chromatographic separation.

References:

[1] US EPA method 551.1, “Determination of Chlorination Disinfection Byproducts, Chlorinated Solvents, and Halogenated Pesticides/Herbicides in Drinking Water by Liquid-Liquid Extraction and Gas Chromatography with Electron Capture Detection”

[2] Richardson, S; Plewa, M; Wagner, E; Schoeny, R; DeMarini, D “Occurrence, genotoxicity, and carcinogenicity of regulated and emerging disinfection by-products in drinking water: A review and roadmap for research” Mutation Research 636 (2007) 178–242

[3] Thurman, M; Goolsby, D; Meyer, M; Kolpin, D “A Reconnaissance Study of Herbicides and Metabolites in Surface Water of the Midwestern United States…” Environ. Sci. Technol. 26 (1992) 2440-2447

[4] Magni, P; Porzano, T; “Concurrent Solvent Recondensation Large Sample Volume Splitless Injection” J. Sep. Sci. 26 (2003) 1491-1498

 

 

3 Responses to “Analysis of Halogenated Organic Contaminants in Water Using GC-MS and Large Volume Injection”

  1. jaap says:

    Nice work Rebecca!. What happens between peak 17 and 18?

  2. Rebecca Stevens says:

    Hi Jaap, Thanks for the comment.

    The elevated baseline in that area is a m/z 75 interference from a solvent impurity.
    Full scan of the offending peak in the solvent blank gives a highest probability hit for tert-butyl hydroperoxide.

    I chose to monitor m/z 75 despite the interference to preserve sensitivity for 1,2,3-trichloropropane (#18).
    There are other ion choices for this analyte which are free from interference and could be used at the expense of abundance.
    When reviewing quantitation data I did have to correct the integration of 1,2,3-trichloropropane more often than for the other analytes, probably as a consequence of being on that “bump” of elevated baseline.

    I was somewhat surprised by the intensity of the tert-butyl hydroperoxide signal, although when making 50 uL injections even low level solvent impurities will be apparent. The QC test listed on the solvent bottle is done by GC-ECD and so wouldn’t have been sensitive to this impurity.

    My theory is that tert-butyl hydroperoxide is formed during production of the MTBE. To synthesize MTBE isobutylene is reacted with methanol in the presence of acid. When isobutylene is exposed to oxygen under the same conditions it can be oxidized to tert-butyl hydroperoxide and other similar compounds.

  3. nike ragazzo says:

    Very well written article. It will be supportive to anyone who usess it, including yours truly :). Keep up the good work – looking forward to more posts.

Leave a Reply