Methods of analysis by the U.S. Geological Survey National Water Quality Laboratory-Determination of 86 volatile organic compounds in water by gas chromatography/mass spectrometry, including detections less than reporting limits

This report presents precision and accuracy data for volatile organic compounds
(VOCs) in the nanogram-per-liter range, including aromatic hydrocarbons, reformulated
fuel components, and halogenated hydrocarbons using purge and trap capillary-column
gas chromatography/mass spectrometry. One-hundred-four VOCs were initially tested.
Of these, 86 are suitable for determination by this method. Selected data are provided for
the 18 VOCs that were not included. This method also allows for the reporting of
semiquantitative results for tentatively identified VOCs not included in the list of method
compounds. Method detection limits, method performance data, preservation study
results, and blank results are presented.
The authors describe a procedure for reporting low-concentration detections at
less than the reporting limit. The nondetection value (NDV) is introduced as a statistically
defined reporting limit designed to limit false positives and false negatives to less than 1
percent. Nondetections of method compounds are reported as ?less than NDV.? Positive
detections measured at less than NDV are reported as estimated concentrations to alert the
data user to decreased confidence in accurate quantitation. Instructions are provided for
analysts to report data at less than the reporting limits. This method can support the use of
either method reporting limits that censor detections at lower concentrations or the use of
NDVs as reporting limits. The data-reporting strategy for providing analytical results at
less than the reporting limit is a result of the increased need to identify the presence or
absence of environmental contaminants in water samples at increasingly lower
concentrations.
Long-term method detection limits (LTMDLs) for 86 selected compounds range
from 0.013 to 2.452 micrograms per liter (?g/L) and differ from standard method
detection limits (MDLs) in that the LTMDLs include the long-term variance of multiple
instruments, multiple operators, and multiple calibrations over a longer time. For these
reasons, LTMDLs are expected to be slightly higher than standard MDLs. Recoveries for
all of the VOCs tested ranged from 36 (tert-butyl formate) to 155 percent
(pentachlorobenzene). The majority of the compounds ranged from 85 to 115 percent
recovery and had less than 5 percent relative standard deviation for concentrations spiked
between 1 to 500 ?g/L in volatile blank-, surface-, and ground-water samples. Recoveries of 60 set spikes at low concentrations ranged from 70 to 114 percent (1,2,3-
trimethylbenzene and acetone). Recovery data were collected over 6 months with
multiple instruments, operators, and calibrations.
In this method, volatile organic compounds are extracted from a water sample by
actively purging with helium. The VOCs are collected onto a sorbent trap, thermally
desorbed, separated by a Megabore gas chromatographic capillary column, and finally
determined by a full-scan quadrupole mass spectrometer. Compound identification is
confirmed by the gas chromatographic retention time and by the resultant mass spectrum,
typically identified by three unique ions. An unknown compound detected in a sample
can be tentatively identified by comparing the unknown mass spectrum to reference
spectra in the mass-spectra computer-data system library compiled by the National
Institute of Standards and Technology.