Gschwend%20thesis.pdf

"
CSi was injected with the split closed (K. Grob and
K. Grob, Jr., 1974). The split was opened (10: I ratio)
after 30 sec. The helium carrier gas flow rate was
3 ml/min at 20°C. The injector and detector were
operated at 250°C. The temperature program was:
2Q-.22"C isothermal for 8 min, then 3°C/min to 250°C.
GC-MS analyses and structural identifications
were performed on a Finnigan Model 3200 equipped
with a Data System 600 computer. A Finnigan
Model 9500 gas chromatograph equipped with a Finnigan
splitless injector and a 25 m x 0.32 mm i.d.
glass capilary SE 52 column was interfaced to the
MS with an all-glass. transfer line, held at 250°C,
allowing coaxial introduction of reagent gases for
chemical ionization (CI-MS) (Blum and Richter,
1975, 1977). Mass spectra were recorded at the rate
of one per 1. sec from 35 to 350 amu at 70 e V and
90°C source temperature for EI; and at one per
1. sec over a range of 60350 amu at 130 e V and
90°C source temperature for CI. Methane was used
as the chemical ionization reagent gas. Compounds
were identified by comparison of GC and GC-MS
properties with those of authentic standards or library
spectra as indicated in Table 2.
Method characteristics
Blanks. Since most VC in seawater were present
below the 10 ng!kg level, much below the levels previously
reported for other samples (K. Grob and G.
Grob, 1974), a premium was placed on achieving sensitivity
and low, reproducible blanks. Figure IA
shows a representative gas chromatogram ofthe trap
extract from stripping a moist 5 1 flask containing
prestripped air. As usual, 40 ng I-chloro-octane (No.
2(0) were added to the trap before extraction and
its peak height corresponded to 10 ng!kg in water
. usually processed. This typical instrument blank
shows quantitative recovery of the standard and the
presencè of very few peaks, corresponding to less than
. i ng each VC!kg seawater introduced by the stripping
. procdure from the apparatus iud by subsequent
manipulations. Contamination from laboratory air
can erratically be a serious problem. Pouring pre-
~ stripped water from one flask to another in the
laboratory introduced numerous contaminants at the
10 ng!kg leveL. However, stripping five liters of 'CD
air' (sampled by emptying a flask filled with prestripped
water at CD) does not give these laboratory
contamination problems.
Recovery. TIie. recovery of internal standard~ relative
to I-chloro-octane (No. 200) was always above
85%. We restrip all samples immediately after the first
strip and again after 24 hr to monitor recovery and
compound generation. The internal standards are
never detectable in these restrips. Figure IB is typical
of additional recovery of VC on restripping a relatively
non-turbid water sample after initial strip
(Fig. 2C). For most compounds little material above
the blank occurs in the rest rip. However, substantial
Volatile organic çornpounds in coastal seawater
quantities of n-pentadecane (No. 417) and n-heptadecane
(No. 515) frequently occur in the second strip,
and the yield of dimethyldisulfide (No. 37) and
dimethyltrisulfide (No. 149) typically increases, especially
after 24 hr (Fig. i C). A few samples that were
especially turbid or contained -high total levels of
volatiles yielded substantial recoveries on restripping.
This is ilustrated in Fig. i D, the restrip of the turbid
sample shown in Fig. 2B. The VC concentrations
reported in this paper are those recovered in the first
2 hr stripping only.
Recovery experiments were also conducted in
duplicate using two different strippers, traps and analysts.
Different seawater samples were spiked with a
test mixture of compounds encompassing a variety
of chemiCai functionalities and volatilities within the
expected range of the technique, Samples were spiked
at 5, 10, 50, and 100 ng of each compound!kg seawater.
Acetone solutions of the mixture were used
to introduce the spike. The two lower level experiments
utilized exhaustively prestripped low-turbidity
CD water, whereas the two higher level experiments
used untreated CD water of low turbidity. The recovery
for each compound was determined by comparing
GC peak areas (normalized to I-chloro-octane, No.
200) determined by a Columbia Scientific Instruments
"Supergrator 3" electronic integrator for the sample
with peak areas from GC analysis of the test mixture.
The test mixture composition and the average recoveries
are given in Table i. For each compound, the
fraction recovered was independent of the level added,
suggesting the absence of 'threshold effects' at these
levels. The poor recovery of the naphthalenes is entirely
due to poor extraction recovery from the charcoal
traps (35%) as shown by trap spiking experiments.
Although I-octanol is also lost to some extent
, on the traps (58% recovery), its extremely poor total
, recovery is probably primarily due to its high water
solubility.
Reproducibilty. In order to assess the reproducibility
of sampling, stripping, and analysis as applied
to environmental samples, duplicate seawater samples
were obtained on five separate occasions and processed
concurrently on two different strippers by two
different analysts. Precision waS estimated by measuring
the heights of twenty of the larger peaks relative
to the I-chloro-octane (No. 2(0) peak height. With
the few exceptions listed below, these peak height
ratios did not vary between duplicates by more than
15%. Peak height ratio differences of up to 30% were
shown by dimethyldisulfide (DMDS, No. 37), demethyltrisulfide
(DMTS, No. 149), nonanal (No. 225),
decanal (No. 275), and n-heptadecane (No. 515).
We find that removing salt and other deposits from
seawater samples from the charcoal traps by the procedure
of Grob and Zürcher (1976) after every few
samples maintains good reproducibility. The moderately
large headspace over our samples was adopted
to minimize the salt spray transport onto the traps. .
We have experienced irreversible trap clogging from
95