See: SO-GasEx cruise report, Section 5.3.3 ppgs 19-20

Operation description:
Sampling times and locations:
Sampled all CTD casts where Niskin bottles were tripped.
One deep cast to 4600m, two mid-depth casts to 1500m, four
shallow casts to 100m, and other casts usually to 500m.
See CTD cast logs and bottle files for specific times, locations, and bottles for each cast.
See event log for times and locations when sampled underway seawater line.

Overall sampling strategy:
Normally collected one sample from each depth that might contain
the released SF6, plus three - six samples deeper. All available
depths were sampled on the first four casts and on any later cast
that was shallower than 500m. If multiple Niskins were tripped
at a given depth, hydrography sampled only one of these. Duplicate
samples were drawn from a Niskin, normally in the mixed layer, on
nearly every cast. The first Niskin of a cast that was sampled
was the deepest Helium sample, and the sampling continued in order
to the surface Niskin. The deeper Niskins for SF6 sampling were
done next, starting at the deepest.

Sampling technique:
The discrete samples were collected in 550 ml borosilicate glass bottles
with ground glass stoppers. A sample bottle was shaken with 30-40 ml of
water; then inverted and rinsed with 10-20 ml of water. The bottle was
filled from the bottom with ~800 ml of bubble free water. The stopper
was put in the bottle ensuring that no gas bubble was enclosed. After
all the samples were collected from a cast, a rubber band was slipped
over the stopper. If the samples were not going to be analyzed within
10 hours, the glass bottles were stored in water.

Analytical method:
The samples were analyzed on an instrument patterned after Law et.al. [1998]
and built in 1998 at AOML. About 269 ml of sample water was sucked into an
evacuated chamber through a showerhead. The SF6 that entered the headspace
during filling and the SF6 that remained dissolved in the water were stripped
with ultra high purity nitrogen onto a Carboxen 1000 trap held at -68 C.
After a little more than 3 minutes of purging, the trap was isolated and heated
to 150 C. The trapped gases were swept onto a 1.5 m x 0.3 cm OD molecular sieve
5A column. The SF6 was separated from oxygen and other gases and was measured
with an electron capture detector (ECD).

The detector was calibrated using six standards:
- cyl#CA2071- 5.7 pptrillion (v/v)
- cyl#CA2089- 55.1 pptr
- cyl#CA3452- 112.3 pptr
- cyl#CA2060- 166.6 pptr
- cyl#CA2056- 345.0 pptr
- cyl#CA2093- 1109.0 pptr

The custom software includes a chromatography package from WillStein Software,
which is used for acquisition of the ECD output and to reintegrate peaks.
The SF6 peaks were reintegrated for all blank analyses, water analyses and the
analyses of 5.5 pptr standard. The larger standards had very favorable signal
to noise ratios.

The pattern of analyses include initial and final groups of standards bracketing
groups of water analyses. The water samples from each cast were typically split
into two groups, each covering the entire water column. Within a group, the
deepest water samples were analyzed before the more concentrated surface water
samples.

The Typical Analyses Sequence:
- 2-3 gas blanks
- 15-20 gas standards spanning the entire range
- gas blank
- 1-2 stripper blanks
- 7-9 water analyses
- stripper blank
- 1 analysis of 166.6 pptr standard
- stripper blank
- 7-9 water analyses
- [repeat analyses of stripper blanks, 166.p pptr std, and water samples two or three times]
- stripper blank
- gas blank
- 5-15 standards
- 2 gas blanks

The stripping chamber was evacuated and flushed with nitrogen before the first use.
If the stripping chamber was not going to used for more than 6 hours, it was rinsed
with fresh water and then evacuated and flushed.

Instrument details:
The ECD gas chromatograph is a Shimadzu Mini 2.
The custom software was written with LabView 5.1 in 1998 and revised in 2002.
The reference for a description of the instrument is: Law, C.S, A.J. Watson,
M.J. Liddicoat, and T. Stanton, Sulphur hexafluoride as a tracer of biogeochemical
and physical processes in an open-ocean iron fertilization experiment,
Deep Sea Research II, 45, 977-994, 1998.

Operation Log - During Cruise
5 - 6 March 2008
Attempts were made to use this discrete instrument to quantitate the amount of
SF6 in water actively being bubbled in the dosing tank. All attempts resulted
in peaks much larger than full scale for the ECD. The excessive amounts of SF6
were easily swept out of the instrument.

Station 3
one sample of a duplicate pair is stored for 23 hours before analyses - precision
of duplicates was good

Station 5
one sample of a duplicate pair is stored for 24 hours before analyses,
another sample of a different duplicate pair is stored for 52 hours.
Precision of the 52-hour duplicates was poorer than normal.

Station 8
These samples were stored overnight before analyses as were some earlier
stations and many later stations. This was the first station for which
some outgassing bubbles in some samples was noticed. These tiny bubbles
(<0.1 ml total volume) were not present when the bottle was sealed and
did not seem to bias the SF6 results.

Station 10
The sampling order was more discontinuous than normal to accommodate noble
gas sampling. For the collection of samples from the underway system after
station 10, the outflow tube of the MicroTSG in the hydro lab sink was put
in the bottom of a sample bottle and left to overfill the bottle for at least
2 minutes (flow at 1.3 liter/min).

13 March 2008
During analysis of sample bottle 15 (stn 9, N 22) the Labview program was in its
'pause and prompt' state longer than usual. The analytical system evacuates the
stripping chamber and then presents a prompt screen until a "start" button is
pushed by the analyst. The analyst has to draw some of the sample through the
inlet tube and solenoid valve before pushing the "start" button. If the 'pause
and prompt' state is long, the chance of outside air leaking into the chamber is
greater (potential higher blank).

15 March 2008
In an effort to have comparable results from the underway and discrete SF6 analytical
systems, the single SF6 gaseous standard for the underway system (58.2 pptr) was run
on the discrete system and three of the discrete standards (166.6, 55.1, and 5.7 pptr)
were run on the underway system.

18 March 2008
Both system and stripper blanks are significantly higher. The second dosing tank is
being bubbled on the afterdeck, and the lab air was enriched in SF6. No leaky fittings
could be found on the analytical system despite a day of searching. After ~48 hrs the
blanks decreased, but remained higher than at the start of the cruise.

22 March 2008
Last night was the first station (16) on the second patch. I took a sample from the
underway system to quickly check the strength of ECD signal. When the sample was run,
the peak went offscale. The ECD was set for 2nAmp standing current through station 14.
I decreased the setting to 1 nAmp. To span the largest peak from the water analyses
of station 16, I had to trap 6 regular gas loops (1.606 ml) of the most concentrated
standard (1109 pptr). I am concerned about peak shape changes while trapping that many
loops sequentially, so I put the second gas sample loop valve back on the system.
[it was taken off during a previous experiment.] I made a large loop (~7 ml) and created
new analytical methods to adequately flush the loop while and after filling it with
standard or blank gas. I will have to determine the exact volume of the new gas loop
back at AOML.

25 March 2008
Try again to locate the source of the elevated blanks, especially the stripper blanks.
Still, could not find any fitting that leaked. At the end of a day of analyses,
the chamber is rinsed with fresh water to reduce the corrosion on the solenoid valves
attached to the drain and vacuum connections. I stripped a sample of the fresh water
on the ship and discovered that it was as concentrated as the surface 'in-patch' water
(>200 fmole/liter). The chamber will continue to be rinsed with fresh water, but it
will go through a couple evacuate-and-purge cycles before being left for the night.
[no reduction in the blanks developed, but the procedure was continued.]

25 March 2008
During analyses of sample bottles 202 (stn 25, N 3), 206 (stn 25, N 9) and 10
(stn 26, N 18) the Labview program was in its 'pause and prompt' state longer than usual.

26 March 2008
During analysis of sample bottles 203 (stn 27, N 8) the Labview program was in its
'pause and prompt' state longer than usual.

28 March 2008
During analysis of sample bottles 2 (stn 32, submerged pump) the Labview program was
in its 'pause and prompt' state longer than usual.

29 March 2008
During analyses of sample bottles 2 (stn 35, N 7) and 17 (stn 35, underway system)
the Labview program was in its 'pause and prompt' state longer than usual.

30 March 2008
During analysis of sample bottles 202 (stn 37, N 6) the Labview program was in its
'pause and prompt' state longer than usual. During analysis of sample bottles 206
(stn 37, N 10) there seemed to be air in the inlet line and the water level in the
stripping chamber seemed lower than usual. Maybe the inlet tube was not submerged
as low as it should have been [but how would the Pt wire water sensors stop the water
flow automatically - splashes with gas going in?].

31 March 2008
During sampling, the stopper on bottle 203 (stn 41, N 3) was removed ~3 minutes after
sampling so some more a little more water from the niskin was gently added to the very
top of the bottle and the stopper reinserted.

03 April 2008
During sampling, the stopper on bottle 203 (stn 41, N 3) was removed ~3 minutes after
sampling so some more a little more water from the niskin was gently added to the very
top of the bottle and the stopper reinserted.

For Blanks and Cals See: SO-GasEx Discrete SF6 Blanks and Calibration Curves

Disc_SF6_stn.csv