GCI TECH NOTES©
Volume 11, Number 1
A Gossman Consulting, Inc.
Publication January 2006
This is part of a
series of GCI Tech Notes focusing on the early
development of the hazardous waste fuels programs during the early
1980s. I was hired as the facility manager for the first
commercial
hazardous waste operation at a cement plant in early 1980. Many
of the
developments in storage, processing, testing and use of hazardous waste
fuels were the result of work done at a handful of plants in the early
and mid 80’s. Look for issues to include topics on storage, lab
testing methods, processing and the impact of HWF on cement product
quality and production.
by
David Gossman, Gossman Consulting, Inc.
From the very beginning the plan for the
first
commercial hazardous waste fuel facility, located at the General
Portland
Cement plant in Paulding, Ohio and operated by Systech Corp. was that
every
batch of waste fuel that was burned would be tested for PCBs and a
variety of
volatile and semivolatile organic compounds. The challenge that I faced
was
that only one gas chromatograph (GC) was purchased to do the job and I
made the
early decision that we would do no testing “after the fact.” In other
words,
all testing had to be done on a truck before it was off-loaded. Testing
after
the load was in the tank was, and continues to be, only a way to
increase
liability, and certainly does not avoid it. The GC was the only piece
of lab
equipment that had been purchased prior to my being hired. Luckily,
Systech had
purchased the top end GC on the market at the time equipped with all
the best
options. I had an HP 5880 Level 4 instrument with dual terminals, the
Basic
programming language, an auto-sampler, dual FIDs, dual ECDs and a tape
drive
for method and data storage. Remember this was before the advent of the
PC –
crude integrators were the best that was available and auto-samplers
were
previously unheard of. This instrument was way out in front. Wow, did I
have a
toy!
At that time there were no capillary columns
or
internet to learn what the options were. The best source of information
on a GC
was the Supelco catalog – not a bad source even today. I set up the GC
with
2mmID packed glass columns, using both a nonpolar column for the
volatiles and
semivolatiles (V/S) and the standard PCB column of the day. I took the
extra
step of packing the injection ports with glass wool to capture the
nonvolatile
residue from the hazardous waste fuel samples. We used soft graphite
ferrules
to insure good seals on the columns. Each of the programs for obtaining
separation of the V/S and PCBs was about 20 minutes long and the
auto-sampler
would only inject one sample at a time. Add in a cool-down time and
sample prep
and the whole process was well over the 30 minute target for getting
the
testing done. What was to be done?
I hacked the computer control of the GC
before
anyone had invented the word “hack,” fooling the computer into starting
a run
on the rear injection port for the PCB run. I then aborted the run,
moved the
auto-sampler automatically to the front port and injected the next
sample into
the V/S-FID column. Then an integrator run started on the front FID
detector
along with a ramped temperature program. A second run was started on
the rear
ECD detector about 10 minutes later as the PCBs started coming off the
column.
We got excellent separation of both PCBs and V/S compounds and the
whole system
was programmed to run around the clock. Truck and tank samples were run
during
the day as they were prepped and standards were run overnight. The
first time
the HP service guys saw the system operate they about had a coronary.
Later
they convinced themselves that it was pretty cool that they had built a
system
with capabilities they did not even know about.
In those early days robotics in laboratories
or
anywhere else were practically unknown. It was not uncommon for an
industrial
or regulatory approval tour member to get the shock of his or her life
when the
GC would automatically inject and start the next sample analysis
without anyone
touching the equipment. A couple years later we got a second GC – a PE
Sigma
115 with a separate computer control system (a 3600 Data System). I
hacked that
system as well and again had the PE people start out by telling me that
what I
had done could not be done – later they, too, congratulated themselves
for
building such a versatile system.
The HP system
allowed me to write a Basic program
that performed a crude pattern recognition analysis of the ECD trace
for PCB
congener patterns. The latter PE system allowed me to take graphical
overlay
software that PE had developed for IR analysis and modify it with the
assistance of my wife Sue to perform overlays and pattern recognition
on the PCB
congener chromatograms.
Sample prep was the other major struggle at
that
time. S/V prep was accomplished by diluting the sample in carbon
disulphide. CS2
is a nearly universal
solvent and does not show up
to an appreciable extent on a FID. PCBs were not so easy. The standard
PCB
preps of the time were based on PCBs in oil. Because of the presence of
both
chlorinated solvents and polar solvents the normal prep using a
florisil column
would not work. The solvent mix would at times create a situation where
the
PCBs would hang up on the prep column – and it took too long. Another
factor
was the use of hexane – volatile, flammable and toxic. I went another
direction. We diluted the sample in isooctane and then washed the
sample with
concentrated sulfuric acid. The final step was to dry the sample over
anhydrous
sodium sulfate. Additional washings with sulfuric acid could be done if
there
was any indication of continued interferences. The entire prep could be
done in
test tubes and auto-sampler vials with relatively small amounts of
sample and
reagent and took less than 5 minutes.
In the end,
sample prep took about 5 minutes and the
run time on the GC was about 20 minutes. That allowed another 5 minutes
for
data interpretation and review and met our goal of a 30 minute sample
turnaround.
Later developments for the S/V method
included
digestion of the sample with concentrated sulfuric acid and/or
saturated
solutions of sodium hydroxide in methanol to destroy selected organic
compounds
and aid in the positive identification of GC peaks.
Another development was the use of both FID and TCD detection
and
the use of the ratio of the response factor – long before the
availability of
table top mass spec detectors, today’s method of choice.