Evaluation of the Churn Splitter for Inclusion in the Division Protocol for the Collection and Processing of Surface-Water Samples for Subsequent Determination of Trace Elements, Nutrients, and Major Ions in Filtered Water

In Reply Refer To:                                       April 21, 1994
Mail Stop 412


Subject:  Evaluation of the Churn Splitter for Inclusion in the Division
          Protocol for the Collection and Processing of Surface-Water 
          Samples for Subsequent Determination of Trace Elements, Nutrients,
          and Major Ions in Filtered Water


The new inorganic  protocol for filtered water (Office of Water Quality
Technical Memorandum 94.09) requires the use of a modified churn splitter for
purposes of compositing cross-sectional whole water samples.  The new protocol
was developed for the constituents and reporting limits listed in Table 1.  
For the majority of the trace elements, the reporting limit is 1 microgram per
liter (ug/L).  However, the reporting limit for iron (Fe), aluminum (Al), and
zinc (Zn) is  3 ug/L.  This limit was raised well beyond current quantification
capabilities because these trace elements are commonly present in most field 
and laboratory working environments and, therefore, are very difficult to
eliminate as contaminants.  

Sample contamination typically falls into two categories--consistent or
erratic.  The cleaning, handling, and processing procedures included in the 
new protocol are designed to limit consistent contamination to less than half
the stated reporting limits.  The same procedures are also selected because
they substantially reduce the chances of erratic contamination.  The quality
assurance/quality control (QA/QC) procedures and guidelines incorporated in 
the protocol are intended to provide adequate checks on potential sample
contamination.  Further, the QC data generated under the protocol are intended
to provide defensible environmental data of known quality.  Such information 
is a requisite for data interpretation.  


Since the new protocol was published, questions have arisen concerning the data 
supporting the use of the churn splitter.  The purposes of this memorandum are
to provide the justifying data and to certify the churn splitter for use with
the protocol.


As is stated in the protocol, use of the churn requires (1) putting a limited
diameter funnel in the lid, (2) enclosing the churn inside two sealable
polyethylene or polypropylene bags and placing these inside a churn carrier,
and (3) replacing the existing spigot valve with a new one from the Quality of
Water Service Unit (Ocala).  An appropriate number of field blanks (defined in
the protocol) should be run when any samples are collected.  Furthermore,
because the protocol was developed for filtered samples, an additional field
blank should be run (including passing water through the spigot) when
unfiltered samples are collected.


During the initial development of the protocol, a series of laboratory tests
evaluated the effectiveness of preconditioning the processing equipment with
deionized water (DIW) rather than with native water.  Samples from five streams
were processed and analyzed, and the results compared.  Before filtering
samples, a blank was run through the entire processing system (churn splitter,
pump tubing, 142-millimeter (mm) plate filter, and a 142-mm 0.45-um MFS filter)
to evaluate whether the filtrates might be contaminated from the processing
equipment.  The equipment was thoroughly cleaned between each processed sample
using the procedure provided in the protocol.  After all the samples had been
run, the system was cleaned one final time, and a final blank was run in the
same manner as the first.  The results for all the blank samples are in Table
2.  The data listed in the rows marked "Conditioning Blank" represent duplicate
aliquots of DIW used to condition the processing system.  The data listed in
the rows marked "Equipment Blank" come from the two separate aliquots of DIW
that were actually passed through the processing system.  Equipment Blank 1 
was run at the beginning of the tests, whereas Equipment Blank 2 was run after
all environmental samples had been processed.  Little or no contamination was
detected in blanks from the processing equipment.  


During the development of the office- and field-cleaning procedures, questions
arose about potential "contaminant" carryover if a sample was obtained at a
highly contaminated site followed by collection of a sample at a relatively
pristine site.  An evaluation of this potential problem, which would also
provide a test for the proposed field-cleaning procedures, was designed and
implemented.  The two test sites were (1) Davis Mill Creek, in Copper Hill,
Tennessee, and (2) Broad River at Bell, near Elberton, Georgia.  Flow at the
Davis Mill Creek site is highly contaminated by acidic discharges from abandoned
copper mines, as well as by effluent from a chemical company.  The Broad River
site is in a rural, agricultural area.  Both sites had been used extensively
for previous studies on the evaluation of dewatering equipment.  

The Davis Mill Creek site was sampled first.  All the appropriate sampling 
and processing equipment had been cleaned and packaged in appropriate
noncontaminating plastic containers in the office the previous day.  Upon
arrival at the site, a series of environmental subsamples was collected using 
a weighted bottle and composited in a standard 14-L churn splitter.  The
composite sample was processed using a peristaltic pump, silicon pump tubing, 
a GeoTech 142-mm nonmetallic filtering system, and a 142-mm 0.45-um MFS
filter.  The filtrate was split between two bottles and acidified with Ultrex
nitric acid.  After the processing was completed, all equipment was thoroughly
field cleaned.  A new filter was placed in the filter holder, preconditioned
with DIW, and a field blank was processed.  The equipment was then repackaged
in plastic bags for the drive to the Broad River site.  Upon arrival at the
second site, the procedure used at Davis Mill Creek was repeated.  A sample 
was collected, composited, and processed.  The system subsequently was
disassembled and field cleaned, and a final field blank was processed.  

The data for the actual samples and the subsequent blanks are presented in
Table 3.  The Davis Mill Creek site contained substantial quantities of
"dissolved" iron (Fe), manganese (Mn), cobalt (Co), copper (Cu), and zinc
(Zn).  The analytical data for the subsequent blank indicate that the field
cleaning was sufficient to remove any traces of the processed sample as
indicated by the low and/or "less than" concentrations of Fe, Mn, Co, Cu, and
Zn.  The source(s) for the measurable silver (Ag) in the blank is unknown.  
The Broad River sample contained little or no detectable trace elements or
major ions, and the subsequent blank was essentially as clean as the one
processed at the Davis Mill Creek site.  Again, as with the Davis Mill Creek
blank, Ag was detected, the source(s) of which is also unknown.  


During the development of the inorganic protocol, results from a series of
filtration tests indicated that the type of filter used to process whole-water
samples could have a substantial effect on "dissolved" trace-element
concentrations (Horowitz and others, 1992).  A further evaluation of this
phenomenon was planned and carried out at three sites (Mississippi River at 
St. Francisville, Tangipahoa River at Robert, and Big Creek at Pollack) in
Louisiana.  The tests entailed the following: (1) all the equipment was cleaned
following the procedures detailed in the protocol; (2) upon arrival at the
site, a field blank using inorganic blank water (IBW) obtained from the Quality
of Water Service Unit (Ocala) was processed and preserved following the
procedures outlined in the protocol; and (3) a field sample was collected,
processed, and preserved following the procedures outlined in the protocol.
The data for the field blanks using capsule filters are provided in Table 4;
the data indicate that the office-cleaned equipment and the field-collection
and processing procedures used with the IBW are capable of limiting
contamination to acceptable levels.  


Based on mutual interest in further evaluating the effects of filtration
artifacts on "dissolved" trace-element concentrations, a series of tests 
were planned by representatives of the USGS/WRD, Environment Canada, and the
Canadian Geological Survey.  The actual experimental work was carried out by
Environment Canada at their St. Lawrence Center in Montreal.  Five different
samples were collected for processing: (1) a sample from an acid mine drainage
site, (2) a sample from a peat bog, (3) a sample containing a high suspended
sediment concentration from the St. Lawrence River, (4) a sample containing 
a low suspended-sediment concentration from the St. Lawrence River, and (5) 
a sample from a near-neutral or alkaline river site.  The samples were
collected and brought to Montreal for processing.  All the processing equipment
(churn splitter, pump tubing, and various filters and filter holders) was
cleaned per the protocol.  Actual processing was carried out inside a laminar
flow hood in a laboratory.  Before processing each sample, a blank was run
through each system.  After each environmental sample was processed, the system
was recleaned following the field-cleaning procedures described in the
protocol.  Before running the next sample, a new equipment blank was
processed.  This continued until all the samples had been processed through 
a variety of filtration devices.  The acid mine drainage sample was processed
first, followed by the peat bog sample, followed by the others.  

The filtrates resulting from all the processed samples and blanks were split and 
subsequently analyzed by both the USGS  National Water Quality Laboratory (NWQL) 
and the Canadian Geological Survey (Ottawa).  The chemical data from both
facilities are comparable.  Table 5 contains USGS/NWQL-generated chemical data
for both the blanks, as well as the acid mine and peat bog samples (2 of the 
5 samples run during the experiment) processed with capsule filters.  These 
two samples were selected because they were run in sequence and represent 
the worst-case scenario of a sample containing relatively low trace-element
concentrations following the processing of one containing relatively high

The data indicate that (1) the processing equipment started out at acceptably
clean levels, (2) the acid mine drainage sample contained substantial amounts
of selected trace elements (Mn, Ni, Cu, Zn, and Cd), (3) there were elevated 
Al and Zn levels in the acid mine drainage blanks, (4) the cleaning procedure
readily removed residues from the acid mine sample before processing the next
blank, (5) the peat bog sample did not contain excessive amounts of trace
elements, and (6) the residues from the peat bog sample also were readily
removed before processing the next blank (Table 5).  The source(s) for the
elevated Zn concentration in the first equipment blank is unknown; however,
based on the results from the other blanks run during this study it should 
be viewed as erratic rather than consistent contamination (that is, it did 
not result from problems associated with the actual cleaning procedures because
it did not show up in the other blanks).  The Al concentration (1.4 ug/L) in
the second equipment blank is elevated, but is viewed as being at an acceptable
level because it is less than half the Al reporting limit (3 ug/L).  The
source(s) of the elevated B levels in the blanks and samples is unknown at 
the present time.  


Data cited in this memorandum indicate that the cleaning procedures (office 
and field) incorporated in the new protocol limit consistent contamination
associated with the churn splitter to concentrations less than one half the
reporting limit.  The extent to which some trace elements in blank samples 
were detected is typical of erratic contamination detected during normal
quality-control tests.  The data support the view that the cleaning procedures 
outlined in the protocol are appropriate for rendering the churn splitter
sufficiently clean for use at the reporting limits listed in Table 1.  

                               REFERENCE CITED

Horowitz, A.J., Elrick, K.A., and Colberg, M.R., 1992, The effect of membrane
   filtration artifacts on dissolved trace element concentrations:  Water
   Resources, v. 26, p. 753-763.

                                      David A. Rickert
                                      Chief, Office of Water Quality

Attachment (see hard copy)

This memorandum refers to Office of Water Quality Technical Memorandum 94.09.

Key Words:  Analysis, protocol, surface water

Distribution:  A, B, S, FO, PO, AH