United States
Environmental Protection
Agency
Health Effects Research
Laboratory
Cincinnati OH 45268
Research and Development
                                   EPA-600/S1-81-062  Feb. 1982
Project  Summary
Organic  and  Organotin
Compounds  Leached  from
PVC  and  CPVC  Pipe
Edward A. Boettner, Gwendolyn L.  Ball, Zane Hollingsworth, and Romulo
Aquino
  The primary objective of this research
program was to determine whether or
not organic  forms of tin existed in
potable water samples after contact
with either polyvinyl chloride (PCV) or
chlorinated polyvinyl chloride (CPVC)
pipe containing organotin heat stabi-
lizers. Analytical techniques to identify
and quantitate organotin compounds
at concentrations below one part per
billion (ppb) were explored and refined.
Four test protocols were used to
simulate the  extraction process. First,
during  method  development, frag-
mented pipe samples were digested
with extractant  water in  order to
obtain high surface-to-volume ratios.
resulting in higher organotin concen-
trations than would be encountered in
the field. Second, extractant water
was continuously pumped through an
8.9-meter rectangular closed loop of
one-inch internal diameter pipe having
an inner surf ace area of 0.71 m2(1100
in2}, maintained  at constant temper-
ature. This system was used to provide
sequential samples  over a 22-day
period, during which time the  ex-
tractant water was  completely  re-
moved  for analysis and  replaced at
two- to four-day intervals. Third, a
miniature pipe system attached to a
laboratory faucet was used primarily
to study the fate of the solvents
incorporated in cements used to join
segments of  pipe. Fourth, incubation
of short lengths of pipe with extractant
water served as a convenient method
to study changes in leach rates as
experimental parameters such as pH
were varied.
  Commercial pipe samples stabilized
with dialkyltin-bis-isooctylthioglyco-
late  compounds were tested. The
extractant water was prepared using
doubly-distilled water and had hard-
ness, buffer,  and chlorine contents
similar to that of "Standard" extract-
ant water recommended by the Na-
tional Sanitation Foundation  in its
February 1977 Standard 14.
  Organotin analysis at these very low
levels utilized hydride derivatization,
followed by collection of the hydrides
on glass bead or OV-1 traps immersed
in liquid nitrogen, and detection of tin
by atomic absorption spectrophotom-
etry as each hydride eluted sequentially
from the trap after the liquid nitrogen
was  removed. The exact organotin
species from  which these hydrides
arose were not identified,  although
there was evidence that they existed in
ionic form in the extractant water. No
evidence of the sulfur-containing part
of the stabilizer was found in the
extractant water.
  The results showed  that alkyltin
species were extracted from the
tested PVC and CPVC pipes by water.
The amount of dimethyltin (as the
dichloride) leached from PVC into pH
5 extractant water at 37°C was 35
ppb for day one, and decreased from

-------
approximately 3.0 to 0.25 ppb per 24
hours, in a biphasic manner, from days
2 through 22. The amount of dibutyltin
(as the dichloride) leached from CPVC
into  pH 5 extractant water at 72°C
was 2.6 ppb for day one, and decreased
from  1.0 to 0.03 ppb per 24 hours,
again in a biphasic manner, from days
2 through 21.
  Volatile organic solvents (methyl
ethyl ketone, tetrahydrofuran, and
cyclohexanone were monitored) used
in the sealing cements applied to PVC
and  CPVC pipe  joints  continued  to
leach into water supplies for more
than 14 days using the miniature pipe
system. The  quaritities  ranged from
10 ppm to 10 ppb during the 15 days
of sampling.
  Sufficient toxicological data are not
available to assess the  health signif-
icance  of the very low levels  of
organotin and cement solvent chem-
icals found in this study.
  This Project Summary  was devel-
oped by EPA's Health Effects Research
Laboratory.  Cincinnati,  Ohio,  to an-
nounce  key findings  of the research
project  that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).

Introduction
  Approximately 800,000 metric tons of
polyvinyl chloride (PVC) and co-polymer
were produced for  pipe and conduit
applications in 1977. A portion of this
was  used in potable water supply sys-
tems because of cost  and  handling
advantages over other materials, mainly
metals.  Use  of PVC  and  chlorinated
polyvinyl chloride (CPVC)  pipe  for
potable  water raises  the  question of
whether chemicals  of a toxic nature
diffuse  from  the  pipe into the  water
supply,  and if so, in  what quantities.
Organotin compounds used as thermal
stabilizers in PVC  and  CPVC pipe
formulations are of particular concern if
leached, due to their toxicities.
  The types of PVC and CPVC used for
pipe are rigid materials with little or no
plasticizer. They  contain  compounds
such  as thermal stabilizers, lubricants,
fungicides, fillers, and pigments to aid in
processing and prolonging useful life.
PVC undergoes complete dehydrochlo-
rination at 300°C, and partial dehydro-
chlorination  at lower temperatures.
Thus, a thermal stabilizer must be added
to prevent undue degradation  of the
material during processing. The pres-
ence of hydrogen chloride,  or  the
chloride  radical, accelerates the  de-
composition of PVC  and CPVC.  An
important function of a thermal stabilizer
is  to  react with any chloride formed
during processing to eliminate  it.
Stabilizers used with PVC and CPVC in
the United States  are organometallic
salts of tin, calcium, zinc, calcium-zinc,
and  antimony.  Organotin compounds
are the most widely used for stabilizing
PVC  and  CPVC potable  water pipe.
Specific compounds used  include
methyl-,  butyl-, and  octyltin esters,
particularly of lauric,  maleic, and
thioglycolic acids.  Nine manufacturers
have organotin  stabilizers accepted by
the  National Sanitation Foundation
(NSF) for  use in PVC and CPVC potable
water  pipe formulations. Organotin
stabilizers are used in the range of 0.3 to
1.5 parts per 100  parts resin for PVC
pipe  and fittings, and in the range of 1.5
to 3.5 parts per 100 parts resin for CPVC
pipe  and fittings.
  Whenever a liquid and solid phase are
in contact, there  is opportunity for
components of the liquid phase to plate
out onto the solid, and for components
of the solid to be leached into the liquid
phase. It  is quite possible that potable
water could leach some compounds
from PVC and CPVC pipes, particularly
additives that are mixed with  the
polymer  and not chemically reacted
with it. Factors influencing the leaching
process include pH, temperature, ionic
composition of the water, exposed
surface area and surface porosity of the
pipe  material,  water solubility of the
polymer additives,  the ability  of these
additives to migrate from within the pipe
to the surface, and the reactivity of the
additives with each other and with the
polymer.  Organotin compounds theo-
retically react with the polymer during
stabilization so that only their reaction
products  are available for leaching, in
the form  of alkyltin chlorides. However,
excess stabilizer is usually present in
the pipe.
  The primary objective of this research
was  to  determine whether or  not
organotin compounds were introduced
into  potable water when  using PVC or
CPVC pipe containing organotin com-
pounds  as stabilizers.  A secondary
objective was to determine the quantities
of other selected  organic compounds
which might be leached from these pipe
systems into  the water supply.  The
approach used was to  first  establish
analytical methodology capable of
determining the  extracted chemicals
present at very low levels, and then to
incorporate that methodology  in a
system to quantify the organotins and
other compounds in water supplies
after contact with PVC and CPVC  pipe.
The  results obtained are a necessary
component  of the  data  needed by
toxicologists  to determine  whether or
not the use of plastic pipe for potable
water distribution presents a  health
hazard.

Materials and  Instrumentation
  Asa result of a letter to compounders,
pipe fabricators, and manufacturers of
tin  stailizers,  we received 17  pipe
samples from eight manufacturers and
six resin  samples from which some of
the  pipe samples were formulated.
Eight stabilizer samples were received.
There were only two negative responses,
and 19 manufacturers failed to respond.
  Two of the pipe samples were used in
pipe-loop experiments. The two samples
chosen were representative of widely-
used PVC and  CPVC pipe compounds.
They incorporated primarily dimethyltin-
and  dibutyltin-bis-isooctylthioglycolate
stabilizers, respectively,  although sta-
blizer formulations may contain mixtures
of alkyltin isooctylthioglycolates. In
fairness to manufacturers, and in
accordance with our letter requesting
samples, all pipe and stabilizer samples
are referred to  by code.
  PCV and CPVC Schedule 80  gray
fittings were used in joining the PVC
and  CPVC loops, respectivley. These
fittings met NSF standards. Pieces to be
joined were cleaned using a commercial
cleaner-primer containing  2-butanone
and  cyclohexanone. The PVC loop was
assembled using a commercial solvent
cement containing tetrahydrofuran, 2-
butanone, cyclohexanone, N, N-di-
methylformamide, and  dissolved PVC
resin. The CPVC  loop was assembled
using a  similar  product containing
tetrahydrofuran, 2-butanone, cyclohex-
anone, and dissolved CPVC resin. The
cleaner  and  cements were purchased
locally and  met  NSF and American
Society  for Testing  and Materials
standards. Solvent  cement and pipe
samples were analyzed by emission
spectroscopy prior to use  to establish
that they contained tin stabilizers.
  Organotin  reference compounds
were purchased from: K & K Rare and
Fine Chemicals, Life Science Division of
ICN  Pharmaceuticals, Inc., Plainvlew,
New York;  Alfa Division, Ventron
Corporation, Danvers, Massachusetts;
Eastman Organic Chemicals, Rochester,
New York; and the National Bureau of

-------
Standards,  Washington,  D.C. All  were
used as received, with the exception of
dibutyltin dichlonde which was recrys-
tallized from hexane prior to use. Meth-
yltin trichloride was the only compound
having major impurities. Minor impurities
totaling <2% (w/w) were often detected
by gas chromatography,  and all of the
organotin chlorides analyzed by  mass
spectrometry had detectable organotin
bromide contamination. The five  com-
pounds used in preparation of  quanti-
tative standards for this study  had no
detectable (>1 % w/w) impurities. Both
samples of dibutyltin-bis-(2-ethylhex-
anoate) had certified tin contents, but
there  was no assurance that all the tin
existed in the organic form. There is a
general  lack of certified organotin
reference compounds.
  Aqueous solutions of alkyltm  halides
in the <0.1  - 10 ppm range  were
prepared fresh daily by serial dilution of
1000-ppm  stock solutions. Water was
used as solvent for  the methyltin halide
stock  solutions, while acetone  or 95%
(v/v) ethanol was  used  as solvent for
stock  solutions of the  other  alkyltin
halides. Aqueous standards of organotin
compounds have often been prepared
by dilution  of high-concentration  stock
solutions in ethanol, acetone, or  other
water-miscible solvent. There was
considerable potential for solvent-
induced transformations  of the  original
organotin compound to occur with this
method of standard preparation. These
transformations  generally did not
hinder the analysis of alkytin species, as
the tin-carbon bond was quite stable.
These transformations did, however,
preclude analysis of the anionic  portion
of the compound.
  Aqueous standard solutions of tetra-
hydrofuran (Fisher  T-397), and  methyl
'ethyl  ketone  (Fisher  M-209)  were
prepared by serial dilution of an aqueous
stock solution containing  100 /jL of each
chemical   per  100 mL of solution.
Cyclohexanone  (Baker G032) standard
solutions  were prepared  by direct
injection of  microhter  quantities of
solvent into water. None of these
compounds had detectable (>1%  w/v)
impurities when analyzed by gas
chromatography.
  The extractant water  composition
was similar to that recommended in the
February  1977 National Sanitation
Foundation Standard 14. Each  liter
contained  111  mg calcium chloride
(Matheson Coleman & Bell CX170)with
hardness equivalent to 100 mg calcium
carbonate  per  liter, 84 mg  sodium
bicarbonate (Matheson Coleman & Bell
SX0320)  as  buffer, and  1.0 mg  of
chlorine (0.5 mL of a stock 2 g chlorine
per liter  solution, prepared by adding
7.65  mL  of  Chlorox,  a commercial
5.25% (w/v) solution of sodium hypo-
chlonte, to 200 mL of water). A pH of
approximately 5 was obtained  by
bubbling  the solution with carbon
dioxide (Linde Bone Dry grade).
  A Jarrell-Ash  JA 82-270 atomic
absorption unit with a Jarrell-Ash FLA-
10 Flameless Atomizer  was used for
both flameless and hydride-generation
analyses. A  Hewlett-Packard 5730-
series gas chromatograph  with flame
ionization detector and a  Hewlett-
Packard 3380A integrator  were used.
All columns were 4-mm  internal di-
ameter. The mass  spectrometer used in
this study was an  Associated Electrical
Industries MS-30, a  double-beam,
double-focusing  magnetic sector in-
strument. It was  interfaced to a Pye
series 104 gas chromatograph via  a
silicone  membrane separator, and
coupled  to an Associated  Electrical
Industries DS-30 data system.

Experimental Procedures

Analytical Methods
  Two closed-system hydride generation
techniques for the  detection and quanti-
tation  of alkyltin  species  in environ-
mental samples have been proposed.
Braman developed a detector which
measured SnH-band emission as the
sample was passed through a hydrogen-
rich hydrogen-air  flame. Hodge devel-
oped  a method using a  hydrogen-rich
quartz-tube burner for the atomic
absorption analysis of alkyltin hydrides.
The applicability of the hydride-genera-
tion methods  of Hodge and Braman to
the analysis of both potable water, and
water extracts of plastic pipe intended
for potable water  transport was thor-
oughly investigated. Our apparatus had
only  minor dimensional differences
from that used by  Hodge.
  The hydride generator was a standard
125-mL gas washing bottle to which a
septum port  for  addition  of sodium
borohydride was added 1  1/2-2 inches
above the base. The sample was purged
with  helium to  volatilize any hydrides
formed and to transfer them efficiently
to a trap. The hydride trap consisted of a
20-cm, 4-mm internaldiameterglassU-
tube packed with 12 cm of 60-70 mesh
glass  beads with glass wool plugs. This
trap was immersed in liquid nitrogen to
collect hydrides purged from the hydride
generator. The liquid nitrogen was then
removed and the trap allowed to warm
to room temperature to release the
trapped hydrides  according  to  their
boiling  points. The  trap had  to  be
immersed in an ~80°C water bath in
order to release tributyltin hydride. The
hydride  burner consisted of  a  small
Vycor tube  positioned  in the atomic
absorption spectophotometer so that
radiation from the hollow cathode lamp
passed down the length of the tube. At
the midpoint of this tube, air plus helium
and sample entered through lines at the
front, and hydrogen entered through a
line at the rear. These gases mixed at
the center of the tube and travelled to its
ends, where they were burned in small
flames to produce atomic tin which was
then quantitatively determined by the
magnitude of its absorbance at 224.6,
235.5, or 286.3nm.
  Investigation  of  alternative packing
materials  for  the hydride trap was
conducted to improve the separation of
organotin hydrides. The use of 18 cm of
10%(w/w)OV-1 on 80/100 mesh Gas-
Chrom Q improved separation of meth-
yltin  hydrides. After the methyltins
eluted, the trap was heated to 145°C to
elute the di- and tributyltin hydrides.
  Both standard calibration curves and
the method of standard additions were
used  with  the hydride-generation
atomic absorption technique. Calibration
curves were linear over an  approxi-
mately  20-fold  concentration range
which varied with the instrument gain.
The lowest range used in this work was
0.05  -  1.0  ppb. Results of triplicate
analyses at  the  same organotin  con-
centration in distilled water generally
varied  by less than  10%.  Due  to
suspected  matrix effects, the method of
standard additions was  selected for
quantitative  organotin analysis. Dupli-
cate or triplicate analyses of the sample
and each  addition were performed.
Standard aqueous  organotin solutions
for  addition were  prepared  by  serial
dilution  of an ethanol  stock solution.
  Grignard  alkylation is one of the
classic reactions of organic synthesis,
involving addition of an alkyl group to a
compound through an alkylmagnesium
halide intermediate. Grignard butylation
and  methylation were  applied by
Meinema to produce tetraalkyltin com-
pounds from various undefined alkyltin
species  in the aqueous  environment.
The reaction worked especially well for
alkyltin  halides,  which  were  used  to
prepare  tetraalkyltin standards.  Other
alkyltin compounds should react directly.

-------
or after addition of acid since tin bonds
with elements other than carbon are
easily cleaved.
  Although initial analyses of butylated
water extracts of  PVC and CPVC pipe
showed considerable promise, they also
suffered  from the disadvantage that
tetrabutyltin derived from inorganic tin
could not be distinguished from tetra-
butyl tin derived from butyltin stabilizer.
Since the use of pentyltin compounds as
stabilizers had not been  reported, we
prepared pentyltin derivatives instead of
butyltin derivatives.
  An  interfering  gas  chromatograph
peak  coeluted with  the  dibutyltin
derivative due to a derivatization by-
product  of the  complexing  agent,
tropolone. To overcome this problem, a
two-step extraction was developed.
First,  an extraction of a hydrobromic
acid-acidified water sample with ben-
zene was used to recover dibutyltin and
tributyltin species. The acidified water
sample was then extracted with benzene
containing  0.05% (w/v)  tropolone to
recover the monobutyltin and inorganic
tin species.  The two extracts were
derivatized  separately, using n-pentyl-
magnesium bromide (1.5 -  2.5 M in
ether, Alfa 87296). The derivatives were
analyzed by gas chromatography on a
column of  10% (w/w) OV-1  on Gas-
Chrom Q.
   The limit of detection of the Grignard
alkylation technique, corresponding to
approximately 1.5 - 3 ug alkyltin species
in 1 L water, was not as low as that of
the hydride-generation atomic absorp-
tion  method.  However,  the Grignard
alkylation method provided  a  sample
suitable for analysis by mass spectrom-
etry, allowing verification of the presence
of alkyltin species.

Extraction Experiments
   Fragmented pipe samples were used
to  provide  a large surface area for
determining what  compounds were
water extractable.  One- to two-inch
internal  diameter pipe was sawed into
four- to  six-inch  lengths and  cleaned
before (and after)  fragmenting.  The
plastic was fragmented by freezing it in
liquid nitrogen, wrapping it in clean
duck canvas, and  hammering on a hard
supporting surface. This process resulted
in  irregularly-shaped, 0.35-cm thick
pieces, 0.25 to 1.5 cm in  length and
width. About 400 grams of pipe frag-
ments were placed in a 500-mL Pyrex
bottle which  was then filled with
extractant water (approximately 200
mL).  A 5-cc  air space was left for the
expansion of bottle contents on heating.
Each bottle was sealed with a ground
glass stopper. A blank, consisting of a
bottle filled only with extractant water,
was run  with each  set of fragmented
pipe samples.
  The basic component of the pipe-loop
system was a 1.71 -m (5.6-foot) by 2.74-
m (9-foot) rectangular closed loop of
one-inch  internal diameter pipe having
an inner surface area of 0.71 m2 (1100
in 2).The loop of 8.9-m (29.2-foot) total
length  was formed using conventional
NSF-accepted  elbows  and solvent
cement as described in  the  Materials
and Instrumentation section. Test water
was circulated in the loop by a centrif-
ugal pump (Process Controls  Co.,
Livonia, Ml 48152. Pump: Dayton Model
6K122. Pump head: Liquiflo, Warren,
NJ 07060) having only wetted stainless
steel and Teflon parts. A galvanized pipe
loop  of the same length and internal
diameter  was mounted adjacent to the
plastic  loop and utilized an  identical
pump.
  Each loop was filled and  drained by
means of a fill tube that fit directly into
an access port at the bottom of the pump
head. This port was  the lowest point in
the system. Another port at the top of
the pump head, the highest point in the
system,  served  as a  vent and an
overflow  indicator to signal  when the
loop was full. Extractant water volumes
required  to fill  the control, PVC and
CPVC loops were 4.8.4.1, and 3.9 liters,
respectively. Volume differences prob-
ably reflect variations in the nominal
internal diameters  of the pipes. The
loops  were housed in  a controlled-
temperature heated test  chamber, built
of sheets of polystyrene insulation.
Heating elements were incorporated to
heat the chamber to 37°C when testing
the PVC loop and 72°C when testing the
CPVC loop.
  A report by Wang  and Bricker stating
that cement solvents could be detected
in water six to eight  months  after
installation of a PVC pipe system, led us
to set  up a miniature pipe system to
address this  problem specifically. The
system was  attached to a   laboratory
faucet and was  assembled using  a
galvanized  fitting, two gray  PVC con-
nectors,  four gray PVC  elbows, three
0.102-m  (4-inch) pieces of PVC sample
#9A, and two 0.40-m (16-inch) pieces of
PVC sample #14A. The last component
of the  system was  a brass  gate  valve
positioned over a  sink. There were 10
solvent-cemented joints  in this 1.22-m
(48-inch) miniature pipe system. The
inner  surface area of the plastic was
approximately 0.1  m2(151 in2), and the
volume of the system was approximately
600  mL. The NSF-accepted solvent
cleaner-primer and  solvent cement,
described in the Materials and Instru-
mentation section, were used.
  When the gate valve was closed,
water  resided in the system under
normal line pressure. When the water
was turned off at the faucet, the gate
valve could be opened and water from
the system  collected  in a  Pyrex bottle,
undiluted for an analysis, after a known
residence time. Samples were  taken
periodically over a 2-week period after
4- to  20-hour residence times in the
system at normal  line pressure. At all
other  times, water flowed through the
system  at  approximately  4 liters per
minute. Water to serve as a blank could
be drawn from the same line using the
faucet at the sink.
  In the pipe incubation  method, one-
inch PVC or CPVC pipe was cut into 2-
inch lengths, placed  in a wide-mouth
Pyrex jar, and covered with  extractant
water with  no headspace in the con-
tainer. The  amount of pipe and water
was such that there was 4mL of water
per square inch of exposed pipe surface,
as  recommended by NSF in their
standard testing  protocol. Incubation
was carried out at 25°C with continuous
shaking for periods of time  up to  168
hours. The water was  removed for
analysis and replaced with fresh water
every 24 hours, unlike the pipe loop
system where the water was removed
for analysis and replaced at 48- to 72-
hour intervals.

Results and Discussion

Analysis  of Pipe  Fragment
Extracts
  A fragmented sample of CPVC #6 was
digested in  pH 5.7 extractant water for
72 hours at 82 ± 3°C and the resulting
water sample was reacted with sodium
borohydride. The  pH  of the extractant
water, adjusted  by  bubbling carbon
dioxide through it, varied  during initial
tests as bubbling rates and times were
explored. The 82°C  temperature was
the maximum reasonable temperature
for testing  CPVC  pipe. The extractant
water from the CPVC #6 digestion was
then extracted with  diethyl ether and
the concentrated ether  extract was
analyzed by gas chromatography. The
chromatogram, obtained on a 3% (w/w)
OV-17 column held  at an isothermal
temperature of 90°, revealed six peaks.

-------
none of which appeared in the blank.
The  sample was reanalyzed by  gas
chromatography/mass spectrometry.
The two major peaks were identified as
octanol isomers. No evidence of either
tin or chlorine was found in any peak.
  Pipe fragment extracts of PVC sample
#14A and  CPVC sample  #14C were
analyzed by the Grignard alkylation
procedure. The butylated extract of PVC
sample #14Acontained dibutyldimethyl
tin and tributylmethyl tin, indicating the
presence of methyltm species in  the
water. A butylated extract of CPVC #14C
contained tetrabutyl tin. A methylated
extract of the same sample contained
dibutyldimethyl tin,  indicating  the
presence of butyltin species in the water
after contact with  CPVC pipe.  All
identifications were confirmed by mass
spectrometry.

Miniature Pipe System
  The continuous-flow pipe system was
set up for collection of water samples for
cement solvent analysis. Analysis was
performed by direct injection of a 5-juL
portion of the water sample onto a  gas
chromatograph column packed with 1 %
(w/w) SP-1000 on 60/80 Carbopack B,
maintained at an isothermal temperature
of 170°C. Standards for tetrahydrofuran,
methyl ethyl ketone, and cyclohexanone
analysis were prepared by direct injec-
tion of known amounts of the solvents
dissolved in water.
  N, N-dimethylformamide  was  not
analyzed. Cyclohexanone was only
detectable at the 1 -2 ppm level in Day 1
samples. The only samples  in which
ppm-quantities of solvents were detected
in a flowing system were taken on Day
1. On subsequent days, one sample was
taken after a 6-9 hour residence time in
the system. Flow was maintained at ~2-
4 liters per  minute at all other times.
  Three  conclusions  were apparent
from this experiment. First, ppm quanti-
ties of tetrahydrofuran and methyl ethyl
ketone (<0.5 -  5 ppm each),  and  ppb
quantities of methyltin species (0.1 -0.8
ppb dimethlytin) were detectable more
than two weeks after installation of the
system and after over 10,000 gallons of
water had flowed through  the system.
Second, there  was no doubt  that  the
substances  tested could  have been
detected longer than two weeks. Third,
there was the distinct possibility that
our building or municipal water supply
contained low-ppb levels  of organotin
species,  since blank samples drawn
from an adjacent faucet showed three
unidentified alkyltins.
  Analysis for these same cement
solvents was  performed  on water
samples collected from the pipe loop
experiments (next  section). However,
the results from  this miniature pipe
system were  more  meaningful  in
demonstrating the unexpected persist-
ence of the solvents, inasmuch as the
amount of water used in the miniature
pipe system was orders of magnitude
greater than the amount of water used
in a pipe loop.

Pipe Loop Experiments
  In each experiment a control loop of
one-inch  internal  diameter galvanized
pipe,  assembled  using galvanized
fittings, was employed and sampled
concurrently to provide blank samples.
Only one sample from a control loop had
detectable amounts of organotin spe-
cies, which have been subtracted from
the reported  results.  The pH  of the
extractant water used to fill each loop
was 5.0 + 0.2.
  The first experiment  utilized PVC
sample #14A  pipe, assembled  using
solvent-cemented PVC elbows and one
metal connector. The stabilizer used in
formulating this pipe was dimethyltin-
bis-isooctylthioglycolate. A total of nine
samples were collected from each loop
during the exposure period of 21 days
and 20 hours. Each loop was filled with
extractant water, the water was contin-
uously pumped through the loop for 24-
to 96-hour periods, the entire  water
sample from the loop was drained into a
glass carboy, and the loop was refilled
with extractant water. Each sample
collected from each loop was analyzed
for  organotin species and for cement
solvents on the day it was collected.
  The water samples contained small
amounts of solid particles, less than 1 -
millimeter in length and present in two
physical forms. Some were opaque and
irregular, as though fragmented from
the pipe. Others were translucent flakes
or spheres, similar  in appearance  to
flakes of dried solvent cement obtained
from around the top of the container.
Because this paniculate matter affected
the reproducibility of repeated runs on
each sample, the samples were filtered
prior to analysis.
  Quantitative  alkyltin analyses were
carried out using hydride-generation
atomic absorption and the method  of
standard additions. Results of organotin
analysis of the nine samples from the
PVC sample #14A pipe loop are given, in
several different  forms,  in  Table  1.
Monomethyltin was only found in the
sample from the first contact period,
where it gave a peak height of about one
twentieth that of the dimethyltin.
Results for the dimethyltin  and tri-
methyltin  species are  expressed in
terms of their respective chlorides. The
ppb/contact  time (1-4  days) is given,
along  with  the  concentration in  ppb
standardized to a 24-hour time period
and the cumulative  amount  of each
compound, in /ug,  leached during the
course of the test. Statistical information
on the precision of analysis, in terms of
the  mean  and standard deviation, is
given in the ppb/contacttime column. A
total of 223.6 ug dimethyltin dichloride
and of 71.6 ug trimethyltin  chloride
leached into the extractant water during
this pipe loop experiment.
  Results of cement solvent analyses of
the nine samples from the PVC sample
#14A pipe loop are shown in Figure 1.
Cyclohexanone was only detectable in
the first five samples (at greater than 1
ppm), but tetrahydrofuran  and methyl
ethyl ketone were detectable in all nine
samples. This again demonstrated the
unexpected persistence of these water-
soluble chemicals.
  The second experiment utilized CPVC
sample #14C pipe,  assembled using
solvent-cemented  CPVC elbows  and
connectors.  The stabilizer  used in
formulating  this pipe was primarily
dibutyltin-bis-isooctylthioglycolate. Fill-
ing, operating, and sampling the loop
were performed as described  for the
PVC pipe loop. A galvanized control loop
was tested concurrently. Seven samples
were collected from  each  loop during
the 21-day exposure  period. The CPVC
loop extractant  water samples were
filtered before analysis to  remove the
persistent particle contamination.
  In the first few analyses of samples
from this  CPVC  loop, an interference
eventually  attributed to  chlorine was
encountered in the hydride-generation
analyses.  After  tracing  the problem,
samples were dechlorinated by adding a
stoichiometric amount of sodium thio-
sulfate as part of the analytical process.
However, this did not solve the inter-
ference problem completely, in that it
was found that the amounts of alkyltin
hydrides formed were not proportional
to the amounts  of sample used. The
analytical  procedure  was changed to
work around this problem. The altered
procedure  involved  running  all  the
samples under the same instrumental
conditions, using the same amount of
sample for each analysis. Likewise, the
same volumes  of extractant water

-------
Table 1.
Sample
Results of Organotin Analyses of Samples from the PVC Sample #14A Pipe Loop*
Total Contact Dimethyltin species, as (CHskSnClz Trimethyltin species, as (CH3)aSnCr
Number Elapsed


1
2
3
4
5
6
7
8
9
Time,
days
1
3
6
8
10
13
16
20
22
Time of
Sample,
days
1
2
3
2
2
3
3
4
2

1

35. ±2.0
4.8 + 0.5
1.0 ± 0.2
1.0 + 0.1
5.6 ± 0.6
3.9 + 0.6
1.5 + 0.1
1.0 + 0.1
0.7 + 0.1

2

35.
2.4
0.33
0.50
2.8
1.3
0.50
0.25
0.35

3

143.5
163.2
167.3
171.4
194.4
210.4
216.6
220.7
223.6

1

13. ± 1.2
1.7 ±0.1
0.21 ±0.04
0.14 + 0.01
0.90 ±0.09
0.7 5 ±0.11
0.39 ± 0.01
0.24 ± 0.03
0.10 ±0.02

2

13.
0.85
0.07
0.07
0.45
0.25
0.13
0.06
0.05

3

53.3
60.3
61.2
61.8
65.5
68.6
70.2
71.2
71.6
*The surface area of pipe in contact with extractant water was approximately 1100 in 2. In comparing these results with results
 of other tests it is imperative to take into account the ratio of exposed pipe surface area to extractant water volume, which in this
 case was approximately one square inch to 3.7 mL.

* Column 1:  Results expressed in terms of ppb/contact time; mean + standard deviation.
 Column 2:  Results expressed in terms of ppb/24 hours; mean.
 Column 3:  Results expressed in terms of cumulative fig leached during the 22-day study; mean.
containing known amounts of the
alkyltin chlorides (dimethyl, trimethyl,
monobutyl, and dibutyl) were run as
standards. One milliliter of 0.07% (w/v)
sodium thiosulfate solution was added
to both samples  and standards to
neutralize  the chlorine effect.
  The contact period and  analytical
results are listed in Table 2. A total of
35.7 //g dibutyltin  dichloride  leached
into the extractant water during this
pipe loop study. Dimethyltin dichloride
and trimethyltin chloride were also
observed in some of the CPVC pipe loop
water samples.
  Water samples from the CPVC sample
#14C  pipe loop were also analyzed for
three of the solvents introduced by the
solvent cement. Results (Figure 2) were
about the  same as in the PVC experi-
ment, although the cements used were
slightly different for the two polymers.

Pipe Incubation Results
  Organotin chloride concentrations in
water samples analyzed in the inde-
pendent pipe incubation study of CPCV
sample #4 were remarkably similar to
those in the CPVC sample #14C pipe
loop study. A rapid initial decrease in
alkyltins was noted, followed by  a
secondary increase  and subsequent
decrease. The same species (monobutyl-,
dibutyl-, dimethyl-, and trimethyltin)
were  found. The reason for the sec-
ondary maximum organotin concentra-
tion may be understood more fully after
completion of the independent pipe
incubation studies.
Quality Assurance  Testing
  This   project  was   conducted  in
accordance with EPA Quality Assurance
requirements. Quality assurance sam-
ples were analyzed along with the PVC
sample #14A pipe loop samples for both
methyltin species and for cement
solvents.  Retention times and peak
heights produced by four aliquots of a
working standard containing the three
butyltin chlorides, analyzed successively
by the hydride-generation  atomic ab-
sorption procedure, varied by less  than
10%.

Conclusions and
Recommendations
1. Alkyltin species were extractable
   from PVC and CPVC pipes by water.
   The amount of  dimethyltin (as the
   dichloride) leached from PVC into pH
   5 extractant water at 37°C was 35
   ppb for day one, and decreased from
   approximately 3.0 to 0.25 ppb per 24
   hours  in  a  biphasic manner  from
   days 2 through  22. The amount of
   dibutyltin (as the dichloride) leached
   from  CPVC  into pH 5  extractant
   water at 72°C was 2.6 ppb for day
   one, and decreased from 1.0 to 0.03
   ppb per 24 hours, again in a biphasic
   manner, from days  2 through 21.
   Both results  were  obtained when
   pumping extractant water continually
   through an  8.9-meter  rectangular
   closed loop of one-inch internal
   diameter pipe having  an inner
   surface area of  0.71 m  (1100 in2),
   over a 22-day period, during which
time the  water was completely
removed for analysis and replaced at
two- to four-day intervals.
a.  The  exact organotin compounds
   from  which these alkytin species
   arose were indefinite, although
   there was evidence  that  they
   existed in  ionic form  in  the
   extractant water.
b.  No evidence of the sulfur-con-
   taining  part of the stabilizer was
   found in the extractant water.
c.  The  dynamics of the extraction
   process were  not fully resolved,
   but the  rate of decrease at the end
   of 21 days indicated that detec-
   table amounts (0.005 ppb per 24
   hours contact time) of organotin
   species would be present for long
   periods of time  under normal
   service.
d.  Experiments should be performed
   to determine  leaching rates at
   other pH and temperature condi-
   tions. Testing should  be carried
   out over longer time periods (6-12
   months) to determine whether or
   not extraction of organotin species
   is a continuing process.
e.  Future studies should be directed
   toward determining whether these
   levels present a  human health
   hazard. Therefore, further analyt-
   ical   work, as outlined in  (d),
   should  be performed in conjunc-
   tion  with toxicity testing of the
   extractant water samples, or
   equivalent synthetic samples at
   the low-ppb level.

-------
     10,000
      5,000
I
      1,000
       500
I
        TOO
         50
         10
                                            Methyl ethyl ketone
                                            Tetrahydrofuran
                                            Cyclohexanone
                              \   \
                                     \  \ \ \
                                                           \   \  \ \ \  \
            10
50       100
     Time (hours)
                                                 500
1,000
igure 1.
Standardized cement solvent concentrations observed in PVC sample
#14A pipe loop water samples.
             of CPVC  pipe when  the pentyl
             derivatives were used,  rather than
             the methyl and butyl  derivatives
             recommended by Meinema.
             Other  conventional   methods   of
             concentrating, separating, and ident-
             ifying the organotin compounds  of
             interest did not prove to be readily
             applicable. These included high-
             pressure  liquid chromatography,
             ion-pairing chromatography, flame-
             less  atomic absorption, and thin-
             layer chromatography.  Mass spec-
             trometry was useful in conjunction
             with  the  hydride  and Grignard
             derivatization methods for identi-
             fication of various constituents.
  Volatile organic solvents (methyl
  ethyl  ketone, tetrahydrofuran,  and
  cyclohexanone were monitored),
  used in the sealing cements applied
  to PVC and CPVC pipe joints, con-
  tinued to leach into water supplies
  for more than  14 days using the
  miniature pipe system. The quantities
  ranged from 10 ppm to 10 ppb during
  the 15 days of sampling.
  The hydride-generation methods of
  Hodge, Bra man, and others permitted
  organotin analysis at concentrations
  below 0.01  ppb.
                              a.  Matrix effects were encountered
                                 using hydride generation  Al-
                                 though the major effect resulted
                                 from  chlorine in the sample  and
                                 was overcome by adding sodium
                                 thiosulfate to the sample, con-
                                 siderably more work is needed to
                                 understand  and control  more
                                 subtle  matrix effects. A single
                                 analytical procedure may not be
                                 suitable for all sample matrices.
                              The Grignard derivatization method
                              was  applicable  to  the  organotin
                              species encountered in water extracts

-------
Table 2.    Results of Organotin Analyses of Samples from the CPVC #14C Pipe Loop"
Sample Total
Number Elapsed
Time,
days
1
2
3
4
5
6
7
1
3
6
10
14
17
21
Contact
Time of
. Sample,
days
1
2
3
4
4
3
4
Alkvltin species, as
(CH^i2SnCI^ (CHsJsSnCI^ C^H^SnCI^ {CArl^^SnCl2
1
0.61
0.34
0.06
0.12
0.12
2
0.61
0.17
0.02
0.03
0.03
3
2.4
3.7
3.9
4.4
4.9
4.9
1
0.30
<0.01
2
0.30
<0.01
3
1.2
1.2
1.2
1
0.63
0.66
0.39
0.08
2
0.63
0.33
0.13
0.02
3
2.5
5.1
6.6
6.9
6.9
1
2.6
2.0
0.84
2.2
0.80
0.60
0.12
2
2.6
1.0
0.28
0.55
0.20
0.20
0.03
3
10.1
21.1
29. t
32.1
35.1
35.:
 *The surface area of pipe in contact with extractant water was approximately 1100 ins. In comparing these results with result.
  other tests it is imperative to take into account the ratio of exposed pipe surface area to extractant water volume, which in i
  case was approximately one square inch to 3.5 mL

+ Column 1: Results expressed in terms of ppb/contact time.
  Column 2: Results expressed in terms of ppb/24 hours.
  Column 3: Results expressed in terms of cumulative ug leached during the 21 -day study.

—Below the detection limit.

-------
I
g

-------
Edward A  Boettner, Gwendolyn L  Ball, Zane  Hollingsworth,  and Romulo
  Aquino are with the  School of Public Health,  University of Michigan, Ann
  Arbor, Ml 48109
Nancy S. Ulmer is the  EPA Protect Officer (see below}.
The complete report, entitled "Organic and Organotm Compounds Leached from
  PVC and CPVC Pipe," (Order No  PB 82-108 333; Cost $11 00, subject to
  change) will be available only from
        National Technical Information Service
        5285 Port Royal Road
        Spring field, VA 22161
        Telephone  703-487-4650
The EPA Project Officer can be contacted at
        Health  Effects Research Laboratory
        U S Environmental Protection Agency
        Cincinnati,  OH 45268
                               10
                                                                        •fr U S. GOVERNMENT PRINTING OFFICE  982/559 -092/3377

-------
33
m
-I

3J
2

-0
o
in
O
O
C
                 1 > TJ
                 >
-------