A Simple Method for the Analysis
of Polychlorinated Biphenyls in
Ambient Air
R. J. Siscanaw* and T. M. Spittler
U.S. Environmental Protection Agency, Region I,
60 Westview Street, Lexington, Massachusetts 02173
-------
ABSTRACT
The development and utilisation of a method to measure ambient polyehlori-
nated biphenxja (PCfis) at ng/nJ level is presented. Thia procedure involves trap-
ping the PCBs on a florIail abaorbant at a flow, rate of 15-25 I/mln. Sampling
tine for the analysis vill vary vith the area under study. Trapping efficiency
la greater than 95?. Samples are eluted vith hexane. Concentration steps involve
• Kuderna-D&nich apparatus, 2 ball micro Snyder column, and a nitrogen evapora-
tion to a volume not lesa than 50 ul. Confirmation is done with porchlorination
by antimony pentachloride to decachlorobiphenyl (DCS). Thia procedure van used
on i»n irdoor area, en insinerator that vus handling kr.ovn PCB material, two
capacitor manufacturers, and a landfill site. These results along vith Borne
correlation data with the polyuretliane foam method are included in this report.
INTRODUCTION
Prior to. 1971, PCBs were used as plaaticizers, dielectric fluid in electri-
cal transformers and capacitors, sealants,lubricants, hydraulic fluids, rubber,
vanishes, inks, adhesives, etc. Today, they are still being used in closed elec-
trical applications. Because of their broad extensive use and stable chemical
properties, PCBs have been found at various levels throughout our environment.
The bioTcagoifieation and toxicity of PCBs is well documented. Some induced effects
are hepatonas (1,2), changes in hepatic microsonal onzynes (3,4), reproductive
dysfunction (5), hn.patic porphyria (6), lower etona jjlcbalin level (7), and a
possibility of tunouorigenesls in the liver (8). The current OSHA health standard,
U.S. Code of Federal Regulations, 1974, for an eight-hour time-voighted average is
0.5 -rig/m for chlorobiphenyl (54?) and 1.0 mg/m for chlorophenyl (1*2%).
There are various methods for determining PCBs in ambient air. Basically,
there are 3 modes of collection: first, itnpingers, using a liquid absorbent such
as ethylane glycol (9); second, coated solid material such as glycerine on a glass
fiber filter (id) or OV-17 on ceramic saddles (ll); third, solid absorbents, such
as florisil (12) or polyurethane foam (13). Florisil is inexpensive, accessible,
and can be bakecl at elevated temperatures. There is no background problem with
florisil. I/ess glassware and shorter time of analysis are achieved because no
Eoxhlet apparatus is necessary for cleaning and extraction. Cleaning involves
rinsing the florisil and baking it at 550°C. The PCBs are directly elutod vith
haxane. Routine laboratory pumps are used in the collection. Two important pro-
perties of florisil are the high trapping officienciea and loading capacity for P.CBs.
This is a simple procedure that does not tio up its personnel and gives onbient
level detectability.
-------
METHOD
Reagents and Apparatus
(A) FOB and KB standards—U.S. Environmental Protection Agency, Quality Assurance
Section, Research Triangle Park, N.C.
(B) Specific chlorinated blphenyls—Analabs, Inc.
(C) Antimony pentachloride—J.T. Baker Chemical Co., reagent grade.
(D) Floriail—PR grade, 60-100 mesh, U.S. Environmental Protection Agency,
Quality Assurance Section, Research Triangle Park, N.C.
(E) Solvents—Burdlck & Jackson Laboratory, distilled in glass.
(F) Trap—150 ml, coarss, glass fritted funnel, ASTM 40-60.
(C) Chro'naflex column—Kontes, 7 ram, size 22.
(H) Oas chromatograph—Varian 2100 equJppsa vith a Hi63 electron capture detector.
Chroiiatographie conditions for PCDsi 6 ft glass, 1.5? OV-17/1.95? QF-1 on
100-120 Gas Chrom Q, 190 C, 40 ml/min. Conditions for DOB analysis: 6 ft
glass, 5* OV-210, 100-120 Gas Chrom Q, 210 C, 40 ml/min.
(I) Gas chromatograph-mass spectrometsr—Finnsgan 1015 sA vith a glass Jet
separator coupled to a Digital PDP8. Conditions for DCB analysis: Gas
chro-natographic conditions, similar to (G) except the helium flov, 30 ml/nlnj
mass opectroaetsr, mass scan range of 494-504 m/e, integration time, 1250 msec.
Procedure
Wash glassware vith chromic acid, rinse with tap water, acetone, hexane, and
then bake overnight at 550'C. Place 12 groins of florisil, that has been previously
baked overnight at 550*C, in a 150 ml glass fritted funnel. Rinse twice with 50 nl
of 15? luethylene chloride in liexane and twice with 50 nl of hexane. Dry the traps
vith nitrogen. Controls and blanks are taken at this point. A control is a
raicroliter volume of an aroclor standard in hexane onto the florisil. A black
consists of 3 rinses of 50 nil of hoxane. Set aside the blanks until the collection
is completed. Dry the traps, including the controls, with nitrogen. Wrap the
prepared traps in clean aluminum foil and place then in a dessicator for transport
into the field. Any site suspected of high PCB concentration should have a
back-up trap.
•k
In the field, the sampler seta up the apparatus as outlined in Figure 1 (14).
Perforate the aluminum foil covering the top of the trap. This is to prevent the
wind from disturbing the florisil layer. Collect the sample at a rate of 15-25
l/min. Sampling time is variable depending upon the anticipated PCD concentration.
Our normal sampling time is 4-6 hr. Monitor the flow rote hourly. Aftor collec-
tion, wrap the traps again in clean aluminum foil and place in the dessicator to
b« transported tack to the laboratory.
Extract the samples and controls with 3 rinses of 50 ml of hexane. Concen-
trate with a Kudqrna-Danish apparatus and a 2 ball micro Snyder coluam to 1 ml.
Usually, at this point, one goes directly to a micro florisil cloan-up. Pack a
chromaflex column with 1.6 gra of florisil (baked at 130 C) follow with 1.6 gra
of sodium sulfate. V%sh with 50 ml of hexane and elute the sample with 10 ml of
hexane. Again concentrate with a 2 ball micro Snyder column to 1 ml. The sample
is now ready for injection into the gas chromatograph. If needed, further concen-
trating can be dpne by nitrogen evaporation.
Perchlorination is used as a confirmation tool. Our procedure IB similar to
one recently published by Crist and Moscman.(15). There are some differences In
the two methods. In our procedure, the solvent exchange portion is 1-2 ml of
chloroform, '^erchlorinate a final volume of 0.2 ml at 150*C, overnight. Add
1 ml of 6H hydrochloric acid. Extract the DCB with 7 rinses of 5 nl of hoxane,
directly out of the hydrolysis tube. This number of rinses is used because some
of the samples would Jell in the extraction. Add 2 drops of methanol in the concen-
trating step.
-------
RESULTS AtlD DISCUSSION
D
The degree of retention of PCBs in the florisil trap vas determined by placing
2 pairs of florisil traps in series. Aroclors 1221 and 1242 were npihed onto the
first traps ujsing a miorosyringe. No break-through vas observed after « 2 hr
operation. This vas repsated for 13 hr using approximately 600 ng each of
aroolors 1221 and 1242. No definite PCB peaks were observed. However there
vas a BTiall amount of background, loss thc.u 5? of the PCB peaks. . This nny have
been due to leakage at ths junction between the 2 glass traps. Modification of
this junction was dons by sealing it consecutively with teflon tape, papar adhe-
sive taps, and coating this with liquid plastic. As of the present, the largest
quantity trapped in a single run is 1.6 ug ar.d*0.05 ug of aroclors 1242 and 1254
respectively, collected over a 3.5 hr period at a rate of 27 l/min. No break-
through was observed. This is illustrated in Figure 2. This back-up systea is
used routinely at sites of any suspected high concentrations,
Sines one is usually dealing with ng levels of PCBs, any contamination will
interfer wit^ the pattern and chlorinate during perchlorlnation. There should be
an organic tnap on the nitrogen used for concentrating the sample. Ojr laboratory
.installed flqrisil filled pasteur pipettes on the end of the multi-concentrator.
Besides contamination one should also be concerned with the loss of lower chlori-
nated biphanyls due to volatilization. Vie concentrated a standard solution of aro-
clor 1221 in hexana to near drynaes volumes using some very narrow tapered tubes.
Total volune^ reduced to were approximately 10 ul and 20 ul with the recoveries
•"•-
of 60? end 90?, respectively. As a result, one should carefxilly control the
nitrogen flow rate to prevent any splashing on the walls of the vessal, taper
the bottons of the hydrolysis tubes, and monitor the final volume.
1ft th regard to psrchlorination, a sample blank raust be run to subs tract out
any background DCB (17). We run an extra control to check the parchlorinetion.
VH.th our procedure the average recovery values for these controls are; aroclor
-------
A-
TIME
ht
z-
6 8 (MIN)
1016, 95?t 2, 5, 4' trichlorobiphsnyl, 90/8, and 4, 4' dichlorobiphenyl, 85)6. Our
recoveries for biphenyl, aroclor 1221, and 2 nonoehloroblphenyl ware low, approxi-
mately 50?, 60J5, and 80J8, respectively. A biphenyl standard in chloroform was
psrehlorinatgd directly, tvice, vith the satas low recovery. So it is doubtful
th»t thass recoveries are totally due to volatility. In these biphenyl psrchlori-
nation chromatograms, there uas only one major D3B peak. No sizable broraonona-
chlorobiphenyl peak was found (16). The biphenyl may be involved with another
competing side reaction besides anlamony broraotetrechlorlde (17). This could b»
enhanced by our chronic acid wash leaving a chromium oxide residue. Unfortunately,
this was not investigated. Data on the perchlorination method for confirmation
of rea?. air samples that were done in duplicate ore listed in Table 1. All of
these sav.plqs were aroclors 1016 or 1242. The KB concentration was converted
into a PCS concentration in the table for comparison. Most of the DCD samples
were analyzed on the gas chromatograph. Two downwind samples from an incinerator
had to be analyzed on the gas chromatograph-nass spectrometer. Even with the
micro florieil clean-up, upon pei'chlorination the background level was too high.
A micro-scale alkali treatment was attempted with little success (18). A mass
spectrometer was used as a specific detector for this determination. Figure 3
is mass spectrum of DCS.
Application
Four examples are presented to illustrate the various PCB patterns that may
be found. In most cases, the pattern appears to shift toward the more volatile,
•*v
less chlorinated, components. Also, the difference between 1016 and 1242 is that
1016 has a smaller amount of pentachloroblphenyls and hexachlorobiphenyls than
12/2, see Figures 2 and 4. At low levels these were reported out as 1016/1242.
The first example is a sample that was done inside our laboratory during the
month of November 1976, and in January 1977. Both analysis were done in dupllcst
These samples were taken from the same room. Major differences are the locations
-------
Table 1. Confirmation by Perchlorination (ng/ra3)
Direct Injection
38
58
150
240
20
40
110
226
Ul
529
Perchlorination
28
55
208
202
20
20
83
95
499
650
LJ
-------
the roo-o and the laboratory's heating Swing the nonth of January. November
results voro 23£ and 284 ng/m3 and January values were 4/tl and 529 ng/n3. In
all ths TCD patterns there was a shift toward the more volatile components.
This is illustrated in Figure 4..
>•.
Two electrical manufacturers that are known to be using fCSa were investi-
gated in September 1978. The results are listed in Table 2 and the patterns are
shown in Figure 5- Here the pattern is a close match to aroclor 32/i2.
A landfill that has received a large amount of PCB waste was tested in January
1973, and later in September 1978. These samples are interesting because of the
changes in the patterns. The early Fall seuple shows a shift toward the heavier,
more chlorinated componsnts as compared to the'January r.ar.ple. This would imply
a relative depletion of the low chlorinated biphenyls et the landfill site along
with the temperature Influence on the pattern. These.results are listed in Table
3 and the patterns are in Figure 6. The back-up at the landfill site for the
September sampling shows no break through.
An incinerator handling known PCB waste was analyzed in the Winter of 1977.
The pattern shows a slight shift toward the more volatile components. The values
are listed In Table It and the patterns are shown in Figure 7.
There was some correlation work done with an independent laboratory using
a polyurethane foam method of collection as described by Bidleman and Olney (19).
The major problem here Is the different sampling times. The polyurethane foam
method uses the hi vol air sampler. Its sampling time is only a fraction of the
"*s»
time needed for the florisil method. For the data presented, the sampling times
for the polyurethane and florisil methods are 15 Bin and 3-4 hr respectively.
This data is given in Table 5. The result of 5 ng/m3 on ths polyurethane foam
tiethod was a downwind sample and the corresponding upwind sample was 19 ng/m3.
No florisil sample was taken at the upwind location. This could have been due
to th-2 ohort sampling time.
8
-------
Table 2. Capacitor Manufacturer
PCB CONCENTRATION (ng/m3)
Ssraples
Plant A
Upwind
Downwind
Back-up*
Plant B
Upwind
Downwind
1242 . 1254
41 . ND
301»r 259 9», 9
ND ND
18 ND
743, 824 24, 38
ND—not detected
B
-------
Table 3. Landfill Results
PCB CONCENTRATION (n
Samples
January
On Site
Downwind
September
Upwind
On Site
Downwind
1016/1242
28, 24
18, 12
27
334, 703
18, 21
1254
ND, ND
ND, ND
ND
33, 23
ND, ND
-------
Table ^. Incinerator
Samples
PCB Concentration (ng/m3)
Day One
Upwind
Downwind
Day Two
Upwind
Downwind
38,
150,
20,
110,
58
240
20
95
8
-------
(FIGURE CAPTIONS)
Table 5. Comparison of
Florisil and Fblyurethane Foam Kathods (r.g/ra3)
Florisil
28, 24
18, 12
703, 774
30, 32
Polyurathane Foso
•21
13
490
5
Figure 1. Florisil trap. (A) prepared trap with perforated aluminum foil
(ft) ball and socket joint (C) rotonster (D) pump vlth exhaust
.. bose. . .
Figure 2. Trapping efficiency. (A) top trap (B) aroclor 1242 (C) bottom trap.
Figure 3. Mas»» spectrum of decachlorobiphenyl.
Figure L. Indoor study. (A) laboratory room (B) aroclor 1016.
Figure 5- Capacitor manufacturer. (A) upward (B) downwind (diluted 7 x A,C)
(C) blank.
Figure 6. Landfill site. (A) January test (B) September test (diluted 7 x A,C)
(C") back-up to B.
Figure 7. Incinerator. (A) downwind (diluted 2 x B) (B) upwind.
-------
REFEHEtlCES
(1) Kimbroush, R.D., Under, R.E. (1974) J. Nat. Can. Inst., 53 (2), 547-549
(2) Allen, J.R., Abraha^on, L.J. (1973) Arch, of Environ. Cont. & Toxicol, 1,
269-280
•.
(3) Chen, T.S., DuBois, K.P. (1973) Toxicol. App. Phara., 26, 504-512
U) Hldetoshii'Y., Nooki, 0., Seitaro, S. (197S) Chen. Phann. Bull., 26 (4),
1215-1221
(5) Barsotti, D.A., Karlar, R.J., Allen, U.R. (1976) Rd. Cosmet. .Toxieo.
14, 99-103
(6) Goldstern, J.A., Hickman, P., Jue, D.L. (1974) Toxicol. App. Pharn. 27,
/37-/4B
(7) Thomas, D.T. Hinsdell, R.D. (1978) Toxicol. App. Pharn. 44, 41-51
(8) Kinbrough, R.D., Squire, R.A., Under, E.E., Strandbert, J.D., Nontali,
R.J., Burse, M.W. (1975) J. Nat. Can. Inst. 55 (6), 1453-1456
(9) Manual of Analytical Methods for the Analysis of Pesticide Residue in
Hunan and Environmental Samples (1977) D.S. E.P.A., Health Effects
Research laboratory, Research Triangle Park, N.C.
(1.0) Toshiichi, 0., Takizawa, 1., Kinagawa, K., Sugal, R., Kifure, I. (1974)
Japan Soc. Air Poll. 9 (2), 214
(11) Harvey, G.R., Steinhaver, W.G. (1974) Atnos. Envir., 8, 777-782
(12) Clam, C.S., Chan, H.S., Hoff, G.S., (1975) Anal. Chen. 47 (13), 2319-20
(13) Lewis, R.G., Browi, A.R., Jackson, M.D., (1977) Anal. Chem. 49, 1668-71
(14) Illustration, (1978) Anal. Chem. 50, 544-
(15) Crist, H.L., Moseman, R.F. (1977) J. Assoc. Off. Anal. Chen. 60 (6),
1277-1281
(16) Huokins, J.S., Suanson, J.E., Stalling, D.L. (1974) J- Assoc. Off. Anal.
Chem., 57 (2), 416-417
(17) Trotter, V., Young S., (1979) J. Assoc. Off. Anal. Cham. 58, 466-468
(18) Young, S.J., Burke, J.A. (1972) Bull. Envir. Cont. fc Toxicol., 7 (3),
160-167.
(19) Bidleman, T.F., Olney, C.E. (1974) Bull. Envir. Cont. t Toxicol 11 (5),
442-450.1
------- |