^y vVp \^-y\ v

August 12, 1985

MEMORANDUM

SUBJECT: An Action Level for 2.3.7.8-TCDD in Air

FROM: Billy J. Fairless, Ph.D.Chief, EMCM/ENSV

TO: John C. MicklundDirector, Environmental Services Division

Morris Kay recommended that I review the April 1, 1985, letterfrom Dr. Houk (copy attached) to see if the letter contained informationthat would permit calculation of an action level for 2,3,7,8-TCDD inair. It appears to me that if we acknowledge the assumptions Dr. Houkidentified in the letter and the fact that our calculations are basedon these assumptions, then the necessary data are available. My calcu-lations are as follows:

° An acceptable daily dose of TCDD, as documented in the residentialrisk document by Kimbrough, et. a 1 . , is 0.028 pg/kg/day for a risk ofone case of cancer per million of people exposed.

° An acceptable daily dose of TCDD for a 55.6 kg person would,therefore, be 0.028 pg/kg/day x 55.6 kg = 1.56 pg/day.

0 If we assume the air coming in contact with the skin contains amaximum average of 0.2 x 10~12 grams/cubic meter of air (the approximateanalytical detection limit and our experience to date), that the skincomes in contact with an average of 10 M- of air each day and absorbs1% of the dioxin in this volume of air (see Attachment 3 of the Falk toWagoner letter), then we may calculate that the maximum amount of dioxinabsorbed through the skin would be:

1 x 10 ' M3 of air x 2 x 10-13 g dioxin x 1 x 10-2 percent =day M- air absorbed

2 x 10~1 4 g dioxin absorbed or 0.02 pico grams per dayday

0 If a person breaths/exchanges an average of 12 M3 of air per dayand 100 percent of the TCOD is absorbed In the lungs (see Attachment 3of the March 1, 1985, letter to David Wagoner from Dr. Henry Fa1k. alsoattached), then the average daily concentration over a 70-year lifetimeexposure that would result in one cancer per million exposed would be:

1.56 pg/day - 0.02.pg/day (skin absorption) 2 0.13 pg/M3

——12 MS/day

0 Assuming that any cleanup activity at a given site would notcontinue for more than three years, and that small children could possiblybe exposed 100 percent of that time, then the acceptable daily averageover the three-year time period would be:

0.13 pg x 70 years » 3.03 or approximately 3 pg ,M7 3 years PP

I would recommend that the Region review the logic outlined onthe previous page and, if we are in agreement, that we use a monthly C\J

average of 3.0 pg/M3 as the action level of 2,3,7,8-TCDD in ambient 0air, which would trigger pollution abatement actions when exceeded, o

0This number would not apply to concentrations at the immediate work ^

site, which could be considerably higher without a comparable increasein risk because the workers wear protective clothing. Since I do nothave data on the degree of protection provided by the protection clothing,I have made no attempt to calculate that action level. I would estimate,however, that the level B protective clothing climates at least 99 percentof the potential exposure and, therefore, that the action level would beabove 500 pg/M3 for concentrations in the immediate vicinity of removalactivities.

Attachment

.-•«», .^ >*

? /%? DEPARTMENT OF HEALTH & HUMAN SERVICES

i^ttL ____________

t-.. Public Health Service^pf ". Afletic^or Toxic Substances

' •'";.'. f^nfl 'Disease Registry

- '^//MemorandumS f ' " r - ^ ' '•• /''. f-

Datc .December 12 . 1 9 8 5 ; rc":{ri':^ -^ '" u5

' "t^

From Acting DirectorOffice of Health Assessment, ATSDR

Subject Health Consultation, Dioxin Ambient Air Levela Off-file

To Mr. Ed SkowronskiPublic Health AdvisorEPA Region VII

r

BACKGROTIND ' ^C\J

The Environmental Protection Agency (EPA) Region VII 0office requeated a review and concurrence vith procedure! Qestablishing an action level for 2,3»7»8-tetrachloro-dibenzodioxin (TCDD) in air off-«ite at UPL »ite» vhere 0

remedial cleanup is underway* This document i« based on a ^reviev of written material •ent to us by Dr«~ Fairless and aconference in Atlanta vith him*

DISCUSSION

The letter to Mr* Skovronski from Dr. Fairless and thememorandum from Dr.' Fairless to Mr* Wicklund propose amethod to calculate an action level triggering pollutionabatement actions at dioxin cleanup si tes* The procedureest imates an acceptable daily average exposure to dioxin inairborne soil blown off-site* This procedure does notadequately address our concerns about an action level forpotential off-si te exposure'*'

The proposed method vould use a monthly average of dioxinin ambient air to trigger dust abatement actions* By thismethod, disturbance of high dioxin concentrations couldallow off-si te movement of highly contaminated dust earlyin a month, but no response would be possible until dailysamples were averaged at the end of the month. Averagingthe daily concentrations over a month could allow severaldays of very high off-site levels to be acceptable ifbalanced by a sufficient number of days of low levels*

The risk assessment method by which Centers for DiseaseControl (CDC) established a level of concern for2,3,7,8-TCDD in residential soil is based in part ontoxicological data derived from animal experiments oflifetime exposure risks for cancer. The virtually safe

Page 2 - Mr.' Edward J'; Skowronski

dose for added cancer risk of 1 / 1 » 0 0 0 » 0 0 0 was calculated tobe 28 fg/kg body weight/day for continuous lifetimeexposure;' The proposed method to estimate cancer risk orvirtually safe dose for exposure less than continuouslifetime exposure would assume that all the exposure occursin some shorter period and that exposure would be zero forthe remaining years of life. This method ignores thepotentially large effect of dose rate on estimates of toxicor carcinogenic effect'*' Therefore, we cannot concur withuse of the proposed method for estimating an acceptabledaily exposure based on an excess cancer riak7 Based onour current knowledge of the method of action for2,3,7,8-TCDD, a chemical believed to be a cancer^promoter,for short tern exposures to low levels of this chemicaloff-site in airborne duat,' we have an immediate andquantifiable concern for possible chronic toxic effects.Ultraviolet light exposure causes TCDD to start to break rn

down very rapidly, greatly reducing the potential for long 0term exposure on fugitive dust particles'*' This in no way odiminishes our concern for increased cancer risk resulting Qfrom cumulative exposureT

0

The following section calculates exposures that animal dataindicate may not produce toxic ef fects in humans*"

A review of toxic e f fec ts in animals showed that a noobservable ef fect level (NOEL) for laboratory rodents hasnot been found at concentrations of orally administereddioxin generally above 0'«'01 ug/kg body weight/day*: Somequestionable data suggest that a NOEL may exist in rats andmice at concentrations below 0.001 ug/kg bw/day.

Nonhuman primates are more susceptible than humans toTCDD.'- Oral doses of dioxin at 1 .8 ng/kg/day (1 .8 E-9g/kg/day) to rhesus monkeys for 7 months, or one dose atI'.O ug/kg resulted in significant deleterious healthe f fec ts * Since this concentration is not a NOEL, astandard practice is to divide this number by 1000 toestimate a tolerable total daily dose* The tolerable totaldaily dose by this method is 1.8 E-12 g/kg bw/day.'ERA assumes an average ventilation rate for humans of 0.33m /kg bv/day.' Further assuming that a resident isexposed for 24 hours per day, and that 100 percent ofinhaled dioxin is absorbed from the fugutive dust, thefollowing table lists off-si te levels of dioxin measured.during clean-up at the Quail Run site and calculatespotential exposure to dioxin in dust.'

Page 3 - Mr. Edward J'. Skowronski

ESTIMATED DAILY EXPOSURE TO DIOXIN IN All

VENTILATION RATE DIOXIN LEVEL DAILY DOSE DIOXIM

M3/KG/DAY G/M3 G/M3/DAY

0.330.-330.330.330.33

1-.818478459.4209 ."40 511-.-12

E-12E-12E-12E-12E-12

6.-001.63.113.103.67

E-13E-12E-12E-12E-12

Therefore, to achieve a dose equal to or less than thetolerable total daily dose ( 1 . 8 E-12 g/kg bw/day), theconcentration of dioxin in air should be less than 5'.5 —E-12/M3 (5.-5 pg/M-);' ^

3 °We therefore believe the suggested 3 pg/n of airborne —diozin is a reasonable action level for potential off-siteshort-term exposure. Short term in this eontezt means —)

periods of a few months'.' This calculation considers only °noncancer e f fec ts from short-term ezposure, because of thehigh carcinogenic potential of TCDD, we vould wish to keepezposure to fugitive diozin containing dust emissions tothe lowest level achievable.

0

RECOMMENDATIONS

1'; Continue to aaintain the lowest possible levels of dustemission while performing soil renovalT

2.' Avoid dust-generating operations on days when winds areblowing from the worksite toward nearby occupiedresidences.

3. Residents who live near sites where soil removal isoccurring should remain inside on days when winds areblowing in their direction and are likely to expose themto dust from the worksite.

4. We recommend that at least one air sample be taken inthe nearby residential area on any day chat the locationis downwind of active soil disturbing activities. Thisprovides a record of the most direct measure of exposureto adjacent populations.

5. A s s e s s potential of f -s i te exposure to dioxin on airbornedust using a dual system, including the one currently inoperation to quantify dioxin and an additional one usingreal-time air monitors (RAM) that provide instantaneous

Page 4 - Mr." Edward J. Skowronski

measurement of dust levels* While this second methoddoes not measure actual airborne concentrations ofdioxin, it gives a real-time estimate of efficiency ofwork practices* If the best dust suppressiontechniques are used, then no lower level of airbornediozin could be achieved'. If dioxin levels measuredoff-site exceed the action level, then two optionsexist: (1) To try a new dust suppression technique; or(2) to cease operations;

Dioxin levels in the air obtained from the siteevaluation show that all measured dioxin levels occurredwith airborne,particulate concentrations in excess of140 ng/(ug?)m •• Because EPA and NIOSH guidelines fortotal airborne dust are in the 20 to 300 ug/m and (\Jabove range, fugitive dust levels should be kept well ^below the 100 mg/m range that was frequently reportedin September 1985» at Quail Run. Dust concentrations °below this were often achieved and paniculate °concentrations below this should be maintained;* There 0are insufficient data to assume that elevated dioxin olevels will not occur at dust levels below 100 mg/a »but this is a good working assumption*' Additional datacollected should be used to validate this hypothesis*

6*' Monthly averages from quantitative samplers should notbe used to determine excessively high ambient levels ofdioxin'. Use the shortest time practical to analyzeairborne levels.

7. Dust samples should not be combined, but analyzedseparately";'

We hope this information is useful to you*

^Y^a. - )W?- Stephen Margolis, Ph.D.

KT* 1

Mr. Ed SkowronskiCenter of Disease ControlAtlanta, Georgia 303.13

Dear Mr. Skowronski:

I appreciate you taking the time to get CDC employees to review andcomment on my first draft. I thought it appropriate to address several oftheir comments in a letter to you rather than just the memo for which weare asking concurrence. The comments you provided by phone and my response mare: ^

1. Should we recalculate the risk using an exposure time longer tnantwo years? °

0This is a valid point and I have modified my recommendation accordingly, o

However. I believe 10 years is excessive because we are not consideringpeople who have lived in a contaminated area, but only people livingnear where we are performing a clean-up action. There is no reason tobelieve that any of these people have ever experienced any exposure todioxin other than that which might result from our clean-up activities.(^e do not expect any exposure from this route eitner, but are trying toagree on an action level as a precautionary step.) People who wereliving in the contaminated area have relocated and are not now exposedto ambient air in the vicinity of the site or have elected not to relocateand to accept any risk from their decision.

2. Should we recalculate the risk using a factor to include dioxinthat might be absorbed through tne skin?

Our concern is for people who may be exposed only to polluted air leavinga site (rather than contaminated soil and polluted air as explained above).Any exposure Dy this route would occur either by inhalation or as youcorrectly point out by absorption through the skin. One can estimate theamount of dioxin that might be absorbed through tne skin several differentways. The method that appears to me to be the most reliable and which9ives tne highest estimated concentration (O.U<; pg/day) is included inmy revised ineino.

^J/^Q^/eSJ-'^n .,' (0 • 3 -S^

£~MC.f1

^/a/J/10

3. Is there a significant difference between summer and winterexposure via absorption through the skin?

Io light of our response to Item 2 above, this question is notpertinent.

4. What are the work practices at the site?

Generally a site is divided into sections of approximately 5.000 sq. ft.each. Each section is sampled and determined to be clean or polluted at the951 confidence level. Approximately 4" of soil 1s removed from pollutedsegments, bagged and hauled to secure storage and the resulting surfaceis resampled. The process is repeated until all segments are clean atwnich time restoration occurs. '

One network of air monitors is located in the active working area toinsure worker safety. Another network of air monitors is located at the "site boundary to detect any air pollution leaving the site. There are at tr\

least four monitor sites in the ambient air network. One of these always 0has a duplicate set of monitors. (Currently we are operating six sites/ oseven monitors at Quail Run.) Twenty-four hour composite samples are Qcollected from each monitor daily for the first 14 days of operationafter which samples are taken from at least an upwind and a downwind <-site. We are requesting concurrence on an action level for ambient airleaving a site.

Currently dust is being controlled by use of water sprinklers and t>

by using closed containers when possible.

5. Are tnere other controls that could be employed?

One can think of several additional actions that might help controlthe release of dust/dioxin from an action site. but to my knowledge nonehave been demonstrated to be effective in the field.

6. Should we attempt to differentiate between people living neara site or "passiny through" contaminated air?

If we protect the people living near a site then we will also protectany people just "passing through" an area near a site. It is our objectiveto protect all people living near a site.

Sincerely yours,

Billy J. Fairless. Ph.D.. ChiefEnvironmental Monitoring & Compliance Br.Environmental Services Uivision

•f UCP^RT.MtNT OF Hl.Al.rH S. HUMAN SERVICES

"C~c"-" eo. .'ST. : : .-.

c-wsT/n^• . H<S^P

PuDlw Mr-ilth Str/ice IC.C

....——————£J^Cf"i;<". fcf OiseJ&e Cunt.'i;!a-i...r..GA3n3?3

April 1, 1985

Mr. Morris KayRegional AdministratorEnvironmental Procection Agency Region VII324 Ease lith ScreecKansas City, Missouri 64106

Dear Mr. Kay:

In response to your concerns about Ur. FalK's leccer of Marchrisk assessment model to determine an adequate clean up levelMobile Homes, I would like co clarify some issues and provideapproches coward this project.

1 regarding chein Quail Runsome alternative

inI would first like co ra-tfmphasize chat thtf risk assesamenc model provided^yDr. Falk is applicable only co the posc-clean-up wipe sanpie evaluation o£0mobile hones located ac the Quail Run Mobile Home Parx. It should not be 0vieued as a, recommendation 'for accepcable levels of indoor residential Qtetrachlorodibenzo-p'-dioxin (TCDD) contamination or as applicable to any _other dioxin sice. It is certainly not applicable to, and does noc replace ormodify, previous staceaencs on dioxin levels in residential soil.

We tuvr had additional CiraL* co evaluate this model, since chc March 1 letterand cuntinutf co have concerns regarding the detection level of the wipesamples. We are .concerned that wipe sampling may not be a sensitive methodfor evaluating contamination inside the oooile homes, as Ur. Lybargerindicated in the Kansas City inter-agency meeting on March 14. As a result,we have explored other mechanisms for evaluating the habitability of mobilehomes following clean-up by EPA. A logical and perhaps easier approach toevaluating mobile home habitability would be to determine the total aaount ofdust remaining in the mobile hone (assuming there is no continuing source orresevoir or contamination) rather than evaluating the concencracion of TCUDpresent on the dust. If there is no potential for continuing exposure todioxin after the clean-up, then future health risks would be detennined by tnetotal amount of TCDD in the hone, not the concentration of TCDD in householddusc. We can richer simply estimate the maximum anounc of dust that couldremain vichouc ham co trailer residents, even if an individual were co comeinto conc.icc wich all the remaining dusc- A clean-up strategy concentratingon che cotJl amount of dust remaining in die mobile home requires Che saaeSCringen; proccdurtiS presently in your clean-up plan, piua some aechanisa ofaddressing dire contained in insulation aacerial and the insulation space.

An acceptable daily aos«i of TCDD, as documented in the redidencial riskassesscient document by Kimbrough, et al, is .028 pg/kg/day. Considering anaverage weight 01 55.b kg over d 70 year lifetimii, a cocal of 39,7au pg

''(dppn'-ci-.Jcely -0,000 pg) would be an acceptable lifetime doae correspondinoto an excriss lifetime cancer risk of 1 case per million. The higti«iscconcent ration or TCUD in trailer vacuum cleaner bags was approximately

Page 2 - Mr. Morris Kay

10 ppb. Ac a concentration of 10 ppb, 10,000 pg of TCDO would be present pergram of dust in the mobile home, and therefore, no more than 4 grams of dustcould be safely present in the mobile home. Assuming that all previouslytested mobile homes determined to contain less than 1 ppb of TCDD in vacuum

' cleaner dusc contained an average concentration of .5 ppb, then similarcalculations would predict that no more than 80 gm of dust could safely bepresent for these mobile homes.

We suggest the following clean-up strategy:

A. A recent vacuum cleaner dust sample analysis for TCOD should be performedto assist in determining the extent of cleanup for each mobile home.Mobile homes with dust concentrations less than 1 ppb should be cleaned bya method outlined in Part B (below), and mobile homes with dustconcentrations, either by recent or previous vacuum samples, between 1 and10 ppb should be cleaned by a method outlined in Part C (below)* Mobiltfhomes with concentrations (either by recent or previous vacuum sanples)°igreater than 10 ppb should be individually evaluated to determine if a^r>efforts beyond chose described in Part C should be taken. It is possiblethat if excrecely high levels are discovered in any mobile home, theusefulness of cleaning that home should be questioned*

0

B. For trailers deterained to contain less than 1 ppb of TCDD in vacuum bagdust:1. Clean as per the EPA prepared protocol for the Quail Run Mobile homes.2. Seal particle board floors.3. Seal all openings into the insulation space.4. Test for cleanliness by HEPA vacuuming all interior surfaces to

determine that less than 80 gm of dust remains in the living area*5. Clean Mobile Home exterior.

C. For trailers previously determined to contain between 1 and 10 ppb of TCDUin vacuum bag dust:1. Clean as per the EPA prepared protocol for the Quail Run Mobile Homes.2. Seal particle board floors.3. New (less than 2 years old at the time of cleanup would be a ,

reasonable time period) undamaged trailers should have all openingsinto insulation spaces sealed. Older (greater than 2 years) or newand damaged trailers, should have the insulation replaced and theinsulation space cleaned.

4. Test for cleanliness by HEPA vacuuming all interior surfaces todeceraine that less than 4 gn of dust remains in the living area.

5. Clean Mobile Hone exterior.

This clean-up plan is based upon mathematical modeling, and similar caveatscowards ics applicability to actual situations, as posed in the residentialsoil risk assessment, should apply. Ue believe Chat this approach doescontain two additional safety features. First, it is unlikely that any person

"would fi incacc all the dust remaining in the mobile home, and second, thetesting procedure of vacuuming dust is in itself an additional clean-up seep.

Page 3 - Mr. Morris Kay

We believe that che use of this plan is a satisfactory way of assessing mobilehome cleanliness and safecy for inhabitation, without che use of costly TCDDanalysis although a few TCDD wipe samples per trailer may be useful inassuring the mobile home residents.

We are currently reviewing the results of samples conducted on the trialcleanup mobile home* A prelminary review of these results seems to confirmthat cleaning the mobile homes is a viable option, and we will be providing amore comprehensive review of your report.

We would be happy to-discuss our mobile home cleanup recommendations withyou. We hope to work towards a timely completion of this project in a mannersuch that both CDC and ERA can be confident of protecting the health of themobile home residents.

•' / ^\ i ^~{y^. \n1^ mVernon Houk, M.D. 0Director 0Center for Environmental Health Q

^* •

. . . 0

V E R T A C C H E M I C A L C O R P O R A T I O N S I T E

m

m

m

c

io

3i

ID

(n

nc

n

r>

n

-i•-<

in

en

X)

iTio

o'l

cio

tnn

n-i

x-i

-iT

ino

Jm

oo

—

ci n

rn

uin

r-i'

nn

a)

-< -

o o

--i

-o

rn m

-<

r"c

i>-im

-<

03

)3

)0

3)3

)0

-ix

xu

i^

!?• m

s

min

mo

rm

z

rnu

i 31

m p

i -i

D

m:^

zrn

ci3

3ic

-il

>i>

33

01

m

o

DA

) -^

nm

o?

n-u

mm

0

33

0

c 3

i -i

n -

a •-'

pi o

31

-o

n

" n

T-

~n

+ c

3 o

x m

31 n

•-'

n^

in

-<

31 z

D D

^

-ir

-ir

-o

nx

i o

^

31

3^

" 31 o

7

.031 -i n

G) o

30

m o

— a

o i>

r>

—

x (

n m

r>

s

m «

- •-'

•-'

nm

-i

m

c 3

1 31

•n

7; r r

m

a

lc

na

zD

nm

m

-<

na

ou

n-

ir

rr

3

z-

i3

3

m

»-

"o

7:

t-

mr

n

D

3-\

•z.

-T)

-f -i

G)

<

-i

tri 31

D C D Z

^- *• •. •3.

0 Z -1 x

I> n 7-

r' D y

n

-^>•

•«[.n

»-• •-'

%1

*•

CP»•'

>-* ri.i

•-'*

-'>-•

—••-

l »-"

'-*•-•

•— •

~' •-» o

o c

' o

IT'

1-' 03 o

o (f

f i-^

>-* '-•

i-*

•-»

»— c' o

o o

•

M

C

-10

<

000000 0 0 0 0 0 0 0 0 0 0 0 0

0

0 0000000 000 0 0 0 0

000000000000000000

0

0 0 0 0 0 0 0 0 000 0 0 0 0

-fr-

u c

r> cri

ro

(.n L

I (-n

w o

CD

O.J

aio

*£)

CD c

r« •--

•N|

*• -

si o

[B

(Ti o

m o

cr>

u3

-si

•-'

^ n

-' O.J

4?

ro -

-.i cr'

i

>-'

^ r

'-' ••-

o'

x'

o L

TI o o

o i-

n ^ o

OJ u

ui

o o

o

a?

c..i r>

.' m u

^

CD -

LI

CD tn

^-

ro

o O

J M

-^ ^

ri-i

(T'

LH

-i-

••

••

••

••

'•

••

••

••

• •

• .

..

..

..

. .

..

..

..

. pi

o o

o o

o o

(.n

o e

n (.n o

o o

o

i n;' o

o

on o

o u

-^

o C

D -si

-P-

ro

o o

o

o o

o o

o

0 0

0 0

00

00

00

0 0

0 0

(.1

LH

0 0

(.1 0

0 0

L1

0) •

-• ^ 0

0

0 0

17'

0 0

0 0

0

^^

Y\^

^'^

\^

\<

s.\'s

.Y

Y\^

\

>''

S

Y's

^.^

.'s

-s

.'s

.'^

's

.Y

\

\\^

Y

N.

33

'3

:^

33

33

33

3a

33

-D

-D

C-

0

U

IX

IIIIX

I

XO

X

DU

XX

X

co

oo

oo

oo

oo

oo

oo

rr

ia

sa

o

i?

oo

oo

oo

oo

o

z»

o

rir

ioQ

o

T'.z

/'z

zz

^z

zz

^z

zz

rr

cc

<

cc

cc

cc

cc

c:<

c

<<

crc

- c

-•'-

i-'-

i-i-

i-i-

j-i-

i-i-

i-t^

rrs

a

x\ v

s •J

O •s

-xi

v v

s »

31

3?

:n

31T

rT

xT

-i-

rx

TiT

iiim

m

o

(t> •

-• *

c

i-n i

-->

L"

II'

0 f

-T

t (P

+

P

!c

•

y m

-<

0

7.

0

01

I

-1

^3

)o

a o

mc

-^ -

, 3

31

riU

) -1 -I

1-1

'

0 0

^E

--

-T

31

X

l Z

m i

n (n

•f-

T> •

o

\

a-, -

s:

ui •

-•ni

--.

om

a

+A

n "•

--^ u i;

-i

-.i

r;-*

c.

*•

-t-

-1

U;

0 0

C

C-

' IT -i-i i" D a•-t

c,o

-•

s-

0

•:

-^

^

ou •D

•v D -i m^

M *

c.i'

cr' I-

J -si

U3

i-

p;>

LP

•'J

LH n

;'

o

-si (.ii

(.i

^ c

r' r"

o ^

r..

^ o

0

00

05

9

(T>

• r,

ۥ

-IG

) o

m m

m c

-< -<

03

3

0 ;

0 -0

CO

T» 0

-< D

s?" c

o D

n -( -i

p z

M

u)

G)

iir o

•D

•o

r en

( (D 0 ro ui U}

0:

• 05 0

~>|

*-

03

l»J

0-J

0)

.!>

1 (D

C

'X-4

ua

ifl r

v C».J

ru (P

-si

ro

^ r

np.i

on

-^ u

i on

|T!

in •

o (D

ry

o u

i T;

..

..

..

(0

Cp i

fl (T

I (T>

C- 0

M »

- (>J

-t-

0 0

G)

•T

1tn

oi>

tfi£

:D

i?"

c -

i TI

--'

xi X

i a

D

mm

Tix

-Ti-

ior

D •

m

m 2

3

n ^

xi

m-i

o <

in

o a

3r

D r

< cf!

rn -

i c;

r> r

T|

z m

m a

i/u

ro LI CP u) •si

>-»

•U+-

ro

M r

u ^

mCD

-

(T

H

OS

33

ifl

ID

^J ^

01 3

ro e

n o

m

01 o

o "

'i t-r

•i1' o

• 2

..

..

..

_)

•1 -

Ul

P. 0

-f-

I0

t»J

»0

•1 0

W0

00

0 0

0 0

0 C 0 0

0 0

0

»-•

X(T

'-

r(D

mff

l (T

l»-•

y.

m'

to a

—

0IP

01 n m »-i

H.I

rI-'

"3 M

tfi 0

7-

-

2 P

Ln ^

0)

0:

h'

• *•

o»

r.^

«t.n

>-

i-

(,J (

T> H

.i c

o o

-i

ui o

o c

L1

0 0

10

'-< 0

0 C

(T'

0 0

0 0

C O

C

-L1

0 C

' 0

0 0

0 0

>S

"'S

'•»

'~,

^~

" ^

•»

.

3

:•? .I

:-4 3

3 3

0 0

Cl 0

D L

7 0

Z 7

'7

'? 2 .'' Z

-1

.•-

;-<

_(

^ .-

I

T I

T

-I:

1r

.1

- • ' . ' . ; . .

'.: •- • '

' EPARTMENT OF HEALTH & HUMAN SERVICES •Public Health Service

Centers for Oiseise ControlAtlanta GA 30333

March 1. 1985

Mr. David WagonerDirector, Air and WasCe Managenent DivisionRegion VII, Environmental Protection Agency726 Minnesota AvenueKansas City, Kansas 66101

7Dear Mr. Wagoner, '

This letter is to follow up on the meeting held on February 14, 1985 at theCenter for Environmental Health (CEH), Centers for Disease Control (CDC) in <—Atlanta. In attendance were John Uddle of Clinical Chemistry Division, CEH,< fMark McClanahan of the Superfund Implementation Group, CEH, and Richard QHoffman, Dave Forney, Paul Stehr, and Jeffrey Lybarger, from Chronic Diseases-Division (CDD), CEH, and Bill Keffer and Carl Blongren of your office. Wediscussed in some detail EPA's proposed cleanup of mobile homes from Quail Run-Mobile Hone Park in Gray Summit, Missouri. This letter contains a descriptionand elaboration of CDC's understanding of the six major areas of discussion.

For the sake of credibility vis a_ vis the former residents of Quail Run, wewere requested to supply an observer during the cleanup process in order toverify that the work was done properly and completely. We do not believe thatwe should maintain a representative onsice. Risk management decisions andmeasures lie within the special expertise of EPA; therefore, EPA should beresponsible for the quality of their work and should be prepared to documenttheir activities. ; - .- , . .: •. • - :, • • ; - , • • " . " . . • • ' - . • - •••: • • . .. • -

' ' . » » . _ . . • - -

We are in basic agreement with the cleanup protocol which has been suggested(see Attachment 1). However, we suggest the inclusion of a fewadditions/revisions. Based on a visual inspection by CDC and MissouriDivision of Health officials on February 25, 1985, there is evidence ofextensive accumulation of dirt in the insulation material in the walls and thefloor (which could serve as a significant reservoir of contaminateddirt/dust). One possible option would be to remove all insulation, thoroughlyclean wall interiors, and put into place new insulation. Also, since theparticle board floors are permeable, the floor boards could be sealed (e.g.,with a waterproof polyurethane coating) after both the interior and exteriorsurfaces have been thoroughly cleaned.

It was suggested during our discussions that an experiment be conducted with acontaminated trailer to test the efficacy of the proposed cleanup procedures.CDC agrees that such a trial should be conducted, using--essentially theprocedures which have been suggested by EPA, as amended. This trial shouldrnclude pro-cleanup sampling of dust and wipe samples (co confirm and documentthe location and levels of contamination as a reference for interpretation ofpost-cleanup sampling results), extensive movement and activity of the

trailer, which Is likely to shake extraneous material loose (if present),followed by post-cleanup sampling. It is our recommendation that wipe samplesand dust samples be collected using sampling methods described in the attachedERA collection protocol (see Attachment 2) before cleanup, immediately aftercleanup before taking the trailer offsite, and after movement stress andrunning the ventilation system for at least 48 hours from the followinglocations:

a) I vertical surface sample on a wall in every room, one meter fromthe floor, in the center of the wall,b) 1 horizontal surface sample in each room from either a window sillor the center of the floor before new carpeting is put in place. Twoadditional samples should be taken in the kitchen from the foodpreparation area (i.e., counter cop) and in a cabinet,c) 1 sample from the central furnace ventilation duct,d) sample of insulation materials beneath a window in each of the 4exterior walls (if insulation is not replaced and wall interiors arenot cleaned), . - . . . • , . .,and .. :- • . ' . . . '. .• . =. ... '•-•• ::-•.-•' . "• -.'• . :• •CM

•e) I sample from the furnace air filter.'" "-'" '

This sampling protocol will require the collection of a total of 54 samples(i.e., 18 in each of the three phases), assuming an average size of 5 rooms 0per trailer. 0

0Following completion of a successful trial (i.e., no contamination found above Qthe criteria limit described below in point 5), then each trailer should besanpled after cleaning according to the following sampling scheme:

a) 1 vertical surface sample on a wall in every room, one meter fromthe floor, in the center of the wall,b) 1 horizontal surface sample from a window sill or the center ofthe floor before new carpeting is put in place (plus one additional

.sample in the kitchen from the food preparation area), ; . .,.• - " c) 1 sample from the central furnace ventilation duct, . •

and . ^ .......„.• -' ..^••. • s . - . . " . - . . ... .-- •• " d) 1 sample from the furnace"air filter.'/ /:.'''-•-'•"'.'" '•."••-' '.'.'..•'•• , ' '^ '"; 'This protocol would require the collection'of at least "13 samples for each''' • " ' ' • 'trailer.

Since we suspect that there is a potential for sequestered dust, we recommendthat all trailers should be cleaned following the complete protocol—includingthose which were previously found to be contaminated at levels below 1 part ' ^ .":* -per billion (ppb). . / ' • .' . - . '* '*• • • -; '.- ;''-"'

There are no relevant precedents for interpreting the exposure risk innonoccupational settings for surface contaminations. Thus, the interpretationof wipe sampling results requires making extensive assumptions regardingexposure Via denaal contact for which we have no supportive data. Althoughthese results cannot easily be converted to a measure of concentration (e.g.,ppb) which can be entered into our existing risk assessment models, we haveattenpted to do so utilizing our best estimates. Attachment 3 details the

assumptions and conclusions of this special risk assessment which was preparedby professional staff from the Investigations Section of the Special StudiesBranch, COD, CEH. Based on these results, and the fact chat EPA has reportedthat their best-obtainable detection limit for these samples will be 4picograms per centimeter^ (cm2)—or 0.2 ppb, the most reasonable approachis to attempt to clean to a nondetectable level on dust-free surfaces* Afterthe cleanup procedures are completed, settled dust on surfaces should becollected, as indicated in the above-recommended sampling protocol. These datahave been compared to our special risk assessment conclusions and should beused as a basis for justifying further remedial actions, if necessary. Whilewe anticipate that these results will show nondetectable levels of TCDD, theacceptance of this fact oust be tempered by the knowledge chat-there is likelyto be so little mass of sample material that the accuracy of the laboratoryanalyses will have wide confidence intervals. Even in these more familiarterms of weight per weight concentration measures, it is not clear whether themodel represents a reasonable and scientifically supportable risk assessment

. since we have few or no data on which to base required assumptions regardingkey input parameters (such as degradation of the contamination levels overtine, amounts of dust with which an individual may come in contact during ^various periods of their life, and absorption rates through the skin and ' t

gastrointestional tract from dust). In light of these facts, we suggest that oif any one post-cleaning sample is found to have a detectable level of TCDD Q(i.e., above 4 pg/cffl2 or 0.2 ppb), then EPA should completely repeat the _cleaning of the entire trailer (following the attached protocol) in aniterative process until this criterion of no detectable amounts in any samples-is met.

We were requested to provide to residents of the mobile homes a certificationthat the trailers are safe for human habitation following the cleanupactivities. CDC is willing to provide a letter which includes the following

.. items: • > • • . s • . , : . . . . . . . • -...-.- ». .. . ...;-. - • .1) a listing of all cleanup activities reportedly done by EPA,

• . - • 2) a listing of all post-cleanup sampling results, .and - .,..,-.- - ... ..- . -. 3) a statement indicating that the cleanup performed by EPA was reportedly

extensive and that the sampling results (presumably all of which werebelow detection limits) would lead us to expect that there is very little,if any, risk of significant future exposures to ICDD.

^ However, it must be made clear that we cannot certify that the trailers are..' absolutely without contamination or that there is no (i.e., zero) health risk' to the inhabitants. Because of limitations in analytical techniques, as well

as our current knowledge of the long- and short-term toxicity of dioxin, themost responsible statement that we can make is that some nonquantifiable riskof adverse health effects nay remain—although this will likely be so small asto be negligible.

As in our previous involvement with the interpretation of environmental datain advisories for public health protection in this situation of dioxincontaminations in Missouri, the above recomnendations are to be considered

open to further discussion in light of newly acquired information and/or newly.proposed remedial actions. We suggest that the principal parties in thissituation ( i . e . , ERA, CDC, Missouri Division of Health, and Department ofNatural Resources, as well as the affected citizens at the Quail Run MobileHome Park) be involved in these future deliberations.

Sincerely yours,

Henry Talk, M.D.ChiefSpecial Studies BranchChronic Diseases DivisionCenter for..Environmental Health

. ' • ' ' • .' • Attachment 1 ' • ' . ' - •

EPA REGION 7PROCEDURE FOR CLEANING TRAILERS AT QUAIL RUN

^ February 20, 1985

I. Interior Preliminary

Remove entire contents - refrigerator, stove, furniture, clothing, dishes,radios, TVs, toys, drapes, water heater, mirrors, light fixtures, electricalpanel covers, vent covers, etc.

A. First stage cleaning for interior!

I* With carpet in place, vacuum the ceiling, walls, windows,sills, cabinets, and floors using the Minuteman HEPA Vacuum(see enclosed insert).

p2. Remove carpet and padding starting at one end of trailer

working toward the other end. Vacuum the floor as the carpetis being removed to avoid grinding dirt into the exposed ^loor,

B. Second stage cleaning of interior; ^0

1. Scrub the ceilings, walls, windows, sills, cabinets, and Ooorsusing a Trisodium Phosphate solution in water with the oappropriate brush to remove all particulates. See insert,-<ortypes of brushes to be used.

2. Allow interior to dry.

C. Third stage cleaning of interior*

1. Dry vacuum entire area once again to remove any loosenedparticles from washing.

II. Cleaning of Trailer Duetwork

with trailer elevated on one end, a steam cleaner nozzle tip will be insertedon the high end. Ouctwork to be steamed until clean. Also an appropriatebrush for the ductwork will be used to insure proper cleaning, (see insert)

A. Cleaning of Contents:

1. All porous materials will be destroyed and containerized fordisposal. Examples are sofas, fabric chairs, mattresses, etc.

2. Steam clean all metal parts of appliances such as refrigerator,stove, water heater, etc. Areas not readily accesible orsensitive to water to be blown out with compressed air gun.These include coils o-f refrigerator, TV components, small

«, electrical components, etc.

3. Any wooden or -formica tables considered salvageable shall becleaned with hot soapy water solution, wooden tables to be

( wiped with lemon oil after drying.

4. Vinyl chairs to be cleaned with hot soapy water and allowed todry.

5. Toys, dishes, silverware* etc. to be cleaned with hot soapywater.

6. Clothing, bedding, etc. shall be laundered twice or dry cleanedas required.

7. Contents to be stored in an uncontaminated area a-fter cleaning.

,t

III. Cleaning Exterior of Trailer

A. Remove insulation and'cloth barrier -from underside of trailer ^andcontainerise for disposal.

1. Clean windows and frames using hot soapy water and appropriatebrushes. 0

02. Seal windows with visqueen plastic in preparation to steaa

(

clean exterior.

3. Steam clean roo-f, sides and undercarriage o-f trailer usingbrushes as needed to remove soil.

B. After initial cleaning, trailer is to be allowed to dry then movedto relatively clean area and steam cleaned once again as it leavesthe site paying special attention to tires and undercarriage.

C. Clean trailers to be stored in an uncontaminated area.

Attachment 2

DEMONSTRATION PROJECT

Sample Walls - 4 SamplesClean Interior - Then seal entries.Clean ExteriorSample interior and exterior locations and Numbers per Centers -ForDisease Control*s request.o Area/Sampleo - Pado Solvents or H2024 Hour Analysis.Data returned and reviewed by the Centers -for Disease ControlIf no reason to fail trailer then take to gate, pressure wash exterior.Drive X miles over roads.Return to Fenton.Hook-up power and vent'i 1 at.i on as per Centers -for Disease Controlrequest. r-~Re-sample per Centers for Disease Control « -Centers for Disease Control Review data ^

000

I - ^•--•:—-^-.-?-- •••'•.-.--•-.

I - -.'" - '. /.-:. - ' ••• ' - ; ' - . ' . .

Attachment 3

SPECIAL RISK ASSESSMENT FOR CLEANUP CRITERIAAT QUAIL RUN MOBILE HOME PARK, MISSOURI

ASSUMPTIONS

Useful lifetime of a trailer is 20 years.

Children comprise the highest risk group; therefore, this risk assessmentis calculated for exposures during the first 20 years of life.

As an individual ages, they are likely to spend an increasingly smallerproportion of their time in the home. Therefore, we have assumed the

. .following: .. '.-. ' ' - r - ' . . . . . ;. ..';.;..• •.„ ": . . . • oo• •" -'• • - • ' . . - • - • -• ; * •• •.'': ••• Proportion of .Time' • ''•• '.:. •. : . . . '• ' .'";

Age Spent in Borne0-2 years old 1002 ' °2-6 years old 802 06-15 years old 672 0> 15 years old ' 402. . . Q

0

The indoor environmental half-life of TCDD is 2 years because of dilution( . of existing indoor contamination levels, breakdown due to ultraviolet

light and volatilization,-and regular cleanings.

There is no effect of seasonal exposures because exposures will take place' ' • in an indoor environment. . :" • .

Routes of uptake are through:• : ' '•"-". V .."/Inhalation—dust concentration of 0*5 ng/m3, air, average^of 12 n^ ...,,.;>\ ";•.. ,.'J':- ,- ',-air exchanged per day, 100X absorption of TCDD in alveoli;-": .-.•• • „ - • ".

Ingestion—uptake of 2 grams dust/dirt per day in children less than' "'" . ' or equal to 5 years old and 100 milligrams in children greater than 5

years old, 507. absorption of TCDD in gastrointestinal tract; andAbsorption through skin—contact with of 2 grams dust/dirt per day.

Y .'• -: ,. .in children less than or equal to 5 years old and 100 milligrams in . - .•t, :. •;.--..' .'..'children greater than 5 years old, 12 absorption through skin. . " ' . •• . •

One gram of dust will cover an area of 50 cn2, so that a factor of 0.05can be used to convert measurements quantitated in terms of picograns (pg)per cm2 to an equivalent measurement in terns of parts per billion (ppb).

EXAMPLE4 pg/cm2 X 0.05 » 0.2 ppb

or20 pg/cm2 X 0.05 - 1 ppb

CONCLUSION

Using these Input parameters, we ran the computerized simulation model* Theresults indicate that even at the suggested analytical detection limits (i.e.,4 pg-/cm^ or 0*2 ppb), the excess lifetime risk of .developing cancer to anexposed individual ranges between 0*40 and 20.21 per million. Even when thecritcal assumed amounts of indoor dust/dirt ingested and deposited on skin arehalved, the range of excess risks of developing cancer is from 0*20 to 10.16per million. Therefore, given the analytical technique limitations, wesuggest that a level of 4 pg/cn2 (or 0.2 ppb) be used as a level-of concernin interpreting the post-cleanup sampling results.

?

r

' ' ' ' • • •' •' •• '•r -'' •: '. • '

• • . Ch

^s-0

0

. - .. 01 0

.. A ••' • '-• - • ,. •-.- . * •>• -

;;• -.•,-.-.. -. ^'.-:-:::.^. .•..;-.. - „-•'.. •-. i , • • : : , . . , : _ ..,,, . . . .

'-r. - *.••<••.« • - " .,- . - • - . - - - . , , •

.A. .:;.- ..,.-^^. ^ .•!.. • ^ y.,'?- : • • • - - . '•--•:•'< ".• : '•-•'••

•^.- •\y'_ •'[[• ;•-.•• { •.•r •. . • • • • ' • •

•-•" :." "" •" " • • • ;"":...• • ' " • • -. •-'•• .'. '?.

ut.r^ni.Ml.NT OF HIALTH & HUMAN SERVICES Public Health Service

Ccnwfk for Diu>ai>> ControlA:!.i,-.i.,GA3033;

April 18, 1&H5

Mr. Morris KayRegional AdministratorEnvironmental Protection Agency Region VII324 Ease llth StreetKansas City, Missouri 64106 <

Dear Mr. Kay:0

In responsti co your request for final criteria for dioxin cleanup of Che 0mobile hones ac the Quail Run Mobile Home Park in Gray Summit, Missouri, WB_suggest che attached criteria. We will also recommend co the MissouriDivision of Health (MDH) Chat they enclose the attached criteria in their c>letters co the owners of the cleaned mobile homes. ^

0Any deviation from the criteria should be reported to and evaluated by theMissouri Division of Health before the issuance of the State's letter to thetrailer ovner.

If you have additional concerns, please contact me.

Sincerely yours,

^-J- dW<-Henry Falk, M.D.Chief, Special Studies BranchChronic Diseases DivisionCenter for Environmental Health

...... n^iArot.R VICES : ; .' Public Health Service

Centers for Oisea&e ControlAilsnu GA 30333

Criteria fur the Clean-up of Mobile HomesQuail Run Mobile Hone Park

Gray Summit, MissouriApril 22, 1985

A. A recent vacuun cleaner dust sample analysis for TCDD should be performedto assist in determining the extenc of cleanup for each mobile home.Mobile hoaes wich dust concentrations less than 1 ppb should be cleaned bya method outlined in Pare B (below), and mobile homes with dustconcencracions.T either by recent or previous vacuum samples, between 1 and10 ppb should be cleaned by a method outlined in.Part C (below). Mobilehomes with concentrations (either by recent or previous vacuum samples)greater than 10 ppb should be individually evaluated to determine if anyefforts beyond chose described in Part C should be taken. It is possiblethat if extremely high levels are discovered in any mobile home, theusefuinesti of cleaning that home should be questioned. • -r-

LnB. For trailers determined to contain less than 1 ppb of TCDD in vacuum ^bag

dust: ' . . . 0

1. Clean as per the EPA prepared protocol for the Quail Run Mobile Ho'nes.2. Seal particle board floors. 03. Seal all openings into the insulation space.A. Test for cleanlineys by HEPA vaosmaing all interior surfaces to

determine that less than 80 gm of dust remains in the living area.5. Clean Mobile Hone exterior*

C. For trailers previously determined to contain between 1 and 10 ppb of TCDDin vacuum bag dust: . . - - . -; • -• .,

1. Clean as per the EPA prepared protocol for the Quail Run Mobile Homes.2. Seal.particle board floors.3. N'cw (less than 2 years old at the time of cleanup would be a reasonable

cine period) undamaged trailers should have all openings into.insulation spaces sealed. Older (greater than 2 years) or new andamaged trailers, should have the insulation replaced and the insulationspace cleaned.

4. Test for cleanliness by HEPA vacuuming all interior surfaces todetermine chat less than 4 gm of dusc remains in the living area.

5. Clean Mobile tionse exterior.

Reprinted from ENVIRONMENTAL SCIENCE A TECHNOLOGY, Vol. 21, Page 550, June 1987Copyright © 1987 by the American Chemical Society and reprinted by pcrmis»ion of the copyright owner.

Procedures Used To Measure the Amount of2,3,7,8-Tetrachlorodibenzo-p-dioxin in the Ambient Air near a Superfund SiteCleanup Operation

Billy J. Falriess,* Dale I. Bates, Jody Hudson, Robert D. Kleopfer, Thomas T. Holloway. and Debra A. Morey

U.S. Environmental Protection Agency, Region VII, Kansas City, Kansas 66115

Tony Babb

Air Quality Services, IT Corporation, Knoxvilte, Tennessee 37914

• Sampling procedures and analytical procedures aredescribed that were successfully used to monitor for2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD) in airsamples collected near a Superfund site cleanup operation.Measured concentrations of 2,3,7,8-TCDD in air samplesare related both to an action level (3.0 pg per standardcubic meter of air) and to a calculated no observed effectlevel (5.5 pg/m3). The study concluded that it is possibleto collect reliable data for 2,3,7,8-TCDD in air at concen-trations that are below the action level specified by theCenters for Disease Control. Data quality was definedrelative to the quality control procedures described in thestudy. There was no apparent relationship between par-ticulate matter in the air and 2,3,7,8-TCDD in the air.

IntroductionFor some animal species, 2,3,7,8-tetrachlorodibenzo-p-

dioxin (2,3,7,8-TCDD) is one of the most toxic syntheticcompounds known (1). The LD50 has been reported tobe 0.6 ^g/kg of body weight for the guinea pig, 115 g/kgfor the rabbit, and 22 ^g/kg for the rat (1). However,2,3,7,8-TCDD does not appear to be nearly as toxic tohumans as it is to other animal species (2). Some of thechlorinated dioxin isomers are, for all practical purposes,not toxic at all. The LD50 for 2,7-dichlorodibenzo-p-dioxinis approximately 2.0 g/kg of body weight for the rat (1).These facts require that any analytical procedure used tomeasure for the concentration of 2,3,7,8-TCDD must beisomer specific.

In addition to being toxic, 2,3,7,8-TCDD is also a sus-pected carcinogen. Recently, R. Kimbrough and otherssuggested the extremely low number of 2.8 X 10'14 g/kgof body weight per day as being the dose responsible fora cancer risk of one in one million (3). Previous studieshave detected concentrations of labeled dioxin in the airin the range of 10"15 g/m3 (7). Several studies have shownthe presence of dioxins in emissions from incinerators atsimilarly low concentrations. Attempts to measure dioxinconcentrations in ambient air at these levels require verysensitive analytical procedures.

In the early 1970s, 2,3,7,8-TCDD was formed as a by-product in the production of hexachlorophene and/orAgent Orange in a small facility in southwest Missouri. Itis known that, unlike some other sources of dioxin, thisprocess resulted in a very high fraction of the 2,3,7,8-TCDDisomer being formed. It is now believed that the 2,3,7,8-TCDD was produced by condensation of 2,4,5-trichloro-phenol. The phenol resulted from alkaline treatment of1,2,4,5-tetrachlorobenzene as shown in eq 1. Waste ma-

N»OH2CsH30Cl3 — CizHACL + 2HC1 (1)

terials containing the dioxin byproduct were mixed withused oil and subsequently applied to roads and other

surfaces for dust control (4).One of the places where the waste material from

southwest Missouri was applied was a trailer park near StLouis, MO. The site covers an area of approximately 11acres. It has an irregular shape as shown by the site map(Figure 1). The surface and subsurface soils at the sitewere tested and were found to contain 2,3,7,8-TCDD atconcentrations above 1.0 ^g/kg of soil (1.0 ppb) at mostlocations within the site. Therefore, a mitigation plan wasprepared to control exposure to the contaminated soil (5).The plan called for the removal of the contaminated sodfrom the surface and for storage of the removed matetifliin a safe location on-site until detoxification proceduresare available.

The EPA has a responsibility under the ComprehensiveEnvironmental Response, Compensation and Liability gt(CERCLA or Superfund) to clean up hazardous waste safesknown to be contaminated with toxic or hazardous ma-terials. A second responsibility is that agency cleanupactivities must not constitute a source of pollution formembers of the general public who are in the immediatevicinity of the cleanup operation. At the beginning of thiswork, it was not apparent that it would be possible withthe available technology to meet both of these responsi-bilities for the site in question. This paper describes theconsiderations that went into the design and operation ofthe network and our findings during the first 5 months ofoperation. Some of the specific questions we had to ad-dress and hoped to answer are listed below:

Can the concentration of 2,3,7,8-TCDD in ambient airbe measured at levels that cause an insignificant risk tothe public?

Can a cost-effective network that measures the averageamount of dioxin in the air be implemented over time, witha high degree of confidence? If not, are other optionsavailable that will provide acceptable answers?

If dioxin is found in the air, will it be absorbed to par-ticulate matter that is trapped on a filter or will it be inthe vapor phase or both?

Relative to local weather conditions and the normalscatter in any measurement, how many sampling sitesshould the network contain? How many samples shouldwe collect at each sampling site each day? How longshould the network be operated?

What are the best available procedures to collect asample? What are the best analytical procedures? Whatreliability can be expected from these procedures?

Is there any relationship between the amount of par-ticulate matter in the air and the amount of 2,3,7,8-TCDDin the air?

Is there a measurable cause-effect relationship betweencleanup operations and measured concentrations of2,3,7,8-TCDD in the air? Is any of the 2,3,7,8-TCDD inthe air coming from background sources?

'550 Environ. Sci. Technol.. Vol. 21. No. 6, 1987 0013-936X/87/0921-0550S01.50/0 © 1987 American Chemical Society

CONTAMINATEDAREA

SAMPLELOCATION

Figure 1. Diagram of the site. During pollution abatement activities.approximately 6 in. of sol would be removed from a section (5000square feet) with a backhoe and placed In plastic bags. The newsurface wodd then be tested for dioxin and either be declared deanor another 6 In. of sol would be removed. Air poliition coutj originatefrom either the sol handing operations or truck traffic wtthtn the site.The six monitoring locations relative to the contaminated area areshown In the figire (M. 0, D, K. L, and E). The prevailing wind Is frommonitor E to monitor 0.

If significant concentrations of 2,3,7,8-TCDD are foundin the air, can models be used to predict the path of anyplume that might be caused by the cleanup operations?

Are any of the pollution abatement actions more ef-fective in controlling the release of dioxin than others?

Experimental ProceduresDesign of the Monitoring Network. When designing

the monitoring network, we assumed that we needed ameasurement detection limit in the range of 0.1-1.0 p/m3

in order to obtain reliable measurements at the 5 p/m3

level The response from the Centers for Disease Control(CDC) contained their recommendation of 5.5 p/m3 as anestimated no observed effect level (NOEL) and theirconcurrence with our recommendation of 3.0 p/m3 as a"warning" or action level (6). Realizing that the total costof sample collection and analyses would probably exceed$1000 per sample, our objective was to collect the minimumnumber of samples that would have a good (95% confi-dence level) probability of showing an exceedance of theaction level if one occurred and of showing that no ex-ceedance occurred if indeed one did not occur. On thebasis of prior experience, we estimated that we couldmonitor for the concentration of 2,3,7,8-TCDD in ambientair at these levels with an accuracy of 25% relativestandard deviation on the basis of the recovery of fieldspikes. We assumed that the upwind and downwind sitedata would each have a normal distribution. The numberof measured concentrations (samples) needed at eachsampling location was then calculated as a function of thedifference between the action level (true mean) and themean of a given number of measured values.

The statistical calculation is outlined as follows. Theaverage concentration of all samples collected at a site isused as an approximation to the true average concentration(which would be obtained from continuous monitoring ofthe air over a long period of time). The uncertainty of thatapproximation is indicated by the 95% confidence limitsas calculated by

95% confidence limits = ? ± .g^-^s/Vra)

where V = the sample average, n = the number of samplesat each sampling location, s = the standard deviation ofsamples collected at that sampling location, and to.97s.n-i

^

in0

Table I. Estimation of Number of Samples Needed To MeetStudy Objective

no. ofsamplesper site

?142856

((Whirl

2.4472.1602.0521.96

95^< % ) •

12%

11153

> confidence li• (ample aver

RSDof24%

221496

miteage*

36%

3321149

s the tabulated Student's t value for n - 1 degrees offreedom. The width of the 95% confidence interval fora site depends on the number of samples collected and thescatter among the data from those samples, as is shownin Table I.

A comparison of the experimental data with the actionlevel will result in a conclusion that no exceedance occurredif the 95% confidence interval is completely below theaction level The data will demonstrate that an exceedancedid occur if the 95% confidence interval is completelyabove the action level. The data will demonstrate thatmore samples would be needed in order to show conclu-sively whether or not the true average concentration ex-ceeded the action level if the action level is within the 95%confidence level

We elected to initially collect 14 samples at each mon-itoring site. With the 14 data values and the assumptionsdescribed above, a definitive conclusion would then be 0reached if the data were 14% either above or below the Qaction level

In order to obtain sufficient data to be able to assess the '--effects ofon-site activities on the off-site ambient air undervariable wind conditions, the monitoring network had tobe designed to provide long-term monitoring of the air ator near the property boundaries of the site. On the basisof the physical configuration of the site (see Figure 1),which is long relative to its width, we decided that aminimum of six fixed monitoring locations would beneeded to ensure consistency throughout the study andto have one upwind and one downwind sample for mostdays. Our major objective, while deciding on the appro-priate number of monitoring sites in the network, was to -TninimiTp the probability that contaminated air would passbetween the monitors without being detected. The threelargest variables affecting this probability were the size ofthe area within the site that might be releasing dioxin, thelocal variation in the wind direction during a samplecollection time period, and the duration of the study.Considering that dioxin might be released from approxi-mately one-fifth of the site at any given time from trucktraffic and construction work, that the wind direction wasvery seldom in one direction for 80% of a 24-h period, andthat the study would require approximately 200 workingdays, we concluded that six monitoring sites would besufficient. An on-site meteorological monitoring stationwas incorporated into the network design for the purposeof obtaining adequate wind speed and direction data.

A site visit was made prior to finalizing the networkdesign to obtain detailed information on the topographyand to choose the specific sampling locations relative toanticipated activities at the site. The specific locations forthe air samplers were selected so that they would be nearbut just inside the perimeter fence for security purposes,would be consistent with the accepted siting guidance forcriteria pollutant monitoring, would provide permanentplacement throughout the life of the project (i.e., thesamplers would not have to be relocated during the course

Environ. Sci. Technol., vol. 21, No. 6, 1987 551

Figure 2. Modified Ht-Vol sampler (General Metal Works. Inc.. ModelPS-1) used to collect 2.3,7,8-TCDD samples.

of the excavation activities at the site), and would provideadequate coverage for most wind directions (6). Thespecific sampling locations that were selected are shownin Figure 1.

Collection of Representative Samples. To collectrepresentative samples, we used commercially availablemodified high-volume air samplers that employ both afilter for collecting particulate matter and a solid adsorbentfor collecting vapors. A diagram of the air sampler(General Metal Works, Inc., Model PS-1) is shown inFigure 2. In operation, a known volume of air (calculatedfrom the flow rate and time of sampling) is drawn througha dual-chambered sampling module (see Figure 3) andexhausted to the air via a 10-ft exhaust duct. The upperportion of the sampling module holds a 4 in. diameter glassfiber filter, which collects the particulate matter, and thelower portion consists of a cylindrical glass cartridge (65

. Lowtt CMItu

mm x 125 mm) containing a 3 in. long solid adsorbent,which entraps selected vapor-phase compounds. Poly-ether-type polyurethane foam (PUF) plugs were utilizedas the solid adsorbent material (7).

To measure the very low concentrations of dioxin re-quired in this project, a large sample volume of air wasrequired. Since the air samplers are only capable of pro-viding a flow rate of approximately 0.280 m'/min, it wasdetermined that the samples should be collected on a 24-hbasis (±15%). This time and flow will give a sampled airvolume of 300-400 m3 of air.

The samplers were placed on 1 m high platforms toobtain samples of the ambient air in the breathing zone.The samplers were operated daily except on Sunday. Ifexcavation occurred on Sunday, the samplers were oper-ated on that day as well.

We assumed the wind direction could not be predictedfor the 24-h sampling period, nor would it be constant overthat period of time. Therefore, all of the samplers wereoperated to collect samples each day. During the first 14days of sampling, all of the samples were analyzed for bothparticulate matter and 2,'3,7,8-TCDD to obtain base-linedata for the site. Pollution abatement activities were oc-curring at the site during this time period. Subsequentto this initial sampling period, only one upwind and^Shedownwind sample were submitted for 2,3,7,8-TCT)Danalysis each day. The selection was based on tha-pre-vailing wind direction and the amount of particulatematter collected on each filter for the sampling peSbd.

Comparability. For a risk assessment, we estimatedthe mnTimimi amount of dioxin a person just off-site wouldexperience during the time of the cleanup activitiesr-^Vedecided to use a daily averaging process because an averageis more comparable to an action level that is based onchronic effects than is a single value (8). We decided toaverage the data values from each monitor separately(rather than average data from different monitors) becausethat would be more representative of exposure for someoneliving near that monitor. Since we wanted to be able totake pollution abatement actions as quickly as possibleafter the data were available, we decided to use a 14-dayrunning average (average concentration of the most recent14 days), which would be calculated daily. In the interestof safety, we elected to use the detection limit as a mea-sured value when calculating the running averages for allsamples that did not contain a measurable concentration

Gutet

Figure 3. Detailed illustration of the sampling head module (see Figure 2) showing the fiberglass filter used to collect parUculates and the glasscylinder/PUF used to collect vapors.

552 Environ. Sci. Technol., Vol. 21. No. 6. 1987

of 2,3,7,8-TCDD. Finally, since none of the analyses ofnondownwind samples showed any measurable 2,3,7,8-TCDD, we decided to use an average of these numbers forthose days when data were not available. A daily datapoint might be missed for a specific monitor if that monitorwas neither upwind nor downwind of the site or if no workwas occurring at the site due to bad weather or for anonworking Sunday.

General Procedures. Particulate matter (PM) and2,3,7,8-TCDD sampling utilized samplers as previouslydescribed. Polyurethane foam (PUT) cartridges wereprecleaned by the laboratory and shipped to the on-sitesample coordinator in sealed, glass sample jars. Filterswere obtained from the PUP sampler manufacturer in lotsof 100. Five percent of the filters were run as blanks toensure acceptable detection limits. Sample modules werecollected and new samples started each morning prior tothe start of any remedial activity on the site. Surgicalgloves and Teflon-tipped forceps were used to remove theglass fiber filter and the PUF cartridge from the samplemodule. The glass fiber filter was placed on a calibratedbalance and weighed to the nearest 0.1 mg. The glass fiberfilter was then folded in half twice (sample side inward)and placed in the glass cartridge on top of the PUF plug.The sample cartridge was then wrapped in aluminum foil(to shield it from sunlight) and placed in the sample jarfrom which it came.

The concentration of the paniculate matter (PM) foreach sample was calculated immediately and used in theselection of samples for 2,3,7,8-TCDD analyses.

Analytical Procedures. All samples were analyzed byan EPA contractor in accordance with a Region VIIstandard method entitled Determination of 2^,7JB-TCDDin Air Samples Using Gas Chromatography-Mass Spec-trometry (9).

The samples were spiked with internal (["Ci;]-2,3,7,8-TCDD) and surrogate (["CIJ-^J^-TCDD)standards of isotopically labeled 2,3,7,8-TCDD. The sam-ples (filter, PUF, and glass cylinder) were then extractedwith methylene chloride in a Soxhiet apparatus. The ex-tracts were cleaned with silica gel, modified silica gel,alumina, and carbon prior to the analyses by high-resolu-tion gas chromatography (GC) and low-resolution massspectrometry (MS).

The gas chromatography column was a 30 m x 0.32 rnmLd. fused silica capillary DB-5 with a 0.25-Mm film thick-ness. Calibration was done by tabulating the peak heightsor peak areas from triplicate injections of 2,3,7,8-TCDDvs. the internal standard. Quantification is based on theresponse of native TCDD relative to the isotopically la-beled TCDD internal standard. Performance is assessedon the basis of extensive quality assurance requirements.These include a requirement for accuracy of surrogateanalyses on each sample.

Quality Assurance. To minimize sample handlingand/or contamination in the filed, two sampling modulesequipped with quick-release connectors were acquired foreach air sampler. The availability of two complete sam-pling modules, which were numbered for ready identifi-cation, made it possible to simply exchange a dean sam-pling module for the module containing the sample in thefield. The modules were transported to and from thesampling sites individually wrapped in plastic bags andstored in a closed container (an ice chest customized tohold the modules upright). This practice enabled allsample handling (disassembly of module and placementof the filter, PUF, and glass cartridge in a sample con-tainer) and sampling module preparation (cleaning and

assembly) to occur in a controlled environment.A five-point calibration curve (equation) for each sam-

pler was obtained initially by using a magnehelic gauge anda calibrated orifice. The equation for each sampler wasthen used to calculate sample volumes. The calibrationwas repeated monthly, and a one-point check was per-formed every other week. A flow audit on 25% of the airsamplers in use was performed monthly by an individualother than the normal operator and with a different cal-ibrated orifice. The maximum acceptable difference be-tween the reported (sampler) and actual (calibrated) flowswas established as ±7%.

Laboratory GC/MS instrument calibration consisted ofan initial three-point calibration conducted in triplicate.The mean, standard deviation, and percent relativestandard deviation (RSD) of the relative response factor(RRF) for 2,3,7,8-TCDD were calculated at each of thethree concentration levels, as well as overall. Acceptablecalibration required an RSD of 10% or less at each indi-vidual level, as well as overall Every 8 h, the calibrationwas to be verified through the analysis of a low-levelstandard solution. The percent difference of the RRF forthe continuing calibration from that of the overall mean Lf^RRF of the initial calibration could not be more than 10%. [T\

Both system and performance audits were included in _the air monitoring plan to ensure that the establishedprocedures were actually being followed. The audit process 0provided the means for continually evaluating the quality Qof the data being generated, identifying apparent problemsquickly, and making in-process changes to correct apparent —problems.

The use of quality control samples was included as aroutine means of tracking the precision and accuracy ofthe data generated and of detecting problems relating tothe quality of the data reported. Throughout the samplingeffort, one field blank, one blank to be spiked by thelaboratory, and one field-spiked sample were submittedfor analysis with every 17 actual samples. In addition,during the initial 14-day sampling period, a second airsampler was collocated with one of the perimeter samplersto collect duplicate field samples.

Results, Discussion, and ConclusionsData Quality. We have grouped the data quality in-

formation into qualitative information and quantitativedata (70). On the basis of the results of our field andlaboratory audits, we concluded that the procedures de-scribed above were being followed as written and that, withthe exception of two data points, all data were acceptablerelative to the qualitative variables.

Two concentration levels were utilized in the perform-ance evaluation (PE) samples. A high-level PE of 11.6 ngwas used to monitor bias above the exposure limit. Thethree data points generated at this level gave a mean valueof 10.99 ng or 95% recovery. Control limits at the 95%confidence interval were 9.65-13.6 ng. Twenty-six medi-um-level performance audit samples containing 5.8 ng wereanalyzed. Mean recovery was 5.29 ng or 91% recovery. A95 % confidence interval of 4.35-6.09 ng was determined.

The field blanks consisted of a filter, PUF, and glasscylinder to which only surrogate and internal standardsolutions were added. The 19 field blank analyses resultedin no detectable quantities of 2,3,7,8-TCDD.

The laboratory fortified 19 samples with 5 ng of2,3,7,8-TCDD. The mean recovery of these was 4.75 ngor 95% recovery.

Nine audits of the sampler flows were performed duringthe studv and gave a standard deviation of flow differenceof 1.72%.

Fnuwnn C^i T^^fiAl \ff\l fn ki e

Table II. Breakthrough Study Results

sampling time (h)/air volume(m') ± 5%, %

0/0 12/180 24/360 48/720 72/1080

Table III. Distribution ofTCDD and TCDF between theFilter and PUF

percent found on the filterand PUF at sampling times of

81.2 93.7 88.92A7.8-TCDD MPR*RSD

1,2,3,4-TCDD MPRRSD

2.3.7,8-TCDF MPRRSD

89.6 64.79.1 2^5

92.8 70.85.47 5.92

92.4 74.46.20 6.86

1.91 5.71 3.9093.6 100 92.97.91 2.59 11.2

99.1 112 99.45.43 9.20 8.95

'MPR • mean percent recovery, RSD • relative standard devi-ation.

Considering the variances in the measurements of boththe volume of ail sampled and the amount of diozin foundin the sample, we estimate that the data are accurate towithin ±12% of the reported values at the 95% confidencelevel.

Is 2,3,7,8-TCDD in the Vapor Phase or on Particu-late Material? Two experiments were conducted to tryto determine what fraction of the 2,3,7,8-TCDD and2,3,7,8-TCDP (2,3,7,8-tetrachloro-p-dibenzofuran) wouldpass through the filter and what fraction would remain onthe filter under the sampling procedures described above.We assume that the materials were in the vapor phasewhen they passed through the filter.

The first experiment was designed to evaluate the po-tential for analyte breakthrough of 2,3,7,8-TCDD,1,2,3,4-TCDD, and 2,3,7,8-TCDF. The experiment con-sisted of spiking clean filters with 10 ng each of 2,3,7,8-TCDD and 2,3,7,8-TCDF and 13.8 ng of 1,2,3,4-TCDD in200 JiL of hexane. Ambient air was then drawn throughthe samplers for varying lengths of time with identical flowrates. At least three samples were collected for each timeperiod. The PUF and filter were analyzed together asdescribed above. The results of the experiment are shownin Table II.

From these data we conclude that the sampling proce-dures are effectively collecting all of the 2,3,7,8-TCDD inthe sampled air and that our field spikes can be used foran accurate estimate of method bias. These conclusionsare supported by work recently reported by DeRoos et aL(7).

In the second experiment, a solution from a standardcontaining both 2,3,7,8-TCDD and 2,3,7,8-TCDF (tetra-chloro-p-dibenzofuran) was placed on the glass fiber filter.Uncontaminated air was then passed through the sampleras described above for 17.5 min, 2 h and 24 min, and 24h. The filters and PUFs were then analyzed separately.The average total (filter and PUF) percent recovery fordioxin was 112% with a standard deviation of 1.3%. Theaverage percent recovery of furan was 90% with a standarddeviation of 11%. The resulting data are shown in TableIII. It was obvious from this experiment that the dioxinwas very slowly migrating from the filter to the PUF. Weconclude that analyses of only the paniculate matter oronly the vapor phase in the sample would both give erro-neous results. It also appears that the furans are moreeasily transferred from the filter to the PUF than are thedioxins.

Effectiveness of Sampling Methodology. The datadescribed above provide a basis for concluding that thesampling equipment and procedures were adequate for thisstudy. Clearly, the sampler must capture both particulateand vapor-phase materials if the data are to be repre-sentative of the air sampled. The volume of air sampledwas just large enough to give adequate data precision (themaximum RSD at the monitor showing highest concen-

1440mm

17.5min

144min

TCDDfilterPUF

TCDFfilterPUF

1000

6337

974

1090

8218

0100

C.C • O.S021

^0

LH

0s.a 3.1 < • s.a r.t -.» —»

2A7.S-TCOO (PIcogr—— pw CuUe M*tf> —

ng— 4. Data from a single monitor where both the 2,3,7,8-TCDOand PM concentrations were above the detection ftntt. Ther«S>noretobto correlation.

trations of 2,3,7,8-TCDD was approximately 22%) in the3-5 pg/m3 concentration range. The samplers were reliable(no downtime), and they maintained the initial flowcharacteristics very well (maximum change in measuredflow was approximately 2%). The ability to very easilychange the sampling modules made field work much moreconvenient and probably improved data precision consid-erably.

Are There Any Relationships between WeatherConditions and 2,3,7,8-TCDD in the Air? At no timeduring this study did we observe a measurable concen-tration of 2,3,7,8-TCDD at a nondownwind monitor. Ap-proximately 200 samples were collected and analyzed for2,3,7,8-TCDD. Ten of those samples contained concen-trations of 2,3,7,8-TCDD above the detection limit Eachof the positive concentrations occurred on days whencleanup activities were occurring in the immediate vicinityof the monitor and when the cleanup activities were up-wind of the monitor for most of the day. From these datawe conclude that there is a measurable relationship be-tween wind direction and concentrations of 2,3,7,8-TCDDin the air. We also conclude that the 2,3,7,8-TCDD wefound in the air was originating at the location of thecleanup work and that the contamination remains in theair only for a short time period. Any background of2,3,7,8-TCDD is below the method detection limit

We were unable to find a model that would, in ourhands, accurately give the experimental results we ob-tained.

Can Particulate Matter Concentrations Be Used ToPredict 2,3,7,8-TCDD Concentrations? Figure 4 is agraph showing the relationship between particulate matterconcentrations and 2,3,7,8-TCDD concentrations. Al-though there may be a very general relationship, it is ap-parent that one could not reliably predict 2,3,7,8-TCDDfrom the measured particulate matter concentrations. Itis also obvious from the data in the graph that there is no

554 Environ. Sci. Teehnol.. Vol. 21, No. 6. 1987

TOO

<w castWtB crrKT icvtiEE

§&

•?S•t;£'M&i

s!51

3.00

gSQz

^ -< » 0 • ^

^ ..vDA1ES

noun 5. Data points are representative (we only show every fifth day(or clarity) averages of measured or estimated (see text) aocenlialioni(or the most recent 14 days. A downwind sample Is one I which themonitor (0 In Figure 1) was downwind of the cleanup activity at thestte during the test day of the 14-day averaging period. A nonsampftigvalue is shown for those days when the concentrations at site 0 wereestimated rather than measured on the last day of the averaging period.The average of all measured (or estimated) concentrations for 14consecutive days is plotted on the y axis against the last day of the14-day averaging period on the x axis tor monitor 0 (see Figure 1).Detection limits were taken as measured concentrations. II a mea-surement was not taken tor a given day because the monitor was notdownwind or upwind on that day, an estimate of the concentration wasmade by averaging all of the nondownwind values to date. Sinceconcentrations above the detection imtt (usually in the range 0.4-0.8p/m3) were obtained only when a monitor was downwind of the cite,al estimated values were averages of detection limits. The data arepresented In this manner to ikJStrate our best estimate of the maximumexposure of any off-site population that might have resulted from thiscleanup activity.

particulate matter concentration that could be used as acontrol limit for 2,3,7,8-TCDD. The particulate matterconcentrations referred to were obtained by the proceduresdescribed in this paper. They are not equivalent to thetotal suspended particulate (TSP) values usually obtainedwith a Hi-Vol sampler. However, a good relationship be-tween TSP by the EPA-approved procedure and themeasured PM concentrations has been demonstrated forthis study.

Since dioxin analyses are so expensive relative to TSPanalyses, it may be beneficial in future studies to look fora 2,3,7,8-TCDD/TSP relationship under different condi-tions.

Was the General Public Exposed to SignificantConcentrations of 2,3,7,8-TCDD as a Result of ThisAction? Figure 5 is a graph showing the 14-day runningaverage and the associated 95% confidence levels of thoseaverages for the monitoring site with the highest concen-trations. As can be seen, all of the 14-day averages werewell below the warning level and the NOEL at all times.We would emphasize that the averages were calculatedwith the assumption that the detection limit was a mea-sured concentration for all samples that did not have ameasurable concentration of 2,3,7,8-TCDD. Therefore, theactual exposure was probably even less than the maximumpossible indicated by the graph. From this data, we con-clude that concentrations of 2,3,7,8-TCDD that cause aninsignificant risk to the public can be measured in ambientair using the procedures described above. We also con-clude that during this study the public was not exposedto a significant concentration (5.5 pg/m3 for a "fewmonths") of 2,3,7,8-TCDD at any time.

Can Monitoring Costs Be Reduced in FutureStudies? We estimate the total cost of the study de-scribed above was approximately $295000 or approxi-mately $1500 per sample. After reviewing the data fromthis study, we see no way to substantially reduce the costswithout additional data. It does not appear to be possibleto use particulate matter as an indicator parameter fordioxin, thus reducing the analytical costs. We would notrecommend reducing the size of the monitoring networkfor a site of this size. We believe the network should bein operation any time pollution abatement activities areoccurring at the site. We doubt that efforts to improvedata quality would be cost-effective. We are investigatingsample breakthrough for longer sampling times both forPUF and for other solid absorbents. An option that mightresult in a reduced cost would be to reduce the time spentto obtain base-line data from 14 days to a shorter timeperiod. This kind of decision is properly a role of man-agement since costs must be balanced against the increasedrisk of obtaining a false positive and of unknowingly ex-posing the public to a significant amount of 2,3,7,8-TCDD.The results of this study should make that decision easierto reach.

r""-Acknowledgments

We express our appreciation to Morris Kay and JohnWicklund for providing resources and encouragement and u

to the many EPA and EPA contractor employees who have 0participated in this effort by doing their assigned work in .—.an efficient and professional manner.

Registry No. 2,3,7,8-TCDD, 1746-01-6. °

Literature Cited(1) Esposito, M. P.; Drake, H. M.; Smith, J. A.; Owens, T. W.

Dioxins: Sources, Exposure, Transport and Control, In-dustrial Environmental Research Laboratory, Office ofResearch and Development, U.S. EPA: Cincinnati, OH,June 1980; VoL I, pp 147-199; EPA-600/2-80-156.

(2) Tachiriey, F. H. Sci. Am. 1986, 254, 29-35.(3) Kimbrough, R., as referenced in April 1,1985, letter from

Dr. Veroon Houk, Center for Environmental Health,Centers for Disease Control, to Morris Kay, Regional Ad-ministrator, EPA, Region VII.

(4) Kleopfer, R. D. Chemosphere 1985,14,739.(5) Emergency Planning and Response Branch, U.S. EPA,

Region VII Quail Run Mitigation Plan; Emergency Plan-ning and Response Branch, US. EPA, Region VIL KansasCity, KS, 1983.

(6) Code of Federal Regulations 40, Part 50.11, Appendix B.(7) DeRoos, F. L.; Tabor, J. E.; Miller, S. E.; Watson, S. C.;

Hatchel, J. A. Evaluation of an EPA High-Volume AirSampler for Polychlorinated Dibemo-p-dioxins and Po-lychlorinated Dibemofurans; Environmental MonitoringSystems Laboratory, Methods Development and AnalysesDivision, U.S. Environmental Protection Agency: ResearchTriangle Park, NC, 1985; Contract No. 68-02-4127.

(8) Margolis, S. "Health Consultation, Dioxin Ambient AirLevels Off-Site'; memorandum to Ed Skowronski, PublicHealth Advisor, EPA, Region VII, 1985.

(9) U.S. EPA, Region VII Determination of 2^,7,8-TCDD inAir Samples Using Gas Chromatography-Mass Spec-trometry, U.S. EPA, Region VII; Kansas City, KS, Aug.1985.

(10) Bates, D.; Hollowa", T. T. Operating and Quality Assur-ances Procedures Manual; Environmental Monitoring andCompliance Branch, U.S. EPA, Region VII: Kansas City,KS, Nov. 11, 1982.

Received for review June 9, 7986. Revised manuscript receivedDecember 9, 1986. Accepted February 16, 1987.

Environ. Sci. Technol., Vol. 21, No. 6, 1987 555

Chemosphere, Vol.18, Nos.1-6, pp 141-148. 1989Printed in Great Britain

0045-6535/89 $3.00 + .00Perqainon Press pic

EVALUATION OF METHOD PERFORMANCE FOR MEASURIN3

2,3,7,8-TETRACHLOBODIBENZO-P-DIOXIN IN AMBIENT MR

Jody L. Hudson* and Debra A. Morey

U.S. Environmental Protection Agency, Region VII

25 Funston Road, Kansas City, Kansas 66115

Abstract

00High volume sampling in combination with GC/MS analytical techniques have been used to

measure 2,3,7,8-Tetrachlorodibena>-p-dioxin (2,3,7,8-TCDD) in ambient air at numerous sites

within Region VII, U.S. EPA. Extensive use of these procedures have resulted in a large

database of information which has been used to evaluate and document the method's performance.0

Introduction ^

Ambient air conitoring for 2,3,7,8-TCDD has been conducted during contaminated soil

removal operations at several dioxin sites located within the state of Missouri. The mea-

surement method used consisted of high volume sampling techniques to simultaneously collect

particulate and vapor phase 2,3,7,8-TCDD on a Glass Fiber Filter (GFF) and Polyurethane Foam

(PUF) adsorbent, respectively, with sample recovery achieved using Soxhiet extraction and

analysis completed using either capillary gas chroniatography/mass spectrometry (GC/MS) or

gas chromatography with tandem mass spectrometry (GC/MS/MS). Although similar methods have

been used by others to measure dioxins, PCB's, and pesticides in air (1-5), little

information was available regarding the method's performance when it was initially selected

to measure 2,3,7,8-TCDD at sites undergoing removal operations. Extensive use of the method

within itegion VII has resulted in compiling a large database from which the method's perfor-

mance has been evaluated. In addition to data from the collection and analysis of over

1000 air samples, the informational database contains the results of specific quality control

procedures that were incorporated into the monitoring program designed to monitor and evaluate

the method's performance. These QC procedures have contributed data for over 130 QC samples

and 80 sampler flowrate audits. These data, coupled with experience gained from performing

ambient air monitoring operations for 2,3,7,8-TCDD at four removal sites over a three year

141

142

period, have provided the opportunity to rigorously evaluate and document the method's

performance under a variety of field conditions.

Method Description

Sample Collection: Air samples were collected using Model PS-1 PUF samplers manufac-

tured by General Metal Mbrks as shown in Figure 1. The samplers utilize a dual-chambered

nodule designed to hold a 4 inch diameter filter in the upper chamber and a 2.5 by 5 inch

cylindrical glass cartridge in the lower chamber as shown in Figure 2. Both Glass Fiber

Filters (GFF) and Quartz Fiber Filters (QFF) were used successfully to collect particulatephase 2,3,7,8-TCDD. Preference «es given to the GFF's, however, due to the observed ability

to achieve approximately 10% higher flowrates than were obtained when using the QFF's. The

PUF's, which are contained in the glass cartridges, are the polyether type having a density

of 0.02 g/cn>3. Both the GFF's and PUF's were obtained from the manufacturer of the Model

PS-1 PUF samplers. Although others run the PUF through an intensive clean-up procedure

prior to sample collection (1-3), it was found to be unnecessary when monitoring for the

single 2,3,7,8-TCDD isoroer at concentrations of interest (i.e., 3.0 pg/m3 action level).

Since the GFF/PUF media was demonstrated by repeated analyses to be sufficiently clean from

interfering background contamination, they were used as is from the vendor.0

Although identical general criteria were used in designing the ambient air monitorifi?

network for all sites, each was tailored to the specific conditions unique to each site.O

Briefly, the monitoring network at each dioxin site was comprised of multiple fixed siteO

sai-pling stations strategically positioned to provide effective fenceline monitoring for

all wind directions with each monitoring station consisting of a sampling unit mounted atop

an elevated platform. Details of general and specific criteria used to design these networks

have been previosly described by Fairless, et al (6).

Figure 1. Schematic of Model PS-1 PUF Sampler (General Metal Works, Inc.)

143

Figure 2. Schematic of Model PS-1 Sampler Module

r. Lawn CmctM

.. ,,»««, r .'S"

With the exception of holidays and some weekends, the air monitoring networks wereoperated continuously while removal operations were in progress. All monitoring stations _comprising a network were operated on a constant, identical and concurrent schedule.Samples were collected while operating the sampling units at their maximum flowrates whichvaried with electrical power fluctuations, ambient temperature, and varying flow restrictioncaused by batch to batch variability in the GFF's. Sample collection flowrates were measureafrom calibration curves generated for each sampler and found to range from 0.20 m3/min. to0.30 m^/min. with a mean of approximately 0.27 m3/min.

f • 'The first air monitoring operation conducted utilized sampling durations of 24 hours.

In an effort to reduce monitoring costs without sacrificing the effectiveness of the moni-toring program, the use of extended sampling durations which would result in fewer samplesto be collected and analyzed were investigated. A study was conducted to determine theeffect of increased sampling duration and air sample volume on analyte breakthrough. Sampleswere spiked by adding approximately 10 ng each of 2,3,7,8-TCDD, 1,2,3,4-TCDD, and2,3,7,8-tetrachlorodibenzofuran to the GFF's. Ambient air was then sampled using the modelPS-1 PUF samplers over durations of 0, 12, 24, 48, and 72 hours at a flowrate of approxi-mately 0.25 in3/min. Samples were prepared In triplicate for each represented duration andblank samples here collected concurrently to measure the amount of any native analytescontributed by the ambient air.

Itie collected GFF/PUF's were analyzed and the percent recovery data for each durationwere compared. As the data show in Figure 3, the mean percent recoveries were similar withthe exception of the 12 hour duration samples. Although the reason for the lower recoveryof the 12 hour samples is uncertain, the study results demonstrated that sampling durations

of at least 72 hours at a flowrate of 0.25 mVnin. can be used without experiencing break-

through of collected 2,3,7,8-TCDD. The data also indicate that matrix effects due to in-creased particulate loading associated with increased sampling durations had no significantimpact on analyte recoveries.

144»-.,.*••/» .-i"

Figure 3. Breakthrough Study Results

Legendo a.37.«-TCao^ 1.2.3 < - TCDD

0 i.i7«-TC»r

VO

0

0

0

Since the results o£ the breakthrough study showed that sampling durations in excess

of 24 hours could be used successfully, sampling durations for subsequent ambient air moni-toring operations were extended to 48 hours, and eventually to 72 hours. This measure not

only resulted in a significant reduction in monitoring costs, but also provided the benefit

of lower method detection limits due to the increased sampled air volumes.r-A

Samples were collected at the conclusion of the sampling duration by detaching the mod-

ules from the samplers using a quick release mechanism. A second set of modules containing

clean CTT/PUF media were attached and the samplers were immediately restarted for a subseguent

sampling period. Collected modules were transferred to an onsite sample processing laboratory

where the sample media were unloaded and packaged for transport to the ofEsite analytical

laboratory for extraction and analysis. Ihe monitoring network continued to operate in this

manner for the duration of the contaminated soil removal project.

Sample Analysis: Air samples were analyzed according to the analytical methods contained

in the Statement of Mark for EPA contract WA87-J060 (7). Briefly, these procedures were asfollows. All glassware was initially cleaned with aqueous detergent, and then rinsed with tapwater, deionized water, acetone, methylene chloride, and benzene. The entire glass cartridgecontaining the sample which consisted of both GFF and PUF plug were transferred to a Soxhiet

extraction apparatus and spiked with internal standard 13^ 2,3,7,8-TCDD and surrogate^Cl, 2,3,7,8-TCDD compounds. The spiking solution was added in a manner that maximized

145

surface area coverage. The samples were extracted with either benzene or methylene chloride

usin9 approximately 250 ml of the selected solvent for 16 cycles. The extract was concen-

trated to about two to three inL and the solvent exchanged to either hexane or isooctane and

then subjected to cleanup procedures if interferences were thought to be present. IM modi-

fied cleanup variations were available (7). The cleanup procedures were modifications of

those used for routine dioxin work, and had the advantage of being able to provide faster

sample throughput. Most air samples analyzed routinely required modified Option A cleanup.

Modified Option D cleanup was typically not required, but was available for extremely diffi-cult samples, Modified Option A cleanup varied slightly when used in conjunction with conven-

tional low resolution GC/MS as opposed to GC/MS/MS. The main difference being that the

GC/HS/MS version was slightly more abbreviated as the GC/MS/MS can tolerate more interferences

and still meet required QC and detection limit requirements. Both versions eitployed two

columns. The first consisted of a layer of silica gel and anhydrous sodium sulfate (for low

resolution GC/MS, a layer of sulfuric acid impregnated silica gel was also added). The

second column consisted of activated alumina topped with anhydrous sodium sulfate. The

modified Option D cleanup was the same for both analytical instruments. It consisted of a

column containing a mixture of Carbopack C and Celite-545. Immediately prior to analysis,C\j

the extract (either directly from extraction or following cleanup), was concentrated to a \0

final volume of about 25 uL. Analysis was done by either conventional low resolution GC/M£>

or GC/MS/MS. The GC column used was a 15 to 30 meter DB-5 fused silica capillary column. Q>

Other columns specific to 2,3,7,8-TCDD can be used; however, this column was chosen for i1(S^

ability to quickly analyze 2,3,7,8-TCDD when it has been demonstrated to be the only isoma^

present. While this column cannot separate 2,3,7,8-TCDD from its three closest eluting

isomers, 1,2,3,4-TCDD, 1,2,3,7-TCDD, and 1,2,3,8-TCDD, for our purposes it was adequate asthe sites under evaluation were all contaminated from a single source which was documented

as only containing the 2,3,7,8-TCDD isomer. We therefore assumed that any positive dioxin

response in the area of 2,3,7,8-TCDD was the 2,3,7,8-TCDD isomer.

Instruments used for this work were initially calibrated in triplicate at three levels

using a standard mixture containing 2,3,7,8-TCDD, the internal standard, and surrogate spike.

The percent relative standard deviation (%RSD) of the relative response factor (RRF) for

2,3,7,8-TCDD was evaluated at each level and was not allowed to exceed 10%. In addition

the overall %RSD of all levels combined could not exceed 101. A single point calibration

verification check was run each 12 hours that samples were analyzed. The percent difference

(ID) of the RBF for 2,3,7,8-TCDD could not exceed 104 from the initial calibration to thedaily calibration verification or else all analyses had to be discontinued and the instru-

ment recalibrated. Surrogate recovery limits were 60-140%. The average surrogate recovery

observed for all samples analyzed was 119%.

Data for each sample analyzed went through an analytical validation process which in-cluded an evaluation of the instrument calibration, internal standard response, surrogate

recovery, qualitative identifications, quantitation, method blank results, field blank re-sults, matrix spike recoveries, and performance evaluation sample results (7). samples were

146

identified as containing 2,3,7,8-TCDD if the following criteria were met: 1) a peak waspresent at mass 320 and 322 (or daughter ions 257 and 259 for GC/MS/MS) within +/- 3 secondsof the internal standard, 2) response at mass 320 and 322 (or 257 and 259 for GC/MS/MS) wasat least 2.5 times the noise level, and 3) the ion ratio for m/z 320/322 was between .69-.90(for GC/MS/MS the ion ratio must be within +/-104 of the mean of the 2,3,7,8-TCDD ion ratioscalculated from the initial calibration). Although not inherent for identification, theinternal standard response for each sample had to be at least 10 times the background noiselevel and also meet ion ratio criteria (for GC/MS/MS no ion ratio criteria is applied as onlyone ion is monitored for the internal standard).

Performance Summary Results

The air monitoring operation incorporated specific QC procedures designed to assessthe method's performance in terms of detection limit, precision, and accuracy, or bias. mSince the method's performance parameters were dependant on both the collection and analyt-ical measurement processes, the QC procedures were designed to evaluate each of these meas-urement components.

0

The sample collection, or field component of the measurement process involved the meaSurenent of the air volume sampled. The precision and accuracy of the air volume measurementswere evaluated by performing flowrate audits on each Madel PS-1 sampler every two weeks whilethe monitoring projects were in progress. Flowrates were audited using a calibrated orificereference standard. Accuracy was expressed as a mean percentage and calculated as the ratioof the indicated flowrate to the audited flowrate. Precision of the field component was alsoexpressed as a percent calculated as the relative standard deviation (RSD) of the pooled accu-racy measurements. Accuracy and precision summaries based on 86 flowrate audits are presentedin Table 1. As the data show, the field component of the measurement process contained asmall positive bias. Since for a given measured mass of 2,3,7,8-TCDD, the concentrationdecreases with increasing air sample volume, the positive bias in the air sample volumemeasurement resulted in a negative bias in the concentration measurement.

The analytical component of the measurement process involved the quantitation of the massof 2,3,7,8-TCDD present in the GFF/PUF air samples. Specific QC procedures to evaluate detec-tion level, precision, and accuracy, included the preparation and analysis of field blanks andmulti-concentration performance evaluation (PE) samples. Of the 83 GFF/PUF field blanks ana-lyzed, none contained detectable quantities of 2,3,7,8-TCDD. A total of 54 PE samples dosedwith a mass of 2,3,7,8-TCDD ranging from 0.295 ng to 11.6 ng were prepared in the on-sitesample processing laboratory by injecting standard solutions containing 2,3,7,8-TCDD intothe PUF. PE samples were submitted blind to the analytical laboratory for analysis. The

lowest mass used that met the criteria for detection as previously described was 0.70 ngwhich was detected in only two of four cases. The minimum mass used that oonsistantly met•the criteria for detection was 1.12 ng. Since no PE samples were analyzed with a mass between

0.70 ng and 1.12 ng, the true analytical detection level is assumed to be at some point be-tween these two values.

The analytical component of accuracy was expressed as mean percent recovery computed asa ratio of the reported mass to the known mass. Prior to computing precision, the percentrecovery data were grouped according to mass and an ANOVA statistical procedure was used totest for differences which may have been due to concentration dependancy. Although lowerpercent recoveries and larger %RSD's were observed with the lowest mass range evaluated, theresults of the ANOVA failed to show significant differences at the 0.95 level of confidence.Consequently, a single expression of precision for all masses evaluated was determined bypooling percent recovery data for all PE samples, and computing the »BSD. The performanceparameters for the analytical component of the measurement process are summarized in Table 1.As the data show, the analytical component contained a bias of -9.7% for the mass range evalu-ated. This bias contributed directly to a negative bias in the concentration measurements.

In addition to being evaluated separately, the parameters of method performance for thefield and analytical components were combined into single expressions so that the total methodperformance could be evaluated. As the data show in Table 1, the total method accuracy anderf-precision were 88.8% and 11.5%, respectively, with a total measurement bias of -11.2%. \T)

0———————————————————————————————————————————————————————————————————Q^

0Table I. Method Performance Summary -

Component Detection Level Accuracy3 Precision0 Bias

FieldAnalytical

Total

H/A0

1 ng0.85 pg/n^

101.7%

90.3%

88.8%

1.92%

11.3%11.5%

+ 1.7%

- 9.7%- 11.2%

a Accuracy expressed as ratio of experimental value to known value; ^ Precision expressedas %RSD; c Not Applicable; d Calculated assuming a theoretical air sample volume of 1170m3(equivalent to mean air sample volume for 72-hour sampling duration) .

Conclusions

The performance of the method used to measure 2,3,7,8-TCDD in ambient air during removalactions at contaminated sites has been evaluated and shown to be both accurate and preciseat concentrations near the method's detection limit of approximately 1 pg/m3. Both the fieldand analytical components of the measurement process contributed a negative bias to the con-

148

centration measurements. Although within acceptable levels, an evaluation of the individual

contributions of the field and analytical components of accuracy and precision to the overallperformance shows that the major source of bias and variability are associated with theanalytical component. The sample collection component of the measurement process was foundto contribute only a small fraction to the total method bias and variability. In additionto having good accuracy and precision, the method was shown to be resistant to generatingfalse positive measurements as indicated by the field blank results.

MENTION of trade names or ccmnercial products does not constitute endorsement or reconnendation '

for use.f

References Cited

(1) Ooapendiun of Methods for the Determination of Toxic Organic Chemicals in Ambient Air

(Method T09), U.S. Qwironmental Protection Agency, April, 1984, 600/4-84-027 [f\

\0(2) Jackson, M. D. and Lewis, R. G. In "Sampling and Analysis of Toxic Ocganics in the Afe^

mosphere1'; Verner, S. S., 53.; American Society for Testing and Materials: Philadeldua,

PA, 1980: pp 36-47; ASTM Special 'technical Publication 721 —

0(3) Lewis, R. G.; Jackson, M. D. Anal. Chem. 1982, 54, 592-594

(4) teller, C. D.; Bidleman, T. F. Atmospheric Bwironment 1984, 18, 837-835

(5) Smith, R. M.; O'Keef, P. W.; Hilker, D. R.; Aldous, K. M. Anal. Chem. 1986, 58, 2414-2420

(6) Fairless, B. F., et al; Bivironmental Science & 'technology 1987, 21, 550-555

(7) Contract Laboratory Program, U.S. EPA, Statement of Work for Rapid-Turnaround Dioxin

Analyses, Multi-media, Itov. 1986; Revised June, 1987; Contract Laboratory Program,

U.S. Environmental Protection Agency, Washington, D.C.