Environ. Sci. Technol. 1991, 25, 427-436
(29) Leckie, J. 0.;Appleton, A. R.; Ball, N. B.; Hayes, K. F.; Honeyman, B. D. Adsorptive Removal of Trace Elements
Ridley, W. P.; Dizikes, L. J.; Wood, J. M. Science 1977,197, 329-332.
from Fly-ash Pond Effluents onto Iron Oxyhydroxide; EPRI-RP-910-1;Electrical Power Institute: Palo Alto, CA,
Topping, G.; Davies, I. M. Nature 1981, 290, 243-244. Craig, P. J.; Rapsomanikis,S. Environ. Sci. Technol. 1985,
1984. (30) Fuller, C. C.; Davis, J. A. Nature 1989, 340, 152-154. (31) Spycher, N. F.; Reed, M, H. Geochim. Cosmochim. Acta 1989,53, 2185-2194. (32) Moore, J. N.; Ficklin, W. H.; Johns, C. Enuiron. Sci. Technol. 1988,22,432-437. (33) Ferguson, J. F.; Gavis, J. Water Res. 1972, 6, 1259-1274. (34) Cutter, G. A. Deep Sea Res., submitted. (35) McBride, B. C.; Wolfe, R. S. Biochemistry 1971, 10, 4312-4317. (36) Sanders, J. G. Chemosphere 1979, 3, 135-137. (37) Ecological Research Associates. Reuter, J. E.; Slotton, D. G.; Goldman,C. R. McLaughlin Mine, Homestake Mining Co, Annual Monitoring Report, July 1,1988-June 30,1989. (38) Balistrieri, L. S.; Murray, J. W. Geochim. Cosmochim. Acta 1982,26, 1041-1052.
19, 726-730. Cooke, T. D.; Bruland, K. W. Enuiron. Sci. Technol. 1987, 21, 1214-1219. Andreae, M. 0. Anal. Chem. 1977,49, 820-823. Howard, A. G.; Comber, S. D. W. J . Appl. Organomet. Chem. 1989, 3, 509-514. Murphy, J.; Riley, J. Anal. Chim. Acta 1962, 31-36. Waslenchuk, D. G.; Windom, H. L. Estuarine, Coastal Mar. Sei. 1978, 7, 455-464. Knox, S.; Langston, W. J.; Whitfield, M.; Turner, D. R.; Liddicoat, M. I. Estuarine, Coastal Mar. Sci. 1984,6234539. Langston, W. J. Can. J. Fish. Aquat. Sci. 1983,40,143-150. Reimer, K. J.; Thompson, J. A. J. Biogeochemistry 1988, 6, 211-237. Andreae, M. 0.;Klumpp, D. Enuiron. Sci. Technol. 1979, 13, 738-741.
Beauchemin, D.; Bednas, M. E.; Berman, S. S.; Mclaren, J. W.; Siu, K. W. M.; Sturgeon, R. E. Anal. Chem. 1988, 60, 2209-2212.
Petersen, M. L.; Carpenter, R. Geochim. Cosmochim. Acta 1986, 50, 353-369.
Frost, R. R.; Griffin, R. A. Soil Sci. SOC.Am. J . 1977, 41, 53-57.
Received for review January 8,1990. Revised manuscript received September 11,1990. Accepted October I , 1990. Funding for our As project came from U.C. Seed Funds, Contract 403151 -05397-5, and via R. Flegal from the U.C. Toxic Substances Research and Teaching Program. The field program was funded by a grant to G. Gill and K.B. from EPRI Contract RP2020-9.
Evaluating the Multimedia Fate of Organic Chemicals: A Level I11 Fugacity Model Donald Mackay” and Sally Paterson Institute for Environmental Studies, University of Toronto, Toronto, Ontario, M5S 1A4, Canada
rn A multimedia model is developed and applied to selected organic chemicals in evaluative and real regional environments. The model employs the fugacity concept and treats four bulk compartments: air, water, soil, and bottom sediment, which consist of subcompartments of varying proportions of air, water, and mineral and organic matter. Chemical equilibrium is assumed to apply within (but not between) each bulk compartment. Expressions are included for emissions, advective flows, degrading reactions, and interphase transport by diffusive and nondiffusive processes. Input to the model consists of a description of the environment, the physical-chemical and reaction properties of the chemical, and emission rates. For steady-state conditions the solution is a simple algebraic expression. The model is applied to six chemicals in the region of southern Ontario and the calculated fate and concentrations are compared with observations. The results suggest that the model may be used to determine the processes that control the environmental fate of chemicals in a region and provide approximate estimates of relative media concentrations.
Introduction There is an incentive to develop methods of calculating, or modeling, the multimedia environmental fate of organic chemicals that are or, may be, in frequent commerical use or are inadvertently discharged to the environment. Examples are pesticides, PCBs, wood preservatives, and byproducts of incineration. A series of fugacity models has been developed for this purpose (1-4), which use physical-chemical properties, reactivity, transport characteristics, and extent of release to produce a comprehensive picture of environmental behavior. In well-defined situations the models can be used to estimate prevailing con0013-936X/91/0925-0427$02.50/0
centrations under steady- and unsteady-state conditions. The fugacity concept has been applied by several groups internationally to a variety of chemicals and environmental conditions (5-10). In this study we describe a novel, relatively simple, four-compartment level I11 fugacity model which, we believe, will prove useful for elucidating the fate of chemicals and may even have the potential to estimate approximate relative concentrations in a specified region. There are two significant departures from previous fugacity models. First, the number of key environmental media is reduced to four (air, water, soil, and bottom sediment), permitting simple algebraic solution of the steady-state equations. These compartments are treated as assemblies of subcompartments, which may consist of air, water, solid, and biotic matter. Equilibrium (equifugacity) is assumed to apply within each compartment, Le., between subcompartments, but not between compartments. Second, early evaluative fugacity models (2) treated only diffusivetransfer processes (such as volatilization/absorption), while this model includes nondiffusive or “one-way” processes in which chemical is conveyed between media by association with material that is undergoing intermedia transfer (e.g., wet deposition). Partitioning, reaction, and diffusive- and nondiffusivetransfer processes are generally treated as described in more recent level I11 models (3) and are only briefly reviewed here. In total, the model treats partitioning between 10 subcompartments, 4 compartmental total reaction (degradation) rates, 13 intermedia-transfer processes, and advective flows in air and water. These transfer processes are expressed in terms of D values by use of previously established relationships. Each D value contains two terns, a Z value, which is chemical specific (and
@ 1991 American Chemical Society
Environ. Sci. Technol., Vol. 25, No. 3, 1991 427
Table I. Properties of t h e Compartments: Volume Fractions, Organic Carbon ( O C ) Fractions, Densities (kg/m3), Volumes (m3), Areas (m2), a n d Depths (m) subcompartmts 0')vol fractns $I,, fractn air water solids biota OC in (1) (2) (3) (4) solids
bulk compartmt
(i)
air (1) 1.0 water (2) 0 soil (3) 0.2 sediment (4) 0 density (subcompartmt) 1.19
0 1.0 0.3 0.7
1000
2 x 10-11 5 x 104 0.5 0.3 2400n
0 1 x 104 0 0 1000b
0.2
0.02 0.04
density volumes (bulk compartmt) S. Ontario eval 1.19 1000 1500 1420
4 x 1014 4 X lo'* 1.2 X 1O'O
2 x 109 2 X lo7 6 X lo4
8 X lo8
4000
areas S. Ontario eval 2 x 1011 8 X 10'O 12 X 10'O 8 X 10'O
1x
io6
4 6 4
lo5 lo5 lo5
X X X
depths 2000 50 0.1 0.01
Soil solids. bAquatic biota.
is obtained from physical-chemical data), and kinetic or transport rate terms such as a mass-transfer coefficient, diffusivity or advective flow rate. It is important to clarify the present objectives, assumptions and claims. The primary aim is to establish the general behavior profile, or likely fate of a specified chemical in a specified region. We believe that this can be achieved by transport parameters that are common to all chemicals. It is appreciated that quantities such as diffusivities may vary from chemical to chemical by a factor of -4, but for this purpose single values may be adequate. A second aim is to estimate order of magnitude concentrations from defined discharge rates. Ideally, the estimated concentrations should be compared with measured values to validate the model. Unfortunately this is not presently possible because of lack of discharge data and uncertainty about many of the model parameters. A good fit could be fortuitous. Accordingly, we focus primarily on the concept behind the model and regard the actual estimations as more illustrative than definitive. It is hoped that more definitive models can be developed and validated, especially on a site-specific basis, using accurate parameters for chemical and environmental properties and discharge rates. Fugacity, 2 Values, and D Values. Partitioning of chemical between phases can be described by the equilibrium criterion of fugacity f (Pa),which can be considered to be the partial pressure of a chemical in a phase and is related to concentration C (mol/m3) by the expression
c = fZ
(1)
where Z is the fugacity capacity (mol/m3.Pa). Methods of calculating 2 values for environmental phases have been reviewed by Mackay and Paterson ( I ) . Chemicals move between phases by both diffusive and nondiffusive processes. The diffusive flux N (mol/h) between two phases, l and 2 , can be described by
N
= D(fi- f2)
(2)
where D is a transfer coefficient with units of mol/h.Pa, and f l and f 2 are the phase fugacities. The difference between f l and f 2 determines the direction of diffusive flux (but not nondiffusive flux) that takes place from high to low fugacity. D is a function of 2 values, interfacial areas, and diffusion properties in adjacent phases ( 2 ) . Nondiffusive or one-way transfer processes between phases, for example, wet or dry particle deposition from the atmosphere to soil or water, or suspended sediment deposition or resuspension, can also be described by a transport parameter as N = GC = GZf = Df (mol/h) (3) where G is the volumetric flow rate (m3/h) of the transported material (e.g., sediment), and D again has units of mol/Pa.h. 428
Environ. Sci. Technol., Vol. 25, NO. 3, 1991
The diffusive and nondiffusive D values can be summed for all transfer processes from phase 1to 2 (DI2)and phase 2 to 1 ( D 2 J and the net flux from 1 to 2 then becomes N = D12f1- D2Jz (4) First-order reaction processes in a phase can be described by N = h~vc = hRVZf = DRf (mol/h) (5) where hRis a first-order rate constant (h-l), Vis the phase volume (m3),and DR again has units of mol/Pa.h. In summary, all rates of chemical transport and transformation can be expressed as products of fugacities and D values. D values can be summed and compared when the processes apply to the same source phase fugacity. This introduces a valuable algebraic simplicity to the mass balance equations described later. Model Description Environmental Media. Four major bulk compartments are defined each of total volume, Vi,-air, water, soil, and sediment-subscripted 1-4. These consist of a combination of subcompartments of pure and particle phases of volume, Vi,, and are defined as a volume fraction, 4i, or V,/Vi, of the bulk phase. For example, bulk water consists of "pure" water, suspended solids, and fish; soil consists of mineral matter, organic matter, water, and air; air consists of pure air and aerosols; sediment is mineral matter, organic matter, and water. This environment can be evaluative in nature as described in previous papers ( 3 , I I ) or can represent an actual region. We apply the model here to (i) an environment scaled to represent approximately southern Ontario and (ii) an evaluative environment of area 1 km2 with similar phase proportions. The southern Ontario region is considered to extend from the Great Lakes system in the south to the French River in the north and from the Quebec border in the east to a line extending from Lake St. Clair up through Lake Huron. The water compartment, which covers approximately 40% of the area, includes all of Lake Ontario, the northern half of Lake Erie, and the eastern one-third of Lake Huron as well as all other lakes and rivers included within the defined boundaries. The average water depth is assumed to be 50 m. The air height has been reduced to an arbitrary 2000 m from 6000 m as suggested by Neely and Mackay (11) to reflect conditions of atmospheric accessibility in southern Ontario. The suggested compartment volumes are given in Table I. It is assumed that equilibrium exists within each compartment; Le., a common fugacity applies. This implies that the times required to reach equilibrium within a compartment are short compared to times required to reach equilibrium between compartments. Fugacity Capacities. For each subcompartment, a Z value or fugacity capacity is calculated. These values are then combined with the volume fractions to give an overall
Table 11. Definition of 2 Values subcompartments
0’)
Z value
air (Zil) water (Ziz)
ref
R = 8.314(Pa m3/mol K) T = absolute temp K H = Henry’s law constant (Pa m3/mol) Cs = aqueous solubility (mol/m3) P = vapor pressure (Pa) x i ] = fraction organic carbon KO, = organic carbon partition coefficient = 0.41 Kow pii = density of solids (kg/L) PLa= liquid vapor pressure (Pa) Kow = octanol/water partition coefficient
1/RT 1/H or Cs/P
soil, sed solids, water partic
(Zi3)
aerosols (Z1J biota (fish) (Z24) compartments air (1) water (2) soil (3) sediment (4)
xijKo,y,/H 6 X 106/(PLSRn 0.048pi,Kow/H z.= &..Z.. J 11 11 z: = 211 + 613213
1 1
13 12 13
z2 = 2 2 2 + $23223 + 6 2 4 2 2 4 2 3 2 4
= $31231 + = 64zZ42 +
63Zz32
+ 63&33
$43243
Table 111. Physical-Chemical Properties of Selected Chemicals compound
molec mass, g/mol
vapor pressure, Pa
solub, g/m3
hexachlorobiphenyl“ (6-CB) benzene benzo[a]pyrene (BAP) hexachlorobenzene (HCB) mirex trichloroethylene (TCE)
350 (15) 78.11 (16) 252.3 (16) 284.8 (16) 545.59 (16) 131.5 (16)
0.0005 (15) 12700 (17) 7.3 X (16) 0.0015 (17) 0.00013 (19) 9870 (3)
0.0035 (15) 1780 (17) 0.0038 (18) 0.005 (18) 0.00007 (19) 1100 (3)
1% Kow 6.8 (15) 2.13 (18) 5.98 (18) 5.47 (18) 6.89 (3) 2.29 (3)
Average properties of isomers.
Z value for the compartment as illustrated in Table 11, Le., Z, is the sum of the terms. ( i ) Air. The air consists of air and particulate matter (aerosols). The fugacity capacity of air, Zll, is calculated as 1/RT where R is the gas constant (8.314 Pa.m3/mol-K) and Tis the absolute temperature. Aerosol particles occupy a volume fraction, $13 of 2 X lo-”, equivalent to -50 pg/m3. Partitioning between particulate and gaseous air phases is inversely proportional to the vapor pressure (12) with the dimensionless particulate-air partition coefficient expressed as
Ksl = 6
X
106/P,s = ZI3/Zl1
(6)
where PLs is the chemical’s subcooled liquid vapor pressure (Pa). The fugacity capacity for air particulates is calculated from eq 6 and the overall Z value is calculated as shown in Table 11. (ii) Water. Water consists of pure water containing suspended sediment particles and fish. For pure water Z2, is 1/H ( I ) where H is the Henry’s law constant defined in Table 11. The 2 values for the particle (Z23)and biota or fish (Z24)phases are calculated from the dimensionless phase-water partition coefficient, which is estimated from the organic carbon content of the particulates (0.2) or the lipid content of fish (0.048) and KO,(13). These Z values are combined with their appropriate volume fractions, $23 of 5 X (equivalent to -12 mg/L) and 4 2 4 of lo4, to give the compartment value shown in Table 11. It should be noted that the volume fraction of “fish” is unrealistically high and is an attempt to represent the inclusion of other biota, especially microorganisms. (iii) Soil. Following Jury et al. (14),soil is assumed to contain a solid phase (volume fraction $33 of 0.5) combined with air (431 or 0.2) and water ($32 of 0.3). The fugacity capacity for the solid phase, Z33,is calculated as for water particles from the product of the pure soil-water partition coefficient and the Z value for water. The fraction organic content in the solid phase x33 is 0.02, and the density p33 is 2.4 kg/L. The 2 values of solids, air, and water are combined, as shown in Table 11, with their volume fractions to give an overall fugacity capacity.
(iv) Sediment. This compartment consists of solids and pore water with volume fractions of $@and $42, with values of 0.3 and 0.7, respectively. As in the case of soil, the Z value for solids (Z43)is calculated from organic carbon content ( x d 3= 0.04) and density p43 = 2.4 kg/L, and the combined bulk Z value is deduced as in Table 11. The parameter values used above should be regarded as only illustrative or typical values. More accurate, better justified values are needed in any site-specific model. Level I Calculations. Although the primary purpose of this paper is to describe the level I11 model, it is useful to include a description of the simpler level I and I1 models [which are already widely used for estimating chemical fate (5-10)], the level I model being illustrated at this point. By use of these volumes and Z values, and an assumed total amount of chemical present, M moles, it is possible to calculate its equilibrium partitioning in this multimedia environment. A mass balance gives M = CViCi = fCViZi (7) The prevailing fugacity can thus be calculated as M / ViZi,followed by bulk concentrations, individual subcompartment concentrations (as Z f ) ,and amounts as CV or VZf. This provides a picture of the dominant media into which the chemical partitions and provides a first, approximate estimate of relative concentrations. The primary difficulty is estimating M . Two approaches can be used. First is to gather concentration data for various media and calculate and sum amounts for the region of interest. Second, if an emission or use rate is available e.g., 1000 kg/year, and an approximate environmental persistence can be estimated, e.g., 0.3 year, then the amount present will be their product or 300 kg. If concentration data are available for one medium (e.g., fish), it is possible to select M by trial and error to obtain the reported value for fish. There is usually order of magnitude uncertainty about M . An illustration of such a calculation is given for PCBs in southern Ontario with properties shown in Table 111. It is estimated that M is 1.75 X lo5 kg from which the fugacity, concentrations, and amounts can be calculated as shown in Table IV. The results show that because of Environ. Sci. Technol., Vol. 25, No. 3, 1991 429
Table IV. Level 1 Distribution of a PCB in Southern Ontarioo
bulk bulk bulk bulk
air water soil sediment
“Total amount 5 Pa.
X
concn
amount, mol
5 ng/m3 1.6 ng/L 8.6 ng/g 10.9 ng/g
5900 1.8 x 104 4.4 x 105 3.5 X lo4
%
1.2
3.6 88.2
7.1
lo5 mol or 1.7 X lo5 kg; fugacity 2.95 X
Table V. Level 2 Distribution of a PCB in Southern Ontarion input output
emissions advection fugacity residence time of chemical amount
compartment
concn
bulk air
5.0 ng/m3
bulk water bulk soil bulk sediment
8.3 ng/g
bulk air bulk water
45 mol/h or 1.38 X lo5 kg/yr 10.9 mol/h or 3.34 X lo4 kg/yr 2.85 X Pa 1.2 yr
4.8
X
lo5 mol
reaction
1.5 ng/L 10.5 ng/g
0.035 6.4 0.5
loss rate. molih advection
47 0.03 (to stratosphere) 1.4 (leaching) 0.4 (burial)
advective flow rate, m3/h
resid time, days
inflow concn, mol/m3
3.3 x 10’2 3.3 X lo8
5 500
3 x 10-12
3 x 10-9
Advective flow rates for the evaluative environment are a factor of 2 X lo5 (the ratio of the areas in Table I) lower than those of southern Ontario.
its hydrophobic nature the highest concentrations of PCBs are found in the biotic, soil, and sediment solid phases. The greatest amount will be found in soil due to its large volume. Chemical Loss Processes. Two loss processes are considered in level I1 and I11 calculations, advection, i.e., loss by outflow, mostly in air or water, and degrading reactions. Air and water advective flow rates and residence times from which they are calculated are suggested in Table V, from which advective D values (DA) can be calculated as GZ, where G is the flow rate (m3/h) and Z applies to the flowing phase, either pure or bulk as desired. The air and
water residence times are assumed to be 5 and 500 days, respectively. The air mass residence times correspond to an approximate average wind speed of 3 m/s. The water residence time is an estimated mean for the varied lake and river sizes in southern Ontario ranging from a few days for small rivers to approximately 7 years for Lake Ontario. Also included are three “pseudo-advection’’ processes. Transfer from air to higher altitude is assumed to occur at a velocity of 0.01 m / h or approximately 90 m/year, as discussed previously ( 3 ) . Leaching from soil to groundwater is set for illustrative purposes at 40% of the rain rate on soil. Sediment burial (or net sedimentation) is assumed to be 0.3 mm/year solids and equals the difference between sediment deposition and resuspension rates. These “velocities” are included in Table VI. Other values representative of other regions can be substituted. Reaction rates are included as first-order rate constants applicable to the compartment or subcompartments. These rate constants or half-lives are chemical specific and usually result from a consideration of the rates of processes such as biodegradation, photolysis, oxidation, and hydrolysis, individually and in total. The D value (DR)is calculated as discussed earlier. Level I1 Calculations. If emission rate data are available, a steady-state, equilibrium calculation can be completed. The emissions rate should include point and nonpoint sources, and to it should be added advective inflow at a suitable “background” concentration or fugacity (fB1). Reaction rates are also included if data are available. A mass balance over the entire environment gives NT = NE + C D A ~ = B~(CPA + ,C&) (8) where N E is the local emission rate into all media and NT the total input rate (mol/h). The prevailing fugacity can be calculated, followed by concentrations, amounts, and inflow and loss rates. It is illuminating to examine the relative importance of these loss processes. The total amount present M can be calculated and used to estimate the chemical’s overall “persistence” or average residence time 70 as MINT (h). This time is a combination of the reaction time or persistence and the advection residence time. It can be shown that the reaction time 7 R is M / (fCDR),the advection time 7 A is M / ( f x D A ) and ,
+
1/7,, = 1 / 7 ~ 1 / 7 ~
The shorter residence time, representing the faster process, dominates. Table V gives an illustrative calculation for PCBs in southern Ontario for an assumed combined emission and
Table VI. Assumed Transport Parameters parameter air-side MTC over water water-side MTC transfer rate to higher altitude rain rate (m3 rain/m2 areaah) scavenging ratio dry deposition velocity air-side MTC over soil diffusion path length in soil molecular diffusivity in air molecular diffusivity in water water runoff rate from soil solids runoff rate from soil water-side MTC over sediment diffusion path length in sediment sediment deposition rate sediment resuspension rate sediment burial rate leaching rate from soil to groundwater 430
Environ. Sci. Technol., Vol. 25, No. 3, 1991
symbol kVA
kvw
US
2 L‘P
kSA
y3 BMA BMW
UWW Us, kYW y4
UDX URX UBX
UL
(9)
value 3 m/h 0.03 m/h 0.01 m / h (90 m/yr) m/h (0.85 m/yr) 9.7 X 200 000 10.8 m/h (0.003 m/s) 1m/h 0.05 m (half soil depth) 0.04 m2/h (3) 4.0 X lo4 m2/h (3) 3.9 X m/h (0.34 m/yr) 2.3 X m3/m2.h (0.0002 m/yr) 0.01 m/h 0.005 m (half sediment depth) 4.6 X m3/m2.h (0.0004 m/yr) 1.1 X m3/m2.h (0.0001 m/yr) 3.4 X m3/m2.h (0.0003 m/yr) 3.9 X m3/m2.h (0.34 m/yr) (40% of rain rate)
Table VII. Calculation of D Parameters compartment air (1)-water (2)
air (1)-soil (3)
individual D
process diffusion rain wet deposition dry deposition diffusion rain wet deposition dry deposition
DV =
l/(l/kVAAlZZIl
total D
+ ~/~VWAIZZZZ)
DQW = A12UQZ22 DDW= A ~ z U Q Q @ J I ~ Z I ~ Dpw = A12Upq413Z13 Ds = ~ / ( ~ / ~ S A A I + S ZYI3I/ ( A 1 3 ( ~ A 3 ~ 1+ 1 BW3Z22)) DQS
=
A12UQZ22
DDS = A13U~Qq413Z13 Dps = A13Upq413Z13
Di2 D21
= DV DQW+ DDW+ DPW = DV
D13 D31
= DS + DQS+ DDS+ DPS
D32
= DSW + DWW =0
= DS
soil (3)-water (2) sediment (4)-water (2) reaction advection
soil runoff water runoff diffusion deposition resuspension either bulk phase or pure phase bulk phase
Dsw = A13UswZ33 DWW= A13UWWZ22 DY = l / ( l / k Y W A 2 4 Z 2 2 + Y,/Bw4A24Zzz) DDX = GDXZ23 D R X = GRXZ43 D R ~= kRiVLZl DRIJ
=
DAL=
‘RiJ
‘iJZi,
GzZ~
advection rate of 171000 kg/year. Approximately 18% of this total is due to background inflow in air and 2% in water. The results show the same equilibrium distribution as level I. However, the total amount present is reduced to 169 000 kg by reaction and advection. The dominant removal process is advective flow in air, which transfers 84% of local emissions and inflow to adjacent regions. Although assumed specific reaction rates are similar in soil and sediment, the removal rate is higher in soil (11.5%) due to its greater volume. Water advection accounts for 2% of total removal and sediment burial and reaction each account for approximately 1’70. The overall residence time is 1.2 years with a reaction persistence of 8 years. This behavior profile contains the assumption that a common fugacity prevails throughout the environment. Clearly this is in error because fugacities tend to be higher in media that receive direct discharges and lower in those that have relatively rapid loss mechanisms. In particular, it appears that the air often experiences a low fugacity because of the high rate of advective loss. To remedy this assumption it is necessary to include expressions for interphase transport conductivities or resistances and introduce the emissions on a medium-specific basis. We now treat this more complex level I11 calculation. Intermedia Transport Parameters. Assumed transfer parameters necessary for calculation of intermedia fluxes are given in Table VI, areas being given in Table 1. These parameters are converted into D values as shown in Table VII. A brief summary of their derivations follows, details being given elsewhere ( 3 ) . (i) Air-Water. Diffusive processes of volatilization and absorption are described by the parameter Dv, where 1/Dv = l/DVA + l/Dvw = l/(kVAAlZZll) + l/(kVWAlZZZZ) (10) where k V A and kvw are air- and water-side mass-transfer coefficients (MTC) (m/h), DVA and Dvw the transfer parameters (mol/h.Pa), and A,, is the interfacial area (m2). The same mass-transfer coefficients are assumed to apply to all chemicals. Nondiffusive chemical transfer by rain dissolution and wet and dry deposition to water are described by the appropriate G,Z, products as DQW,DDW, and Dpw, respectively, as described by Mackay et al. (12). The transfer parameters necessary to calculate the net flux between bulk air and water become air-water: D12 = Dv + DQw+ DDW + Dpw (11) water-air: D2, = Dv (12) total flux: N,, = D12fl - Dzlfi (13)
D23 024
D4Z DRi DRt DAi
= DY + DDX = DY + D R X = kRiViZi
=
C(kRLJViJZiJ)
=
G~zz
It is appreciated that the use of single values for deposition velocity and scavenging ratio is simplistic, but in the interests of brevity single values are used for all chemicals. (ii) Air-Soil. Diffusion from soils can take place in the air or water phases in parallel as described by Fick’s first law as flux = A,,BAC/Y (mol/h)
(14)
where B is an effective diffusivity in air (B,) or water (Bw,) (m2/h),AC is the concentration difference (mol/m3),and Y is the diffusion path length (m) to which the difference applies. B is calculated as in the model developed by Jury et al. (14) using the Millington and Quirk expressions and air and water molecular diffusivities common to all chemicals. A mass-transfer coefficient kSA (of 1m/h) combined with the air fugacity capacity and area characterizes the air-side boundary layer diffusion parameter DSA. Since the resistance 1/DsA is in series with the two parallel resistances in soil-air and water, 1/DAD and 1/ DWD,the total D value for soil-air diffusion becomes Ds = ~ / ( ~ / D s+A DAD
+ DwD))
(15)
Nondiffusive transfer of chemical by atmospheric deposition is included to give total D values of air-soil: D,, = Ds
+ DDS + Dps + DQS
soil-air: D3, = Ds
(16) (17)
where DDS, Dps, and DQS are parameters for wet and dry deposition and rain washout from air to soil and are calculated as for air-water transfer but adjusted to reflect the soil-air interfacial area. (iii) Water-Sediment. Water-sediment diffusion is calculated similarly to soil-air diffusion by calculating an effective diffusivity in sediment pore water. The total diffusive transport coefficient, Dy, between sediment and water is the series combination DY = 1/(1/Dxw
+ 1/Dyw)
(18)
Diffusion in pore water is described by the parameter Dxw and in the water boundary layer by the parameter Dyw DYW = kYWA24Z22
(19)
where k Y W is the water-side mass-transfer coefficient. Transfer by suspended sediment deposition and resusand pension are described by “velocities” UDx (4.6 X URx (1.1X m3 of solids/m2 area.h or m/h, which are Environ. Sci. Technol., Vol. 25, No. 3, 1991
431
-
n
where
'p
4 2 G X
~ 0 0 0 0 0 h
+ -
g m
g 5 200000 DT1 = D12 t Dl3 t Dhi t D R ~ t Dsr DT2 = D21 t Dz, t DM t Dm DT3 = D31 t D3, t D R t~ DL DT, = D42 t D R t~ DBX Figure 1. Algebraic solution of model equations.
combined with sediment area to give flow rates G D X and GRX. The respective transfer parameters become DDX
=
GDXZ23
(20)
DRX
=
GRXZ43
(21)
h
a,
v
Inclusion of these parameters gives overall D values for sediment-water transfer water-sediment: D24 = D , sediment-water: D,z = D, m 0 0 0 0 0
x x x
x
cr. m m
0 0 CD
fib
s
0
0-
m o m i o o h
(22) (23)
(iv) Soil-Water. One-way transfer of chemical from soil to water includes water runoff, Dww, and associated soil loss, Dsw, calculated from appropriate GZ values, i.e.
Dww = GwZ32
(24)
Dsw = GsZ33
(25)
The parameters for total transfer become soil-water: D32 = Dww
h
+ DDX + DRX
+ Dsw
water-soil: DZ3= 0
(26) (27)
The water and solids runoff rates from soil set at 40% of the rain rate on soil are expressed per unit area of soil and represent fairly heavy runoff conditions. Runoff and leaching are, of course, episodic processes, but the model treats them as steady state, Le., an average rate applies. The expressions defining the D values are summarized in Table VII. Level I11 Calculations. Steady-state mass balance equations can be set up for each of the four bulk phases that incorporate emissions, transfer between adjacent phases, reaction, and advection. The four equations take the form
NE^ + D~ifsi- fi[CDij -I-DK + D R ~+] Zj[D,ifjl
=0 (28)
The solution for the four fugacities is algebraic, as shown in Figure 1. It is important to reemphasize that in all cases single typical values have been assumed for all kinetic terms; i.e., they are not chemical specific. There is thus 432
Environ. Sci. Technol., Vol. 25, No. 3, 1991
0.008 I
0.0
“0,)
Flow and reaction time= 5 years Persistence = 7.8 years
._.-.-. +
/ O . 107 /
_.-. 4.329
10.013 Leaching to Groundwater
diffusion wet particle dep. d r y particle dep. rain sediment resuspension sediment deposition
-
*-*
E M IS S IONS
TRANSFER
- - - + REACTION -.-.+ ADVECTION (rnol/hl
Figure 2. Level 3
/
N 3 .aa
1Sediment 2.95 Burial
f=fugacity(Pa) c=concentration (mol/m3 1 m=amount ( m o l ) ‘IO= percentage of total amount
m
distribution of hexachlorobiphenyl in southern Ontario.
no need for inclusion of more chemical-specific data in this version of the level I11 calculation. The dependence of flux on chemical properties is contained entirely in the 2 values, which comprise part of the D values. No claim is made that the assumed coefficients are applicable to other environments; indeed there is considerable doubt about some of the values in southern Ontario. The user is encouraged to modify these values in the light of experience. The values given should be regarded as only reasonable first estimates. The model contains approximately 40 environmentspecific parameters (volumes, velocities, diffusivities, etc.), which are assumed to apply to all chemicals. In many cases the values are not well established (e.g., sediment resuspension rate); thus, the user should regard these parameters as initial order of magnitude estimates, which will be subject to change in the light of experience. Chemical-specific property parameters are used only to estimate 2 values. Reaction rate constants are both chemical and environment specific. Finally, emission rates are region specific and are often the most poorly quantified of all variables. The steady-state level I11 output includes all fugacities, concentrations, in mol/m3 and conventional units, and transfer rates for individual and combined processes. A total mass balance can be assembled, as illustrated for PCBs in Figure 2, from which the overall chemical persistence can be calculated as in level 11. Figure 2 illustrates that most PCB introduced into the air is advected from the region, but there is appreciable
wet and dry deposition, which in the case of water is offset by volatilization. PCB present in water is subject to deposition and advective loss. The dominant loss processes are advective loss in air, reaction in soil, advection in water, reaction in sediment, and sediment burial. The overall persistence is 5 years, with the reaction persistence being 7.8 years. Most PCB is found in soils and sediments. The actual relative emissions to air, water, and soil are not known, but the rates are regarded as being reasonable for illustrative purposes. The PCB fugacities in soil, water, and sediment are fairly Pa, but the air fugacity is a factor of 10 similar at less due to rapid advective loss. The corresponding concentrations are in fair agreement with observed levels as reported in Table IX. The model is believed to be reproducing the generally accepted fate of PCBs.
-
Application to Other Chemicals Models such as this cannot be “validated” in the same sense that a simpler physical or chemical model can be validated. Environmental concentrations vary in time and space, and few reliable data are usually available. Reaction rate constants vary diurnally and seasonally. Emissions are rarely known accurately. There are variations and uncertainties in transport rate parameters, such as deposition and resuspension rates. Few environments are at steady state as described here. The best that can be hoped for is that the model corresponds with order of magnitude fidelity to fragmented observations for a variety of chemicals of quite different properties and pathways. Environ. Sci. Technol., Vol. 25, No. 3, 1991 433
4
1
'4
g 5
8
O
x
8 m
N
0
3 3
434
2 8
Environ. Sci. Technol., Vol. 25, No. 3, 1991
The chemical-to-chemical variation should be described entirely by the chemicals' properties and emission rates and not by adjustable parameters. But even this modest model capability is invaluable for predicting the likely behavior of new chemicals and for estimating order of magnitude concentrations. The model was run for selected chemicals of varying properties (Table 111);environmental concentrations were calculated and compared with observations. A common problem is estimation of reasonable emission rates. Assumed emissions, background inflow concentrations, and degradation rate constants are listed in Table VIII. Table IX lists the calculated compartment concentrations and fugacities and gives some reported values for comparison. Benzene's fate is dominated by atmospheric processes because of its high volatility. Predicted values of benzene for air and water concentrations are generally low by a factor of 10. It is probable that this discrepancy reflects the prevalence of measurements in urban areas where there is a much higher than average fuel use. A good match of model estimates with observations is obtained if it is assumed that emissions are a factor of 10 higher. This could represent more intense emission of benzene in urban areas. The estimated sediment, soil, and fish concentrations are very low; thus it is unlikely that monitoring these compartments will be useful. No concentration data could be found for these comparments. Benzo[a]pyrene is mainly an atmospheric pollutant, which is deposited in association with particulate matter, resulting in widespread contamination of soil and sediments. Emissions and an air advective inflow concentration of 5 ng/m3 are based on data reported by Canviro (22). It has been the subject of multimedia modeling by Ryan and Cohen (23). As with benzene, predicted concentrations tend to be low and it is likely that regions in which there is higher than average fuel usage, especially diesel fuel usage, experience concentrations that are larger by a factor of 10 than those estimated by the model. Emissions of hexachlorobenzene are assumed to be 80% into water and 10% each into soil and air. This results in a relative distribution that is in agreement with reported concentrations. HCB is primarily of concern as an aquatic pollutant with a high bioconcentration potential. The emissions were unknown and were chosen to produce levels of the correct order of magnitude. It is not known if these emissions are correct; thus, the results should be regarded as indicating the general fate, but not the concentrations. The main routes of entry of mirex to the Great Lakes have been via the Niagara and Oswego Rivers from production plants in the United States. It has been extensively monitored in Lake Ontario. Sediment and suspended sediment concentrations are well quantified, but water concentrations are now generally low; for example, Yin (31)reported 2 pg/L. Warry and Chan (25) estimated loadings of 13-20 kg/ year to Lake Ontario in association with suspended matter. Good agreement was found between predicted and reported water concentrations for these emissions. Sediment concentrations are underpredicted, which is probably due to previous contamination. Mirex is interesting because it represents a situation in which there was a high emission rate in the past, but present emissions are low. If the model is run for two emission scenarios corresponding to concentrations of inflow water differing by a factor of 500, then good agreement is obtained with concentrations experienced in the 1970s and with prevailing concentrations. The steady-state model can thus be used with appropriate caution to elu-
cidate unsteady-state behavior, provided that account is taken of the response time of the system, which in this case is 12.5 years. Trichloroethylene (TCE), a major industrial solvent, is introduced mainly by evaporation into the air. It is assumed that a small percentage (1%)is introduced to the aquatic environment. The Ontario emissions are scaled from U S . production data (26). A background atmospheric inflow of 80 ng/m3 is included. The resulting concentrations are in good agreement with data from La Jolla, CA, and Liverpool, England, and the model predictions of Cohen and Ryan (26).
Discussion The model provides a relatively simple, rapid method of establishing the relative importance of environmental fate processes for specific chemicals using minimum data. It is believed to give an adequate characterization of the dominant phases of accumulation. For example, mirex is 62% in soil and 35% in sediments, while benzene is 46% in air and 54% in water. These percentages are controlled by the emission sources and the chemical’s properties. An overall persistence or residence time can also be estimated that ranges from 6 days for benzene to 12 years for mirex. This persistence is a function of partitioning and rates of reaction and advection. The relative importance of advective inflow and local emissions as determinants of prevailing Concentrations can be estimated. The evaluative 1-km2version can be used to determine the dominant environmental partitioning pathways and the persistence of a chemical, using illustrative rather than real emissions. The computed absolute concentrations then have no significance. Since many jurisdictions have a common interest in estimating the environmental fate of chemicals, but the regions corresponding to these jurisdictions have differing ratios and natures of air, water, soil, and sediments, it is useful to have a common evaluative test system that can be used to intercalibrate environmental assessments. The combination of evaluative and region-specific environments thus provides a convenient system for international assessment of chemicals. A useful feature of both models is that a sensitivity analysis is possible by which the effects of varying properties of the chemicals and the environment can be explored. This establishes which properties should be estimated most accurately and can provide an estimate of likely error in concentration as a function of error in property. The model has been described in steady-state form, but it can be written in differential equation form and solved numerically to give the response in the region to timevarying emissions. The easiest solution is that using constant, annually averaged parameters and allowing only discharges and other inputs to vary. More difficult is the treatment of seasonal variation of parameters such as temperature and episodic events such as rainfall. If emission estimates are available, the model can be used to estimate concentrations. We cannot presently suggest a level of accuracy, but it is certainly no better than order of magnitude (factor of 10). It is believed that the model does give order of magnitude information about relative media concentrations, especially between water and sediment. The model’s steady-state nature of course precludes it from describing unstead.y-state conditions. The model is probably most reliable for substances, such as PCBs, that are persistent and widely dispersed in the environment. It can be used as an interpretive tool to explore temporal and spatial variations in concentrations. For example, it appears that mirex has been discharged
a t two time periods at emission rates that have differed by a factor of approximately 500. Since the model is linear, concentrations are linearly related to total (point and nonpoint) emissions; thus, there is no need to run the model repeatedly to explore such changes. Because certain contaminants such as benzene and BAP are emitted with considerable spatial variation, it can be misleading to use average emissions and concentrations. It may be possible to develop “rules”, based on observation, which may be region specific, that one would expect to encounter local concentrations that are perhaps a factor of 10 or 20 higher than average. There are three compelling incentives to test, improve, and ultimately validate multimedia models of this type. First, they would provide guidance about likely prevailing environmental concentrations in the various media which result from present or possible future emissions. This is useful when designing analytical monitoring and identification programs. The present model should be regarded as only providing relative concentration data. Second, they could provide guidance as to the proximity of the concentrations to those that are judged to be of toxicological or aesthetic significance. Generally, the estimated concentrations will be (or should be) well below lethal concentrations, but the quotient of effect-to-estimated concentration provides a numerical expression of the “safety factor”. Chemicals with low quotients are presumably of greatest concern and thus require priority treatment. The combination of the model with toxicological concentration information thus provides a method of establishing priorities among the large number of chemicals in present use and helps to assign priority to new chemicals. Third, the concentrations could provide a starting point for estimation of human exposure. Air inhalation and water ingestion rates can be combined with the appropriate media concentrations to estimate exposure by these routes. The relative importance of the various routes then becomes obvious. The major weakness in this model is the exclusion of vegetation. Persistent organic compounds, such as pesticides, herbicides, PCBs, and dioxins, have been found to partition into grains (34),fruits and vegetables (30,35),pine needles (36),and moss and lichens (37). They thus become directly available to herbivores and humans through the ingestion of grains, fruits, and vegetables, and indirectly through the consumption of domestic animals and milk. If partitioning into vegetation can be quantified, estimates can then be made of human intake by ingestion of these foods and fish in appropriate “market-basket’’ ratios. Such an exposure assessment can be extended further to include physiologically based pharmacokinetic models (38, 39) to predict partitioning, accumulation, and persistence in various animal or human tissues after prolonged, continuous exposure. Ultimately, it is believed, it will be possible to combine environmental, human exposure, and pharmacokinetic models in an overall process of tracking the pathways of a chemical from its sources, to its distribution in various environmental media, such as air, water, soil, sediment, and food, to its availability or exposure to animals and humans, and finally to target tissues. The model presented here contributes one small step in this process. Finally, we believe that the model can be, and should be, improved by obtaining better estimates of certain transport and partitioning parameters. For example, the mass-transfer coefficients and diffusivities are average values applied to all chemicals. Chemical-specific and Environ. Sci. Technol., Vol. 25, No. 3, 1991 435
region-specific values could be used instead to increase accuracy. Most important, the model should be tested more thoroughly against a variety of chemicals of widely differing properties. The principal difficulties in accomplishing this are the lack of reliable emission data, environmental concentration data, reaction rate constants, and physical chemical properties.
Conclusions A regional fugacity model has been developed, tested with six chemicals, and found to give a reasonable picture of environmental partitioning, reaction, advection, and transport characteristics. Estimated relative concentrations in the media are in order of magnitude accord with reported values, but there are considerable uncertainties in the emission rates on which these concentrations depend. Copies of a IBM-PC compatible computer program diskette and documentation are available on request from the authors. Registry No. 6-CB, 26601-64-9; BAP, 50-32-8; HCB, 118-74-1; TCE, 79-01-6; PCB, 92-52-4; benzene, 71-43-2; mirex, 2385-85-5.
Literature Cited Mackay, D.; Paterson, S. Environ. Sci. Technol. 1981, 15, 1006. Mackay, D.; Paterson, S. Environ. Sci. Technol. 1982, 16, 654.
Mackay, D.; Paterson, S.; Cheung, B.; Neely, E. B. Chemosphere 1985, 14, 335. Mackay, D.; Paterson, S. In Pollutants in a Multimedia Environment; Cohen, Y., Ed.; Plenum: New York, 1986; pp 117-131. Klein, A. W. In Compendium of Environmental Exposure Assessment Methods for Chemicals; OECD Hazard Assessment Panel, Paris, in press. Wood, W. P.; Nold, I. A,; Lefler, J. Enuironmental Partitioning Model (Draft Documentation); Office of Toxic Substances, U S . EPA: Washington, DC, 1982. Calamari, D.; Vighi, M.; Bacci, E. Chemosphere 1987, 16, 2359. Zitko, V.; McLeese, D. W. Evaluation of Hazards of Pesticides used in Forest Spraying to the Aquatic Environment. Can. Tech. Rep. Fish Aquat. Sci. 1980, No. 985. Sloof, W.; Cleven, R. G. M. J.; Janus, J. A.; Ros, J. P. M. Draft Basis Document Copper. RIVM Report No. 758474003, RIVM: Bilthoven, T h e Netherlands, 1987. Yoshido, K.; Shigeoka, T.; Yamauchi, F. Toxicol. Environ. Chem. 1987, 15, 159. Neely, W. B.; Mackay, D. In Modelling the Fate o f Chemicals in the Aquatic Environment; Dickson, K. L., Maki, A. W., Cairns, J., Jr., Eds.; Ann Arbor Science: Ann Arbor, MI, 1982; pp 127-143. Mackay, D.; Paterson, S.; Schroeder, W. H. Enuiron. Sci. Technil. 1986, 20, 810.
436
Environ. Sci. Technol., Vol. 25, No. 3, 1991
(13) Mackay, D. Enuiron. Sci. Technol. 1982, 16, 274. (14) Jury, W. A.; Spencer, W. F.; Farmer, W. J. J . Environ. Qual. 1983, 12, 558. (15) Shiu, W. Y.; Mackay, D. J . Phys. Chem. Ref. Data 1986, 15, 911. (16) Verschueren, K. Handbook of Environmental Chemicals, 2nd ed.; Van Nostrand Reinhold Co.: Toronto, 1983. (17) Mackay, D.; Shiu, W. Y. J. Phys. Chem. Data 1981,10,1175. (18) Miller, M. M.; Wasik, S.P.; Huang, G.-L.; Shiu, W. Y.; Mackay, D. Enuiron. Sci. Technol. 1985, 19, 522. (19) Smith, J. H.; Mabey, W. R.; Bohonos, N.; Holt, B. R.; Lee, S. S.; Chou, T.-W.; Bomberger, D. C.; Mill, T. Enuiron-
mental Pathways of Selected Chemicals in Freshwater Systems, Part 11: Laboratory Studies; EPA Report 60017-78-074; 1978. (20) Shiu, W. Y.; Doucette, W.; Gobas, F. A. P. C.; Andren, A.; Mackay, D. Enuiron. Sci. Technol. 1988, 22, 651. (21) Corpus Information Services, Toronto, Canada, 1988. (22) Canviro Consultants Ltd. Reuiew of Benzo[a]pyreneOccurence,Human Exposure and Health Effects;Report to Ontario Ministry of the Environment, Toronto, 1985. (23) Ryan, P. A.; Cohen, Y. Chemosphere 1986, 15, 21. (24) Beck, J.; Hansen, K. Pestic. Sci. 1974, 5 , 41. (25) Warry, N. D.; Chan, C. H. J. Great Lakes Res. 1981, 7,394. ( 2 6 ) Cohen, Y.; Ryan, P. A. Enuiron. Sci. Technol. 1985,19,412. (27) Canviro Consultants Ltd., Review of PCB Occurrence, Human Exposure and Health Effects; Report to Ontario Ministry of the Environment, Ontario, 1985. (28) Hazardous Contaminants Coordination Branch, Ontario Ministry of the Environment, 1986; personal communication. (29) Clark, T.; Clark, K.; Paterson, S.; Norstrom, R.; Mackay, D. Environ. Sci. Technol. 1988, 22, 120. (30) Davies, K. Chemosphere 1988, 17, 263. (31) Yin, C.-Q. Ph.D. Dissertation, College of Environmental Science and Forestry, State University of New York, Syracuse, NY, 1985. (32) Holdrinet, M. V. H.; Frank, R.; Thomas, R. L.; Hetling, L. J. J. Great Lakes Res. 1978, 4 , 69. (331 Eitzer. ~ ~ ,B. D.: Hites. R. A. Int. J . Anal. Chem. 1986,27,215. (34) Briggs, G. G.; Bromilow, R. H.; Evans, A. A. Pestic. Sci. 1982, 13, 495. (35) Iwata, Y.; Gunther, F. A. Arch. Enuiron. Contam. Toxicol. 1976, 4 , 44. (36) Gaggi, C.; Bacci, E.; Calamari, D.; Fanelli, R. Chemosphere 1985, 14, 1673. (37) Villeneuve, J. P.; Fogleqvist, E.; Cattini, C. Chemosphere 1988, 17, 399. (38) Paterson, S.; Mackay, D. Toxicol. Appl. Pharmacol. 1986, 82, 444. (39) Paterson, S.; Mackay, D. Environ. Toxicol. Chem. 1987,6, 395. ~
Received for review September 13, 1990. Accepted September 26,1990. We are grateful to the Hazardous Contaminants Coordination Branch, Ontario Ministry of the Environment for funding this work
