La historicidad perdida

V apour - particle partitioning of SOCs in the atm osphere was first described by Junge (1977), w ho used the gas-solid, linear Langm uir isotherm theory

w hich states, th at-th e rate, of ad sorption-of .a .com pound to a surface is proportional to the com pound's vapour pressure and the am ount of surface area available:

0 = c0/(P + c0) (2.1)

W here 0 is the fraction of the total atm ospheric concentration of a

c o m p o u n d so rb ed to the aerosol, P is the v ap o u r p ressu re of th at

com pound, 0 is the concentration of the aerosol surface area (cm2 cm-3 air) an d c is a param eter depending upon the sorbate molecular weight, surface concentration necessary for m onolayer coverage and the heat of desorption of th at com pound. Junge's approach to gas-particle partitio n in g p roved difficult. H ow ever, since finding the concentration of an aerosol surface area (cm2 cm-3 air) to param eterise the distribution process proved to be a difficult task, Yamasaki et al. (1982) got around the problem by assum ing th at the surface area is linearly related to the total suspended particulate (TSP jig m -3) in the atm osphere. By using TSP, Junge's isotherm approach could be a p p lied th ro u g h the use of a com pound- an d tem p eratu re- d ep en d en t therm odynam ic partition coefficient of the form:-

K =

F/TSP (2.2)

A an d F rep resen t the equilibrium concentrations of a com pound in the v ap o u r and particulate phase respectively (ng m -3). F is typically defined as

the filter retained concentration and A the adsorbent retained concentration of an air sam pling system. The adsorbent is typically located dow nstream of the filter in any air sam pling system. SOC air sam pling is discussed in Section 2.6. The quantity F/TSP (ng fig"1) represents the therm odynam ic activity o n /in the particulate matter. The constant K can be view ed as the e q u ilib riu m ratio of A to F/T SP, i.e. as the equilibrium ratio of the concentration of a com pound in the gas phase to that in /o n the particulate m atter. Yamasaki et al. (1982) chose to invert Equation (2.2) to show that an increasing partition coefficient, K, denotes decreasing partition to the solid p h a se. A lth o u g h th is is the in v erse of the u su a l co n v en tio n for param eterising a two com partm ental system the partition coefficient Kp is now usually expressed as:-

K = F/TSP = F = K-l

A A(TSP) (2.3)

V arious studies have been selected w hich can be divided up into three distinct observations to support Equation (2.3) in validating vapour-particle distributions of SOCs in the atmosphere.

O bservation 1: The com pound dep en d en t Kp values, at typical am bient tem p eratu res, for a range of PAHs have been found to be rem arkably sim ilar in different cities including: Osaka, Japan (Yamasaki et al., 1982), P o rtlan d , OR (Ligocki and Pankow , 1989) and Chicago, IL (Cotham and Bidlem an, 1992).

Observation 2: At a certain tem perature, such as 20 °C, a linear relationship w as found betw een log Kp and log P°l [where log P°L is the sub-cooled liquid vap o u r pressure of that compound] (Foreman and Bidleman, 1987). This experim ental data supports the theory of Pankow (1987), w here Kp should correlate w ith com pound vapour pressure at a certain tem perature.

O bservation 3: Log Kp plotted against inverse tem perature (1/T ) (T =

Kelvin) for a variety of PAHs sampled in Osaka (Yamasaki et a l, 1982) gave sufficient linearity to determine the heats of desorption (Hd) [from the slope of the line]. The same plots carried out for PAHs sam pled in Chicago IL air produced similar H d values (Kreiger and Hites, 1994).

The use by Yamasaki et a l (1982) of TSP as a surrogate m easurem ent of the am ount of surface area available for adsorption by vapour phase molecules is therefore acceptable for the Langm uir isotherm approach and allows the use of E quation (2.3). Furtherm ore, the partition coefficient Kp is a strong fu n ctio n of atm o sp h eric tem p eratu re (w hich affects a co m p o u n d 's volatility) and w hen Kp is plotted vs. 1/T then the linear regression takes the form (Pankow, 1987)

- Log Kp = m /T + b (2.4)

W here m and b are the slope of the line and the Y- intercept respectively. These are effectively thermodynamic expressions of the partitioning process.

Several studies have found strong correlations betw een log (F /T S P )/A and 1 /T for a range of SOCs. T is taken as the ambient tem perature during the sam pling event. Hoff et al. (1992b) examined the partitioning for a variety of pesticides and found that there was increased sorption to particulate m atter w ith decreasing tem p eratu re for cis-chlordane, y-H C H , 4,4'-DDT and e n d o s u lp h a n (the slope of the line d e p en d in g on the com pound). Interestingly, no apparent correlation was noted for a-H C H and heptachlor, in d icatin g th at these com pounds rem ain w holly in the v ap o u r state at am bient tem peratures.

Pankow (1987) predicted that the value of m and b obtained by the linear regression over some am bient tem perature range will be given by:-

m = H d / 2 . 3 0 3 R - T a m b / 4 . 6 0 6 ( 2 .5 )

b = log ( A t s p t o / 2 7 5 [ M / T a m b P - 5 ) + 1 / 4 . 6 0 6 ( 2 .6 )

W here H d is the heat of desorption (KJ mol-1) from a surface, R is the gas constant and Tamb (K) is the centre of the ambient tem perature range for the regression. For b (the Y-intercept) Atsp is the surface area of the total

su sp en d ed particulate (cm2 cnrr3), t0 is the molecular vibration time (s) and

M is the m olecular w eight (g m ol-1). The slope and the intercept are therefore strongly dependent upon H d and the surface area of the particulate

(so rb en t) A t s p respectively. H eats of d eso rp tio n (from atm o sp h eric particulate surfaces) have been calculated from plots of log (F /T S P )/A vs. 1 /T for a variety of SOCs, including PAHs (Yamasaki et al., 1982, Kreiger and Hites, 1994), organochlorine pesticides (Cotham and Bidleman, 1992; Hoff et al., 1992b) and PCBs (Bidleman et a l, 1986; Hoff et a l, 1992b).

The therm odynam ic approach was utilised by Storey and Pankow (1991), w ho studied the partitioning of PAHs to several m odel aerosols (graphitic carbon, sodium chloride, silica and alum ina) as well as stan d ard urban particulate m atter (UPM). Good agreem ent between log K p ,s vs. 1/T plots for the graphitic carbon aerosol and UPM supports the theory that partitioning to atm ospheric aerosol is adsorptive and non-specific in nature. K p ,s was a

surface area corrected K p , derived for the different aerosols from Equation (2.6). A greem ent betw een UPM and the other three sorbents, however, was n o t as good as w ith graphitic carbon, indicating that different aerosol com positions m ay affect the vapour to particulate partitioning.

Pankow (1991) im proved the simple linear regression (SLR) plots (Log Kp vs. 1/T ) for individual com pounds, to derive a common factor for a whole com pound class such as the PAHs. This was done on the basis that similar co m p o u n d s sorbing to the same particulate m atter should possess very sim ilar Y -intercepts [b value, E quation (2.4)]. This com m on y-intercept reg re ssio n (CYIR) uses a m ean b value, calculated from in d iv id u a l com pounds, to plot the regression. For a class of com pounds such as the

PA Hs, the use of a com m on b value (or Y intercept) has p ro v ed m ore reliable in evaluating heats of desorption. These in tu rn are m ore highly c o rre lated w ith the heats of v ap o u risatio n d e riv e d from lab o rato ry experim ents.