of conductivities (0 to 23 mmho/cm) between the
where K is close to the octanol water partition co-
matrices. These results, however, are consistent
efficient (Kow) for many analytes (Zhang and Pawl-
with those observed for BTEX where salt concen-
iszyn 1993a, Louch et al. 1992, Dean et al. 1996).
trations below 1% did not significantly affect re-
For BTEX compounds, the ratios of Kow/K ranged
sults (Arthur et al. 1992c). However, salt satura-
from 0.9 to 2.7 (Zhang and Pawliszyn 1993a). Kow
tion has been used successfully to enhance re-
and KH have been measured for white phospho-
sponse for BTEX by headspace SPME (MacGilliv-
rus (Spanggord et al. 1985). The estimates are Kow
= 1200 and KH = 2100 atm-cm3/mole. Using these
ray et al. 1994).
The mean mass of white phosphorus detected
estimates for Kow and KH, the predicted mass of
by headspace SPME for the combined data set
white phosphorus sorbed to the SPME phase was
(Table 5) is 35.4 pg. To prepare each 25-mL aque-
ous standard at 0.144 g/L, 3600 pg of white phos-
ous solution and the two size vials used in these
phorus was added. Therefore, the mass removed
studies (Table 6).
by headspace SPME was 0.98% of the initial mass.
Table 6. Predicted mass (pg) of white
These data provide further confirmation that head-
phosphorus sorbed to the SPME
space SPME may be performed on the same sam-
phase during headspace SPME based
ple prior to solvent extraction without biasing the
on model by Zhang and Pawliszyn
concentration estimates obtained by solvent
(1993) and estimates of Kow and KH
by Spanggord et al. 1985.
extraction. For water samples, sequential SPME
solvent extraction of the same aliquot would be
Size of
Volume (mL)
Theoretical
appropriate for samples with white phosphorus
vial
Aqueous
mass sorbed
concentrations greater than 0.1 g/L. For water
(mL)
Headspace
solution
(pg)
samples with lower concentrations, a larger ali-
40
1
39
104
quot (500 mL) of water is needed to provide ade-
5
35
102
10
30
100
quate preconcentration using ether extraction.
15
25
97
Results using these various extraction options will
20
20
93
be discussed later in relation to field samples.
25
15
87
30
10
77
35
5
57
Model predictions
120
1
119
105
Zhang and Pawliszyn (1993a) derived the fol-
10
110
104
lowing equation for the mass of analyte sorbed
20
100
103
30
90
102
40
80
100
space SPME of water samples:
50
70
98
60
60
95
K 1K 2C0V1V2
n=
70
50
92
K 1K 2V1 + K 2V3 + V2
80
40
87
90
30
80
100
20
69
where
K1 = SPME phase/gas partition coef-
110
10
49
ficient
Predictions are shown for two sizes of extrac-
K2 = gas/water partition coefficient
tion vials used for these studies and various
C0 = initial concentration of the anal-
volumes of headspace. Numbers in bold cor-
yte in the aqueous solution
respond to volumes used in this study.
V1, V2, and V3 = volumes of the SPME phase,
K1K 2C0V1V2
n=
= mass of analyte
aqueous solution and the head-
K1K 2V1 + K 2V3 + V2
space.
sorbed by the SPME phase at equilibrium
during headspace SPME.
The gas/water partition coefficient (K2) is derived
V1= volume of SPME phase = 6.12 104
from the Henry's Law constant (KH)
mL*
V2= volume of aqueous solution
KH
V3= volume of headspace
K2 =
.
C0= 0.144 g/L
RT
K1= SPME phase/gas partition coefficient
K2= gas/water partition coefficient = KH/
The SPME phase/gas partition coefficient (K1) can
RT = 0.09
be estimated from the relationship
K1K2 ≈ KOW = 1200.
K = K1K2
* Personal communication, Supelco, 1996.
7