CAN A SOLUTE’S AQUEOUS SOLUBILITY OR OCTANOL–WATER PARTITION COEFFICIENT BE USED TO PREDICT RP-SPE PERCENT RECOVERIES?

230



Trace Environmental Quantitative Analysis, Second Edition



Phenol



log Kow



Phenol

o-Methyl phenol

o-Chlorophenol

2,6-Dimethyl phenol

p-Chlorophenol

2,4-Dichlorophenol

p-Trimethyl silyl phenol



1.46

1.95

2.15

2.36

2.39

3.08

3.84



Clearly, the effect of either chloro or methyl substitution serves to significantly

increase the octanol–water partition coefficient. This increase in hydrophobicity is

an important consideration in developing an RP-SPE method that is designed to

isolate one or more phenols from a groundwater sample. Those substituted phenols

with the greater values for KOW would be expected to yield the higher percent

recoveries in RP-SPE. The substitution of a trimethyl silyl moiety in place of

hydrogen significantly increases the hydrophobicity of the derivatized molecules.

This use of a chemical derivative to convert a polar molecule to a relatively nonpolar

one has important implications for TEQA and will be discussed in Chapter 4.



59. WHAT IS AQUEOUS SOLUBILITY ANYWAY?

The amount of solute that will dissolve in a fixed volume of pure water is defined

as that solute’s aqueous solubility, Saq. It is a most important concept with respect

to TEQA because there is such a wide range of values for Saq among organic

compounds. Organic compounds possessing low aqueous solubilities generally have

high octanol–water partition coefficients. This leads to a tendency for these compounds to bioaccumulate. The degree to which an organic compound will dissolve

in water depends not only on the degree of intramolecular vs. intermolecular interactions, as we saw earlier, but also on the physical state of the substance (i.e., solid,

liquid, or gas). The solubility of gases in water is covered by Henry’s law principles,

as discussed earlier.

A solute dissolved in a solvent has an activity a. This activity is defined as the

ratio of the solute’s fugacity (or tendency to escape) in the dissolved solution to that

in its pure state. The activity coefficient for the ith solute, γ iw , relates solute activity

in pure water to the concentration of the solute in water, where the concentration is

defined in units of mole fraction, xw, according to

i

xw =



moles i



∑ moles



i



γ iw ≡



i

aw

i

xw



i



We usually think of aqueous solubilities in terms of the number of moles of the

ith solute per liter of solution. Alternative units commonly used to express aqueous

© 2006 by Taylor & Francis Group, LLC



Sample Preparation Techniques



231



solubility include milligrams per liter or parts per million and also micromoles per

liter. The concentration in moles per liter can be found by first considering the

following definition:

Ci =



xi (moles/mole total)

Vmix ( L /mole total)



Vw is the molar volume of water (0.018 L/mol) and Vi is the molar volume of a

typical solute dissolved. The molar volume of the mixture, Vmix, is found by adding

the molar volume contributions of the solute and solvent, water, according to

Vmix,w = 0.2 xi,w + 0.018 xw

This expression can be further simplified to

Vmix,w = 0.182 xi,w + 0.018

Because the mole fraction for solutes dissolved in water is very low, xi,w < 0.002,

which corresponds to about 0.1 M, the molar volume for the mix can be approximated

at 0.018, and hence the above equation for the concentration in moles per liter can

be expressed as

sat

Cw =



sat

xw

x sat

≅ w ( moles/L )

Vmix,w 0.018



The above equation converts the mole fraction of a solute dissolved in water,

sat

up to saturation conditions, xw , to the corresponding concentration of that solute in

units of moles per liter. By knowing the molecular weight of a particular solute, it

is straightforward to obtain the corresponding concentration in milligrams per liter.

For hydrophobic solutes, γw is greater than 1. When enough solute has been

added to a fixed volume of water until no more can be dissolved, the solution is

said to be saturated and a two-phase system results. This is a condition of dynamic

equilibrium whereby the solute activity in both phases is equal. Denoting o as the

organic phase, we have

aw = ao

so that

xw γ w = xo γ o

For immiscible liquids in water, the mole fraction of solute in itself and its

activity coefficient approximate 1, so that

xw γ w = 1



© 2006 by Taylor & Francis Group, LLC



232



Trace Environmental Quantitative Analysis, Second Edition



and the mole fraction, xw , is related to the aqueous activity coefficient by

xw =



1

γw



and in terms of logarithms,

log xw = − log γ w

The purpose of deriving this is to show that attempts to measure the aqueous

solubility, xw , are, in essence, attempts to estimate the activity coefficient, γw . This

relationship applies only to liquid solutes. Solid or crystalline solutes require an

additional term where

log xw = log



xc

− log γ w

x sel



where xc/xsel represents the ratio of the solubility of the crystal to that of the hypothetical supercooled liquid.90 The additional term above is approximated by the term

–0.00989 (Tmp – 25) for rigid molecules at 25°C and –[0.01 + 0.002(n – 5)]Tmp for

long-chain molecules, with n monomers per polymer and Tmp the melting point of

the solid. Octanol–water partition coefficients can also be estimated from solute

retention times using, for example, reversed-phase HPLC.

Aqueous solubility for various solutes can be estimated by applying any of the

techniques in the following four categories:91

1. Methods based on experimental physico-chemical properties, such as

partition coefficient, chromatographic retention, boiling point, and molecular volume.

2. Methods based directly on group contributions to measured activity

coefficients.

3. Methods based on theoretical calculations from molecular structures,

including molecular surface area, molecular connectivity, and parachor.

4. Methods based on combinations of two or more parameters that can be

experimentally measured, calculated, or generated empirically. These

include the solubility parameter method and the UNIFAC technique (a

method based on linear solvation energy relationships and on the use of

multivariate statistical methods).

However, for many of the very hydrophobic solutes of interest to TEQA, the

aqueous solubility is extremely low. For example, iso-octane, a nonpolar hydrocarbon,

will dissolve in water up to 0.0002% at 25˚C.92 Higher-order aliphatic hydrocarbons

have such low values for their aqueous solubilities that it is impossible to directly

measure this physico-chemical property.



© 2006 by Taylor & Francis Group, LLC



Sample Preparation Techniques



233



60. WHAT DOES A PLOT OF KOW VS. AQUEOUS

SOLUBILITY TELL US?



Absorbance



Figure 3.24 is a qualitative plot of the logarithm of the octanol–water partition

coefficient, KOW , against the logarithm of the aqueous solubility, Saq , of a selected

number of organic solutes of environmental interest. This qualitative plot makes it

very evident that polar solutes with appreciable aqueous solubility exhibit low values

of KOW . Hydrophobic solutes that are not soluble in water to any great degree exhibit



Volume of

water passed

through SPE

column

VrDMP



VrDBP



FIGURE 3.24 Plot of the logorithm of the octanol–water partition coefficient vs. the logarithm

of the aqueous solubility for selected organic compounds of enviro-chem/enviro-health interest.

(Adapted from Chiou, G. et al., Environmental Science and Technology, 11(5): 475–478, 1977.)



© 2006 by Taylor & Francis Group, LLC



234



Trace Environmental Quantitative Analysis, Second Edition



TABLE 3.10

Degree of Correlation between the Octanol–Water

Partition Coefficient and Aqueous Solubility for Four

Classes of Organic Compounds

Class



log KOW



Aromatic hydrocarbons

Unsaturated hydrocarbons

Halogenated hydrocarbons

Normal hydrocarbons



0.786–1.056

0.250–0.908

0.323–0.907

0.467–0.972



log

log

log

log



Correlation Coefficient

Saq

Saq

Saq

Saq



0.995

0.993

0.993

0.999



large values of KOW . The importance of knowing KOW values as this pertains to the

practice of TEQA and to environmental science in general has been stated earlier:

In recent years the octanol/water partition coefficient has become a key parameter in

studies of the environmental fate of organic chemicals. It has been found to be related

to water solubility, soil/sediment adsorption coefficients, and bioconcentration factors

for aquatic life. Because of its increasing use in the estimation of these other properties,

KOW is considered a required property in studies of new or problematic chemicals.93



Much effort has been expended in finding a linear equation that relates

octanol–water partition coefficients for any and all solutes to the aqueous solubility

in octanol-saturated water. Two useful equations have emerged in which the logarithm of the aqueous solubility, log Saq, is, in general, linearly related to log KOW:90

For liquid solutes:

log K OW = 0.8 − log Saq



(3.37)



log K OW = 0.8 − log Saq − 0.01(Tmp − 25)



(3.38)



For crystalline solutes:



Yalkowsky and Banerjee90 have reviewed specific studies and summarized the

relationship in Equations (3.37) and (3.38) as applied to liquid organic compounds

and crystalline organic solids. The results are shown in Table 3.10 for four classes

of liquid hydrocarbons. A plot of log KOW against log Saq reveals a slope close to the

value of –1 for each of these four classes of hydrocarbons. The differences in the y

intercept reflect differences in the activity coefficients as defined in Equation (3.36).



61. CAN WE PREDICT VALUES FOR KOW FROM ONLY

A KNOWLEDGE OF MOLECULAR STRUCTURE?

Yes, and the predictions are quite good. Two different methods emerge from a host

of others and are most commonly used to predict octanol–water partition coefficients

for the many organic compounds that exist. One approach is to calculate KOW from

© 2006 by Taylor & Francis Group, LLC



Sample Preparation Techniques



235



a knowledge of structural constants, whereas the second approach requires that a

chemical’s partition coefficient be measured between a solvent other than octanol

and water, KSW . KOW can then be calculated from linear regression equations that

relate log KSW (for a particular solvent) and log KOW. Two forms of the structural

constant approach are most popular: (1) the Hansch π hydrophobic character of

substitutents approach and (2) the Leo fragment constant approach. The Hansch π

approach is based on the assignment of a value for πX as the difference between

octanol–water partition coefficients for a substituted vs. unsubstituted or parent

compound. Mathematically, this difference can be stated as follows:

X

H

π X = log K OW − log K OW



For example, the log KOW for chlorobenzene is 2.84, whereas that for benzene

is 2.13, and thus

π X = 2.84 − 2.13 = 0.71

and the πX for the substituent chlorine on the monoaromatic ring is 0.71. The Hansch

π approach has been recently criticized because it ignores hydrogens attached to

carbon, and this led to some erroneous values for a few aliphatic hydrocarbons.

The Leo fragment method has emerged as a more powerful way to predict log

KOW. The working mathematical relationship is

log K OW =



∑ f + ∑F

i



i



(3.39)



j



j



where the logarithm of the octanol–water coefficient for a particular organic compound is found by algebraically summing the contributions due to the ith structural

component (f ) (building block or functional group), overall components, and the

algebraic sum of the jth factor (F) due to intramolecular interactions caused largely

by geometric or electronic effects. For complex molecules, it is desirable to have a

known log KOW value for a given compound and use the fragment approach to add

and subtract structural and geometric/electrical contributions such that

log K OW (unknown ) = log K OW ( known )

−



∑



removed



f +



∑ f − ∑ F +∑F



added



removed



(3.40)



added



Thus, two ways emerge to calculate log KOW using the Leo fragment method.

One is to build the entire molecule from fragments and factors using Equation (3.40),

and the other is to calculate log KOW from structurally related compounds. To

illustrate how one goes about calculating the log KOW for a specific organic compound, we return to TCMX. We begin by first using Equation (3.39) to establish the

log KOW for benzene by using previously published fragment factors. For carbon

© 2006 by Taylor & Francis Group, LLC



236



Trace Environmental Quantitative Analysis, Second Edition



that is aromatic, a fragment factor of 0.23 has been established. Hence, because

benzene consists of six aromatic carbons that in turn are bonded to six hydrogens,

log K OW (C 6 H 6 ) = 6 fc (aromatic ) + 6 f hφ

= 6 (0.13) + 6 (0.23)



(3.41)



= 2.16

This prediction is quite close to the well-established and observed value for

benzene, which is 2.13. Using this established value for benzene, we begin to view

our molecule of interest, TCMX, as a substituted benzene. Theoretically, we can

arrive at TCMX by removing all six aromatic hydrogens and then adding two methyl

groups and four chlorine groups. Stated mathematically,

φ

φ

φ

log K OW ( TCMX) = log K OW (C6H6 ) − 6 f H + 2 fCH3 + 4 fCl



(3.42)



The methyl groups are viewed as being composed of an aromatic carbon and

three aliphatic bonded hydrogens and a fragment constant. Equation (3.42) is modified as follows:

φ

φ

log K OW ( TCMX ) = log K OW (C6H6 ) − 6 f H + 2  fCφ + 3 f H  + 4 fCl







(3.43)



Upon substituting the values for the various fragment constants into Equation

(3.43), we get

log K OW ( TCMX ) = 2.13 − 6 (0.23) + 2[(0.13) + 3(0.23)] + 4(0.19)

= 6.15

Such a large value for the octanol–water partition coefficient for TCMX suggests

that the compound exhibits a very low aqueous solubility. Note that the substitution

by hydrogen by methyl and chloro groups significantly increases the hydrophobicity

of the molecules. The addition of a methyl group is seen to add about 0.5 of a log

unit, irrespective of the compound involved. For example, substituting a methyl for

hydrogen in benzene contributes the same, as if a methyl were substituted for

hydrogen in cyclohexane.

As long as the building blocks lead to a complete molecular structure, only

structural fragment constants are necessary, as in the TCMX example. However, for

molecules that are more complex, the molecular components begin to exert significant influences on one another through steric, electronic, and resonance intramolecular interactions. These interactions affect both the aqueous and octanol activity

coefficients, as ability of groups to rotate about a carbon–carbon single bond,

chair/boat confirmation in cyclohexane, and carbon-chain branching in aliphatic

structures. Geometric effects increase aqueous solubility, decrease KOW, and hence

their F factor contributes negativity to Equation (3.40). Electronic effects due to

© 2006 by Taylor & Francis Group, LLC



Sample Preparation Techniques



237



electronegativity differences between the elements of both atoms that comprise a

polar covalent bond, such as a carbon–chlorine bond, decrease aqueous solubility,

increasing KOW , and hence their F factor contributes positively to Equation (3.40).

Other electronic effects include nearby polyhalogenation on carbon, nearby polar

groups, and intramolecular hydrogen bonding. These other factors also contribute

in a positive manner to Equation (3.40).



62. ARE VALUES FOR KOW USEFUL TO PREDICT

BREAKTHROUGH IN RP-SPE?

The capacity factor for an RP-SPE cartridge when pure water is used, kW , discussed

′

earlier, has been found to be closely related to KOW and is related by94,95

kW = 0.988 log K OW + 0.02

′

Hence, merely obtaining octanol–water partition coefficient values from various

tabulations in the literature can yield a predictive value for kW because we have

′

shown that kW is directly related to the breakthrough volume, Vr .

′



63. WHERE CAN ONE OBTAIN KOW VALUES?

The resources cited are available in most university libraries in order to find these

values, as well as to obtain good introductions. Leo et al.96 published a comprehensive listing and followed this with a more systematic presentation.97 Even earlier,

Hansch et al.98 published their findings. Lyman et al.99 have published a handbook

on the broad area of physico-chemical property estimations. These references’

sources also provide detailed procedures for calculating and estimating KOW.



64. CAN BREAKTHROUGH VOLUMES BE DETERMINED

MORE PRECISELY?

A more precise way to determine the breakthrough volume for a particular analyte

on a given SPE sorbent, Vr , is to use reversed-phase HPLC. The SPE sorbent is

efficiently packed into a conventional HPLC column. The analyte of interest is injected

into the instrument, while the mobile-phase composition is varied. This is accomplished by varying the percent organic modifier, such as acetonitrile or methanol,

and measuring the differences in analyte retention time. Extrapolation of a plot of

k′ vs. percent organic modifier to zero percent modifier yields a value for the capacity

factor for that analyte at 100% water. This capacity factor is represented by

kW ,SPE,HPLC , where the subscript refers to 100% water. The breakthrough volume is

′

related to the capacity factor in 100% water according to

Vr = V0 (1 + kW ,HPLC )

′

V0 is the void volume of the SPE column and kW (SPE, HPLC) is the capacity

′

factor of the analyte eluted by water. This capacity factor should theoretically be

© 2006 by Taylor & Francis Group, LLC



238



Trace Environmental Quantitative Analysis, Second Edition



FIGURE 3.25 Plot of log k′HPLC vs. the percent MeOH in a MeOH-water mobile phase.



the same for any technique used, as V0 is found by knowing the porosity of the

sorbent and the geometric of the SPE column or sorbent bed in the cartridge. The

capacity factor of a given analyte, k′ (SPE, HPLC), is generally obtained from HPLC.

From an HPLC chromatogram, k′ (SPE, HPLC) is obtained by taking the ratio of

the difference in the retention time between an analyte and its void retention time

to the void retention time. Expressed mathematically,

kHPLC =

′



tR − t0

′

′

t0

′



where tR represents the retention time for the analyte of interest under the HPLC

′

chromatographic conditions of a fixed mobile-phase composition. t0 represents the

′

retention time for an analyte that is not retained on the stationary phase. t0 also

′

relates to the void volume in the column.

Log kW ,SPE,HPLC is obtained by extrapolation of a plot of log k′ vs. percent MeOH

′

to zero. This plot is shown in Figure 3.25. A linear relationship exists between the

logorithm of the capacity factor, k′ (SPE, HPLC), and the percent MeOH concentration for a specific organic compound in HPLC. Because kW ,SPE,HPLC has also been

′

shown to be related to the octanol–water partition coefficient, KOW, kW ,SPE,HPLC , and

′

hence Vr , can be obtained for a given RP-SPE sorbent by this approach.



65. HOW DOES SPE RELATE TO TEQA?

Because environmental matrices are largely air, soil, and water of some sort, the

objective in TEQA is to isolate and recover a hydrophobic organic substance from



© 2006 by Taylor & Francis Group, LLC



Sample Preparation Techniques



239



a more polar sample matrix. RP-SPE can directly remove organic contaminants and

represents the dominant mode of SPE most relevant to TEQA. Normal-phase SPE

(NP-SPE) also has a role to play in TEQA. NP-SPE has served as an important

cleanup step following LLE using a nonpolar solvent to remove polar organic

interferences. This was accomplished in large-diameter glass columns instead of the

familiar barrel type of SPE cartridge. Hydrophobic organic contaminants found in

the environment can be easily transferred from the native matrix to the RP-SPE

sorbent using procedures that will be discussed below.

Since 1985, the analytical literature has been replete with articles that purport

good to excellent recoveries of various organic compounds that have been isolated

and recovered by applying RP-SPE techniques. The rise in popularity of RP-SPE

between 1985 and the present has served to limit the growth of both LLE and

supercritical fluid extraction (SFE). Hydrophobic organic contaminants are partitioned from the environmental matrix to the surface layer of the organo moiety. The

contaminants are then eluted off of the sorbent using a solvent with the appropriate

polarity. This eluent is then introduced into the appropriate determinative technique,

such as a gas or liquid chromatograph.

Reversed-phase SPE is applied principally in two areas of TEQA. The first

application is in the determination of organochlorine pesticides (OCs) in connection

with the determinative technique of GC using a chlorine-selective detector. The

second application is in the determination of priority pollutant semivolatile organics

that are recovered from drinking water. Offline RP-SPE coupled to a determinative

technique such as GC using an element-selective detector is a very powerful combination with which to achieve the objectives of TEQA. To illustrate, this author’s

study of various organophosphorous pesticides (OPs) from spiked water using

RP-SPE will now be discussed.

Organophosphorous pesticides are used primarily as insecticides, and in mammals,

they act as cholinesterase inhibitors. A number of OPs are listed as priority pollutants.

EPA Method 8141A from SW-846, Revision 1, September 1994, is an analytical method

that uses LLE coupled to GC with a phosphorous-selective detector and a megaborecapillary column to separate and quantitate some 49 individual OPs. The method is

highly detailed and points out numerous challenges to be encountered during implementation. OPs are all structurally related in that a phosphate, thiophosphate, or dithiophosphate has been esterified. Either methyl or ethyl esters occupy two of the three

functional groups around central pentavalent phosphorous, whereas a complex organo

moiety is esterified to the third functional group. The generalized structure is as follows:

O

P



O



O

O



R



Generalized molecular structure

for organophosphorous pesticides



© 2006 by Taylor & Francis Group, LLC



240



Trace Environmental Quantitative Analysis, Second Edition



This complexity gives rise to a wide range of compound polarities. These

polarities are largely governed by the nature of this third esterified functional group.

The multisample nature of the conventional SPE vacuum manifold enabled enough

replicate extractions to permit a statistical evaluation to be accomplished. A multiinjection procedure using a liquid GC autosampler enabled enough replicate injections

to be made as well. These data were plentiful enough to enable equations first introduced in Chapter 2 to be used. The following flowchart describes part of a multisample/multi-injection study using two dissimilar GC megabore-capillary GC columns

while spiked samples are passed through a conditioned octyl-bonded silica, C8:

Spike 200 µL of 10 ppm

each OP into



70 mL of distilled,

deionized water to

simulate drinking

water



Pass the spiked

sample through a

conditioned C8

sorbent



Elute with

4 mL MTBE

and divide

eluent into

three GC

vials



DB

608



DB-5



DB608



DB-5



DB608



DB-5



Three % recoveries for each OP is obtained on the DB-608 GC column

while three % recoveries for each OP is obtained on the DB-5 GC column

for each C8 bonded sorbent SPE cartridge used



© 2006 by Taylor & Francis Group, LLC