Artificial receptor-facilitated solid-phase microextraction of barbiturates

Article abstract of DOI:10.1021/ac980587o

A receptor for barbiturates, N,N′-Bis-[6-(2-ethylhexanoylamino)-pyridin-2-yl]-isophthalamide, was designed to dissolve in plasticizers of poly(vinyl chloride) (PVC). Microextractions using receptor-doped films of PVC were carried out as a function of rece

Full text of DOI:10.1021/ac980587o

Anal. Chem. 1999, 71, 2146-2151
Art ific ia l Re c e p t o r-Fa c ilit a t e d S o lid -P h a s e
Mic ro e x t ra c t io n o f Ba rb it u ra t e s
Shu Li, Lifa ng Sun, Yongs oon Chung, a nd Ste phe n G. We be r*
Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260
6
A receptor for barbiturates, N,N′-Bis-[6 -(2 -ethylhexa-
noylamino)-pyridin-2 -yl]-isophthalamide, was designed to
dissolve in plasticizers of poly(vinyl chloride) (P VC).
Microextractions using receptor-doped films of P VC were
carried out as a function of receptor concentration. The
effect of the concentration of the receptor on extraction
yield is considerable for barbiturates that have significant
binding to the receptor but negligible for very similar
molecules that do not bind to the receptor strongly. Thus,
it is the receptor’s ability in molecular recognition, not
its generic ability as an H-bonding cosolvent, that is
important. On the other hand, NMR data show that the
receptor self-associates. A simple, approximate analysis
is given to extract the amount of active receptor from the
data. Receptor-enhanced extractions of barbiturates from
urine are compared to extractions using a phosphate ester
as solvent.
of other chemical species in the sample. One way to achieve
7
selectivity is to use biological elements, such as antibodies, in
8
9
extractions. Membranes and solid-phase materials with immu-
noaffinity have been applied successfully to the analysis of complex
matrixes, such as environmental10,11 or biological fluids.12 The
advantage of using antibodies is that high binding constants with
selectivity to the target molecules can probably be achieved.
Another route to selectivity is the use of synthetic receptors. In
contrast to antibodies, synthetic receptors are small and simple
molecules. The advantage of using synthetic receptors is that they
are usually robust and predictable because of their simple
structure. They may also be available in large quantities for low
cost. Artificial receptors that gain their specificity and binding
strength from hydrophobic effects work well in water, whereas
those that function based on hydrogen bonding work well in the
organic phase. Here we report on an approach with a synthetic
receptor1
3-15
for the extraction of barbiturates into organic phases.
To avoid saturation of the receptor population and to increase the
extraction efficiency, the concentration of receptor sites in the
extraction medium needs to be high. On the other hand, to
achieve selectivity in an extraction requires the use of poor
solvents. By using poor solvents, the extraction of unwanted
species by virtue of their solubility in the solvent is minimized.
Thus, the general problem exists to develop receptors that can
function in, or for that matter even dissolve in, very poor solvents.
Receptor 1 a16 N,N′-bis-[6-(butyrylamino)-pyridin-2-yl]-isophthala-
mide (Figure 1) is effective in chloroform17 but fails in plasticized
PVC because it is not very soluble.18 In the current work, we use
solid-phase microextraction (SPME)19,20 using plasticized poly-
In the quest for more information from increasingly complex
1
samples, such as those generated by library or mixture synthesis,
2
3
those from the environment, or from biological samples, enor-
mous attention has been paid to increasing the resolution of
separations and to increasing the sensitivity and specificity of
detectors. At the same time very little attention has been paid to
improvements in sample preparation. Microextraction is an
extraction method in which both the quantity of consumed sample
and produced extract are small. It is especially designed for
subsequent analysis by methods such as gas chromatography,4
5
and capillary electrophoresis that have low mass detection limits.
We report the first microextraction carried out with an artificial
receptor. Furthermore, the extraction procedure is “green” as it
uses no volatile organic solvents.
(
6) Giddings, J. C. Unified Separation Science; John Wiley & Son, Inc.; New
York, 1991; p 8.
7) Hennion, M.-C.; Barcelo, D. Anal. Chim. Acta 1 9 9 8 , 362, 3-34.
(
If an extraction medium is selective, the extraction medium
should at equilibrium contain relatively more of the analyte than
(8) Dombrowski, T. R.; Wilson, G. S.; Thurman, E. M. Anal. Chem. 1 9 9 8 , 70,
1969-1978.
(
9) Pichon, V.; Rogniaux, H.; Fischer-Durand, N.; Rejeb, S. B.; Le Goffic, F.;
(
1) For example, recent papers are: Booth, R. J.; Hodges, J. C. J. Am. Chem.
Soc. 1 9 9 7 , 119, 4882-4886; Marx, M. A.; Grillot, A.-L.; Louer, C. T.; Beaver,
K. A.; Bartlett, P. A. J. Am. Chem. Soc. 1 9 9 7 , 119, 6153-6167, Flynn, D. J.;
Crich, J. Z.; Devraj, R. V.; Hockerman, S. L.; Parlow, J. J.; South, M. S.;
Woodard, S. J. Am. Chem. Soc. 1 9 9 7 , 119, 4874-4881; Studer, A.; Hadida,
S.; Ferrito, R.; Kim, S.-Y.; Jeger, P.; Wipf, P.; Curran, D. P. Science 1 9 9 7 ,
Hennion, M.-C. Chromatographia 1 9 9 7 , 45, 289-295.
(10) Ferrer, I.; Hennion, M.-C.; Barcel o´ , D. Anal. Chem. 1 9 9 7 , 69, 4508-4514.
(11) Pichon, V.; Chen, L.; Hennion, M.-C.; Daniel, R.; Martel, A.; Le Goffic, F.;
Abian, J.; Barcelo, D. Anal. Chem. 1 9 9 5 , 67, 2451-2460.
(12) Hennion, M.-C.; Cai, J. Y. Anal. Chem. 1 9 9 6 , 68, 72-78.
(13) B u¨ hlmann, P.; Badertscher, M.; Simon, W. Tetrahedron 1 9 9 3 , 49, 595-
599.
2
75, 823-826; Dunayevskiy, Y. M.; Vouros, P.; Wintner, E. A.; Shipps, G.
W.; Carell, T.; Rebek, J. Proc. Natl. Acad. Sci. U.S.A. 1 9 9 6 , 93, 6152-6157.
2) Clement, R. E.; Yang, P. W.; Koester, C. J. Anal. Chem. 1 9 9 7 , 69, 251R-
(14) Linton, B.; Hamilton, A. D. Chem. Rev. 1 9 9 7 , 97, 1669-1680.
(15) Rebek, J. J. J. Mol. Recognit. 1 9 9 2 , 5, 83-88.
(16) Chang, S. K.; Hamilton, A. D. J. Am. Chem. Soc. 1 9 8 8 , 110, 1318-1319 .
(17) Valenta, J. N.; Dixon, R. P.; Hamilton, A. D.; Weber, S. G. Anal. Chem. 19 94,
66, 2397-2403.
(18) Valenta, J. N.; Weber, S. G. J. Chromatogr. 1 9 9 6 , 722, 47-57.
(19) Zhang, Z.; Yang, M. J.; Pawliszyn, J. Anal. Chem. 1 9 9 4 , 66, 844A-853A.
(20) Eisert, R. and Pawliszyn, J. Crit. Rev. Anal. Chem. 1 9 9 7 , 27, 103-135.
(
(
2
87R.
3) Anderson, D. J.; Guo, B.; Xu, Y.; Ng, L. M.; Kricka, L. J.; Skogerboe, K. J.;
Hage, D. S.; Schoeff, L.; Wang, J.; Sokoll, L. J.; Chan, D. W.; Ward, K. M.;
Davis, K. A. Anal. Chem. 1 9 9 7 , 69, 165R-229R.
4) Pawliszyn, J.; Liu, S. Anal. Chem. 1 9 8 7 , 59, 1475-1478.
5) Li, S.; Weber, S. G. Anal. Chem. 1 9 9 7 , 69, 1217-1222.
(
(
2146 Analytical Chemistry, Vol. 71, No. 11, June 1, 1999
10.1021/ac980587o CCC: $18.00 © 1999 American Chemical Society
Published on Web 04/21/1999

COCH(CH
as measured by FAB-MS. The NMR titration data are obtained
by titrating 0.5 mL of receptor CDCl solution (40% w/ w) with
CDCl . The chemical shifts of the two pairs of amide NH protons
of the receptor were monitored as the concentration changed. The
association constant for 1 b and phenobarbital in CHCl was
2 3 2 2 2 3
CH )CH CH CH CH ). The molecular weight is 600
3
3
3
determined by UV spectrophotometry according to Connors.22
Standard solutions were prepared by adjusting an aqueous
solution of 10 barbiturates with 20 mM phosphate to pH 7.5.
Spiked urine was prepared by mixing the standard solution of
barbiturates with urine (1:9 or 1:49 v/ v). The pH of the spiked
urine was adjusted with sodium acetate and acetic acid to pH 5.5.
Spiked serum was prepared by injecting into a vial containing
lyophilized bovine serum, 25 µL of anhydrous acetic acid and 1
mL of the barbiturate standard solution, and bringing the total
volume with D. I. water to 5 mL as designated by the serum
producer. The pH of spiked serum is about 4.6.
The procedures for SPME device preparation, extraction, back
extraction, and analysis of the extract by CE have been described.5
Briefly, a stainless steel rod (o.d. 1.1 mm) was dip-coated to form
membranes of dioctylphthalate- or chloroparaffin-plasticized PVC,
either with or without receptor 1 b after poly(vinyl chloride-co-
vinyl acetate-co-maleic acid) was coated as the primer. Extractions
were carried out by immersing the membrane-coated rod for 5
min in 3.5 mL of quiescent solutions of standard or spiked sample.
Back-extraction was carried out by inserting the extraction rod
into a Teflon tube containing 5 µL of phosphate (20 mM, pH 11.5)
and allowing it to sit for 15 min. A portion of the back-extraction
solution was vacuum injected for 4 s to a CE instrument (Isco
Fig u re 1 . Receptors and barbiturates. Receptor 1a: R ) n-propyl,
receptor 1b: R ) 3-heptyl, receptor 1c, 1d: macrocyclic receptors.
Mephobarbital: X ) O,Y ) CH3; thiopental: X ) S,Y ) H; Other
barbiturates: X ) O, Y ) H.
3
850 capillary electropherograph, Isco Inc., Lincoln, NE). The
(
vinyl chloride) (PVC) as the nonvolatile, reuseable extraction
separation column was a 75 µm i.d. fused silica capillary (Poly-
micro Technologies) with total length of 75 cm, 50 cm to the
window. The separation potential was 25 kV. The separation buffer
was 50 mM Tris adjusted with Tapso to pH 7.8.
medium. We demonstrate that a receptor can be compatible with
plasticizers that are poor solvents. Furthermore, the H-bonding-
based molecular receptor acts as a specific receptor rather than
as a polar cosolvent. Self-association of the receptor in the poor
solvent occurs and decreases its efficacy.
The volume of the extracting phase was estimated by the
determination of the dioctyl phthalate in solution following
redissolution of the deposited film. A rod coated with a dioct-
ylphthalate-plasticized membrane was soaked in 3.5 mL of THF
while ultrasonicating for 30 min. The concentration of dio-
ctylphthalate in the solution was determined by CE (15 mM total
borate, 50 mM SDS, pH 9.0) with UV detection. The density of
the plasticized PVC membrane was estimated to be 1.0 g/ mL.
The concentration of the receptor in the membrane was calculated
from the concentration of PVC, plasticizer, and receptor in the
coating solution and the density of membrane.
EXPERIMENTAL SECTION
All of the chemicals, unless specified otherwise, were obtained
from Aldrich (Milwaukee, WI) or Sigma (St. Louis, MO). Chlo-
roparaffin was obtained from Fluka (Ronkonkoma, NY). Santicizer
1
41 is a gift from Monsanto (St. Louis, MO).
Preparation of 1 b, N,N′-bis-[6-(2-ethylhexanoylamino)-pyridin-
-yl]-isophthalamide (Figure 1), adapted from Hamilton’s proce-
2
dure,21 is as follows. The diamine resulting from condensation of
equiv of 2,6-diaminopyridine and isophthaloyldichloride was
2
incubated for 8 h at room temperature with racemic 2-ethylhex-
anoyl chloride. The crude product was purified by chromatography
RESULTS AND DISCUSSION
Replacing the butyryl groups in 1 a with 2-ethylhexanoyl
groups, 1 b, is effective in increasing the solubility of the receptor.
The solubility of receptor 1 a in CHCl3 is about 4 mM, in
dioctylphthalate it is 1.2 mM, and in chloroparaffin it is about 300
µM.18 1 b (a mixture of diastereomers) is highly soluble in
chloroform: solutions up to 40 wt % have been prepared with no
evidence of precipitation. We have not seen evidence of its
precipitation either in dioctylphthalate or chloroparaffin mem-
branes at compositions of tens of weight percent. 1 b has a very
2 2
on silica gel using 5% THF in CH Cl as the eluent. After solvent
evaporation, 1 b was obtained as a pale yellow solid, mp 50-52
1
°
C. H NMR (17.5 mM in CDCl
3
) δ 8.59 (2H, s, isophth-CONH),
8
.49 (1H, br s, isophth-2H), 8.12 (2H, d, pyr-3H), 8.06 (2H, d,
isophth-4,6H), 8.04 (2H, d, pyr-5H), 7.88 (2H, s, CH CONH), 7.78
2H, t, pyr-4H), 7.64 (1H, t, isophth-5H), 2.19 (2H, m, COCH),
.73 (4H, m, COCH(CH CH )CH ), 1.57 (4H, m, COCH(CH CH )-
CH CH ), 1.33 (4H, m, COCH(CH CH )CH CH CH ), 1.33 (4H,
m, COCHCH CH ), 0.98 (6H, t, COCHCH CH ), 0.93 (6H, t,
2
(
1
2
3
2
2
3
2
2
2
3
2
2
2
2
3
2
3
(
21) Chang, S. K.; Engen, D. V.; Fan, E.; Hamilton, A. D. J. Am. Chem. Soc. 1 9 9 1 ,
13, 7640-7645 .
(22) Connors, K. A. Binding Constants, The Measurement of Molecular Complexes
Stability; John Wiley and Sons: New York, 1987; pp 147-157.
1
Analytical Chemistry, Vol. 71, No. 11, June 1, 1999 2147

Ta b le 1 . Co n c e n t ra t io n Fa c t o rs (p f) o f Un d o p e d
Me m b ra n e s a n d t h e Ma x im u m No rm a lize d
P re c o n c e n t ra t io n Fa c t o rs (n p f) o f Re c e p t o r-Do p e d
Me m b ra n e s in Dio c t ylp h t h a la t e a n d Ch lo ro p a ra ffin
pf
npf
a
c
b
c
CP ( sc DOP ( sc
CP ( s
DOP ( s
butabarbital
butalbital
amobarbital
0.083 ( 0.004 0.36 ( 0. 02
0.046 ( 0.002 0.31 ( 0.02
0.162 ( 0.008 0.81 ( 0.03
50 ( 7
47 ( 5
42 ( 8
25 ( 2
19 ( 1
16 ( 1
29 ( 2
15 ( 1
26 ( 1
11 ( 1
23 ( 1
phenobarbital 0.028 ( 0.001 0.159 ( 0.007 42 ( 4
pentobarbital 0.21 ( 0.01 0.91 ( 0.03 40 ( 6
0.049 ( 0.002 0.181 ( 0.008 42 ( 2
0.29 ( 0.01 1.45 ( 0.07 31 ( 5
aprobarbital
secobarbital
allobarbital
thiopental
0.030 ( 0.002 0.093 ( 0.003 19 ( 4
1.48 ( 0.05
6.6 ( 0.4
2.7 ( 0.2
2.5 ( 0.8 1.1 ( 0.8
1.6 ( 0.3 2.0 ( 0.2
mephobarbital 1.17 ( 0.06
a
CP, chloroparaffin. b DOP, dioctylphthalate. c s, standard deviation.
Fig u re 2 . Extraction of phenobarbital (X), secobarbital (]), me-
phobarbitral (O), and thiopental (4) with DOP membranes doped with
various concentrations of receptor.
low melting point around 50 °C. Thus, it is possible, without
sacrificing the ease of synthesis, to create a barbiturate receptor
that is compatible with nonpolar solvents. It functions as a
receptor. The formation constant for the formation of the complex
with phenobarbital is 3.3 × 10 M in chloroform. Therefore, we
have applied it to SPME.
4
-1
We will define the term “preconcentration factor” (pf) as the
ratio of concentration in the solution injected into the analytical
instrument, in this case capillary electrophoresis, to that in the
original solution. We will define the term “normalized preconcen-
tration factor” (npf) as the ratio of the pf in a receptor-enhanced
extraction to that without the receptor. As will be shown below,
npf’s are receptor-concentration dependent and display a maxi-
mum. Table 1 shows values of pf and maximum npf values for
the barbiturate solutes. The data show that chloroparaffin- and
dioctylphthalate-plasticized PVC matrixes are poor solvents,
reflecting the behavior of the pure plasticizers.18,23 However, the
barbiturates can all be extracted to some degree into the receptor-
free solvents. There is modest selectivity displayed by such a
system based on hydrophobicity.
Fig u re 3 . Extraction of phenobarbital (X), secobarbital (]), me-
phobarbital (O) and thiopental (4) with CP membranes doped with
various concentrations of receptor.
The effectiveness of the receptor depends on the solvent.23 The
The values of npf are all greater than unity; thus the receptor-
doped systems are all more efficient than solvent alone, although
to various degrees. The least affected (minimum npf) barbiturates
are thiopental and mephobarbital. These results are consistent
with NMR binding studies on analogous macrocyclic receptors,
chloroparaffin-based membrane displays pf values that are lower
and npf values that are higher than those in dioctylphthalate. It is
expected that the lower the polarity of the solvent, the more the
partitioning of the polar drug will be dependent on the formation
of a complex. The data in Table 1 are in agreement with that
expectation.
The final question concerns self-association of the receptor.
Self-association is more prevalent at high concentrations than low
concentrations; thus, the determination of the npf as a function
of the receptor concentration should alert us to the extent of self-
association. Figures 2 and 3 plot the npf for dioctylphthalate and
chloroparaffin membranes for four barbiturates against the
concentration of receptor in membrane. The data for the other
six barbiturates tested are not shown on the graph but generally
fall in the area between phenobarbital and secobarbital. It can be
seen that, within the range of 0-0.1 M receptor, the extraction
efficiency increases approximately linearly with the concentration
of receptor in the extraction membrane. Indeed, it can be shown
that this is expected under some reasonable assumptions. We will
assume that the receptor is in excess so that the distribution of
1
c or 1 d; the binding constants for thiopental and mephobarbital
16,21
K (M-1) for phenobarbital, 2.0 × 10 ; barbital,
5
are relatively low
6
5
2
2
.0 × 10 ; mephobarbital, 6.8 × 10 ; and thiobarbital, 7.4 × 10 .
The marked difference in values of npf that reflect qualitatively
the differences in the measured binding constants allows us to
comment on the nature of the receptor’s activity. 1 b could act as
a general H-bonding cosolvent. All of the compounds in the study
can act as both hydrogen bond donors and acceptors. Thiopental
and mephobarbital are unique in that they do not have the native
malonylurea ring. If 1 b were acting as a simple cosolvent, one
would not expect the values of npf to be so sensitive to molecular
structure. Therefore, it is clear that 1 b is acting primarily as a
molecular receptor, not as a generic HBA/ HBD cosolvent.
(
23) Valenta, J. N.; Sun, L.; Ren, Y.; Weber, S. G., Anal. Chem. 1 9 9 7 , 69, 3490-
495.
3
2148 Analytical Chemistry, Vol. 71, No. 11, June 1, 1999

in the system. To speak in at least semiquantitative terms to the
issues just raised, a semiquantitative treatment of the NMR data
is required to determine to what degree the receptor is associated
at each concentration. This question is not answered easily and
accurately by considering all of the possible equilibrium reactions,
since the number of different “products” is vast. However, a simple
but practically useful model can be derived through reasonable
approximations. Each receptor has four H-bond donating func-
tional groups (HBD, n
a
) 4), and six H-bond accepting functional
groups (HBA, n ) 6). Each HBA group is assumed to react one-
b
to-one with an HBD group. Equilibrium is assumed to be achieved
for the (HBA + HBD) reaction with a self-association constant K,
where HA denotes all of the unbound HBD sites, B denotes all of
the unbound HBA sites, B:HA denotes all of the bound sites.
Fig u re 4 . The experimental and calculated chemical shifts of H1
square, left axis) and H2 (circle, right axis). Calculated values are
shown as the solid line.
(
HA + B ) B:HA
K ) (B:HA)/ (HA)(B)
There are nominally three types of basic sites and two types
of acidic sites in the receptor. There are therefore six different
equilibria that exist. This makes for a difficult analysis. We will
assume that the acidic sites all have the same strength and so do
the basic sites. It is known that the H-bonding accepting ability
for pyridine and carbonyl groups in general,24 and for amide
carbonyl groups in particular25-27, are similar. Thus, we are
justified in assuming that the bases are the same. For the acid
sites, for which we have the NMR data, we can calculate a value
of the constant K for each set of data and compare the values of
K to see how good our assumption is.
the solute does not depend on the solute concentration. We
estimate the maximum solute concentration in the membrane
during extraction to be a few mM, whereas the receptor concen-
2
tration is on the order of 10 mM. Thus, excess receptor is a good
assumption. We know that the membranes do not reach equilib-
rium in the extraction time allowed;5 however, the primary
determinant of the extractability of these molecules of very similar
molecular weight is the driving force based on the free energy
difference of the solute in the feed phase and the extracting phase.
Our previous data show that under the conditions used here, the
extraction yield is about 1/
that obtained at equilibrium. We make
3
The bound fraction of HA, Pbound, is thus expressed in terms
the approximation that the system is in equilibrium so that the
mathematical treatment is simple.
of K, the self-association constant, n
receptor, n , the number of HBA per receptor, and C
concentration of receptor.
a
, the number of HBD per
b
r
, the molar
The equilibrium value of npf is a function of, K
constant, [R org, the total concentration of the receptor in the
organic phase, D , distribution coefficient, and Φ, the phase ratio
as shown in eq 1.
f
, the formation
t
]
c
Pbound
)
2
(
1/ (KC ) + n + n ) ( (1/ (KC ) + n + n ) - 4n n
r
a
b
x
r
a
b
a
b
1
+ D Φ
c
npf ) (1 + K [R ]org)
(1)
f
t
2na
1
+ D Φ(1 + K [R ]org)
c t t
(
3)
(∼1) and the phase ratio (∼3 × 10-4)
18
Since the product of D
c
is small in our case, npf can be approximated as eq 2.
Given Pbound < 1, and n
a
) 4, n
b
) 6 for this receptor
npf ≈ 1 + K [R ]
(2)
f
t org
2
1
KCr
1
Pbound
)
+ 1.25 -
+ 1.25 - 1.5 (4)
(
x
)
8
8KCr
Equation 2 shows that the slopes in Figures 2 and 3 are binding
constants. Thus, the small slopes of mephobarbital and thiopental
shown in Figures 2 and 3 are the direct result of small binding
constants of those solutes to the receptor. The slopes for the other
The chemical shift of each of the amide protons can be
expressed as a concentration weighted average of the bound and
free chemical shifts as shown below.
2
3
drugs are in the range 10 -10 , which is lower than one would
18,21
expect,
even taking into account the factor of 3 due to the lack
of achieving equilibrium. Furthermore, in contradiction to the
prediction of eq 2, the extraction efficiency no longer increases
beyond about 0.1 M. It is likely that both the low estimate of the
binding constant and the shapes of Figures 2 and 3 are explained
by self-association of the receptor.
δ ) δboundPbound + δfree(1 - Pbound
)
(5)
(24) Jeffrey, G. A. An Introduction to Hydrogen Bonding; Oxford University
Press: New York, 1997; p 56-78.
25) Huyskens, P. L., Luck, W. A. P., Zeegers-Huyskens, T., Eds. Intermolecular
Forces An Introduction to Modern Methods and Results; Springer-Verlag:
New York, 1991; p 23.
(
1
To test this, H NMR was performed on solutions of 1 b in
(
(
26) Kamlet, M. J.; Abboud, J_L. M.; Abraham, M. H.; Taft, R. W. J. Org. Chem.
chloroform-d. Figure 4 shows the chemical shifts of the two
distinct amide protons as points. The increase in the chemical
shifts with concentration indicates that self-association does exist
1
9 8 3 , 48, 2877-2887.
27) Li, J.; Zhang, Y.; Ouyang, H.; Carr, P. W. J. Am. Chem. Soc. 1 9 9 2 , 114,
9813-9828.
Analytical Chemistry, Vol. 71, No. 11, June 1, 1999 2149

Ta b le 2 . Ca lc u la t e d Bin d in g Co n s t a n t s a n d Ch e m ic a l
S h ift s
H1
H2
K (M- ) ( s
1
a
0.48 ( 0.05
8.92 ( 0.02
0.79 ( 0.06
9.92 ( 0.06
a
δbound ( s
a
s: standard deviation.
Fig u re 6 . Extraction of barbiturates from spiked urine. Peak
assignments: (1-5) pento-, buta-, seco-, amo-, apro-barbital; (7)
butalbital; (9) thiopental; (10) phenobarbital. Membranes a, c, d:
plasticized with DOP and doped with receptor (90 mM); (b) plasticized
with phosphate ester and without receptor. Barbiturate concentrations
(a), (b) about 0.5 µM for each barbiturate except mephobarbital which
is 0.2 µM, and thiopental which is 1 µM; (c) barbiturate concentrations
are about 5 times those in (a) and (b); (d) blank urine.
Fig u re 5 . The concentration of receptor with all binding sites free.
Thus eqs 4 and 5 can be combined to yield an equation for
the observed chemical shift that depends on four parameters,
namely, C
shift at the lowest receptor concentration yields δfree, and C
known. Then, nonlinear regression of the observed δ on C can
r
, K, δbound, and δfree. The measured value of the chemical
r
is
r
be used to determine values of K and δbound. Calculated values of
K and δbound from the two sets of chemical shifts are listed in Table
2
. Calculated chemical shifts are shown as solid lines in Figure 4.
The fits are excellent. The small difference in the self-association
constants K justifies the assumption that there is not a large
difference in the HBD strength between the two amide proton
donors. To be concrete, the implied difference in H-bond strengths
Fig u re 7 . Extraction of barbiturates from spiked bovine serum.
Comparison of extraction using plasticized membranes (a) DOP
doped with receptor (90 mM) and (b) phosphate ester with no
receptor. (1-7) pentobarbital, butabarbital, secobarbital, amobarbital,
aprobarbital, mephobarbital, butalbital; (9) thiopental. About 4.4 µM
for each barbiturate except mephobarbital, which is 1.3 µM, and
thiopental, which is 8.8 µM.
is RT ln (K
H
2
/ K
H
1
) ), or ∼0.5RT.
The foregoing calculations can be used to determine the
amount of receptor that is available for binding to barbiturate.
We will assume that there is a statistical mixture of receptors,
some of which are H-bonded to neighbors and others that are
not. We will further assume that if a receptor is not involved in
HB donation, and if its pyridine groups are not involved in HB
accepting (i.e., all of the six H-bond sites involved in barbiturate
binding are free), then that molecule can accept a barbiturate as
a guest. Recall that we have already made the assumption that
the receptor is in excess. This allows us to ignore the shift in the
self-association equilibrium brought about by the presence of the
barbiturate.
above. This fraction, f, is given by eq 6.
4
2
2
f ) (1 - PHBD) (1 - / PHBD)
(6)
3
It should be noted that both binomial coefficients are unity;
therefore, they do not appear explicitly in eq 6.
Given K as an average of the equilibrium constants from Table
2
, eq 4 can be used to calculate the average probability of finding
Figure 5. shows calculated concentrations of the free receptor.
First, although a quantitative comparison is not possible because
the solvents are different in the extraction and NMR experiments,
there is a remarkable similarity between the npf curves (Figures
2 and 3) and the curve in Figure 5. Second, self-association can
be a serious problem. At the high concentration end, near 0.6 M,
only about 1% of the receptor is “free”. We conclude that self-
association exists, indeed is prevalent.
a bound state of HBD (Pbound in eq 4 is denoted as PHBD in the
following to differentiate from the probability of bound state for
HBA). Since each HBD must be paired with an HBA, the
probability that an HBA is occupied is just 2/
PHBD because each
3
molecule has six HBA sites for the four HBD sites that it has.
The binomial distribution can then be used to calculate the fraction
of the receptor population that satisfies the criteria mentioned
2150 Analytical Chemistry, Vol. 71, No. 11, June 1, 1999

Examples of the effectiveness of the receptor-doped SPME
rods are shown in Figure 6 and Figure 7 which compare the
dioctylphthalate/ receptor-based system to extractions based on
a phosphate ester, which is a good solvent for phenobarbital.5,18,23
Electropherograms from spiked urine are shown in Figure 6, and
those from spiked bovine serum are shown in Figure 7. Obviously,
in these two cases the signal-to-noise ratios are improved when
the receptor is added to the membrane extractor. In the urine
chromatograms there are two peaks near 7.3 and 8.3 min whose
concentration increases due to the receptor’s presence. We do
not know what those peaks are, but their behavior shows that
there may be compounds in complex samples that cannot be
discriminated against with a receptor-doped solvent.
extraction. The advantage of molecular selectivity comes from
selectively preconcentrating analytes. We cannot say with confi-
dence that the molecular selectivity is absolute, however. The
extraction devices still contain a solvent, and although it is poor,
it can still dissolve solutes that do not bind to the receptor. There
may be other compounds that interact with the barbiturate
receptor. Nonetheless, if the complexity of a sample can be
reduced before its entry into a separations/ detection system, then
lower detection limits and/ or higher speed analysis will result.
ACKNOWLEDGMENT
We thank the NSF for financial support through Grant CHE-
9
710213.
CONCLUSION
The concept of using a doped, poor, nonvolatile solvent for
Received for review May 27, 1998. Accepted March 10,
999.
the extraction of organics from aqueous samples is feasible. There
are many advantages to this. There are no volatile organic solvents
in the extraction, and there is some molecular selectivity in the
1
AC980587O
Analytical Chemistry, Vol. 71, No. 11, June 1, 1999 2151