Covalently Bonded Polymeric Zwitterionic Stationary Phase for Simultaneous Separation of Inorganic Cations and Anions

Article abstract of DOI:10.1021/ac9804083

A novel polymer-based zwitterionic separation material was synthesized by a two-step reaction, in which 2-hydroxyethyl methacrylate-ethylene dimethacrylate copolymer beads were first activated with epichlorohydrin, whereafter (2-dimethylamino)ethanesulfonic acid inner salt was coupled to the epoxide groups on the activated beads in a quaternizing reaction. The resulting material carried strong/strong charge zwitterionic pendant groups, whose charge properties did not change over a wide pH range. It was capable of separating inorganic anions and cations both independently and simultaneously using aqueous solutions of perchloric acid or perchlorate salts as eluent It was found that the retention of cations and anions depended on the individual ions in the analyte, not on type of salt injected, as reported for ODS columns dynamically coated with zwitterionic detergents using water only as eluent. This study also revealed that both the concentration and the water structure related properties of the eluent ions were important for the retention of ionic species. The mechanism appears to involve the interaction of the chaotropic anion component with the quaternary ammonium group, resulting in an increasing cation-exchange capacity of the zwitterion up to a perchloric add eluent concentration of c.a. 10 mM. Above this concentration the material appears to become saturated, and the retention times of both cations and anions become practically independent of the HClO4 concentration.

Full text of DOI:10.1021/ac9804083

Anal. Chem. 1999, 71, 333-344
Covalently Bonded Polymeric Zwitterionic
Stationary Phase for Simultaneous Separation of
Inorganic Cations and Anions
Wen Jiang and Knut Irgum*
Department of Analytical Chemistry, Umeå University, S-901 87 Umeå, Sweden
single eluent. A particularly interesting concept is the separation
mode termed zwitterion-pair chromatography, which was intro-
duced by Knox and Jurand in the early 1980s.8 Their technique
makes use of long-chain amino acids as zwitterion-pair agents,
which adsorb onto the stationary phase by hydrophobic interaction
and form quadrupolar ion pairs with zwitterionic compounds in
the mobile phase. Optical isomers,8b nucleotides,8c and several
other biochemical compounds8d have been separated using this
principle.
However, despite its potential application, this zwitterion-pair
chromatography generally requires extensive equilibration of the
system and suffers from the relatively poor reproducibility that is
often seen for separation modes based on dynamic modification.
Covalently bonded stationary phases are thus preferred instead
of dynamically coated surfaces, especially if the mobile phase
contains an organic solvent. Yu et al.9,10 have consequently
described chemistries for covalent surface modification of silica,
to obtain both weak or strong zwitterionic functionalities. These
bonded zwitterionic silica stationary phases have been shown to
possess a high separation power and are capable of simultaneously
separating organic anions and cations.
More recently, Hu et al. utilized zwitterionic detergents to
dynamically coat an ODS reversed-phase column. Very good
separations of inorganic anions and cations were achieved, using
pure water as eluent.7,11 When the ODS column was coated with
a bile salt to form a weak/ strong zwitterion stationary phase, the
separation of several anions and cations was achieved. However,
simultaneous separation of halide anions and divalent cations has
not been shown.6 Meanwhile, several other zwitterionic separation
media capable of separating organic or inorganic substances have
also been described in the literature,12,13 all being based on the
matrix of silica.
A novel polym er -ba sed zwitterionic separation material
was synthesized by a two-step reaction, in which 2 -hy-
droxyethyl methacrylate-ethylene dimethacrylate copoly-
mer beads were first activated with epichlorohydrin,
whereafter (2 -dimethylamino)ethanesulfonic acid inner
salt was coupled to the epoxide groups on the activated
beads in a quaternizing reaction. The resulting material
carried strong/ strong charge zwitterionic pendant groups,
whose charge properties did not change over a wide pH
range. It was capable of separating inorganic anions and
cations both independently and simultaneously using
aqueous solutions of perchloric acid or perchlorate salts
as eluent. It was found that the retention of cations and
anions depended on the individual ions in the analyte,
not on type of salt injected, as reported for ODS columns
dynamically coated with zwitterionic detergents using
water only as eluent. This study also revealed that both
the concentration and the water structure related proper-
ties of the eluent ions were important for the retention of
ionic species. The mechanism appears to involve the
interaction of the chaotropic anion component with the
quaternary ammonium group, resulting in an increasing
cation-exchange capacity of the zwitterion up to a per-
chloric acid eluent concentration of ∼1 0 mM. Above this
concentration the material appears to become saturated,
and the retention times of both cations and anions become
practically independent of the HClO4 concentration.
Several techniques such as connected cation- and anion-
exchange columns, mixed-bed ion-exchange columns, zwitterionic
micelle-coated columns, ion-exclusion chromatography mixed with
cation-exchange chromatography, and complexing techniques1-7
have been applied in order to separate inorganic anions and
cations simultaneously in a single chromatographic run with a
(8) (a) Knox, J. H.; Jurand, J. J. Chromatogr. 1 9 8 1 , 203, 85-92. (b) Knox, J.
H.; Jurand, J. J. Chromatogr. 1 9 8 2 , 234, 222-224. (c) Knox, J. H.; Jurand,
J. J. Chromatogr. 1 9 8 1 , 218, 341-354. (d) Knox, J. H.; Jurand, J. J.
Chromatogr. 1 9 8 1 , 218, 355-363.
(9) Yu, L. W.; Floyd, T. R.; Hartwick, R. A. J. Chromatogr. Sci. 1 9 8 6 , 24, 177-
182.
(10) Yu, L. W.; Hartwick, R. A. J. Chromatogr. Sci. 1 9 8 9 , 27, 176-185.
(11) (a) Hu, W. Z.; Takeuchi, T.; Haraguchi, H. Anal. Chem. 1 9 9 3 , 65, 2204-
2208. (b) Hu W.; Haddad P. R. Trends Anal. Chem. 1 9 9 8 , 17, 73-79.
(12) (a) Nesterenko, P. N.; Elefterov, A. I.; Tarasenko, D. A.; Shpigun, O. A. J.
Chromatogr., A 1 9 9 5 , 706, 59-68. (b) Elefterov, A. I.; Kolpachnikova, M.
G.; Nesterenko, P. N.; Shpigun, O. A. J. Chromatogr., A 1 9 9 7 , 769, 179-
188.
* Corresponding author: (e-mail) kim@chem.umu.se.
(1) Small, H.; Miller, T. E. Anal. Chem. 1 9 8 2 , 54, 462-469.
(2) Yamamoto, M.; Yamamoto, H.; Yamamoto, Y.; Matsushito, S.; Baba, N.;
Ikushigo, T. Anal. Chem. 1 9 8 4 , 56, 832-834.
(3) Pietrzyk, D. J.; Brown, D. M. Anal. Chem. 1 9 8 6 , 58, 2554-2557.
(4) Tanaka, K.; Ohta, K.; Fritz, J. S.; Matsushita, S.; Miyanaga, A. J. Chromatogr.,
A 1 9 9 4 , 671, 239-248.
(5) Padarauskas, A.; Schwedt, G. Fresenius’ J. Anal. Chem. 1 9 9 5 , 351, 708-
713.
(6) Hu, W. Z.; Haraguchi, H. Anal. Chem. 1 9 9 4 , 66, 765-767.
(7) Hu, W. Z.; Tao, H.; Tominaga, M.; Miyazaki, I.; Haraguchi, H. Anal. Chim.
Acta 1 9 9 4 , 299, 249-256.
(13) Chou, T. Y.; Yang, M. H. J. Liq. Chromatogr., Relat. Technol. 1 9 9 6 , 19,
2985-2996.
10.1021/ac9804083 CCC: $18.00 © 1999 American Chemical Society
Published on Web 12/12/1998
Analytical Chemistry, Vol. 71, No. 2, January 15, 1999 333

A problem with silica is the limited operational range 2 < pH
< 7. At high pH silica is hydrolyzed, while at low pH the bonded
phase is lost, also through a hydrolytic process.14a Bulky and
hydrophobic ligands protect the material from hydrolysis,15 most
likely because they disturb the hydrogen-bonding pattern in the
vicinity of the surface or prevent water from reaching the
underlying silica structure altogether. Highly hydrophilic materials
are consequently not protected by this mechanism, making
functional silica with ionic groups attached through short-chain
alkyl spacers vulnerable to hydrolysis.
The aim of this study was consequently to develop and evaluate
a polymer-based separation medium with permanent strong/ strong
zwitterionic character. This paper thus describes the functional-
ization of cross-linked 2-hydroxyethyl methacrylate polymer
particles with covalently bonded zwitterionic groups, yielding
materials with fixed interaction sites at the surface. Their
chromatographic properties were evaluated by independently and
simultaneously separating inorganic anions and cations using a
single column under isocratic conditions, and both spectroscopic
and sorptive characteristics were measured.
Midland, MI) sulfonic acid strong cation exchanger in the H+ form
at a flow rate of ∼2 mL/ min. The purified solution was precipitated
twice from boiling water/ ethanol and dried at 50 °C for 24 h in a
1
vacuum oven. The purity of the DMAES was determined by H
NMR (400 MHz, Bruker) using D2O as solvent (δ 2.3 (2H, CH2);
2.5 (2H, CH2); 1.9 ppm (6H, (CH3)2N)).
Activation of 2-Hydroxyethyl Methacrylate P olymer Beads
with Epichlorohydrin. A 5-g sample of 12 µm Spheron 300 cross-
linked 2-hydroxyethyl methacrylate polymer beads (Chemapol,
Brno, Czech Republic) was suspended in 30 mL of 50% aqueous
NaOH in a 100-mL ground-glass-neck flask and stirred for ∼1 h
at room temperature until a uniform suspension was obtained.
The resulting suspension was kept for 18 h below 10 °C,
whereafter 10 mL of dioxane was added to the flask under slow
stirring for 30 min at room temperature. A mixture of 25 mL of
epichlorohydrin and 15 mL of dioxane was then filled into the
flask, and the activation was allowed to take place under slow
stirring for 2 h at 40 °C, then for an additional 2 h at 60 °C. The
resulting particles were filtered on a glass filter and washed to
neutral conditions with a large quantity of Milli-Q water, methanol
(3 × 100 mL), and acetone (3 × 100 mL) and finally dried for 18
h at 40 °C in a vacuum oven.
EXPERIMENTAL SECTION
Reagents and Solutions. Dimethylamine (40% in water),
WARNING: Epichlorohydrin and 2,3-epoxypropyl methacrylate
are suspected carcinogens and should be handled with due care.
Functionalization of Epichlorohydrin-Activated Hydroxy-
ethyl Methacrylate (HEMA) P articles. DMAES (2 g; 0.013 mol)
was dissolved in 20 mL of aqueous (0.2 M) phosphate buffer (pH
8) in a 50-mL glass tube. The mixture was thereafter adjusted to
pH 8 with 5 M NaOH under stirring. A 2-g sample of the activated
polymer beads was added to the solution under slow stirring and
reacted at 50 °C for 90 h. The reacted beads were thereafter
washed with water, methanol, and acetone on a glass filter under
weak suction and dried at 50 °C for 18 h in a vacuum oven. This
was the final procedure; experiments at different reaction tem-
peratures and pH were conducted in the similar way.
epichlorohydrin (98%), 2-bromoethanesulfonic acid sodium salt
(97%), 2,2′-azobisisobutyronitrile (AIBN; 95%), ethylene glycol
dimethacrylate (EDMA; 90%), 1-dodecanol (99.5%), and potassium
iodide (p.a.) were purchased from Fluka (Buchs, Switzerland).
Cyclohexanol (99%) and 2,3-epoxypropyl methacrylate (GMA,
glycidyl methacrylate; 97%) were from Aldrich (Steinheim, Ger-
many), while dioxane, acetone, sodium perchlorate, sodium
chloride, sodium sulfate, calcium chloride, and potassium nitrate
were of analytical grade and obtained from Merck (Darmstadt,
Germany). Ethanol (99.7%) was from Solveco Chemical AB,
Stockholm and perchloric acid (70% in water) from Riedel-de-Hae¨n
(Seelze, Germany). Sodium hydroxide solution (50% w/ w; p.a.),
sodium bromide (p.a.), and methanol (HPLC grade) were from
J. T. Baker (Deventeer, Holland). Water was purified by a Milli-Q
water purification system (Millipore, Bedford, MA), and always
had a conductivity of less than 0.1 µS/ cm at the tap. Eluents and
stock solutions of sample ions were prepared by dissolving the
salts or acids directly in Milli-Q water.
Synthesis of 2 -(Dimethylamino)ethanesulfonic Acid
(DMAES). 2-Bromoethanesulfonic acid sodium salt (10.88 g; 0.05
mol) was dissolved in 100 mL of water in a 250-mL Erlenmeyer
flask. Dimethylamine (12.40 g; 0.11 mol) was added to the above
solution, and the mixture was allowed to stand for 45 min at room
temperature and then reacted at 70-80 °C for 18 h under refluxing
conditions. After the solution was cooled to 40 °C, ∼2 g of
granulated charcoal (Hopkin & Williams, Essex, England) was
added, and the mixture was boiled for 15 min without a refluxing
condenser. The mixture was cooled, the charcoal was allowed to
settle, and the supernatant solution was thereafter filtered (What-
man GF/ A) under weak suction. Further purification was carried
out by letting the solution pass through a 150 mm × 40 mm i.d.
glass column packed with Dowex 350C UPN (Dow Chemical,
Characterization of the Zwitterionic Materials. The degree
of functionalization was determined by elementary analysis, using
a Leco (Saint Joseph, MI) CHN-1000 to determine the nitrogen
content, while a Leco SC-432 was used to determine the content
of sulfur. Infrared spectra of the particles during the different
stages of the functionalization procedure were obtained using an
ATI Mattson Genesis Series FT-IR instrument on tablets pressed
from ground, dried polymer and KBr. The FT-Raman spectra were
recorded with a Perkin-Elmer System 2000R near-infrared FT-
Raman spectrometer, which was equipped with a InGaAs detector.
The sampling accessories for solids supplied by Perkin-Elmer were
used. All spectra were recorded in the 90° optical geometry and
with a 500-mW laser power. The final spectra, collected at a
resolution of 4 cm-1, were the average of 16 scans. The 13C CP/
MAS NMR spectra were recorded on a Bruker AMX2-500 NMR
spectrometer operating at a 13C NMR frequency of 125.77 MHz.
The samples were packed into 4-mm ZrO2 rotors equipped with
Kel-F caps. The spinning rate was 8 kHz. A total of 1024 transients
comprising 2K data points were collected with a standard CP/
MAS pulse sequence using a contact time of 5 ms, a spectral width
of 300.04 ppm, and a delay between pulse sequences of 5.0 s. The
FIDs were zero-filled to 4K real points, and a 20-Hz exponential
line broadening was applied prior to Fourier transformation. A
(14) (a) Poole, C. F.; Poole, S. K. Chromatography Today; Elsevier: Amsterdam,
1991; pp 323-324. (b) Ibid. pp 371-375.
(15) Sagliano, N.; Floyd, T. R.; Hartwick, R. A.; DiBussolo, J. M.; Miller, N. T. J.
Chromatogr. 1 9 8 8 , 443, 155-172.
334 Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

rotor packed with adamantane was used to set up the Hartman-
Hahn condition and to set the reference frequency of the chemical
shift scale (by using the downfield peak in the adamantane
spectrum at 38.3 ppm). A Metrohm (Herisau, Switzerland) model
632 pH meter was used to monitor the pH conditions.
number of micromoles contained in the sum of the pump latency
and column void volumes. The pump latency was determined as
the volume that had to be pumped for the conductivity signal to
reach 50% of the plateau level of the programmed step change,
with the column disconnected from the system and replaced by
a zero dead volume union. The transition was reasonably sharp,
but had two or three small jags in the transition interval because
the step change had passed through the pump. The column void
volume was determined after all the chromatograms had been
acquired, by weighing the water-filled column before emptying,
the dry column blank, and the dry recovered packing material.
The packing material density used in these calculations was
determined to be 1.13 by pycnometry, using the above-mentioned
values and the measured volume of the column blank.
Evaluation of the functionalization reaction was done on
polymer particles prepared from a monolith precursor mixture
containing 24% (w/ w) GMA, 16% EDMA, 54% cyclohexanol, and
6% 1-dodecanol. To this mixture, which had been deaerated by
purging with helium for 10 min, was added 0.4% (w/ w) AIBN
shortly before polymerization, which took place in sealed glass
tube molds at 60 °C for 16 h. The resulting monoliths were
thereafter removed from the molds by breaking the glass, cut in
cubic pieces of approximately 2-3-mm sides, and then Soxhlet
extracted with methanol for at least 24 h. The extracted monolith
pieces were dried, ground in a mortar, and dry sieved. The fraction
smaller than 200 mesh (74 µm) was used for functionalization
following the procedure described in the paragraph above.
Chromatographic Evaluation. The zwitterionic functionalized
2-hydroxyethyl methacrylate polymer particles were dry packed
in a 150-mm by 4-mm-i.d. column made from poly(ether ether
ketone) (PEEK; Upchurch Scientific, Oak Harbor, WA) using a
glass rod to compact the bed by tapping carefully between
stationary phase additions. The chromatographic system used for
evaluation consisted of an SP8700XR gradient LC pump (Spectra-
Physics, San Jose, CA), a Rheodyne (Berkeley, CA) 7010 sample
injection valve with a 20-µL stainless steel injection loop, and an
LDC (Laboratory Data Control, Riviera Beach, FL) Conducto-
Monitor electrolytic conductivity detector with a Conductocell
7011. Attempts to measure the pH of the eluate in experiments
using water only as the eluent were done with a Metrohm
combination pH microelectrode (model 10 406 3043) mounted in
a specially designed conical cell machined from poly(methylpen-
tene) (TPX) that directed the eluate flow axially toward the
electrode membrane, provided a narrow measurement channel,
and had an exit above the reference electrode frit. Chromatograms
were recorded on an SS 250F strip-chart recorder (Sekonic). All
chromatographic evaluations were carried out at room tempera-
ture (22 ( 2 °C).
Frontal Chromatography. The column used for chromato-
graphic evaluation was subsequently used in frontal chromatog-
raphy experiments to determine the dynamic capacity for per-
chlorate and chloride ions at a flow rate of 0.5 mL/ min. The
column was first thoroughly washed with deionized water until a
steady baseline was obtained, whereafter the pump was pro-
grammed to deliver a step gradient to 2 mM hydrochloric acid,
sodium perchlorate, or sodium chloride or different concentrations
(1-20 mM) of perchloric acid. Conductivity detection was used
to monitor the breakthrough of the acid or salts. When a stable
conductivity level was obtained after the breakthrough, the pump
was programmed to deliver a step gradient back to deionized
water, whereafter the decrease in conductivity was recorded.
Water continued to be pumped until a stable baseline was obtained
before commencing the next experiment. The frontal chromato-
grams presented in Figures 12 and 13 have their conductivities
normalized to the plateau level after breakthrough. A horizontal
offset was applied to correct for total system dead volume by
shifting each recorded curve left for a distance equivalent to the
Sorption Isotherms. Static sorption isotherms were deter-
mined on the material from the column used in the frontal
chromatography experiments. The packing material was sus-
pended in 20 mL of deionized water in a beaker which was placed
on a water bath maintained a temperature of 25 ( 0.1 °C. The
beaker was furnished with a magnetic stir bar and covered with
Parafilm M (American National Can, Chicago, IL) except for a
small hole for the pipet tip, to minimize evaporation of water
during the titration. Electrolytic conductivity in the suspensions
was measured using a platinum black electrode cell, connected
to a Metrohm model 660 conductivity meter operated at 2-kHz
excitation frequency. The cell and instrument assembly had been
calibrated (cell constant determined to be 0.80) with 10 and 100
mM KCl in degassed, deionized water. To the thermostated
suspension was then repeatedly added 100-µL aliquots of 5 mM
perchloric acid using a calibrated autopipet. The conductivity was
traced on a strip-chart recorder, and the values after each addition
were taken from the digital readout when the readings had
stabilized. The suspension was then poured into a glass filter
gooch, washed by slow suction with a large amount of deionized
water (∼1 L applied over 30 min), and the experiment repeated.
Similar experiments were conducted with sodium perchlorate,
hydrochloric acid, and sodium chloride, following extensive
washing between each titration. A titration was also done using
100 mM perchloric acid to determine the saturation capacity. A
total of 70 additions were made with the 5 mM titrands and 50
additions with 100 mM HClO4. Calibration curves for each acid
and salt without suspended packing material were recorded within
the measured conductivity range, as were “blanks” using Separon
S-300 HEMA beads in the same particle range. The linear portion
of both the HEMA blank curves and the titration with perchloric
acid after saturation showed that the presence of suspended
packing material in the conductivity cell decreased the slope of
the conductivity curves to 83.4 and 84.2% of the values obtained
for the corresponding additions to solutions without suspended
material. The concentrations determined from calibration curves
in water were therefore multiplied by 1.19 to calculate the actual
concentrations in solution with suspended material present.
RESULTS AND DISCUSSION
P reparation of the P olymeric Zwitterionic Stationary
P hase. In this study, the Spheron particles, which carry 2-hy-
droxyethyl functional groups, were initially reacted with epichlo-
Analytical Chemistry, Vol. 71, No. 2, January 15, 1999 335

Scheme 1. Representation of the Synthesis Reactions Leading to the Zwitterionic Separation
Material Based on the Cross-Linked 2-Hydroxyethyl Methacrylate Polymer Beads
Table 1. Effect of the pH of the DMAES
rohydrin to introduce a surface coverage of epoxide groups.16-19
The reaction of epoxy groups with nucleophilic functional com-
pounds is a frequently used route in the synthesis of functional
polymers.17-23 In this case, we chose to use the inner salt of
DMAES to couple with epoxide groups in order to obtain a per-
manent zwitterionic functionality, because the quaternization re-
action is known to proceed most efficiently when the tertiary amine
is used as the salt.24 The reactions are outlined in Scheme 1.
As the separation on zwitterionic media is dependent on
simultaneous attractive and repulsive electrostatic interactions
between the zwitterionic ligand and analytes, it is a weaker
interaction than ion exchange.11 A high degree of functionalization
was thus needed to get sufficient retention. It has been shown in
the literature that the efficiency of the epichlorohydrin activation
step is controlled mainly by the concentration of epichlorohydrin
in the reaction mixture, the reaction temperature, and the reaction
time.17,18 We therefore selected a relatively high reaction temper-
ature, a high epichlorohydrin concentration, and a long reaction
time to maximize the degree of activation.
The reaction conditions for coupling of DMAES with epoxide
groups were investigated during the initial parts of this study,
where we were using a copolymer of glycidyl methacrylate and
ethylene dimethacrylate instead of the epichlorohydrin-activated
HEMA particles. In those studies, we found that the reaction
temperature and the pH were the major factors controlling the
degree of functionalization in the reaction between the epoxide
groups and DMAES. The relationship between pH and the degree
of functionalization is presented in Table 1. We found that both
the N and S contents increased with increasing pH in the range
below pH 9, above which they both started to decrease. Maximum
functionalization was thus obtained at pH ∼9. If the incorporation
of zwitterionic functionality had proceeded without side reactions,
we would have expected a 32.06:14.01 ratio between the weight
Functionalization Solution on the Nitrogen and Sulfur
Contents When Beads from Pulverized Porous
Poly(GMA-EDMA) Monoliths Were Reacted at 70 °C for
24 h
a
a
b
c
pH
% NEA
%SEA
%Scalc
%SEA:% Scalc
3.73
4.80
6.43
7.10
8.22
9.15
10.27
10.91
ndd
0.024
0.090
0.81
1.42
1.97
2.51
0.98
0.45
0.13
0.53
0.82
1.14
2.07
1.65
1.71
0.30
1.21
1.88
2.61
4.74
3.78
3.91
0.30
0.67
0.76
0.75
0.53
0.26
0.12
a
% N and % S determined from elemental analysis. b Stoichiometric
% S calculated from NEA according to the molar weight ratio of 32.06:
14.01. c Amount of S incorporated in relation to the stoichiometric ratio.
d None detected.
Table 2. Effect of Reaction Time on the Nitrogen and
Sulfur Contents of Beads from Pulverized Porous
Poly(GMA-EDMA) Monoliths Functionalized with
DMAES at 40 °C and pH 8a
a
a
b
c
reaction time h
% NEA
%SEA
%Scalc
%SEA:% Scalc
20
40
90
0.12
0.16
0.22
0.15
0.37
0.49
0.27
0.37
0.50
0.56
1.00
0.98
a
See Table 1 for legend.
percent S and N found in the resulting materials. However, under
these reaction conditions, we never reached a unit molar ratio;
the measured molar ratio peaked at 0.75:1 only in a narrow pH
range between 6 and 8. A possible explanation for this deviation
from stoichiometry between S and N is that high pH promotes
an elimination of sulfoethylene moieties due to Hoffman degrada-
tion,25 a reaction that is known to affect quaternary ammonium
groups on anion exchangers at elevated temperatures.
(16) Porath, J.; Janson, J. C.; Låås, T., J. Chromatogr. 1 9 7 1 , 60, 167-177.
(17) Turkova´, J.; Bla´ha, J.; Hora´cek, J. M.; Vajcner, J.; Frydrychova´, A.; Coupek,
J. J. Chromatogr. 1 9 8 1 , 295, 165-179.
(18) Matsumoto, I.; Mizuno, Y.; Seno, N. J. Biochem. 1 9 7 9 , 85, 1091-1098.
(19) Ponte´n, E.; Glad, B.; Stigbrand, M.; Sjo¨ gren, A.; Irgum, K. Anal. Chim. Acta
1 9 9 6 , 320, 87-97.
New experiments were therefore carried out at a lower reaction
temperature (40 °C) while keeping pH at 8 and maintaining
otherwise identical reaction conditions. The effect of the coupling
reaction time on the degree of functionalization under these
conditions is shown in Table 2. At this lower temperature, the
percent S determined from the elemental analysis and the percent
S calculated from percent N were nearly identical in the reaction
time interval between 40 and 90 h. From these results it can be
concluded that little or no side reaction was taking place at this
(20) (a) Turkova´, J.; Bla´ha, K.; Malan´ıkova´, M.; Vancurova´, D.; Svec, F.; Kalal, J.
Biochim. Biophys. Acta 1 9 7 8 , 524, 162-169. (b) Hradil, J.; Svec, F. React.
Polym. 1 9 9 0 , 13, 43-53. (c) Tennikova, T. B.; Bleha, M.; Svec, F.;
Allmazova, T. V.; Belenkh, B. G. J. Chromatogr. 1 9 9 1 , 555, 97-107.
(21) (a) Glad, B.; Irgum, K. J. Membr. Sci. 1 9 9 2 , 67, 289-293. (b) Glad, B.;
Irgum, K.; Ho¨ rstedt, P. J. Membr. Sci. 1 9 9 5 , 100, 171-179.
(22) (a) Svec, F.; Fre´chet, J. M. J. Anal. Chem. 1 9 9 2 , 64, 820-822. (b) Svec, F.;
Fre´chet, J. M. J. Chromatogr., A 1 9 9 5 , 702, 89-95. (c) Svec, F.; Fre´chet,
J. M. Bio. Technol. Eng. 1 9 9 6 , 49, 355-363.
(23) Ponte´n, E.; Viklund, C.; Irgum, K.; Bogen, S. T.; Nilsson-Lindgren, Å. Anal.
Chem. 1 9 9 6 , 68, 4389-4396.
(24) Small, H. Ion Chromatography; Plenum: New York, 1989; pp 45-46.
(25) March, J. Advanced Organic Chemistry, 3rd ed.; John Wiley & Son: New
York, 1985; pp 906-908.
336 Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

spectrum for an unfunctionalized HEMA-EDMA sorbent are
presented in Figure 2. After activation with epichlorhydrin, two
new peaks appear at 850 and 910 cm-1 in the FT-IR spectrum,
attributable to the epoxide group. After further reaction with
DMAES, these two peaks disappear, and peaks instead appear at
1039 cm-1 in the FT-IR spectrum and at 1032 cm-1 in the Raman
spectrum. These peaks can be assigned to an alkanesulfonate
group. The valley at 1200 cm-1 in the FT-IR spectrum decreases,
indicating that a smaller peak characteristic of hydrated alkane-
sulfonate ion might be present in this region but obscured between
the intense peaks at 1157 and 1261 cm-1. A twin peak at 1182/
1200 cm-1 is also observed in the Raman spectrum close to this
region. Quaternary ammonium groups show few useful vibrational
spectroscopy features.26 The trimethylammonium groups give rise
to peaks in the area 900-980 cm-1, depending on the nature of
the fourth substituent.27 However, we have not been able to find
vibrational spectral data in the literature on dimethyldialkaneam-
monium groups, least of all sulfoethylbetaine zwitterions, where
vibrational features have been assigned to the methyl groups on
the quaternary ammonium group. A Raman spectrum of a water-
soluble N-3-(sulfopropyl)-N-(methacryloxyethyl)-N,N-dimethylam-
monium betaine polymer in the range 800-1500 cm-1 is presented
by Schultz et al.,28 showing no prominent features apart from a
sulfonic acid peak at 1044 cm-1. The peak at 1001 cm-1 in the
Raman spectrum of the fully functionalized material can therefore
not be unambiguously assigned to the methyl groups on the
quaternary ammonium group. A distinct peak appeared at 3052
Figure 1. Overlaid FT-IR spectra of the unfunctionalized Spheron
300 cross-linked HEMA sorbents (lower), after epichlorohydrin activa-
tion (middle), and after final coupling of DMAES to the activated
particles (upper).
cm-1 in the Raman spectrum with a smaller peak at 3036 cm-1
,
both assigned to the asymmetric methylene groups in the
zwitterion spacer. Signals from methyl and methylene groups
bound to tertiary nitrogen were not observed in the 2760-2830
cm-1 region, so the presence of weak anion-exchange function-
alities could not be established by vibrational spectroscopy.
A final attempt was made to positively verify the existence of
a quaternary ammonium functionality, this time with 13C CP/ MAS
NMR spectroscopy. The spectra obtained from unfunctionalized
Separon S-300 HEMA and the zwitterionic polymer are presented
in Figure 3. The peaks in the functionalized material are broad
and characteristic of a solid material with restricted movement of
the polymeric chains. In (meth)acrylate-based polymers containing
sulfopropylbetaine zwitterions, the methylene carbons in R posi-
tion to the quaternary ammonium groups are reported with shifts
between 65 and 68 ppm,29a,b the methyl groups on the quaternary
ammonium group with shifts of 48 29 or 53.6 ppm,29b and the
methylene groups R to the sulfonic acid with shifts between 50
and 52 ppm.29a,b However, in the CP/ MAS spectrum of our solid
materials, we could not confidently assign any of peaks in this
region because they were broadened along with the quaternary
and â-carbons of the methacrylic polymer backbone.29b Erhmann
et al.29b also failed to see the signals characteristic of the carbons
attached to the nitrogen and sulfur in a strong/ strong zwitterionic
group in a solid-state CP/ MAS spectrum of a 0.88:0.12 copolymer
temperature. However, the reaction at 40 °C for 90 h reached only
20% functionalization compared to the reaction at 70 °C for 24 h.
Although these experiments were carried out on the GMA-
EDMA system, we assume that the reaction conditions for the
spacer-extended and thus less sterically hindered epoxide groups
on the epichlorohydrin-activated HEMA beads would follow a
similar trend and the functionalization be at least equally effective.
Based on the GMA-EDMA experiments, we therefore chose a
relatively low reaction temperature and long reaction time (50 °C
for 90 h) as the final reaction conditions, because the temperature
rather than the reaction time appears to be the critical factor
leading to the loss of a sulfonic acid functionality with a
concomitant production of net charge in the material due to
orphaned tertiary amine sites.25
Elemental analyses of the material produced in this reaction
showed 1.08, 0.94, 1.08, and 1.14% N and 2.11 and 1.91% S,
indicating a S:N stoichiometry in the range 0.85-0.90. Total
zwitterionic functionalization, determined from the percent S, was
thus ∼600 µmol/ g of dry material, translating into a total amount
of ∼500 µmol of zwitterionic functionalities in the separation
columns. Notably, the S:N stoichiometry elemental analysis also
indicates that the material could contain a significant number of
weak anion-exchange sites, which show up when the materials
are titrated with strong acids (see below).
(26) Colthrup, N. B.; Daly, L. H.; Wiberley, S. E. Introduction to Infrared and
Raman Spectroscopy, 3rd ed.; Academic Press: San Diego, 1990.
(27) Hume A. S.; Holland, W. C.; Fry F. Spectrochim. Acta 1 9 6 8 , 24A, 786-8.
(28) Schulz, D. N.; Peiffer, D. G.; Agarwal, P. K.; Larabee, J.; Kaladas, J. J.; Soni,
L.; Handwerker, B.; Garner, R. T. Polymer 1 9 8 6 , 27, 1734-1742.
(29) (a) Kathmann, E. E. L.; DeAnna, D. D.; McCormick, C. L. Macromolecules
1 9 9 4 , 27, 3156-3161. (b) Ehrmann, M.; Galin, J.-C.; Meurer, B. Macro-
molecules 1 9 9 3 , 26, 988-993.
FT-IR spectra of the Spheron 300 cross-linked HEMA sorbents
before and after activation, and after final coupling of the DMAES
to the activated particles, are shown in Figure 1, while a Raman
spectrum for the zwitterionic functionalized material along with a
Analytical Chemistry, Vol. 71, No. 2, January 15, 1999 337

Figure 2. Overlaid FT-Raman spectra of unfunctionalized Separon HEMA S 300 (lower) and Spheron 300 after activation and final coupling
of DMAES to the activated particles (upper).
Figure 4. Separation of mixture containing 1 mM each of NaCl and
Figure 3. Overlaid 13C CP/MAS NMR spectra of unfunctionalized
CaCl2 using pure water as eluent at a flow rate of 0.6 mL/min with
Separon HEMA S 300 (lower) and Spheron 300 after activation and
direct electrolytic conductivity detection.
final coupling of DMAES to the activated particles (upper).
salts of anions ranging in retention times in normal anion
of n-butyl acrylate and dimethyl(3-methacrylamidopropyl)(3-sul-
fopropyl)ammonium betaine,29b which means that zwitterionic
groups may very well be present in the material at the concentra-
tion indicated by the elemental analysis although CP/ MAS NMR
failed to verify their existence.
Attempts To Separate Inorganic Anions and Cations Using
Water Only as Eluent. In our initial attempts at using these
materials for separation, we tried using water only as eluent, to
obtain the detection advantages inherent in the use of an
essentially nonconductive eluent.7,11 An equilibration overnight was
required for the conductivity signal to reach baseline, which
settled slightly above 1 µS/ cm. This should be compared to the
conductivity of ∼0.25 µS/ cm measured by the detector with the
column disconnected. When 1 mM standard solutions of sodium
exchange from fluoride, chloride, nitrate, and iodide to thiocyanate
were injected, they all eluted at the same retention time with nearly
identical peak shape in a conductivity detector, represented by
the first peak in Figure 4. Notably, all sodium salts showed small
dips in conductivity shortly before and after the peak. When
calcium salts were injected, the positive peak appeared at a longer
retention time, now with a dip in conductivity after the positive
peak of nearly equal magnitude (see the right peak of Figure 4).
The water used as eluent had been purified with state-of-the-art
equipment and should therefore not give rise to a background
conductivity significantly above the autoprotolysis of water. We
therefore cannot explain these dips at present, since we see no
reason why the conductivity detector signal should go below the
equilibrium conductivity observed when the column had been
338 Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

Table 3. Retention Times of Inorganic Cations with 2
mM Perchloric Acid at 0.6 mL/min as Eluent
cation
tR (min)
cation
tR (min)
Li+
Na+
K+
3.40
3.40
3.40
3.40
5.38
15.0
Zn2+
Mg2+
Ca2+
Cu2+
Ba2+
5.46
6.10
6.15
6.70
7.20
+
NH4
Mn2+
Ce3+
2-
retention order SO4 < F- < Cl- < NO2- < Br- < NO3- < I-.
This order is the same as that obtained by Hu et al., who used
ODS silica that was dynamically coated with zwitterionic surfactant
and water as mobile phase. We thus conclude that this polymeric
zwitterionic stationary phase has a similar ability to separate
common inorganic anions, when aqueous solutions of sodium
perchlorate are used as eluents.
Next, we attempted the separation of cations by injecting 1
mM aqueous solutions of different chloride salts, this time using
2 mM perchloric acid as the eluent. The retention values presented
in Table 3 show that the monovalent cations are eluted faster than
the divalent cations, which in turn elute ahead of a trivalent cation,
represented by Ce3+. However, there was no difference in
retention time among the monovalent cations and only small
differences between the divalent cations tested. This means that
cations of different charge can be separated by this kind of
zwitterionic stationary phase using low eluent concentrations but
that the selectivity between cations of the same charge suffers
from a similar lack of resolution as in the other kinds of isocratic
ion-exchange chromatography.12b
The system peak31 appears at the retention time of the eluent
compound (perchloric acids or its salts). Unlike the pH- and
concentration-dependent retention shifts seen for the system peak
in single-column ion chromatography with carboxylic acid eluents,31a
the retention volume for the system peak in this system was
largely independent of eluent concentration (2.96, 2.57, 2.84, 2.88,
and 2.88 mL, respectively, for 1, 2, 3, 5, and 8 mM HClO4).
Furthermore, we observed no change in system peak retention
volume when perchlorate alkali metal salts were used as eluents
instead of perchloric acid (2.76, 2.58, and 2.71 for eluents
containing 2 mM LiClO4, NaClO4, and KClO4, respectively). With
2 mM perchloric acid eluent, the system peak magnitude was not
noticeably affected by injection of ions eluting before the system
peak, while injection of later eluting iodide ion at 1 mM concentra-
tion diminished the negative system peak to approximately half
its height compared to when deionized water was injected.
Furthermore, anions eluting before the system peak gave positive
signals, while iodide, eluting after the system peak, showed a
negative signal. On increasing the perchloric acid eluent concen-
tration in the experiments that produced the retention data
presented in Figure 8 below, it was also evident that the sensitivity
for iodide increased as the peak moved closer to the system peak.
Figure 5. Chromatogram resulting from the injection of 1 mM each
of several inorganic anions, using 2.5 mM NaClO4 at a flow rate of
0.6 mL/min as eluent with direct electrolytic conductivity detection.
Peak identifications: (1) SO42-, (2) F-, (3) Cl-, (4) NO2-, (5) Br-, (6)
NO3-, (7) system peak, and (8) I-.
extensively equilibrated with Milli-Q water. Attempts to measure
pH in the eluted band proved unsuccessful at the solute concen-
trations used in Figure 4 due to the zero ionic strength and lack
of buffering capacity in the pure water eluent. When 100 mM CaCl2
was injected, we observed a decrease in pH, followed by an
increase above the baseline, at the same retention and with the
same shape as the peak in the experiment with 1 mM injection
and conductivity detection. Because of the exceptionally high
concentration of the injected salt, we abstain from drawing any
conclusions from this observation. What is clear is that these
polymeric materials behaved differently compared to ODS silica
columns dynamically modified with strong/ strong zwitterionic
detergents, where separation of anions is obtained using water
only as the eluents.7,11
Separation of Inorganic Anions and Cations Using P er-
chlorate Eluents. When separation using water only as eluent
failed, we searched for alternative eluents containing various salts
and acids, which could accomplish separation on this material.
The compounds tested as eluents were 5 mM solutions of copper
sulfate, sodium phosphate buffer at pH 7, nitric acid, sulfuric acid,
phthalic acid, and 2 mM solutions of sodium thiocyanate, sodium
perchlorate, perchloric acid, oxalic acid, and citric acid, all chosen
on the basis on their use in zwitterionic chromatography6 and
single-column cation- and anion-exchange chromatography.30 Of
the tested eluents, perchloric acid was superior in terms of eluent
strength and the position of the system peak. We therefore chose
aqueous solutions of perchlorate/ perchloric acid as the eluent
system for the continued investigations.
To identify the ion separation properties of the zwitterionic
separation material with this eluent system, different salt mixtures
were introduced as standard samples. Figure 5 shows a chro-
matogram from the separation of a mixture containing 1 mM each
of sulfate, chloride, fluoride, bromide, iodide, nitrite, and nitrate
ions, all as sodium salts, with 2.5 mM sodium perchlorate as
mobile phase. The mixture was separated within 8 min in the
(31) (a) Jackson, P. E.; Haddad, P. R. J. Chromatogr. 1 9 8 5 , 346, 125-137. (b)
Hayakawa, K.; Kato, A.; Yamamoto, A.; Miyazaki, M. Anal. Sci. 1 9 9 2 , 8,
25-29. (c) Nishimura, M.; Hayashi, M.; Yamamoto, A.; Horikawa, T.;
Hayakawa, K.; Miyazaki, M. J. Chromatogr., A 1 9 9 5 , 708, 195-201. (d)
Yamamoto, A.; Matsunaga, A.; Mizukami, E.; Hayakawa, K.; Miyazaki, M.
J. Chromatogr. 1 9 9 3 , 644, 183-7.
(30) Gjerde, D. T.; Friz, J. S. Ion Chromatography; Hu¨ thig: Heldelberg, 1987; pp
133-182, 208-229.
Analytical Chemistry, Vol. 71, No. 2, January 15, 1999 339

This is in accordance with the findings of Hayakawa31b and can
be explained if eq 3 in their paper is used when Ie, the dissociation
constant of the eluent species, is set to 1 for perchloric acid:
∆Hc ) B(λs- + λH+)[∆S] +
B(λe- + λH+)Ie[∆S]ks′/ (ke′ - ks′) (1)
∆S is the change in sample concentration, ∆Hc is the signal
intensity, λe-, λs-, and λH+ are the equivalent conductivities of the
eluent, sample ion, and hydrogen ions, respectively, and B is a
constant. Accordingly, cation peaks eluting before the system peak
were negative, which is rationalized by the signal polarity behavior
reported by Schill and Crommen.32 Calcium constitutes a special
case, where the signal changed from positive to negative sensitivity
through zero, as the eluent concentration increased and the peak
increased in retention time. This change in detection polarity took
place at a retention time after the system peak and needs further
studies to be explained.
Simultaneous Separation of Inorganic Cations and An-
ions. The above results told us that this polymeric zwitterionic
material can act as a separation medium for anions and cations,
respectively, with perchlorate/ perchloric acid as eluent. It was
therefore intriguing to see whether anions and cations could be
separated simultaneously with a single eluent in a single run.
Further evaluations were therefore carried out by separating a
mixed aqueous solution containing 1.5 mM calcium chloride and
potassium iodide, 1 mM sodium chloride and potassium nitrate,
and 0.5 mM sodium sulfate, using 2 mM perchloric acid as mobile
phase. The resulting chromatogram is presented in Figure 6,
verifying that these anions and cations could be separated in one
run with a single mobile phase.
Figure 6. Chromatogram showing a simultaneous separation of
inorganic anions and cations using 2 mM HClO4 at a flow rate of 0.6
mL/min as eluent with electrolytic conductivity detection. The solution
injected contained 1.5 mM each of calcium chloride and potassium
iodide, 1 mM each of sodium chloride and potassium nitrate, and 0.5
mM sodium sulfate. Peak identifications: (1) SO42-, (2) Cl-, (3) Br-,
(4) Na+ and K+, (5) NO3-, (6) system peak, (7) Ca2+, and (8) I-.
With these promising results, we had to ascertain that the
retention times for the anions and cations were determined by
the individual ions and not by the type of salt injected, as reported
for ODS silica dynamically coated with strong/ strong zwitterionic
surfactants using pure water as eluent. On those stationary phases,
the ions in the sample elute as ion pairs and the same ion may
therefore appear at several retention times with different coun-
terions, depending on the total ionic composition of the sample.7
This was tested by injecting 1 mM solutions of NaCl, NaNO3,
CaCl2, and Ca(NO3)2. In each case, the anions and cations eluted
at their characteristic retention times, independent of the coun-
terion. When the salts were mixed in a concentration matrix and
injected together, the peak heights varied according to the injected
concentrations of each ion, eluting at the retention times indicated
-
in Figure 7. Although the Na+ and NO3 peaks overlap in this
chromatogram, it was evident from these experiments that the
ions eluted as such, not as ion pairs comprising the injected anions
and cations. The use of an ionic eluent therefore relieves the need
for the sample ions to form ion pairs in the elution process.
Effect of Eluent Concentration. To further investigate the
separation conditions with perchlorate eluents, we continued with
a study of the effect of mobile-phase concentration on the retention
of anions and cations. Due to the difficulty of assessing an exact
t0 in this system, we have chosen to report retention volumes
instead of k′ values. Figure 8 thus shows the relationship between
Figure 7. Chromatograms from a simultaneous separation of
inorganic anions and cations using 2 mM HClO4 at 0.8 mL/min as
eluent with direct electrolytic conductivity detection. To the left is a
mixture of 1 mM NaCl and 1 mM Ca(NO3)2, and to the right a mixture
of 1 mM CaCl2 and 1 mM NaNO3.
the concentration of perchloric acid and the retention volume of
several inorganic ions with the same analyte mixture as in Figure
6. The general tendency seen for the anions was that their
retention decreased with increasing perchloric acid concentration.
Thiocyanate ion was also injected at 1 mM concentration but was
not seen at any of the eluent concentrations tested. The retention
(32) Schill, G.; Crommen, J. Trends Anal. Chem. 1 9 8 7 , 6, 111-6.
340 Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

Figure 9. Sorption isotherm obtained by conductometric titration
of the zwitterionic material used in establishing the retention volumes
in Figure 8 and the frontal chromatograms in Figures 12 and 13.
Figure 8. Effect of mobile-phase concentration on the retention
times of inorganic anions and cations using perchloric acid at 0.6 mL/
min as eluent: SO4 ([), Cl- (9), Br- (2), NO3 (b), I- (×), Na+
2-
-
(O), K+ (4), and Ca2+ (0).
time for iodide was >25 min using 1 mM HClO4 as eluent, and it
decreased in a manner normal for anion-exchange separations up
to a concentration of 8 mM, indicative of competitive ionic elution.
Increasing the perchloric acid concentration above this level had
little effect on the retention of I- in the tested interval. The other
anions also showed this general behavior on increasing the
concentration of HClO4: an initial decrease followed by almost
identical retention times at 10 and 20 mM HClO4. This observation,
paired with the relatively large difference in retention time
between the anions even as the effect of increasing the eluent
concentration had ceased at the highest HClO4 concentration
tested, indicates that mechanisms other than “conventional” ion
exchange must be operational. As the ions elute according to
increasing chaotropic properties in the Hofmeister series, i.e., in
the order of smaller hydrated radii,33 it is reasonable to assume
that an exclusion mechanism can be in effect. Partitioning with
the polymeric substrate cannot be ruled out either, since the
hydrophobic character of an ion is also connected to the level of
hydration.
More curiously, the k′ of Ca2+ increased sharply with increasing
concentration of mobile phase. An increase in retention was also
evident for Na+ and K+, although the effect was substantially
smaller and the increase in retention leveled off at an eluent
concentration of ∼5 mM perchloric acid. Increase in retention
with increasing eluent concentration is contrary to retention theory
in conventional ion exchange,34a and we therefore carried out a
series of experiments to characterize the interaction between
perchlorate ion and the zwitterionic material.
Figure 10. Sorption efficiency for perchloric acid (upper curve) and
hydrochloric acid (lower curve) upon titrating the zwitterionic material
recovered from the column below the saturation region.
mg of dry weight) recovered from the column used in preparing
the chromatograms and Figure 8 with 100 mM HClO4 is shown
in Figure 9 and indicates that the sorption reached a plateau level
of 37 µmol of acid at 4 mM perchloric acid in the aqueous phase,
above which the material appeared to be saturated. We then
performed similar titrations with perchloric acid and hydrochloric
acid below the saturation level to determine whether this uptake
of acid was specific for perchloric acid. The titration curves,
presented in Figure 10, show that perchloric acid is sorbed
considerably more efficiently than hydrochloric acid, and although
perchlorate ion forms strong ion pairs with amines, both these
strong acids should be able to titrate weakly basic groups
efficiently. Sorption of acids can thus not be solely due to an acid-
base titration of amine groups on the zwitterionic material. A
“blank” titration of unfunctionalized HEMA-EDMA material was
consequently done to ascertain that the acid consumption was
not due to functional groups present in the methacrylate polymer
matrix. The results of this titration are shown in Figure 11.
Although some acid is consumed, it is evident that the sorption
Sorption Isotherms Obtained by Titration. A sorption
isotherm obtained by conductometric titration of material (842
(33) Cacace, M. G.; Landau, E. M.; Ramsden, J. J. Q. Rev. Biophys. 1 9 9 7 , 30, p
241-277.
(34) (a) Haddad, P. R.; Jackson, P. E. Ion Chromatography-Principles and
Applications; Elsevier: Amsterdam, 1990; p 133. (b) Ibid. p 23. (c) Ibid. pp
152-156.
Analytical Chemistry, Vol. 71, No. 2, January 15, 1999 341

Figure 11. Conductivity “blank” titration of a unfunctionalized
Separon HEMA S300 at low perchloric acid concentration (filled
squares) with titration data for the zwitterion-functionalized material
added for comparison (open triangles).
of strong acids seen in Figures 9 and 10 is due to a property
introduced in the functionalization. We also made an attempt at
measuring pH in the suspension during titration with acids.
Addition of salt was impossible since it would have disturbed the
experiment, so the ionic strength was therefore close to nil during
most parts of the titration and we could not obtain data that are
reliable enough to report.
Figure 12. Breakthrough curves resulting from the frontal chroma-
tography experiment using step changes from pure water to perchloric
acid at different concentrations. Curves at higher concentrations
should be given less significance, because of the difficulties of exactly
determining the dead volume in the system. This effect becomes more
significant when the concentration is increased.
Frontal Chromatography. The dynamic capacity for perchlo-
rate and chloride ions was determined by frontal chromatography,
applying a step change to the respective solution after the column
had been equilibrated with deionized water until the background
was within 1 µS/ cm of the deionized water used for washing. The
breakthrough curves for perchloric acid, compensated for the total
dead volume in the system, are presented in Figure 12. We ran
the tests up to 20 mM HClO4, as higher concentrations were
deemed irrelevant in view of the data presented in Figure 8. The
breakthrough profiles were sharp at all concentrations, and the
amounts sorbed on the zwitterionic material were essentially
constant up to a concentration of 5 mM, above which an increase
appeared to be taking place. However, the unexpected lag seen
in the curves before the conductivity starts to return to the
baseline on switching back to water indicates that the curves from
the two highest concentration are unreliable. This flaw is caused
by the way we had to compensate for the dead volume, wherein
the highest concentrations with their faster breakthrough were
more affected by errors in estimation of the efficient dead volume.
Unable to exactly measure the dynamic t0 value, we based the
correction on measurement of the total liquid volume in the
system, which deviates slightly from the t0 of a porous materials
under dynamic conditions.14b In gel-type materials, there is also a
kinetic factor involved since diffusion in gels is comparatively slow.
It was still found to be the least erroneous way of introducing
this correction. At 2 mM HClO4, the amount sorbed on the 908
mg of dry polymer (recovered from the column afterward) was
23 µmol, which translates into 0.035% N for comparison with the
elemental analysis discussed above. Thus, approximately 3-4%
of the total nitrogen exists in a form that preferentially sorbs strong
acids. In terms of column capacity, this is roughly in the same
range as the separation columns normally used for anion chro-
matography, which means that elution of anions with pure water
should not be possible unless there is an ancillary, “endogenous”
elution process in effect. It would also be hard to explain how
perchloric acid can be washed off the column with pure water in
a relatively short time (cf. the decay portions of Figure 12).
Although the breakthrough for the salts occurred close to the
dead volume, as shown in Figure 13, there was a distinct difference
between NaCl, which had practically no retention, and NaClO4,
where ∼5 µmol had to be pumped through the column before
breakthrough took place. This difference in affinity also existed
for the acids, as seen in Figure 10. In view of the overall degree
of functionalization measured by elemental analysis, the distribu-
tion coefficients for perchlorate ion are surprisingly low, evident
also from the low concentration needed to reach saturation in the
titrations and to effectuate elution in Figure 8. Elution, therefore,
did not take place by conventional ion exchange, and a reasonable
hypothesis is that the strong field of the zwitterion mediates
dissociation of water, forming H+ and OH- that act as a transient
extraneous source of eluting ions. This would result in a “self-
elution” and can rationalize why ions are eluted as ion pairs when
deionized water is used as eluent, as well as the elution of acids
attached to weakly basic sites in the vicinity of a zwitterion. Apart
from the circumstantial evidence presented here, we have not yet
been able to carry out experiments to verify this hypothesis, but
we plan to synthesize and evaluate a series of materials based on
342 Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

pattern seen in Figure 8. These observations were explained by
the combined effect of the preference of anion-exchange sites for
perchlorate and the position of this ion as a moderately strong
chaotrope in the Hofmeister series,33 which is a vaguely defined
and rather empirical way of explaining salt and solvent effects on
interfacial phenomena. In the current system, the slight tendency
-
of ClO4 to serve as a ligand in metal complexes36 makes it an
ideal eluent ion. The leveling off in retention for Na+ and K+ can
thus be explained by normal ion-exchange behavior, if we assume
that the concentration of charged sites on the zwitterionic material
increases concomitantly with the eluent strength exerted by the
hydronium ions, according to the following equation (eq 5.61 in
ref 34c):
1/ y(Q/ y)x/ y[Eym+
]
(2)
-(x/ y)
km′ ) (w/ Vm)(KM,E
)
where w is the weight of the stationary phase, Vm is the volume
of the mobile phase, KM,E is the ionic selectivity coefficient, Q is
the capacity of the stationary phase, E is the eluent concentration
(in this case H+), and x and y are the charges of the solute ion
and eluent ion, respectively. Equation 2 dictates that the retention
should remain constant if there is a corresponding increase in
the number of charged groups on the column and the eluent
concentration, thus explaining the constant retention of the
monovalent cations at eluent concentrations above ∼5 mM.
However, when the surface charge density is low, deviations from
this linearity tend to occur for a multiple charged ion, because an
increasing distance between the ion-exchange sites makes it more
difficult to interact with several ionic sites simultaneously, which
is required for a multivalent ion to adhere to this theory.34c This
provides an explanation of why the retention of Ca2+ increased
more strongly with increasing [HClO4].
Figure 13. Breakthrough curves resulting from frontal chromatog-
raphy experiments, using step changes from pure water to 2 mM of
sodium chloride, sodium perchlorate, and perchloric acid.
weak anion exchangers, where we will gradually increase the
density of strong zwitterionic groups.
It then remains to explain the increased retention of cations
upon increasing the eluent concentration when HClO4 was used
to effect elution. This must be caused by a mechanism strong
enough to overcome the induction of net positive charge on the
material through protonation of the weak amine sites. We attribute
this to an interaction of ClO4-, both with the weak amine and
with the zwitterions. The perchlorate ion has a uniquely high
retention among monovalent anions on strong anion exchangers,34b
indicative of a strong interaction with the quaternary ammonium
groups. The strong ion pairs formed will therefore compensate
for the protonation of amines, rendering the weak anion-exchange
sites tied up in ion pairs. A strong interaction with ClO4- is likely
to exist also in a sulfobetaine-type zwitterion, whereby the
perchlorate ion could get “trapped” by strong interaction in the
quaternary ammonium layer. When the accompanying counterion
has a low interaction with the sulfonic acid group, which is the
case with the strongly hydrated hydronium ion, an increasing
concentration of perchloric acid will induce a net negative charge
on the ion exchanger, resulting in increased retention for cations.
This is supported by observations made in an entirely different
system by Ferreira et al.,35 who saw a uniquely high increase in
aggregation size for micelles formed by N-hexadecyl-N,N-dimethyl-
3-ammonium-1-sulfonic acid (a “Zwittergent” similar to those used
in dynamically coated zwitterion chromatography) when sodium
perchlorate was added to the aqueous medium. They also
observed an initial increase in the binding of a cationic anesthetic
agent to the micelles when adding perchlorate ion followed by a
leveling off at higher concentrations, analogous to the elution
CONCLUSION
The newly synthesized polymeric zwitterionic stationary phase
is an interesting alternative to previously described silica-based
zwitterionic separation materials, since its pH stability allows it
to be used with acid eluents. Thus, by using perchloric acid, we
have found it suitable for the separation of anions and cations
both independently and simultaneously, whereby the retention
times are characterized by the individual ions, not by the salt
injected. The mobile-phase concentration affects the retention
substantially, but in a manner that is quite different from
conventional ion-exchange chromatography. It therefore appears
as if these materials can provide more options for controlling the
individual retention of ions, should materials with both zwitterionic
and conventional ion-exchange groups be prepared and used in a
mixed retention mode. Work is therefore in progress synthesizing
several different covalently bonded polymeric zwitterionic materi-
als differing in spacer length between the anionic and cationic
sites, surface group densities, etc. The retention investigation was
complicated by the likely presence of weak anion-exchange sites
in the column, as indicated by the deviation from stoichiometry
in the elemental analyses, strengthened by frontal chromatography
and the titration experiments. On the other hand, neither FT-IR,
Raman, nor 13C CP/ MAS NMR were able to positively verify or
(35) Ferreira, G. S. S.; Pe´rigo, D. M.; Politi, M. J.; Schreier, S. Photochem.
Photobiol. 1 9 9 6 , 63, 755-761.
(36) Cotton, F. A.; Wilkinson G. Advanced Inorganic Chemistry, 5th ed.; John
Wiley & Son: New York, 1988; p 568.
Analytical Chemistry, Vol. 71, No. 2, January 15, 1999 343

rule out the presence of tertiary amino groups. Although we take
into account that a significant fraction of groups in the column
may have weak anion-exchange properties, the material still
showed a typical zwitterionic retention behavior.7,11
us to use their elemental analysis equipment. The authors are
also grateful to Per Persson for help with the Raman spectroscopy,
to Jo´n Hauksson for acquiring the CP/ MAS NMR spectra, and
to Camilla Viklund and Einar Ponte´n for valuable help and
discussions during the course of this work.
ACKNOWLEDGMENT
Received for review April 13, 1998. Accepted October 19,
1998.
This work was supported by The Swedish Natural Science
Research Council, Grant K-KU 8724-315. The National Swedish
Laboratory for Agriculture Chemistry is acknowledged for allowing
AC9804083
344 Analytical Chemistry, Vol. 71, No. 2, January 15, 1999