Study of the Behavior of Water-Soluble
Vitamins in HILIC on a Diol Column
2010, 71, 751–759
Andreas E. Karatapanis, Yiannis C. Fiamegos, Constantine D. Stalikas&
Laboratory of Analytical Chemistry, Department of Chemistry, University of Ioannina, 45110 Ioannina, Greece; E-Mail: [email protected]
Received: 21 December 2009 / Revised: 12 February 2010 / Accepted: 19 February 2010
Online publication: 11 April 2010
Introduction
Abstract
An effort has been made to investigate the chromatographic behavior and to understand the
basic mechanisms in HILIC-based separation of water-soluble vitamins with highly varied
properties on a diol column. The water content of the mobile phase is of utmost importance
because it directly affects the type and extent of interactions of the solutes with the stationary
phase and with the buffered mobile phase. A mixed-mode partitioning-surface adsorption
mechanism enables most precise description of their chromatographic retention and separation. The point at which surface adsorption becomes apparent, however, depends on the
properties of the solutes on the given stationary phase, and on the presence of buffer salt
ions. Adjustment of mobile phase pH and use of different buffer salts can be used to modify
electrostatic interactions among the solutes, active silanols, and counter-ions. The role of
hydrogen bonding was clarified by substitution of ACN by solvents with moderate to strong
hydrogen bonding potential. Analytes which are neutral at the working pH start to interact
with the stationary phase when the ACN content is increased to 80%. Negatively charged
analytes are adsorbed on the stationary phase when the ACN content is approximately 86%,
because augmentation of the counter-ions weakens electrostatic repulsion by the active
silanol groups. On the other hand, the electrostatic attraction of thiamine contributes
significantly to its retention even when using mobile phases with high water content.
Keywords
Column liquid chromatography
Hydrophilic interaction liquid chromatography
Diol column
Chromatographic behavior
Retention models
Water-soluble vitamins
Original
DOI: 10.1365/s10337-010-1564-3
0009-5893/10/05
The term hydrophilic interaction liquid
chromatography (HILIC) was first
coined by Alpert, who used the method
for separation of proteins, peptides,
amino acids, oligonucleotides, and
carbohydrates [1]. Conceptually, the
mechanism of HILIC primarily involves
‘partitioning of polar analytes between
the mobile phase and a layer of mobile
phase enriched with water and partially
immobilized on the stationary phase’ [1].
HILIC is finding new and innovative
applications every day and has come to
the forefront in much recent literature.
Over the years, HILIC has been proved
to be a useful technique for retention
and separation of polar compounds,
providing an alternative separation
mechanism to traditional reversed-phase
liquid chromatography (RP-LC). HILIC
retention is based on the interactions
between a polar stationary phase (e.g.
bare silica, aminopropyl, cyano, diol)
and aqueous–organic mobile phases [2].
Increasing the polarity of the solutes
leads to stronger interactions with the
environment of the stationary phase and
to increased retention, which favors their
separation. In contrast with RP-LC,
elution in HILIC is facilitated by the
aqueous component of the mobile phase.
Chromatographia 2010, 71, May (No. 9/10)
Ó 2010 Vieweg+Teubner Verlag | Springer Fachmedien Wiesbaden GmbH
751
responsible for their retention in the
HILIC process by evaluating critical
chromatographic conditions and retention models.
Experimental
Chemicals, Reagents,
and Solutions
Fig. 1. Molecular formulae and pKa values of the WSV [37]
Partitioning was originally proposed as
the retention mechanism in HILIC.
Analytes do not interact directly with the
stationary phase, although there is evidence that surface adsorption may play
an important role in retention [2]. The
mechanism of retention in HILIC has
not been fully established, however,
because complex interactions occur among
the polar solute, the solvent and counterions of a buffered mobile phase, the polar
stationary phase, and the residual silanol
groups [3, 4].
Conditions useful in the optimization
of HILIC separation, for example
mobile phase organic content and organic
modifiers, mobile phase pH, buffer concentration, column chemistry and temperature, have been emphasized in many
publications, which reveal how these
752
conditions affect separation selectivity
[4–12].
The water-soluble vitamins (WSV)
are a group of compounds with hydrophilic character; this suggests they would
be retained under HILIC conditions.
Because of their structural complexity
(Fig. 1) and their diverse chemical and
physicochemical properties, their chromatographic separation, and study of
their behavior under HILIC conditions
would be both challenging and
demanding. We recently developed and
validated a method based on isocratic
HILIC separation for analysis of WSV in
pharmaceutical preparations [13]. In the
work discussed in this paper, an effort was
made to investigate the behavior of
selected WSV on a diol column to obtain
insight into the assorted mechanisms
Ammonium acetate, acetic acid, thiamine, nicotinic acid, nicotinamide, pyriDL-dithiothreitol
(DTT),
doxine,
triethylamine (TEA), and trifluoroacetic
acid (TFA) were obtained from Fluka
Chemie (Buchs, Switzerland). Ammonium formate, sodium acetate, and
riboflavin were obtained from SigmaAldrich Hellas (Athens, Greece) and
formic acid from Scharlau Chemie
(Barcelona, Spain). Aqueous ammonia
(25%) and HCl (37%) were purchased
from Merck (Darmstadt, Germany)
and L-ascorbic acid and ammonium
chloride were from Reidel-de Haën
(Seelze, Germany). HPLC-grade acetonitrile (ACN), methanol (MeOH), isopropanol (IPA) and tetrahydrofuran
(THF) were purchased from Fisher
Scientific (Leicester, UK). Doubledistilled water, filtered through a 0.45 lm
nitrocellulose membrane, was used
throughout. All chemicals and solvents
were of analytical-reagent grade.
Stock solutions of the water-soluble
vitamins (500 mg L-1) were prepared by
dissolving appropriate amounts in
ACN–water (1:1). Dithiothreitol was
used as stabilizer for L-ascorbic acid
solutions. To achieve complete dissolution of riboflavin, addition of aqueous
ammonia was necessary.
All stock solutions were stored
refrigerated in dark vials. Working
standard solutions were prepared daily
by appropriate dilution of the concentrated stock solutions with mobile phase.
Chromatography
Liquid chromatographic analysis of the
water-soluble vitamins was conducted
with a Shimadzu (Duisburg, Germany)
HPLC system comprising an LC20AD
pump, a CTO 10AS column oven, and
Chromatographia 2010, 71, May (No. 9/10)
Original
an SPD-M20A diode-array detector.
Injections were made by means of a
manual 7725i Rheodyne (Cotati, CA,
USA) injector with a 20 lL sample
loop. A GL Sciences (Tokyo, Japan)
HILIC Inertsil, silica-based, diol column
(150 mm 9 4.6 mm, 5 lm particle size)
was used for separation of the analytes.
LC-solution software, version 1.21 SP1,
was used for data analysis and processing. Detection wavelengths were 254 and
272 nm.
The mobile phases used for HPLC
were prepared by mixing 100% water
containing one of the salts ammonium
acetate, ammonium formate, ammonium
chloride, or sodium acetate at different
concentrations and pH (component A)
with 90:10 (v/v) ACN–water containing
the same salts at the same concentrations
as used in A (component B). The pH of
component A was adjusted by adding
acetic acid to the ammonium acetate and
sodium acetate-containing solutions,
formic acid to the ammonium formatecontaining solution, and hydrochloric
acid to the ammonium chloride-containing solution. These acids were
appropriately diluted before use for pH
adjustment. The same volumes of the
acids were added to component B at
each pH. Mobile phases were delivered
isocratically at a flow-rate of 0.6 mL
min-1.
The dead/void time (t0), a measure of
the dead/void volume of the column, was
determined by injecting aqueous ACN
and monitoring the baseline disturbance.
Results and Discussion
Type of Organic Modifier
The chromatograms obtained by use of
different organic solvents are compared in Fig. 2. The eluotropic strength
of the modifiers tested for the compounds under study follows the order
ACN < THF < IPA < MeOH. The
vitamins tend to ‘‘bunch-up’’ and
become poorly separated with alcoholcontaining mobile phases. Use of THF
resulted in adequate retention and separation but it was inferior to ACN. The
solvent order given above is not consistent with the eluotropic strength
Original
Fig. 2. Effect of organic modifier on separation of the WSV. Conditions: Inertsil diol column
with 10:90 (v/v) aqueous ammonium acetate buffer (10 mM, pH 5.0) —organic solvent as mobile
phase; flow rate 0.6 mL min-1; column temperature 25 °C. 1 = nicotinamide, 2 = pyridoxine,
3 = riboflavin, 4 = nicotinic acid, 5 = L-ascorbic acid, 6 = thiamine
reported for these modifiers on silica
(THF < ACN < IPA < MeOH) [14]
or with their polarity, which increases
in the order IPA < THF < MeOH
< ACN (if one considers that partitioning is the primary separation mechanism in HILIC). These observations
indicate that the retention behavior of
the solutes cannot be anticipated or
predicted by use of eluotropic strength
or solvent polarity alone. These properties are not only a function of the mobile
phase but also depend on the chemical
structure of the solute and the supporting material of the column.
Another means of satisfactory
explanation of the retention behavior
of the solutes is based on the significant differences among the hydrogenbonding capacities of the modifiers,
compared with water. MeOH is characterized by high propensity to interact
as a strong proton donor (a = 0.93)
and as a strong proton acceptor
(b = 0.62), whereas ACN is a weak
proton acceptor (b = 0.31) and much
weaker proton donor (a = 0.19). Because of its strong hydrogen-bonding
ability, MeOH competes for interactions with the solutes partitioned into
the water-enriched layer, resulting in
poor retention. MeOH has also been
reported to perturb the formation of the
water-enriched layer on the stationary
phase surface by replacing water molecules and producing a more hydrophobic stationary phase [4]. As a
consequence, analytes with hydrogenbonding potential, for example pyridoxine, riboflavin, nicotinic acid, and
ascorbic acid, are poorly retained.
Retention was also poor when ACN
was replaced with IPA, because the
latter has strong hydrogen-bonding
capacity (a = 0.76, b = 0.95). Resistance to reduction of the retention of
thiamine was observed for both alcoholcontaining mobile phases. This was
more pronounced with IPA (Fig. 2),
because of its demonstrated strong ionexchange interactions, which contribute
to the retention mechanism. The elution
profile of THF resembles that of ACN,
although most analytes were more
strongly retained when ACN was used.
THF has no proton-donor properties
(a = 0.00) but it accepts protons more
strongly (b = 0.55) than ACN; this
could account for the lower retention of
the solutes.
Chromatographia 2010, 71, May (No. 9/10)
753
Fig. 3. Effect of pH on the retention factors (k) of the WSV. Mobile phase ACN–water 90:10
(v/v); flow rate 0.6 mL min-1; column temperature 25 °C
Effect of Mobile Phase pH
Ammonium acetate at a concentration
of 10 mM was used throughout. At
pH 3.0 (which is outside the pH
range of ammonium acetate), ammonium formate at the same concentration
was tested. It should be noted that the
pH values of the mobile phases refer to
the aqueous portion. The effect of pH on
the degree of ionization both of the solutes and of the stationary phase should
be considered to explain the different
behavior. Empirical calculations reveal
that for every 10% increase in ACN the
pH of the aqueous ammonium acetate
increases by approximately 0.3 pH units
[15]. With regard to the ionizable analytes and the residual silanols, in the
same manner the pKa values of weak
bases decrease with increasing organic
content whereas those of weak acids
increase [16]. Because the pH shift for
the aqueous–organic mobile phase used
and the pKa shift for the acidic analytes
on addition of ACN are almost proportional [15], a compromise can be reached
to study the behavior of WSV in the
aqueous–organic mobile phase, taking
into consideration the aqueous pKa
754
values only, as if we are working in a
purely aqueous environment.
Figure 3 shows the retention factors
obtained for the WSV at different pH
with ACN–water 90:10 (v/v) as mobile
phase. Retention factors for riboflavin,
nicotinamide, and pyridoxine remained
almost unaffected as the pH of the mobile phase was increased from 3.0 to 6.0.
Riboflavin is not charged in the pH
range 4.0–6.0 (pKa * 10.2, weakly
acidic amido group). In water, the pKa
values of nicotinamide are *3.3 (pyridinyl group) and *9.2 (weakly acidic
amide group) and those of pyridoxine
are *5.0 (pyridinyl group) and *8.9
(ring OH). Although pyridoxine and, to
a lesser extent, nicotinamide are positively charged in the pH range 3.0 to 4.0,
retention data are indicative of weak or
absent electrostatic interactions with the
residual silanol groups of the column.
This behavior can be attributed to silanols’ lack of significant deprotonation
and possible charge delocalization, by
resonance of the pyridinyl groups of the
progressively polar molecules.
The retention factors of nicotinic
acid, L-ascorbic acid, and thiamine were
dramatically increased by increasing
mobile phase pH from 3.0 to 6.0. Nicotinic acid has two pKa values
(pKa1 * 2.2 for the carboxyl group and
pKa2 * 4.8 for the pyridinyl group) and
hence, in the pH range 3.0–4.0 the
carboxylic acid functionality is deprotonated, the amine functionality is partially protonated, and the molecule exists
as a zwitterion. At pH >5.0, the amine
functionality is deprotonated and the
molecule is negatively charged. L-Ascorbic acid, also, has two pKa values in
water (pKa1 * 4.1 and pKa2 * 11.7). At
mobile phase pH 3.0, retention of nicotinic acid and L-ascorbic acid is similar
to that of pyridoxine and riboflavin,
causing co-elution of these four analytes
(Fig. 3). On increasing the mobile phase
pH to 4.0, L-ascorbic and nicotinic acids
are charged to a different extent and
discriminated from pyridoxine and
riboflavin, because they are more polar.
At mobile phase pH 5.0, nicotinic and
L-ascorbic acids are fully resolved with
inversion of the order of elution and
with L-ascorbic acid the later-eluting
compound. At this pH, ascorbic acid
becomes fully deprotonated and, hence,
more polar than at pH 4.0. This may
account for the inversion in the order of
elution of the peaks for this pair of
analytes when changing from pH 4.0 to
pH >5.0. Thiamine is positively
charged—it carries a permanent positive
charge on thiazolium ring (pKa1 of
the pyrimidinyl group *4.80)—thus,
undergoing ion-exchange interactions
with residual silanols which leads to
greater retention.
Effect of Type and
Concentration of Buffer
Different buffer salts, ammonium
acetate, ammonium formate, ammonium
chloride, and sodium acetate at a concentration of 10 mM and pH 5.0, were
also evaluated to examine the affect of
different counter-ions on retention of the
WSV. The mobile phase containing 80%
ACN was chosen to compensate for
solubility limitations. The analytes
carrying acidic functionality (i.e. nicotinic
acid and L-ascorbic acid) had slightly
different retention times. The tendency
Chromatographia 2010, 71, May (No. 9/10)
Original
of retention times in ammonium formate
to be slightly longer can be attributed to
the different eluent strength of the formate and acetate ions. This phenomenon
was more striking when 90% ACN was
used; under these conditions the ammonium ions form a more complete electrical layer on the surface of the
stationary phase, reducing its electrostatic repulsion of the acidic molecules
[17, 18]. When ammonium acetate was
used instead of ammonium formate the
positively charged thiamine was less retained. Ion association with the acetate
anion results in greater solubility in the
bulk eluent, which explains the moderate
decrease in the retention of thiamine
[19]. Nicotinamide, pyridoxine, and
riboflavin, being neutral at the working
pH, were unaffected.
Retention for the acidic WSV was
much less with ammonium chloride than
with ammonium acetate and ammonium
formate, because of the stronger competition of chloride anions with silanols
for the ammonium ions. When the mobile phase containing sodium acetate was
used the retention of the acidic analytes
increased substantially. The opposite
was observed for thiamine, and nicotinamide, pyridoxine, and riboflavin were,
again, unaffected. This behavior reflects
the more effective competition of Na+
than NH4+ for the ion-exchange sites of
the stationary phase.
Finally, use of TFA (0.1%, v/v) resulted in weak retention, with retention
factors ranging from 0.6 to 1.8 when the
ACN content was 90%; this resulted in
co-elution of most of the analytes. An
explanation of this behavior is the ionpairing strength of TFA, which results in
a decrease of the hydrophilicity of the
basic analytes [20, 21]. For the acidic
analytes, also, hydrophilicity is reduced,
because of their protonation at low mobile phase pH (*2.5).
The effect of buffer concentration on
retention and separation was investigated by varying the concentration of
ammonium acetate from 5 to 20 mM
with different amounts of ACN, keeping
the aqueous mobile phase at pH 5.0
(Fig. 4). For mobile phase of composition ACN–water 90:10, an increase in
buffer concentration from 5 to 20 mM
led to a moderate increase in the retenOriginal
Fig. 4. Effect of buffer concentration on separation of the WSV. Conditions: Inertsil diol
column with 90:10 (v/v) ACN–water containing 5, 10, or 20 mM ammonium acetate (pH 5.0) as
mobile phase; flow rate 0.6 mL min-1; column temperature 25 °C. Peak assignment as for Fig. 2.
Peaks 5 and 6 at 20 and 5 mM, respectively, do not elute in the time scale of the chromatograms
tion of nicotinic acid. For L-ascorbic acid
the retention was substantial—it did not
elute within 45 min. In contrast, the
retention of nicotinamide was unaffected
and a slight increase in the retention of
pyridoxine and riboflavin was observed.
Adopting the theory of the water-enriched liquid layer, at high mobile phase
organic content the salt prefers to reside
in this layer. In addition, higher salt
concentrations drive more solvated salt
ions into the water-rich layer. This
would result in an increase in the
hydrophilicity of the aqueous layer,
promoting retention of the analytes by
the partition mechanism [8]. Apart from
the effect of the salt on the hydrophilicity
of the mobile phase, the prolonged
retention with increasing ionic strength
of the mobile phase is evidence that extra
electrostatic effects are involved in the
retention of charged analytes compared
with neutral ones. Finally, the retention
factor of thiamine decreased substantially with increasing buffer concentration. This observation clearly indicates
the occurrence of ion-exchange interactions with the residual silanol groups of
the stationary phase. An increase in
buffer concentration would lead to augmentation of the counter-ion (ammo-
nium) concentration in the mobile phase,
leaving fewer active silanol groups
available to interact with thiamine, thus
reducing its retention time.
Modeling the Dependence
of HILIC Retention on the
Acetonitrile Content of the
Mobile Phase
Many attempts have been made to
determine whether the mechanism of
retention in HILIC is partitioning or
surface adsorption or both [5, 9, 12, 22,
23]. Secondary dipole–dipole and
hydrogen-bonding interactions or ionexchange interactions between ionizable
solutes and the mobile phase and/or
stationary phase have also been proposed [3, 4, 8, 20, 24, 25]. When a separation is based on partitioning, the
relationship between the retention factor
k and the volume fraction of the stronger
solvent in the mobile phase is given by
Eq. 1 [2, 26, 27]:
log k ¼ log korg Su
ð1Þ
where korg is the retention factor for the
weaker (i.e. organic) component only as
mobile phase, u is the volume fraction of
Chromatographia 2010, 71, May (No. 9/10)
755
Fig. 5. Plots of log k against volume fraction of water in the mobile phase. Conditions: Inertsil
diol column with ACN–10 mM aqueous ammonium acetate, pH 5.0, as mobile phase; flow rate
0.6 mL min-1; column temperature 25 °C
the stronger solvent, and S is the slope of
the plot of log k against u when fitted to
a linear regression model.
For surface adsorption the relationship can be expressed by Eq. 2 (the
Snyder–Soczewinski model) [2, 28, 29]:
log k ¼ log kw AS
log u
nw
ð2Þ
where kw is the solute retention factor
with pure water as mobile phase, AS and
nw are the cross-sectional areas on the
surface occupied by the solute molecule
and water molecule, respectively, and u
is the mole fraction of the stronger solvent (i.e. water) in the mobile phase.
Jin et al. [30] recently proposed the
following
model
which
properly
describes the HILIC retention of nucleosides on some polar column packings:
ln k ¼ a þ b ln u þ cu
ð3Þ
where a is a constant related to the
interaction energy between solutes and
the stationary and mobile phases, b is a
coefficient related to direct analyte–
stationary phase interactions, and c is
related to the interaction energy between
solutes and solvents.
756
To model the HILIC retention of
WSV on the bonded diol column, chromatography was performed using isocratic elution with 10 mM ammonium
acetate buffer at pH 5.0 with ACN content varying from 50 to 90% (v/v). For
all six compounds there was little or no
retention when the mobile phase contained 50% (v/v) H2O whereas for ACN
greater than 70% (v/v), the HILIC
behavior of the column could be observed. The retention times were inversely proportional to the water content
of the mobile phase and increased with
the polarity of the solute, which is
characteristic of typical hydrophilic
interactions [13].
Retention models are usually compared on the basis of how well the
respective equation fits retention data for
the solutes obtained over a range of
mobile phase compositions. Plots of log
k against water content and the logarithm of water content are given in
Figs 5 and 6. Neither log–linear plots
(0.9587 < R2 < 0.9938) nor log–log
plots (0.9817 < R2 < 0.9988) were
satisfactory for all the vitamins, hinting
at the absence of pure partitioning or pure
surface adsorption mechanisms. Replotting the log–linear data for nicotinamide,
pyridoxine, and riboflavin (neutral at the
working pH) for water content ranging
from 20 to 50% (v/v) resulted in a
significantly improved linear relationship (0.9900 < R2 < 0.9929). Likewise,
linear relationships for ascorbic acid,
nicotinic acid, and thiamine (charged
analytes) were significantly improved
(0.9905 < R2 < 0.9991) by replotting
the log–linear data with the water content ranging from 16 to 50% (v/v). These
results indicate there is a change in the
relative contributions of partitioning and
surface adsorption mechanisms, with the
latter being more evident for highly organic mobile phases.
It is difficult to locate a turning point,
accurately applicable for all analytes, at
which the contribution of surface
adsorption to the overall retention
mechanism begins to increase. For
highly organic mobile phases (i.e. >80%
v/v), a pronounced difference in polarity
between the aqueous layer and the bulk
mobile phase is established. The WSV,
which are neutral at pH 5.0, remain in
the aqueous layer longer. In this way,
they are more amenable to hydrogen
bonding interactions with the surface of
the stationary phase. The different contribution of surface adsorption for the
charged analytes can be attributed to
electrostatic interactions (attractive or
repulsive) with the negatively charged
silanol groups in the stationary phase.
As the organic content increases further
(>85%) more salt ions enter the aqueous layer. Consequently, the ammonium
ions more effectively shield the negatively charged silanol groups and the
repulsive interactions that nicotinic and
ascorbic acids experience at high aqueous content become less pronounced.
This enables the analytes to interact
further with the surface of the stationary
phase by hydrogen bonding. On the
other hand, the electrostatic attraction of
thiamine to the stationary phase is still
evident with highly aqueous mobile
phases. This is because of its greater
capacity to compete with ammonium
ions, which interact with charged silanol
groups.
Equation 3, which proved workable
for nucleosides [30], seems to fit the data
Chromatographia 2010, 71, May (No. 9/10)
Original
for the entire range of water content of the
mobile phase, despite the complex role of
solute–solvent–stationary phase interactions. Although statistical analysis of
residuals is the most rigorous test of the
degree-of-fit, more simply, the mean
residual and the residual standard deviation can be used, which should be close to
the standard deviation of ln k values. The
numerical values of the constants a, b, and
c for the analytes under study are listed in
Table 1, with goodness-of-fit test data,
which demonstrate the fitness of the
model. The specific values of the coefficient c again indicate suggest that the
WSV studied can be divided into two
distinct groups. Nicotinamide and pyridoxine, which carry no charge at the
working pH, have negative values close to
zero. In contrast, charged molecules are
characterized by low negative values.
Although uncharged, riboflavin lies
between the two groups, possibly because
the larger number of polar functional
groups compared with the two neutral
compounds can be solvated by the water
molecules.
Taking into account that:
1. polar analytes partition between the
water-rich layer on the surface of the
polar stationary phase and the relatively hydrophobic mobile phase [1]
and
2. when highly organic mobile phases
are used the analytes are able to
directly interact with the stationary
phase
we can explain the chromatographic
behavior of the WSV. The interaction of
polar functionality with the surface of
the stationary phase is a logical explanation of the different retention behavior
of nicotinamide, pyridoxine, and riboflavin. Nicotinic and L-ascorbic acids are
both at their ionic form and, thus, are
more hydrophilic than the first three
eluting compounds. The repulsive interactions these two analytes experience
from the active silanol groups become
weaker for highly organic mobile phases,
as mentioned above, and their retention
is significantly increased above 80% (v/v)
ACN. In addition, the order of elution of
nicotinic acid and ascorbic acid is explained by the presence of functional
Original
Fig. 6. Plots of log k against the logarithm of the volume fraction of water in the mobile phase.
Conditions as for Fig. 5
groups (for example hydroxyl groups)
with greater hydrogen bonding potential
in the L-ascorbic acid moiety. Finally,
electrostatic attraction of thiamine by
the silanols of the stationary phase [13]
causes it to elute with the longest retention time.
Effect of Column Temperature
Column temperature is an important
condition in liquid chromatographic
analysis because it affects mobile phase
viscosity, solute partitioning, and diffusivity [31]. Van’t Hoff described the
relationship between retention factor (k)
and column temperature (T) by use of
the equation:
DH DS
ln k ¼ þ
þ ln /
RT
R
ð4Þ
where DH° and DS° are the change of
enthalpy and entropy between the mobile and the stationary phases, R is the
gas constant, and / is the phase ratio.
This equation should sensibly apply to
HILIC mode, if the retention involves
partitioning. The effect of column temperature on retention of the WSV was
investigated by varying the column
temperature from 15 to 45 °C at the
constant mobile phase composition
ACN–water 90:10 containing 10 mM
ammonium acetate (pH 5.0). A significant decrease in retention was observed
for nicotinic and L-ascorbic acids, a
moderate decrease for pyridoxine and
riboflavin, and a slight decrease for
nicotinamide and thiamine. The increase
in the pKa values of the acidic compounds
with temperature lessens the difference
between their polarity and that of the
mobile phase. This could be a reasonable
explanation of the striking reduction in
the retention of nicotinic and ascorbic
acids and the less pronounced reduction
of thiamine.
A plot of ln k as a function of 1/T is
more linear (R2 = 0.9847–0.9966) for
nicotinamide, pyridoxine, and riboflavin
than for the other compounds (Table 2).
This is similar to the behavior usually
observed in RPLC, pointing to partitioning as the major force driving
retention [32–34]. Deviation from linearity was observed for nicotinic acid
and L-ascorbic acid, and was more
pronounced for thiamine; this is attributed to abovementioned electrostatic
Chromatographia 2010, 71, May (No. 9/10)
757
Table 1. Regression data for the WSV by use of Eq. 3
Compound
aa
Nicotinamide
Pyridoxine
Riboflavin
Nicotinic acid
L-Ascorbic acid
Thiamine
-1.29
-1.46
-2.03
-0.64
0.74
0.34
ba
±
±
±
±
±
±
0.23
0.12
0.26
0.28
0.28
0.23
-0.655
-0.954
-1.39
-1.37
-1.08
-1.91
ca
±
±
±
±
±
±
0.093
0.050
0.11
0.15
0.19
0.29
-0.69
-0.36
-1.52
-4.78
-8.07
-7.66
±
±
±
±
±
±
0.37
0.19
0.42
0.61
0.76
0.91
R2
RESmean
RESSD
0.9951
0.9991
0.9985
0.9986
0.9986
0.9952
0.024674
0.012975
0.025509
0.035012
0.043962
0.041138
0.017
0.009
0.023
0.019
0.029
0.034
a
± standard error, evaluated at P = 0.05
RESmean is the mean of the residuals and RESSD is the standard deviation of residuals; the standard deviation of ln k ranges from 0.0095
to 0.016
Table 2. Squared correlation coefficients and enthalpies of retention of the WSV
Compound
Nicotinamide
Pyridoxine
Riboflavin
Nicotinic acid
L-Ascorbic acid
Thiamine
a
b
ACN–water 80:20a
ACN–water 90:10a
ACN–water 90:10b
R2
DH° (kJ mol-1)
R2
DH° (kJ mol-1)
R2
DH° (kJ mol-1)
0.9975
0.9939
0.9957
0.9926
0.9934
0.9924
-6.67
-9.04
-11.9
-13.6
-21.9
-8.00
0.9966
0.9900
0.9951
0.9847
0.9870
0.9524
-5.65
-8.53
-10.4
-9.53
-19.9
-6.14
0.9983
0.9937
0.9932
0.9929
0.9973
0.9848
-8.47
-13.0
-16.9
-16.4
-32.6
-8.44
CH3COONH4 10 mM, pH 5.0
CH3COONH4 10 mM, TEA 20 mM, pH 5.0
interactions [35]. Interestingly enough,
linear behavior (R2 = 0.9926–0.9981)
was observed for all the compounds
tested when the ACN content was
reduced from 90 to 80% (v/v) with 10 mM
ammonium acetate, in the absence of
TEA. Further investigation of the effect of
column temperature on the retention
behavior of thiamine showed that addition of TEA (20 mM) to the mobile phase
containing 90% ACN resulted in greater
linearity (R2 = 0.9848), suggesting the
occurrence of an ion-exchange mechanism. These data are consistent with the
blocking action of residual silanol groups
by TEA.
On the basis of the retention,
enthalpy values were calculated for all the
compounds from the slope of van’t Hoff
plots; these are listed in Table 2. For all
the analytes the enthalpy of retention
was negative, ranging from -5.65 to
-19.93 kJ mol-1 although positive
values have been reported in HILIC [36].
Our results indicate that transfer of the
solutes from the mobile phase to the
stationary phase is an exothermic process
[8] and the solutes become less retained as
758
column temperature is increased. Addition of TEA to the mobile phase
increased the absolute values of the
retention enthalpy of the analytes,
because of the decrease in electrostatic
interactions.
Conclusion
The mechanism of retention of WSV in
HILIC is multimodal. Apart from partitioning, specific interactions, that may
predominate, depend strongly on the
structure of the analytes, the nature of
the stationary phase, and the composition of the mobile phase. Deviation
from the partition mechanism when
highly organic mobile phases are used is
evidence of the occurrence of both
partitioning and surface adsorption.
Data gathered from this study indicate
that above 80% ACN there is a change
in the relative contributions of these
two mechanisms and adsorption becomes more significant. The contribution of the adsorption mechanism for
the charged WSV depends on the extent
to which electrostatic effects, attractive
or repulsive, are present as mixed-mode
effects. Analytes which are neutral
under the working conditions are not
affected by changing buffer composition
or pH; they start to interact with the
stationary phase when the amount of
ACN is above 80%. In addition, high
ACN content is required for negatively
charged analytes to interact with the
surface of the stationary phase, because
electrostatic repulsion by the negatively
charged silanols is weakened. For the
positively charged thiamine, adsorption
contributes significantly to the overall
mechanism in the range of ACN composition studied. Hydrogen bonding
contributes substantially to retention in
HILIC, which is confirmed by the
reduction of retention when switching
to organic solvents with stronger
hydrogen bonding capacity, for example
alcohols. The satisfactory linearity of
the Van’t Hoff equation for all the
compounds tested when the water content is increased from 10 to 20% (v/v),
is an additional convincing argument
for the contribution of the adsorption
Chromatographia 2010, 71, May (No. 9/10)
Original
mechanism when highly organic mobile
phases are used.
Finally, the models of partitioning
and adsorption, per se, commonly discussed in the literature, are not appropriate for describing the behavior of
WSV in the whole range of mobile
phase water content. Another model,
which considers solute–solvent–stationary phase interactions, can describe
more precisely their multimodal chromatographic behavior discriminating the
compounds studied according to their
charge.
References
1. Alpert AJ (1990) J Chromatogr 499:177–
196. doi:10.1016/S0021-9673(00)96972-3
2. Hemström P, Irgum K (2006) J Sep Sci
29:1784–1821. doi:10.1002/jssc.200600199
3. Tanaka H, Zhou X, Masayoshi O (2003)
J Chromatogr A 987:119–125. doi:
10.1016/S0021-9673(02)01949-0
4. Li R, Huang J (2004) J Chromatogr A
1041:163–169. doi:10.1016/j.chroma.2004.
04.033
5. Guo Y, Huang AH (2003) J Pharm Biomed Anal 31:1191–1201. doi:10.1016/
S0731-7085(03)00021-9
6. Quiming NS, Denola NL, Saito Y,
Catabay AP, Jinno K (2008) Chromarographia 67:507–515. doi:10.1365/s10337008-0559-9
7. Pack BW, Risley DS (2005) J Chromatogr
A 1073:269–275. doi:10.1016/j.chroma.
2004.09.061
8. Guo Y, Gaiki S (2005) J Chromatogr A
1074:71–80. doi:10.1016/j.chroma.2005.
03.058
Original
9. Valette JC, Demesmay C, Rocca JL,
Verdon E (2004) Chromatographia
59:55–60. doi:10.1365/s10337-003-0139-y
10. Vikingsson S, Kronstrand R, Josefsson M
(2008) J Chromatogr A 1187:46–52. doi:
10.1016/j.chroma.2008.01.070
11. Wu JY, Bicker W, Lindner W (2008) J
Sep Sci 31:1492–1503. doi:10.1002/jssc.
200800017
12. Liu M, Chen EX, Ji R, Semin D (2008) J
Chromatogr A 1188:255–263. doi:10.1016/
j.chroma.2008.02.071
13. Karatapanis A, Fiamegos Y, Stalikas C
(2009) J Sep Sci 32:909–917. doi:10.1002/
jssc.200800525
14. Meyer VR (ed.) (2004) In: Practical highperformance liquid chromatography, 4th
edn. Wiley, Chichester
15. Kazakevich Y, LoBrutto R (eds.) (2007)
In: HPLC for pharmaceutical scientists,
Wiley, New Jersey
16. Gagliardi LG, Castells CB, Ráfols C,
Rosés M, Bosch E (2007) Anal Chem
79:3180–3187. doi:10.1021/ac062372h
17. Alpert AJ (2008) Anal Chem 80:62–76.
doi:10.1021/ac070997p
18. Takegawa Y, Deguchi K, Ito H, Keira T,
Nakagawa H, Nishimura SI (2006) J Sep
Sci 29:2533–2540. doi:10.1002/jssc.2006
00133
19. Wang XD, Li WY, Rasmussen HT (2005)
J Chromatogr A 1083:58–62. doi:
10.1016/j.chroma.2005.05.082
20. Strege MA (1998) Anal Chem 70:2439–
2445. doi:10.1021/ac9802271
21. Daunoravičiusl Ž, Juknaite I, Naujalis E,
Padarauskas A (2006) Chromatographia 63:373–377. doi:10.1365/s10337-0060768-z
22. Naidong W (2003) J Chromatogr B
796:209–224. doi:10.1016/j.jchromb.2003.
08.026
23. Le Rouzo G, Lamouroux C, Bresson C,
Guichard A, Moisy P, Moutiers G (2007)
J Chromatogr A 1164:139–144. doi:
10.1016/j.chroma.2007.06.053
24. Yoshida T (2004) J Biochem Biophys
Methods 60:265–280. doi:10.1016/j.jbbm.
2004.01.006
25. Berthod A, Chang SSC, Kullman JPS,
Armstrong DW (1998) Talanta 47:1001–
1012. doi:10.1016/S0039-9140(98)00179-9
26. Snyder LR, Kirkland JJ, Glajch JL (eds.)
(1997) In: Practical HPLC method
development, Wiley–Interscience, New
York
27. Poole CF (ed) (2003) The essence of
chromatography. Elsevier, Amsterdam
28. Snyder LR, Poppe H (1980) J Chromatogr 184:363–413. doi:10.1016/S00219673(00)93872-X
29. Nikitas P, Pappa-Louisi A, Agrafiotou P
(2002) J Chromatogr A 946:33–45. doi:
10.1016/S0021-9673(01)01536-9
30. Jin G, Guo Z, Zhang F, Xue X, Jin Y,
Liang X (2008) Talanta 76:522–527. doi:
10.1016/j.talanta.2008.03.042
31. Hao Z, Xiao B, Weng N (2008) J Sep Sci
31:1449–1464. doi:10.1002/jssc.200700624
32. Philipsen HJA, Claessens HA, Lind H,
Klumperman B, German AL (1997) J
Chromatogr A 790:101–116. doi:10.1016/
S0021-9673(97)00739-5
33. Kayillo S, Dennis GR, Wormell P,
Shalliker RA (2002) J Chromatogr A
967:173–181. doi:10.1016/S0021-9673(02)
00753-7
34. Tran JV, Molander P, Greibrokk T, Lundanes E (2001) J Sep Sci 24:930–940. doi:
10.1002/1615-9314(20011201)24:12<930:
AID-JSSC930>3.0.CO;2-2
35. Guo Y, Srinivasan S, Gaiki S (2007)
Chromatographia
66:223–229.
doi:
10.1365/s10337-007-0264-0
36. Dong L, Huang J (2007) Chromatographia 65:519–526. doi:10.1365/s10337007-0200-3
37. Gokel GW (ed) (2004) Dean’s handbook
of organic chemistry, 2nd edn. McGrawHill, New York
Chromatographia 2010, 71, May (No. 9/10)
759