Content uploaded by Daniel W. Armstrong
Author content
All content in this area was uploaded by Daniel W. Armstrong on Oct 18, 2017
Content may be subject to copyright.
Gone in Seconds: Praxis, Performance, and Peculiarities of Ultrafast
Chiral Liquid Chromatography with Superficially Porous Particles
Darshan C. Patel,
†
Zachary S. Breitbach,
†
M. Farooq Wahab,
†
Chandan L. Barhate,
†
and Daniel W. Armstrong*
,†,‡
†
Department of Chemistry and Biochemistry, University of Texas at Arlington, Arlington, Texas 76019, United States
‡
AZYP LLC, 700 Planetarium Place, Arlington, Texas 76019, United States
*
SSupporting Information
ABSTRACT: A variety of brush-type chiral stationary phases
(CSPs) were developed using superficially porous particles (SPPs).
Given their high efficiencies and relatively low back pressures,
columns containing these particles were particularly advantageous for
ultrafast “chiral”separations in the 4−40 s range. Further, they were
used in all mobile phase modes and with high flow rates and
pressures to separate over 60 pairs of enantiomers. When operating
under these conditions, both instrumentation and column packing
must be modified or optimized so as not to limit separation
performance and quality. Further, frictional heating results in axial
thermal gradients of up to 16 °C and radial temperature gradients up
to 8 °C, which can produce interesting secondary effects in
enantiomeric separations. It is shown that the kinetic behavior of
various CSPs can differ from one another as much as they differ from the well-studied C18 reversed phase media. Three
additional interesting aspects of this work are (a) the first kinetic evidence of two different chiral recognition mechanisms, (b) a
demonstration of increased efficiencies at higher flow rates for specific separations, and (c) the lowest reduced plate height yet
reported for a CSP.
For much of 3 decades, the development and study of
chromatographic enantiomeric separations have been
dominated by investigations focused on selectivity. This is
not surprising given the unique position of chiral separations in
chromatography where conventional strategies used for all
other molecules are completely ineffective for enantiomers.
Hence, the highest impact studies involved conceiving,
understanding, and optimizing the use of new and better chiral
selectors.
1−20
Numerous thermodynamic and mechanistic
studies as well as evaluations of solvent and additive effects
continue even today.
21−24
As the field of “chiral separations”has matured, it has focused
on other, more typical chromatographic concerns including
speed, efficiency, and kinetic effects. While “chiral separations”
are ultimately affected by the same parameters as achiral
separations, they can have some idiosyncrasies (vide infra)as
compared to the most extensively studied systems which
typically involve reversed phase liquid chromatography on C18
or analogous stationary phases.
The demand for fast and efficient achiral separations
provided the impetus for researchers to explore new avenues
to increase throughput of chiral screening and analysis. Welch
et al. first used multiparallel chiral screening and method
development systems that provided method development times
of ∼1h.
25
Hamman et al. used supercritical fluid chromatog-
raphy (SFC) at high flow rates, short columns, and a gradient
to obtain a 2.5 min screening method.
26
Shortly after, Ai and
co-workers developed a bonded sub-1 μm mesoporous silica
based cyclodextrin chiral column and published a few 1−6 min
enantiomeric separations.
27
Concurrently, Gasparrini et al.
studied a bonded brush-type (pi-complex) phase using sub-2
μm fully porous particles (FPPs) and demonstrated a few
normal phase enantiomeric separations in the 15−40 s
range.
28−30
More recently, superficially porous particles (SPPs) for
achiral separations have allowed for column efficiencies
comparable to sub-2 μm FPPs while using conventional
HPLCs and column hardware.
31,32
There have been numerous
empirical and theoretical comparisons of these approaches
when used in a reversed phase (C18) format.
33−35
SPPs are
able to decrease all contributions to band broadening (i.e.,
longitudinal diffusion, eddy dispersion, and stationary phase
mass transfer contributions).
35
Initially it was thought that
better packing of SPPs was due to their having narrower
particle size distributions than FPPs, but it was later shown that
better packing homogeneity across the column (i.e., from wall
to center of the bed) is largely responsible for the decreased
eddy dispersion contribution.
36,37
Since, SPP columns are
Received: February 20, 2015
Accepted: May 6, 2015
Published: May 6, 2015
Article
pubs.acs.org/ac
© 2015 American Chemical Society 9137 DOI: 10.1021/acs.analchem.5b00715
Anal. Chem. 2015, 87, 9137−9148
This is an open access article published under an ACS AuthorChoice License, which permits
copying and redistribution of the article or any adaptations for non-commercial purposes.
generally better packed than FPP columns, they can yield
reduced plate heights of 1.3−1.5 for columns packed with
conventional achiral stationary phases, whereas FPP based
columns typically have reduced plate heights greater than 2.0.
33
Also, the shell thickness of SPPs leads to a shorter trans-particle
path length which can decrease mass transfer contributions to
band broadening for large molecules with small diffusion
coefficients and smaller molecules that have slow adsorption−
desorption kinetics.
31,38,39
This is particularly important at
higher flow rates.
The possible benefits of SPPs in other important but more
specialized areas of LC are less explored. Chankvetadze
compared a polysaccharide based chiral selector coated on
FPPs and SPPs in both nano-LC and HPLC.
40,41
In the latter,
an obvious decrease in enantiomeric selectivity was noted for
the SPP based material. Gritti and Guiochon’s theoretical
treatment of the same polysaccharide based chiral selector
indicated that a 10% gain in resolution (Rs) could be possible
due to the decreased plate heights afforded by the SPPs.
42
However, this estimated value was based on an assumption that
the SPP based polysaccharide column would have a similar
enantiomeric selectivity value as the analogous FPP based
column which, as noted, has not been obtainable to date. Most
recently, Spudeit et al. presented the first successful and
practical SPP CSP.
39
This work showed that brush-type chiral
selectors (i.e., isopropylcarbamated cyclofructan 6) have a
higher chiral selector load (per surface area). This plus the
observed increase in column efficiency for the SPP based CSP
resulted in improved enantiomeric separations, while maintain-
ing the same enantiomeric selectivity as FPP based CSPs.
39
Further, the SPP CSP maintained this performance with 50−
75% lower retentions. When comparing constant retention
modes, the Rsvalues obtained using the SPP column were far
greater than the FPP columns. It was also noted that as flow
rates increased (e.g., to 3 mL/min), the resolution per analysis
time greatly improved for the SPP column (by 18−67%)
meaning that high-throughput screening would benefit from
such columns.
39
In this work, a broad range of analyte types and polarities
including pharmaceuticals, catalysts, peptides, amino acids,
primary amines, and biaryls among others were baseline
separated on a variety of SPP brush type CSPs that are very
effective for ultrafast chiral separations (∼4−40 s range). It is
demonstrated that they can be performed in any mobile phase
conditions or mode, i.e., reversed phase, normal phase, polar
organic, and HILIC. Finally, the practice of ultrafast chiral LC
often produces interesting and unusual consequences that must
be recognized, dealt with, and/or properly understood for
optimal performance.
■EXPERIMENTAL SECTION
Materials. All HPLC solvents and reagents for reactions
were purchased from Sigma-Aldrich (St. Louis, MO). Cyclo-
fructan-6 (CF6) and cylcofructan-7 (CF7) derivatized with
isopropyl carbamate and dimethylphenyl carbamate groups,
respectively, were synthesized by AZYP LLC (Arlington, TX).
The 2.7 μm superficially porous silica particles with a surface
area of 120 m2/g and pore size of 120 Å were provided by
Agilent Technologies (Wilmington, DE). The core is 1.7 μmin
diameter and the surrounding shell thickness is 0.5 μm. The
fully porous 2.1 and 3 μm silica particles were purchased from
Daiso (Tokyo, Japan) and Glantreo (Cork, Ireland),
respectively. The 2.1 μm fully porous particles have a surface
area of 479 m2/g and pore size of 91 Å, whereas the 3 μm fully
porous particles have a surface area of 300 m2/g and pore size
of 100 Å. Tröger’s bases were obtained as indicated in the
literature.
43
All solvent mixtures are given in (v/v).
Synthesis of Stationary Phases. All reactions were
carried out in anhydrous solvents under a dry argon
atmosphere. The synthetic procedures for the six stationary
phases employed in this work have already been pub-
lished.
10,13,16,18,44
The first chiral stationary phases explored
were cyclofructan based as they have recently proven to be
unique chiral selectors.
18,45−50
The cyclofructan-6 derivatized
isopropyl carbamate (CF6-P) and cyclofructan-7 dimethyl-
phenyl carbamate (CF7-DMP) were bonded to silica particles
under anhydrous conditions as described previously.
18
The 2-
hydroxypropyl-β-cyclodextrin bonded silica (CD-HP) was
synthesized via a proprietary bonding procedure.
10,44
Macro-
cylic antibiotics vancomycin, teicoplanin, and teicoplanin
aglycone were covalently attached to silica surface as described
by Armstrong et al.
13,16
Each of the above chiral selectors were
bonded to 2.7 μm SPPs. The 2.1 and 3 μm fully porous
particles were functionalized with CF6-P. The CHN analyses of
the phases and chiral selector coverage per surface are provided
in Table S2 in the Supporting Information.
Each stationary phase was slurry packed with a pneumatically
driven Haskel pump (DSTV-122) into 10 cm ×0.46 cm i.d., 5
cm ×0.46 cm i.d., and 3 cm ×0.46 cm i.d. stainless steel
columns (IDEX Health and Science, Oak Harbor, WA). See the
Supporting Information for the packing method (Figure S1).
Commercial LARIHC CF6-P, LARIHC CF7-DMP, Chirobiotic
V, Chirobiotic T, Chirobiotic TAG, and Cyclobond I 2000 HP-
RSP columns (fully porous 5 μm particles, 25 cm ×0.46 cm
i.d.) which were used for comparative purposes were provided
by AZYP LLC, Astec, and Supelco/Sigma-Aldrich.
Instrumentation. All ultrafast separations were performed
on an Agilent 1290 Infinity series UHPLC system (Agilent
Technologies, Santa Clara, CA) equipped with a quaternary
pump, an auto-sampler, and a diode array detector. Routine
separations were performed on an Agilent 1260 HPLC
equipped with a quaternary pump, an auto-sampler, and a
diode array detector. An inlet filter was installed between the
pump outlet and the auto-sampler injection valve to prevent
clogging of 75 μm tubings. For fast separations, the data
collection rate was set at 160 Hz with a response time of 0.016
s, unless otherwise stated. The thermostated column compart-
ment and the column switching 6-port valve were bypassed to
minimize the length of connection tubings. The instrument was
further optimized to reduce extra-column effects by using an
ultralow dispersion kit from Agilent (P/N 5067-5189). The kit
consists of an ultralow dispersion needle and needle seat, two
75 μm i.d. stainless steel connection tubings, and a detector
flow cell with a volume standard deviation V(σ) of 0.6 μL.
Alternatively, 75 μm i.d. polyether ether ketone (PEEK)
nanoViper connection tubings (Thermo Fisher Scientific, MA)
were also employed. The instrument was controlled by
OpenLAB CDS ChemStation software (Rev. C.01.06 [61],
Agilent Technologies 2001−2014) in Microsoft Windows 8.1
(see the Supporting Information for the calculation of peak
parameters). The reported percentages of mobile phases (m.p.)
are listed as volume/volume (v/v).
Axial Temperature Gradient in Mobile Viscous Fric-
tional Heating. The effect of viscous frictional heating of the
mobile phases in the SPP columns was studied by wrapping the
column in an insulating sheet (at room temperature) and
Analytical Chemistry Article
DOI: 10.1021/acs.analchem.5b00715
Anal. Chem. 2015, 87, 9137−9148
9138
inserting a Mastech thermocouple MS8222H (Pittsburgh, PA)
inside the column outlet with the help of a screw cap. The flow
rate was varied and the resulting temperature was monitored
after 10 min of equilibration.
■DISCUSSION
Figure 1A provides comparisons of different particle size fully
porous particles (FPPs) and superficially porous particles
(SPPs) which have the same bonded chiral selector (via the
same chemistry) and with the same mobile phase. These
chromatograms were generated using conventional HPLCs
with conventional conditions and column sizes (i.e., 1.0 mL/
min flow rate and 5 cm ×0.46 cm i.d. columns). Clearly using
the same mobile phase, the SPP-based CSP provided both the
greatest efficiency and shortest analysis time as compared to all
FPPs, including the 2.1 μm particles (Figure 1A). However,
according to Gritti and Guiochon, a better comparison of such
columns is realized when resolutions (RS) are compared at
constant retention (Figure 1B).
42
They also indicated that a
SPP’s core to particle diameter ratio (ρ) can be related to its
gain in resolution. Specifically ρvalues between 0.5 and 0.95 at
constant retention factor (k) can slightly improve the
resolution. Interestingly, recent work on a brush-type CSP
showed a resolution increase of 20%.
39
In Figure 1B, the
increase in resolution of the SPP-CSP over both 3 and 2.1 μm
FPPs is ∼30%, which is quite impressive. The SPP-CSP used
here had a ρvalue of 0.63 (see Experimental Section), which is
within the optimal range (vide supra).
42
A direct comparison of
the efficiencies, reduced plate heights, and tailing factors of
current commercial columns (5 μm particles) and the
analogous CSPs on 2.7 μm SPPs is given in Table 1. The 3−
4-fold increase in efficiencies is impressive but not totally
unexpected given the smaller SPP particle diameter. However,
the reduced plate heights of the SPPs also are up to 2 times
smaller and with comparable or better peak symmetries. The
reduced plate height (h) of 1.6 for the CF7-DMP SPP is the
smallest reported for any CSP on any particle to date. Given
these results, it is clear that SPP based CSPs should be
particularly advantageous for ultrafast chiral separations.
In the literature, the current accepted time limit for being
labeled as an ultrafast chromatographic separation seems to be
∼1 min.
51,52
This is probably a reasonable choice since typical
HPLC autoinjectors cycle at ∼1 injection per min (or down to
0.5 min for UHPLC). Hence in ultrafast LC, the chromato-
graphic separations can be completed more quickly than new
samples can be injected (by conventional injection devices).
Figure 2 and Table 2 show over 60 such baseline enantiomeric
separations. The table covers a wide structural variety of
enantiomers. Most separations are <40 s and almost a quarter
of those are on the order of 10 or fewer seconds. Furthermore,
these are accomplished in all mobile phase modes, i.e., normal
phase, reversed phase, and polar organic modes and on a variety
of bonded CSPs. Theoretically, we could screen ∼90 to 360
chiral analytes per hour which could use less solvent than any
Figure 1. Enantiomeric separations of BINAM on CF6-P bonded to
SPPs and FPPs at 1.0 mL/min, Tcol = 25 °C. All columns were 5 cm
×0.46 cm in dimensions. (A) Constant MP mode, MP = 92:8
heptane−ethanol. (B) Constant retention mode, MP = (i) 82:18
heptane−ethanol, (ii) 85:15 heptane−ethanol, (iii) 82:18 heptane−e-
thanol, and (iv) 92:8 heptane−ethanol.
Table 1. Comparison of Theoretical Plates/Meter (N/m),
Reduced Plate Height (h), and USP Tailing Factor Using a
Standard Achiral Probe 1,3-Dinitrobenzene with 70:30
Heptane−Ethanol at Reduced Velocity of 4.5 (1 mL/min for
2.7 μm SPP, 0.6 mL/min for 5 μm FPP)
stationary phase N/m
a
htailing factor
b
Stationary Phases Bonded to 2.7 μm SPPs
CF6-P
c
172 000 2.2 1.1
CF7-DMP
d
221 000 1.6 1.2
teicoplanin
d
165 000 2.3 1.0
teicoplanin aglycone
c
133 000 2.8 1.3
vancomycin
c
173 000 2.1 0.9
hydroxypropyl-β-cyclodextrin
d
181 000 2.0 1.3
Commercial Columns Packed with 5.0 μm FPPs (25 cm ×0.46 cm)
LARIHC CF6-P 70 000 2.9 1.1
LARIHC CF7-DMP 59 000 3.4 1.2
Chirobiotic-T 54 000 3.7 0.9
Chirobiotic-TAG 50 000 4.0 1.1
Chirobiotic-V 57 000 3.5 0.9
Cyclobond I 2000 HP-RSP 37 000 5.4 1.1
a
N/m calculated by half height method.
b
USP tailing factor T=W0.05/
2f, where W0.05 = peak width at 5% peak height and f= distance from
the leading edge of the peak to the peak maximum at 5% peak height.
c
Dimensions of these columns were 10 cm ×0.46 cm.
d
Dimensions of
these columns were 5 cm ×0.46 cm.
Analytical Chemistry Article
DOI: 10.1021/acs.analchem.5b00715
Anal. Chem. 2015, 87, 9137−9148
9139
other current method. However, this is restricted to ∼60 to, at
most, 120 samples/hour because of instrumental autoinjector
limitations. Certainly, this is not the first nor the only example
of chromatographic potential being limited by instrumental
deficiencies.
51,53
Indeed, as discussed in the following para-
graphs, the separations shown in Figure 2 and listed in Table 2
cannot be achieved under standard HPLC conditions used for
Figure 1.
Effect of Packing on Columns Used for Ultrafast
Chiral LC. Accomplishing ultrafast separations in HPLC
generally requires higher flow rates, higher pressures, and
shorter columns. Consequently, both the column packing
quality and permeability are important. Commercial packing
procedures are usually trade secrets. When packing identical
columns with different slurry solvents, it was found that the use
of a “well dispersed”slurry produced columns of >2.3×higher
efficiencies and slightly different permeability according to
Darcy’s law (see Figures S1 and S2 and the associated text
followed in the Supporting Information).
Detector Sampling Rates and Response Times. The
detector sampling rate (a.k.a. sampling frequency, data
acquisition frequency or rate, etc.) and the detector response
time become increasingly important for rapidly eluting analytes
and highly efficient separations as demonstrated with SPPs.
Under certain circumstances, peak shapes, peak width, and
baseline noise can vary considerably as a result of detector
settings. There is some debate as to the exact cause and nature
of these effects.
54
We will address this debate in a subsequent
communication but will only present the empirical results, as it
impacts enantiomeric separations herein. Figure 3 shows the
effect of detector sampling rate and response time (for an
Agilent 1290 UHPLC) on the efficiency (N), resolution (RS),
and baseline noise for six ultrafast enantiomeric separations
performed under otherwise identical conditions. Note that with
Agilent ChemStation software, the detector sampling rate and
response times are coupled and the operator cannot
independently change or “unpair”these two parameters. The
observed effects are the combined result of these two
parameters. At the lowest sampling rate and longest response
time (bottom curve, Figure 3), the separation is not discernible,
the apparent efficiency and resolution is poor, but there is little
baseline noise. The separation parameters improve tremen-
dously as the sampling rate increases and the coupled time
constant decreases up to about the 80 Hz curve. Concurrently
the noise level increases (see 80×zoom in Figure 3). The
default setting on this instrument is 2.5 Hz. It should be noted
that with other instruments (Dionex and Shimadzu, for
example) the operator can independently set these detector
settings which could relate in an array of unwanted or
suboptimal combinations. It is apparent that to maintain high
Figure 2. Representative ultrafast enantiomeric separations on each of 6 chiral stationary phases: (A) vancomycin SPP (3 cm ×0.46 cm), MP =
methanol, 4.95 mL/min, Tcol =60°C; (B) teicoplanin aglycone SPP (3 cm ×0.46 cm), MP = methanol, 4.70 mL/min, Tcol =60°C; (C)
hydroxylpropyl-β-cyclodextrin SPP (5 cm ×0.46 cm), MP = 97:3:0.3:0.2 acetonitrile−methanol−TFA−TEA, 4.75 mL/min, Tcol =60°C; (D)
teicoplanin SPP (3 cm ×0.46 cm), MP = 40:60 water−methanol, 3.00 mL/min, Tcol =22°C; (E) CF7-DMP SPP (3 cm ×0.46 cm), MP = 90:10
heptane−ethanol, 4.80 mL/min, Tcol =22°C; (F) CF6-P SPP (10 cm ×0.46 cm), MP = 70:30:0.3:0.2 acetonitrile−methanol−TFA−TEA, 4.50
mL/min, Tcol =22°C.
Analytical Chemistry Article
DOI: 10.1021/acs.analchem.5b00715
Anal. Chem. 2015, 87, 9137−9148
9140
efficiency and good resolution when doing ultrafast separations
that detector coupled sampling rates should be ≥40 Hz and
response time ≤0.13 s (Figure 3). For enantiomeric separations
<10 s, even higher rates and lower times are needed. If one is
simply screening samples and concentration is not a factor, the
choice of detector settings are straightforward (e.g., 80 or 160
Hz). However, if one is examining either very low amounts of
an analyte or enantiomeric purities, the higher baseline noise
(top curve in Figure 3) can obscure low level enantiomeric
impurities (e.g., <1% and especially <0.1%) and decrease the
accuracy and precision of the measurement.
Extra Column Band Broadening Effects on Ultrafast
Separations. It is well established that extra column band
broadening is a concern when using short and/or narrow-bore
columns that often are packed with smaller diameter particles,
as in UHPLC.
55
In this regard, chiral separations are no
different, especially when doing ultrafast separations where it is
essential to maintain high efficiencies. Figure 4 illustrates this
assessment. A “stock”UHPLC was tested (top chromatogram,
Figure 4) and then the “extra column parts”of the instrument
were replaced with smaller volume versions. Using the variance
(σ2) calculated from second moment analysis, intrinsic column
efficiencies were calculated in each case, reflecting the true
column efficiency of 4750 plates for a 20 s separation (see the
Supporting Information). The σratio
2was also calculated using
the relationship σratio
2=σsystem
2/σcolumn+system
2. As can be seen, a
complete system optimization produced a decrease in the extra
column variance ratio from 26% to 3% and this resulted in an
ultrafast enantiomeric separation that went from ∼71 000
plates/m and a resolution of 1.4 to ∼94 000 plates/m and a
resolution of 1.7.
Kinetic and Thermal (Frictional) Considerations. Both
the general topic of column efficiency and the more specific
issue of frictional heating have been considered for columns
containing small particles (e.g., <2 μm diameter) and for
narrow bore columns.
56
Most of these studies focused on
reversed phase C18 based column formats.
57−61
There are few
kinetic studies on small particle and SPP chiral stationary
phases (CSPs) and none on the effect of frictional heating on
these CSPs.
40−42
As stated previously, CSPs are subject to the
same thermodynamic and kinetic constraints as other column
types. However, the manifestation of these kinetic terms can
Table 2. continued
a
All separations were performed on an Agilent 1290 UHPLC instrument optimized for low extra column volume. See Supporting Information for
more information on RS. Column dimensions for all separations were 3 cm ×0.46 cm and column temperature was ambient (∼22 °C) unless
otherwise stated.
b
T = teicoplanin, TAG = teicoplanin aglycone, CF7-DMP = Cyclofructan-7 dimethylphenyl carbamate, CF6-P = Cyclofructan-6
isopropyl carbamate, V = vancomycin, CD-HP = hydroxypropyl-β-cyclodextrin.
c
Dimensions of column = 5 cm ×0.46 cm.
d
Dimensions of column
=10cm×0.46 cm.
e
Data for the 1st eluted enantiomers.
f
Tcol =60°C.
Analytical Chemistry Article
DOI: 10.1021/acs.analchem.5b00715
Anal. Chem. 2015, 87, 9137−9148
9144
differ as much from one CSP to another as they do from
conventional C18 or silica gel stationary phases. Likewise, the
effect of frictional heating and column temperature gradients
has been evaluated and discussed for C18 reversed phase
columns.
57,59,61
For SPP-based CSPs, differences as well as any
peculiarities can be revealed by any of the related kinetic plots
(van Deemter, reduced van Deemter, or Knox).
56
For the
purpose of this discussion, we will use the standard Giddings’
coupled van Deemter equation of:
=+ + + +
−
⎛
⎝
⎜⎞
⎠
⎟
H
B
u
Cu C
AC
11
SSM
M
1
(1)
where His the height equivalent to a theoretical plate, Ais the
eddy dispersion term, Bis the longitudinal diffusion term, CSis
stationary phase mass transfer, CSM is mass transfer in the
stagnant mobile phase (sometime treated as “short range”eddy
dispersion), CMis the moving mobile phase mass transfer term,
and uis the linear velocity (m/s) of the mobile phase.
62
Figure 5 shows four unique sets of van Deemter plots done
in the (A) polar organic mode, (B) normal phase mode, and (C
and D) in the reversed phase mode under two different
temperature conditions. Each set of curves contains one pair of
enantiomers and at least one achiral test molecule. The
experimental conditions are given in the legend. The solvent
temperature at the column outlet was measured at different
linear velocities and mobile phase modes (see the Experimental
Section and Supporting Information).
The “polar organic”plots in Figure 5A indicate what some
would consider to be a normal “well behaved”system. The
achiral void volume marker (1,3-dinitrobenzene) has the lowest
H at all linear velocities above ∼0.5 mm/s and the flattest rise
at higher velocities. The least retained (first eluted) enantiomer
and a retained achiral analyte (nicotinamide) had almost
identical efficiencies at all linear velocities and similar, slightly
greater slopes at higher linear velocities. The most retained
enantiomer is generally thought to have the greatest resistance
to stationary phase mass transfer as it is subject to a greater
number of associative stereochemical interactions and often
reorientation of the enantiomer.
22,63
This appears to be so as
the Hmin is at a slightly lower linear velocity for the second
enantiomer compared to the first enantiomer and the achiral
probe, indicating an increase in the CSterm.
Figure 5B shows the analogous plots for the enantiomers of
Tröger’s base as well as retained and unretained achiral probe
molecules in the normal phase mode. The relative kinetic
behaviors of these molecules are quite different than those in
Figure 5A. The plots of the enantiomers are almost identical at
all linear velocities. However, this behavior is believed to be
related to two different things, one of which relates to the
stereochemical recognition mechanism while the other is
related to general column properties. The similar kinetic
behaviors of the two enantiomers indicate that chiral
recognition is likely due to the presence of repulsive (steric)
interactions rather than multiple associative interactions with
one of the enantiomers. For example, the minimum 3-point of
interaction needed for chiral recognition could come from one
associative interaction plus 2 steric interactions with one of the
enantiomers. The only requirement of this model is that the
total energy of association be greater than that of the combined
steric repulsive interactions. Such systems have been proposed
previously, but this is the first time kinetic data has been used to
support such a scenario.
22,63
Also important is the relative behavior of the retained and
unretained achiral analytes in Figure 5B which is opposite to
that in Figure 5A. The unretained void volume marker (1,3,5-
tri-tert-butylbenzene) has worst efficiency at all linear velocities
but a flatter rise than the enantiomers at higher linear velocities.
The retained achiral molecule (1,3-dinitrobenzene) exhibited
the highest efficiency at all linear velocities and had the flattest
rise at higher linear velocities. This type of behavior has been
reported previously in a few instances for well packed, high
efficiency columns.
55,64
The van Deemter curves in Figure 5B
were produced using a standard HPLC with a conventional
injector, tubings, column compartment, and detector flow cell.
When the extra column effects were minimized (Figure 4), the
observed efficiencies of the 1,3,5-tri-tert-butylbenzene and 1,3-
dinitrobenzene were nearly identical. This clearly illustrates the
pronounced effects of extra column band broadening on
Figure 3. Effect of detector sampling rate and response time on
efficiency (N) and resolution (Rs) in ultrafast chromatographic
separations. BINAM analyzed on CF7-DMP SPP (3 cm ×0.46 cm),
MP = 90:10 heptane−ethanol, 4.0 mL/min, Tcol =22°C;1Hz=1s
−1.
Figure 4. Optimization of Agilent 1290 UHPLC for ultrafast
separations by replacing stock parts with low extra column volume
alternatives. Tröger’s base analyzed on CF7-DMP SPP (5 cm ×0.46
cm), MP = 70:30 heptane−ethanol, 2.5 mL/min, Tcol =22°C. Percent
extra column contribution is expressed as σratio
2=σsystem
2/σcolumn+system
2
(see the Supporting Information for moment analysis). (A) Stock
condition: stock injection needle and needle seat, 170 μm i.d.
connection tubing (22 cm total) with IDEX 10-32 finger tight fittings,
and a 1.0 μL detector flow cell. (B) Optimized conditions: ultralow
dispersion needle and needle seat, 75 μm i.d. nanoViper connection
tubing (22 cm total), 0.6 μL detector flow cell.
Analytical Chemistry Article
DOI: 10.1021/acs.analchem.5b00715
Anal. Chem. 2015, 87, 9137−9148
9145
observed efficiencies in such van Deemter curves. Indeed, the
highest efficiency column (CF7-DMP with a reduced plate
height (h= 1.6)) was chosen for this example in an ultrafast
format. Clearly under these conditions, one must be aware at all
times of extra column effects and how they can generate
apparent anomalous behaviors.
56
Figure 5C,D is for the same reversed phase enantiomeric
separation and the same retained achiral analyte (1,3-
dinitrobenzene). The only difference in these two series of
experiments was that the column in Figure 5C was in a
thermostated, temperature controlled, “still air device”set at 25
°C, while for Figure 5D the column was in ambient conditions
(22 °C). It is well-known that teicoplanin chiral selectors
strongly and selectively bind D-amino acid enantiomers and
that this leads to greater resistance to mass transfer and broader
peaks. This is confirmed by the upper plots for the more
retained D-homophenylalanine in Figure 5C,D. Indeed no H
minima vs linear velocity can be identified from these plots and
the efficiencies are lower than those in the other mobile phase
modes. It should be noted that such efficiencies can be greatly
improved by judicious use of specific additives, but that is not
the subject of this work. As in the polar organic mode, the
curves for the first eluted (least retained) enantiomer and the
achiral retained analyte (1,3-dinitrobenzene) are quite similar
to one another and both show minima in the 0.5−1 mm/s
region.
Perhaps the most striking aspect of these plots is the trend
shown in Figure 5D. At linear velocities higher than ∼2.5 mm/
s, the efficiencies of both enantiomers and 1,3-dinitrobenzene
begin to improve significantly. This effect is most pronounced
for the more retained D-homophenylalanine. It is well
documented that two types of temperature gradients develop
(axial and radial) when there is significant frictional
heating.
57,59,61,65
Eluents with the heat capacity and density of
mobile phases used in Figure 5 (acetonitrile, heptane, and
water) and operating pressures above 300 bar can easily
generate axial temperature differentials of 10 °C.
61
In fact, when
the flow averaged temperature was measured at the column
outlet at various linear velocities in three different modes, the
axial temperature differences ranged from 11 to 16 °C (see the
Supporting Information Tables S3−S5). This axial variation in
fast separations does not contribute to an increase in peak
width. On the other hand, the peak efficiency is significantly
affected by radial temperature gradients which change local
viscosities, velocity profiles, and diffusion coefficients of
analytes.
61,65
Afirst order “approximation”of the maximum
radial temperature difference ΔTRwhich can develop between
the column center and the column wall is given by
λ
Δ
=
()
T
uR
4
P
z
R
d
d
2
rad (2)
where uis the superficial flow velocity in m/s (obtained by
dividing the volumetric flow rate by the total cross sectional
area of the column), dP/dzthe change in pressure in the
direction of the column axis (z) per unit length in N/m3,Rthe
column radius in m, and λrad is the approximate thermal
conductivity of the mobile phase in the radial direction in W/m
°C.
65
Figure 5. van Deemter plots for chiral and achiral analytes in polar organic mode, normal phase, and reversed phase on 2.7 μm SPP CSPs. (A) CF6-
P SPP (10 cm ×0.46 cm i.d.), MP = 80:20:0.3:0.2 acetonitrile−methanol−TFA−TEA, Tcol =25°C (thermostated). (B) CF7-DMP SPP (10 cm ×
0.46 cm), MP = 90:10 heptane−ethanol, Tcol =25°C (thermostated). (C) Teicoplanin bonded SPP (5 cm ×0.46 cm), MP = 90:10 water−
methanol, Tcol =25°C (thermostated). (D) Tcol =22°C (not thermostated), other conditions were identical to part C. See the Supporting
Information for temperature effects on selectivities. The kvalues reported are for a flow rate of 1 mL/min.
Analytical Chemistry Article
DOI: 10.1021/acs.analchem.5b00715
Anal. Chem. 2015, 87, 9137−9148
9146
For example, in the normal phase mode (Figure 5B), the
thermal conductivity of the heptane−ethanol mixture is
approximately 0.13 W/m°C.
66
At low linear velocities, (1.67
mm/s or 1 mL/min, ΔP= 80 bar), the magnitude of the
maximum radial temperature difference is 1 °C; however, as the
linear velocity is increased to 5 mm/s (3 mL/min), the pressure
drop is significant (250 bar), and the calculated maximal radial
temperature gradient is 8 °C. Note that eq 1 is generally used
for first order approximations, it has been shown that the
calculated radial temperature gradients can overestimate the
observed radial gradients because it ignores the compressibility
of the eluent. Consequently the actual energy generated in the
column is reduced by a factor of 2/3.
59
On the other hand, as in
Figure 5D, when a water-rich mobile phase is in use (thermal
conductivity of 0.55 W/m°C), a linear velocity of 1.67 mm/s (1
mL/min) generated a back pressure of 112 bar due to higher
viscosity. The calculated value of ΔTRis only 1 °C, and at
higher linear velocities, e.g., 5 mm/s (3 mL/min), a radial
temperature difference of only 4 °C is developed. Also note
than the axial temperature difference in Figure 5B,D was similar
(∼12 °C). However, the data used in Figure 5B was from a
thermostated column (walls ∼25 °C) while Figure 5D was not
thermostated. Though, since heptane (Figure 5B) is far more
compressible than water (Figure 5D), the energy produced is
reduced by 2/3. However, it is clear from Figure 5D, that there
are other factors, as in some chiral separations when resistance
to mass transfer effects are more pronounced. In these
interesting cases, such as a high thermal conductivity water
rich mobile phase, the gain in efficiency from an improvement
in mass transfer at higher axial temperature gradients is enough
to visibly counter any smaller losses in efficiency due to radial
temperature gradients and eddy dispersion. This possibility was
noted early on by Halász
61
and is apparent in Figure 5D. See
the Supporting Information for detailed temperature measure-
ments and calculations.
■CONCLUSIONS
The results of this study, indicate that (1) SPPs are
advantageous for ultrafast and high efficiency chiral separations,
(2) enantiomeric separations on the order of few seconds are
now feasible in all mobile phases with bonded brush type CSPs,
(3) kinetic behaviors can sometimes be used to shed light on
chiral recognition mechanisms, (4) CSPs can show quite
different kinetic profiles from each other and from achiral
systems, (5) ultrafast chiral separations require optimized
detection and minimization of extra column effects, (6)
frictional heating effects must be accounted for in ultrafast
separations as they can manifest themselves in disparate ways
and to different degrees for various CSPs and mobile phase
modes, (7) efficiencies and separation speeds for chiral analytes
can now exceed those in capillary electrophoresis. Also it is
feasible to expect that (8) SPPs may be advantageous for
preparative separations when their high efficiencies, faster
analyses times, and reduced solvent consumption compensate
for lower chiral selector loading, (9) ultrafast SPP-CSPs may be
attractive as the second dimension in 2D-LC because of their
greater selectivity and orthogonality to conventional achiral
stationary phases, and (10) real-time monitoring of product
formation in asymmetric synthesis is possible with ultrafast
chiral separations.
■ASSOCIATED CONTENT
*
SSupporting Information
Peak parameter calculations, surface coverage of chiral selectors
on silica, column permeability calculations, determination of
extra-column contributions, and frictional heating measure-
ments data. The Supporting Information is available free of
charge on the ACS Publications website at DOI: 10.1021/
acs.analchem.5b00715.
■AUTHOR INFORMATION
Corresponding Author
*Phone: (817) 272-0632. Fax (817)-272-0619. E-mail:
sec4dwa@uta.edu.
Notes
The authors declare no competing financial interest.
■ACKNOWLEDGMENTS
The authors would like to acknowledge Agilent Technologies
for providing the superficially porous particles. We also
acknowledge the support of AZYP, LLC, Arlington, Texas.
■REFERENCES
(1) Davankov, V. A.; Rogozhin, S. V. J. Chromatogr. A 1971,60, 280−
283.
(2) Mikes, F. E.; Boshart, G. J. Chromatogr. A 1978,149, 455−464.
(3) Pirkle, W. H.; Finn, J. M. J. Org. Chem. 1981,46, 2935−2938.
(4) Okamoto, Y.; Honda, S.; Okamoto, I.; Yuki, H.; Murata, S.;
Noyori, R.; Takaya, H. J. Am. Chem. Soc. 1981,103, 6971−6973.
(5) Allenmark, S.; Bomgren, B.; Borén, H. J. Chromatogr. A 1983,
264,63−68.
(6) Armstrong, D. W.; Ward, T. J.; Armstrong, R. D.; Beesley, T. E.
Science 1986,232, 1132−1135.
(7) Okamoto, Y.; Kawashima, M.; Yamamoto, K.; Hatada, K. Chem.
Lett. 1984, 739−742.
(8) Armstrong, D. W.; DeMond, W. J. Chromatogr. Sci. 1984,22,
411−415.
(9) Okamoto, Y.; Aburatani, R.; Fukumoto, T.; Hatada, K. Chem.
Lett. 1987, 1857−1860.
(10) Stalcup, A. M.; Chang, S. C.; Armstrong, D. W.; Pitha, J. J.
Chromatogr. A 1990,513, 181−194.
(11) Armstrong, D. W.; Stalcup, A. M.; Hilton, M. L.; Duncan, J. D.;
Faulkner, J. R., Jr.; Chang, S. C. Anal. Chem. 1990,62, 1610−1615.
(12) Pirkle, W. H.; Welch, C. J.; Lamm, B. J. Org. Chem. 1992,57,
3854−3860.
(13) Armstrong, D. W.; Tang, Y.; Chen, S.; Zhou, Y.; Bagwill, C.;
Chen, J.-R. Anal. Chem. 1994,66, 1473−1484.
(14) Laemmerhofer, M.; Lindner, W. J. Chromatogr. A 1996,741,
33−48.
(15) Maier, N. M.; Nicoletti, L.; Lammerhofer, M.; Lindner, W.
Chirality 1999,11, 522−528.
(16) Berthod, A.; Chen, X. H.; Kullman, J. P.; Armstrong, D. W.;
Gasparrini, F.; D’Acquarica, I.; Villani, C.; Carotti, A. Anal. Chem.
2000,72, 1767−1780.
(17) Gasparrini, F.; Misiti, D.; Rompietti, R.; Villani, C. J.
Chromatogr. A 2005,1064,25−38.
(18) Sun, P.; Wang, C. L.; Breitbach, Z. S.; Zhang, Y.; Armstrong, D.
W. Anal. Chem. 2009,81, 10215−10226.
(19) Sun, P.; Armstrong, D. W. J. Chromatogr. A 2010,1217, 4904−
4918.
(20) Han, X.; He, L.; Zhong, Q.; Beesley, T. E.; Armstrong, D. W.
Chromatographia 2006,63,13−23.
(21) Péter, A.; Török, G.; Armstrong, D. W.; Tóth, G.; Tourwé,D.J.
Chromatogr. A 1998,828, 177−190.
(22) Davankov, V. A. Chirality 1997,9,99−102.
(23) Gasparrini, F.; Misiti, D.; Villani, C. Chirality 1992,4, 447−458.
Analytical Chemistry Article
DOI: 10.1021/acs.analchem.5b00715
Anal. Chem. 2015, 87, 9137−9148
9147
(24) Oberleitner, W. R.; Maier, N. M.; Lindner, W. J. Chromatogr. A
2002,960,97−108.
(25) Sajonz, P.; Schafer, W.; Gong, X.; Shultz, S.; Rosner, T.; Welch,
C. J. J. Chromatogr. A 2007,1145, 149−154.
(26) Hamman, C.; Wong, M.; Aliagas, I.; Ortwine, D. F.; Pease, J.;
Schmidt, D. E., Jr.; Victorino, J. J. Chromatogr. A 2013,1305, 310−319.
(27) Ai, F.; Li, L.; Ng, S.-C.; Tan, T. T. Y. J. Chromatogr. A 2010,
1217, 7502−7506.
(28) Cancelliere, G.; Ciogli, A.; D’Acquarica, I.; Gasparrini, F.;
Kocergin, J.; Misiti, D.; Pierini, M.; Ritchie, H.; Simone, P.; Villani, C.
J. Chromatogr. A 2010,1217, 990−999.
(29) Kotoni, D.; Ciogli, A.; Molinaro, C.; D’Acquarica, I.; Kocergin,
J.; Szczerba, T.; Ritchie, H.; Villani, C.; Gasparrini, F. Anal. Chem.
2012,84, 6805−6813.
(30) Cavazzini, A.; Marchetti, N.; Guzzinati, R.; Pierini, M.; Ciogli,
A.; Kotoni, D.; D’Acquarica, I.; Villani, C.; Gasparrini, F. TrAC, Trends
Anal. Chem. 2014,63,95−103.
(31) Gritti, F.; Guiochon, G. LC−GC N. Am. 2012, 586−597.
(32) Omamogho, J. O.; Nesterenko, E.; Connolly, D.; Glennon, J.
LC−GC N. Am. 2012,63−69.
(33) DeStefano, J. J.; Langlois, T. J.; Kirkland, J. J. J. Chromatogr. Sci.
2008,46, 254−260.
(34) Gritti, F. LC−GC N. Am. 2014, 928−940.
(35) Broeckhoven, K.; Cabooter, D.; Desmet, G. J. Pharm. Anal
2013,3, 313−323.
(36) Bruns, S.; Stoeckel, D.; Smarsly, B. M.; Tallarek, U. J.
Chromatogr. A 2012,1268,53−63.
(37) Gritti, F.; Farkas, T.; Heng, J.; Guiochon, G. J. Chromatogr. A
2011,1218, 8209−8221.
(38) Dolzan, M. D.; Spudeit, D. A.; Breitbach, Z. S.; Barber, W. E.;
Micke, G. A.; Armstrong, D. W. J. Chromatogr. A 2014,1365, 124−
130.
(39) Spudeit, D. A.; Dolzan, M. D.; Breitbach, Z. S.; Barber, W. E.;
Micke, G. A.; Armstrong, D. W. J. Chromatogr. A 2014,1363,89−95.
(40) Fanali, S.; D’Orazio, G.; Farkas, T.; Chankvetadze, B. J.
Chromatogr. A 2012,1269, 136−142.
(41) Lomsadze, K.; Jibuti, G.; Farkas, T.; Chankvetadze, B. J.
Chromatogr. A 2012,1234,50−55.
(42) Gritti, F.; Guiochon, G. J. Chromatogr. A 2014,1348,87−96.
(43) Weatherly, C. A.; Na, Y.-C.; Nanayakkara, Y. S.; Woods, R. M.;
Sharma, A.; Lacour, J.; Armstrong, D. W. J. Chromatogr. B 2014,955−
956,72−80.
(44) Armstrong, D. W.; Li, W.; Chang, C. D.; Pitha, J. Anal. Chem.
1990,62, 914−923.
(45) Jiang, C. X.; Tong, M. Y.; Breitbach, Z. S.; Armstrong, D. W.
Electrophoresis 2009,30, 3897−3909.
(46) Zhang, Y.; Breitbach, Z. S.; Wang, C. L.; Armstrong, D. W.
Analyst 2010,135, 1076−1083.
(47) Perera, S.; Na, Y.-C.; Doundoulakis, T.; Ngo, V. J.; Feng, Q.;
Breitbach, Z. S.; Lovely, C. J.; Armstrong, D. W. Chirality 2013,25,
133−140.
(48) Padivitage, N. L. T.; Dodbiba, E.; Breitbach, Z. S.; Armstrong,
D. W. Drug Test. Anal. 2014,6, 542−551.
(49) Woods, R. M.; Patel, D. C.; Lim, Y.; Breitbach, Z. S.; Gao, H. Y.;
Keene, C.; Li, G. Q.; Kurti, L.; Armstrong, D. W. J. Chromatogr. A
2014,1357, 172−181.
(50) Woods, R. M.; Breitbach, Z. S.; Armstrong, D. W. LC−GC N.
Am. 2014,32, 742−751.
(51) Fekete, S.; Kohler, I.; Rudaz, S.; Guillarme, D. J. Pharm. Biomed.
Anal. 2014,87, 105−119.
(52) Wang, X.; Carr, P. W.; Stoll, D. R. LC−GC N. Am. 2010,28,
932−942.
(53) Elgass, H.; Engelhardt, H.; Halász, I. Fresenius’Z. Anal. Chem.
1979,294,97−106.
(54) Guiochon, G.; Sepaniak, M. J. Anal. Chem. 1991,63,73−73.
(55) Ma, Y.; Chassy, A. W.; Miyazaki, S.; Motokawa, M.; Morisato,
K.; Uzu, H.; Ohira, M.; Furuno, M.; Nakanishi, K.; Minakuchi, H.;
Mriziq, K.; Farkas, T.; Fiehn, O.; Tanaka, N. J. Chromatogr. A 2015,
1383,47−57.
(56) Usher, K. M.; Simmons, C. R.; Dorsey, J. G. J. Chromatogr. A
2008,1200, 122−128.
(57) de Villiers, A.; Lauer, H.; Szucs, R.; Goodall, S.; Sandra, P. J.
Chromatogr. A 2006,1113,84−91.
(58) Fountain, K. J.; Neue, U. D.; Grumbach, E. S.; Diehl, D. M. J.
Chromatogr. A 2009,1216, 5979−5988.
(59) Gritti, F.; Guiochon, G. Anal. Chem. 2008,80, 5009−5020.
(60) Gritti, F.; Martin, M.; Guiochon, G. Anal. Chem. 2009,81,
3365−3384.
(61) Halász, I.; Endele, R.; Asshauer, J. J. Chromatogr. A 1975,112,
37−60.
(62) Giddings, J. C. Dynamics of Chromatography: Principles and
Theory, 1st ed.; CRC Press: Boca Raton, FL, 2002.
(63) Boehm, R. E.; Martire, D. E.; Armstrong, D. W. Anal. Chem.
1988,60, 522−528.
(64) Gritti, F.; Guiochon, G. J. Chromatogr. A 2010,1217, 6350−
6365.
(65) Poppe, H.; Kraak, J. C.; Huber, J. F. K.; van den Berg, J. H. M.
Chromatographia 1981,14, 515−523.
(66) Qun-Fang, L.; Rui-Sen, L.; Dan-Yan, N.; Yu-Chun, H. J. Chem.
Eng. Data 1997,42, 971−974.
Analytical Chemistry Article
DOI: 10.1021/acs.analchem.5b00715
Anal. Chem. 2015, 87, 9137−9148
9148
