Journal of the American Chemical Society
Article
γ-aminobutyric acid, phenibut, can also be well separated
affording α/Rs = 1.74/2.46 (Figure 5b). 3,4-Dihydroxypheny-
lalanine (Dopa), a drug for the treatment of Pakinson’s disease,
can be separated with α/Rs = 1.65/1.76 (Figure 5c). Besides,
CE-1-based CSP can also separate the agricultural chemical
glufosinate ammonium (phosphinothricin) affording α/Rs =
1.84/1.40 (Figure 5d). These results demonstrate the potential
versatility of CE-1 for the separation of diverse chiral drugs.
To gain further insight into the separation of chiral
molecules, we prepared a column packed with Me4L1 and
evaluated its chiral recognition ability as CSP. The as-prepared
column can separate amino acids with good to excellent
separation resolution, affording α/Rs = 1.19/0.91 to 1.83/3.37
(Figure S14b and Table 1). This suggested that the crown
ether moieties play a significant role in the separation process.
However, as compared to the CE-1 column, the Me4L1 column
typically showed lower separation factor and resolution. This
may be attributed to the fact that the framework can exert extra
confinement environment and thus lead to an enhancement of
enantiospecific interaction. To examine the importance of
crown ether moieties, we have synthesized an isoreticular
framework OET-1 with OEt groups instead of crown ether
moieties. OET-1 also showed excellent chemical stability
However, the OET-1 packed column cannot separate any of
the analytes under the same conditions (Figure S14c and Table
1). Besides, control experiments showed that the pure C18
column does not have the separation ability toward amino
acids (Figure S14e). This proves the significant role of crown
ether groups in enantioselective recognition. To study the
confined space effect of the framework, we used the bulky 3,5-
dibenzyloxy-1-phenylalanine methyl ester with a size of 14 Å ×
16 Å as the analyte (Figure S11). It can be separated in the
Me4L1 packed column with α/Rs = 1.21/1.19, but cannot be
baseline separated in the column packed by CE-1 with a
maximum channel size of 8.5 Å × 14.9 Å (Figure S14d). Taken
together, we can speculate that the crown ether groups
combined with the confined space of the crystalline framework
exert an important influence in chiral recognition.
To examine the loading capacity of the CSP without
compromising resolution, a loading test was carried out using
different inject masses. When the loading was increased from 4
to 20 μg of each racemate, Phg can still achieve baseline
separation (Figure 5b). Besides, the chromatographic peak area
of each single antipode rises in line with the increase of the
injected mass (Figure S13). It is observed that the retention
time of the enantiomer became closer with increasing injection
mass, although it did achieve baseline resolution. The result
may be due to the strong interaction between amine species
and crown ether groups.
To better understand the separation process, the thermody-
namics of Phg-OMe separation was investigated at temper-
atures ranging from 288 to 308 K (Figure 5c). It was observed
that the retention times of the analytes were decreased as the
column temperature increased. The van’t Hoff plots for the
analytes exhibited excellent linear correlation, and this
suggested no changes of the interactions during the separation
process (Figure 5d). The molar adsorption enthalpy/entropy
changes ΔadsHm/ΔadsHm of (D)- and (L)-protonated Phg-OMe
are −3.64 and −11.30 kJ/mol, respectively. The differences
Δ(ΔadsHm)L−D and Δ(ΔadsSm)L−D are calculated as −7.66 kJ/
mol and −70.66 J/(mol K), and this suggests that the
separation process is enthalpy driven. In addition, the
adsorption enthalpy results are consistent with the order of
retention times and the elution sequence of the D- and L-
enantiomers.
To microscopically elucidate the interactions between the
crown ether group in (S)-CE-1 and L/D-protonated Phg-OMe,
density functional theory (DFT) calculations were performed.
As shown in Figure S15, there are two configurations of the
crown ether groups in (S)-CE-1 with the dihedral angles of
two center phenyl rings at about 64° or 112°, respectively.
However, geometric optimization of the host models (Figures
S17 and S18) suggests that only the configuration with the
dihedral angles of two center phenyl rings at 64° could provide
adequate space for the favorable binding of L/D-enantiomers.
Figure 6 depicts the simplified structures of the most stable
Figure 6. Side (upper) and top (lower) views of the lowest-energy
structures of L- and D-protonated Phg-OMe with (S)-CE-1. The
Gibbs binding energies at 298 K are indicated in the parentheses. The
lengths (in Å) of hydrogen bonds involved in the host−guest
interactions and the dihedral angles of the two center phenyl rings are
labeled. The MOF frameworks surrounding the host−guest
complexes are omitted for clarity. C, gray; H, white; O, red; and N,
dark blue.
conformers of L/D-protonated Phe-OMe with (S)-CE-1. The
corresponding structures with the surrounding frameworks can
be found in Figure S20. Both the L- and the D-protonated Phg-
OMe bind to the crown ether group in (S)-CE-1 through three
hydrogen bonds. At room temperature, the predicted Gibbs
binding energy for the L-enantiomer is 2.9 kcal/mol higher
than that for the D-enantiomer. This demonstrates a marked
chiral recognition of the L-enantiomer relative to the D-
enantiomer by (S)-CE-1, as attributed to the favorable host−
guest binding through hydrogen bonds at optimum distances
and collinear orientations in the confined framework.
CONCLUSIONS
■
We have designed and prepared three chiral crown ether
embedded Zr-MOFs with a flu network topology based on
three enantiopure 1,1′-biphenol-derived tetracarboxylate link-
ers. The obtained porous Zr-MOFs displayed highly chemical
stability and can serve as robust CSPs for RP-HPLC
separation. Thus, a range of racemic N-containing compounds
including amino acids, amino esters, and some drugs were well
separated under acidic eluent conditions, with separation
G
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX