Full Papers
spectively (see the Supporting Information for the surface area
of the materials, Figure SI-22 and Table SI-3). However, in this
case, the difference in pore volume did not result from the
presence of bulky functional groups, but rather from the use
of longer ligands. Moreover, MIL-140(Zr)-type frameworks are
hydrophobic owing to the zirconium oxide nature of their sec-
ondary building units. Neither the polar LA nor water mole-
cules were thus expected to have a high affinity for their hy-
drophobic pore walls.[32] However, the larger channel size of
MIL-140C(Zr) compared to MIL-140A(Zr) allowed LA to enter
the channels with minimal contact with the pore walls, which
could account for the higher affinity for LA of MIL-140C(Zr)
(46 Lmolꢀ1) than of MIL-140A(Zr) (14 Lmolꢀ1).
Figure 3. Isotherms for adsorption of LA on UiO-66(Zr) at 298 K from aque-
ous solutions at natural pH value (grey) and from aqueous solutions adjust-
ed to pH 3.7 with Ca(OH)2 (red). The drawn curves are a guide to the eye.
The highest measured LA uptake by MIL-101(Cr)-type frame-
works (15-17 wt%) did not vary much with the pore function-
alization of the frameworks. The relative loss in pore volume in
the presence of bulky functional groups was substantially
lower for mesoporous materials (e.g., MIL-101) than for micro-
porous ones [e.g., UiO-66(Zr)]. Interestingly, the same relation
as for UiO-66(Zr)-type frameworks was found between LA affin-
ity and hydrogen-bonding properties of the pending groups:
MIL-101(Cr)-NH2 (11 Lmolꢀ1)>MIL-101(Cr)-NO2 (9 Lmolꢀ1)>un-
functionalized MIL-101(Cr) (6 Lmolꢀ1). Nevertheless, the much
higher affinity of LA for UiO-66(Zr) (35–73 Lmolꢀ1) than for
MIL-140(Zr) (14–46 Lmolꢀ1) and MIL-101(Cr)-type frameworks
(6–11 Lmolꢀ1) suggests that besides the effect of the functional
group, the nature of the secondary building units must play
a key role in the LA adsorption process.
lactate [Ca(Lact)2] complexes. The small amount of calcium de-
tected arose most probably from adsorption on the external
surface of the MOF particles.
This proves that UiO-66(Zr) can selectively adsorb LA from
the typical buffered medium. Nevertheless, the combined use
of acid-stable MOF adsorbents and pH-tolerant microorga-
nisms, such as yeasts, opens a perspective on LA production
and isolation in more acidic conditions, which would eventual-
ly eliminate the need for neutralization of the fermentation
broths with calcium bases.[6]
Rietveld refinement on UiO-66(Zr) loaded with lactic acid
In short summary, for the selective adsorption of LA from
aqueous solutions, MOFs based on hydroxylated zirconium
clusters are the most promising materials. Combining a high
affinity with a high capacity and stability, UiO-66(Zr) was select-
ed for further investigation of the adsorption and desorption
processes.
To gain more insight in the position of LA in the pores, a Riet-
veld refinement was performed on the PXRD pattern of UiO-
66(Zr) with 21 wt% LA loading (corresponding to an equilibri-
um concentration in solution of 0.075m). The refinement evi-
denced the presence of electron density in the framework that
could be assigned to LA (Figure 4). It suggested the presence
of approximately 3.6 LA molecules and 4.6 terephthalate struc-
tural ligands per cluster in the structure.
Lactic acid adsorption from buffer solutions
In the course of the LA fermentation, a calcium base is typically
added to the fermentation broth to maintain the pH value in
a range suitable for microbial growth. Therefore, the isotherm
for LA adsorption on UiO-66(Zr) was measured from solutions
buffered at pH 3.7, near the pKA of LA (3.86) (Figure 3). The iso-
therm reaches a plateau at lower equilibrium concentrations
and at a three times lower uptake for adsorption from buffer
solutions (12 wt% at 0.1m) than from non-buffered solutions
(40 wt% at 0.25m). Interestingly, the pH value of the solution
rose during adsorption (0.25 and 0.15 units for batch adsorp-
tion experiments from 0.015 and 0.15m buffer solutions, re-
spectively). This observation could be explained by the adsorp-
tion of LA in its protonated form, disturbing the equilibrium of
the LA dissociation reaction in solution and thus increasing the
pH value. Notably, adsorption of lactate molecules complexed
with calcium ions could be excluded. Elemental analysis was
performed for this purpose on UiO-66(Zr) to determine its cal-
cium content after a batch adsorption experiment from buffer
solution. A Ca2+/LA ratio of 1:14.3 was measured, which is sub-
stantially larger than the 1:2 expected ratio for neutral calcium
Two possible positions were identified for the adsorbed LA
(Figure 5). The first position is occupied by about 75% of the
LA and the second position by the remaining LA. Interestingly,
both positions of LA are only meaningful in terms of steric
demand if there is no ligand present, that is, if LA occupies
missing linker sites, often referred to as defects in the litera-
ture.[47] For both positions, the LA does not enter the primary
coordination sphere of the zirconium atoms and is most likely
hydrogen-bonded to the framework (displayed as blue
dashes). The parent UiO-66(Zr) material contains approximately
two missing linker sites per cluster, as evidenced by thermo-
1
gravimetric analysis (TGA) and H NMR spectroscopy (Figure
SI-23).
The refinement further revealed the presence of a crystalline
terephthalic acid phase corresponding to one terephthalate
molecule per cluster. This observation suggests that some
framework degradation (i.e., terephthalate removal) took place
during the initial LA adsorption. Given the extensive washing
(24 h soxhlet extraction in ethanol) and activation (12 h at
1508C) of the material prior to the adsorption experiment, no
&
ChemSusChem 2016, 9, 1 – 9
4
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!