The separation of styrene and ethylbenzene on
MOFs: analogous structures with different
adsorption mechanisms
Michael Maesa, Frederik Vermoortelea, Luc Alaertsa, Sarah Couckb Christine E. A.
Kirschhocka, Joeri F. M. Denayerb, Dirk E. De Vosa*
a
Centre for Surface Science and Catalysis, Katholieke Universiteit Leuven, Arenbergpark 23,
B-3001 Leuven (Belgium)
b
Department of Chemical Engineering, Vrije Universiteit Brussel, Pleinlaan 2, B-1050
Brussel (Belgium)
*
Corresponding Author: dirk.devos@biw.kuleuven.be, Tel: +32 16 321639, Fax: +32 16
321998
Table of contents
Single compound adsorption isotherms
S2
Regeneration of the columns
S3
Batch experiments on [Cu3(BTC)2]
S4
Calculation of apparent adsorption enthalpies
S5
Zero-coverage adsorption enthalpy
S7
Crystal structure analysis
S9
S1
Single compound isotherms on MIL-47 and MIL-53
Figure S1. Single compound adsorption isotherm of styrene (St) and ethylbenzene (EB) on
MIL-47 at room temperature: uptake (wt%) as a function of equilibrium concentration (M).
Figure S2. Single compound adsorption isotherm of styrene (St) and ethylbenzene (EB) on
MIL-53 at room temperature: uptake (wt%) as a function of equilibrium concentration (M).
S2
Regeneration of the columns
Regeneration of the columns is performed by flushing with 100 ml of pure aliphatic solvent
(i.c. heptane). As soon as the breakthrough of the mixture is complete and the effluent
concentrations of ethylbenzene and styrene are constant, pure heptane is pumped over the
column under the same experimental conditions. 1 ml of effluent is collected after every 10
ml of flushing and analyzed on GC for traces of ethylbenzene or styrene. The GC, equipped
with a FID detector can detect concentrations of hydrocarbons as low as 0.06 mM. The signal
gradually decreases during the flushing and it takes approximately 100 ml of pure heptane to
obtain a pure effluent free of both styrene and/or ethylbenzene (see Table S1 for regeneration
data on MIL-47). After this regeneration step, a second cycle of separation and regeneration is
started and this procedure is repeated for up to 10 times. No difference in capacities, elution
volumes, preferences and separation factors is observed between the different cycles, proving
that both MIL-47 and MIL-53 are fully regenerable.
Table S1. Concentrations of ethylbenzene (EB) and styrene (St) during the regeneration with
100 ml of pure heptane after the breakthrouh experiment performed in Figure 2a on MIL-47
(equimolar mixture of 0.047M St and 0.0047 M of EB in heptane).
elution volume (ml) concentration (M)
EB
St
0
0.046
0.046
20
0.018
0.045
40
0.006
0.015
60
0.001
0.004
80
0.000
0.001
100
0.000
0.000
S3
Batch experiments with [Cu3(BTC)2]: adsorption of an ethylbenzene-styrene mixture
contaminated with small amounts of toluene and ortho-xylene
As has been reported by Alaerts et al. in Angew. Chem. Int. Ed., 2007, 46, 4372, [Cu3(BTC)2]
shows little preference for compounds like o-xylene, especially when compared with
ethylbenzene. As the competitive batch results in Table S2 show, the preference of this
material displays the order:
styrene > ethylbenzene > o-xylene > toluene
These data were collected from those available in the abovementioned reference and from
additional batch experiments performed on a [Cu3(BTC)2] sample that was synthesized
according to the procedure in Patent WO 049892 A1, 2004.
Table S2. Separation factors αi,j calculated from the uptakes from binary mixtures (50:50) of
o-xylene (oX), toluene (TOL), ethylbenzene (EB) and styrene (St) out of heptane on
[Cu3(BTC)2] at a concentration of 0.028 M for each compound. The separation factor αi,j is
given as the prefence of compound i (left column) over compound j (right column) according
to the formulas provided in the experimental section of this paper. (Alaerts et al. in Angew.
Chem. Int. Ed., 2007, 46, 4372).
i
j
St
EB
EB
oX
oX
TOL
αi,j
5.9
1.4
4.4
S4
Calculation of apparent adsorption enthalpies
The relative retention of the different compounds depends on their adsorption equilibrium
constants. In liquid phase chromatography, the equilibrium constant Ki of a given compound
reflects the competition between the molecules of the mobile phase and the compound under
consideration for adsorption on the surface of the adsorbent. Thus, the following relationships
can be written:
α styrene / ethylbenzene =
µcorr, styrene
µcorr, ethylbenzene
=
K styrene
K ethylbenzene
(4)
with:
− RT ln K styrene = ∆Gads ,styrene − ∆Gads ,heptane
= ∆H ads,styrene − ∆H ads ,heptane − T (∆S ads,styrene − ∆S ads,heptane )
(5)
A similar equation can be written for ethylbenzene. Combining eqns 4 and 5 gives:
K styrene
K ethylbenzene
 ∆H ads ,styrene − ∆H ads ,ethylbenzene ∆S ads ,styrene − ∆S ads ,ethylbenzene 

= exp −
+
RT
R


(6)
As such, a weak dependence of the separation factor on temperature means that the adsorption
enthalpies of St and EB have comparable values, whereas a strong dependence of the
separation factor on temperature means that the adsorption enthalpies of St and EB have
significantly different values.
Based on the liquid phase pulse chromatographic data, changes in enthalpy are calculated
according to literature (Y. Zhang; V. McGuffin, Journal of Liquid Chromatography &
Related Technologies, 2007, 30, 1551). The retention factor k is defined as:
k=
(µi − µtc )
µtc
(7)
S5
and is obtained from single compound injections on the column. This retention factor is
related to the changes in molar enthalpy (∆H) and molar entropy (∆S) through the van ‘t Hoff
equation:
ln k = −
∆H ∆S
+
RT
R
(8)
with R being the gas constant and T the absolute temperature. Plotting ln k vs 1/T results in a
linear curve as shown in Figure S3 for MIL-47. ∆H can then be calculated from the slope of
this curve (slope*R = - ∆H); the obtained value is interpreted as an apparent adsorption
enthalpy as it comprises the total change of enthalpy caused by the interaction of both the
adsorbate and the solvent with the framework.
Figure S3. Representative graph of the logarithm of retention factor versus inverse
temperature for styrene (St) and ethylbenzene (EB) on a column of MIL-47. Linear regression
curves are plotted on the data to determine the slope and intersection.
S6
Zero-coverage adsorption enthalpy
Gas phase adsorption at low degree of pore filling was studied using the pulse
chromatographic technique with a 15 cm column (0.22 cm internal diameter) packed with
pellets (500 – 630 µm) of MIL-47 or MIL-53 (see also R. A. Ocakoglu, J. Denayer, G. Marin,
J. Martens, G. Baron, J. Phys. Chem. B 2003, 207, 398). Adsorption equilibrium Henry
constants were calculated from the first moment of the chromatographic response curves.
Adsorption enthalpies were obtained from the temperature dependence of the Henry
adsorption constants. The corresponding van ‘t Hoff plots are given in Figures S4 and S5. On
MIL-53, the difference in Henry adsorption constant is significant, with a preference for
adsorbing styrene. On MIL-47, both components adsorb almost equally.
Figure S4: van ‘t Hoff plot for the adsorption of ethylbenzene and styrene on MIL-53.
S7
Figure S5: van ‘t Hoff plot for the adsorption of ethylbenzene and styrene on MIL-47.
S8
Crystal structure analysis
The samples for X-ray powder diffraction (XRPD) were prepared by immersing 0.3 g MOF
crystallites in a pentane solution containing 0.15 M styrene or ethylbenzene at room
temperature. The crystallites were stirred to allow adsorption to take place. After two hours,
pentane is evaporated from the crystallites under inert nitrogen atmosphere. After evaporation
of pentane, the loaded crystallites were sealed in a capillary. Capillaries were measured on a
Stoe Stadi MP with focusing monochromator (CuKα1) in transmission geometry with a
position sensitive detector (6 °2θ) in a range between 3 and 80 °2θ and a resolution of 0.01
°2θ. Measurements occurred at room temperature. Rietveld refinement was performed with
the GSAS software (see: A. Larson, R. Von Dreele, General Structure Analysis System
(GSAS), Los Alamos National Laboratory Report LAUR, 2004, 86 -748).
MIL-47 loaded with styrene
The obtained pattern for MIL-47 could not be described with the spacegroup and cell
dimensions mentioned in literature (K. Barthelet, J. Marrot, D. Riou, G. Férey, Angew. Chem.
2002, 114, 291; Angew. Chem. Int. Ed. 2002, 41, 281). A symmetry analysis lowering the
symmetry from Pnma first to a cell tripled along channel direction a and then to Pn21a (a=
20.4563(33)Å, b= 15.6609(19)Å and c= 14.4447(16)Å; estimated standard deviations of the
last reported digits in brackets) allowed description of almost all reflections. Some reflections
not indexed in this setting were identified as scattering by small amounts (3 wt%) of the
structure in Pnma symmetry without tripling along a (lattice constants of a= 6.8285(7)Å, b=
14.5274(18)Å and c= 14.0135(18)Å; see Table 3 in main text). This high symmetry phase
from now on will be referred to as ‘parent structure’.
S9
The lattice of the tripled structure was assembled using terephthalate molecules as rigid
bodies and inserting V and O atoms at the positions corresponding to the original lattice
connectivity. Overall parameters, like lattice constants, background or profile parameters were
freely refined. Tilt of the terephthalate molecules and atomic positions of the framework also
were refined. Adsorbed molecules were first searched for by analysis of observed and
difference electron density and then inserted as rigid bodies as can be seen from Figure S.6
and S.7 (RGB).
Figure S6. Observed electron density at the adsorption sites of the two localized styrene
molecules.
S10
Figure S7. Refined powder pattern of MIL-47 loaded with styrene. Tick marks: dark grey:
tripled phase (Pn21a) ; light grey: parent phase (Pnma).
The tripled phase was satisfactorily described using two styrene positions opposite to each
other. Position and occupation numbers of the guest molecules were at first freely refined but
occupation numbers were fixed after they converged on 1.
The phase with the parent symmetry was analyzed next. The framework was refined based on
the original publications (K. Barthelet, J. Marrot, D. Riou, G. Férey, Angew. Chem. 2002,
114, 291; Angew. Chem. Int. Ed. 2002, 41, 281), replacing linkers with rigid body units.
Styrene pairs were inserted on positions corresponding to those found in the tripled system
and then left to refine in tilt and position, which led to a good description of the powder
pattern with slightly increased R-values. Despite the very low amount of the parent Pnma
S11
phase next to the phase with broken symmetry (Pn21a), the improvement of the refinement
indicates that the found positions of the styrene pairs, which still closely resemble the
arrangement found in the tripled phase, are quite feasible. As was expected, occupation
numbers of the parent phase by styrene molecules refined to much smaller values compared to
the tripled phase (at most 10% in terms of elemental mesh). This implies that in presence of
only a few pairs of styrene the molecules can be distributed evenly over the framework
without hindering each other. Therefore at these low loadings the tripling of the lattice
constant caused by packing of the styrene pairs in the channels does not occur.
The final refinement led to satisfactory R values; so the reduction of symmetry and the
position of the guest molecules can be assumed to be reliable (Rp=0.0180, Rwp=0.0290,
expected Rexp =0.0302)
MIL-53
Analysis of the diffraction patterns of MIL-53 loaded with styrene and ethylbenzene revealed
in both cases the open mesh structure in the space group Pnma to be dominant. The closed
pore structures in space group Cc were present only in small amounts below 5%. Linker
molecules of the framework and guest molecules were treated as rigid bodies, whereof
position and tilt were refined. The ethylbenzene molecule was constructed in such a way that
the ethylgroup could rotate freely. After the unit cell and framework geometry were described
adequately, difference electron plots were inspected to identify the most likely site for the
guest molecules. Both types of molecules were found inside the channels almost parallel to
the pore walls in a position between two linker molecules. The main difference between the
guest molecules was the observation that in the case of the ethylbenzene the ethylgroup was
turned out of the plane of the phenyl ring, pointing towards the framework. For both
S12
molecules the occupation number approached 25%, which is the highest possible loading
without steric hindrance between neighboring molecules. The final refinement led to
satisfactory R values, so the position of the guest molecules can be assumed to be reliable
(Rp=0.0341, Rwp=0.0485, expected Rexp =0.0250 for ethylbenzene; Rp=0.0246, Rwp=0.0354,
expected Rexp =0.0140).
S13
Figure S8. Refined powder pattern of MIL-53 with ethylbenzene (top) and styrene (bottom).
Tick marks: dark grey: open mesh phase (Imma); light grey: closed mesh phase (Cc).
S14