Energy Dispersive X-ray Diffraction Studies 102

It has been shown that many different molecules can be adsorbed within the pores of MIL-53(Fe) (Chapter 1) from simple alcohols to larger guests such as pyridine and lutidine.1, 19 The results in this section show that the pores of the MIL-53(Fe) framework are capable of accommodating larger guests such as benzothiophene, benzothiazole and indole (Figure 3.13) which are the three guest molecules chosen for these experiments.

Figure 3.13: Three N/S containing heterocycles used for liquid phase adsorption studies

The uptake of these three bulky molecules was followed via energy dispersive X-ray diffraction (EDXRD). The experiments were performed in two different solvents to help study the effects of the solvent interactions. Isopropanol, a hydrophilic solvent known to be taken up by MIL-53(Fe) and heptane, a long-chain hydrophobic solvent known to interact poorly with MIL-53(Fe), were chosen as the two experimental solvents. Prior to these experiments, high-resolution X-ray diffraction (HRXRD) was used to obtain the unit cell parameters of MIL-53(Fe)[guest]. Solutions of the guest molecules in isopropanol were injected into capillaries containing MIL-53(Fe) powder. The capillaries were then centrifuged to concentrate the powder at the end of the capillary. The unit cell parameters obtained from a full Rietveld refinement (performed by N. Guillou, Institut Lavoisier) are shown below in Table 3.2.

- 103 -

Table 3.2: Unit cell parameters for MIL-53(Fe)[guest]. Values obtained from full Rietveld refinement of high-resolution XRD data.

Guest MIL-53(Fe) [benzothiophene] MIL-53(Fe) [benzothiazole] MIL-53(Fe) [indole] a/ Å 15.8023(5) 15.7727(7) 16.0647(6) b/ Å 14.5731(4) 14.6253(6) 14.2624(5) c/ Å 6.91279(9) 6.9026(2) 6.8769(2) V/ Å3 1591.93(7) 1592.3(1) 1575.6(1)

S. G. Imcm Imcm Imcm

Isopropanol is known to only give the monoclinic half-open phase (C2/c,

V ≈ 1200 Å3)19 when adsorbed within the pores, this is in contrast to the molecules we

are studying which give the fully-expanded orthorhombic phase (Table 3.2). The different symmetries of the framework induced by different guests mean that the progression of the uptake can be followed using powder X-ray diffraction. Heptane was chosen as an alternative solvent as it is more representative of the conditions for guest uptake from petrochemicals. When MIL-53(Fe) is stirred in heptane, preliminary tests show that it is not adsorbed by MIL-53(Fe) within the first 60 minutes; Bragg peaks from a secondary phase are only visible after 1.5 hours and after 10 hours adsorption is still not complete (Figure 3.14). Therefore, the framework should remain in its initial phase before addition of the guest molecules.

- 104 -

Figure 3.14: Adsorption of heptane by hydrated MIL-53(Fe)

For the initial tests two types of experimental procedure were used to follow the uptake of the guest molecules:

Static experiments- the guest solution and solvent were added straight to the test tube at the beginning of the experiment and data collection was started immediately. Final concentration = 5.2 M and the number of molar equivalents of guest present at end of the experiment = 10

Syringe pump experiments- the guest solution (10 mL) was added to a syringe and a calibrated syringe pump was used to control the addition of the solution to the test tube at a rate of 10 mL/hr. Final concentration = 0.67 M and the number of molar equivalents of guest present at end of the experiment = 5.6.

The contour maps shown below plot the energy of the Bragg peaks as a function of time. The Bragg peaks visible in the contour plots were indexed using the crystallographic data recorded using high-resolution XRD, the conversion of the EDXRD data from energy to d-spacing is shown in tables in the Appendix.

Figure 3.15 shows the results of the preliminary experiments performed in isopropanol, the contour plots confirm that all three guests cause the framework to expand fully. The

- 105 - uptake of benzothiophene and indole was studied using the static experimental technique whereas the uptake of benzothiazole was followed using the syringe experimental technique; the expansion of the framework was too rapid to be followed using the static technique. Using the syringe technique, hydrated MIL-53(Fe) was suspended in isopropanol prior to the addition of the guest molecules with the result that the water present in the pores of hydrated MIL-53(Fe) was displaced by isopropanol; demonstrated by the presence of the Bragg peaks of the monoclinic half-open phase observed in the contour plot (Figure 3.15b). The timescale required for the completion of the benzothiophene and indole experiments provides evidence that benzothiophene is taken up more quickly by hydrated MIL-53(Fe); conversion to the fully-open benzothiophene phase is complete after 100 minutes whereas conversion to the fully- open indole phase takes more than 260 minutes.

In all three cases, the expansion from the initial phase to the fully-open phase is not gradual; the material is present in either one phase or the other. However, for benzothiazole, after expansion to the fully-open phase there is a slight shift of the Bragg peaks. The (110) peak of the fully-open phase shifts to lower energy and the (200) peak shifts to a higher energy, this is consistent with an expansion of the framework. It is suggested that this slight expansion is due to minor rearrangements of the guest molecules after adsorption. Benzothiazole contains S and N heteroatoms which are both able to interact with the host framework; this will be discussed later in more detail.

- 106 -

Figure 3.15: Contour plots showing the uptake of guest molecules in isopropanol a)

benzothiophene (static experiment, 5.2 M) b) benzothiazole (syringe experiment, 0.67 M) and c) indole (static experiment, 5.2 M).

The static tests, where the MIL-53 remains in the hydrated phase prior to guest adsorption, gave the first evidence that the framework does not expand via the half-open

- 107 - phase during the uptake of these bulky guest molecules. This is in contrast to the behaviour seen for some of the molecules that were previously studied. It is suggested that this is due purely to their bulkier size; one heterocycle molecule is too large to be able to occupy a contracted form of the framework, they require the full volume of the available pores of MIL-53(Fe).

To make these experiments more applicable to the potential industrial application, the separation of heterocycles from petroleum fuels, they were repeated, using both isopropanol and heptane, at lower concentrations of guest to see the effect, if any, on the ability of the framework to remove these molecules from solution. The concentration of the guest-solvent solution was chosen based on the maximum concentration of indole that could be dissolved in heptane. The maximum concentration was calculated to be 0.39 M, which was the concentration used in the syringe. Due to the solution being added to an additional 3 mL of solvent in the test tube the final concentration was 0.3 M. This corresponds to 2.5 molar equivalents of the guest per molar equivalent of host; the guest was always in excess at the end of each experiment.

For each guest the experiments were performed using both hydrated and dehydrated MIL-53(Fe) to investigate the effect that the water, contained in the pores of the framework, had on the uptake of these molecules. For the experiments using dehydrated MIL-53(Fe), which was dehydrated using the method described in Chapter 2.2.1.4, anhydrous isopropanol was used. The addition of the guest solutions was controlled by a syringe pump and set to a constant speed of 10 mL/hr as outlined above for this experimental technique.

The contour plots in Figure 3.16 show the uptake of the guest molecules by dehydrated MIL-53(Fe) in dry isopropanol; the initial phase at the start of each contour is the

- 108 - monoclinic, C2/c, half-open phase, ~1200 Å3. The contours show that uptake of benzothiazole by dehydrated MIL-53(Fe) in dry isopropanol is much faster than the uptake of benzothiophene. The experiment with benzothiophene never reaches completion in the time period studied (350 minutes); there is some of the initial half- open phase still present. Only the first 60 minutes of the indole experiment is shown as it was found that under these conditions, using low guest concentration and dehydrated MIL-53(Fe), the indole was not adsorbed. The test tube was re-examined in the beam periodically and after 28 hours there was still no evidence for any conversion to the fully-open phase (Figure A.1).

- 109 -

Figure 3.16: Contour plots showing the behaviour of dehydrated MIL-53(Fe) in dry isopropanol upon the addition of a) benzothiophene, b) benzothiazole and c) indole.

The contour plots for the heptane experiments performed with dehydrated MIL-53(Fe), Figure 3.17, show that the initial phase at the start of each experiment is the monoclinic,

C2/c, dehydrated phase, ~900 Å3; indicating that heptane was not adsorbed prior to

guest addition. The contours show a similar trend as was seen for isopropanol; benzothiazole and benzothiophene are taken up by MIL-53(Fe) quickly while indole is

- 110 - not taken up at all. However, the difference between benzothiazole and benzothiophene is less distinct than was seen in isopropanol; evidence for the fully-open phase can be seen very early-on in both experiments but benzothiazole continues to a higher conversion percentage than benzothiophene (this can be seen more clearly in the extent of conversion graphs in Figure 3.19).

- 111 -

Figure 3.17: Contour plots showing the behaviour of dehydrated MIL-53(Fe) in heptane upon the addition of a) benzothiophene, b) benzothiazole and c) indole.

The results for the uptake of these heterocycles by hydrated MIL-53(Fe) in isopropanol show the same trend as was seen for dehydrated MIL-53(Fe) in isopropanol. Benzothiazole was adsorbed by the pores much more quickly than benzothiophene and indole was not taken up by the framework at all. In the contour plots for benzothiophene and indole there are Bragg peaks present that are due to the hydrated phase which was caused by the incomplete conversion of the hydrated phase to the half-open isopropanol phase before the experiment was started. This is more visible in the indole contour plot where the evolution of the hydrated (200) peak into the half-open (200) peak can be clearly seen. Experiments using hydrated MIL-53(Fe) and heptane were not possible

- 112 - with the conditions used for the syringe pump experiments as efficient stirring could not be achieved due to the incompatibility of hydrophobic heptane with the hydrated solid.

Figure 3.18: Contour plots showing the behaviour of hydrated MIL-53(Fe) in isopropanol upon the addition of a) benzothiophene, b) benzothiazole and c) indole.

It is clear from these experiments that benzothiophene and benzothiazole are still taken up by MIL-53(Fe) at these lower concentrations whereas indole is not. It can also be observed that the removal of the water molecules from the pores of MIL-53(Fe) did not affect the order of adsorption based on the rate of uptake; the order of preference seen for MIL-53(Fe) was always benzothiazole > benzothiophene > indole.

- 113 - The contour plots above show that not all experiments went to completion as the peaks of the initial phase are still present in small amounts. To gain more information from these experiments the data were converted to extent of conversion curves which allows the data sets to be compared more directly.

As it was found that not all experiments reached completion, converting the data to extent of conversion graphs could not be done by normalising to the Bragg peak intensity of the final product. However, the contour plots show that evolution of pore opening does not proceed via any intermediate phases; there is only evolution from the initial to the fully-open phase. Therefore, assuming that the transition from one phase to the other is directly related, the decay of the initial phase can be used as an indication of the growth of the fully-open phase. This method also gives an indication of the percentage of completion of each experiment.

The strongest Bragg peak from the starting phase was integrated and then the complement of the normalised data (Equation 3.1) was used to create the extent of conversion graphs.

Equation 3.1

(xn is the normalised data and xc is the complement of the data).

These data have been plotted as a function of time and also as a function of concentration. As the concentration of the guest solution was known and the rate of addition was controlled, the concentration at any given time can be calculated. This comparison is important as it is the behaviour of MIL-53(Fe) towards low

- 114 - concentrations of guest molecules that is of interest in practical applications such as the removal of contaminant molecules.

From the graphs in Figure 3.19 it can be concluded that benzothiazole is taken up faster than benzothiophene under all conditions. Benzothiazole is taken up marginally slower from heptane than from isopropanol; it takes longer for MIL-53(Fe) to reach its maximum conversion. In contrast, the initial rate of uptake of benzothiophene is quicker in heptane compared to isopropanol but the final conversion maximum is approximately equal for both solvents when using dehydrated MIL-53(Fe). These graphs highlight that evolution of the structure continues after complete addition of the guest solution at 60 minutes. It should be noted that the experiments performed with hydrated MIL-53(Fe) were not studied for long enough so have not reached equilibration but information about the general trend of uptake can still be obtained. The results for indole have not been included as the EDXRD studies showed that indole was not adsorbed by MIL- 53(Fe) under the conditions used.

- 115 -

Figure 3.19: Extent of conversion as a function of time for the slow addition of benzothiophene and benzothiazole to a) dehydrated MIL-53(Fe) and isopropanol, b) dehydrated MIL-53(Fe) and heptane and c) hydrated MIL-53(Fe) and isopropanol.

- 116 - From extent of conversion graphs versus guest concentration (Figure 3.20) it can be seen that at the point of complete addition (0.3 M) benzothiazole is always between 85- 95% completely converted to the fully-open phase whereas benzothiophene only reaches between 65-80% of full conversion. Comparison of the lower concentrations between the heptane and isopropanol experiments shows that there is a clear solvent effect that influences the initial uptake of the guest. For both benzothiophene and benzothiazole the experiments in heptane show pore opening commences as low as 0.05 M whereas pore opening is only seen between 0.08 - 0.1 M in isopropanol.

- 117 -

Figure 3.20: The extent of conversion as a function of time for the slow addition of

benzothiophene and benzothiazole to a) dehydrated MIL-53(Fe) and isopropanol, b) dehydrated MIL-53(Fe) and heptane and c) hydrated MIL-53(Fe) and isopropanol.

- 118 - Table 3.3 shows the percentage of the framework that was converted to the fully-open phase after complete addition of the guest solution; complete addition was reached after 60 minutes, the concentration at this point was 0.3 M. Also given in the table is the concentration at which the material is first observed to adsorb the guest molecules.

Table 3.3: The maximum conversion percentages immediately after complete addition at 0.3 M (60 minutes) and the concentration at which MIL-53(Fe) first takes up the guest molecule.

Guest MIL-53(Fe) Solvent Conversion at

0.3 M/ % Initial Uptake Concentration/ M Benzothiophene Dehydrated Isopropanol 68 0.11 Heptane 79 0.039 Hydrated Isopropanol 64 0.16 Heptane - - Benzothiazole Dehydrated Isopropanol 93 0.08 Heptane 85 0.056 Hydrated Isopropanol 94 0.11 Heptane - -