3-1. Introduction
In this chapter, I report about one of the unveiled support effects of MOFs, i.e., an electronic interaction between the loaded metal NPs and MOF supports. In conventional material such as oxide-supported metal catalysts, the support strongly affects the catalytic activity of metal NPs through charge-transfer1 which accelerates catalytic reaction on the heterointerface between metal NPs and oxides.2 In particular, some researchers recently reported that electronic interactions between metal NPs and an oxide support can control catalytic activities; this is referred to as the EMSI.3 However, such oxides are unsuitable for electron donation to metal NPs because the valence band level of the oxide support, which is mainly composed of O 2p orbitals, is too low to increase the electron density on the metal NP surface. This limits the modification range of the electronic states of supported metal NPs, whereas the electron-rich environments in metal NPs are also important in some cases.4 Considering that the valence orbitals of MOFs are mainly derived from organic ligands, the energetic states of MOFs are potentially tunable through the ligand design. Therefore, I believe that MOF supports provide a great opportunity to achieve a wider range of control of catalytic activities in heterogeneous catalysis through the electronic interaction with metal NPs.
As a model reaction of this investigation, CO oxidation reaction in gas phase was selected.
This reaction is known as a reaction with a highly sensitivity to electronic state of loaded Pt NPs.5 Note that Pt/MIL-101, Pt/MIL-121 and Pt/MIL-125 were not included in this investigation because of some reasons such as its catalytic activity of the bare support and an overlapping in XPS signal from Pt and the center metals, which disturbs the accurate evaluation.
Then, the 4 types of Pt/MOFs (Pt/Zn-MOF-74, Pt/Mg-MOF-74, Pt/HKUST-1 and Pt/UiO-66-NH2) were selected as catalysts in this work.
44
Detailed studies on the electronic states of the Pt/MOF series were carried out using X-ray photoelectron spectroscopy (XPS), ultraviolet photoelectron spectroscopy (UPS) and density functional theory (DFT) calculations, which indicated that the electronic interaction between Pt NPs and a MOF support, i.e., electronic interaction.
To examine how the electronic states of the Pt NPs influence catalytic reactions, performances of the Pt/MOFs for a CO oxidation reaction were evaluated. The CO oxidation reaction was conducted in a fixed-bed flow reactor, loading catalyst powder that includes the same amount of Pt. I found that the catalytic activity on the reaction apparently differ according to the types of the MOF supports. Moreover, the order of the reaction temperature is the same as that observed in the electronic states of Pt NPs.
Through these experiments, I show the first report on systematic investigation of support effect of MOF, i.e., electronic interaction.
3-2. Experimental
3-2-1. Evaluation of electronic properties of Pt/MOFs.
X-ray photoelectron spectroscopy (XPS)
Electronic states of loaded Pt NPs were estimated by XPS measurements with ULVAC-PHI PHI 5000 VersaProbe II (Al-K, h = 1486.6 eV). Binding energies obtained by XPS were calibrated with C1s spectra of carbon tape on the sample holder at 284.6 eV. In particular, I focused on electrons of 4f7/2 (70–73.5 eV) and 4f5/2 (73.5–78 eV), which are known for appearing significant spectrum as XPS spectra of Pt NPs. Note that unidentified broad peaks around 75–78 eV were necessary to be used for the fitting of HKUST-1 data, which is probably attributable to Cu3p peaks.6 Obtained spectra were analyzed by least squares method under the assumption that the Pt NPs include Pt0, Pt2+ and Pt4+.
45
Ultraviolet photoelectron spectroscopy (UPS)
UPS measurement was also performed using the ULVAC-PHI PHI 5000 VersaProbe II (He I, h = 21.2 eV) with a bias (–6.0 or –11.0 V) applied to the sample holder. As mentioned below, to estimate working function of samples from UPS spectra, I have to detect “zero momentum energy = Ecutoff”. However, the value of energy is usually too low to be detected, this time too.
Then, the applied bias (–6.0 or –11.0 V) helps us detect it by accelerating photoelectrons (i.e.
increasing the energy of detected photoelectrons). Note that, the zero momentum energy represent an energy of photoelectron which barely beyond a barrier of ionization energy, among the photoelectrons generated from UV irradiation. As shown in Figure 3-1, to ensure electronic conduction of the surface of our samples and a sample holder, fixation of samples was performed by not carbontape but metal mask and washer. Gold wire (25 m) was also used to avoid charge-up on the surfaces of the samples. Ionization potentials (EI) obtained by the UPS were estimated by the following equation.
EI = h ( HeI: 21.2 eV) – (Ecutoff – EHOMO)
Here, EHOMO represents a momentum energy of photoelectron emitted from the top level of valence band. Simple scheme explaining the relationship among them are shown in Figure 3-2.
Figure 3-1a. Photos of (left) mask, (center) washer and (right) sample holder used for UPS measurement.
46
Figure 3-1b. A photo of the sample holder with all parts (with HKUST-1).
Figure 3-2. Theory of calculation of an ionization energy from UPS spectrum.
47
Density functional theory calculation
Calculations in this study were performed with plane-wave density functional theory implemented in the Vienna ab initio simulation package.7–10 The Perdew–Burke–Ernzerhof exchange-correlation functional was employed using projector-augmented wave pseudopotentials. I used the graphical visualization software package VESTA to analyze and visualize the optimized geometries11 as shown in Figure 3-3. Spin-polarized calculations were performed with 400 eV cutoff energy. Geometry optimization was performed and all atomic coordinates and cell parameters were fully relaxed. The geometry optimization was performed with k-point sampling at the gamma points. Monkhorst–Pack k-point sampling was used to estimate the local potential of the investigated metal–organic frameworks (MOFs). The k-point mesh for Mg-MOF-74 was 2 2 8, for Zn-MOF-74 2 2 8, for UiO-66-NH2 2 2 2, and for HKUST-1 2 2 2. The ionization potentials of the MOFs were calculated as a difference between the vacuum level and the top level of valence band of the MOFs. The value of the vacuum level was taken at the center of the void or pore of each MOF structure. For Mg-MOF-74 and Zn-MOF-Mg-MOF-74, this point was selected with coordinates of 0.0, 0.0, 0.5 relative to the cell parameters. For HKUST-1 and UiO-66-NH2, this point was selected with coordinates of 0.5, 0.5, 0.5 relative to the cell parameters as shown in Figure 3-4. This approach has been reported in the literature.12
48
Figure 3-3. Optimized geometries of (a) Zn-MOF-74, (b) Mg-MOF-74, (c) HKUST-1 and (d) UiO-66-NH2, used for the DFT calculation. Zn, Mg, Cu, Zr, C, N, and O are represented by gray, orange, blue, green, brown, purple, and red, respectively.
Figure 3-4. A scheme of procedure of setting vacuum level in calculation (an example of HKUST-1). (left) A face which was calculated for potential energy. (right) An energy level regarded as vacuum level.
49
3-2-2. Characterization of the catalytic performance
H2-pulse chemisorption
Active surface areas of loaded Pt NPs were estimated by H2-pulse chemisorption measurements with BELCAT-A (BEL Japan, Inc.) using 30–50 mg of samples. This measurement was conducted at room temperature.
CO oxidation reaction condition A (for T50)
The CO oxidation reaction with the condition A was performed with ~12 mg of catalysts, containing the same Pt amount, using BELCAT-A (BEL Japan, Inc.) as shown in Figure 3-5.
The gas products were constantly monitored using BEL-MASS quadrupole mass spectrometer (BEL Japan, Inc.), while heating to 300 C at a rate of 1.5 C min–1. A reaction gas mixture of 1% CO, 28% O2, and 71% He was passed over the catalysts at room temperature before the reaction. Total flow rate and gas hourly space velocity (GHSV) were 50 ml min–1 and 250000 ml g–1 h–1, respectively. The CO conversion was calculated according to the following equation:
CO conversion (%) = (([CO]inlet – [CO]out) / [CO]inlet) × 100
where [CO]inlet and [CO]out represent the inlet and outlet concentrations of CO, respectively.
CO oxidation reaction condition B (for turnover frequency)
The CO oxidation reaction with the condition B was performed with 50 mg of catalysts, containing the same Pt amount using tubular quartz reactor in tubular furnace. The gas products were analyzed using thermal conductivity detector (TCD) gas chromatography (Shimadzu, GC-8A) equipping an active carbon column (GL science, mesh 60/80, ID 3 mmφ, 2 m). A reaction gas mixture of 2% CO, 18% O2, and 80% He was passed over the catalysts at room temperature before the reaction. Total flow rate and gas hourly space velocity (GHSV) were 50 ml min–1 and 60000 ml g–1 h–1, respectively. The CO conversion was calculated according to the same equation with the condition A.
50
Figure 3-5. (left) scheme of sample tube and (right) a photo of reaction device for CO oxidation reaction
3-3. Results and Discussion
Figure 3-6 shows the resulting iso-electron density surfaces of Vienna ab initio calculation using plane-wave DFT theory. As a result, ionization energies of Zn-MOF-74, Mg-MOF-74, HKUST-1 and UiO-66-NH2 werecalculated as 4.37, 4.22, 5.18 and 5.55 eV, respectively. This result suggests that there is significant difference in electron-donating abilities (i.e. the top level of valence band) among the 4 MOFs as I expected. In terms of actual measured value, Figure 3-7 indicates the resulting UPS spectra with the applied bias. From analyses of the UPS spectra, ionization energies were estimated as 5.2, 5.7, 6.0 and 6.7 eV, respectively. These values are almost consistent with the ones from DFT calculations. Ionization energies determined by DFT calculation and UPS measurement are summarized in Table 3-1. Note that there is a certain difference (about 1.0 eV) between the results of the DFT calculation and UPS measurement.
The differences were probably derived from the assumption which was used for the DFT calculations. As I mentioned in experimental section, I employed the energy values at the center of the void point of each MOF structure as the energy of the vacuum level, which possibly causes underestimation of each vacuum level. From the both results, obtained ionization
51
energies of Zn-MOF-74 and Mg-MOF-74 were remarkably higher than those of the others.
Considering their structures, such higher ionization energies are probably derived from the existence of electron-rich ligands with electron-donating hydroxyl groups (i.e. 2,5-dihydroxyterephthalic acid (bdc-(OH)2)).13 Meanwhile, the relatively lower ionization energy of UiO-66-NH2 is attributed to the high valent Zr4+ ion which can stabilize neighbor electrons through electrostatic interaction, whereas center metals of the other MOFs are 2+.
Figure 3-6. Iso-electron density surfaces obtained from the DFT calculations of MOFs, (a) Zn-MOF-74, (b) Mg-Zn-MOF-74, (c) HKUST-1, and (d) UiO-66-NH2.
52
Figure 3-7. UPS spectra of (a) Zn-MOF-74, (b) Mg-MOF-74, (c) HKUST-1, and (d) UiO-66-NH2.
Table 3-1. Ionization energy determined by calculation and measurement, respectively.
Up to here, the estimated ionization energies suggested that electron-donating ability of Zn-MOF-74, Mg-Zn-MOF-74, HKUST-1, and UiO-66-NH2 have a tendency: Zn-MOF-74 ≈ Mg-MOF-74 > HKUST-1 > UiO-66-NH2. So I estimated that electronic states of loaded Pt NPs are different depending on electronic interaction with MOF supports.
To examine the surface electronic states of them, I performed XPS measurements of Pt/Zn-MOF-74, Pt/Mg-Pt/Zn-MOF-74, Pt/HKUST-1 and Pt/UiO-66-NH2. I chose this measurement to evaluate the electronic states of Pt NPs in the Pt/MOFs because the energy scale of X-ray used
53
for XPS measurement is more suitable than that of UV for the observation of an elemental core-level binding energy which reflects an effect from an electronic interaction.14
Considering the experimental condition of CO oxidation reaction, as other researchers reported, Pt0 species is the strongest candidate as an active site for CO oxidation occurring running under gas flows of CO and O2 (in particular, molar ration is O2 >> CO).15 Note that Figure 3-8 indicates how to calibrate obtained results from each C1s spectrum. In the spectra, the largest peaks in the region of 238–287 eV were almost the same among all of the samples, which belongs to signals from carbon tapes. The smaller peaks at 289 eV which are varied by the types of MOFs were identified to be C=O and O-C=O in the MOFs. Then, I chose the peaks from carbon tapes for calibration of XPS spectra of Pt.
Figure 3-8. (a) Calibrated XPS spectra of C1s from Pt/MOFs, (b) An example of curve fitting analysis (Pt/Zn-MOF74, C1s)
54
The obtained XPS spectra of Pt4f on Pt/Zn-MOF-74, Pt/Mg-MOF-74, Pt/HKUST-1, and Pt/UiO-66-NH2 are shown in Figure 3-9. These were composed of spectra for Pt4f7/2 (70–73.5 eV) and Pt4f5/2 (73.5–78 eV) and analyzed by the least squares fitting method, using curves, calculated assuming the existence of Pt0, Pt2+, and Pt4+ species. From this result, I determined the binding energies of Pt04f7/2 on Pt/Zn-MOF-74, Pt/Mg-MOF-74, Pt/HKUST-1, and Pt/UiO-66-NH2 as 70.83, 71.07, 71.21, and 71.72 eV, respectively. The surface electronic states of loaded Pt NPs which were estimated from the XPS spectra of Pt are clearly different among the samples. As I already mentioned the Pt NPs are almost the same in the amount and average diameters (i.e. there is no size effect on their electronic states). Hence, I can say that the differences of surface electronic states of the Pt NPs are surely derived from the type of MOF supports.
Figure 3-9. XPS spectra of Pt4f of (a) Pt/Zn-MOF-74, (b) Pt/Mg-MOF-74, (c) Pt/HKUST-1 and (d) Pt/UiO-66-NH2.
55
From comparison with the binding energy of bulk Pt (71.2 eV16, represented by dot line) in Figure 3-10, especially when I focus on Pt/Zn-MOF-74 and Pt/Mg-MOF-74, there may be electronic interactions between Pt NPs and each support because the positions of Pt0 peaks from Pt04f of them are shifted to the lower than that for bulk Pt. This observation suggested that the certain charge-transfer from MOF-74 supports to Pt NPs possibly happens. Meanwhile, the positions of a peak for Pt/UiO-66-NH2 was higher than that for bulk Pt, indicating that charge-transfer from Pt NPs to UiO-66-NH2 happens. From these results, I revealed that the relationship of electronic states of the loaded Pt NPs consistent with the estimated tendency of electron-donating ability, i.e. ionization energy, of the MOF supports. This indicates that there were charge-transfer interactions between Pt NPs and MOF supports. This is the first systematic investigation about electronic interaction between loaded metal NPs and MOFs, which strongly suggest that electron-donating ability of MOF supports directly influences on the electronic states of loaded metal NPs in M/MOF composites.
Figure 3-10. Fitted Pt04f7/2 XPS spectra of Pt/MOFs. The dot line indicates the binding energy of bulk Pt0 of 4f7/2 from reference 16.
To see how the electronic states of Pt NPs influence on catalytic reactions, I studied
56
performances of the Pt/MOFs (Pt/Zn-MOF-74, Pt/Mg-MOF-74, Pt/HKUST-1, and Pt/UiO-66-NH2) for the CO oxidation reaction. Catalytic activity of Pt for the CO oxidation reaction is highly sensitive to their electronic states, in other words, the reaction activity is expected to depend on the charge transfer between Pt and catalytic support.5 Figure 3-11 indicates a brief description of the reaction. A reaction steps of the CO oxidation reaction on a Pt NPs loading catalyst was proposed as described below:
(i). CO molecules rapidly adsorb on active sites of the surface of Pt NPs at room temperature.
(ii). By raised temperature, the adsorbed CO molecules desorb from the Pt NPs.
(iii). O2 molecules adsorb on the active site which is opened by CO desorption.
(iv). The adsorbed CO and O2 molecules react to CO2.
Here, rate determining step of this reaction is expected as step (ii) procedure because it is the only step which needs heating to high temperature. As an evidence for that, the CO oxidation reaction can be promoted by decreasing the CO-adsorption energy, which is achieved by increasing a charge density of the surface of the Pt NPs.1 Therefore, I expected the tendency of catalytic activities of Pt/MOFs is Pt/Zn-MOF-74 > Pt/Mg-MOF-74 > Pt/HKUST-1 > Pt/UiO-66-NH2 for the CO oxidation reactionbecause the result of XPS measurement suggested that the tendency of electron densities of the Pt NPs is also Pt/Zn-MOF-74 > Pt/Mg-MOF-74 >
Pt/HKUST-1 > Pt/UiO-66-NH2.
57
Figure 3-11. Reaction model of CO oxidation reaction with a NPs catalyst
In the beginning, the CO oxidation reaction was performed using a fixed-bed flow reactor, loading catalyst powder that includes the same amount of Pt, which was described above as condition A in the experimental section. Figure 3-12 shows the temperature dependence of CO conversion obtained from a real-time monitoring with mass spectrometry. All of these samples showed catalytic activity for the CO oxidation reaction and 100% conversion of CO was observed at a temperature range from 180 to 210 °C, indicating that the surface of the loaded Pt is not poisoned and active for catalytic reactions. Besides, all of MOFs without Pt NPs were also applied for the catalysis, as a sequel, I confirmed that they don’t have catalytic activities for the reaction under the same condition. In the Figure 3-12, I defined each reaction temperature as T50 at temperature of half conversion, which is observed in the vicinity of a sharp rise of the CO conversion. The T50 of Pt/Zn-MOF-74, Pt/Mg-MOF-74, Pt/HKUST-1, and Pt/UiO-66-NH2 were determined as 181.1, 195.5, 201.5, and 209.8 C, respectively. The values of T50 apparently differ according to the types of the MOF supports. Remarkably, the order of the reaction temperature is the same as that observed in the electronic states of Pt NPs (the estimated surface electron densities of Pt04f7/2 of the loaded Pt: Pt/Zn-MOF-74 > Pt/Mg-MOF-74 > Pt/HKUST-1 > Pt/UiO-66-NH2).
58
Figure 3-12. Temperature dependence of CO oxidation reaction
To see this relationship clearer, I made a graph of the binding energy of Pt04f7/2 obtained by XPS vs T50 in Figure 3-13. Then, I can clearly see that the binding energies of Pt0 and T50 have direct proportional relationship. This means that higher electron density of Pt NPs resulting from charge-transfer interaction from MOF supports results in its higher catalytic activity on CO oxidation reaction.
59
Figure 3-13. Relationship between binding energies of Pt0 and T50
I also calculated turnover frequency (TOF) of Pt/MOFs on CO oxidation reaction under a low temperature elevation–rate condition using gas chromatography (see experimental section). To calculate TOF, active surface areas on loaded Pt NPs were determined by H2 pulse chemisorption (summarized in Table 3-2). Figure 3-14 shows the calculated TOF of Pt/MOFs, which have a tendency: Pt/Zn-MOF-74 > Pt/Mg-MOF-74 >
Pt/HKUST-1 > Pt/UiO-66-NH2. This tendency is consistent with the result of XPS, which proves that the charge-transfer interaction between Pt NPs and MOFs makes influence on the loaded Pt NPs. These results also composed to a fact that an electron-rich surface is suitable on the CO oxidation reaction5. In this section, from the results of DFT, UPS, XPS and catalysis as summarized in Table 3-3, I can say that the charge-transfer interaction is able to enhance a catalytic activity of Pt NPs on CO oxidation reaction, i.e. a reaction depending on electronic states of a catalyst. This is the first systematic investigation of modulation of catalytic activities of loaded metal NPs by
60
differences of electron-donating ability of MOFs. This work clearly show that MOFs can act as a modulator of electronic states of loaded metal NPs to control activity for heterogeneous catalysis.
Table 3-2. The amount of adsorbed H2 on the loaded Pt NPs, measured by pulse chemisorption.
Figure 3-14. (a) TOFs of Pt/MOFs for CO oxidation reaction at each temperature (condition B) and (b) Arrhenius plots of the TOFs.
61
Table 3-3. Ionization energy (eV), Binding energy of Pt04f7/2 (eV) and T50 °C
3-4. Conclusion
In conclusion, the electronic properties of the Pt/MOFs (Pt/Zn-MOF-74, Pt/Mg-MOF-74, Pt/HKUST-1, and Pt/UiO-66-NH2) were evaluated by theoretical calculations, UPS, and XPS studies. These measurements revealed that Pt NPs in the Pt/MOFs have different electronic states originating from the charge-transfer interaction between Pt NPs and the MOF supports having different electron-donating ability. The results of CO oxidation catalysis showed that the catalytic activities of the loaded Pt NPs are directly modulated by the type of MOFs. I demonstrated that the electronic band energy of the MOF supports apparently affects the activity of the loaded Pt NPs for heterogeneous catalysis due to the charge transfer interaction. The knowledge gained in this study should contribute to fine-tuning the activity of various metal catalysts by choosing appropriate central metals, ligands, or functional groups of the MOF supports. This results had been reported on international academic journal17.
3-5. References
1. SWang, F.; Ueda, W.; Xu, J. Angew. Chem., Int. Ed. 2012, 16, 3883–3887.
2. Rachmady, W.; Vannice, M. A. J. Catal. 2000, 192, 322–334.
3. (a) Bruix, A.; Rodriguez, J. A.; Ramirez, P. J.; Senanayake, S. D.; Evans, J.; Park, J. B.;
62
Stacchiola, D.; Liu, P.; Hrbek, J.; Illas, F. J. Am. Chem. Soc. 2012, 134, 8968–8974. (b) Vayssilov, G. N.; Lykhach, Y.; Migani, A.; Staudt, T.; Petrova, G. P.; Tsud, N.; Skála, T.;
Bruix, A.; Illas, F.; Prince, K. C.; Matolı´n, V.; Neyman, K. M.; Libuda; J. Nature mater.
2011, 10, 310–315.
4. Kitano, M.; Inoue, Y.; Yamazaki, Y.; Hayashi, F.; Kanbara, S.; Matsuishi, S.; Yokoyama, T.; Kim, S. W.; Hara, M.; Hosono, H. Nature Chem. 2012, 4, 934–940.
5. Chua, Y. P. G.; Gunasooriya, G. T. K. K.; Saeys, M.; Seebauer, E. G. J. Catal. 2014, 311, 306–313.
6. M. T. Anthony, M. P. Seah, Surf. Inter. Anal. 1984, 6, 107–115.
7. G. Kresse, J. Hafner, Phys. Rev. B 1993, 47, 558–561.
8. G. Kresse, J. Furthmüller, Phys. Rev. B 1996, 54, 11169–11186.
9. G. Kresse, J. Furthmüller, Comput. Mater. Sci. 1996, 6, 15–50.
10. P. E. Blöchl, Phys. Rev. B 1994, 50, 17953–17979.
11. K. Momma, F. Izumi, J. Appl. Crystallogr. 2011, 44, 1272–1276.
12. K. T. Butler, C. H. Hendon, A. Walsh, J. Am. Chem. Soc. 2014, 136, 2703–2706.
13. Butler, K. T.; Hendon, C. H.; Walsh, A. J. Am. Chem. Soc. 2014, 136, 2703–2706.
14. Shpiro, E. S.; Dysenbina, B. B.; Tkachenko, O. P.; Antoshin, G. V.; Minachev, Kh.
M., J. Catal., 1988, 110, 262–274.
15. (a) Hendriksen, B. L. M.; Frenken, J. W. M. Phys. Rev. Lett. 2002, 89, 046101. (b) Qadir, K.; Kim, S. H.; Kim, S. M.; Ha, H.; Park, J. Y. J. Phys. Chem. C, 2012, 116, 2405424059.
16. (a) Parkinson, C. R.; Walker, M.; McConville, C. F. Surface Sci. 2003, 545, 19–33.
(b) Contour, J. P.; Mouvier, G.; Hoogewys, M.; Leclere, C. J. Catal. 1977, 48, 217228.
17. Yoshimaru, S.; Sadakiyo, M.; Staykov, A.; Kato, K.; Yamauchi, M. Chem. Commun., 2017, 53, 6720–6723.
63