Jeong Hwan
Lee‡
a,
Donghui
Jo
b and
Suk Bong
Hong
*a
aCenter for Ordered Nanoporous Materials Synthesis, Division of Environmental Science and Engineering, POSTECH, Pohang 37673, Korea. E-mail: sbhong@postech.ac.kr
bLow-Carbon Petrochemical Research Center, Korea Research Institute of Chemical Technology (KRICT), Daejeon 34114, Korea
First published on 21st February 2024
The synthesis of PST-24 with a relatively high Al content (Si/Al = 47), a disordered medium-pore zeolite, using the seeding technique and its catalytic properties for the low-temperature (250 °C) dehydration of isobutanol to isobutene are presented. This PST-24 was found to be considerably more active and stable for isobutanol dehydration than H-ZSM-5 and H-ferrierite with similar Si/Al ratios, the two most widely studied catalysts for this reaction, revealing its high potential as a new dehydration catalyst for biomass-derived isobutanol. While variable-temperature IR spectroscopy with adsorbed isobutanol reveals stronger interactions of isobutanol molecules with the PST-24 framework than with the latter two zeolite frameworks, 1H–13C CP MAS spectroscopy shows the formation of soft coke only in the former catalyst.
A wide variety of solid acids, including amorphous silica-alumina, mixed-metal oxides and hydroxyapatites, have been investigated for alcohol dehydration so far.5–9 Owing to its cost-effective availability, for example, γ-alumina has been extensively studied for isobutanol dehydration.10–12 However, a significant decrease in dehydration activity by water is the serious drawback of this catalyst, despite its high selectivity toward isobutene:12 at 285 °C, isobutanol conversion decreased from 53–61 to 28–42% upon co-feeding of 15 wt% water.10
On the other hand, aluminosilicate zeolites with channels and cavities of molecular dimensions have been of fundamental interest in the dehydration of isobutanol to isobutene, because of their well-known shape- and surface-selective properties. For example, de Reviere et al. examined the effect of zeolite framework topology on this reaction and showed that the proton form (H-ZSM-5) of the medium-pore high-silica (Si/Al = 24) zeolite ZSM-5 (framework type MFI) outperformed large-pore zeolites H-Y (FAU; Si/Al = 39) and H-mordenite (MOR; Si/Al = 11). Although the latter two catalysts showed negligible isobutanol conversion at 240 °C and 5 bar, the conversion over H-ZSM-5 was found to be ca. 50%, probably due to the optimal fit of isobutanol and reaction intermediates within its 10-ring channel intersections.13 H-ferrierite (FER), a former industrial catalyst for the skeletal isomerization of n-butenes to isobutene,14 was also reported to be highly selective for isobutanol dehydration to produce linear butenes.15–17 This medium-pore zeolite exhibited similar activity in both the presence and absence of 10% water in the feed even after 15 h, except for a nominal decrease in isobutanol conversion (∼10%) at early time on stream.15
PST-24 is a new disordered medium-pore zeolite synthesized via the so-called excess fluoride approach18 using the pentamethylimidazolium (PMI+) cation as an organic structure-directing agent (OSDA).19 Its pore architecture is exceedingly unique in that the intracrystalline channel dimensionality varies locally from two-dimensional (2D) to bi-level 2D to 3D. While the composite cas-zigzag building chains in PST-24, which are connected by ‘disordered’ double 5-ring (d5r) columns, are ordered throughout the whole framework structure, the d5r column pairs can adopt two different arrangements that ‘close’ and ‘open’ 10-ring (6.1 × 3.5 Å) channels along the [101] direction as illustrated in Fig. 1. This has led us to suggest three representative ordered polytype structures, denoted as PST-24A (P21/c), PST-24B (P-1) and PST-24C (P2/c), where the second one is the alternating structure of the first and third ones.19 Consequently, PST-24 has 10-ring pockets ([4.56·66·82·10] cages; PST-24A) and short 10-ring channels along [101] (PST-24C) (Fig. 1), in addition to parallel straight 10-ring (5.8 × 5.4 Å) and 8-ring (4.8 × 3.1 Å) channels along [010] and sinusoidal 8-ring channels along [001]. Such a pore structure provides the distinctive void space in the spectrum of medium-pore cavities where the channel intersections of ferrierite and ZSM-5 belong, rendering PST-24 attractive as an efficient isobutanol dehydration catalyst.
However, the lowest Si/Al ratio (or the highest Al content) of PST-24 that can be achieved by direct synthesis is about 200.19 Such difficulty in incorporating more Al atoms into its framework, which is often observed in zeolite synthesis under normal fluoride conditions (HF/OSDA = 1.0),20 can limit the potential of this disordered zeolite as a solid acid catalyst or catalyst support. Here we show that the Si/Al ratio of PST-24 can be lowered as low as 50 when 4 wt% of calcined PST-24 with Si/Al = 200 is added as seed crystals to a PMI+-containing synthesis mixture. We also show that the proton form of this PST-24 zeolite is considerably more active and stable for the low-temperature (250 °C) dehydration of isobutanol to isobutene than H-ZSM-5 and H-ferrierite with similar Si/Al ratios. The origin of the high isobutanol reactivity of H-PST-24 has been investigated by using variable-temperature IR spectroscopy with adsorbed isobutanol.
As-made zeolites were calcined in air at 550 or 600 °C for 8 h to remove the occluded OSDAs and refluxed twice in 1.0 M NH4NO3 (2.0 g solid per 100 mL solution) for 6 h followed by calcination at 550 °C for 4 h in order to ensure that the zeolite was completely in its proton form. Two ZSM-5 zeolites with Si/Al = 47 and 95 and one ferrierite with Si/Al = 33 were obtained from Clariant and Tosoh, respectively, and converted to their proton form in a similar way described above. All zeolite catalysts studied here will be denoted by adding ‘(n)’ to their common name, where n is the bulk Si/Al ratio.
Isobutanol adsorption–desorption and/or dehydration were monitored by variable-temperature IR spectroscopy. A self-supporting zeolite wafer of 13 mg (1.3 cm diameter) was activated at 450 °C under vacuum (10−3 Torr) for 2 h inside a home-built IR cell with CaF2 windows, contacted with isobutanol vapor of 5.0 Torr at room temperature for 30 min, and evacuated at the same temperature under vacuum to a residual pressure of 10−3 Torr for another 30 min to remove physisorbed isobutanol. Then, the IR spectra were recorded on a Thermo-Nicolet 6700 FT-IR spectrometer with increasing temperature up to 300 °C, with a temperature interval of 5 °C and a holding time of 10 min at each temperature. The spectrum of gas-phase isobutanol was also recorded at room temperature and 5.0 Torr.
Si/Al ratio in the synthesis mixture | Addition of seedsb | Productc | Si/Al ratio in the productd |
---|---|---|---|
a The composition of the synthesis mixture is 0.5PMIOH·1.5HF·xAl2O3·1.0SiO2·5H2O, where x is varied between 0.01 ≤ x ≤ 0.2. All syntheses were carried out under rotation (60 rpm) at 175 °C for 14 days. b A small amount (4 wt% of the silica in the synthesis mixture) of calcined PST-24 with Si/Al = 200 was added as seed crystals. c The phase appearing first is a major phase. d Determined by elemental analysis. | |||
50 | No | PST-24 | 200 |
50 | Yes | PST-24 | 90 |
20 | No | PST-24 + amorphous | |
20 | Yes | PST-24 | 53 |
10 | No | SSZ-50 + amorphous | |
10 | Yes | PST-24 | 47 |
7.5 | No | Amorphous + SSZ-50 | |
7.5 | Yes | SSZ-50 | |
5.0 | No | SSZ-50 + unknown | |
5.0 | Yes | SSZ-50 |
Fig. 2 shows the PXRD patterns of H-PST-24(47) and H-PST-24(200) synthesized here. Comparison with the literature data reveals that both of them are phase-pure and highly crystalline,19 as further evidenced by the N2 micropore volumes in Table 2. The 27Al MAS NMR spectra of their as-made and proton forms are compared in Fig. 3. Like those of the as-made form, the spectra of the proton form exhibited no noticeable 27Al resonances around 0 ppm corresponding to extra-framework Al species, suggesting the high thermal stability of PST-24. However, this does not mean the complete lack of extra-framework Al species, because IR spectroscopy with adsorbed with pyridine indicates the presence of a non-negligible amount of Lewis acid sites in both H-PST-24(47) and H-PST-24(200) zeolites (Table 2) such as NMR-invisible Al in highly distorted extra-framework positions due to high quadrupolar coupling constant causing substantial signal broadening.24 We also note that their 27Al MAS NMR spectra show two resonances around 59 and 56 ppm in the tetrahedral Al region, but the spectra of their as-made form exhibit only one tetrahedral resonance around 58 ppm. It thus appears that upon OSDA removal, the eleven crystallographically distinct Al sites in the average structure of PST-24 are changed to two groups of Al sites with similar T–O–T angles. We speculate that interactions between the OSDA and the zeolite framework in as-made PST-24 would be strong enough to offset differences in the T–O–T angle of crystallographically distinct Al sites.
Catalyst IDa | Crystal shape and sizeb (μm) | N2 BET surface areac (m2 g−1) | Micropore volumec (cm3 g−1) | Acidityd (μmol pyridine g−1) | NH3 uptakee (mmol g−1) | |
---|---|---|---|---|---|---|
Brønsted | Lewis | |||||
a The values in parenthesis are the bulk Si/Al ratios determined by elemental analysis. b Determined by SEM. c Calculated from N2 adsorption data. d Determined from the intensities of the IR bands of retained pyridine at 1545 and 1455 cm−1 after desorption at 300 °C for 30 min, respectively.23 e Determined from NH3 TPD measurements. | ||||||
H-ZSM-5(50) | Spheres, 2.0–3.0 | 390 | 0.11 | 60 | 17 | 0.26 |
H-ZSM-5(95) | Spheres, 1.0–3.0 | 370 | 0.10 | 30 | 15 | 0.11 |
H-ferrierite(33) | Cuboids, 0.3–1.0 | 420 | 0.14 | 62 | 52 | 0.29 |
H-ferrierite(130) | Rectangular plates, 1.0 × 3.0 × 0.1 | 340 | 0.12 | 32 | 3 | 0.08 |
H-PST-24(47) | Aggregated rods, 0.05 × 0.5 | 410 | 0.13 | 47 | 39 | 0.26 |
H-PST-24(200) | Rectangular plates, 1.5 × 0.25 × 0.06 | 360 | 0.12 | 13 | 4 | 0.13 |
![]() | ||
Fig. 3 27Al MAS NMR spectra of the as-made (left) and proton (right) forms of PST-24(47) (bottom) and PST-24(200) (top). |
Fig. 4 shows the NH3 TPD profiles of three pairs of H-PST-24, H-ZSM-5 and H-ferrierite zeolites with relatively low and high Si/Al ratios. All the TPD profiles of zeolites with low Si/Al ratios (i.e., H-PST-24(47), H-ZSM-5(50) and H-ferrierite(33)) are characterized by two desorption peaks with maxima in the temperature ranges 280–330 and 410–440 °C, corresponding to NH3 desorption from weak and strong acid sites, respectively. No significant differences in the NH3 uptake (i.e., the density of acid sites) were found (Table 2), in good agreement with the similarity in their Si/Al ratios. The same trend is observed for the profiles of three zeolites with higher Si/Al ratios, although there are no weak acid sites. Therefore, the catalytic results obtained from these three pairs of zeolites would reflect mainly the effects of the geometrical constraints imposed by each of the zeolite pore structures.
Fig. 5 shows isobutanol conversion and isobutene yield as a function of time on stream (TOS) in isobutanol dehydration at 250 °C and 24.8 h−1 isobutanol WHSV over six zeolite catalysts with different framework structures and/or Si/Al ratios. These catalytic data were obtained in the presence of 9% water vapor in order to check whether biomass-derived isobutanol can be directly used without further separation of concomitant water. We were not able to observe any noticeable amount of diisobutyl ether, a bimolecular dehydration product, over all zeolite catalysts studied, indicating that none of their pore structures are large enough to catalyze the SN2-type dehydration of isobutanol.25
From the catalytic results in Fig. 5, it is clear that the initial isobutene yield, as well as the initial isobutanol conversion, is notably higher in the order H-ferrierite < H-ZSM-5 < H-PST-24, regardless of their Si/Al ratio. The excellent performance of H-PST-24 in isobutanol dehydration is further validated by the fact that the initial isobutene yield (61%) of H-PST-24(200) is considerably higher than the value (37%) of H-ZSM-5(50). This isobutene yield is even higher than that (52%) of phosphorus-modified ZSM-5 with a significantly higher Al content (Si/Al = 13) obtained at higher reaction temperature (280 vs. 250 °C) and a lower WHSV (7.4 vs. 24.8 h−1) than our conditions.17 We note that the initial isobutene yield and n-butenes/isobutene ratio (5.2% and 5.1 at conversion of 33%, respectively; Fig. 5) of H-ferrierite(33) are comparable to those (4.3% and 4.1 at conversion of 50%, respectively) of the same structure type of zeolite with a similar Si/Al ratio (28) at the same reaction temperature (250 °C) reported by Daele et al.,16 imparting the reliability of our catalytic results. We also note that both the isobutanol conversion and isobutene yield of all H-ZSM-5 and H-ferrierite catalysts decrease with TOS and reach a steady state after 3 h on stream. However, this is not the case of H-PST-24(50) and H-PST-24(200) because their conversion and yield decrease gradually over the period of TOS studied here. For example, the isobutene yield (35%) of H-PST-24(47) at 10 h on stream is still twice higher than that (17%) of H-ZSM-5(50). It is worth noting that the activity loss of the zeolite catalysts would result from coke deposition, leading to a decrease in the number of accessible active sites rather than changes in the reaction mechanism or in the nature of active sites. This can be supported by the fact that a conversion decrease accompanied no changes in product selectivity.
Fig. 5 also compares selectivities to isobutene and n-butenes (i.e., 1-butene, trans-2-butene and cis-2-butene) as a function of TOS in low-temperature isobutanol dehydration over all zeolite catalysts described above. It can be seen that while the initial isobutene selectivities (ca. 75%) of H-ZSM-5(50) and H-HZM-5(95) are higher by ca. 10% than those (61 and 66%, respectively) of H-PST-24 zeolites with similar Si/Al ratios, all of these four catalysts maintain almost constant isobutene selectivities over 10 h on stream. However, both H-ferrierite(33) and H-ferrierite(130) were found to show much lower isobutene selectivities (≤23%) from the beginning of the reaction. In fact, they were characterized by much higher selectivities to n-butenes that should be a result of the isomerization of isobutylcarbenium cations to linear carbocations. This shows that H-ferrierite is structurally much less selective for low-temperature isobutanol dehydration than H-ZSM-5 and H-PST-24.
We next investigated the regenerability of H-ZSM-5(50) and H-PST-24(47), the two most representative catalysts in this study, by repeating the isobutanol dehydration run under the same reaction conditions as those described above, but over a longer period (30 h) of TOS. After each run, the used catalyst was regenerated by calcination at 550 °C in dry air for 8 h. The initial isobutene yield of H-ZSM-5(50) decreases to only ca. 15% of its fresh form after three regeneration cycles. As shown in Fig. 6, however, the yield of H-PST-24(47) is still over 80% of the fresh catalyst. Thermal analysis reveals that the amount (4.7 vs. 9.1 wt%) of organic species deposited during isobutanol dehydration at 250 °C for 30 h on stream is considerably higher in used H-ZSM-5(50) than in H-PST-24(47). Also, 1H–13C CP MAS NMR spectroscopy shows that used H-PST-24(47) gave no detectable resonances around 130 ppm due to the formation of hard aromatic coke (Fig. S1†), unlike used H-ZSM-5(50) and H-ferrierite(33), explaining its higher regenerability. However, because differences in the acidic properties of H-PST-24(47) and H-ZSM-5(50) are negligible (Table 2), it is still unclear what causes the formation of aliphatic coke only in this zeolite catalyst. One possible explanation is the unique pore structure of PST-24, although further study is necessary.
To understand why H-PST-24 is considerably more active for low-temperature isobutanol dehydration than H-ZSM-5 and H-ferrierite, we compared the IR spectra of H-ZSM-25(50), H-ferrierite(33) and H-PST-24(47) before and after isobutanol adsorption at room temperature. As shown in Fig. 7, isobutanol adsorption on these zeolites led to the appearance of two sharp bands at 2965–2968 and 2879–2882 cm−1 and one sharp band at 1472–1474 cm−1 that could be assigned to the C–H stretching and deformation modes of the molecules adsorbed on the zeolite Brønsted acid sites (BASs; bridging Si–OH–Al groups), respectively.26 It should be noted that the intensities of IR bands from adsorbed isobutanol are much weaker on H-ferrierite(33) than on H-ZSM-5(50) and H-PST-24(47). Fig. 7 also shows that after isobutanol adsorption, the intensity of the BAS band around 3610 or 3600 cm−1 is again significantly weaker in the former zeolite than in the latter two zeolites. Therefore, we can conclude that the intracrystalline diffusion of isobutanol with a kinetic diameter of 5.4 Å (ref. 27) in ferrierite at room temperature is severely restricted, probably due to the small 10-ring size (4.2 × 5.4 Å) compared to ZSM-5 and PST-24.19,28 As previously reported by Daele et al.,16 in fact, the poor isobutene selectivity and yield of both H-ferrierite(33) and H-ferrierite(130) can be rationalized by considering that isobutanol dehydration occurs mostly on their external BASs.
It is also remarkable that the broad band around 3500 cm−1 in the spectrum of H-ZSM-5(50) assigned to internal Si–OH groups29 migrates to 3260 cm−1 after isobutanol adsorption (Fig. 7), indicative of intermolecular hydrogen bonding. This suggests that isobutanol can be adsorbed on not only terminal Si–OH groups but also internal ones. We also carried out variable-temperature isobutanol-IR measurements on H-ZSM-5(50) and H-PST-24(47) and monitored the decrease in C–H stretching and C–H deformation band intensities of adsorbed isobutanol at temperatures up to 300 °C (Fig. S2†). When normalized against the spectra obtained at room temperature, H-PST-24(47) was found to retain a relatively higher amount of isobutanol than H-ZSM-5(50) in the entire temperature range studied (Fig. 8). This indicates a stronger interaction of isobutanol with BASs in the former zeolite, partly explaining why H-PST-24(47) shows a considerably higher conversion in isobutanol dehydration at 250 °C than H-ZSM-5(50) (Fig. 5).
To better understand the pore structure effect of H-PST-24(47) and H-ZSM-5(50) on their isobutanol dehydration activity, we performed periodic density functional theory calculations using the exchange–correlation functional suggested by Perdew–Burke–Ernzerhof (PBE)30 and with the additional empirical dispersion correction proposed by Grimme31 implemented in the Vienna ab initio simulation package (VASP).32 Unfortunately, the simultaneous alteration of both the transition state and intrazeolitic active site model did not allow us to capture the most energetically favorable transition state for this reaction, revealing unexpected complexity in calculations. Given the basic principle that catalytic conversion of reactant begins with its adsorption on the active sites of heterogeneous catalysts, on the other hand, it is not difficult to expect that the strong adsorption of reactants can lead to a high conversion as shown in Fig. 5, which is also the case of various oxygenated molecules.33,34 Therefore, we speculate that the unique pore architecture of H-PST-24 rather than its acidic properties is mainly responsible for stronger isobutanol adsorption and thereby higher activity, making this zeolite an efficient catalyst for the dehydration of biomass-derived isobutanol.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4cy00063c |
‡ Present address: Department of Chemical Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802, United States. |
This journal is © The Royal Society of Chemistry 2024 |