Direct Conversion of 1,3-Butanediol to 1,3-Butadiene over ZSM-22 Catalysts: Influence of the Si/Al Ratio

Abstract

ZSM-22 zeolites with different Si/Al ratios (38, 50, 80) were prepared via a hydrothermal synthesis method, investigated for the catalytic dehydration of 1,3-butanediol (1,3-BDO) to butadiene (BD) at 300 °C. The catalytic performance of the synthesized materials was related to their properties and compared to a commercial ZSM-22 zeolite (Si/Al = 30). ZSM-22 (50) exhibited a quick decline in conversion, a lower BD selectivity, and higher propylene selectivity compared to the other materials, which could be attributed to the presence of strong Lewis acid sites and silanol nests. The Lewis sites favor the cracking of the intermediate 3-buten-1-ol (3B1OL) into propylene, while the silanol nests interact with the free hydroxyl group of 3B1OL, potentially inhibiting further dehydration towards BD. The highest initial BD yield of 74% was observed over ZSM-22 (80), while the highest initial BD productivity of 2.7 g_BD·g⁻¹_cata·h⁻¹ was achieved over ZSM-22 (38). After 22 h time on stream (TOS), c-ZSM-22 and ZSM-22 (38) outperformed previously reported catalysts from the literature, with productivities amounting to 1.3 g_BD·g⁻¹_cata·h⁻¹ and 1.2 g_BD·g⁻¹_cata·h⁻¹, respectively, at a site time of 6.6 mol_H+·s·mol⁻¹_1,3-BDO.

1. Introduction

Short-chain olefins (C₂–C₄) are among the most important platform chemicals in the petrochemical industry and are predominantly produced by steam cracking. They are fundamental building blocks for producing various plastics and synthetic rubbers. Their global demand has been steadily increasing over the years, driven by their widespread applications in the electronic and automotive industries, among others [1]. Among these olefins, 1,3-butadiene (BD) is a key building block in the petrochemical industry. It is predominantly used as a precursor in the production of styrene-butadiene rubber (SBR), polybutadiene rubber, and acrylonitrile butadiene styrene polymer (ABS) [2,3]. The ethylene demand heavily affects the current supply and price of butadiene [4]. More than 90% of butadiene is produced as a by-product of naphtha steam cracking in ethylene plants, which is accompanied by a high carbon footprint with cradle-to-gate CO₂ emissions amounting to 1.2–1.8 ton_CO2/ton_BD [5]. However, recent technological advancements in North America have rendered shale gas as an economically more favorable feedstock, while European crackers are transitioning to liquefied petroleum gas as a primary feedstock [6,7]. These trends have significantly reduced the global supply of BD. Hence, the butadiene market faces a dual challenge of ensuring a sufficient butadiene supply while mitigating carbon emissions, making the development of a bio-based alternative technology of prime importance.

The ethanol-to-butadiene (ETB) route via the Lebedev process is a widely investigated alternative for BD production. However, this complex process involves several catalytic reactions, including dehydrogenation, aldol condensation, hydrogenation, and dehydration [8]. Moreover, the reaction generally occurs at high temperatures (>400 °C) over promoted (Zr, Ag, Cu or Au) MgO/SiO₂ catalysts exhibiting low BD productivities of 0.2 g_BD·g⁻¹_cata·h⁻¹ and BD yields of approximately 50% [2,9]. In contrast, obtaining BD from butanediol (BDO) isomers (1,4-, 2,3-, and 1,3-BDO) involves only two consecutive dehydrations, which facilitates the catalyst design for this type of reaction. Additionally, it has been demonstrated that these BDO isomers can be obtained from the microbial fermentation of sugars [10,11,12,13,14,15,16,17], positioning the acid-catalyzed dehydration of BDO as a promising sustainable alternative for BD production. Since cyclodehydration of 1,4-BDO and pinacol rearrangement of 2,3-BDO yield tetrahydrofuran and methylethylketone as major products, respectively [18,19,20,21,22], 1,3-BDO becomes an attractive feedstock for the on-purpose production of 1,3-butadiene. Moreover, this specific butanediol isomer can be produced via gas fermentation by Cupriavidus Necator using CO₂ as the sole carbon source [23], highlighting the potential value of 1,3-BDO as a sustainable feedstock for the on-purpose BD production.

The catalytic dehydration of 1,3-BDO into BD proceeds via two consecutive dehydration steps, generating unsaturated alcohols (UOLs) as intermediates: 2-buten-1-ol (2B1OL), 3-buten-1-ol (3B1OL), and 3-buten-2-ol (3B2OL). However, several side reactions can occur depending on the catalyst properties, including 1,3-BDO dehydrogenation into methyl vinyl ketone (MVK), see Figure 1. MVK hydrogenation results in methyl ethyl ketone (MEK) and 2-butanol. Moreover, 2B1OL and 3B1OL can be hydrogenated, resulting in the formation of 1-butanol, which can be converted into butanal via dehydrogenation. Finally, 3B1OL is prone to cracking via the reverse Prins reaction, generating propylene and formaldehyde [24,25]. Various solid acid catalysts have been investigated for the catalytic dehydration of 1,3-BDO, including silica-alumina, Al-SBA-15, and zeolites [24,25,26,27,28,29,30,31,32]. These works have implemented several strategies to fine-tune the catalyst properties to increase the BD yield. According to the work of Pera-Titus [28], Brønsted acid sites with a medium strength are responsible for a high BD selectivity. This hypothesis is also supported by the reports of Padro et al. [30] and Sato et al. [31], which demonstrated that Brønsted acid sites favor the dehydration of the UOLs towards BD, while Lewis acid sites facilitate the cracking of 3B1OL. The best-performing materials from these works are tungstophosphoric acid supported on SiO₂ (initial BD yield and productivity amounting to 75% and 1.92 g_BD·g⁻¹_cata·h⁻¹, respectively, at 300 °C) [30], WO₃-modified SiO₂ (BD yield and productivity after 5 h TOS of 73.4% and 0.17 g_BD·g⁻¹_cata·h⁻¹, respectively, at 300 °C) [31], and Ag-modified silica-alumina (BD yield and productivity after 5 h TOS of 68.9% and 1.33 g_BD·g⁻¹_cata·h⁻¹, respectively, at 250 °C) [32]. Despite the improvements in BD yields and selectivities, the overall productivity remains limited, particularly in long-term experiments. Most catalysts are evaluated within the first 5 h time on stream TOS, making it difficult to assess their potential for industrial application. Notably, Y₂Zr₂O₃ has been reported at 375 °C with BD yields of 95% and 87% after 10 and 30 h TOS, respectively. Nevertheless, its productivity remains low, with values of 0.38 and 0.35 g_BD·g⁻¹_cata·h⁻¹ at 10 and 30 h TOS, respectively [33].

Additionally, in the case of zeolites, the topology has been identified as a determining factor for BD selectivity. Lee et al. [25] demonstrated that zeolites with a 1D and 2D topology, such as ZSM-22 and FER, suppress 3B1OL cracking, thereby increasing the BD selectivity. Although BD yields over 65% have been reported over ZSM-22 (Si/Al = 160) and FER (Si/Al = 130), the BD productivity remains relatively low (e.g., productivity of 0.59 g_BD·g⁻¹_cata·h⁻¹ after 10 h TOS over FER (Si/Al = 130) at 300 °C). In contrast, ZSM-5 has bigger pores compared to ZSM-22 and FER, which makes it possible for 3B1OL to coordinate its hydroxyl group with the negatively charged pi cloud of the C=C double bond, causing this intermediate to be less active for dehydration. This observation positions 1D and 2D zeolites as better alternatives for BD production from 1,3-BDO. However, no further investigations into these types of zeolites has been carried out.

Pera-Titus and coworkers have thoroughly investigated the influence of the Si/Al ratio in ZSM-5 on the dehydration behavior of 1,3-BDO [28], aiming at unraveling the correlation between the acid properties and the catalytic performance to improve the BD yield. These authors reported that higher Si/Al ratios are favorable for BD selectivity. Moreover, higher Si/Al ratios are favorable for the catalyst stability. For instance, ZSM-5 (Si/Al = 130) achieved a BD yield of 60% at 300 °C with a BD productivity amounting to 0.48 g_BD·g⁻¹_cata·h⁻¹ after 8 h time on stream (TOS) [25]. However, no studies on the influence of the acid properties of 1D or 2D zeolites on the catalytic dehydration of 1,3-BDO have been conducted yet. It is, therefore, an interesting strategy to investigate the influence of acid density in ZSM-22 zeolites on BD selectivity.

In this study, the influence of the acid properties of ZSM-22 on the catalytic dehydration of 1,3-BDO is investigated. Via an in-house recipe from our research group, ZSM-22 zeolites are synthesized with a similar crystal shape and size as investigated for the dehydration of 1,3-BDO by Lee et al. [25], but with lower Si/Al ratios, i.e., 38, 50, and 80. The structural properties are thoroughly analyzed through X-ray diffraction (XRD), scanning electron microscopy (SEM), Ar sorption, inductively coupled plasma mass spectroscopy (ICP-MS), ²⁷Al and ²⁹Si solid-state magic angle spinning nuclear magnetic resonance spectroscopy (MAS-NMR), and in situ Fourier transform infrared spectroscopy (FTIR) with pyridine as a probe molecule to determine the acid sites. Moreover, adsorption experiments with 3B1OL as a probe molecule are performed to gain insight into the adsorption behavior of this unsaturated alcohol within a zeolite framework, since this intermediate is prone to cracking. All these properties are compared to commercial ZSM-22, which is used as reference material. Finally, the catalysts are screened over a long period (up to 22 h) to investigate the influence of the acid properties on the BD yield and the long-term productivity behavior is compared to literature-reported catalysts.

2. Results and Discussion

2.1. Characterization of the Zeolite Materials

The normalized XRD patterns of both the commercially available ZSM-22 and the synthesized ZSM-22 zeolites with varying Si/Al ratios are depicted in Figure 2. All samples exhibit diffraction patterns consistent with those reported by Zhai and coworkers [34], characterized by well-defined peaks at 2θ = 8.1°, 20.3°, 24.2°, 24.6°, 25.7°, and 35.6°, which are indicative of the TON topology [35]. It is worth noting that no additional sharp peaks corresponding to impurities such as ZSM-5 and cristobalite were observed in the range of 2θ = 5–30°. This indicates that the addition of seeds during synthesis promotes the formation of pure ZSM-22. The average crystal size of the zeolites determined by the Scherrer equation for 2θ = 35.6° amounts to 77 nm for c-ZSM-22, 63 nm for ZSM-22 (38), 69 nm for ZSM-22 (50), and 76 nm for ZSM-22 (80), which is summarized in Table S1. The smaller value for the average crystal size of ZSM-22 (38) and ZSM-22 (50) can be attributed to the higher alkaline content in the precursor gel, which increased the number of nucleation sites [36].

Figure 3 presents the SEM images of c-ZSM-22, ZSM-22 (38), ZSM-22 (50), and ZSM-22 (80). The commercial sample (see Figure 3a) consists of short prismatic crystals with dimensions ranging between 100 and 600 nm in length (more detailed in Figure S3.2), which are densely stacked together into agglomerated clusters with overall dimensions between 3 and 10 µm (Table S1). The crystal width, which is approximately 60–80 nm, is consistent with the values calculated by the Scherrer equation. In contrast, the ZSM-22 zeolites synthesized in this work (Figure 3b–d) exhibit a different morphology. They consist of discrete nanorods stacked together to form elongated needle-like bundles with dimensions between 2.5 and 4 µm in length. The nanorods themselves have a length of ~200 nm and widths between 60 and 100 nm (Figure S3.2), which is in good agreement with the calculated widths by the Scherrer equation. As can be seen from Figure 3b–d, the nanorods are closely aligned, resulting in dense needle-like bundles exhibiting widths between 250 and 500 nm, which is consistent with previous observations by Zhai and coworkers [34]. However, they reported smaller bundles, which could be attributed to the lower rotation speed of 50 rpm used during the synthesis in their work. The rotation speed of 100 rpm in our work could facilitate the deposition of zeolite crystals on the wall of the Teflon liners, which favors particle growth. The dense needle-like bundles further agglomerate into spherical clusters of about 10 µm width (Figure S3.1).

The normalized ²⁷Al MAS NMR spectra of the ZSM-22 zeolites, as depicted in Figure 4a, exhibit a dominant peak at 58 ppm. This is characteristic to tetrahedrally coordinated Al species, indicating that the majority of the Al atoms is successfully incorporated into the zeolite framework. Additionally, a broad signal centered at 8 ppm for ZSM-22 (50), which is less pronounced in the other materials. This could be associated with octahedrally coordinated Al, i.e., extra-framework alumina, which acts as Lewis acid sites [37]. After normalization, the ²⁹Si MAS NMR spectra can be deconvoluted into seven peaks (Figure S4). Two peaks are located in the region between −99 and −105 ppm, which are attributed to Si(1Al) coordination, while five bands are located between −106 ppm and −115 ppm, corresponding to Si(0Al) coordination [38,39]. The analysis of the deconvoluted bands reveals Si/Al ratios of 32, 40, 53, and 87 for c-ZSM-22, ZSM-22 (38), ZSM-22 (50), and ZSM-22 (80), respectively, which is consistent with the ICP-MS results (

Table 1. Physicochemical properties of the catalysts used in this study.

Sample	Si/AlICP−MS	Ar-Sorption			Brønsted Acid Sites [µmol/g]		Lewis Acid Sites [µmol/g]
		SBET [m2/g]	Vtot [cm3/g]	Vmicro [cm3/g]	Tevac = 150 °C	Tevac = 300 °C	Tevac = 150 °C	Tevac = 300 °C
c-ZSM-22	30	216	0.09	0.07	128	110	14	9
ZSM-22 (38)	38	280	0.12	0.09	136	110	27	15
ZSM-22 (50)	50	289	0.12	0.09	105	86	29	24
ZSM-22 (80)	80	291	0.11	0.09	87	71	26	13

). The lower Si/Al ratios compared to the targeted ratio in the synthesis gel are also reported for other ZSM-22 zeolites [34,35], and can be explained by the presence of K⁺ cations in the synthesis mixture. In the presence of hydroxides, the Si-O-Si and Si-O-Al bonds are assembled around alkali cations such as the K⁺ ions originating from the applied KOH, to form aluminosilicate oligomers. Due to the high charge density of these cations, a higher amount of Al atoms is incorporated into the zeolite framework to compensate for the positive charge. Moreover, the dissolution of colloidal silica is slow, which results in a lower incorporation of Si in the zeolite framework [40,41].

Figure 4b presents the argon sorption isotherms of all ZSM-22 zeolites investigated in this study. For clarity, the isotherms of ZSM-22 (50) and ZSM-22 (80) have been lifted by 0.5 and 1.0 mmol/g, respectively. According to the IUPAC classification, all zeolites exhibit type Ia isotherms, characterized by a steep Ar uptake at very low pressures (p/p° < 0.05). This type of isotherm is related to microporous materials predominantly composed of narrow micropores (pore widths below 1 nm). In addition, all self-synthesized materials exhibit a small hysteresis loop at high relative pressures (p/p° > 0.90), suggesting the presence of intercrystalline mesopores, which are formed by the dense aggregation of the zeolite particles, as depicted in Figure 3. These types of isotherms are typically observed in ZSM-22 zeolites obtained via the hydrothermal synthesis method [34,35,42,43,44,45,46]. The textural properties of all materials (specific surface area and pore volumes) are listed in Table 1. The in-house prepared ZSM-22 samples exhibit a comparable specific surface area, which is higher than the commercially available ZSM-22 zeolite. Additionally, their micropore volume exceeds that of c-ZSM-22 by more than 25%.

The normalized FTIR spectra of the zeolites in the region corresponding to the ν(OH) stretching vibrations are presented in Figure 4c, and their deconvoluted spectra are reported in the Supporting Information (Figures S5.1–S5.4). All samples exhibit a prominent band between 3750 and 3720 cm⁻¹, indicating the presence of isolated silanol groups [40,47,48,49,50]. The c-ZSM-22 and ZSM-22 (38) spectra contain a sharp peak centered at 3743 cm⁻¹. In contrast, the spectra of ZSM-22 (50) and ZSM-22 (80) exhibit a broad band around 3735 cm⁻¹, with a tail extending towards lower wavenumbers related to looser stretch vibrations. The band can be decomposed into three main peaks at 3742 cm⁻¹, 3730 cm⁻¹, and 3710 cm⁻¹ for all zeolite materials. According to the literature [48,51], the bands between 3745 cm⁻¹ and 3740 cm⁻¹ are assigned to isolated silanols on the external surface of the zeolites, while the bands between 3730 and 3700 cm⁻¹ correspond to isolated silanol groups in the micropore channels. The band around 3730 cm⁻¹ is related to hydroxyl groups linked to Si connected to Al atoms (silanol-Al), which are sometimes hypothesized to be tri-coordinated in the zeolite framework, i.e., Lewis acid sites [48,51]. It is worth noting that c-ZSM-22 and ZSM-22 (38) mainly exhibit external silanol groups, while the self-synthesized materials primarily contain isolated silanol groups inside the zeolite channels, which is also reflected in the long tail towards lower energy-stretching vibrations. Moreover, the band at 3743 cm⁻¹ is more pronounced in the deconvoluted spectra of c-ZSM-22 and ZSM-22 (38) in Figures S5.1 and S5.4. Additionally, all samples exhibit a strong band at 3600 cm⁻¹, which corresponds to isolated bridged hydroxyl groups (Al-O(H)-Si) within the zeolite framework, and are attributed to strong Brønsted acid sites [47,48,49,52,53]. Furthermore, a broad band at 3530 cm⁻¹ is observed for the materials synthesized in this work, which is associated with strong hydrogen bonding of internally located Si-OH groups. These silanol groups are typically formed due to framework defects, such as the absence of a T atom in the zeolite framework, and are commonly referred to as silanol nests [50,52]. It is noticeable that the amount of silanol nests decreases in the following order—ZSM-22 (50) > ZSM-22 (80) > ZSM-22 (38)—which is reflected in the values of the peak areas (summarized in Table S1) of the deconvoluted bands of Figures S5.1–S5.4.

The normalized FTIR spectra of the zeolites recorded after pyridine adsorption at 150 °C and subsequent evacuation at 300 °C are depicted in Figure 4d. The interaction of pyridine with the zeolite surface results in several characteristic peaks between 1700 cm⁻¹ and 1400 cm⁻¹. The coordination of pyridine to Lewis acid sites results in two bands at 1454 cm⁻¹ and 1618 cm⁻¹, while the protonation of pyridine on Brønsted acid sites results in two peaks at 1545 cm⁻¹ and 1639 cm⁻¹. An additional band at 1489 cm⁻¹ can also be detected, which is attributed to overlapping contributions of pyridine interacting with Lewis and Brønsted acid sites [54]. Due to the overlap of the ν_8a bands between 1600 cm⁻¹ and 1640 cm⁻¹, deconvolution and integration of the signals at 1454 cm⁻¹ and 1545 cm⁻¹ is used to determine the amount of Lewis (A_L) and Brønsted (A_B) acid sites, respectively. Table 1 summarizes the results after pyridine evacuation at 150 °C and 300 °C. All zeolites have an A_B/(A_B+A_L) ratio between 0.77 and 0.90 after evacuation at 150 °C, which increases to a value between 0.78 and 0.92 when the temperature is increased to 300 °C, reflecting the Brønsted nature of the acid sites. Interestingly, ZSM-22 (50) contains a similar amount of acid sites as c-ZSM-22 and ZSM-22 (38), but contains more Lewis acid sites. Additionally, the acid strength can be estimated by comparing the amounts of pyridine remaining on the acid sites after evacuation at 150 °C and 300 °C, i.e., by the A_B,300/A_B,150 and A_L,300/A_L,150 ratios for Brønsted and Lewis acid sites, respectively, which is presented in Table S1 [30]. All zeolites exhibit a comparable, high Brønsted acid strength, which is reflected by the A_B,300/A_B,150 ratios between 0.81 and 0.86. However, while c-ZSM-22, ZSM-22 (38), and ZSM-22 (80) exhibit a comparable medium Lewis acid strength, i.e., an A_L,300/A_L,150 ratio with values between 0.50 and 0.65, ZSM-22 (50) contains stronger Lewis acid sites, reflected by the A_L,300/A_L,150 ratio of 0.82.

Three unsaturated alcohols are generated during the dehydration of 1,3-BDO to BD, i.e., 2B1OL, 3B1OL, and 3B2OL. It is known that 3B1OL is the only one of the intermediates that can coordinate its hydroxyl group to its internal C=C bond within the zeolite pores, which could make the molecule less active for further dehydration towards BD (Figure S6) [25]. To investigate the adsorption behavior of this unsaturated alcohol in ZSM-22, the adsorption was monitored by in situ FTIR experiments. Figure 5 presents the normalized FTIR spectra of 3B1OL adsorbed on all the catalysts investigated in this work. The bands observed between 2750 cm⁻¹ and 3100 cm⁻¹ correspond to the ν(CH) stretching vibrations of the unsaturated alcohol and are consistent with previous literature [25]. In the ν(OH) stretching region, several additional bands can be observed compared to the reference spectrum. Upon the adsorption of 3B1OL at room temperature followed by evacuation, the band at 3743 cm⁻¹ remains unchanged, indicating no interaction between the isolated silanol groups and 3B1OL. In contrast, the band between 3600 and 3590 cm⁻¹, related to Brønsted acid sites, has almost completely disappeared, suggesting a strong interaction between the unsaturated alcohol and these acid sites [55]. Additionally, a sharp peak is noticeable in the spectra of c-ZSM-22 at 3695 cm⁻¹, while this appears to be a shoulder in ZSM-22 (38). This peak could be associated with the interaction between the hydroxyl group of the unsaturated alcohol and the silicate wall inside the zeolite channels [56]. Furthermore, a broad band centered at 3540 cm⁻¹ emerges in all samples, which is assigned to weakly perturbed hydroxyl groups. It has been proposed that this band is associated with the interaction of the hydroxyl group of the alcohol and the Brønsted acid site of the zeolite [55]. Interestingly, the intensity of this band does not directly correlate with the acid density, since a more prominent band is observed in the order of ZSM-22 (80) > ZSM-22 (50) > ZSM 22 (38) > c-ZSM-22, suggesting that additional effects within the zeolite pores are present. One possible explanation is the formation of hydrogen bonds between the 3B1OL molecules. Di Iorio et al. [57] reported that in more hydrophobic zeolites, a band near 3510 cm⁻¹ becomes more pronounced upon 2-butanol adsorption. They attributed this band to the formation of hydrogen bonds between alcohol dimers because similar behavior has been observed for short-chain alcohols (C1–C4) in non-polar solvents. Since ZSM-22 (80) contains less Al, it is, therefore, more hydrophobic, and similar intermolecular interactions might contribute to the increased intensity of the band located at 3540 cm⁻¹.

Finally, a broad band centered at 3300 cm⁻¹ is observed in ZSM-22 (50) and ZSM-22 (80), while this is less pronounced in ZSM-22 (38) and absent in c-ZSM-22. Di Iorio and coworkers [57] observed a similar band between 3300 and 3400 cm⁻¹ for 2-butanol adsorption in Beta zeolites. They attributed this peak to hydrogen bond formation between the hydroxyl group of the alcohols and zeolite defects such as Si-OH groups, resulting in the formation of dimeric and polymeric 2-butanol, resembling a liquid-like phase of bulk 2-butanol within the zeolite framework. Similarly, the band centered at 3300 cm⁻¹ in Figure 5 resembles the ν(OH) stretching vibration related to hydrogen bonding in liquid 3B1OL. Since ZSM-22 (50) and ZSM-22 (80) contain more silanol nests than ZSM-22 (38), it can be suggested that there might be an interaction between the silanol nests and the free hydroxyl group of 3B1OL, resulting in a disordered liquid-like phase of hydrogen bonds.

2.2. Catalytic Results: Dehydration of 1,3-Butanediol into Butadiene

Figure 6 presents the 1,3-BDO conversion and product selectivities as a function of time on stream (TOS) during the vapor phase dehydration of 1,3-BDO over the zeolite materials at 300 °C, while

Table 2. Catalytic performance and butadiene productivity of all zeolites used for the dehydration of 1,3-BDO at 1 h and 22 h TOS. The reaction is performed at 300 °C, and the reactant feed consists of 20 wt.% 1,3-BDO in dioxane. The feed rate for c-ZSM-22 and ZSM-22 (38) was 1.5 g/h, while this was 0.8 g/h for ZSM-22 (50) and ZSM-22 (80).

Zeolite	TOS [h]	Conversion [%]	Selectivity [%]					Productivity [gBD·g−1cata·h−1]
			PE	BD	BuAL	MEK	3B1OL
c-ZSM-22	1	100	10	71	0	18	0	2.5
	22	61	10	58	1	11	21	1.3
ZSM-22 (38)	1	100	10	73	0	16	0	2.7
	22	58	9	61	1	12	17	1.2
ZSM-22 (50)	1	99	16	68	0	14	1	2.0
	22	31	14	55	2	11	18	0.5
ZSM-22 (80)	1	99	10	74	0	15	0	1.4
	22	57	10	62	1	11	16	0.7

lists the relevant catalytic performance indicators of the zeolites after 1 h and 22 h TOS. In all cases, the major products detected were butadiene, propylene, 3-buten-1-ol, and methyl ethyl ketone, while only small amounts (selectivity < 1.5%) of butanal were observed. Additionally, no 3-buten-2-ol nor 2-buten-1-ol were present in the product pool. During the acid-catalyzed dehydration of 1,3-BDO, the secondary alcohol 3B2OL is more favorable to be dehydrated. Therefore, it is evident that 3B2OL is the least probable reaction intermediate, which could explain its absence in the reactor effluent. Additionally, according to Zaitsev’s rule, dehydration of 1,3-BDO results in the formation of the most substituted unsaturated alcohol, i.e., 2B1OL, compared to 3B1OL. However, no 2B1OL is observed in the product pool, which could be related to its high reactivity towards further dehydration into butadiene in the presence of Brønsted acid sites [27,58]. The presence of propylene is related to the cracking of 3B1OL, while the formation of MEK is directly formed from 1,3-BDO after a cascade of dehydration and dehydrogenation-hydrogenation reactions (Figure 1) [24,30].

Both commercial ZSM-22 and ZSM-22 (38) achieve full BDO conversion during the first 5 h TOS, which then gradually decreases as the reaction proceeds. The initial BD selectivity over c-ZSM-22 and ZSM-22 (38) amounts to 71% and 73%, respectively. After 15 h TOS, the commercial sample stabilizes at a conversion of 62%, which is maintained for the remaining 7 h. However, ZSM-22 (38) reaches a stable conversion of 70% after 9 h TOS, which is maintained for 8 h before further deactivation occurs. After 22 h TOS, the conversion reaches a value of 58% over ZSM-22 (38). Once a stable regime of product selectivity is obtained, only minor differences in product distribution are observed between c-ZSM-22 and ZSM-22 (38). For instance, c-ZSM-22 exhibits a lower BD and MEK selectivity than ZSM-22 (38), but produces more 3B1OL. No difference in propylene or butanal selectivity is observed. The similar behavior in the product distribution might be associated with the nature, strength, and number of acid sites. It is known that Lewis acid sites favor the cracking of 3B1OL, while Brønsted acid sites promote its further dehydration to BD [30]. Since both catalysts exhibit a similar distribution of acid sites (Table 1), a similar product selectivity can be expected. However, ZSM-22 (38) exhibits weaker Lewis acid sites, which contribute to a reduced 3B1OL cracking, resulting in a slightly higher BD selectivity (Table 2). Although both zeolites exhibit a stable conversion regime, deactivation of ZSM-22 (38) is observed. This could be attributed to the bigger particle size of ZSM-22 (38) (2–4 µm), which is built up from stacked nanorods to form elongated needle-like bundles. As a result, ZSM-22 (38) exhibits longer diffusion path lengths for reactants and reaction products. Therefore, it might be more difficult for the coke precursors to reach the external surface, which could lead to deactivation by coke buildup inside the zeolite channels [59].

ZSM-22 (80) exhibits a similar behavior to that of c-ZSM-22 and ZSM-22 (38). Full 1,3-BDO conversion is achieved within the first 6 h TOS, with an initial BD selectivity of 74%. However, a rapid decline in conversion is observed, reaching 68% after 11 h TOS. This is followed by a gradual decrease to 57% at 22 h, similarly to the trend observed for ZSM-22 (38), which could be attributed to the similar acid properties, i.e., acid strength and the number of Brønsted acid sites. The catalyst deactivation could be associated with the morphology of ZSM-22 (80). Similarly to ZSM-22 (38), bigger particles are observed for ZSM-22 (80), which might hinder the diffusion of reactants, reaction products, and coke precursors [60,61]. Moreover, ZSM-22 (80) contains a higher amount of silanol nests (Figure 4c and Table S1), which could trap these coke precursors and enhance carbon deposition inside the pores, decreasing the accessibility towards other acid sites [28,59]. Additionally, as depicted in Figure 5, hydrogen bonding of 3B1OL (peak at 3300 cm⁻¹) occurs, which is suggested to be related to the interaction of 3B1OL’s hydroxyl group with silanol nests. This could result in trapping of this unsaturated alcohol, making the hydroxyl group of 3B1OL not available for dehydration. Consequently, 3B1OL might crack, which can react further to produce coke precursors, which the silanols could retain.

In contrast to the other zeolites, ZSM-22 (50) can only maintain full conversion during the first 3 h TOS, after which it rapidly declines, reaching a value of 40% after 9 h TOS. The conversion keeps decreasing during the overall experiment and reaches a value of 31% after 22 h TOS. Furthermore, this zeolite exhibits an overall lower BD selectivity compared to the other catalysts. This is accompanied by a notably higher selectivity towards propylene (C₃₌), suggesting a shift in the reaction pathway, possibly due to differences in acid site distribution or strength. As depicted in Figure 3, ZSM-22 (50) comprises particles of size similar to those in ZSM-22 (38), resulting in a longer diffusion path length compared to c-ZSM-22. Moreover, ZSM-22 (50) contains an increased amount of isolated silanol groups, which are known to be related to catalyst deactivation due to coke formation during the conversion of methanol to olefins [60]. Additionally, more silanol nests can be observed in ZSM-22 (50) (Figure 4c and Table S1), which are responsible for trapping and stabilizing coke precursors. Similarly to ZSM-22 (80), the interaction between 3B1OL and the silanol nests might cause the hydroxyl group of 3B1OL to be less active for dehydration, and favor cracking and formation of coke precursors. The combination of those defects with longer diffusion path lengths in ZSM-22 (50) compared to c-ZSM-22 could explain the rapid decrease in conversion. As can be seen in Table S1, ZSM-22 (50) contains a higher amount of Lewis acid sites, which are also stronger (A_L,300/A_L,150 ratio of 0.82) compared to c-ZSM-22, ZSM-22 (38), and ZSM-22 (80) (A_L,300/A_L,150 ratio of 0.65, 0.55, and 0.50, respectively). This could lead to more cracking of 3B1OL instead of dehydration, which explains the increased amount of propylene over this catalyst.

To obtain a more accurate comparison of the catalytic performance of the zeolite materials, the productivity of each catalyst was calculated and compared both among the tested zeolites and with previously reported catalysts. Both the productivity at early TOS and after 22 h TOS were considered and are listed in Table 2. Initial productivities of 2.5 g_BD·g⁻¹_cata·h⁻¹ and 2.7 g_BD·g⁻¹_cata·h⁻¹ are observed for c-ZSM-22 and ZSM-22 (38), respectively, outperforming all previously reported catalysts. The initial productivities of ZSM-22 (50) and ZSM-22 (80) are comparable to those observed for commercial SiO₂/Al₂O₃ and H₃PW₁₂O₄₀/SiO₂. Although ZSM-22 (38), ZSM-22 (50) and ZSM-22 (80) exhibit comparable 1,3-BDO conversion and BD selectivity at 1 h TOS, a notable difference can be observed in their productivity. This is attributed to the lower acid site density in the two latter catalysts compared to ZSM-22 (38), as summarized in Table 1. Since productivity is expressed per gram of catalyst, the reduced number of active sites in ZSM-22 (50) and ZSM-22 (80) leads to an overall lower productivity, despite similar activity. It is important to note that productivity values calculated at (near) complete conversion represent a lower estimate. This implies that a smaller amount of catalyst could potentially achieve the same conversion and, consequently, a higher productivity.

Although the BD yield of all ZSM-22 zeolites decreases over time, and higher BD yields can be achieved over other types of catalysts (e.g., Y₂Zr₂O₇ and WO₃/SiO₂), the overall BD productivity of ZSM-22 (50) and ZSM-22 (80) after 22 h TOS remains comparable to other zeolite materials documented in the literature (e.g., 0.48 g_BD·g⁻¹_cata·h⁻¹ over ZSM-5 (130) after 8 h TOS and 0.59 g_BD·g⁻¹_cata·h⁻¹ over FER (130) after 10 h TOS) [25,28]. Moreover, the BD productivities of c-ZSM-22 and ZSM-22 (38) after 22 h TOS surpass all previously reported catalysts. Furthermore, since productivity in the literature is mostly reported between 0 and 10 h TOS, the values observed in this work demonstrate the enhanced performance of ZSM-22 zeolites over prolonged reaction times. Consequently, these catalysts could serve as a more efficient option for sustainable BD production.

3. Materials and Methods

3.1. Materials

A commercial sample of NH₄-ZSM-22 (<0.1 wt.% Na₂O) was obtained from Bonding Chemical (Si/Al = 30–40). This zeolite was calcined under air at 550 °C (heating ramp of 1 °C·min⁻¹ from room temperature) for 4 h to obtain the protonic form, i.e., H-ZSM-22. This sample is referred to as c-ZSM-22. Potassium hydroxide (KOH, 85%), aluminum sulfate (Al₂(SO₄)·nH₂O, 98%), 1,6-diaminohexane (DAH, 98%), colloidal silica (Ludox AS-40, 40 wt.% in water), ammonium nitrate (NH₄NO₃, 98%), and 1,3-butanediol (99.5%) were purchased from Merck Life Science. 1,4-dioxane (99.9%) was purchased from Chem-Lab Analytical. All chemicals are used without any further purification.

3.2. Catalyst Preparation

In this study, ZSM-22 zeolites with a targeted Si/Al ratio of 45, 70, and 100 were synthesized using the hydrothermal synthesis method reported by Zhai et al. [34], with some slight adjustments. First, three homogeneous solutions were prepared: KOH, DAH, and Al₂(SO₄)·nH₂O were dissolved in water under continuous stirring (400 rpm). Additionally, colloidal silica was diluted in water under the same stirring conditions. Subsequently, the DAH solution was added to the KOH solution, followed sequentially by the Al₂(SO₄)·nH₂O solution and diluted colloidal silica, with 10 min of continuous stirring at 400 rpm between each addition. The molar composition of this precursor gel was xAl/90Si/yKOH/27DAH/3600H₂O with x = 2, 1.3, or 0.9 and y = 15, 15, or 12 to obtain a Si/Al ratio of 45, 70, or 100, respectively. After all the solutions were added, the pH was verified to ensure this was between 12 and 13. The resulting mixture was then aged for 3.5 h under continuous stirring at 200 rpm, after which seeds (c-ZSM-22) were added to the precursor gel. The amount of seeds was 5 wt.% with respect to the Si content in the gel. The suspension was then stirred for another 30 min to achieve complete homogenization. Crystallization was carried out in 100 mL Teflon-lined stainless-steel autoclaves under continuous stirring at 100 rpm using a Nabertherm oven and a MIXcontrol 20 stirring plate. ZSM-22 (45) was crystallized for 48 h at 160 °C, while ZSM-22 (70) and ZSM-22 (100) were crystallized for 36 h at 150 °C. After crystallization, the zeolites were recovered, washed three times with distilled water, dried at 100 °C for 20 h, and calcined under air (Nabertherm muffle furnace) at 550 °C (heating ramp 1 °C·min⁻¹) for 8 h to remove the structure-directing agent. The K-ZSM-22 zeolites were then three times ion-exchanged using 1 M NH₄NO₃ (10 mL/g zeolite) for 2 h at 50 °C under continuous stirring (200 rpm). NH₄-ZSM-22 was then recovered via centrifugation and washed until the pH of the washing water was neutral. Finally, the zeolites were dried at 100 °C for 20 h and calcined under air at 550 °C (heating ramp of 1 °C·min⁻¹ from 100 °C) for 8 h to obtain H-ZSM-22.

3.3. Catalyst Characterization

The crystal structure of the zeolites was identified by X-ray powder diffraction (XRD) using a Bruker D8 Advance diffractometer (Bruker, Billerica, MA, USA) with Cu Kα radiation (λ = 1.54 Å). The XRD patterns were collected in a 2θ range of 5–90° with a scan rate of 1°/min and steps of 0.02°. The Scherrer equation was used to estimate the average crystal size of the zeolite particles

$$\mathrm{d}={\displaystyle \frac{\mathrm{K}\mathsf{\lambda }}{\mathsf{\beta }\mathrm{c}\mathrm{o}\mathrm{s}\mathsf{\theta }}}$$

(1)

where d is the average crystal size in nanometers, K is a dimensionless shape factor with a value of 1, λ is the wavelength of the X-rays (0.154 nm), β is the full-width at half maximum, and θ is Bragg’s angle. In this study, 2θ = 35.6° was chosen, which relates to the [002] crystallographic plane.

Scanning electron microscopy (SEM) with a JSM-IT800 (JEOL, Tokyo, Japan) was performed to gain insight into the crystal morphology and size of the zeolites.

The Si/Al ratio was determined for all samples through inductively coupled plasma mass spectrometry. Before analysis, the sample was added to a Teflon liner with a mixture of fluoric acid and aqua regia, which was then heated at 110 °C for one hour to digest the sample completely. After cooling, the fluoric acid was neutralized with a solution of boric acid. The prepared solution was then analyzed using a 7900 ICP-MS from Agilent Technologies (Waltham, MA, USA).

Solid-state magic angle spinning nuclear magnetic resonance spectroscopy (MAS-NMR) was used to determine the local ²⁷Al and ²⁹Si environments in the zeolites. All MAS NMR experiments were performed on a Bruker AVANCE 500 NB, using 4 mm rotors and a rotational speed of 12 kHz. ²⁷Al MAS NMR was recorded at 130.30 MHz, using Al(NO₃)₃ as a reference for the chemical shift. ²⁹Si MAS NMR was performed at 99.35 MHz, using tetramethyl silane (TMS) as a reference for the chemical shift. The framework Si/Al ratio can be calculated by using the following equation

$$\mathrm{S}\mathrm{i}/\mathrm{A}\mathrm{l}={\displaystyle \frac{\sum _{\mathrm{n}=0}^4{\mathrm{I}}_{\mathrm{S}\mathrm{i}\left(\mathrm{n}\mathrm{A}\mathrm{l}\right)}}{\sum _{\mathrm{n}=0}^4\frac{\mathrm{n}}{4}\times {\mathrm{I}}_{\mathrm{S}\mathrm{i}\left(\mathrm{n}\mathrm{A}\mathrm{l}\right)}}}$$

(2)

where n is the number of adjacent Al nuclei, and I is the peak area of the corresponding spectrum signal in the ²⁹Si MAS NMR spectrum.

Argon sorption experiments were performed at −186 °C using a Micromeritics 3FLEX instrument to determine the zeolites’ specific surface area and pore volume. Prior to the analysis, the catalyst sample (100 mg) was degassed at 350°C overnight under vacuum to remove any adsorbed impurities. The specific surface area was determined via the Brunauer–Emmett–Teller (BET) method, applying the criteria of Rouquerol to increase the accuracy of the measurements. The micropore surface area and volume were calculated by the t-plot method, and the total pore volume (V_tot) was determined by the amount of adsorbed argon at a relative pressure (p/p°) of 0.97.

In situ Fourier transform infrared spectroscopy (FTIR) experiments were performed on a Nicolet iS60 spectrometer (Thermo Fisher Scientific, Waltham, MA, USA)with a resolution of 4 cm⁻¹. The samples were pressed under 1 ton into self-supported wafers (20 mg, 16 mm diameter) before loading to a carrousel, which is a transmission-type in situ cell designed to accommodate up to 12 self-supported sample disks. The samples were then activated at 450 °C for 2 h (temperature ramp 1 °C·min⁻¹) under a 10⁻⁶ Torr vacuum. In situ FTIR of dehydrated samples was acquired at 150 °C (reference) to reveal the acid sites and silanol groups. The quantification of the Brønsted and Lewis acid sites was performed by Py-FTIR. The probe molecule was adsorbed at 150 °C and an equilibrium pressure of 1 Torr. The spectra were subsequently recorded after evacuation at 150 °C and 300 °C. The FTIR spectra were obtained by subtracting the reference from the spectra of interest. The amount of acid sites was determined using the integrated bands situated at 1454 cm⁻¹ (Lewis) and 1545 cm⁻¹ (Brønsted), and the following molar extinction coefficients: ε(B) = 1.67 and ε(L) = 2.22 cm·µmol⁻¹ [62]. In addition, after the same pretreatment, in situ FTIR of 3-buten-1-ol was performed at room temperature. The unsaturated alcohol was adsorbed at room temperature at an equilibrium pressure of 0.5 Torr for 0.5 h, followed by desorption (10⁻³ Torr) at the same temperature to remove physiosorbed 3B1OL, whereafter the spectra were collected.

3.4. Catalytic Dehydration Experiments

The catalytic experiments were performed on a high-throughput kinetics set-up consisting of tubular reactors with a length of 0.85 m and an inner diameter of 22 mm at 5 bar. All catalysts were pelletized, sieved to 125–150 µm, and dried overnight at 120 °C to remove adsorbed water before loading them in the middle of the reactor between layers of inert α-alumina (140 µm). The reactor was then heated to the reaction temperature, i.e., 300 °C, and kept for at least 4 h under a continuous N₂ (Air Liquide) flow of 10 NLh⁻¹. The catalyst mass was varied between 50 and 60 mg and diluted with the same inert α-alumina until 20 wt.% to avoid hot spots in the catalyst bed. A liquid reactant feed of 20 wt.% 1,3-BDO in 1,4-dioxane (mass flow rate varied between 0.8 and 1.5 g/h) was mixed with N₂ (flow rate varied between 9 and 16 nL/h) and fed over the catalyst bed. All experiments were conducted at an identical site time. The reactor effluent was analyzed on-line by GC-FID (Thermo Fisher Scientific Trace 1310) equipped with a 100 m PONA column using methane as an internal standard. To ensure intrinsic kinetics were measured, engineering correlations were used [63,64]. A carbon balance exceeding 80% is obtained for all experiments, which is in accordance with reported values in the literature [26,27,28]. The reproducibility of the catalytic test was verified by conducting the dehydration of 1,3-BDO in threefold for c ZSM 22, as shown in Figure S1. Based on the consistent results, this reproducibility level was considered representative of the other zeolite catalysts. Post-reaction characterization could not be applied due to the dilution of the catalyst with inert α-alumina, which prevents physical separation and meaningful analysis. Nevertheless, a visual inspection of the catalyst bed after reaction (as illustrated for c-ZSM-22 in Figure S2) reveals a distinct black coloration, indicative of coke deposition on the zeolite, which is consistent with catalyst deactivation.

Definitions

The difference in acid densities between the zeolites was compensated for by comparing the materials based on site time:

$$\mathrm{S}\mathrm{i}\mathrm{t}\mathrm{e} \mathrm{T}\mathrm{i}\mathrm{m}\mathrm{e} [\mathrm{m}\mathrm{o}\mathrm{l}{\mathrm{H}}^{+}·\mathrm{s}·{\mathrm{m}\mathrm{o}\mathrm{l}}_{1,3\text{-}\mathrm{B}\mathrm{D}\mathrm{O}}^{-1}]={\displaystyle \frac{\mathrm{W}}{\mathrm{F}}}{\mathrm{c}}_{\mathrm{a}}$$

(3)

With W the mass of the zeolite loaded into the reactor, F the molar flow rate of 1,3-BDO, and C_a the concentration of acid sites as obtained from Py-FTIR. The site times of c-ZSM-22, ZSM-22 (45), ZSM-22 (70), and ZSM-22 (100) amounted to 6.6, 6.6, 7.0, and 8.3 mol_H+·s·mol⁻¹_1,3-BDO, respectively.

The conversion (X) of 1,3-BDO and the carbon selectivities (S) towards the products (i) were expressed as

$${\mathrm{X}}_{1,3\text{-}\mathrm{B}\mathrm{D}\mathrm{O}}[\%]={\displaystyle \frac{{\mathrm{F}}_{1,3\text{-}\mathrm{B}\mathrm{D}\mathrm{O}}^0-{\mathrm{F}}_{1,3\text{-}\mathrm{B}\mathrm{D}\mathrm{O}}}{{\mathrm{F}}_{1,3\text{-}\mathrm{B}\mathrm{D}\mathrm{O}}^0}}\times 100\%$$

(4)

$${\mathrm{S}}_{\mathrm{i}}[\%]={\displaystyle \frac{{{\mathrm{c}}_{\mathrm{i}}\ast \mathrm{F}}_{\mathrm{i}}}{4\left({\mathrm{F}}_{1,3\text{-}\mathrm{B}\mathrm{D}\mathrm{O}}^0-{\mathrm{F}}_{1,3\text{-}\mathrm{B}\mathrm{D}\mathrm{O}}\right)}}\times 100\%$$

(5)

where F⁰_1,3-BDO and F_1,3-BDO are the molar flow rates at the inlet and the outlet of the reactor, respectively. F_i and c_i represent the outlet flow of product i and the carbon numbers per molecule of product i, respectively.

BD productivity (P_BD) was calculated using the following equation

$${\mathrm{P}}_{\mathrm{B}\mathrm{D}}[{\mathrm{g}}_{\mathrm{B}\mathrm{D}}·{\mathrm{g}}_{\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{y}\mathrm{s}\mathrm{t}}^{-1}·{\mathrm{h}}^{-1}]={\displaystyle \frac{54.09\times {\mathrm{X}}_{1,3\text{-}\mathrm{B}\mathrm{D}\mathrm{O}}\times {\mathrm{S}}_{\mathrm{B}\mathrm{D}}\times {\mathrm{F}}_{1,3\text{-}\mathrm{B}\mathrm{D}\mathrm{O}}^0}{90.12\times \mathrm{W}\times 10000}}$$

(6)

where 54.09 and 90.12 are the molecular weights [g·mol⁻¹] of BD and 1,3-BDO, respectively, X_1,3-BDO is the conversion of 1,3-BDO, S_BD is the selectivity towards BD, F_1,3-BDO the mass flow of 1,3-BDO [g·h⁻¹], and W the mass of the catalyst [g].

4. Conclusions

ZSM-22 zeolites with different Si/Al ratios were prepared via a hydrothermal synthesis method, evaluated for the sustainable production of BD by 1,3-BDO dehydration at 300 °C, and compared to a commercial ZSM-22 sample. All synthesized zeolites were characterized by XRD, SEM, Ar sorption, ICP-MS, ²⁷Al, and ²⁹Si MAS NMR, and in situ FTIR with pyridine or 3B1OL as probe molecules. The zeolites displayed the characteristic ZSM-22 topology without impurities and exhibited distinct morphological and textural properties, including bigger particle sizes and higher specific surface areas compared to c-ZSM-22.

The commercial zeolite and both ZSM-22 (38) and ZSM-22 (80) showed a similar trend in 1,3-BDO conversion and product selectivity, i.e., full conversion at early TOS, followed by a decline to approximately 60% conversion, with BD and propylene selectivities around 60% and 10%, respectively, after 22 h TOS. The absence of silanol nests and strong Lewis acid sites in c-ZSM-22 and ZSM-22 (38) favored the dehydration, achieving initial BD productivities of 2.5 and 2.7 g_BD·g⁻¹_cata·h⁻¹, respectively. After 22 h TOS, the productivities of c-ZSM-22 and ZSM-22 (38) decreased to 1.3 g_BD·g⁻¹_cata·h⁻¹ and 1.2 g_BD·g⁻¹_cata·h⁻¹, respectively, outperforming previously reported zeolite catalysts in the literature. Both c-ZSM-22 and ZSM-22 (38) can be seen as potential catalysts for the sustainable production of BD.

In contrast, ZSM-22 (50) exhibited a conversion of 31% after 22 h TOS, with a lower BD and higher propylene selectivity of 55% and 14%, respectively. This behavior is attributed to the presence of strong Lewis acid sites in this zeolite, which favor the cracking of 3B1OL into propylene. Additionally, the silanol nests within the pore channels of ZSM-22 (50) interact with 3B1OL, forming hydrogen bonds that result in a liquid-like structure inside the pores. These silanol nests may reduce the activity of 3B1OL during dehydration and act as traps for coke precursors, leading to decreased 1,3-BDO conversion and BD selectivity.

Future work could focus on optimizing ZSM-22 (38) by reducing the particle size to introduce shorter diffusion path lengths, potentially enhancing the catalyst stability and ensuring higher 1,3-BDO conversion and productivity rates.