Role of Spatial Constraints of Brønsted Acid Sites for Adsorption and

Article pubs.acs.org/JACS

Role of Spatial Constraints of Brønsted Acid Sites for Adsorption and Surface Reactions of Linear Pentenes Stefan Schallmoser, Gary L. Haller, Maricruz Sanchez-Sanchez,* and Johannes A. Lercher* Department of Chemistry and Catalysis Research Center, Technische Universität München, Lichtenbergstraße 4, D-85747 Garching, Germany S Supporting Information *

ABSTRACT: The Brønsted acid sites of H-ZSM-5 and ferrierite reversibly adsborb linear pentenes via hydrogen bonding, rapidly isomerizing the double bond. On H-ZSM-5, dimerization of adsorbed pentenes occurs at a slower rate and leads to pentyl ester covalently bound to the surface. Pentene adsorbed on zeolites with narrower pores, such as ferrierite, remained stable in a hydrogen-bonded state even up to 423 K. Comparing the differential heat of adsorption of 2-pentene on silicalite and ferrierite allowed for the determination of the enthalpy difference between physically adsorbed pentene in ZSM-5 and the localized hydrogen-bonded π-complex at Brønsted acid sites, −36 kJ/mol. The activation energy (35 kJ/mol) for dimerization is almost identical to this enthalpy difference, suggesting that the rate-determining step is associated either with the mobilization of π-bonded 2-pentene or with the equally large activation barrier to form an alkoxy group via a carbenium-ion transition state. In a closed system, the dimerization rate is first order in the concentration of the π-complex that is both in equilibrium with the mobile pentene phase and in production of the carbenium ion that reacts with the mobile pentene. Overall, the alkoxy group is −41 ± 7 kJ/mol more stable than physisorbed pentene, establishing a series of energetically well-separated groups of reactive surface species.

1. INTRODUCTION Solid acids such as zeolites are widely applied in the chemical industry for the conversion of hydrocarbons in reactions such as catalytic cracking, hydrocracking, and alkylation.1−6 These reactions involve alkanes and alkenes as reactants and products, which interact with the zeolite and its Brønsted acid sites (BAS).7,8 The adsorption of alkanes on various zeolites has been the subject of numerous studies, whereas comparatively little is known about the adsorption of alkenes.9,10 Because alkenes undergo isomerization and oligomerization even at low temperatures, reliable data are difficult to obtain from mere adsorption−desorption experiments.7 It is commonly accepted that the interaction of an alkene with the BAS of a zeolite proceeds via two different states (Figure 1).7,11−13 First, a π-complex is formed, which is

frequently referred to as “physisorbed alkene” because, despite the clearly localized interaction (hydrogen bonding) between the π-electrons of the double bond and the BAS, covalent bonds are neither formed nor broken.7,14 Subsequently, the πbonded pentene may be protonated, leading to the formation of a chemisorbed species.7,13 The nature of the resulting chemisorbed species is still debated. It has been proposed that the chemisorbed species is stabilized as covalently bonded alkoxide or as an ion pair involving a free carbenium ion (Figure 1).11,13−16 To specify the possible adsorption states, we strictly use the following expressions: “physisorbed” signifies nondirected adsorption in zeolite pores by dispersion forces; “π-complex” represents a hydrogen bond between the double bond of the alkene and the BAS;12 and “chemisorption” encompasses two different adsorption states, the covalently bonded alkoxide and the ion pair (carbenium ion). With the exception of a series of NMR and spectroscopic experiments, the adsorption and especially the chemisorption of alkenes in zeolites has mainly been addressed theoretically.7,13,14,17,18 In these studies, calculated heats of adsorption (chemisorption and π-complex formation) depend largely on the zeolite model and the theoretical methods employed to describe the long-range, noncovalent interactions. Nieminen et

Figure 1. Illustration of adsorbed species for 2-pentene on a zeolite.

Received: April 12, 2017 Published: June 6, 2017

© 2017 American Chemical Society

8646

DOI: 10.1021/jacs.7b03690 J. Am. Chem. Soc. 2017, 139, 8646−8652

Article

Journal of the American Chemical Society al. report that for H-FER the heat of π-complex formation of alkenes is comparable to the heat of adsorption of alkanes.14 The specific interaction of the double bond with the BAS was presumed to be compensated by the overall reduced nonspecific dispersion interaction (RCH2−CH2R′ has two more C−H fragments than RHCCHR′).14 Nguyen et al. reported for π-complex formation of, for example, 1-pentene on H-ZSM5, a value of −84 kJ/mol (1-pentene) and for chemisorption, a value of −156 kJ/mol (2-pentoxy) using the quantum mechanics−interatomic potential (QM−Pot) hybrid method.7 Domen et al. experimentally studied the adsorption and reaction of butenes on H-ZSM-5 and mordenite.11,12,16,19,20 For H-ZSM-5 at subambient temperatures, a stable π-complex was formed with double-bond isomerization occurring already at 230 K.12,19,21 Isotope-labeling experiments suggested a concerted mechanism for the double-bond isomerization in the absence of a carbenium ion at these temperatures.21−23 Stepanov et al. determined the activation energy for the doublebond shift in linear butenes to be 41 kJ/mol on FER and showed that carbenium ions are involved at 290 K.22,24 Interestingly, Stepanov et al. did not detect a stable alkoxide species in ferrierite, which would be easily observable in 13C MAS NMR.22 Thus, two mechanisms seem to exist for doublebond migration in acidic zeolites, depending on the temperature: a concerted mechanism for low temperatures (T < 230 K) and the classical carbenium-ion mechanism at higher temperatures.19,21,22 Both pathways require the presence of an acidic proton.19,21,22 Isotope experiments demonstrate, in addition, that the high mobility of alkenes exists even at subambient temperatures.19,21 Hence, the π-bonded alkenes cannot be considered to be static but are moving between BAS with a low barrier, on a path stabilized by the zeolite pore walls.12 At room temperature, the π-complex was not stable in H-ZSM-5,14 and linear butenes were gradually transformed to a dimer, which was claimed to be mainly physisorbed via its aliphatic backbone.12 Such a dimerized alkoxy species was also observed for the analogous reaction of 2-methyl-propene on mordenite, when the π-bonded butene was thermally converted into a chemisorbed species upon heating to 207 K.11 Thus, whereas an overall understanding of the elementary processes of alkene adsorption on zeolites is emerging, the quantitative barriers, and, hence, the basis for including this chemistry in the kinetic evaluations of zeolite properties, are lacking. This is largely the result of the substantial reactivity of the adsorptive species under typical experimental conditions. Therefore, we have chosen a new approach to explore the adsorption of pentene on H-ZSM-5 qualitatively and quantitatively, combining in situ IR spectroscopy, gravimetry, and calorimetry to determine not only the nature of the states during adsorption but also the enthalpy of these states considering reactive transformations such as isomerization and dimerization. Aluminum-free ZSM-5 (silicalite) and ferrierite are used to explore specific adsorption modes.

Figure 2. Time course of the surface reaction on H-ZSM-5 upon pulsing of 1-pentene followed by IR spectroscopy (T = 323 K). (a) Difference spectra for different times after pulsing of 1-pentene. The arrow points to the H−CC vibration of 1-pentene (3020 cm−1) that disappears on formation of 2-pentene. (b) Change of concentration of interacting BAS during outgassing (shown for three subsequent pulses; first pulse −13% of all BAS initially covered, second pulse −31%, third pulse −68%) plotted as a function of time. BAS0 signifies the number of (additional) initially covered BAS before starting outgassing.

spectra, characteristic bands are compiled in Section 2.2 of the Supporting Information. It is concluded that linear pentenes rapidly form a π-bonded complex with the protons of H-ZSM5. This is deduced from the initial (t = 0.5 min) decrease in the intensity of the undisturbed O−H vibration of BAS (3610 cm−1), the presence of the characteristic H−CC stretching vibration (3020 cm−1, indicative of the presence of 2pentene),25 the presence of CC stretching vibration (1654 cm−1, not shown), as well as the broad band of the OH groups at 3100 cm−1, which is characteristic of hydrogen-bonding. Because the spectrum of adsorbed pentene showed all bands characteristic of 2-pentene (see Supporting Information for details), the double-bond isomerization is concluded to be equilibrated at this stage. The alkene adsorbed via hydrogen bonding to the double bond reacted further during the next 30 min. The broad peak characteristic of OH groups, as found in hydrogen-bonding pentene, disappeared with time. In parallel, the band of undisturbed BAS was partially restored. The concentration of adsorbed alkenes, as judged by the intensity of C−H vibrations, did not decrease with time (see Figure S.5). This allowed the ruling out of desorption of a fraction of adsorbed (π-complex or physisorbed) pentene during this process. Once the system was equilibrated, approximately 50% of BAS involved in hydrogen bonding had reappeared (Figure 2a). Combining the disappearance of pentene from 50% of the initially interacting BAS and the slight increase in the intensity of CH bands (3000−2700 cm−1; no desorption) led us to conclude that pentene dimerized in this process. The modest increase in the integral intensity of the CH-stretching vibrations compared to the initial adsorption (Figure S.5) is attributed to a slightly higher extinction coefficient for the carbonnormalized branched dimers. Additionally, a minor broad band was observed at 3450 cm−1. Kondo et al. attributed it to a shifted OH band caused by the interaction of BAS with the aliphatic part of an alkene.21 Because the extinction coefficient of a shifted BAS increases upon polarization by an alkane, it is concluded that only a minority of molecules interact in that way.26 The absence of a further broad band characteristic for hydrogen bonding (despite the intense negative peak at 3610 cm−1 in the difference spectra) implies that a fraction of the BAS reacted

2. RESULTS AND DISCUSSION 2.1. Surface Reaction of 1-Pentene at 323 K. In contrast to alkanes, olefins undergo transformations induced by the acid sites of zeolites even at room temperature. The reactivity of 1pentene on H-ZSM-5 was visible in the IR spectra of adsorbed pentene. Figure 2a shows the time-resolved surface reaction at 323 K. Small amounts of 1-pentene were dosed, and the IR spectra were recorded with a minimum time delay at the equilibrated pressure. To facilitate the discussion of the IR 8647

DOI: 10.1021/jacs.7b03690 J. Am. Chem. Soc. 2017, 139, 8646−8652

Article

Journal of the American Chemical Society

Figure 3. Illustration of a possible dimerization pathway of 2-pentene on BAS of a zeolite. Reaction of the π-complex to form a chemisorbed specieshere depicted as a carbenium ionis proposed as the rate-determining step, and the reaction of the carbenium ion with a physisorbed, mobile pentene is fast and irreversible.

with pentene. We hypothesize at this point that this chemical reaction is the formation of a chemisorbed alkene species, specifically the dimer of 2-pentene. As mentioned above, the recovery of 50% of the band at 3610 cm−1 is attributed to dimerization. Figure 2b shows the time course of BAS recovery (based on the intensity of the band at 3610 cm−1) for a sequence of three pulses. If the concentration of BAS recovered is normalized to the initial concentration of BAS interacting with pentene (initial negative band), then it can be determined that the reaction followed the same kinetics in all three pulses, independent of the different initial surface coverages by pentene, independent of the different concentration of BAS not covered by hydrocarbons, and independent of the additional presence of dimers. An exponential rate law describes well the kinetics of BAS recovery (see 2.3.1 in Supporting Information for details). The low concentration of molecules in the pores limits the reaction to dimerization as well as the fact that neither the adsorbed product of 2-pentene dimerization (decene) nor the monomer desorbs under the experimental conditions applied. The above observations are in excellent agreement with those of Kondo et al.12 with regard to adsorption and reaction of 1-butene at room temperature on H-ZSM-5 (Si/Al = 50). Namely, the isomerization between 1- and 2-alkene is equilibrated upon adsorption forming a 1:1 π-complex with BAS, and the π-complex dimerized but did not further oligomerize or polymerize. Thus, let us turn to the quantitative aspects of the adsorption. Figure 3 illustrates the steps involved in the dimerization of pentene, showing one of the many possible isomers that can result as product of this surface reaction. Pentene adsorbed via its π-bond is equilibrated with physically adsorbed pentene. At the temperatures studied, we conclude that the equilibrium lies predominantly on the side of the adsorbed π-complex. Reaction starts from a π-bound alkene with the formation of the chemisorbed intermediate. It is hypothesized that once the monomeric chemisorbed intermediatea carbenium ion or an alkoxy groupis formed, it readily reacts with mobile physisorbed pentene that has desorbed from a π-complex with BAS (Figure 3). Alternatively, it may be that a reactive encounter between a mobile, physisorbed pentene and a πbonded pentene initiates the reaction to form a carbenium ion that reacts further. Overall, this step results in a chemisorbed dimer and one free BAS. Theory estimates that the alkoxide (2-alkene to 3-alkoxide in the range of C4−C8, more or less independent of carbon number) is more stable than the π-alkene by 55−57 kJ/mol.7

Thus, we believe that the main reason for the dimer not to further oligomerize is the greater stability (against migration or activated complex formation or both) of the dimer alkoxide than the monomer π-complex. The liberation of the BAS in the dimerization step can be followed by IR spectroscopy (Figure 2a), and it has been used for experimentally deriving the rate parameters. The fraction of the liberated BAS, Θ*, is defined as the ratio of liberated BAS and initially (t = 0) covered BAS0. Note that Θ* is not the total number of vacant BAS, but just that fraction of the initial πbonded 2-pentene adsorbed from the pulse that becomes vacant as a result of dimerization. Equation 1 is derived by using the reaction pathway summarized in Figure 3 (see Supporting Information section 2.3). 1 − 2Θ* = e−2kt

(1)

Figure 4 shows the results of this fit for 323 K. The value of energy of activation (35 kJ/mol) is close to the experimentally

Figure 4. Kinetics of surface reaction at 323 K fitted by using eq 1 (left) and the corresponding Arrhenius plot (right). Symbols signify experimentally determined values, and lines show results of fitting. The results for 304, 313, and 333 K are shown in Figure S.6.

determined values for the H/D exchange mediated by butenes on zeolite ZSM-5 (31 ± 8 kJ/mol).11,24 The rate constant for the H/D exchange is also comparable (if extrapolated to 323 K)11 to the values determined by this fitting method (k = 0.335 min−1 for T = 323 K). As the H/D exchange proceeds via a carbenium-ion transition state,11 the comparable rates and energies of activation support the proposed reaction pathway of Figure 3, leading us to hypothesize that the rate-determining step in pentene dimerization involves the formation of a C5 carbenium ion. Given the fast dimerization of pentene at low temperatures on ZSM-5, it is not possible to define quantitatively the process 8648

DOI: 10.1021/jacs.7b03690 J. Am. Chem. Soc. 2017, 139, 8646−8652

Article

Journal of the American Chemical Society of adsorption of pentene in a direct way. It is, therefore, necessary to resort to measurements of pentene adsorption on reference materials with less reactivity than ZSM-5, in order to obtain a complete energetic picture of the elementary steps involved in pentene adsorption and dimerization. In the next part, we study the adsorption of linear 1- and 2-pentene on nonacidic silicalite and ferrierite that have an acidity similar to ZSM-5, but also have a very low reactivity for dimerization, with the aims of deducing the energy of physisorption and the π-complex formation from physisorbed pentene, respectively. These values will be used to estimate the enthalpy of alkoxide chemisorption for ZSM-5. 2.2. Estimation of the Enthalpic Contribution from the Formation of the π-Complex. 2.2.1. Adsorption Studies with Silicalite. First, we studied the adsorption of pentene on silicalite, a zeolite isostructural with HZSM-5, but without Brønsted acid sites. As mentioned above, double-bond isomerization occurs faster on ZSM-5 than dimerization and, within 1 min after dosing 1-pentene, 2-pentene was already the dominant species on the zeolite surface (Figure 2). IR spectroscopy allowed differentiation between pentene doublebond isomers by the characteristic H−CC stretching vibration25 (see section 2.2 of Supporting Information). On silicalite, pentene adsorption was completely reversible and double-bond isomerization was not observed. Thus, we conclude that BAS are indispensable for double-bond isomerization. Because 2-pentene was the dominant isomer adsorbed on ZSM-5, its differential heat of adsorption was measured as a function of coverage and a value of −56 kJ/mol was obtained (see Figure 5). This energy corresponds to the enthalpy associated with the interaction of the olefin and the zeolite pore walls not involving BAS.

Figure 6. Difference spectra following adsorption of 1-pentene on HFER-25 at 323 K. Spectra after 1, 2, 3, and 9 min of introducing 0.05 mbar of 1-pentene. Arrows indicate evolution of bands with time.

suggesting that both zeolites have the same acid strength.28 Indications for the formation of an alkoxide or dimerization were not observed. Upon evacuation, the intensity of the BAS O−H band was restored whereas the C−H stretching bands of pentene decreased (see Figure S.7 in Supporting Information). All bands characteristic of π-bonded pentene were stable (for 2 h), and the interaction was also reversible at 423 K (Figure S.8). Hence, only double-bond isomerization occurred on FER at these conditions. The difference in reactivity between H-ZSM-5 and H-FER is attributed to the smaller space at BAS in the FER pores (for comparison, see Table S2), with dmax located in the 8-ring cage for FER (see illustration in Figure S.9), whereas for MFI it is located in the highly accessible intersection of the 10ring pore channels.29,30 Formation of the alkoxide or carbenium ion from the π-complex involves rehybridization of at least one sp2 to an sp3 carbon atom. Therefore, the transition state from the planar (along CC bond) pentene molecule to the carbenium ion/alkoxide will involve a transition-state complex, which occupies significant space in the zeolite confinement.7,13 In MFI, the larger channel size (0.56 × 0.53 nmstraight channel) seems to enable this transformation, whereas in FER, the tighter channel nature (0.54 × 0.42 nm) is hypothesized to be hindering the dimerization at this low temperature. The absence of surface reactions allowed, therefore, the direct measurement of differential heats of adsorption for 2pentene (Figure 7) on H-FER. Three regions can be discerned. For coverages of up to 0.2 mmol/g, a value of −92 kJ/mol was found, decreasing to −62 kJ/mol in the second region, and further decreasing for coverages beyond 0.6 mmol/g, the latter loading corresponding to a volume of 2-pentene which is identical to the micropore volume in agreement with earlier measurements.31 The deconvolution of the O−H region in H-FER during pentene adsorption showed how 2-pentene adsorbs preferentially on BAS located on the more accessible 8- and 10-ring channels (see Supporting Information 2.4), similarly to what has been previously described for n-pentane.32,33 The step in the differential heat of adsorption plot (Figure 7A) coincides with the pressure in which BAS located in 10-ring and 8-ring channels were saturated (Figure 7B, 3609 and 3587 cm−1 respectively). Hence, the value of −92 kJ/mol is assigned to interaction of 2-pentene with those easily accessible BAS, whereas −62 kJ/mol is attributed to sorption of 2-pentene with less accessible BAS in 8-ring cages. Assuming transferability between zeolite structures, the value of pentene’s heat of

Figure 5. Differential heat of adsorption determined at 323 K (2pentene on silicalite).

2.2.2. Adsorption of Linear Pentenes on H-FER. In contrast to H-ZSM-5, the π-complex formed when alkenes are adsorbed on H-FER is stable.27 Thus, the adsorption of olefins on H-FER provides information on the selective contribution of the hydrogen bonding with pentene. Also for H-FER, mainly 2pentenes were observed by IR spectroscopy after 1 min of equilibration (Figure 6), as deduced from the shift of the H− CC stretching vibration to 3020 cm−1 and the band of the CC stretching vibration at 1654 cm−1 (see section 2.4 Supporting Information for details). A negative peak is observed in the difference spectrum at 3600 cm−1 together with a broad band at 3100 cm−1. This shift of 500 cm−1 of the OH-stretching vibration upon hydrogen bonding of 2-pentene is in good agreement with the shift observed with H-ZSM-5, 8649

DOI: 10.1021/jacs.7b03690 J. Am. Chem. Soc. 2017, 139, 8646−8652

Article

Journal of the American Chemical Society

C1−C10 alkanes on silicalite report the heat of adsorption of pentane to be −60 kJ/mol on silicalite, in agreement with the rule of thumb of −12 kJ/mol per carbon on H-ZSM-5.34 Thus, pentane (−60 kJ/mol) and pentene (−56 kJ/mol) have comparable heats of adsorption in the physisorbed state in the absence of BAS. 2.3. Estimation of the Energies Involved in Chemisorption. At this point all of the relevant information is available to derive estimates of the enthalpies for those adsorption states of 2-pentene in H-ZSM-5 that are experimentally not directly accessible. IR experiments showed that the chemisorbed dimer is the dominant surface species, and hence the heat released upon equilibration with linear pentenes corresponds to the formation of a chemisorbed dimer. This heat of dimerization, which includes the heat of chemisorption of the dimer, was measured by calorimetry to be −285 ± 7 kJ/mol. Enthalpic contributions from double-bond isomerization can be neglected because 2-pentene was used in these experiments. Figure 8 shows the steps that need to be considered to deduce the energy levels in this dimerization pathway. For the analysis, we start first from the gas phase without considering the zeolite confinement. The gas phase reaction of C−C bond formation between two pentene molecules was estimated using HSC Chemistry 6.0 software.35 The dimerization reaction of two 2-pentene molecules will result, among other things, in the formation of 4,5-dimethyl-3-octene. As thermodynamic data were not available for these branched olefinic C10 isomers, the analogous reaction with two butene molecules was used in order to estimate the reaction enthalpy for this dimerization reaction (e.g., ΔHr = −88 kJ/mol (2 × 2butene → 3,4-dimethyl-2-hexene)). 3,4-Dimethyl-2-hexene is, in fact, the dimer product of 1-butene identified by IR, on HZSM-5 (Si/Al = 50) at room temperature.12 Like 1-pentene, the isomerization of 1-butene is too rapid (diffusion controlled) to measure, but dimerization occurs in about 15 min and half the initially perturbed OH (3610 cm−1) is recovered, quite parallel to what is shown in Figure 2a. The contribution of the physisorption in the zeolite pores by dispersion forces is estimated for an alkane (and about the same is expected for an alkene as discussed above) to be ca. −12 kJ/

Figure 7. (A) Differential heat of adsorption determined at 323 K (2pentene on H-FER). Lines indicate zones of different adsorption behavior. (Data for three independent experiments are shown.) (B) Results of deconvolution of OH-stretching region of H-FER using four Gaussian functions centered at 3609, 3600, 3587, and 3563 cm−1 shown as functions of increasing pentene coverage (represented by the area of CC stretching vibration), measured at T = 323 K, 0−5 mbar of 1-pentene (see Supporting Information 2.4 for details).

adsorption of −92 kJ/mol, obtained for BAS in 10-ring pores in FER, is used to specify the adsorption in the 10-ring pores of H-ZSM-5. The case of pentane shows that the analogy is justified, because the heats of adsorption of −69 kJ/mol for HFER31 and −70 kJ/mol for H-ZSM-534 were identical. We note in passing that the double bond of 2-pentene may impose some steric constraints for an optimized interaction with BAS in the somewhat narrower H-FER. However, the similar activity for double-bond isomerization of 2-pentenes on H-FER and HZSM-527 suggests that such an influence is marginal. Thus, the heat of adsorption on H-FER, −92 kJ/mol, closely approximates the heat of adsorption on H-ZSM-5. Finally, the value obtained from the reference experiment in silicalite allows the calculation of the net enthalpic contribution of the interaction between BAS in MFI and the CC bond of the olefin. The heat of adsorption of pentene on silicate can be used to estimate the physisorption enthalpy of 2-pentene on MFI, −56 kJ/mol (Figure 5). Thus, with −92 kJ/mol estimated to be the heat of formation of the hydrogen-bonded pentene, the difference (−36 kJ/mol) represents the specific contribution of hydrogen bonding at the BAS. Independent studies of

Figure 8. Schematic illustration of the enthalpic dissection of the dimerization reaction of 2-pentene between the gas phase and confinement within a zeolite ZSM-5. 8650

DOI: 10.1021/jacs.7b03690 J. Am. Chem. Soc. 2017, 139, 8646−8652

Article

Journal of the American Chemical Society

Figure 9. Overall energy diagram for the surface reactions of 2-pentene and the associated energy levels and barriers.

bond isomerization takes place on FER via a π-complex that converts to a carbenium ion for temperatures above 230 K.22,24 The activation energy for this reaction in ferrierite was 41 kJ/ mol. Given that the BAS in ferrierite is somewhat more confined in the 10-ring pore than in ZSM-5 (see Table S.2), that might slightly increase the barrier to carbenium ion formation in ferrierite relative to ZSM-5. Thus, it is reasonable to expect that the barrier to carbenium-ion formation in ZSM-5 may be equal to 35 kJ/mol. Both mobile 2-pentene and the carbenium ion derived from π-bonded 2-pentene are required for dimerization, but the mechanism presented requires that the relative concentrations are π-bonded 2-pentene ≫ mobile pentene > carbenium-ion intermediate. Indeed, only the π-bonded 2-pentene can be observed spectroscopically. Even though the formation of the carbenium ion and of the mobile pentene from π-bonded 2pentene have comparable energy barriers, the mobile-pentene phase would result in an increase in entropy, whereas the rehybridization from sp2 to sp3 in the formation of the carbenium-ion intermediate must result in a decrease in entropy. From the kinetic analysis in section 2.1(Figure 4), the entropic barrier of the dimerization reaction of pentene in ZSM-5 can be calculated by application of the Eyring−Polanyi equation. It was found that the entropy loss is significant, −186 J/mol·K, which supports the hypothesis that the ratedetermining step is the formation of a carbenium-ion intermediate.

mol per C atom.10 The contribution from the π-complex has been estimated above to be −36 kJ/mol for 2-pentene on BAS of ZSM-5, and this value was used to account for the direct interaction of the CC bond of a C10 dimer with the proton as well. This leads to an estimated value of −244 kJ/mol for formation of the π-bonded C10 dimer in H-ZSM-5. The enthalpic contribution from formation of the chemisorbed dimer is hypothesized to cause the difference between −244 kJ/ mol and the experimentally determined value of the adsorption and dimerization process of −285 ± 7 kJ/mol. This energy difference suggests a value of about −41 kJ/mol, with an error bar of ±7 kJ/mol, for the conversion of the π-complex of the dimer into an alkoxide, as shown in Figure 8. This information allows us to complete the enthalpic picture of the adsorption of linear pentenes in the 10-ring pores of HZSM-5 type zeolites (Figure 9). Both theory and experiment suggest that the heat of adsorption of 2-pentene measured for H-FER, −92 kJ/mol, can be transferred to H-ZSM-5. The overall heat of chemisorption for a pentene monomer on MFI is deduced to be −133 kJ/mol (i.e., the sum of the heat of adsorption of the 2-pentene π-complex (−92 kJ/mol) and its conversion to the chemisorbed state −41 kJ/mol, see Figure 8). This is slightly lower than the value calculated by Nguyen et al.7 for 2-pentene → 3-pentoxy for the overall adsorption from the gas phase to alkoxide, −147 kJ/mol. Thus, this would point to an alkoxide as the chemisorbed species for the pentene dimer. Surprisingly, the monomer alkoxide has not been detected in the experiments in this work nor in the preceding experiments of Domen et al.12 The reason for this surface chemistry is unclear at present. Finally, we have determined by analysis of the kinetics of pentene dimerization that the dimerization is an activated process with an energy barrier of 35 kJ/mol (Figure 4). As already noted above, this is about the same as the 36 kJ/mol difference between the physisorbed 2-pentene and the πcomplex, so a possible rate-determining step of dimerization is the formation of the mobile physisorbed 2-pentene. If we hypothesize that the conversion of the π-complex to the carbenium ion is rate determining, this barrier would then be equal to 35 kJ/mol. It has been reported that butene double-

3. CONCLUSIONS Combining IR spectroscopy and calorimetry provided a full description of the elementary steps of the adsorption of a light olefin in H-ZSM-5, exemplified here by pentene. Overall, three states of the olefin have been identified: (i) olefin adsorbed in the pores via dispersion forces, (ii) interactions of the CC bond of pentene with the BAS in the pores (π-complex), and (iii) an alkoxy group in the form of a dimer. The principal nature of these elementary steps has been quantitatively evaluated, combining several measurements and using a MFI sample without BAS to establish the reference value for adsorption via dispersion forces (i.e., −56 kJ/mol). 8651

DOI: 10.1021/jacs.7b03690 J. Am. Chem. Soc. 2017, 139, 8646−8652

Journal of the American Chemical Society

■

Measuring the heat of adsorption of hydrogen-bonded 2pentene on H-FER (−92 kJ/mol) allowed the deduction of the specific interaction between the BAS and the pentene double bond of 2-pentene (π-complex) (i.e., −36 kJ/mol). As we hypothesize that the specific energies of interactions are transferable between zeolites of identical acid strength and comparable pore size, we conclude that the overall heat of hydrogen bonding of 2-pentene in H-ZSM-5 is −92 kJ/mol. The energies involved in the transformation of the interaction of the π-bonded pentene to a surface alkoxy dimer were deduced from calorimetric measurements of the adsorption of pentene on H-ZSM-5 leading to a chemisorbed dimer. The data indicated a further stabilization of the alkoxy group compared to the π-bonded state by −41 kJ/mol. The activation energy to reach this (dimerized) state is 35 kJ/mol. This is very similar to the energy of stabilization via π-bonding with respect to the physisorbed pentene (−36 kJ/mol). The closeness of these two energy steps does not allow a final conclusion about the nature of the rate-determining step for the formation of the dimerized alkoxy group. However, the large loss of entropy calculated for the rate-determining step, together with the absence of dimerization in the narrower pores of H-FER, point toward the sterically demanding rehybridization of sp2 to sp3 orbitals to form a carbenium ion as the most probable barrier. The fact that the energetic stabilization by forming an alkoxy group is 41 kJ/mol, only 5 kJ/mol higher than the dimerization barrier, inevitably leads to a competition between the dimerization and the formation of a monomeric alkoxy group; i.e., eventually all alkenes form a dimer. These values also indicate that, for linear pentenes, the energy level of the sec-pentyl carbenium ion must approximate the level of the π-bonded state. These transformations indicate that alkenes undergo rapid bond formation as long as the steric environment of the acidsite transition state allows the demanding rehybridization of the π-bonded alkenes to carbenium ions. Time-resolved vibrational spectroscopy coupled with pump-probe experiments will be needed to better understand the details of the complex transformations.

■

REFERENCES

(1) Guisnet, M., Gilson, J.-P., Eds. Zeolites for Cleaner Technologies. Imperial College Press: London, 2002. (2) Corma, A. Chem. Rev. 1995, 95, 559. (3) Buchanan, J. S.; Santiesteban, J. G.; Haag, W. O. J. Catal. 1996, 158, 279. (4) Rahimi, N.; Karimzadeh, R. Appl. Catal., A 2011, 398, 1. (5) Jacobs, P. A.; Martens, J. A. Introduction to Acid Catalysis with Zeolites in Hydrocarbon Reactions. In Introduction to Zeolite Science and Practice; Flanigen, E. M., Jansen, J. C., van Bekkum, H., Eds.; Elsevier: Amsterdam, 1991; Vol. 58. (6) Vermeiren, W.; Gilson, J. P. Top. Catal. 2009, 52, 1131. (7) Nguyen, C. M.; De Moor, B. A.; Reyniers, M.-F.; Marin, G. B. J. Phys. Chem. C 2011, 115, 23831. (8) Kissin, Y. V. Catal. Rev.: Sci. Eng. 2001, 43, 85. (9) De Moor, B. A.; Reyniers, M. F.; Gobin, O. C.; Lercher, J. A.; Marin, G. B. J. Phys. Chem. C 2011, 115, 1204. (10) Eder, F.; Lercher, J. A. Zeolites 1997, 18, 75. (11) Ishikawa, H.; Yoda, E.; Kondo, J. N.; Wakabayashi, F.; Domen, K. J. Phys. Chem. B 1999, 103, 5681. (12) Kondo, J. N.; Liqun, S.; Wakabayashi, F.; Domen, K. Catal. Lett. 1997, 47, 129. (13) Bhan, A.; Joshi, Y. V.; Delgass, W. N.; Thomson, K. T. J. Phys. Chem. B 2003, 107, 10476. (14) Nieminen, V.; Sierka, M.; Murzin, D. Y.; Sauer, J. J. Catal. 2005, 231, 393. (15) Boronat, M.; Viruela, P. M.; Corma, A. J. Am. Chem. Soc. 2004, 126, 3300. (16) Kondo, J. N.; Wakabayashi, F.; Domen, K. J. Phys. Chem. B 1998, 102, 2259. (17) Pascual, P.; Ungerer, P.; Tavitian, B.; Pernot, P.; Boutin, A. Phys. Chem. Chem. Phys. 2003, 5, 3684. (18) Haw, J. F.; Richardson, B. R.; Oshiro, I. S.; Lazo, N. D.; Speed, J. A. J. Am. Chem. Soc. 1989, 111, 2052. (19) Kondo, J. N.; Domen, K.; Wakabayashi, F. Microporous Mesoporous Mater. 1998, 21, 429. (20) Yoda, E.; Kondo, J. N.; Domen, K. J. Phys. Chem. B 2005, 109, 1464. (21) Kondo, J. N.; Shao, L.; Wakabayashi, F.; Domen, K. J. Phys. Chem. B 1997, 101, 9314. (22) Stepanov, A. G.; Arzumanov, S. S.; Luzgin, M. V.; Ernst, H.; Freude, D. J. Catal. 2005, 229, 243. (23) Boronat, M.; Viruela, P.; Corma, A. J. Phys. Chem. A 1998, 102, 982. (24) Stepanov, A. G.; Luzgin, M. V.; Arzumanov, S. S.; Ernst, H.; Freude, D. J. Catal. 2002, 211, 165. (25) Armaroli, T.; Finocchio, E.; Busca, G.; Rossini, S. Vib. Spectrosc. 1999, 20, 85. (26) Makarova, M. A.; Ojo, A. F.; Karim, K.; Hunger, M.; Dwyer, J. J. Phys. Chem. 1994, 98, 3619. (27) Ivanov, P.; Papp, H. Langmuir 2000, 16, 7769. (28) Lercher, J. A.; Noller, H. J. Catal. 1982, 77, 152. (29) Zholobenko, V. L.; Lukyanov, D. B.; Dwyer, J.; Smith, W. J. J. Phys. Chem. B 1998, 102, 2715. (30) Miyaji, A.; Sakamoto, Y.; Iwase, Y.; Yashima, T.; Koide, R.; Motokura, K.; Baba, T. J. Catal. 2013, 302, 101. (31) Eder, F.; Lercher, J. A. J. Phys. Chem. B 1997, 101, 1273. (32) Pascual, P.; Boutin, A. Phys. Chem. Chem. Phys. 2004, 6, 2015. (33) van Well, W. J. M.; Cottin, X.; de Haan, J. W.; Smit, B.; Nivarthy, G.; Lercher, J. A.; van Hooff, J. H. C.; van Santen, R. A. J. Phys. Chem. B 1998, 102, 3945. (34) Eder, F.; Stockenhuber, M.; Lercher, J. A. J. Phys. Chem. B 1997, 101, 5414. (35) Ates, A.; Reitzmann, A.; Waters, G. Appl. Catal., B 2012, 119− 120, 329.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b03690. Materials and methods section, supporting results, and information(PDF)

■

Article

AUTHOR INFORMATION

Corresponding Authors

*[email protected] *[email protected] ORCID

Johannes A. Lercher: 0000-0002-2495-1404 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge BU Catalysts, Clariant Produkte (Deutschland) GmbH (former Sued-Chemie AG) for the financial support and fruitful discussions in the framework of MuniCat. 8652

DOI: 10.1021/jacs.7b03690 J. Am. Chem. Soc. 2017, 139, 8646−8652