Supporting Information: Ethene Oligomerization in Ni-containing Zeolites: Theoretical Discrimination of Reaction. Mechanisms

Transcription

1 Supporting Information: Ethene Oligomerization in Ni-containing Zeolites: Theoretical Discrimination of Reaction Mechanisms Rasmus Y. Brogaard and Unni Olsbye Department of Chemistry, Centre for Materials and Nanoscience (SMN), University of Oslo, P.O. Box 1033, Blindern, N-0315 Oslo, Norway. S1

2 Contents S1 Nickel ion sites S3 S2 Benchmark calculations: CO and ethene adsorption on metal ion sites S4 S3 Free energies from harmonic frequencies: full versus partial Hessian S6 S4 Additional potential and free energy profiles S8 S5 Atomic structures S12 S6 Additional thermochemical data S14 References S24 S2

3 S1 Nickel ion sites Figure S1A and B show the two sites that can be created with one Al/Si substitution in the super cell of the AFI framework. The site in Figure S1B is the most stable by 19 kj/mol in electronic energy, as it leads to the least amount of distortion of the framework. Hence, the DFT calculations on species presented in the main manuscript focused on the S1B site. Figure S2 shows the Ni 2+ sites considered in the calculations, having two Al atoms in the same six-membered ring. According to the Löwenstein rule, 2 two Al atoms have to be separated by at least one Si atom in the framework. This leaves four options: Figure S2A and B illustrate the most stable symmetric substitutions, whereas Figure S2C and D represent the least stable asymmetric substitutions. Figure S1: Modeled Ni sites with one Si/Al substitution in the AFI zeolite framework. A) and B) show the two different sites, whereas C) shows the B) site with a hydroxide ligand. Figure S2: Modeled Ni sites with two Si/Al substitutions in the AFI framework. S3

4 S2 Benchmark calculations: CO and ethene adsorption on metal ion sites In order to benchmark the accuracy of the computational setup described in the main manuscript, we have calculated adsorption enthalpies of CO and ethene on singly and doubly charged metal ions in the AFI framework. All zeolite structures were treated in spin-polarized calculations and several starting guesses were tried, to locate the most stable spin states. Enthalpies of the adsorbates were derived using partial Hessian vibrational analysis, displacing only the atoms of the adsorbates (Section S3). The calculated adsorption enthalpies are compared to available experimental data in Table S1. As can be seen, there is quite a large spread in the reported experimental values. The calculated values are generally within this range, with the exception of ethene adsorption on Cu +. It is worth noting that the calculated adsorption enthalpies depend quite significantly on the metal ion site in the AFI framework, as was also observed in earlier computational work on chabazite. 3 This means that the adsorption enthalpies can be expected to depend significantly on the type of zeolite framework. With that in mind, we consider the calculated values to be in satisfactory agreement with the experimentally measured ones. S4

5 Table S1: Experimentally obtained adsorption enthalpies for CO and ethene on extraframework metal ions in zeolites, compared to calculated values in the AFI framework. Metal(I) ions were modeled as in Figure S1 and metal(ii) ions as in Figure S2. The enthalpies were derived using partial Hessian vibrational analysis (PHVA, see Section S3), displacing only the atoms of the adsorbates. All values are in kj/mol. [Cu-CO] + [Cu-ethene] + [Co-ethene] 2+ [Ni-CO] 2+ [Ni-ethene] 2+ BEEF-vdW SSZ-24 Kuroda 4 ZSM-5 Bolis 5 ZSM-5 Borgard 6 Y Klier 7 A 303 K 313 K 296 K 303 K 303 K 135, 156 a 140, 163 a 72 b 68, 71, 86, 145 c 78, 81, 79, 123 d 91, 82, Halliche 8 ZSM-5: 127 US-Y: a : values for sites analogous to Figure S1A and B, respectively. Spin state: S=0. b : value for site analogous to Figure S2A. Spin state S=3/2. c : values for sites shown in Figure S2A, B, C and D, respectively. Spin states: S=1 (A, B) and S=0 (C, D). d : values for sites shown in Figure S2A, B, C and D, respectively. Spin states: S=1 (A, B, C) and S=0 (D). S5

6 S3 Free energies from harmonic frequencies: full versus partial Hessian In this work we derived free energies from partition functions treating gas phase molecules in a rigid rotor harmonic oscillator approximation and zeolite species in the harmonic approximation. In the most straightforward approach, harmonic frequencies are derived from a Hessian including all atoms in the unit cell, termed full Hessian vibrational analysis (FHVA). In contrast, a partial Hessian vibrational analysis (PHVA) includes only a limited number of atoms. In the following we consider the Cossee-Arlman mechanism at the CA-I site described in the main manuscript (Figure 1a), deriving vibrational frequencies in two approaches: 1) FHVA, and 2) PHVA, including only the atoms of the adsorbates. Figure S3 shows the results. As can be seen, there is an increasing difference between the free energy profiles with increasing size of the adsorbates. The differences relate almost exclusively to entropy; enthalpies (relative to the site) calculated using FHVA and PHVA differ by at most 5 kj/mol for the species shown in Figure S3. The difference between FHVA and PHVA originates from the low frequency vibrations that contribute most to the entropy. PHVA works reasonably well for smaller adsorbates containing two to four carbon atoms, where the adsorbate vibrations do not couple extensively to the framework. However, for larger species PHVA starts to deteriorate, returning spurious imaginary frequencies that were not present in FHVA. Hence, PHVA returned one imaginary frequency for [Ni-ethene-hexyl] +. This vibration resembles a rotational, wagging motion of the hexyl chain. When displaying the PHVA profile in Figure S3 this frequency was replaced by 12 cm 1 to limit the entropy contribution to that of a free rotation. 1 We however consider this an unsatisfactory approach and hence rely exclusively on FHVA results when discussing the free energy profile of the Cossee-Arlman mechanism in the main manuscript. Likewise, we employed FHVA when calculating the free energy spans used to discriminate the oligomerization pathways. S6

7 1/2 H 2 H Si O Al O Si C 2 H 5 H C 2 H 5 H C 4 H 9 C 4 H 9 H C 4 H 9 H H C 6 H 13 Figure S3: Gibbs free energy profiles (393 K, 1 atm) for the Cossee-Arlman mechanism of ethene oligomerization on the CA-I site in the AFI zeolite framework, extending Figure 4 of the main manuscript. The profiles were derived from harmonic frequencies calculated using FHVA and PHVA, respectively (see text). S7

8 S4 Additional potential and free energy profiles H Ni 2+ Ni Ni Ni Ni 2+ H Ni Ni Ni Figure S4: Calculated free energy profiles (393 K, 1 atm, PHVA) for ethene dimerization according to the metallacycle mechanism on a neutral nickel atom (black) as well as (blue) and Ni 2+ ions (red) in the AFI zeolite framework. The red structures are representative for all three profiles. In this and following figures, the arrows indicate (de)sorption steps of gas phase molecules. The neutral nickel atom was placed in a super cell of a purely siliceous AFI framework. The ionic sites are illustrated in Figure 1 of the main manuscript (Figures S1B and S2D). The pathway on Ni 2+ ion is associated with the lowest free energy of activation. Note that we derived virtually identical free energy spans of the metallacycle mechanism on Ni 2+ ion from FHVA and PHVA profiles (Table S2). This justifies calculating only the pathway on Ni 2+ in FHVA, considering that the pathways on Ni 2+ and have very similar structures (Figures S9 and S10) and PHVA entropy profiles (Table S6). The metallacycle mechanism on Ni(0) is clearly unfavored and has hence not been considered further. S8

9 H Ni 2+ Ni 2+ Ni 2+ H Ni 2+ Ni 2+ Figure S5: Calculated free energy profiles (393 K, 1 atm, PHVA) for the metallacycle mechanism on the Ni 2+ sites shown in Figures S2C and D, respectively. Despite several attempts, it was not possible to locate the barrier to metallacycle formation for the S2C site. It is however expected to be significantly higher than for the S2D site, based on the instability of the ethene dimer and metallacycle intermediates at the S2C relative to the S2D site. Figure S6: Potential energy profiles for metallacycle formation at Ni 2+ and Ti 2+ sites. The calculations used the same configuration of the site for both metals as shown for Ni 2+ in Figure S2D. S9

10 TDTS H O O O Al Si Al Ni 2+ O O Al Si Al H O O O Al Si Al Ni 2+ Ni 2+ H O O O Al Si Al H Ni 2+ Ni 2+ TDIS Ni 2+ Figure S7: Calculated free energy profiles (393 K, 1 atm, PHVA) for ethene dimerization via proton-transfer, both self-assisted (compare to Figure 3 of main manuscript) and etheneassisted. The calculations employed the same Ni 2+ site as in the main manuscript and placed the zeolite proton on O 1 (see Figure 1b of main manuscript). S10

11 OH H OH OH H OH H OH OH OH O OH H H Figure S8: Calculated free energy profiles (393 K, 1 atm, PHVA) for formation of a Cossee- Arlman site (CA-I, Figure 1a of main manuscript) from [Ni-OH] + (Figure S1c) and ethene. The Ni species are charge-balanced by the AFI framework as shown for [Ni-OH] + in Figure S1C. Ethene adsorbs on [Ni-OH] + and partakes in a migratory insertion reaction, analogously to the Cossee-Arlman mechanism itself. The ethenol product isomerizes to ethanal and leaves behind the [Ni-ethene-H] + species. The tautomerization of ethenol to ethanal (red) is considered to occur in the gas phase in parallel with the last step on the zeolite surface: rotation of ethene and formation of the agostic bond. As can be seen, the reaction is exergonic and associated with a free energy of activation of 65-(-36)=101 kj/mol. S11

12 S5 Atomic structures All structures optimized in this work are available as XYZ files. Figures S9 S11 show structures of selected nickel species. Figure S9: Optimized structures along the metallacycle pathway for ethene dimerization on ions (Figure S4). A) ethene dimer on, B) transition state for C-C coupling, C) the nickelacyclopentane intermediate. Figure S1B shows the site in the AFI framework, which has been omitted here for clarity. Figure S10: Optimized structures along the metallacycle pathway for ethene dimerization of Ni 2+ ions (Figure S4 and Figure 2 of main manuscript). A) ethene dimer on Ni 2+, B) transition state for C-C coupling, C) the nickelacyclopentane intermediate. Figure S2D shows the site in the AFI framework, which has been omitted here for clarity. S12

13 Figure S11: Optimized structures of transition states for reaction steps of migratory insertion in [Ni-alkene-alkyl] + species, derived from the CA-I site shown in Figure 1a of the main manuscript. A) ethene dimerization, γ-hydrogen coordination, B) ethene dimerization, δ- hydrogen coordination, C) ethene-butyl reaction, γ-hydrogen coordination. The AFI zeolite framework has been omitted here for clarity. S13

14 S6 Additional thermochemical data Table S2: Calculated free energy spans G at 393 K, 1 atm, for the metallacycle, protontransfer and Cossee-Arlman mechanisms (Figures 2, 3 and 4 in main manuscript). The free energies are derived using PHVA and FHVA, respectively (Section S3). Mechanism TDIS G, FHVA G, PHVA Metallacycle, Ni 2+ [Ni-1-butene-1-butene] Proton-transfer [Ni-1-butene-1-butene] Cossee-Arlman [Ni-ethene-ethyl] a a : TDIS in PHVA is [Ni-1-butene-butyl] +, resulting in a free energy span of 86 kj/mol. S14

15 Table S3: Calculated thermodynamic data at 393 K, 1 atm, FHVA, for adsorption of ethene and 1-butene on [Ni-1-butene] 2+ and [Ni-1-butene-H] + species in SSZ-24, as well as physisorption in the purely siliceous AFI framework. Sorbent Adsorption of H ads (kjmol 1 ) S ads (Jmol 1 K 1 ) G ads (kjmol 1 ) [Ni-1-butene-H] + a ethene butene [Ni-1-butene] 2+ b ethene butene siliceous AFI ethene 1-butene -30 c -59 c a : homologue of the CA-I site, shown in Figure 1a in the main manuscript. b : same Ni 2+ site as in the main manuscript, Figure S2D. c : electronic energies. Table S4: Calculated thermodynamic data at 393 K, 1 atm, FHVA, of [Ni-alkene-alkyl] + species derived from the CA-I site shown in Figure 1a in the main manuscript. Values (kj/mol) are tabulated relative to [Ni-ethene-ethyl] +, with non-adsorbed alkenes considered to be in the gas phase. Species H (kjmol 1 ) S (Jmol 1 K 1 ) G (kjmol 1 ) [Ni-ethene-butyl] [Ni-1-butene-1-butyl] [Ni-ethene-1-hexyl] S15

16 Table S5: Calculated thermodynamic data at 393 K, 1 atm, FHVA, for the metallacycle mechanism shown in the full line in Figure 2 of the main manuscript. Numbers are tabulated relative to the bare Ni 2+ site. Each column represents a state in the same order (left to right) as shown in the figure. Electronic energy (kjmol 1 ) Enthalpy (kjmol 1 ) Entropy (Jmol 1 K 1 ) Gibbs free energy (kjmol 1 ) S16

17 Table S6: Calculated thermodynamic data at 393 K, 1 atm, PHVA, for the metallacycle mechanisms shown in Figure S4. Numbers are tabulated relative to the bare Ni(0), and Ni 2+ sites, respectively. Each column represents a state in the same order (left to right) as shown in the figure. Electronic energy (kjmol 1 ) Ni Ni(0) Enthalpy (kjmol 1 ) Ni Ni(0) Entropy (Jmol 1 K 1 ) Ni Ni(0) Gibbs free energy (kjmol 1 ) Ni Ni(0) S17

18 Table S7: Calculated thermodynamic data at 393 K, 1 atm, for the Cossee-Arlman mechanism shown in Figure 4 of the main manuscript. Numbers are tabulated relative to the CA-I site (Figure 1a of main manuscript) and were derived using FHVA and PHVA (Section S3). Each column represents a state in the same order (left to right) as shown in Figure 4. Electronic energy (kjmol 1 ) FHVA PHVA Enthalpy (kjmol 1 ) FHVA PHVA Entropy (Jmol 1 K 1 ) FHVA PHVA Gibbs free energy (kjmol 1 ) FHVA PHVA S18

19 Table S8: Calculated thermodynamic data at 393 K, 1 atm, FHVA, for the self-assisted proton-transfer mechanism shown in the full line of Figure 3 of the main manuscript. Numbers are tabulated relative to the Ni 2+ ion (Figure S2D). Each column represents a state in the same order (left to right) as shown in Figure 3. Data are shown for the proton positioned on O 1 (see Figure 1b of the main manuscript). Electronic energy (kjmol 1 ) O 1 -H Enthalpy (kjmol 1 ) O 1 -H Entropy (Jmol 1 K 1 ) O 1 -H Gibbs free energy (kjmol 1 ) O 1 -H S19

20 Table S9: Calculated thermodynamic data at 393 K, 1 atm, PHVA, for the self-assisted proton-transfer mechanism shown in Figure S7. Numbers are tabulated relative to the Ni 2+ ion (Figure S2D). Each column represents a state in the same order (left to right) as shown in Figure S7. Data are shown for the proton positioned on O 1 and O 5, respectively (see Figure 1b of the main manuscript). Figure S7 shows the results from the O 1 position, as it results in the most favorable pathway. Electronic energy (kjmol 1 ) O 1 -H O 5 -H Enthalpy (kjmol 1 ) O 1 -H O 5 -H Entropy (Jmol 1 K 1 ) O 1 -H O 5 -H Gibbs free energy (kjmol 1 ) O 1 -H O 5 -H S20

21 Table S10: Calculated thermodynamic data at 393 K, 1 atm, PHVA, for the ethene-assisted proton-transfer mechanism shown in Figure S7. Numbers are tabulated relative to the Ni 2+ ion (Figure S2D). Each column represents a state in the same order (left to right) as shown in Figure S7. Data are shown for the proton positioned on O 1 and O 5, respectively (see Figure 1b of the main manuscript). Figure S7 shows the results from the O 1 position, as it results in the most favorable pathway. Electronic energy (kjmol 1 ) O 1 -H O 5 -H Enthalpy (kjmol 1 ) O 1 -H O 5 -H Entropy (Jmol 1 K 1 ) O 1 -H O 5 -H Gibbs free energy (kjmol 1 ) O 1 -H O 5 -H S21

22 Table S11: Calculated thermodynamic data at 393 K, 1 atm, PHVA, for formation of a CA-II Cossee-Arlman site from a Ni 2+ ion, as shown in Figure 6 of the main manuscript. Numbers are tabulated relative to the ethene dimer on Ni 2+. Each column represents a state in the same order (left to right) as shown in the profile in the figure. Data are shown for the proton positioned on O 1 and O 5, respectively (see Figure 1b of the main manuscript). The main manuscript includes only results from the O 1 position, as it results in the most favorable pathway. Electronic energy (kjmol 1 ) O 1 -H O 5 -H Enthalpy (kjmol 1 ) O 1 -H O 5 -H Entropy (Jmol 1 K 1 ) O 1 -H O 5 -H Gibbs free energy (kjmol 1 ) O 1 -H O 5 -H S22

23 Table S12: Calculated thermodynamic data at 393 K, 1 atm, PHVA, for formation of a CA-I Cossee-Arlman site (Figure 1a of main manuscript) from a [Ni-OH] + species, as shown in Figure S8. Numbers are tabulated relative to the [Ni-OH] + site shown in Figure S1C. Each column represents a state in the same order (left to right) as shown in the profile in Figure S8. Electronic energy (kjmol 1 ) Enthalpy (kjmol 1 ) Entropy (Jmol 1 K 1 ) Gibbs free energy (kjmol 1 ) S23

24 References (1) Brogaard, R. Y.; Henry, R.; Schuurman, Y.; Medford, A. J.; Moses, P. G.; Beato, P.; Svelle, S.; Nørskov, J. K.; Olsbye, U. J. Catal. 2014, 314, (2) Loewenstein, W. Am. Mineral. 1954, 39, (3) Goltl, F.; Hafner, J. J. Chem. Phys. 2012, 136, (4) Kuroda, Y.; Yoshikawa, Y.; Kumashiro, R.; Nagao, M. J. Phys. Chem. B 1997, 101, (5) Bolis, V.; Busco, C.; Bordiga, S.; Ugliengo, P.; Lamberti, C.; Zecchina, A. Appl. Surf. Sci. 2002, 196, (6) Borgard, G. D.; Molvik, S.; Balaraman, P.; Root, T. W.; Dumesic, J. A. Langmuir 1995, 11, (7) Klier, K.; Kellerman, R.; Hutta, P. J. J. Chem. Phys. 1974, 61, (8) Halliche, D.; Cherifi, O.; Auroux, A. Thermochim. Acta 2005, 434, S24