(465g) Consequences of Diffusion, Acid Strength, and Confinement on Bifunctional Reactions of Alkanes

Hydrocracking and hydroisomerization reactions convert large-molecular-weight
alkane feedstocks to more valuable smaller and more highly branched molecules,
respectively, through cascade reactions mediated by alkene intermediates [1]. Such reactions require presence of
metal function—sufficient to establish alkane-alkene equilibrium for reactants
and to scavenge intermediates and alkene products—that resides within diffusion
distances from kinetically relevant acid function. These distances are seldom
well-defined and are sensitive to synthetic and formulation protocols but, in
part, define concentration gradients of reactant and product alkenes with
significant consequences for rates and product selectivities,
respectively, especially when acid sites reside within channels and voids of
zeolite or zeotype materials (with characteristic
sizes ~0.4-1.1 nm [2]). These
apertures and voids restrict diffusion of reactants and products—and do so
sensitively based on molecule size and shape—and also confer higher reactivity
through stabilizing van der Waals and electrostatic interactions with
carbocation moieties at ion-pair transition states. Resultant concentration
gradients of product alkenes depend on the size of the acid domain and on the density
of protons and lead to product selectivities
reflecting multiple secondary interconversion events occurring before product
alkenes exit acid domains and become hydrogenated at extracrystalline
metal functions. Such diffusion-enhanced secondary reactions are rarely treated
using rigorous kinetic-transport formalisms, and their effects on measured
isomerization and β-scission
rates and selectivities are typically attributed,
unconvincingly and often inaccurately, to transition state selectivity or to effects
of framework structure on acid strength.

Here, we dissect the effects of diffusion, acid strength,
and transition state confinement to assess their respective catalytic
consequences. Kinetic data, benchmarked against density functional theory, are reported
for n-heptane (nH) and 2,4-dimethylpentane (24DMP)
isomerization and β-scission on
physical mixtures of Pt/SiO₂ and zeolites with diverse confining
environments (MFI, MTW, BEA, SFH, FAU) but similar acid strengths and MFI zeotypes with different framework heteroatoms (Al, Fe, Ga,
B) which contain H⁺ of different acid strength within the same
confining environment. To attain intrinsic product formation rates, we assess
the effects of product-alkene concentration gradients within acid domains using
rigorous reaction-transport formalisms that account for diffusion timescales independently
measured from isochoric, isothermal
transient uptake of 2,2-dimethylbutane. We additionally develop and report
descriptors of confinement as van der Waals interaction energies, E_vdw,
for transition states based on Lennard-Jones potential force fields [3] and of acid strength as
deprotonation energy (DPE), defined as that required to heterolytically cleave
the Brønsted O-H bond; these descriptors, together
with thermochemical cycles providing the conceptual framework for interpreting
activation barriers in terms of separable properties of the acid (E_vdw
and DPE), the reactant (proton affinity), and the interacting moieties at the
confined ion-pair transition state, permit systematic interpretation of the
independent effects of acid strength and confinement on intrinsic isomerization
and β-scission rates and selectivities.

n-Heptane and
2,4-dimethylpentane isomerization and β-scission turnover rates (per H⁺; 548 K; 0.5-25 kPa alkane; 60-101 kPa H₂)
on all physical mixtures are proportional to alkane/H₂ molar ratio
(a surrogate for the prevalent pressure of heptenes)
at low values of this ratio and approach constant values as molar ratio
increases. First-order rate constants (k_isom,nHK_prot,nH
for n-heptane isomerization and (k_isom,24+k_β,24)K_prot,24 for combined
2,4-dimethylpentane isomerization and β-scission)
reflect the free-energy differences between their respective confined, ion-pair
transition states and unconfined, neutral precursors (gas phase alkene &
acid sites). Measured ratios of k_β,24/k_isom,24 reflect ubiquitous
secondary reactions that occur within the acid domain before alkene products
leave an acid domain and become alkanes. Similarly, on all acids and for both
reactants, measured product selectivities (the ratio
of the measured formation rate for a given product to the consumption rate of
the reactant) were invariant with reactor residence time or alkene reactant
pressure; these selectivities do not reflect the
intrinsic values of primary formation events because primary alkene products
undergo multiple secondary interconversions before they form their respective
less-reactive alkane analogs. The extent of these diffusion-enhanced secondary
reactions was assessed for each aluminosilicate by varying
the intracrystalline residence time (τ_int)
through monotonic changes in the proton density ([H⁺]) during
desorption of weakly bound NH₃ titrants from the active protons.
Figure 1 shows selectivities measured during NH₃
desorption as a function of τ_int for (a) β-scission products and (b) isomer products during
2,4-dimethylpentane conversion on FAU-Pt/SiO₂ catalyst mixtures.
Product selectivities at low values of τ_int
­ are invariant with τ_int, indicating that such β-scission selectivities
reflect primary chemical events. As τ_int increased, β-scission, n-heptane, 2-methylhexane, and 3-methylhexane selectivities increased, while that for to
2,3-dimethylpentane decreases, indicating the prevalence of secondary alkene
interconversions of primary 2,4-dimethylpentane products as [H⁺]
increases.

 Rigorous
reaction-transport treatments reveal that a Thiele parameter for the primary isoalkene products determines the extent of such
diffusion-enhanced secondary reactions (
, where k_i,isoK_prot,iso
is the first-order rate constant for secondary reaction i of the isoalkenes,
D_iso
is diffusivity of the isoalkenes, and R²
is zeolite crystallite size). Figure 2 shows β-scission selectivities for nH reactant during NH₃ desorption on FAU, BEA,
and MFI (as physical mixtures with Pt/SiO₂; BEA and MFI of different
[H+]R²/D_22DMB
are indicated, where D_22DMB/R²
was measured using transient uptake of 2,2-dimethylbutane) as a function of the
Thiele modulus for n-heptane (
); the dashed curve is the regressed fit to the theoretical
expression describing these secondary reactions (Fig. 2, inset equation, where
 and
) with a single fitting parameter (
) that has the same value for all aluminosilicates. β-Scission selectivities
increased monotonically as
 increases and are a single-valued function of
this parameter; selectivities for all
aluminosilicates lie along the same curve, despite their very different void
sizes, indicating that all frameworks stabilize the two transition states (for β-scission and isomerization) to
the same extent.

These confinement
effects, however, lead to preferential stabilization of both β-scission and isomerization transition states with respect to
their gas phase precursors; the stabilization of transition states by van der
Waals interactions with the framework is reflected in E_vdw, where a better
match between void size/shape and organic carbocation size/shape results in
more negative values. First-order nH isomerization
rate constants decrease exponentially with increasing E_vdw. The effects of E_vdw on activation barriers are attenuated
by energy penalties brought forth by distortions of the framework (E_str)
and the carbocation at the transition state in order to establish more
effective van der Waals contacts, especially as the organic moiety approaches
the size of the confining voids.

Such organic
carbocations, together with the conjugate anion of the zeolite or zeotype, constitute ion pair transition states mediating
isomerization and β-scission
reactions. First-order
isomerization and β-scission
rate constants for 2,4-dimethylpentane, measured on MFI zeotypes
with H⁺ of different acid strength (by varying framework heteroatom,
in order of decreasing acid strength: Al, Ga, Fe, B; DPE = 1226-1292 kJ mol^-1;
in physical mixture with Pt/SiO₂), were corrected for
diffusion-enhanced secondary reactions to obtain intrinsic values. These
first-order rate constants decrease exponentially with increasing DPE,
predominantly a reflection of the less stable conjugate anion at ion-pairs of
weaker acids. The ratio of rate constants k_β,24/k_isom,24, however, increases
with increasing DPE, indicating that β-scission
occurs preferentially on weaker acids. Cations at transition states mediating β-scission resemble “harder” carbenium ions (according to classical acid-base formalisms
[4]) compared to those for
isomerization. These “harder” carbocations interact more effectively with the
conjugate anion, recovering more of the energy required to deprotonate a weaker
acid, and consequently are less sensitive to DPE. Thus, for transition state
structures differing in their tendency to delocalize charge, acid strength can
influence product selectivity through stronger effects of DPE for the “softer”
transition state.

This work
mechanistically assesses the effects of both confinement and acid strength on
intrinsic isomerization and β-scission
turnover rates and selectivities. Methods of
untangling the dramatic and ubiquitous effects of diffusion-enhanced secondary
reactions on measured product selectivities are also
demonstrated. The insights gained from this work can be applied to design
catalytic architectures to obtain desired selectivities
and reaction rates.

References

[1]     H.L. Coonradt, W.E. Garwood, Ind. Eng. Chem. Process
Des. Dev. 3 (1964) 38–45.

[2]      C. Baerlocher,
L. McCusker, Database of Zeolite Structures,
http://www.iza-structure.org/databases/.

[3]      S. Herrmann, E. Iglesia,
J. Catal. 346 (2017) 134–153.

[4]      G. Pearson, J. Am. Chem. Soc. 85 (1963)
3533–3539.

The authors acknowledge technical
discussions with Drs. Stacey I. Zones (Chevron) and William Knaeble
(UC Berkeley), diffusion timescale measurements by Zhichen
(Jane) Shi (UC Berkeley), financial support from Chevron Energy Technology
Company, and computational resources from the National Science Foundation’s
Extreme Science and Engineering Discovery Environment (ACI-1053575).