Control of coordinatively unsaturated Zr sites in ZrO2 for efficient C–H bond activation

Due to the complexity of heterogeneous catalysts, identification of active sites and the ways for their experimental design are not inherently straightforward but important for tailored catalyst preparation. The present study reveals the active sites for efficient C–H bond activation in C1–C4 alkanes over ZrO2 free of any metals or metal oxides usually catalysing this reaction. Quantum chemical calculations suggest that two Zr cations located at an oxygen vacancy are responsible for the homolytic C–H bond dissociation. This pathway differs from that reported for other metal oxides used for alkane activation, where metal cation and neighbouring lattice oxygen form the active site. The concentration of anion vacancies in ZrO2 can be controlled through adjusting the crystallite size. Accordingly designed ZrO2 shows industrially relevant activity and durability in non-oxidative propane dehydrogenation and performs superior to state-of-the-art catalysts possessing Pt, CrOx, GaOx or VOx species.

Hereafter, the precipitate was washed thoroughly with deionized water and dried at 110 °C overnight. The obtained powder was calcined at 110, 200, 250, 300, 350, 400 or 450 °C for 4 h.
The resulting catalysts were denoted as ZrO 2 _1, ZrO 2 _2, ZrO 2 _3, ZrO 2 _4, ZrO 2 _5, ZrO 2 _6, ZrO 2 _7 respectively. The catalysts ZrO 2 _8 to ZrO 2 _14 were prepared in a similar way. Briefly For the preparation of the catalysts ZrO 2 _13 and ZrO 2 _14, the solution was aged at 100 °C for 2 days. For the synthesis of other catalysts, the aging temperature was kept at 80 °C. Hereafter, the solution was transferred into autoclave. Hydrothermal treatment was performed at 180 °C for all catalysts. Crystallization time was 6,12,24,48,72,24, and 48 h for ZrO 2 _8, ZrO 2 _9, ZrO 2 _10, ZrO 2 _11, ZrO 2 _12, ZrO 2 _13, ZrO 2 _14 respectively. After that, all the catalysts were dried at 110 o C overnight and then calcined at 550 °C for 4 h. The catalysts from ZrO 2 _15 to ZrO 2 _21 were prepared in a similar way as ZrO 2 _11, but in the presence of F127 (Aldrich), P123, Diethylenetriamine (99%, Aldrich), Sodium dodecyl sulphate (>98.5%, Aldrich), CTAB (>98%, Aldrich), Dodecylamine (98%, Aldrich) and NH 4 F (>98%, Fluka) with the molar ratio of such additives to Zr 4+ of 0.1, 0.1, 0.5, 0.1, 0.5, 0.5, and 0.3 respectively. The synthesis procedure of ZrO 2 _22 was similar to that of ZrO 2 _15. The only difference was the absence of the template P123. ZrO 2 _23 was prepared similarly as ZrO 2 _21, but the NH 4 F to Zr 4+ molar ratio was 1.4, and the solution was crystallized at 105 °C for 24 h. ZrO 2 _24 was prepared using ZrO(NO 3 ) 2 ·xH 2 O as a precursor. The procedure was as follows. 7.622 g of ZrO(NO 3 ) 2 ·xH 2 O was firstly dissolved in 500 mL of H 2 O. Then, ethylenediamine was added dropwise to the solution until pH reached 11. Hereafter, the solution was stirred for 30 min at room temperature, heated up to 100 °C and stirred for 4 days. The obtained precipitate was filtered, dried and calcined at 550 °C for 4 h. ZrO 2 _25 was prepared using ZrOCl 2 ·8H 2 O (>95%, Fluka) as a precursor. Briefly, required amount of ZrOCl 2 ·8H 2 O was dissolved in deionized water to obtain solution with a concentration of Zr 4+ of 1.0 mol·L -1 . An aqueous solution of ammonia was then added dropwise under stirring until pH reached 9. The precipitate formed was aged overnight, filtered and washed several times with deionized water until no more chloride ions were identified in filtrate (reaction with AgNO 3 ). The solid was dried at 110 °C overnight and calcined at 550 °C for 4 h.

Supplementary Note 2
Temperature-programmed desorption (TPD) of CO was carried out in an in-house developed setup containing eight individually heated continuous-flow fixed-bed quartz reactors.
The following procedure was applied. The catalysts (0.05 g) were initially calcined in air at 550 °C for 1 h, purged with Ar for 15 min and then treated with 57 vol% H 2 in Ar flow for 1 h.
Hereafter, the treated catalysts were cooled down in Ar flow to 250 °C and then exposed to a flow of CO (1 vol% CO in Ar, 10 mL·min -1 ) for 1 h followed by cooling down in Ar flow to room temperature. Finally, the catalysts were heated to 900 °C with a heating rate of 10 K·min -1 in Ar flow (10 mL·min -1 ). Desorbed CO was registered by an on-line mass spectrometer (Pfeiffer Vacuum OmniStar GSD 320) at the atomic mass unit (AMU) of 28.
To investigate acidic properties of ZrO 2 , NH 3 or C 3 H 6 were used as probe molecules for TPD tests. The catalysts (0.05 to 0.2 g) were initially calcined in air at 550 °C for 1 h, purged with Ar for 15 min and then treated with 57 vol% H 2 in Ar flow at the same temperature for 1 h.
After that, the catalysts were cooled down in Ar flow to 120 (for NH 3 adsorption) or to 50 °C (for C 3 H 6 adsorption). Hereafter, they were exposed to a flow of either 1 vol% NH 3 in Ar (14 mL·min -1 ) or 5 vol% C 3 H 6 in Ar (6 mL·min -1 ) for 1 h. The catalysts for NH 3 -TPD were then cooled down in Ar flow to 80 °C, tempered for 2 h and finally heated to 900 °C with a heating rate of 10 K·min -1 . The catalysts for C 3 H 6 -TPD were purged in Ar flow at 50 °C for 12 h to remove weakly adsorbed C 3 H 6 and finally heated to 900 °C with a heating rate of 10 K·min -1 .
Desorbed ammonia or propene was registered by an on-line mass spectrometer (Pfeiffer Vacuum OmniStar GSD 320) at AMU of 15 (NH) or AMU of 41 (C 3 H 5 ) and 40(Ar).
Temperature-programmed reduction (TPR) experiments were performed using a feed with either 5 vol% H 2 or 1 vol% CO in Ar. Catalysts (0.2 g) were firstly calcined in air at 550 °C for 1 h, then cooled down in Ar flow to room temperature. Hereafter, they were heated in a flow of 5 vol% H 2 (10 mL·min -1 ) or 1 vol% CO (10 mL·min -1 ) in Ar from room temperature to 900 °C with a heating rate of 10 K·min -1 . The consumption of H 2 or CO was quantified with an on-line mass spectrometer (Pfeiffer Vacuum OmniStar GSD 320).
The electrical conductivity measurements of the catalysts were performed in a fixed-bed continuous-flow tubular quartz reactor. Firstly, each catalyst was pressed into a dense disc of 7.1 mm in diameter and 1.2-1.4 mm in height to minimize interior grain boundaries. Both surfaces of the disc were covered with a platinum conducting paste (ChemPur, 71% Pt), and dried at 120 °C.
Hereafter, a sample disc was sandwiched between two inert platinum electrodes and arranged inside the reactor. A constant alternating current of 1 kHz was adjusted with an automatically compensating bridge (Wayne Kerr B905). At this mono-frequency regime, the contributions of both surface and bulk conductivity are regarded similar for all samples, owing to the identical treatment of the probed discs. As an equivalent circuit, a parallel connection of a condenser and an ohmic resistance was applied. Electrical conductivity of catalysts was measured at 550 °C in air and N 2 to ensure different partial pressures of oxygen, i.e. 20 kPa and about 10 -4 kPa respectively. The electrical conductivity of the catalysts was calculated according to Where R is the electrical resistance (Ω), and the ratio h/S represents the thickness h (mm) and cross section area S (mm 2 ) of the catalyst disc.

Supplementary Note 3
Catalytic tests with C 2 H 6 , C 3 H 8 , and iso-C 4 H 10 were performed at 1 bar between 550 and 625°C using an in-house developed setup consisting of 15 continuous-flow fixed-bed quartz tubular (length and inner diameter are 465 and 4 mm respectively) reactors operating in parallel.
Tests with CH 4 were carried out at 800°C and 1.25 bar in an in-house developed setup consisting of 6 continuous-flow fixed-bed quartz tubular (length and inner diameter are 330 and 4 mm respectively) reactors operating in parallel. To determine the rate of olefin formation, the tests were carried out at a degree of alkane conversion below 10% using a feed containing 40 vol% C m H 2m+2 (m=1, 2, 3, or 4) in N 2 . The mass (0.05 to 0.3 g) of catalyst and total flow (10 to 40 mL·min -1 ) were varied to fulfil this requirement. The catalysts were firstly heated in N 2 flow up to 550 °C (in case of C 2 H 6 , C 3 H 8 and iso-C 4 H 10 ) or to 800°C (in case of CH 4 ) followed by calcination in air flow for 1 h. Then they were purged with N 2 for 15 min, treated with H 2 or CO flow (57 vol% H 2 or CO in N 2 ) for 1 h and flushed in N 2 for 15 min. Finally, a flow of C m H 2m+2 -N 2 mixture (40 vol% C m H 2m+2 in N 2 ) was passed through the catalyst bed. To investigate the influence of the nature of reducing agent (H 2 or CO) and the effect of reduction time on catalyst activity, the reduction time varied from 0 to 420 min and from 0 to 50 min for H 2 and CO respectively. The initial rate of olefin formation in tests with CH 4 , C 2 H 6 , C 3 H 8 or iso-C 4 H 10 was measured after 600, 600, 190 or 540 s on stream.
To determine an integral propene selectivity (Supplementary Equation 2), catalytic tests were performed at 550 °C for 1 hour. Total flow of a mixture of 40 vol% C 3 H 8 in N 2 was fixed at 10 mL·min -1 , catalyst amount (0.12 to 2.4 g) was varied to achieve an initial propane conversion of 30%. On stream profiles of propene yield and propane conversion in Supplementary Figure 4 were integrated to obtain the number of moles of propene formed (n(C 3 H 6 ) formed ) and the number of moles of propane consumed (n(C 3 H 8 ) consumed ) within 1 hour propane dehydrogenation.
Here, is a volumetric feed flow rate (mL·h -1 ), ( ) is a molar fraction of olefin, is molar volume (22400 mL·mol -1 ), is catalyst amount (g), n with superscripts "in" and "out" is a molar flow of gas phase components (indicated with subscripts " ", "C m H 2m+2 " or "N 2 ") at the reactor inlet and outlet respectively (mol·min -1 ), is reciprocal stoichiometric coefficient for product , ( ) is a molar weight of (g·mol -1 ).

Supplementary Methods
Spin-polarized and periodic density functional theory (DFT) calculations were carried out by using the Vienna ab initio simulation package (VASP) 3,4 . Exchange and correlation were treated within the Perdew-Burke-Ernzerhof generalized gradient approximation (GGA-PBE) 5 . To obtain accurate energies with errors of less than 1 meV per atom, a cutoff energy of 400 eV was used.
Geometry optimization was converged until forces acting on atoms were lower than 0.02 eV/Å, whereas the energy threshold defining self-consistency of the electron density was set to 10 −4 eV.
The Climbing Image Nudged Elastic Band (CI-NEB) method with eight images was applied for finding transition states and minimum energy paths of all reactions 6 Figure 23 indicate that the predictions from both methods support the conclusions of this study.
As XRD analysis proved that our catalysts mainly contain the monoclinic phase, we also  On the basis of the computed C-H dissociative adsorption energy (Supplementary Table 7) we considered the dehydrogenation of propane to propene firstly via methylene C-H activation Since propene has very low adsorption energy (0.13 eV) and can desorb easily from the surface immediately after its formation (D2), we therefore considered H 2 formation directly.
where S rot = R(lnq rot +3/2) (For non-linear molecule) (15) where (16) where where I is the moment of inertia, s is the rotational symmetry number and m is the mass of the molecule.
The translational, rotational, and vibrational and entropic contributions of gas-phase molecules were calculated on the basis of ideal gas models. However, it should be noted that both propane and propene do not follow the ideal gas rule and this makes such a pressure and temperature dependent reaction more complicated than expected on the basis of the ideal gas models. For the adsorbed molecules and transition states on the surface, the rotational and translational contributions were converted into vibrational modes. We also approximated that the PV term is negligible because it is very small with regard to the energetic terms. Hence, the Gibbs energy for the surface species is computed from the simplified Supplementary Equation 17.