The effect of particle size and support on the catalytic performance of supported subnanometer copper clusters was investigated in the oxidative dehydrogenation of cyclohexene. From among the investigated seven size-selected subnanometer copper particles between a single atom and clusters containing 2–7 atoms, the highest activity was observed for the titania-supported copper tetramer with 100% selectivity toward benzene production and being about an order of magnitude more active than not only all the other investigated cluster sizes on the same support but also the same tetramer on the other supports, Al2O3, SiO2, and SnO2. In addition to the profound effect of cluster size on activity and with Cu4 outstanding from the studied series, Cu4 clusters supported on SiO2 provide an example of tuning selectivity through support effects when this particular catalyst also produces cyclohexadiene with about 30% selectivity. Titania-supported Cu5 and Cu7 clusters supported on TiO2 produce a high fraction of cyclohexadiene in contrast to their neighbors, while Cu4 and Cu6 solely produce benzene without any combustion, thus representing odd–even oscillation of selectivity with the number of atoms in the cluster.
Cyclic hydrocarbons have aroused considerable interest for their use as precursors for the production of valuable chemicals, including oxygenated cyclic hydrocarbons such as alcohols,1 ketones,2 and epoxides;3 the synthesis of polymers and agrochemicals;4 or in drug manufacturing by the pharmaceutical industry.1,5 The oxidative dehydrogenation (ODH) of cyclic hydrocarbons represents, among others, a promising approach in petroleum refining and reforming processes, in particular in the conversion of cyclohexane to benzene, through cyclohexene as an intermediate. As a structure-sensitive reaction,6–8 it represents a complex process, including the formation of partially dehydrogenated intermediates, such as cyclohexene,9,10 where the adsorbed cyclohexene is required to accommodate a specific orientation on the top of the catalyst to facilitate the subsequent dehydrogenation step.11,12 Due to the rate-limiting nature of this step, cyclohexene is often used in studies focused on the detailed understanding of cyclohexane dehydrogenation and it can also serve as a model molecule for studying structure-sensitive reactions, in general.13–15 Also of interest is the environmentally benign preparation of adipic acid via catalytic oxidation of cyclohexene that avoids the use of the nitric acid oxidant, which produces the greenhouse gas by-product N2O,1,5 or the production of cyclohexadiene, an important precursor for the manufacture of polycarbonates and polyesters.16 For mentioned reasons, a proper understanding of the reaction pathways and control of ODH rates is important for further implementation in industries.13,17 Moreover, cyclohexene is an intermediate in the benzene–cyclohexane–benzene cycle involving hydrogenation and dehydrogenation steps in perspective liquid organic hydrogen carriers with metal catalysts, including the copper-based ones.18 Cyclohexene and possible products from its oxidative dehydrogenation are shown in Scheme 1.
Numerous studies focused on the elucidation of a catalyst’s performance in the oxidative dehydrogenation of cyclohexene. In the pioneering work of Erkelens et al. published in 1962, a gold film was used to study the reaction of hydrogen and deuterium with cyclohexene, yielding information about the H–D exchange while also producing cyclohexane and benzene, i.e., both hydrogenation and dehydrogenation products, respectively.19 Simultaneous hydrogenation and dehydrogenation of cyclohexene were observed on a Pt(100) single crystal surface under cyclohexene, yielding benzene and cyclohexane in a ratio of about 3:1. In the presence of 15 Torr H2, the production of cyclohexane prevailed at temperatures around 350 K, while above 400 K, benzene production became dominant.20 Dummer et al. investigated TiO2 supported precious metal-based nanocatalysts of 8–10 nm particle size made of mono- and bimetallic Au and Pd. At 423 K, 100% conversion of cyclohexene and 99% selectivity toward benzene were observed on a Pd/TiO2 catalyst.21 In addition to precious metal-based catalysts, several other catalysts were also reported in the literature for the transformation of cyclohexene, such as alumina-supported molybdena catalyst prepared by metal oxide vapor synthesis, which exhibited 50% selectivity toward benzene at 653 K.22
Subnanometer size catalytic moieties can offer potentially new ODH catalysts by exploiting the strongly size-, composition-, and support-dependent properties of these atomic cluster-based materials,23–29 including the ODH of cyclohexene,25,26 the subject of the present study. Despite recent progress, the performance of the existing catalysts toward the formation of valuable intermediates remains rather limited without a stoppage before full dehydrogenation to benzene and often hampered with overoxidation, leading to combustion.
Herein, we present results of the ODH of cyclohexene from a study covering an extended range of Cu particle sizes, including the single atom and clusters from the dimer to the heptamer supported on TiO2, along with a glimpse of the effect of support on catalyst performance using SnO2, Al2O3, and SiO2 as additional supports. The results described below reveal (1) a strong dependence of the activity and selectivity on the size of the clusters, (2) support effect on performance, and (3) indication of odd–even oscillations of selectivity with the number of atoms in the cluster.
Preparation of the catalyst
As a support material, the surface of naturally oxidized N-type phosphorus doped silicon chips (Si) of 500 µm thickness and with an ∼2 nm thick native oxide top layer was coated with thin layers of TiO2, Al2O3, and SnO2 by atomic layer deposition (ALD). A 2 nm thin TiO2 layer was prepared by 90 ALD cycles of alternating exposures to titanium tetrachloride (TiCl4) and H2O, using 0.5 s reactant exposures and purge times of 60 s following TiCl4 and 120 s following H2O at a deposition temperature of 50 °C.30 The SnO2 and Al2O3 films were prepared in an R-200 standard reactor (Picosun). The 1.5 nm Al2O3 film was prepared by using alternating exposures of trimethylaluminum (TMA, EpiValence) as an Al precursor and H2O (EpiValence) as an oxidant, applying 18 cycles. Each cycle comprised a pulse-purge sequence: 0.1 s pulse of TMA–5 s N2 purge–0.1 s pulse of H2O–8 s N2 purge. Both precursors were held in a stainless steel bubbler maintained at 22 °C, the reaction chamber temperature was set to 300 °C, and the reaction chamber pressure was 9 hPa during deposition. The 1.5 nm thick SnO2 film was fabricated by alternating pulses of tetrakis(dimethylamido)tin(IV) (TDMASn, STREM Chemicals) and H2O (EpiValence) precursors and by 16 cycles consisting of a pulse-purge sequence: 1.6 s pulse TDMASn–6 s N2 purge–0.1 s pulse of H2O–9 s N2 purge. The stainless steel bubbler with TDMASn was maintained at 65 and 22 °C for H2O. The reaction chamber temperature was set to 118 °C, and the pressure was 7 hPa during deposition. Ultrahigh purity nitrogen (Messer Technogas, 99.999%) was used as carrier and purging gas of the ALD apparatus. The thickness of the ALD oxide layers was measured by ellipsometry.
A native SiO2 layer on the Si chip was cleaned by sonication with methanol and acetone; then, Al2O3, TiO2, or SnO2 films were deposited by ALD and characterized ex situ, under ambient conditions, by atomic force microscopy (AFM, Dimension Icon, Bruker, USA) in the tapping mode. Silicon cantilever VTESPA-300 with a resonant frequency of ≈300 kHz, a spring constant of k = 0.42 N m−1, and a nominal tip radius of 5 nm (Bruker, USA) was used. Gwyddion, software (v. 2.60), was utilized for AFM image data processing.
Fourier-Transform Infrared (FTIR) spectra were collected in the Attenuated Total Reflectance (ATR) mode with a Si crystal (Nicolet iS10 FTIR spectrometer equipped with a deuterated triglycine sulfate (DTGS) detector with a spectral resolution of 4 cm−1, 32 scans) under ambient conditions. The acquired spectra were processed using the OMNIC software for background subtraction. As depicted in the supplementary material, Fig. S1, the typical IR band in the range assigned to the OH stretch mode31 reveals the presence of terminal OH groups: 3797 cm−1 can be assigned to OH groups coordinated by single Al, the peak at 3735 cm−1 can be assigned to OH bridging groups, and 3647 and 3567 cm−1 can be assigned to OH stretching vibrations of OH connected to a metal atom,32–34 in accord with the reported theoretical32 studies of Al2O3 and IR characterization of the surfaces of Al2O3,34–36 SnO2,37–39 TiO2,40 and SiO2.41
The synthesis of atomically precise Cun clusters was performed in molecular beams in the gas phase, utilizing two vacuum apparatus. Both instruments consist of three interconnected vacuum chambers differentially pumped with turbomolecular pumps backed with vacuum rotary pumps. The charged clusters were produced in the first chamber by means of a magnetron-based sputter source and liquid nitrogen cooled gas aggregation source of Haberland type42 powered by an advanced energy model MDX 500 DC power supply. Argon and helium were used as sputtering and carrier gases, respectively, with a typical total flow in the range of 350–500 SCCM. 1 and 2 in. diameter copper sputter targets were used in the two cluster instruments of the analog design, respectively. The beam of positively charged copper clusters produced in the source chamber was directed onto the entrance orifice of the ion guide chamber and was collected and guided downstream into two differentially pumped chambers by a series of conical and linear octupoles. Then, the molecular beam containing a distribution of cluster sizes entered the mass spectrometer and the output of the apparatus was optimized for the desired cluster size by on-line monitoring the mass spectrum of the clusters produced. (For typical mass spectra of the produced Cun clusters used for the deposition of clusters containing 1–7 atoms, see Figs S2–S8, respectively.) Next, the desired single size cluster was filtered out on the mass spectrometer and directed toward the support in the deposition chamber. The flux of charged clusters landing on the support was measured in real-time by using a picoammeter (Keithley 6487), which was also used to bias the substrate to control the impact energy of the clusters during their landing on the support, keeping their energy below 1 eV per metal atom. Using custom homemade software, written in Python, the deposition current was converted on-line into total charge, providing the number of clusters and atoms deposited, as well as the atomic monolayer equivalent surface coverage. Typical deposition currents were in the order of several nA.
In the magnetron-based apparatus with a 1-in. target,42 single size clusters of Cu3–Cu7 were deposited on TiO2 and SiO2 substrates of 19 × 21 mm2 dimensions, applying two spots, each with a diameter of 0.8 cm (corresponding to an area of 1.0 cm2), at a surface coverage of 10% atomic monolayer equivalent, corresponding to 18.7 ng of Cu metal loading or 1.77 × 1014 Cu atoms. For Cu1–Cu2, 5% of atomic monolayer equivalent surface coverage was applied to avoid agglomeration of the clusters upon deposition. This lower coverage is equal to 9.3 ng of Cu or 8.85 × 1013 atoms of Cu. In the second apparatus, using a magnetron source with a 2-in. target, clusters on SnO2 and Al2O3 substrates were deposited to one spot with a diameter 1.2 cm (corresponding to an area of 1.13 cm2) and amount corresponding to 10% of atomic monolayer equivalent, which is equal to 21.1 ng of Cu or 2.00 × 1014 atoms of Cu. A detailed description of the cluster synthesis process can be found elsewhere.42 After deposition, the samples were transferred under air into the test setup.
Catalyst testing: Temperature programmed reaction (TPRx)
Catalytic testing was performed under continuous flow (15 SCCM) of reactants diluted in helium at a constant pressure of 800 Torr in a custom reaction cell made of alumina alloy (EN AW 6061) and an internal volume of 33 cm3, where the investigated sample was situated on a top of a heater plate. The mixing of reactant gases seeded in helium was realized in a custom mixer setup equipped with mass-flow controllers (Brooks SLA5850). The pressure in the collision cell was monitored using a pressure transducer (Omega PX209) and kept at a constant preset pressure by using a down-stream mass-flow controller (Brooks SLA5850) interconnected through a regulation loop pumped by using the diaphragm pump (Divac 1.4HV3) and controlled by a custom homemade software written in Python. The investigated sample was heated on a boron–nitride heater (Momentive Performance Material Inc.) using a Kepco power supply and the temperature ramp set and regulated using a LakeShore 340 controller in combination with a K-type thermocouple attached to the body of the heater. The body of the reaction cell was cooled by water circulating through its walls at a temperature of 25 °C maintained by an external chiller (Thermo AC200). Prior to the experiment, the temperature on the sample surface was calibrated against the temperature of the body of the heater using a second K-type thermocouple in contact with the surface of a blank silicon chip performed under the same flow of He as applied during catalyst testing. A mass spectrometer with an electron impact ionization source (Pfeiffer Vacuum Prisma Plus QMS 220) was used for online analysis of the composition of gas outgoing from the reaction cell. The gas in between the outlet of the reaction cell and the down-stream mass-flow controller was sampled into the differentially pumped mass spectrometer chamber using an electronic needle control valve (Pfeiffer EVR 116) with the flowrate controlled by a regulator (Pfeiffer RVC 300) combined with a pressure gauge (Pfeiffer PKR 261) to keep a constant pressure set to 5.0 × 10−6 mbar in the mass spectrometer chamber. The mass spectrometer chamber was pumped by using a turbomolecular pump (Pfeiffer HiCube 80 Eco), which typically reached a background pressure of about 2 × 10−8 mbar. All upstream and downstream gas lines in the path from mixer MFCs to the reaction cell and downstream from the cell were heated to 70 °C.
The mass spectrometer was operated in the continuous mass scanning mode (2 scans per minute) in the range from 5 to 100 m/z controlled by PV MassSpec software (Pfeiffer). Electron impact energy for the ionization was set to 70 eV. Sensitivity of the mass spectrometer for desired molecules (cyclohexene, benzene, CO2, CO, and O2) was determined using calibrated gas mixtures (certified analytical grade mixed gas—Siad, Messer, Air Products or Linde). The uncertainty of measured concentration is estimated to be ∼2%. Calibration of cyclohexadiene concentration was calculated using NIST gas phase ion energetics data,43 yielding an estimated error of 10%.
Before the catalytic test and with the sample in the reactor, the reaction cell and all tubings were first evacuated and flushed several times with pure helium. Then, for 2 h, a constant flow of helium (5 SCCM) was maintained through the reaction cell at 800 Torr at 25 °C. After 2 h of flushing with pure helium, the gas was switched to the reactant mixture (maintaining the 800 Torr pressure and 15 SCCM flowrate) and sampling of gas to the mass spectrometer started. The flow reactant mixture was kept for 6 h to stabilize the background in the mass spectrometer before the start of the temperature ramp. The reactant mixture consisted of 0.30% cyclohexene and 0.33% oxygen in helium (i.e., a 1:1.1 cyclohexene to oxygen molar ratio), obtained by mixing 11.25 SCCM of 4000 ppm cyclohexene in helium (Siad) with 0.50 SCCM of 10% oxygen diluted in helium (Messer) and 3.25 SCCM of pure helium (Messer).
The performance of the catalysts was tested in a temperature range from 25 to 400 °C. The temperature program was comprised of eight increasing temperature steps from 50 to 400 °C in the increments of 50 °C followed by eight decreasing steps by 50 °C from 400 °C down to 50 °C. Each temperature set point was approached with a ramp rate of 5 °C min−1 followed with a dwell time of 20 min at each temperature [see Fig. 2(a) for the applied temperature ramp].
Mass spectra of the gas mixture leaving the reaction cell were acquired continuously during the flush of the system with the reactant mixture at 25 °C, the entire temperature ramp, and after the temperature ramp at 25 °C. As a first step of the data analysis, in order to eliminate a possible long-term drift in spectrometer sensitivity, each mass spectrum was normalized to the intensity of the main peak of cyclohexene (m/z = 67) at room temperature (at 25 °C), prior to the start of the temperature ramp. Based on the NIST electron impact mass spectrum database,43 we selected the following mass peaks to observe and quantify a possible reaction product: m/z = 67 amu for cyclohexene, m/z = 78 amu for benzene, m/z = 80 amu for cyclohexadiene, m/z = 98 amu for cyclohexanone, and m/z = 44 amu for CO2. The dominant mass peaks of the products (m/z = 44, 78, and 80) have overlap with fragments of reactant cyclohexene. Since no product formation at room temperature has been reported for the studied reaction, the contributions from cyclohexene fragments (i.e., baselines) were determined from the mass spectrum recorded at 25 °C before the temperature ramp started and after the ramp finished. In the second step of the data analysis, these baselines subtracted from the spectra acquired during the temperature ramp, with the resulting mass peak intensity further referred to as “background-corrected,” as illustrated in Fig. S9 on the example of the evolution of the mass peak intensity of benzene (m/z = 78) for a blank titania support without clusters (referred to as “blank”) and two cluster samples with different size clusters on this same support. Then, average intensities at each temperature were calculated for all m/z from the 40 available data points at each temperature, and the result is shown in Fig. S10 for all titania-based samples, including the ones obtained on the blank titania support (“blank”). The average m/z intensities obtained for the “blank,” which contain contributions arising from the blank support and the exposed internal parts of the reactor, are then subtracted from the data obtained on the cluster-containing sample, thus yielding the sole contribution by the clusters (not shown). Finally, the carbon-based rates of the formation of the individual products are calculated by calibrating the average mass peak intensities to the calibration standards, normalizing to the total number of copper atoms in the sample and multiplying with the number of carbon atoms in the given product [as an example, see for benzene in Fig. 2(b)].
RESULT AND DISCUSSION
AFM images of the ALD deposited thin metal oxide films on the SiO2/Si support are shown in Fig. 1. Roughness analysis was made from 1 × 1 µm2 areas and reveals nanoroughness of Ra = 0.16, 0.23, 0.29, and 0.41 nm for SiO2, Al2O3, TiO2, and SnO2, respectively. Roughness of the SiO2/Si substrate correlates well with the value published by Blasco et al.44 ALD deposited layers were homogenous without aggregation into an isolated island. The nanoroughness is slightly increasing after deposition of metal oxide thin films and copies the SiO2/Si support morphology. AFM imaging does not indicate the presence of crystalline form. Nanograins observed on all samples are typical for oxide growth. We hypothesize that the observed grainy nanomorphology may also contribute to the stabilization of the clusters by providing anchoring sites for the clusters and limit their mobility.
In Figs. 2(b) and 2(c), a carbon-based rate of benzene production per total number of deposited copper atoms (further just rate) is presented for supported cluster-based catalysts. Production of benzene is observed for all the investigated Cu particle sizes and all supports. In the case of Cu clusters on the TiO2 support, benzene production sets on at 200–250 °C (depending on cluster size), and its formation increases gradually with the temperature increase reaching the maximum at 400 °C [Fig. 2(b)]. The exception from the trend is Cu2/TiO2, which do not produce benzene under 350 °C. In general, Cun titania-supported catalysts show a nonlinear dependence on the size of atomic subnanometer clusters with the activity ranked in the order of Cu4 > Cu5 > Cu3 ∼ Cu6 ∼ Cu1 ∼ Cu7 ∼ Cu2. The highest rate of benzene production exhibits a Cu4/TiO2 catalyst with more than four-fold superiority compared to the second most active catalyst Cu5/TiO2. (Note: For simplicity, further in the text, we will refer to Cun/TiO2 catalysts as Cun unless otherwise stated.)
Based on the highest activity of Cu4 on TiO2, we investigated Cu4 clusters deposited on different oxide supports. The rate of benzene production for a copper tetramer on three ALD prepared thin oxide supports TiO2, SnO2, or Al2O3 and on natural oxidized SiO2 is compared in Fig. 2(c). It can be seen that the benzene formation varies with temperature, and it significantly depends on the nature of the substrate. Tetramers on TiO2 and on SiO2 produce benzene starting from 200 °C; the formation of benzene on SnO2 is observed starting at 350 °C, and the Al2O3-supported catalyst performs only at 400 °C. The maximum rate for Cu4/TiO2 at 400 °C is an order of magnitude higher than in the case of Cu4 clusters supported by SnO2, Al2O3, or SiO2. The support effect on the catalytic activity will be discussed in more detail at the end of this section.
To explain the observed results in terms of cluster size-dependent reactivity and selectivity and the effect of support, the performance of Cun/oxide catalysts was compared based on the quantification of the products for individual temperatures in the range of 250–400 °C, as demonstrated in Fig. 3. The series shows the total rate for all the identified products (left column) and selectivity for benzene, cyclohexadiene, and carbon dioxide (right column). The results obtained at 250 °C for Cun/TiO2 catalysts show that all cluster sizes are active except the dimer and trimer [Fig. 3(a)]. Cu4 and Cu6 based catalysts produce benzene with 100% selectivity [Fig. 3(b)], and Cu4 has the highest rate of benzene production of 0.27 atom−1 s−1. On the other hand, Cu1, Cu5, and Cu7 have benzene selectivity below 60% and the benzene rate from 0.1 to 0.2. Interestingly, another valuable product cyclohexadiene is also produced with a rate around 0.10 atom−1 s−1 and a selectivity of up to 40%. Cu7 also produces CO2 at a rate of 0.04 atom−1 s−1 and 15% selectivity, while a small amount of CO2 is detectable for Cu5 at a rate of 0.01 atom−1 s−1. In contrast, Cu4 and Cu6 do not combust to CO2 at all in the whole temperature range; see Fig. S11.
When the temperature rises to 300 °C, the overall behavior of Cun/TiO2 catalysts, in terms of rate and selectivity, remains very similar to previous observations at 250 °C [Figs. 3(c) and 3(d)]. The only change is the sharp increase in the benzene rate for the most active catalyst Cu4/TiO2, rising to 0.92 atom−1 s−1, while its selectivity toward benzene remains at 100% [Figs. 3(c) and 3(d)]. Moreover, at this temperature, Cu3 starts producing benzene at a rate of 0.12 atom−1 s−1 and selectivity 75%, and only the dimer does not produce any detectable products.
Further increasing the temperature to 350 °C and then to 400 °C well reflects the overall tendency toward an increase of the rate of benzene production. The best indicator is the performance of Cu4/TiO2 with the rate jumping to 4 atom−1 s−1 at 400 °C. The cyclohexadiene formation also changes: Hence, Cu7 stop producing cyclohexadiene at 350 °C [Fig. 3(e)] and the same holds for Cu1 and Cu5 at 400 °C [Fig. 3(g)]. At this temperature, Cu5 has the second highest rate of benzene formation with 0.82 atom−1 s−1 and selectivity reaching 97%, with a small amount of CO2 produced. In general, combustion does not play a significant role and the evolution of CO2 production does not follow a consistent trend. In the case of Cu5, CO2 production decreases with temperature whereas increases for Cu2, with Cu2 exhibiting the highest rate of CO2 formation at 400 °C with 0.11 atom−1 s−1 and 40% selectivity. At 400 °C [Fig. 3(h)], Cu1, Cu3, and Cu5 produce benzene with rates of 0.28, 0.51, and 0.82 atom−1 s−1, respectively, and selectivity higher than 90% with a small amount of CO2 (under a rate of 0.02 atom−1 s−1 and selectivity under 7%). On the other hand, the tetramer, hexamer, and heptamer exhibit 100% selectivity for benzene formation at these temperatures. Noteworthy, other products (cyclohexanone) are not detected during measurements.
Our observations about size-dependent reactivity and selectivity fit seemingly well into the picture arising from recent research of sub-nanometer clusters in different reactions.45–48 It was already shown that atomic clusters can exhibit different properties when their size is varied in an atom by atom fashion. Such an effect of size on reactivity can be illustrated on examples of Au,46,49 Pd,45 or Pt clusters.50,51 Indeed, in the smallest size regime, the number of atoms in the cluster and nature and strength of its interaction with the support can shape its geometric structure (planar vs 3D) and alter the number of neighboring atoms (i.e., coordination number), fraction of atoms at the interface with the support, and possibly also the final positioning/anchoring on the support. The support and its affected cluster morphology can also play a crucial role in determining the binding of reactants and the course of the reaction itself.51 In the case of cyclohexane dehydrogenation on Pt and Pt–Pd clusters, ODH occurs on a stepwise elimination of hydrogen and its release in the form of H2, where the efficiency is found to be dependent on the configuration of cyclohexene on the cluster being the rate determining step.52 In the case of Cun clusters on TiO2, the anchoring to support was explained by the formation of specific Cu–O bonds and identifying the interface between the cluster and the support where the decomposition of methanol takes place.53 Such a cluster–surface interaction can influence the electronic structure of the cluster, including charge transfer between Cun clusters and the TiO2 support. Considering the density functional theory (DFT) calculations presented by Tao et al.,53 the extraordinary activity observed for Cu4 on TiO2 may be related to the efficient charge transfer in Cu4/TiO2, leading to a stronger electron-donating ability for activation of reactants in comparison to other clusters and supports interrogated in the present study. In support of this hypothesis, in their another DFT study, Wang et al. revealed that Cu2/TiO2 is the only catalyst in series of Cun/TiO2 (n = 1–14) with no charge transfer observed from the cluster to the support, mainly caused by the strong Cu–Cu sigma bonds in Cu2, not typical for other cluster sizes,54 which may explain the lowest activity observed for the dimer in the present study (see Fig. 3). Additional parameter affecting activity could be the balance between the adsorption and desorption of reactants, intermediates, and final products in the function of cluster size.
Furthermore, Arrhenius plots with a linear fit (see Figs. S13–S15) yield apparent activation energies for benzene as follows: Ea = 48 kJ mol−1 for Cu4/TiO2, 21 kJ mol−1 for Cu5/TiO2, and 41 kJ mol−1 for Cu6/TiO2, not showing a straightforward correlation between effectiveness of benzene production and the apparent activation energy. It implies that, at least in the case of Cu5/TiO2, the reaction bottleneck is not the transition state barrier. In comparison, an activation energy of 89–126 kJ mol−1 was reported for unsupported and supported molybdenia22 and activation energies calculated from the reaction rate for Co27±4 clusters are 73, 66, 71, and 69 kJ mol−1 for MgO, ZnO, TiO2, and Al2O3 supports, respectively, in the abundance of oxygen.25 In comparison, Co27±4 clusters in the presence of trace amount of oxygen show activation energies of 23, 65, and 62 kJ mol−1 for the MgO, ZnO, and Al2O3 supports, respectively.26 Rates on Al2O3 and ZnO supports are similar, but clusters supported on MgO show three times lower activation energy. However, we need to note that in the case of the Co–MgO catalyst, the catalyst transformed into a solid CoOx–MgO solution under reaction conditions. In our experiment, the calculated Ea is lower in comparison to most of the listed ones, which can also contribute to higher selectivity toward dehydrogenated products while preventing combustion.
The results show odd–even oscillations with the atomicity of the clusters in terms of selectivity. Only odd size Cun (Cu4 and Cu6) supported on TiO2 produce benzene with 100% selectivity, in contrast to their neighbors, while Cu5 and Cu7 produce up to 31% cyclohexadiene at lower temperatures. Previous DFT calculation on gas phase copper clusters shows odd–even oscillation for Cun in terms of total magnetic moment (mT), dipole electric moment μ, and HOMO–LUMO energy gap Eg.55 Charged clusters are found more stable when n is odd, and neutral clusters are more stable when n is even. In that case, electron pairing effect plays an important role and the stability of clusters is determined by the presence or absence of unpaired electrons of Cu clusters.55 Recently published studies by Wang et al. and Huang et al. also confirm this fact.54,56 When studying the relative stabilities of Cun clusters on the rutile TiO2 (110) support, clusters with unpaired electrons exhibit stronger interactions with the support as the main reason for odd–even oscillations of cluster stability.54 We believe that odd–even oscillations, especially in charge transfer between the cluster and the support, are probably the main cause of the observed odd–even oscillations in selectivity.
The activity and selectivity observed for the copper tetramer deposited on various supports reveal much different propensities of this cluster on SnO2, Al2O3, and SiO2 than on TiO2 (see Fig. 3). Cu4 on TiO2 has 100% selectivity toward benzene at all temperatures and is the far most active one. The same tetramer on SnO2, Al2O3, and SiO2 is also less selective toward benzene at the expense of combustion. However, when supported on SiO2, at 400 °C, this cluster shows the production of cyclohexadiene with 31% selectivity [see Figs. 3(h) and S12]. It is well established that the support has an essential effect on the geometric and electronic structure of subnanometer clusters, which can influence their reactivity.25,26,47,57,58 DFT calculations for Cu4 in the gas phase predict a planar rhombic structure of tetramers.59 However, when placed on the oxide support, the structure can vary significantly. For example, Cu4 on TiO2 possess a tetrahedron shape,53,54 whereas the tetramer on Al2O3 is planar.60 In addition, charge transfer from Cu4/Al2O3 is lower with respect to TiO2, also resulting in lower stability of the tetramer on the Al2O3 surface.35 To the best of our knowledge, similar data on Cu4 on SiO2 or SnO2 supports are not available.
Based on the obtained results and available literature, there is no doubt that the type of oxide support plays a key role in determining the metal support-interactions and the pathway of the reaction. We believe that concerted changes in the electronic and geometric structure and in charge transfer between Cu4 and various oxide supports are among the main factors that affect the performance of the tetramer, like in the present case of a structure-sensitive reaction.
In the ODH of cyclohexene, a strong effect of the cluster size and the support was observed on both the activity and the selectivity of copper particles made of 1–7 atoms, with a much or entirely suppressed combustion channel, finding the titania-supported copper tetramer possessing the highest activity. The effect of support on catalyst selectivity was highlighted by the copper tetramer, which showed 100% selectivity toward benzene and 31% selectivity toward cyclohexadiene when supported on titania or silica, respectively. Selectivity also showed odd–even oscillations with the atomicity of the titania-supported clusters, with the even numbered Cu4 and Cu6 clusters producing only benzene and no CO2 and the odd numbered Cu5 and Cu7 clusters producing sizable amounts of cyclohexadiene. Based on the available data, it is not possible to unambiguously identify the cause of such different performance with support and cluster size, but the findings show that the subnanometer size range offers the tunability of catalytic performance in an atom by atom fashion to design new catalysts in combination with different supports.
See the supplementary material for Cun mass spectra, time/temperature dependent reactivity data of select products (mass spectrometer signals, temperature-averaged data, representative mass spectra of reactants, and products), and Arrhenius plots.
The authors thank Dr. Michael Pellin for providing the TiO2-coated Si chips and Dr. Alvaro Rodriguez and Dr. Martin Jindra for ellipsometry measurement of the thin oxide layers of Al2O3 and SnO2. The authors acknowledge Argonne National Laboratory for facilitating the use of cluster synthesis and testing equipment for this study. St.V., M.V., J.J., and S.V. acknowledge support from the European Union’s Horizon 2020 research and innovation program under Grant Agreement No. 810310, which corresponds to the J. Heyrovsky Chair project (“ERA Chair at J. Heyrovský Institute of Physical Chemistry AS CR—The institutional approach toward ERA”). The funders had no role in the preparation of the article.
Conflict of Interest
The authors have no conflicts to disclose.
The data that support the findings of this study are available from the corresponding author upon reasonable request.