Photocatalytic ethane conversion on rutile TiO₂(110): identifying the role of the ethyl radical

Abstract

Oxidative dehydrogenation of ethane (C₂H₆, ODHE) is a promising approach to producing ethene (C₂H₄) in the chemical industry. However, the ODHE needs to be operated at a high temperature, and realizing the ODHE under mild conditions is still a big challenge. Herein, using photocatalytic ODHE to obtain C₂H₄ has been achieved successfully on a model rutile(R)-TiO₂(110) surface with high selectivity. Initially, the C₂H₆ reacts with hole trapped O_Ti⁻ centers to produce ethyl radicals , which can be precisely detected by a sensitive TOF method, and then the majority of the radicals spontaneously dehydrogenate into C₂H₄ without another photo-generated hole. In addition, parts of the radicals rebound with diversified surface sites to produce C₂ products via migration along the surface. The mechanistic model built in this work not only advances our knowledge of the C–H bond activation and low temperature C₂H₆ conversion, but also provides new opportunities for realizing the ODHE with high C₂H₄ efficiency under mild conditions.

---

Introduction

Ethene (C₂H₄), as an important basic material for manufacturing diverse consumer products, accounts for about 75% of petrochemical products.^1,2 With the increase of global C₂H₄ demand, environmental and economic issues have become serious problems facing the world because of the energy- and emission-intensive activities for C₂H₄ production (e.g., naphtha steam cracking,^3–5 fluidized catalytic cracking (FCC),^6–8 methanol-to-olefins (MTO)² and Fischer–Tropsch to olefins (FTO)²). The vigorous exploitation of shale gas containing abundant light alkanes has promoted the development of the direct dehydrogenation of ethane (C₂H₆) to C₂H₄. However, compared with the non-oxidative dehydrogenation of the C₂H₆ technique, which is thermodynamically limited with highly endothermic properties, the selective oxidative dehydrogenation (ODH) of C₂H₆ is a promising alternative route for C₂H₄ production due to its autothermal conditions.^1,2,9,11

Although the ODH of C₂H₆ (ODHE) is thermodynamically favored, it is still often conducted under harsh conditions (high temperature and pressure) because of the high chemical stability of the C–H bonds (414 kJ mol⁻¹), resulting in high energy consumption, catalyst deactivation, and over oxidation.^1,2,10 Therefore, various new catalysts (such as boron nitride (BN) based catalysts,^12,13 metal dopants^14,15) and approaches (including CO₂-assisted oxidation,^16,17 chemical looping oxidative dehydrogenation (CL-ODH),^18,19 and so on) were developed to achieve the ODHE process with high selectivity and high efficiency under mild conditions. Among them, photocatalysis, as an emerging technology, can efficiently utilize clean solar energy for the C–H bond activation under mild conditions. Recently, both theoretical and experimental results have shown that TiO₂-based catalysts have potential for C–H bond activation of light alkanes,^20–25 indicating that the photocatalytic ODHE may achieve selective C₂H₄ production under mild conditions. Although both theoretical and experimental studies claimed that alkyl radical intermediates may be formed in the photocatalytic conversion of light alkanes,^20–25 the formation of alkyl radicals is rarely identified due to the sensitivity of the experimental methods, which have confused the fundamental understandings of these reactions. Therefore, illustrating the formation of the ethyl radical in photocatalytic ODHE could play a vital role in understanding the microkinetic mechanisms underlying the reaction.

Herein, we systematically investigated the use of photocatalytic ODHE with rutile-TiO₂(110) using temperature-programmed desorption (TPD), photo stimulated desorption (PSD), and time-of-flight (TOF) methods. The results demonstrate that using photocatalytic ODHE to obtain C₂H₄ can be achieved efficiently on the O atom covered R-TiO₂(110) surface, and the radical intermediate is captured very sensitively. Furthermore, a clear insight into the microkinetic mechanism of the photocatalytic ODHE has been explored.

Experimental

All the TPD experiments were performed with a home-built apparatus, which has been described in detail elsewhere.²⁶ The preparation of the well-ordered R-TiO₂(110) crystal surfaces (Princeton Scientific, 10 mm × 10 mm × 1 mm) was accomplished using cycles of Ar⁺ sputtering and annealing at 850 K in an ultra-high vacuum (UHV). The characterization of the ordering and cleanness of the R-TiO₂(110) surfaces was conducted using low-energy electron diffraction (LEED) and Auger electron spectroscopy (AES), respectively. The density of the oxygen vacancies (O_v) on the surface was measured using H₂O TPD, and was determined to be about 6–7%. The purity of the C₂H₆ and O₂ gases used in the experiment was ≥99.99%. The 355 nm light was produced using a fiber laser (Braze Laser UV), and the pulse time and repetition rate of the laser were ≤15 ps and 400 kHz, respectively. The third harmonic (343 nm) output was produced from a 1030 nm laser (Flare NX laser, Coherent), and the pulse time and repetition rate of the UV laser were 1.5 ns and 200 Hz, respectively. To minimize the increase of surface temperature by the UV irradiation, the maximum power of the laser is 5 mW, corresponding to a flux of 2.1 × 10¹⁶ photons per cm² per s at 355 nm, and 2.0 × 10¹⁶ photons per cm² per s at 343 nm. During the UV light irradiation process, the temperature increase of the surface was less than 2 K. The previously described laser (Braze Laser UV) was used in photocatalytic reactions and the PSD measurements. In order to improve the signal-to-background ratio for sensitively detecting trace signals of the photo-desorbed products, the Flare NX laser (Coherent) was used in the TOF measurements.

For the PSD measurements, the time resolution was set to 0.5 s. If we assume that the ionization of the background residual gases under vacuum will produce 1 × 10⁶ counts per second (cps) at m/z = 29, and the ionization of the radical product will produce 1 × 10⁵ cps for C₂H₅⁺ (m/z = 29), and the signal-to-noise ratio (SNR) the PSD method used here was 10 : 1. If the fluctuation of the signal of m/z = 29 from the background was about 10%, it was hard to determine whether the radical had been produced. However, when the 200 Hz light was used for the experiment with a TOF analyzer, the moment that the laser light arrived at the surface was set as time zero, and the C₂H₅⁺ signal arriving at the detector could be precisely counted in an exact relationship to their arrival time in each pulse. The time interval between every laser shot was 5 ms, and the time resolution was set to 256 ns for the TOF measurements. The background signal of the C₂H₅⁺ signal was produced randomly, and then the background signal collected in each frequency bin was (1 × 10⁶ ÷ 200) ÷ (5 ms ÷ 256 ns) ≈ 0.256 count per 256 ns. However, the production of radical was not random, and it was produced at the time scale of 0.1 ms (see below). Correspondingly, the C₂H₅⁺ product signal collected in each frequency bin is (1 × 10⁵ ÷ 200) ÷ (0.1 ms ÷ 256 ns) ≈ 1.28 count per 256 ns, and the SNR of the TOF method was 1 : 5. As a result, the sensitivity of the TOF method was much higher than that of the PSD method (10 : 1).

Results

The TPD results of the C₂H₆ conversion into C₂H₄

Fig. 1 shows the typical TPD spectra of the mass-to-charge ratios (m/z) of 15 (CH₃⁺), 18 (H₂O⁺), 26 (C₂H₂⁺), 27 (C₂H₃⁺), 28 (C₂H₄⁺ and CO⁺), 29 (C₂H₅⁺ and CHO⁺), 30 (C₂H₆⁺), 31 (CH₂OH⁺) and 43 (CH₃CO⁺) collected on the oxidized R-TiO₂(110) surfaces after adsorbing 0.28 ML (1 ML = 5.2 × 10¹⁴ molecules per cm²) of C₂H₆ followed by 355 nm irradiation for 0 (black lines) and 10 min (red lines). The oxidized R-TiO₂(110) surfaces were prepared by exposing the reduced surfaces to 200 L of O₂ at 300 K.^25,27,28 After surface oxidation, the bridging oxygen vacancies (O_v) will be healed, leaving oxygen atoms on the five coordinated Ti⁴⁺ sites (Ti_5c, O_Ti).²⁵ Before irradiation, only one desorption peak at 129 K appeared in the TPD traces of m/z = 15, 26, 27, 28, 29, 30, and 31, which was attributed to the desorption of the molecular C₂H₆ on the Ti_5c sites (C₂H_6(Ti)).²⁹ No signals from thermocatalytic products suggested that the oxidized R-TiO₂(110) surface was not thermally active for the C–H bond activation of C₂H₆.

Fig. 1 Left: Typical TPD spectra acquired at m/z = 15 (CH₃⁺), 18 (H₂O⁺), 26 (C₂H₂⁺), 27 (C₂H₃⁺), 28 (C₂H₄⁺ and CO⁺), 29 (C₂H₅⁺ and CHO⁺), 30 (C₂H₆⁺), 31 (CH₂OH⁺), and 43 (CH₃CO⁺) after irradiating the 0.28 ML C₂H₆ covered oxidized R-TiO₂(110) surfaces for 0 min (black lines) and 10 min (red lines) by 355 nm at 75 K, respectively. Right: The TPD spectra at the temperature range >140 K are highlighted. The oxidized R-TiO₂(110) surfaces were prepared by exposing the reduced surfaces to 200 L of O₂ at 300 K. The photon flux of 355 nm light was 2.1 × 10¹⁶ photons per cm² per s.

After 355 nm irradiation, a new desorption peak at 330 K was observed in the TPD spectra of m/z = 18 (Fig. 1a), which was contributed by the molecular H₂O desorption on the Ti_5c sites (H₂O_Ti) or recombinational H₂O desorption from the terminal OH groups on the Ti_5c sites (OH_Ti).³⁰ The H atoms of the H₂O product could only be from C₂H_6(Ti), demonstrating that the photocatalytic dehydrogenation of C₂H_6(Ti) occurred on the O_Ti atom covered R-TiO₂(110) surface. Conversely, the reduced R-TiO₂(110) showed no photoactivity for the ODHE process (see Fig. S1, ESI†). Based on our previous results of the photocatalytic oxidative dehydrogenation of propane (C₃H₈, ODHP) on R-TiO₂(110),²⁵ the hole trapped O_Ti⁻ centers rather than hole trapped bridging oxygen atoms (O_b⁻) were the active species for the initial C–H bond activation of C₂H₆, leading to the formation of H₂O_Ti and OH_Ti (Fig. 1a).

The H₂O_Ti formation was accompanied by several desorption features of carbon-containing products, which were observed at 168 K, 365 K, 423 K, 580 K, and 585 K (Fig. 1b). The broad peak (400–700 K) in the TPD traces of m/z = 15, 26, 27, 28, 29 and 30 was due to the desorption of C₂H₆ from the copper blocks, which were used for the mounting tantalum sample holder (Fig. S2, ESI†)The relative intensity ratio of the 365 K peak in the TPD traces of m/z = 15, 29, and 43 were calculated to be 0.80 : 1 : 0 : 14, respectively, which was very close to that of acetaldehyde (CH₃CHO) measured by mass spectrometry (Fig. S3, ESI†). Therefore, this peak could be attributed to the formation of CH₃CHO on the Ti_5c sites.³¹ The tiny peak at 585 K (m/z = 31) was likely to be due to the formation of ethanol (C₂H₅OH).³² In addition, as shown in Fig. 1b, the relative intensities of the 168 K, 423 K, and 580 K peaks in the TPD traces of m/z = 26 and 27 were calculated to be 0.87 : 1 (168 K), 0.89 : 1 (423 K) and 0.88 : 1 (580 K), respectively, which were very close to that of the C₂H₄ sample (Fig. S4, ESI†), and very different from that of the alkanes and alkenes (C_nH_2n (3 ≤ n ≤ 10) and C_nH_2n+2 (2 ≤ n ≤ 10)) found in the NIST database. Therefore, all three peaks could be assigned to the C₂H₄ product, illustrating that the photocatalytic ODHE process to produce C₂H₄ could be realized on oxidized R-TiO₂(110).

Although the structure of C₂H₆ was simpler than that of C₃H₈, and the initial C–H bond activation process for these two molecules was nearly the same, the pathways for C₂H₄ production from photocatalytic ODHE on oxidized R-TiO₂(110) were more complicated than that of the photocatalytic ODHP.²⁵ For the photocatalytic ODHP on R-TiO₂(110), the majority of the propylene (C₃H₆) product can be formed efficiently under UV irradiation at 100 K. Only a tiny amount of the C₃H₆ product is formed at 340 K via the thermal dehydrogenation of the C₃H₇⁻ groups on the Ti_5c sites (C₃H_7(Ti)⁻), whereas, no oxygenated carbon products were produced.²⁵ However, for methane (CH₄) dehydrogenation via either thermocatalysis or photocatalysis,^20,33 the radicals are thought to be suspended above the TiO₂ surface or to enter directly into the gas phase, showing a very high mobility. Therefore, once the radical is produced via photocatalytic ODHE on R-TiO₂(110), it may also migrate on the surface or enter directly into the gas phase, in a similar way to the radical from CH₄ conversion,^20,33 leading to the complicated reaction pathways in the photocatalytic ODHE process.

Evidence of radical formation and the fate of radical

To confirm the formation of the radical intermediate in photocatalytic ODHE on R-TiO₂(110), the PSD signals were collected at m/z = 26, 27, 28, 29, and 30 from the 0.28 ML C₂H₆ covered oxidized R-TiO₂(110) surface during the UV irradiation (Fig. 2). Upon irradiation, no obvious PSD signal was detected at m/z = 30, indicating that no photodesorption of C₂H₆ had occurred. Conversely, sharp increases of the desorption signals at m/z = 26, 27 and 28 were detected immediately when the UV light was on. The relative intensity of the PSD signals at m/z = 26 and 27 was 0.89 : 1, suggesting that the signals were contributed by photo-desorbed C₂H₄ molecules. In addition, a tiny PSD signal also appeared at m/z = 29, which may come from two sources. The first was due to the fragmentation of the C₂H₄ product at m/z = 29 (Fig. S4, ESI†). The other one was the radical product. However, due to the small PSD signal at m/z = 29, it was hard to determine whether a radical was formed.

Fig. 2 The PSD spectra acquired at m/z = 26 (C₂H₂⁺), 27 (C₂H₃⁺), 28 (C₂H₄⁺), 29 (C₂H₅⁺) and 30 (C₂H₆⁺) on the 0.28 ML C₂H₆ covered oxidized R-TiO₂(110) surfaces. The purple dashed line represents the moment when the light is turned on (t = 20 s). The photon flux of the 355 nm light is 2.4 × 10¹⁶ photons per cm² per s.

Subsequently, the TOF method, which can enhance the detection sensitivity significantly by improving the SNR,³⁴ was used to monitor the desorbed products from the photocatalytic ODHE on R-TiO₂(110) during the irradiation. As shown in Fig. 3, the TOF signals at m/z = 27 (C₂H₃⁺), 29 (C₂H₅⁺) and 30 (C₂H₆⁺) were collected. Obvious peaks appeared in the TOF spectra of m/z = 27 and 29. According to the result shown in Fig. 2, the TOF peak at m/z = 27 was due to the desorption of C₂H₄. Interestingly, the relative intensities of the TOF signal at m/z = 27 and 29 were about 3 : 1, which was much smaller than that of the C₂H₄ sample (Fig. S4, ESI†). However, no discernible TOF signal at m/z = 30 suggested that no photodesorption of C₂H₆ had occurred, and this was consistent with the results in Fig. 2. As a result, the large TOF signal at m/z = 29 could only be from the desorption of radical upon irradiation, illustrating that the initial photocatalytic C–H activation of C₂H₆ on R-TiO₂(110) produced the radical.

Fig. 3 The TOF signals of C₂H₄ collected at m/z = 27 (C₂H₃⁺), 29 (C₂H₅⁺) and 30 (C₂H₆⁺) as a function of the flight time when irradiating the 0.28 ML C₂H₆ covered oxidized R-TiO₂(110) surfaces at 343 nm. The photon flux of the 343 nm light is 2.0 × 10¹⁶ photons per cm² per s.

Thus, after the O_Ti²⁻ centers trap the photogenerated holes to form excited O_Ti⁻ centers:

TiO₂ + hv → h⁺ + e⁻ (1)

O_Ti²⁻ + h⁺ → [O_Ti⁻]* (2)

the separated electrons are left on the R-TiO₂(110). Then, because the C₂H₆ was only weakly adsorbed on the surface, the direct hole transfer from R-TiO₂(110) to C₂H₆ was nearly impossible, and the reaction was most likely to occur via the abstraction of H atoms from C₂H₆ by excited O_Ti⁻ centers to produce the radical:

When the radical was produced, it may further dehydrogenate spontaneously into C₂H₄ and H atoms on the O_Ti²⁻/O_b²⁻ sites or be ejected into the gas phase:²¹

Here, it is not certain whether the second C–H bond cleavage needs another hole. The detection of the radical demonstrates that the interaction between the radical and the R-TiO₂(110) was very weak. Therefore, the migration of the radical on the surface may occur easily. Once the radicals migrate on the surface, the C₂H_5(Ti)⁻ groups, ethoxy groups (C₂H₅O_b⁻ and C₂H₅O_Ti⁻) and C₂H₅OH_Ti may be produced via the rebounding of the radicals to the Ti_5c, O_b²⁻, O_Ti²⁻ and OH_Ti⁻ groups:³⁵

Among them, the C₂H_5(Ti)⁻ groups would further dehydrogenate into C₂H₄via a similar thermocatalytic pathway used for the thermocatalytic C₃H₆ formation,²⁵ giving a C₂H₄ desorption peak at 423 K. The C₂H₅O_Ti⁻/C₂H₅OH_Ti will decompose into CH₃CHO_Ti easily upon irradiation:^36,37

C₂H₅O_Ti⁻ + h⁺ → CH₃CHO_Ti + H⁺ (at O_Ti²⁻/O_b²⁻) + e⁻ (9)

C₂H₅OH + h⁺ → CH₃CHO_Ti + 2H⁺ (at O_Ti²⁻/O_b²⁻) + 2 e⁻ (10)

During the reaction steps, most of the steps were hole induced half-reactions, which will leave electrons on the surface. It seems that the photocatalytic ODHE on R-TiO₂(110) will produce excess electrons on the surface. In fact, when the reduced R-TiO₂(110) surface is oxidized by O₂ at room temperature to form the O_Ti covered surface, the surface O_b vacancies will be healed, the excess electrons of R-TiO₂(110) contributed by the vacancies (O_v²⁻) and Ti interstitials (Ti³⁺) under the surface or in the bulk, will be trapped by the dissociated O_Ti atoms to form O_Ti²⁻.^25,27,28 Upon irradiation, after the electron–hole separation, the holes will be trapped at the O_b²⁻ and O_Ti²⁻, forming O_b⁻ and O_Ti⁻. The photogenerated electrons will be trapped by the vacancies and Ti interstitials that gave the electrons to the surface of the O_Ti atoms before. Similarly, in the later reactions, even electrons are left behind, and most of them are probably trapped by vacancies and the Ti interstitials. As a result, although the excess electrons of R-TiO₂(110) did not transfer to C₂H₆ and C₂H₅O_Ti⁻/C₂H₅OH_Ti during the photocatalytic ODHE process, they are trapped by O_Ti atoms initially to form O_Ti²⁻, and most of them go back to the surface after the reactions because the formation of low temperature C₂H₄ and CH₃CHO are whole reactions, not half reactions. The overall reaction via photocatalysis follows eqn (11):

C₂H₆ + O_Ti + hv → CH₃CHO_Ti or C₂H₄ + H₂O_Ti (11)

In addition, the minor reaction pathways of reactions in eqn (5) and (6) may produce excess electrons on the surface.

Furthermore, although the C₂H₅O_b⁻ groups have little photo reactivity,³⁶ the C₂H₅O_b⁻ groups could dissociate to C₂H₄ with a small amount of C₂H₅OH product during the TPD process,³⁸ which was consistent with our TPD result for the C₂H₅OH desorption on the O_v sites of R-TiO₂(110) (Fig. S5, ESI†). Therefore, the C₂H₄ formation at 580 K was due to the thermocatalytic dehydrogenation of the C₂H₅O_b⁻ groups, and the tiny peak at 585 K (m/z = 31) may be assigned to the recombinational C₂H₅OH desorption from the C₂H₅O_b⁻ groups and dissociated protons (H⁺) during the TPD process.

To evaluate the importance of the C₂H₄ production via photocatalytic ODHE on R-TiO₂(110), the formation of carbon containing products and H₂O were monitored using the TPD traces of m/z = 18, 27, 29, and 31 collected on the 0.28 ML C₂H₆ covered oxidized R-TiO₂(110) surfaces as a function of irradiation time (Fig. 4). As the irradiation time increased, the signals of the thermocatalytic products (the 423 K (C₂H₄), 580 K (C₂H₄), and 585 K (C₂H₅OH) peaks) increased very fast and reached plateaus after approximately 60 s irradiation (the green traces). However, the signals of the 168 K (C₂H₄), 330 K (H₂O), and 365 K peaks (CH₃CHO) increased slowly and almost reached plateaus after 600 s irradiation. As discussed, previously, the O_Ti is involved in the formation of 168 K (C₂H₄), 330 K (H₂O), and 365 K (CH₃CHO) products. Here, the coverage of O_Ti atoms on the surface was only about 0.06–0.07 ML, which was strongly dependent on the concentration of the O_v sites.^27,28 In contrast, the coverages of Ti_5c and O_b sites on the oxidized R-TiO₂(110) were nearly 1.0 ML. As a result, the possibility for the C₂H₅ moieties bonding to the Ti_5c and O_b sites to produce C₂H_5(Ti)⁻ and C₂H₅O_b⁻ groups would be much higher than that for the C₂H₅ moieties bonding to O_Ti²⁻/OH_Ti⁻ to produce C₂H₅O_Ti⁻/C₂H₅OH_Ti, resulting in the formation of C₂H_5(Ti)⁻ and C₂H₅O_b⁻ groups much faster than the C₂H₅O_Ti⁻/C₂H₅OH_Ti.

Fig. 4 The TPD spectra acquired at m/z = 18 (H₂O⁺) (a), 27 (C₂H₃⁺) (b), 29 (C₂H₅⁺) (c), and 31 (CH₃O⁺) (d) on the 0.28 ML C₂H₆ covered oxidized R-TiO₂(110) surfaces as a function of irradiation time. The desorption peaks of the carbon containing products are been highlighted in the inserts of (b–d).

From Fig. 4, the yields of H₂O and carbon containing products (C₂H₄, CH₃CHO, and C₂H₅OH) were derived and are plotted in Fig. 5. The total yield of C₂H₄ contained the 168 K, 423 K, and 580 K peaks. With an increasing irradiation time, the difference between the yield of H₂O and carbon containing products became larger and larger. At 20 min irradiation, about 0.041 ML of H₂O was produced, and the yields of C₂H₅OH, CH₃CHO, and C₂H₄ are about 0.004 ML, 0.0075 ML, and 0.02 ML, respectively. Combining the results from Fig. 2 and 3, it was seen that the big difference between the yields of H₂O and carbon containing products was due to the photo-desorbed C₂H₄ and radicals. Therefore, the C₂H₄ product (including the 168 K, 423 K, and 580 K peaks) was the main product in photocatalytic ODHE on R-TiO₂(110).

Fig. 5 The yields of H₂O (red triangles) and carbon containing products (C₂H₅OH (blue circles), CH₃CHO (blue squares), and C₂H₄ (blue stars)) in photocatalytic ODHE on oxidized R-TiO₂(110) as a function of irradiation time, derived from Fig. 4. All the plotted lines are only to guide the eye.

Microscopic kinetics of the C₂H₄ formation

Similar to the case of the other hydrocarbons (C₃H₈, ethylbenzene (C₆H₅C₂H₅, EB)),^25,39 the initial C–H bond activation of C₂H₆ could be induced by hole trapped O_Ti⁻. However, the second C–H bond cleavage in the C₂H₆ conversion on R-TiO₂(110) was still not clear. On the one hand, the radicals may further dehydrogenate into C₂H₄ directly,²¹ suggesting that the whole process only needs one hole. On the other hand, because the radicals can decay into C₂H_5(Ti)⁻ groups via the de-excitation process, the C₂H_5(Ti)⁻ group may also trap another hole to form a radical again for C₂H₄ production. For this process, a total of two holes are consumed for the stepwise C₂H₄ production. Although the involvement of C₂H_5(Ti)⁻ in photocatalytic ODHE does not affect C₂H₄ production macroscopically, the microkinetic mechanism is very distinct.

To confirm whether C₂H_5(Ti)⁻ is involved in photocatalytic ODHE into C₂H₄ on R-TiO₂(110), the PSD signals of C₂H₄ at m/z = 27 were collected from the 0.28 ML C₂H₆ covered oxidized R-TiO₂(110) surfaces as a function of the laser power. As the laser power increased, the intensity of the C₂H₄ PSD signal (the initial data point in each photodesorption experiment) increased significantly. More importantly, the intensity of the PSD signal of the C₂H₄ scaled linearly with the square root of the photon flux (F_hν^1/2) (see the inset of Fig. 6). According to the results of previous work on O₂ photodesorption on R-TiO₂(110)^40,41 and C₂H₅OH photodecomposition on R-TiO₂(110),⁴² such a linear relationship illustrates that the photocatalytic ODHE to C₂H₄ on oxidized R-TiO₂(110) was governed by the second-order electron–hole (h⁺/e⁻) pair recombination kinetics, and only one hole (or one photon) was involved in the complicated process. Furthermore, the pathway of C₂H₄ production from the photocatalytic ODHE on oxidized R-TiO₂(110) with the involvement of C₂H_5(Ti)⁻ could be ruled out, because it needs two holes (or photons). Therefore, the C₂H₄ formation from photocatalytic ODHE on oxidized R-TiO₂(110) occurs in a stepwise manner, in which the C₂H₆ first undergoes the initial C–H bond cleavage to form radicals with the help of hole trapped O_Ti⁻ centers, and is then followed by further spontaneous dehydrogenation of the radicals into C₂H₄ without the involvement of an extra photon or hole. During the C₂H₄ formation process, the initial C–H bond activation is the rate-limiting step.

Fig. 6 The PSD signals of C₂H₄ collected from the 0.28 ML C₂H₆ covered oxidized R-TiO₂(110) surfaces as a function of the laser irradiation power. The intensity of the PSD signals (the initial point when the UV light is turned on) of C₂H₄ has a linear relationship with the square root of the incident light flux (F_hν^1/2), as shown in the inset of this figure.

In addition, as shown in Fig. S6 (ESI),† C₂H₄ can be photo-desorbed on the C₂H₄ covered R-TiO₂(110), indicating that the desorption of C₂H₄ was induced by the photogenerated charge carriers (electron or hole). If the C₂H₄ product from the photocatalytic ODHE prefers to adsorb on the R-TiO₂(110) surface first, and then be photo-desorbed from the surface, at least two photons are consumed for the C₂H₄ formation and desorption processes. However, the whole process was accomplished by one hole (or one photon), thus, once C₂H₄ was formed by further C–H bond cleavage of the radicals, it was preferentially ejected directly into the vacuum rather than being adsorbed on the surface followed by photoinduced desorption. Therefore, only a tiny signal for the C₂H_4(Ti) desorption can be observed during the TPD process.

Discussion

As shown previously, both reduced and oxidized R-TiO₂(110) surfaces do not show thermocatalytic reactivities for C₂H₆ activation. This was very different from the results of the C–H activation of the light alkanes on the PdO(101), RuO₂(110), and IrO₂(110) surfaces,^43–46 in which alkanes adsorb on these surfaces stably by forming strongly adsorbed σ-complex species as the precursor, leading to a weakening of the C–H bond in the alkanes. Similarly, Yue and co-workers proposed that the initial C–H bond activation of C₂H₆ on the M/TiO₂ (M = Pd and Cu) surfaces was realized via the interaction of C₂H₆ with the surface sites to produce surface C₂H₅⁻ groups, and then the C₂H₅⁻ groups trapped the photogenerated holes to form radicals, which converted rapidly into the C₂H₄ product.^22,24 However, although the R-TiO₂(110) has the same surface structure with RuO₂(110) and IrO₂(110), the weak interaction between the C₂H₆ and Ti_5c sites of R-TiO₂(110) inhibit the formation of the C₂H₆ σ-complex. As a result, the H atom abstraction from C₂H₆ by the O_b and O_Ti atoms (or O_b²⁻ and O_Ti²⁻) following the Langmuir–Hinshelwood (L–H) mechanism is difficult. Namely, the initial C–H bond activation of C₂H₆ on R-TiO₂(110) seems to occur ineffectively at the ground state via thermocatalysis.^21,22,24

However, when the O_b²⁻ and O_Ti²⁻ centers trap photogenerated holes, the nucleophilic O_b²⁻ and O_Ti²⁻ convert into electrophilic O_b⁻ and O_Ti⁻ centers,^47,48 which have a stronger ability than O_b²⁻ and O_Ti²⁻ to abstract the H atoms of the small alkanes.^20,21,25 Correspondingly, the study of the photocatalytic EB dehydrogenation on R-TiO₂(110)³⁹ demonstrated that both the O_b⁻ and O_Ti⁻ centers produced by trapping the holes can activate the α-C–H bond of the side chain alkyl groups of EB. In addition, theoretical works also suggest that the O_b⁻ centers formed upon the UV irradiation play a vital role in the C–H bond activation of CH₄ and C₂H₆ on R-TiO₂(110).^20,21 Unfortunately, no product signal of photocatalytic ODHE was detected on reduced R-TiO₂(110) (Fig. S1, ESI†) under 355 nm irradiation, indicating that the O_b⁻ center produced with the 355 nm irradiation finds it difficult to activate the inert C–H bond of C₂H₆ under the current conditions.

In contrast, the EB can be regarded as one H atom of the C₂H₆ molecule substituted by a phenyl group (C₆H₅), in which the C–H bond of the ethyl group can be activated efficiently by the hole trapped O_b⁻ center.³⁹ The difference in the initial C–H bond activation of C₂H₆ and EB by the O_b⁻ center on R-TiO₂(110) may be because of two possible reasons. Firstly, compared with the H atom, the phenyl group as an electron withdrawing group will decrease the electron density of the α-C in the C₂H₅ group via a σ–π hyperconjugation, resulting in weakening of the α-C–H bond. As a result, the α-C–H bond will be activated more easily than the C–H bond of C₂H₆. Secondly, the desorption temperature of EB on R-TiO₂(110)³⁹ was much higher than that of C₂H₆ by about 120 K, indicating that the former has a stronger interaction between the aromatic ring and the surface, which may be more beneficial for energy and charge transfer between adsorbates with the surface, leading to the second C–H bond cleavage of EB.

In addition, for C₃H₈, one H atom of the C₂H₆ molecule substituted by a methyl group (CH₃–), contained two types of C–H bonds (1° and 2°), and its structure was more complicated than that of C₂H₆. However, the C₃H₆ was produced with a high selectivity via photocatalytic ODHP on R-TiO₂(110), and no oxygenates were produced,²⁵ indicating that the reaction intermediates in the photocatalytic ODHP did not show high mobility on the surface. This was most likely due to the different interaction strengths between different intermediates with R-TiO₂(110). According to the desorption temperature of C₂H₆ (Fig. 1) and C₃H₈,^25,29 both C₂H₆ and C₃H₈ were weakly bound to the surface, causing both C₂H₆ and C₃H₈ to easily migrate on the surface. As a result, the initial C–H bond activation of C₂H₆ on R-TiO₂(110) was more likely to follow the O_Ti atom mediated Eley–Rideal (E–R) mechanism,^1,25 forming a movable radical by the abstraction of an H atom from C₂H₆ by an excited O_Ti⁻ center, which may significantly affect the selectivity of the products. The interaction between the radicals, which were worse electron donors, with the Ti_5c sites should be weaker than that of the possible radicals produced in the photocatalytic ODHP on R-TiO₂(110), which tended to form an allyl σ–p hyperconjugation configuration (CH₃CH˙CH₃) with a stronger electron-donating ability.^25,49 Then, the radicals may migrate on the surface more easily, resulting in the formation of additional byproducts via the diffusion and rebounding processes.

Similarly, previous research on the photooxidation of tert-butanol and ketones on R-TiO₂(110) also observed radical ejection,^50–52 which was associated with hole-induced chemistry.^{20,21,25,50,52} According to the work on radical formation from ketone photooxidation on R-TiO₂(110),⁵² two dissociation channels (“fast” and “slow” channels) of the radical desorption were detected. The ‘‘fast’’ radical production was attributed to the prompt dissociation of an internally “hot’’ acetone–oxygen complex (intermediates at the excited state), and the acetone–oxygen complex weakly coupled to the surface. However, the ‘‘slow’’ radical production was assigned to the dissociation of a relaxed acetone–oxygen complex formed via internal vibrational redistribution (IVR), which consumed the available energy for the C–C bond cleavage. In the case of the radical ejection from 2-butanone photooxidation on R-TiO₂(110), the ‘‘slow’’ channel dominated.⁵⁰ As a result, no obvious C₂H₄ product obtained from radical dehydrogenation was detected.⁵⁰

However, for the photocatalytic ODHE on the R-TiO₂ (110) surface, C₂H₆ was also weakly adsorbed on the surface and its structure was very simple without a π-conjugated system, and the IVR process will not occur to make the “hot” radical relax efficiently, thus resulting in the further dehydrogenation of the excited radical into C₂H₄. Compared with these results,^50–52 it was not too difficult to conclude that the IVR process can affect the energy relaxation in excited molecules or ions on the R-TiO₂(110) surface, which can further affect the bond breaking and product formation. The smaller molecules (such as C₂H₆) may inhibit the IVR process in photocatalytic reactions due to having fewer vibrational energy levels than complicated molecules, leading to the high efficiency of bond breaking.

Due to the formation of weakly bonded radical intermediates, it is reasonable that oxygen-containing species are formed via the rebounding between the radicals and the surface O atoms (O_b²⁻ and O_Ti²⁻). However, based on previous research about ODHE over vanadium oxides,^53–55 terminal M=O species (terminal metal oxo, V=O) are the active sites for ODHE, in which either the direct insertion of the C–H bond across the M=O bond resulted in C₂H₅O group formation or C–H bond activation by H abstraction to form M–OH and a transient alkyl radical product may occur. If the C₂H₅O_b⁻ formation occurs via the O_b²⁻ insertion pathway, the C₂H₄ product formed at 580 K by the dehydrogenation of C₂H₅O_b⁻ groups should be observed on both the reduced and oxidized R-TiO₂(110) surfaces. However, the formation of the C₂H₅O_b⁻ groups was only detected on the oxidized surface, suggesting that the C₂H₅O_b⁻ groups were produced via the recombination of the radical and O_b²⁻ rather than the O_b²⁻ insertion pathway. In addition, the direct insertion generally has a high barrier,^54,55 which is difficult for C₂H₆ activation on TiO₂ following the L–H mechanism due to weak adsorption energy. Despite all this, the existence of a direct heterolytic insertion of O_Ti⁻ to C₂H₆ for C₂H₅O_Ti⁻/C₂H₅OH_Ti formation cannot be completely ruled out.

Interestingly, although the initial C–H bond cleavage of EB on R-TiO₂(110) occurred more easily than that of C₂H₆, the possible radical intermediate produced from the initial C–H bond cleavage of EB under UV irradiation preferred to decay to C₈H₉⁻ rather than further dehydrogenate into styrene directly, leading to a low yield of low temperature styrene production.³⁹ In contrast, in the case of C₂H₆ and C₃H₈ activation²⁵ on R-TiO₂(110), the intermediates mainly dehydrogenated into C₂H₄ and C₃H₆ spontaneously, whereas only tiny radicals decay to alkyl groups adsorbed on the Ti_5c sites. This suggested that the photon energy for the C–H bond activation of small alkanes into alkenes via TiO₂ photocatalysis may be utilized more efficiently than that of aromatic EB.^39,56 Furthermore, due to phenyl group substitution, the rate-determining step of the photocatalytic dehydrogenation of hydrocarbons into corresponding alkenes shifts from the initial C–H bond activation to the second further dehydrogenation.^25,39 This result may be more evidence that the IVR process in the larger molecule reduces the available energy for bond breaking. In order to overcome the consumption of available energy by the IVR process, excitation with higher energy photons may be a feasible way. Referring to the results of the recent photocatalytic conversion of EB into styrene on R-TiO₂(100),⁵⁶ the efficiency of the initial α-C–H bond activation is nearly the same at 257 nm and 343 nm, whereas the rate of the β-C–H bond cleavage was strongly enhanced with the photon energy. In contrast, for C₂H₆ and C₃H₈,^21,25 once the initial C–H bond cleavage was activated by the hole derived from the 355 nm photoexcitation with a lower photon energy, the second dehydrogenation was still accomplished quite easily. However, the mobility of the intermediates determined the complexity of the reactions.

Conclusion

In summary, photocatalytic ODHE on the model R-TiO₂(110) surface has been systematically investigated to determine the mechanism of the process. The C₂H₄ formation for the photocatalytic ODHE on the oxidized surface can be achieved via a stepwise manner at 75 K, in which the radical intermediate is detected directly. The hole trapped O_Ti⁻ centers play a crucial role in the initial C–H bond activation of C₂H₆, and only one hole is involved in the cleavage of the two C–H bonds to produce C₂H₄. This result not only illustrates that photocatalysis is very suitable for the inert C–H bond activation of small alkanes, but also provides a novel mechanistic insight into photocatalytic ODHE, which offers new opportunities for the development of novel ODHE pathways with high C₂H₄ selectivity under mild conditions.

Data availability

The data supporting this study is available within the main text and the associated ESI.†

Author contributions

YL and YZ conducted the TPD experiments and analysis. XC and TW helped to conduct the TPD experiments, and to revise the manuscript. FL and QG conceived the idea, designed the experiment, analyzed the data, and co-wrote the paper. XY supervised the research. All the authors discussed the results and commented on the paper.

Conflicts of interest

The authors declare no competing financial interests.

Acknowledgements

This work was supported by the National Key R&D Program of China (Grant No. 2022YFA1503102, 2018YFE0203002), the National Natural Science Foundation of China (NSFC Center for Chemical Dynamics, Grant No. 22173041, 22103033, 22103031, 22173042, 21973037), the Strategic Priority Research Program of Chinese Academy of Sciences (Grant No. XDB17000000), the Shenzhen Science and Technology Innovation Committee (Grant No. 20220814164755002, ZDSYS20200421111001787), the Guangdong Innovative & Entrepreneurial Research Team Program (Grant No. 2019ZT08L455, 2019JC01X091), the International Partnership Program of Chinese Academy of Sciences (Grant No. 121421KYSB20170012), and the Innovation Program for Quantum Science and Technology (Grant No. 2021ZD0303304).

References

1. S. Najari, S. Saeidi, P. Concepcion, D. D. Dionysiou, S. K. Bhargava, A. F. Lee and K. Wilson, Chem. Soc. Rev., 2021, 50, 4564–4605.
2. Y. Dai, X. Gao, Q. Wang, X. Wan, C. Zhou and Y. Yang, Chem. Soc. Rev., 2021, 50, 5590–5630.
3. R. Gudgila and C. A. Leclerc, Ind. Eng. Chem. Res., 2011, 50, 8438–8443.
4. P. Sun, G. Siddiqi, W. C. Vining, M. Chi and A. T. Bell, J. Catal., 2011, 282, 165–174.
5. J.-P. Lange, R. Schoonebeek, P. Mercera and F. Van Breukelen, Appl. Catal., A, 2005, 283, 243–253.
6. V. Zacharopoulou and A. A. Lemonidou, Catalysts, 2017, 8, 2.
7. S. Najari, S. S. Hosseini, M. Omidkhah and N. R. Tan, RSC Adv., 2015, 5, 47199–47215.
8. A. Corma, E. Corresa, Y. Mathieu, L. Sauvanaud, S. Al-Bogami, M. Al-Ghrami and A. Bourane, Catal. Sci. Technol., 2017, 7, 12–46.
9. M. Sun, J. Zhang, P. Putaj, V. Caps, F. d. r. Lefebvre, J. Pelletier and J.-M. Basset, Chem. Rev., 2014, 114, 981–1019.
10. J. T. Grant, J. M. Venegas, W. P. McDermott and I. Hermans, Chem. Rev., 2017, 118, 2769–2815.
11. C. A. Gärtner, A. C. van Veen and J. A. Lercher, ChemCatChem, 2013, 5, 3196–3217.
12. Y. Zhou, J. Lin, L. Li, X. Pan, X. Sun and X. Wang, J. Catal., 2018, 365, 14–23.
13. Y. Honda, A. Takagaki, R. Kikuchi and S. T. Oyama, Chem. Lett., 2018, 47, 1090–1093.
14. T.-K. Cheung and B. C. Gates, J. Catal., 1997, 168, 522–531.
15. S. Wang, K. Murata, T. Hayakawa, S. Hamakawa and K. Suzuki, Chem. Commun., 1999, 103–104.
16. E. Gomez, B. Yan, S. Kattel and J. G. Chen, Nat. Rev. Chem, 2019, 3, 638–649.
17. S. Deng, S. Li, H. Li and Y. Zhang, Ind. Eng. Chem. Res., 2009, 48, 7561–7566.
18. S. Yusuf, L. Neal, V. Haribal, M. Baldwin, H. H. Lamb and F. Li, Appl. Catal., B, 2018, 232, 77–85.
19. Y. Gao, F. Haeri, F. He and F. Li, ACS Catal., 2018, 8, 1757–1766.
20. M. Zhou and H. Wang, JACS Au, 2022, 2, 188–196.
21. X. Wang, L. Wan, Z. Wang, X. Liu, Y. Gao, L. Wang, J. Liu, Q. Guo, W. Hu and J. Yang, J. Phys. Chem. Lett., 2022, 13, 6532–6540.
22. R. Zhang, H. Wang, S. Tang, C. Liu, F. Dong, H. Yue and B. Liang, ACS Catal., 2018, 8, 9280–9286.
23. Y. Jiang, W. Zhao, S. Li, S. Wang, Y. Fan, F. Wang, X. Qiu, Y. Zhu, Y. Zhang and C. Long, J. Am. Chem. Soc., 2022, 144, 15977–15987.
24. L. Song, R. Zhang, C. Zhou, G. Shu, K. Ma and H. Yue, Chem. Commun., 2023, 59, 478–481.
25. F. Li, B. Wang, X. Chen, Y. Lai, T. Wang, H. Fan, X. Yang and Q. Guo, JACS Au, 2022, 2, 2607–2616.
26. Z. Ren, Q. Guo, C. Xu, W. Yang, C. Xiao, D. Dai and X. Yang, Chin. J. Chem. Phys., 2012, 25, 507–512.
27. E. Lira, J. Ø. Hansen, P. Huo, R. Bechstein, P. Galliker, E. Lægsgaard, B. Hammer, S. Wendt and F. Besenbacher, Surf. Sci., 2010, 604, 1945–1960.
28. I. Sokolović, M. Reticcioli, M. Čalkovský, M. Wagner, M. Schmid, C. Franchini, U. Diebold and M. Setvín, Proc. Natl. Acad. Sci. U. S. A., 2020, 117, 14827–14837.
29. L. Chen, R. S. Smith, B. D. Kay and Z. Dohnálek, Surf. Sci., 2016, 650, 83–92.
30. M. A. Henderson, W. S. Epling, C. H. Peden and C. L. Perkins, J. Phys. Chem. B, 2003, 107, 534–545.
31. R. T. Zehr and M. A. Henderson, Surf. Sci., 2008, 602, 2238–2249.
32. R. Sun, X. Liu, X. Chen, L. Che, X. Yang and Q. Guo, J. Phys. Chem. Lett., 2022, 13, 801–807.
33. J. Xie, R. Jin, A. Li, Y. Bi, Q. Ruan, Y. Deng, Y. Zhang, S. Yao, G. Sankar and D. Ma, Nat. Catal., 2018, 1, 889–896.
34. J. J. Lin, D. W. Hwang, S. Harich, Y. T. Lee and X. Yang, Rev. Sci. Instrum., 1998, 69, 1642–1646.
35. J. T. Grant, C. A. Carrero, F. Goeltl, J. Venegas, P. Mueller, S. P. Burt, S. Specht, W. McDermott, A. Chieregato and I. Hermans, Science, 2016, 354, 1570–1573.
36. J. Ø. Hansen, R. Bebensee, U. Martinez, S. Porsgaard, E. Lira, Y. Wei, L. Lammich, Z. Li, H. Idriss and F. Besenbacher, Sci. Rep., 2016, 6, 21990.
37. Z. Ma, Q. Guo, X. Mao, Z. Ren, X. Wang, C. Xu, W. Yang, D. Dai, C. Zhou and H. Fan, J. Phys. Chem. C, 2013, 117, 10336–10344.
38. Y. K. Kim, B. D. Kay, J. White and Z. Dohnálek, Catal. Lett., 2007, 119, 1–4.
39. F. Li, X. Chen, Y. Lai, T. Wang, X. Yang and Q. Guo, J. Phys. Chem. Lett., 2022, 13, 9186–9194.
40. T. L. Thompson and J. T. Yates, J. Phys. Chem. B, 2005, 109, 18230–18236.
41. T. L. Thompson and J. T. Yates, J. Phys. Chem. B, 2006, 110, 7431–7435.
42. Y. Lai, Y. Zeng, X. Chen, T. Wang, X. Yang and Q. Guo, J. Phys. Chem. C, 2023, 127, 1863–1869.
43. J. F. Weaver, C. Hakanoglu, A. Antony and A. Asthagiri, Chem. Soc. Rev., 2014, 43, 7536–7547.
44. V. Fung, F. F. Tao and D.-e. Jiang, Phys. Chem. Chem. Phys., 2018, 20, 22909–22914.
45. C.-C. Wang, S. S. Siao and J.-C. Jiang, J. Phys. Chem. C, 2012, 116, 6367–6370.
46. V. Fung, G. Hu, F. Tao and D. E. Jiang, ChemPhysChem, 2019, 20, 2217–2220.
47. M. A. Henderson, Surf. Sci. Rep., 2011, 66, 185–297.
48. Q. Zhang, Y. J. Li, H. F. Wen, Y. Adachi, M. Miyazaki, Y. Sugawara, R. Xu, Z. H. Cheng, J. N. Brndiar and L. Kantorovich, J. Am. Chem. Soc., 2018, 140, 15668–15674.
49. J. Mullins, J. Chem. Educ., 2012, 89, 834–836.
50. D. Wilson, D. Sporleder and M. White, J. Phys. Chem. C, 2013, 117(18), 9290–9300.
51. C. Walenta, S. Kollmannsberger, C. Courtois, M. Tschurl and U. Heiz, Phys. Chem. Chem. Phys., 2018, 20, 7105–7111.
52. D. Wilson, D. Sporleder and M. White, Phys. Chem. Chem. Phys., 2012, 14, 13630–13637.
53. X. Rozanska and J. Sauer, Int. J. Quantum Chem., 2008, 108, 2223–2229.
54. C. Coperet, Chem. Rev., 2010, 110, 656–680.
55. G.-L. Dai, Z.-P. Liu, W.-N. Wang, J. Lu and K.-N. Fan, J. Phys. Chem. C, 2008, 112, 3719–3725.
56. Y. Lai, Y. Zeng, F. Li, X. Chen, T. Wang, X. Yang and Q. Guo, J. Phys. Chem. Lett., 2023, 14, 6286–6294.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3sc05623f

‡ These authors contributed equally to this work.

---