Introduction

Ethylene (C2H4) is the largest feedstock for producing numerous polymer materials and chemicals in industries1,2,3. As of today, ethylene is primarily produced by the thermal cracking of naphtha and liquefied petroleum gas, suffering from the substantial limitations of intensive thermal input, harsh operation conditions, as well as heavy carbon emission4,5,6. More critically, the depletion of fossil fuels renders the conventional ethylene production not sustainable. As an ideal alternative to fossil fuels, bioethanol can be tremendously produced from sugar cane, starch, and even lignocellulose by fermentation with a global yield of over 109.8 billion liters7,8. It is mainly utilized as a gasoline additive for internal engine combustion with a superfluous supply, which can be facilely amplified if there is a growing market demand. It is of significance to explore a new route for bioethanol valorization. From the molecular perspective, bioethanol shows a similar framework to ethylene, thus holding grand promise for the sustainable production of ethylene9. In principle, ethanol can be transformed into ethylene by simple dehydration (CH3CH2OH → CH2CH2 + H2O). However, strong acids or bases are often required to accelerate the reaction under severe conditions, making the process not eco-friendly10,11.

Photocatalysis presents an emerging strategy to drive chemical reactions by utilizing solar energy12,13,14,15,16. It is noted that, owing to the unique charge-induced characteristics, photocatalysis has great potential in biomass valorization by photoexcited active redox species e.g. electron, hole, and ·OH17. Over recent years, light-driven biomass valorization has been extensively studied by various photocatalysts18,19,20,21. Inexhaustible solar energy offers a promising energy source for photocatalytic bioethanol reforming toward ethylene, however, there have been rare endeavors in this valuable conversion route. What is more, the photocatalytic activity is hampered by low optical absorption and severe photo-generated e-/h+ recombination. Of note, it remains a grand challenge to mediate the C2H5OH-to-C2H4 conversion with the use of active species during photocatalysis, because of the complex reaction pathway and intricate chemical bond network of C-C, C-O, C-H, and O-H in ethanol. More critically, as the photogenerated active species e.g. *OH, *O2- generally possess strong redox ability, it is difficult to regulate the oriented conversion of organic matter22. There is an urgent demand to explore a rational photocatalyst for addressing the issues above.

One-dimensional gallium nitride nanowires (GaN NWs) vertically aligned onto a silicon substrate have emerged as a promising semiconductor platform for photocatalysis due to the following advantages23,24. First of all, GaN NWs onto a silicon substrate are capable of light absorption with alleviated photon scattering. Secondly, the nearly defect-free crystal structure, and fast electron mobility, in combination with the minimized charge diffusion path, are promising for enhancing the e−/h+ separation25. Moreover, the well-defined nanowires offer large surface area for anchoring catalytic sites with high atom efficiency. And the flexible morphology and surface polarity property endow GaN NWs with a tunable coordination environment, providing a unique support for tailoring the steric and electronic properties of catalytic sites26. To date, significant advancements have been made in water splitting, CO2 reduction, and CH4 conversion by utilizing GaN NWs as light absorbers27,28,29,30,31. Despite grand promise, there have been no attempts on exploring suited cocatalysts by cooperating with this emerging material for the light-driven transformation of bioethanol into ethylene.

In this study, a surface-hydrogenated binary Cr-Mn oxide is explored for coupling with GaN NWs (denoted as GaN@CMO-H) to catalyze light-driven ethanol dehydration toward ethylene. The as-designed GaN@CMO-H reports a measurable C2H4 activity of 1.78 mol‧gcat−1‧h−1 with a high turnover number of 94,769 under full-arc concentrated light irradiation of 4 W‧cm−2 without external energy input. A surface-proton migration and replenishment mechanism were proposed to disclose the ethanol dehydration pathway over GaN@CMO-H. In-situ spectroscopic characterizations, deuterium isotope labeling and control experiments, as well as theoretical calculations show that, upon light illumination, the excited electrons of GaN@CMO-H are energetically favorable for the C-O bond cleavage of CH3CH2-OH compared to the O-H bond cleavage, followed by the formation of CH3CH2• and OH- (GaN@CMO-H + CH3CH2OH + e- = GaN@CMO-H + CH3CH2• + OH-). H2O was subsequently formed from -OH and the surface-rich H of CMO-H in the presence of photoexcited holes (CMO-H + OH- + h+ = CMO* + H2O, * represents the dehydrogenated sites). The resultant dehydrogenated sites can be replenished by the further deprotonation of CH3CH2•, thus enabling the formation of C2H4 (CH3CH2• + h+ = CH2CH2 + H+). Eventually, the released protons from CH3CH2• work in synergy with the photoexcited electrons to close the catalysis loop (CMO* + H+ + e- = CMO-H). From the energetic point of view, the surface-hydrogenated property of GaN@CMO-H can significantly facilitate C2H5OH-to-C2H4 conversion by lowering the reaction energy barrier and switching the rate-determining step, thus contributing to the exceptional activity. This study shows a viable strategy for the sustainable production of ethylene with the inputs of bio-derived ethanol and sunlight beyond fossil fuels.

Results

Materials growth and characterizations

GaN@CMO-H was first fabricated by integrating plasma-assisted molecular beam epitaxy growth of 1D nanostructured GaN on a silicon wafer with photo-deposition of the surface hydrogenated binary Cr-Mn oxides (Please see Supplementary Fig. S1 and Method section). For comparison, CrOx-H and MnOx-H were separately deposited onto the GaN surface via the same procedure, which were denoted as GaN@CO-H and GaN@MO-H, respectively. The structural and morphological characterizations of the as-prepared photocatalysts were performed by various techniques. As shown in Fig. 1a and Supplementary Fig. S2, scanning electron microscopy (SEM) images displayed that the epitaxial GaN NWs were vertically aligned on a silicon wafer with an average length of about 800 nm. The decoration of CMO-H did not result in an observed morphology change of nanowires. The mass density of GaN on the silicon wafer was calculated to be ~0.302 mg‧cm−2 (Please see Method section and Supplementary Fig. S3). As characterized by high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM), the lattice space of 0.261 nm is attributed to the (002) plane of GaN (Supplementary Figs. S4–S5)32,33. Notably, the epitaxial GaN NWs exhibited nearly defect-free crystal structure. The well-defined 1D nanostructured morphology is favorable for anchoring cocatalysts with rich sites34. Furthermore, as seen in Fig. 1b, c and Supplementary Figs. S6–S7, a noteworthy core/shell structure was observed for GaN@CMO-H. Herein, the (001) plane of Cr2O3 was detected by high-resolution STEM characterization (Fig. 1c)35. The intimate contact between the CMO-H core and GaN shell, together with the defect-free 1D nanostructure of GaN, is beneficial for separating the photo-excited charge carriers (Fig. 1d and Supplementary Fig. S8). The loaded contents of Cr and Mn were calculated to be 2.84 μg‧cm−2 and 0.59 μg‧cm−2, respectively, by the inductively coupled plasma-atomic emission spectrometer (ICP-AES) (Supplementary Table S1). Moreover, as suggested by the geometric phase analysis (GPA), the GaN NWs core was nearly stress-free even after decorating with cocatalysts (Fig. 1e and Supplementary Fig. S9). Energy dispersive spectroscopy (EDS) elemental mapping images confirmed the successful distribution of Cr and Mn on the surface of GaN NWs (Fig. 1f and Supplementary Fig. S10). Meanwhile, the thickness of the cocatalyst shell was adjustable (Supplementary Fig. S11), thus enabling the optimization of ethanol dehydration.

Fig. 1: Physical and chemical characterizations of the prepared nanoarchitectures.
figure 1

a 45° tilted-view SEM image, (b) HR-TEM image, and (c) HADDF-STEM image of GaN@CMO-H. d Intensity distribution of the HADDF-STEM images of GaN@CMO-H. e GPA pattern of the stress component εxx for GaN@CMO-H based on Fig. 1d. f Elemental mapping images of GaN@CMO-H. g TOF-SIMS characterization of GaN@CMO-H vertically aligned on a silicon substrate. h The fine XPS spectra of O 1 s for GaN@CMO and GaN@CMO-H. i Transient-state PL spectra of bare GaN and GaN@CMO-H.

Full size image

To validate the surface protonation of the coupled GaN@CMO-H, time-of-flight secondary ion mass spectrometry (TOF-SIMS) was conducted. It is observed that GaN@CMO-H demonstrated an abundant signal of H over the surface (Fig. 1g). In contrast, when GaN@CMO-H was annealed in air at 573 K for 3 h, the content of hydroxyl species showed a significant reduction from 53% to 34%, indicating the dehydrogenation of the localized surface (Fig. 1h)36,37. The annealed nanoarchitecture was denoted as GaN@CMO. Furthermore, as characterized by steady-state photoluminescence (PL) spectroscopy, the recombination of photo-induced e-/h+ pairs of GaN was inhibited by decorating with CMO-H (Supplementary Fig. S12)38. And the irradiated GaN@CMO-H can provide energetic enough photogenerated charge carriers to drive ethanol dehydration to produce ethylene. Moreover, by decorating with CMO-H, the average charge lifetime (τaverage) declined from 2.95 ns downward to 2.14 ns, as shown in Fig. 1i and Supplementary Table S2. It is indicative of the faster interface electron transfer, thus electronically favoring the reaction.

X-ray photoelectron spectroscopy (XPS) characterizations were further carried out to investigate the surface chemical compositions of GaN@CMO-H (Supplementary Fig. S13). The typical peaks of N−Ga (397.3 eV) were detected for bare GaN and GaN decorated with various cocatalysts in Fig. 2a39. For GaN@CMO-H and GaN@CO-H, a new peak of N-Cr appeared at 399.6 eV. The fine XPS spectra of Ga 3 d were deconvoluted with a major peak of the Ga-N bond (Fig. 2b). Interestingly, the characteristic peaks of both Ga 3 d and N 1 s of GaN showed notable positive shifts upon decoration with different cocatalysts, which suggested that the electron flowed from GaN to cocatalysts40,41. As exhibited in Fig. 2c, the peaks at around 586.8 eV (Cr 2 p1/2) and 577.1 eV (Cr 2 p3/2) could be attributed to the deposited Cr3+ species42,43. In addition, it was evidenced that the deposited Mn species existed in the oxidation state of MnOx (Supplementary Fig. S14)44. Furthermore, a negative shift of about 0.4 eV appeared when Mn was coupled with Cr species. It suggests the electronic interaction between Cr and Mn, as verified by the positive shift of the binding energy of Mn (Supplementary Fig. S15). Bader charge analysis further validated the electron transfer from Mn to Cr (Fig. 2d). The characterizations above indicated the unique electronic properties of GaN@CMO-H, which plays an important role in the exceptional performance of light-driven bioethanol dehydration toward ethylene and will be studied below.

Fig. 2: X-ray photoelectron spectroscopy characterization and charge density distribution analysis of various architectures.
figure 2

High-resolution XPS spectra of (a) N 1 s and (b) Ga 3 d for GaN, GaN@MO-H, GaN@CO-H, and GaN@CMO-H, respectively. c High-resolution XPS spectra of Cr 2 p for GaN@CMO-H and GaN@CO-H. d Charge density distribution of GaN@CMO-H. The yellow and cyan regions indicate electronic charge gain and loss, respectively, with an isosurface of 0.02008 e/Å3. Ga, green; N, gray; Cr, blue; Mn, purple; O, red; and H, pink.

Full size image

Light-driven ethanol dehydration toward ethylene

Light-driven ethylene synthesis from ethanol over GaN@CMO-H was conducted in a homemade sealed quartz chamber (Supplementary Fig. S16)19. Unless specifically noted, the reactions were performed under illumination by a full-arc 300 W Xenon lamp at 4 W‧cm−2 under atmospheric argon. Among the photocatalysts tested, GaN@CMO-H with an area of circa 1 cm2 showed an optimum ethylene yield of 268.71 μmol during 30 min of concentrated light irradiation. By contrast, CMO-H deposited on silicon substrate was nearly inactive for bioethanol dehydration under the same conditions (Fig. 3a), suggesting that GaN is vital for the reaction. Meanwhile, pristine GaN NWs exhibited a slight C2H4 activity of 0.08 mol‧gcat−1‧h−1, which was considerably improved by the incorporation of cocatalysts. To be specific, GaN@CMO-H exhibits a superior activity compared to both GaN@CO-H and GaN@MO-H (Fig. 3b, c), suggesting the synergetic effect between Cr and Mn in promoting the photocatalytic bioethanol dehydration. Under the optimized condition, an exceptional C2H4 production rate of 1.78 mol‧gcat−1‧h−1 was obtained over GaN@C5M10O-H, outstripping state-of-the-art photo- and thermo- catalytic systems (Supplementary Table S3). Essentially, GaN NWs make a critical contribution to the outstanding performance by offering energetic charge carriers with short diffusion pathway to drive the reaction, and high surface area for enriching active sites of the cocatalyst. What is more, GaN nanowires work in synergy with CrMnOx-H to significantly lower the energy barrier of the C2H5OH-to-C2H4 conversion, which will be studied by DFT calculations.

Fig. 3: The performance of light-driven ethanol dehydration to ethylene.
figure 3

a The influence of various photocatalysts on the performance. The effect of varied (b) Cr and (c) Mn precursor concentrations on the yield rate of ethylene. d The activity comparison between photocatalysis and thermocatalysis. e The influence of surface dehydrogenation on the performance. (f) Stability testing. Experimental conditions: 300 W Xenon lamp; ethanol dosage, 1 mL; atmosphere, 1 atm Ar; catalysts, GaN@CMO-H, ~0.302 mg cm−2, 0.2−1.0 cm2.

Full size image

The light intensity-dependent activity was further measured. As shown in Fig. 3d and Supplementary Fig. S17, the reaction did not occur in the dark without external heating. In contrast, the C2H4 rate was sharply increased from 0.41 mol·gcat−1·h−1 up to 1.78 mol·gcat−1·h−1 when the light intensity was increased from 2 to 4 W·cm−2. And the light-to-ethylene (LTE) conversion efficiency reached 0.17% under light illumination of 4 W cm−2. It primarily arises from the enriched photoexcited charges under higher intensity illumination. The influence of light wavelength on the reaction was also studied. Equipped the full-arc xenon lamp with a visible or an infrared filter, the performance of ethanol dehydration towards ethylene over GaN@CMO-H was not detected. However, when equipped with an ultraviolet filter, a low activity was noticed. These results point to the fact that ultraviolet light excites the photogenerated charge carriers upon GaN@CMO-H to promote the reaction. Notably, once an external heat source was applied for heating the reaction system (433 K, which is limited by the settable maximum temperature of the photothermal stainless steel reactor), the introduction of ultraviolet light can greatly improve the activity compared to that under pure ultraviolet light illumination without external heat source. These findings reveal the synergy between charge carriers induced by ultraviolet light and photo-induced heat as a result of visible light and infrared light played a vital role in the performance of GaN@CMO-H. Considering the photo-thermal effect under concentrated light illumination, the surface temperature of the photocatalyst was measured by an infrared thermal camera. The thermo-catalytic activity of GaN@CMO-H was then evaluated at the corresponding temperatures at various light intensities by an external heating system (Supplementary Fig. S18). It was found that the pure thermo-catalytic turnover frequency (TOF) of ethylene (702.2 h−1 at 553 K) was over one order of magnitude lower than that of photocatalysis with the only input of concentrated light (8267.9 h−1 at 4 W cm−2) (Supplementary Fig. S19). Notably, based on the Arrhenius equation45, the apparent activation energy (Ea) of ethanol dehydration over GaN@CMO-H is sharply diminished from 4.62 eV under dark to 2.88 eV upon the introduction of light (Supplementary Fig. S20), manifesting that the reaction is significantly improved by photogenerated electron-hole pairs. In addition, the results of carbon-based gas product distribution exhibited that the carbon selectivity toward C2H4 driven by photocatalysis reached over 80% along with trace amounts of CO, CH4, and C2H6, which was superior to that driven by thermocatalysis (Supplementary Table S4). The results above suggest that photoinduced charge carriers make a dominant contribution to the exceptional activity and selectivity of ethanol dehydration toward ethylene. In stark contrast, upon annealing in the air at 573 K for 3 h, the dehydrogenated GaN@CMO experienced a sharp C2H4 activity reduction from 1.78–0.52 mol‧gcat−1‧h−1 (Fig. 3e). It is rationalized that the surface hydrogen of the photocatalyst plays a vital role in light-driven ethanol dehydration to ethylene. As a key descriptor, the stability of the photocatalyst was also tested. During one cycle of 4 h, the produced ethylene reached 11.77 ml at 298 K under 1 atmospheric pressure (equivalent to 0.525 mmol) with a turnover frequency of 5051 h−1 (Supplementary Fig. S21). Moreover, the yield rate of ethylene over GaN@CMO-H did not show a remarkable variation during a long-term operation of 11 h, affording a considerable turnover number (TON) of 94,769 (Fig. 3f). This result indicates the robustness of the photocatalyst for the consecutive production of ethylene from bioethanol.

Mechanism of light-driven ethanol dehydration

To gain more insights into the critical role of the surface-hydrogenated photocatalyst, a series of isotope-labeling and control experiments, and in-situ spectroscopic investigations were conducted. The photocatalyst was first labeled with deuterium by conducting the photo-deposition of Cr-Mn in D2O (Denoted as GaN@CMO-D). GaN@CMO-D was then used for catalyzing bioethanol dehydration. It was found that the surface hydrogen content showed a reduction as measured by time-of-flight secondary ion mass spectrometry (TOF-SIMS) (Fig. 4a). More additionally, DOH was detected in the liquid reaction mixture by liquid chromatography-mass spectroscopy (Fig. 4b). However, DOH was not produced in a blank control experiment without ethanol. These findings reasonably validate the participation of the surface deuterium atoms in the C2H5OH-to-C2H4 conversion process, leading to the formation of the dehydrogenated sites (CMO*). Furthermore, based on the infrared (IR) spectroscopy characterization, four infrared peaks at 2988, 2940, 2904, and 2884 cm−1 attributed to the CH3 arising from C2H5OH/C2H5 were observed during the light-driven reaction (Fig. 4c), and it is further verified to be C2H5 by the quasi-in-situ electron paramagnetic resonance (EPR) spectroscopy result (Fig. 4d)46,47,48. Notably, the ethyl radical was not detected by EPR when the reaction operated under pure thermal conditions (Supplementary Fig. S22), which reveals a distinct reaction pathway between thermo-catalysis and photo-catalysis. These results are indicative of the C-O bond cleavage of C2H5-OH induced by photo-generated electrons under light illumination over GaN@CMO-H. The resultant OH- via the scission of C-O was facilely consumed for generating H2O by extracting the surface hydrogen of CMO-H in the presence of holes (OH- + CMO-H + h+ = H2O + CMO*). C2H4 was then yielded from the further deprotonation of ethyl radicals (•C2H5 + h+ =  C2H4 + H+), as indicated by the appearance of the C = C bond characteristics at 1604 cm−2 (Supplementary Fig. S23)49. The released protons participated in replenishing the dehydrogenated sites by photoexcited electrons, thus leading to the achievement of an ideal chemical loop (CMO* + H+ + e- = CMO-H). In-situ XPS characterizations31 suggest that the deposited CMO-H serves as effective electron sinks during photocatalysis (Fig. 4e, f, Supplementary Figs. S24 and 25). It is critical for the superior activity by separating the photoexcited electron-hole pairs and breaking the C-O bond cleavage of CH3CH2OH, as verified by the varied activity between photocatalysis and thermocatalysis in Fig. 3d. The results provide atomic and spectroscopic evidence for outlining the basic surface-proton migration and replenishment reaction pathway of ethanol toward ethylene (Fig. 4g and Supplementary Fig. S26).

Fig. 4: Atomic and spectroscopic investigations of light-driven ethanol dehydration toward ethylene over GaN@CMO-H.
figure 4

a TOF-SIMS signal of D element over GaN@CMO-D before and after the reaction. b The signal of D2O element in the liquid production. c In-situ IR for the ethanol dehydration process over GaN@CMO-H. d Quasi-in-situ electron paramagnetic resonance spectroscopy measurement for ethyl radical. In-situ high-resolution XPS spectra of (e) Ga 3 d and (f) Cr 2 p under illumination. g Schematic illustration of surface-proton migration and replenishment during ethanol dehydration.

Full size image

DFT calculations were further performed to better understand the mechanism of ethanol dehydration at the molecular level (Supplementary Figs. S27–S31). The adsorption behavior of ethanol was first studied. As illustrated in Fig. 5a, bare GaN is efficient for ethanol adsorption with an energy of -1.252 eV. The introduction of CMO-H weakens its interaction with ethanol as indicated by a reduced adsorption energy of -1.064 eV. After annealing, the surface-dehydrogenated GaN@CMO exhibits a maximum ethanol adsorption energy of -1.557 eV, not dynamically favorable for the subsequent reaction steps. The Gibbs free energy diagram of ethanol dehydration toward ethylene was further calculated. As shown in Fig. 5b and Supplementary Tables S5–S7, on bare GaN, the reaction energy of *C2H5OH → *C2H5 + *OH is estimated to be 0.314 eV. The further dehydrogenation of *C2H5 (*C2H5 + *OH → *C2H4 + *OH + *H) was found to be the rate-limiting step of ethanol dehydration to ethylene over bare GaN with a high reaction energy of 0.857 eV. Upon introduction of either CMO-H or CMO, the C-O cleavage of C2H5-OH into *C2H5 and *OH becomes thermodynamically favored. Meanwhile, in sharp contrast with bare GaN and GaN@CMO, the surface-hydrogen property of GaN@CMO-H significantly alters the reaction pathway of ethanol dehydration toward ethylene. To be specific, the surface hydrogen of GaN@CMO-H directly participates in the formation of water by reacting with the *OH released from C2H5OH (GaN@CMO-H + *OH + C2H5* = GaN@CMO* + H2O + C2H5*), which is substantially different from that on bare GaN and GaN@CMO due to the lack of surface hydrogen (C2H5* + *OH = H* + C2H4* + *OH). This hypothesis has been validated by the isotope labeling experiments and in-situ spectroscopic characterizations above. The dehydrogenated sites can be then replenished by the dehydrogenation of *C2H5, in the concurrent generation of C2H4. Of note, compared to GaN@CMO, the reaction energy of the rate-limiting step over GaN@CMO-H was lowered from 0.777 to 0.643 eV and the rate-limiting step was switched from *C2H4 desorption to *H2O desorption. What is more, as shown in Fig. 5c, the reaction energy of the key *H2O intermediate formation over GaN@CMO-H is 0.288 eV, which is much lower than that over GaN@CMO (0.606 eV) and bare GaN (0.857 eV). Moreover, coupling Cr with Mn species regulates the adsorption of key intermediates (Supplementary Fig. S32). The calculated results are consistent with the activity order of GaN@CMO-H > GaN@CMO > GaN. The experimental and theoretical results above demonstrate that the surface-hydrogenated property of GaN@CMO-H is vital in dominating the reaction pathway and promoting light-driven ethanol dehydration into C2H4 (Fig. 6).

Fig. 5: Density functional theory calculations.
figure 5

a The adsorption energy of ethanol, (b) energy profiles for the reaction path of ethanol dehydration, and (c) the formation energy of *H2O intermediate over bare GaN, GaN@CMO, and GaN@CMO-H. The value in the figures indicated the reaction energy for the potential-limiting steps of the forward reaction.

Full size image
Fig. 6: Schematic illustration of ethanol dehydration over the surface-hydrogenated GaN@CMO-H.
figure 6

Under light irradiation, the GaN@CMO-H nanoarchitecture dominates the reaction of ethylene production from biomass-derived ethanol with photoexcited charge carriers.

Full size image

Discussion

In summary, a surface-hydrogenated binary Cr-Mn oxide has been first coupled with GaN nanowires on a silicon substrate for light-driven bioethanol dehydration toward ethylene. The assembled architecture affords a measurable reaction rate of 1.78 mole C2H4 per gram GaN@CMO-H per hour under concentrated light illumination of 4 W cm−2 with a high TON of 94,769 per mole ethylene per CMO-H. In-situ spectroscopic characterizations, isotope-labeling and control experiments, and density functional theory calculations disclosed that at the molecular level, the surface hydrogenated property of the binary CrMn oxides is energetically favorable for breaking the C-O bond of C2H5-OH. More importantly, the hydroxyl intermediate released from the C-O dissociation of C2H5OH was consumed by the surface hydrogen of CMO-H to yield H2O. The dehydrogenated sites of CMO* can be recovered by further deprotonation of •C2H5, thus closing the catalysis loop of C2H4 synthesis with a considerably lowered reaction energy. This work presents a new route for light-driven bioethanol valorization toward C2H4, thus promising to achieve sustainable production of the important chemical bulk beyond fossil fuels.

Methods

Materials

The chemicals were purchased from commercial companies and directly used without any purification.

Synthesis of 1-dimensional GaN nanowires (NWs) on the monocrystalline silicon wafer

Using a radio-frequency plasma-assisted SVTA MBE system under nitrogen-rich conditions, the nearly defect-free GaN NWs were grown on a 4-inch monocrystalline Si (111) wafer substrate. Based on the previous reports19,50, the standard pre-growth processes including high temperature degas and nitridation were first performed. The nanowire growth was carried out by maintaining a constant nitrogen flux and growth temperature of ~700 °C measured by a thermocouple. The nanowire consisted of about 400 nm unintentionally doped GaN and 400 nm p-GaN (Mg-doped) layer. At first, undoped GaN NWs with a height of about 400 nm were grown. Subsequently, the Mg-doped GaN upper parts were grown in multiple steps by switching Mg cell temperatures in the order of 230°C, 250°C, and 270°C (Mg and/or hole concentrations were intentionally varied by changing Mg flux and/or cell temperature). Notably, the GaN NWs had been grown using a continuous growth process without any growth interruption by changing only Mg flux. It was worthwhile mentioning that it was difficult to precisely know the hole density at different Mg fluxes as there was no known and reliable technique to measure Mg and/or hole density in nanowires.

The coupling of surface-hydrogenated CrMnOx with GaN NWs

CMO-H was coupled with GaN NWs by a straightforward one-step photo-deposition process. Typically, the epitaxial GaN NWs vertically aligned onto a silicon substrate were first placed on the bottom of a 0.4 L quartz chamber, followed by the introduction of 36 ml methanol aqueous solution with a CH3OH/H2O volume ratio of 1/6. Desired amounts of the precursor aqueous solutions of Cr(NO3)3·9H2O (0.2 mol/L) (>99.99% trace metals, Sigma-Aldrich) and Mn(NO3)2 (0.2 mol/L) were then injected into the chamber. The chamber was then vacuumed for several times to remove the air in the system and filled with atmospheric argon (99.999%, Shanghai Wetry Standard Gas Analysis Technology Co., Ltd.). After that, the chamber was irradiated by a 300 W xenon lamp without optical filter (AuLight, CEL-HLF300-T3) for 30 min. Finally, the prepared GaN @CMO-H was rinsed with distilled water thoroughly. For comparison, single Cr or Mn species were deposited on Si/GaN NWs by the same method, and the obtained photocatalysts were denoted as GaN@CO-H, and GaN@MO-H, respectively.

Materials characterization

X-ray diffraction (XRD) patterns were collected on a Bruker D8 Advance diffractometer (with Cu Kα, at 60 kV and 80 mA) in continuous scanning mode over a 2θ range of 20–80 degrees. X-ray photoelectron spectroscopy (XPS) was collected by an ESCALAB 250xi non-monochromatic Al anodes, and the binding energy of C 1 s at 284.8 eV was used for the internal calibration. The loading density of the cocatalyst was evaluated by an inductively coupled plasma-atomic emission spectroscopy (AGILENT ICP-OES 730). High-angle annular dark field-scanning transmission electron microscopy (HAADF-STEM) and energy dispersive X-ray spectroscopy (EDS) were performed using a Thermo Fisher Scientific Talos F200X S/TEM, equipped with a Super-X EDS detector and operated at 200 kV. Scanning electron microscopy (SEM) images were conducted using a Quattro ESEM (Thermo Fisher). Transmission electron microscopy (TEM) images were conducted using a JEOL 2100 F microscope at 200 kV. The electron paramagnetic resonance (EPR) measurements were carried out on a Bruker A300 spectrometer under room temperature using 5,5-dimethyl−1-pyrroline N-oxide (DMPO) as the trapping agent. Steady-state photoluminescence (PL) and transient-state PL spectra were tested on the F-4600 and FLS980 fluorescence spectrophotometer (320 nm, Edinburgh Instruments, UK), respectively. The semiconductor bandgap is calculated by the formula (Bandgap = 1240/Wavelength). Time-of-flight secondary ion mass spectroscopy (TOF-SIMS) measurements were performed using an ION TOF ToF SIMS 5−100 (Germany). Liquid chromatography-mass spectrometry (LC-MS) measurement was conducted using an Agilent 1100 + thermos TSQ quantum Ultra AM, USA.

Photocatalytic reactions

Photocatalytic ethanol dehydration was performed in a 0.44 L homemade sealed quartz chamber illuminated by a full-arc 300 W Xenon lamp (AuLight, CEL-HLF300-T3) without an optical filter if not specifically noted. Typically, the prepared GaN@CMO-H was placed at the bottom of the chamber. The system was then evacuated and filled with Ar, followed by injecting 1 ml of ethanol (99.7%, Aladdin). The light intensity was set at 4 W/cm2 if not specifically noted. The gaseous products were measured by a gas chromatograph (GC) equipped with a flame ionization detector (FID) and a thermal conductivity detector (TCD). During the long-term stability test, after performing the first test, the photocatalyst was thoroughly rinsed and evacuated, subsequently filled with Ar. The fresh ethanol was then injected into the chamber, followed by irradiation for the next cycle.

Calculation of reaction performance activity of ethylene production

$${{\rm{n}}}=\frac{{{\mbox{V}}}_{{{\rm{chamber}}}}\times {{\rm{C}}}}{{{{\rm{V}}}}_{{{\rm{m}}}}}$$
(1)
$${\mbox{Gas Yield Rate}}=\frac{{\mbox{n}}}{{\mbox{S}}\times {\mbox{W}}\times {\mbox{T}}}$$
(2)
$${\mbox{TOF}}=\frac{{\mbox{n}}}{{\mbox{S}}\times {\mbox{T}}\times {\mbox{Loading Density}}}$$
(3)
$${\mbox{TON}}={\mbox{TOF}}\times {\mbox{T}}$$
(4)

where Vm is the gas molar volume under standard conditions, 22.4 L/mol. Vchamber is the total gas volume of the reaction chamber, 0.44 L. \({{{\bf{C}}}}\) is the ethylene concentration measured by GC. n is the produced molar amount of ethylene. \({{{\bf{S}}}}\) is the geometric surface of the sample. W is the total weight of the epitaxial of GaN NWs, estimated to be ~0.302 mg/cm2, without considering the weight of trace CrMnOx-H. T is the reaction time. Loading Density of CrMn cocatalyst is evaluated to be ~0.065 µmol/cm2 by ICP-AES. TOF is the turnover frequency. TON is the turnover number.

Calculation of the light-to-ethylene efficiency

The calculation of LTE (light-to-ethylene) efficiency under various light intensity was evaluated by using a 300 W Xe lamp as the light source. As shown in Fig. 3a, after light illumination of 30 min, the evolution amount of ethylene was detected by GC to be 268.71 μmol, and the free energy of output ethylene was then calculated to be 12.065 J as the standard Gibbs free energy of the reaction is 44.9 kJ/mol51. The incident light intensity was 4 W cm−2 with an irradiated area of circa 1 cm2, so that the energy of input light could be calculated to be 7200 J. Finally, the LTE efficient was calculated to be 0.17% by the following equation:

$${{{\rm{LTE}}}}=\frac{{\mbox{Energy of output }}{{\mbox{C}}}_{2}{{\mbox{H}}}_{4}}{{\mbox{Energy of input light}}}=\frac{{\mbox{Amount of }}{{\mbox{C}}}_{2}{{\mbox{H}}}_{4}{\mbox{ evolution }}\times 44.9 {\mbox{ kJ/mol}}}{{\mbox{Light intensity}}\times {\mbox{S}}\times {\mbox{T}}}$$
(5)

where S represents the geometric surface of the sample and T is the reaction time.

Theoretical section

DFT calculations were performed employing the Vienna Ab Initio Simulation Package (VASP 5.4.4)52,53. The interaction between valence electrons and ions were described by the projector-augmented wave (PAW) method54. The exchange correlation of the Kohn-Sham equation was presented by generalized gradient approximation (GGA) with the Perdew-Burke-Ernzerhof (PBE) functional and Grimme’s DFT-D3 method to include the effect of weak van der Waals (vdW) interactions55,56. A plane-wave basic set was used to expand the wavefunctions with the kinetic cutoff energy set to 450 eV. For the calculations of slabs, a 2×2×1 Mokhorst-Pack k-point grid was implemented to sample the Brillouin zone57. The atomic geometric structures were fully relaxed, and the convergence criteria for the atomic forces and the total energies were set to 0.02 eV/Å and 10-5  eV, respectively. To obtain the interactions during the reaction, the transition states were explored by using the climbing-image nudged elastic band (CI-NEB) method58.

As the m-plane of GaN was observed in experiments, we constructed GaN (10-10) surface with a 4 × 3 supercell containing 4 layers to represent the pristine GaN. Besides, the models of Cr4O6/GaN, Cr2Mn2O6/GaN, Cr4O6H6/GaN and Cr2Mn2O6H6/GaN were created based on depositing a Cr4O6 cluster on GaN (10-10). The top view, side view of optimized geometries models of these models are represented in Supplementary Fig. S27. A vacuum spacing of at least 25 Å was set along the normal direction to the surface for all slab models and along all three directions for nanoparticles to avoid image interaction. Moreover, for all slab models, the bottom three layers of GaN were fixed in their bulk positions during structural relaxation. The Gibbs free energy of adsorption ∆G was calculated via the:

$$\Delta {\mbox{G}}={{\mbox{E}}}_{{{\rm{ad}}}}+\Delta {\mbox{ZPE}}-{\mbox{T}}\Delta {\mbox{S}}$$
(6)

where ΔZPE and ΔS are the changes in zero-point energy and entropy of a reactant, respectively. All the energies were corrected by considering under standard conditions (p0 = 1 bar and T0 = 298.15 K). Ead is the adsorption energy defined by:

$${{{\rm{E}}}}_{{{\rm{ad}}}}={{{\rm{E}}}}_{{{\rm{Total}}}}-{{{\rm{E}}}}_{{{\rm{Surface}}}}+{{{\rm{E}}}}_{{{\rm{Adsorbate}}}}$$
(7)

where ETotal, ESurface, and EAdsorbate are the total energies of the optimized surface with adsorbate, adsorbate in the gas phase, and the surface, respectively. The binding of hydrogen atom with the surface, EH is taken as half of the total energy of a gas phase H2 (\(1/2\,{E}_{{H}_{2}}\)).