Visible light-driven C−H activation and C–C coupling of methanol into ethylene glycol

The development of new methods for the direct transformation of methanol into two or multi-carbon compounds via controlled carbon–carbon coupling is a highly attractive but challenging goal. Here, we report the first visible-light-driven dehydrogenative coupling of methanol into ethylene glycol, an important chemical currently produced from petroleum. Ethylene glycol is formed with 90% selectivity and high efficiency, together with hydrogen over a molybdenum disulfide nanofoam-modified cadmium sulfide nanorod catalyst. Mechanistic studies reveal a preferential activation of C−H bond instead of O−H bond in methanol by photoexcited holes on CdS via a concerted proton–electron transfer mechanism, forming a hydroxymethyl radical (⋅CH2OH) that can readily desorb from catalyst surfaces for subsequent coupling. This work not only offers an alternative nonpetroleum route for the synthesis of EG but also presents a unique visible-light-driven catalytic C−H activation with the hydroxyl group in the same molecule keeping intact.

Photocatalytic synthesis is an interesting and important aspect in semiconductor photocatalysis. The authors reported a visible-light-driven dehydrogenative coupling of methanol into ethylene glycol (EG) over nanofoam-modified cadmium sulfide nanorod catalyst. Meanwhile, a synthetic mechanism study was carried out. This work might be suitable for publication in Nature Communications if the authors can address the following concerns properly: 1) The authors "estimated the ratio of photogenerated electrons and holes consumed in product formation by assuming Eqs. 2-4", Page 4; however, for most cases in Table 1, the estimated e-/h+ is smaller than 1 and even 0.87 for TiO2. Proper explanation should be given out on this issue.
2) The authors claimed that "The enrichment of intermediate inside the mesopores of MoS2 foam may be a reason for the enhancement of EG selectivity.", Page 7. Please offer the pore size and distribution information of mesopores MoS2 and make a comparison with the size of intermediate like CH3O･ to clarify this assumption.
3) Based on "assuming the nearby surface sulfur atom as a proton acceptor" and theoretical simulation, a conclusion that "weak adsorption of •CH2OH on catalyst surfaces plays a key role in the formation of EG" was drawn. However, it seems that this assumption lacks generality. For example, the selectivity of HCHO over CuS reached 89% while the selectivity to EG is 0. Please give out proper explanation. 4) The photocatalytic oxidation of methanol is a different perspective of photocatalytic H2 evolution with the presence of hole scavenger. So, it should be comparable. However, in Table 1, the H2 evolution rates over some traditional semiconductor were relatively low, particular for TiO2. How to comprehend these results in addition to the absence of Pt as cocatalyst. 5) Wavelength dependence of quantum yield in EG production should be given to further identify a photocatalytic reaction.

Reply and actions taken:
We thank the reviewer for the comments. We would like first to point out that the C-H cleavage is not an easy task, and it is one of the most important challenges in catalysis. Perhaps even more challenging in our case is the selective C-H cleavage in CH 3 OH over O-H breaking. Our work presents the first visible-light-driven photocatalytic process producing high-value EG from CH 3 OH with good efficiency and high selectivity. Although the focus of this work is the formation of EG via coupling of •CH 2 OH from CH 3 OH, we agree with the reviewer that the mechanism of HCHO formation may not be simple.

2c-ii and Supplementary Table 5)" (please see from Page 8-Line 2 from bottom to
Page 9-Line 2). The path for the formation of HCHO via •CH 2 OH has also been added in Figure 2c (Please see Page 18, Figure 2c).

Response to Reviewer 2
General Comments: The authors describe an original and very interesting visible-light driven homocoupling of methanol to produce ethylene glycol in up to 90% selectivity on a heterogeneous catalyst system consisting of molybdenum disulfide nanofoam-modified cadmium sulfide nanorods. They have conducted detailed mechanistic studies both experimentally and using DFT methods, which demonstrate the preferred activation of C-H bonds over O-H bonds in the system by photoexcited holes on the catalyst. The study is of broad interest to the chemical and chemical engineering communitites as the direct transformation of methanol into multicarbon species in a controlled fashion is very challenging. The unique visible-light driven catalytic C-H activation is also highly interesting at a fundamental level. Moreover, the work demonstrates a chemical route towards ethylene glycol that does not require petroleum.
The manuscript is generally well written, holds high scientific and graphical quality and have only minor spelling/grammar issues. The topic is timely, of high interest and the work will be widely read and cited. Recommendation: Publish as is.

Reply:
We appreciate the very positive evaluation from this reviewer.

Response to Reviewer 3
General Comments: Photocatalytic synthesis is an interesting and important aspect in semiconductor photocatalysis. The authors reported a visible-light-driven dehydrogenative coupling of methanol into ethylene glycol (EG) over nanofoam-modified cadmium sulfide nanorod catalyst. Meanwhile, a synthetic mechanism study was carried out. This work might be suitable for publication in Nature Communications if the authors can address the following concerns properly.

Reply:
We thank the reviewer for the pertinent comments on our work. Our replies to the issues raised by this reviewer and the corresponding revisions are described as follows.

Comment 1:
The authors "estimated the ratio of photogenerated electrons and holes consumed in product formation by assuming Eqs. 2-4", Page 4; however, for most cases in Table 1, the estimated e-/h+ is smaller than 1 and even 0.87 for TiO 2 . Proper explanation should be given out on this issue.

Reply and actions taken:
We thank the reviewer for this comment. The calculated ratio of e -/h + in Table 1 is indeed slightly less than 1.0. We believe that the detection errors of gas/liquid chromatography measurements and the trace amounts of other products are the two main reasons for this issue. The detection errors of gas/liquid chromatography were estimated by testing the standard samples for at least three times. The relative deviation of detection was 4% for gas chromatography and 3% for liquid chromatography. The amount of electrons was calculated from the amount of H 2 and CH 4 detected by gas chromatography, and the amount of holes was calculated by the amount of CO, CO 2 detected by gas chromatography and EG, HCHO, HCOOH measured by liquid chromatography. Thus, the overall relative deviation of the estimated e -/h + in the detection may reach about 7%. In addition, we have observed the formation of very small amount of CH 4 under UV-vis light irradiation. The amount of CH 4 produced on the UV-light-responsive catalyst has been added into Table 1 in the revised manuscript. CH 4 should be formed from CH 3 OH via two-electron reduction process. Thus, we have recalculated the ratio of e -/h + using the following equation: e -/h + = [2×n(H 2 ) + 2×n(CH 4 )] / [2×n(EG) + 2×n(HCHO) + 4×n(HCOOH) + 4×n(CO) + 6×n(CO 2 )] The e -/h + value calculated using this equation for each catalyst is now in the range of 0.90-1.0. Therefore, we think that the calculated ratio of electrons and holes consumed in product formation is reasonably accurate.
We have added the rate of CH 4 formation in Table 1 (please see Page 20, Table   1). The following sentence has been added in the footnote of Table 1 Table   1, footnote). We have further added the following sentences in the revised manuscript to describe the relative errors: "H 2 , CH 4 , CO and CO 2 were analyzed by an Agilent Micro GC3000 equipped with a molecular sieve 5A column and a high-sensitivity thermal conductivity detector. The relative deviation of detection was 4% for gas chromatography and 3% for liquid chromatography" (please see Page 11, Paragraph 2, the last two sentences).

Reply and actions taken:
We thank the reviewer for this useful suggestion. We have performed N 2 adsorption measurements to gain information on pore-size distribution for MoS 2 foam. The obtained pore-size distribution has been displayed in Supplementary Figure 10  We have added the following sentences in the revised manuscript: "The average pore size of MoS 2 nanofoam derived from N 2 physisorption is 26 nm ( Supplementary  Fig. 10), while the size of •CH 2 OH is only 0.29 nm. Thus, the •CH 2 OH radicals can easily diffuse into the mesopores of MoS 2 nanofoam" (please see Page 6-last three lines).

Comment 3:
Based on "assuming the nearby surface sulfur atom as a proton acceptor" and theoretical simulation, a conclusion that "weak adsorption of •CH 2 OH on catalyst surfaces plays a key role in the formation of EG" was drawn. However, it seems that this assumption lacks generality. For example, the selectivity of HCHO over CuS reached 89% while the selectivity to EG is 0. Please give out proper explanation.

Reply and actions taken:
We thank the reviewer for this important comment. To better understand the effect of surface adsorption, CuS (with zero EG selectivity) and ZnS (with 54% EG selectivity) have also been chosen to calculate the adsorption energies of •CH 2 OH on the surfaces. The adsorption energies of •CH 2 OH and CH 3 O• on CuS and ZnS surfaces have been added into the Supplementary Table 5 in the revised manuscript. On the CuS surface, the adsorption energies of •CH 2 OH and CH 3 O• are −1.0 eV and −1.7 eV, respectively. Similar to TiO 2 , with such strong adsorption, •CH 2 OH will readily undergo further oxidation to form HCHO instead of desorption and coupling to form EG. In contrast, the •CH 2 OH and CH 3 O• intermediates have lower adsorption energies of -0.5 eV and -0.6 eV, respectively, on ZnS. Thus, the •CH 2 OH intermediate can desorb from the ZnS surface relatively more easily for further coupling, consistent with our finding that ZnS has a moderate selectivity of EG. All these results indeed show that weak adsorption of •CH 2 OH intermediate on catalyst surfaces plays a key role in the formation of EG.
We have slightly revised Supplementary Fig. 17 Table 1, the H 2 evolution rates over some traditional semiconductor were relatively low, particular for TiO 2 . How to comprehend these results in addition to the absence of Pt as cocatalyst.

Reply and actions taken:
We thank the reviewer for this comment. The TiO 2 used in our work is commercial Degussa P25, which have been widely used in photocatalysis.
The H 2 formation rates on TiO 2 (P25) using methanol as a scavenger in some recent work have been listed in Table R1 and the values are in the range of 0.36-4.6 mmol g cat -1 h -1 . The difference should arise from the different reaction devices and reaction conditions used for the reaction. The H 2 formation rate on TiO 2 (P25) in our work is 2.0 mmol g cat -1 h -1 , which is comparable to the results in the literature. Comment 5: Wavelength dependence of quantum yield in EG production should be given to further identify a photocatalytic reaction.

Reply and actions taken:
We appreciate this constructive comment from the reviewer to help further improve the quality of our manuscript. We have measured the apparent quantum yields using irradiations with different wavelengths (λ = 380, 420, 450, 475, 500, 550 and 600 nm) for the photocatalytic conversion of CH 3 OH to EG over the 5% MoS 2 -foam/CdS catalyst. The results have been added in Supplementary   Fig. 16 in the revised manuscript. The quantum yield of EG was above 5.0% at wavelengths ≤ 450 nm for the MoS 2 -foam/CdS catalyst, and decreased upon increasing the wavelength (Supplementary Fig. 16). The longest wavelength suitable for EG formation was found to coincide with the absorption edge of the MoS 2 -foam/CdS catalyst, which was obtained from the diffuse reflectance UV-vis measurement. This further indicates that the formation of EG is indeed driven by light.
Based on these results, we have added the following new paragraph in the revised manuscript: "We have measured the apparent quantum yields of EG under irradiation with different wavelengths. The quantum yield of EG was above 5.0% at wavelengths not longer than 450 nm for the MoS 2 -foam/CdS catalyst, and decreased upon increasing the wavelength (Supplementary Fig. 16). The longest wavelength suitable for EG formation (~500 nm) was found to coincide with the absorption edge of the MoS 2 -foam/CdS catalyst, which was obtained from the diffuse reflectance