Br−/BrO−-mediated highly efficient photoelectrochemical epoxidation of alkenes on α-Fe2O3

Epoxides are significant intermediates for the manufacture of pharmaceuticals and epoxy resins. In this study, we develop a Br−/BrO− mediated photoelectrochemical epoxidation system on α-Fe2O3. High selectivity (up to >99%) and faradaic efficiency (up to 82 ± 4%) for the epoxidation of a wide range of alkenes are achieved, with water as oxygen source, which are far beyond the most reported electrochemical and photoelectrochemical epoxidation performances. Further, we can verify that the epoxidation reaction is mediated by Br−/BrO− route, in which Br− is oxidized non-radically to BrO− by an oxygen atom transfer pathway on α-Fe2O3, and the formed BrO− in turn transfers its oxygen atom to the alkenes. The non-radical mediated characteristic and the favorable thermodynamics of the oxygen atom transfer process make the epoxidation reactions very efficient. We believe that this photoelectrochemical Br−/BrO−-mediated epoxidation provides a promising strategy for value-added production of epoxides and hydrogen.

Summary -An organic chemistry oriented journal such as the Journal of Organic Chemistry or Organic Letters is more reasonable for this research as it stands. Nature Communications would require a significant amount of additional research.

Response to Reviewer's Comments
(Reviewer's comments in black font. Our response in blue font. Revision in manuscript in yellow background.) Thanks to all reviewers for having given us valuable comments and suggestions on the manuscript of NCOMMS-22-38942 submitted to the journal of Nature

Communications.
To Reviewer #1: The manuscript by Chen et al. reported a high efficiently photoelectrochemical epoxidation research through Br -/BrOmediated route with oxygen atom transfer pathway on α-Fe2O3 anode. In comparison with the reported methods, the PEC Br -/BrOmediated epoxidation system shows clear advantages for satisfied selectivity and Faradaic efficiency for a wide range of alkenes. The authors have verified the epoxidation process mediated by Br -/BrOroute with an oxygen atom transfer pathway, and conducted broad substrates to illustrate the universality of the methods. The manuscript is very interesting and worth to studying for the efficient synthesis of epoxides, however, it still needs to solve some nonnegligible problems before publishing in this journal.
Response: Thanks for the reviewer's precious comments and suggestions. We have revised our manuscript according to the reviewer's suggestions.
The tests of incident photon to current efficiencies (IPCE) and other electrolytes (NaBr, NaBrO3, TBACl and TBAI) were also conducted. The corresponding results and discussions have been supplied in the revised manuscript and supplementary information. The point-by-point responses are presented as follows.
1). The specific structure and elemental state of α-Fe2O3 are necessary to provide for a heterogeneous electrode, even if the same material has been reported in your previous reports, such as XRD, XPS, and TEM, etc., which contribute to understanding deeply the changes of structure and metal valence before and after reaction. Additionally, the same characterizations of the used electrode are needed. TiO2 anode is found to be changed significantly after the Br-mediated PEC reaction. Particularly, in the used TiO2 photoanode, a new absorption in the wavelength range of 400 -650 nm in the UV-vis spectra appears (Fig. R3b).

Response
More O1s and C1s peaks in the XPS spectra are observed (Fig. R4). The SEM image shows that the surface of anode seems to be covered by some amorphous species (Fig. R3d). All these results suggest that some organic byproducts during PEC Br-mediated epoxidation process are formed and deposited on the surface of TiO2, which is consistent with the poorer selectivity and the rapid decay in the PEC activity of TiO2.     2). In this manuscript, above 80% of Faradaic efficiency for epoxidation was achieved on α-Fe2O3 anode, however, what reactions happen for the rest electrons?
Response: Due to high selectivity of epoxide (> 99%), the possibility of other oxidation pathways of the alkene substrates (e.g. to form diol, aldehyde) can be safely excluded. The oxidation of water to produce O2 or/and overoxidation of bromide may account for the rest electrons in our system. To examine the contribution of these two reactions, the gas in the headspace after the PEC reaction was analyzed by gas chromatography (GC). No dioxygen (if formed, its retention time should be at ~ 1.6 min) was detected ( Fig. R5) in GC spectra, which rules out that water oxidation to O2 is a significant contribution to the rest currents. By ion chromatography (IC), some BrO3ions were detected (Fig. R6), which accounts for a FE of 13%. After considering the formation of BrO3 -, the total FE reaches 95%. Therefore, the overoxidation of Brto BrO3ions is dominantly responsible for the lost FE.
In the revised version, we have added the above results and discussion in the revised manuscript (Lines 293-298, Page 7) as "It is also notable that some BrO3were detected by ion chromatograph (IC), accounting for a FE of 13% ( Fig. S50), which indicates that the overoxidation of Brmay occur during Br-mediated epoxidation process. In addition, no O2 evolution was observed (Fig.   S7). These results imply that the loss in FE is mainly attributed to the overoxidation of Br -, rather than to the water oxidation." and revised supplementary information (Fig. S50).

Fig. R5
The GC spectra of the headspace gas of the PEC cell after 4 hours' photoelectrolysis.

Fig. R6
The IC spectra of reaction solution after 2 hours' photoelectrolysis.
3). Hydrogen peroxide is a nonnegligible product during the oxidation process of water molecules, and is also the common oxidant in the epoxidation of alkene.
And oxygen gas evolved on the anode was reduced on the Pt cathode to oxygen species, and then participating in the epoxidation reaction. Whether the author excludes the possibilities in the PEC system?
Response: According to the result of GC (Fig. R5), dioxygen is not formed during the Br-mediated epoxidation process, which means that water oxidation is much less competitive than bromide oxidation on α-Fe2O3 photoanodes, consistent with LSV results in Fig. 2a. Thus, little hydrogen peroxide should be generated during the PEC reaction. As the reviewer pointed out, hydrogen peroxide (H2O2) can act as a common oxidant for the epoxidation of alkenes in the presence of catalysts (such as titanium silicalite-1, metal-based complex).
To further exclude the possibility of participation of H2O2 in our system, we conducted the epoxidation experiment with 10 mM alkene and 10 mM H2O2 on α-Fe2O3. After 4 hours' reaction, no epoxide was detected (Fig. R7). Thus, H2O2 geneated by water oxidation can be excluded as the oxidant for the epoxidation of alkene in our system.
In the revised manuscript, the corresponding description (Lines 254-256, Page 6) as "In addition, the epoxidation in the presence of H2O2 shows poor activity on α-Fe2O3 (Fig. S46), excluding the possibility that H2O2 geneated by water oxidation can act as the oxidant for the epoxidation of alkenes." has been added in the revised main text. More detailed description is provided in the revised supplementary information (Fig. S46). 4). The authors should add the epoxidation-based incident photon to current efficiencies in the PEC system. Fig. R8, the monochromatic incident photon-toelectron conversion efficiency (IPCE) experiments were conducted. IPCE measurement showed that the visible-light activity started at ~600 nm and increased with the shortened wavelength. These efficiencies exhibited the maximum values at a wavelength of 400 nm, and increased with increasing of applied biases, which reached to ~21% at 0.55 V vs. Fc/Fc + .

Response: As shown in
To address the reviewer's concern, we have added the above results in the  Response: According to the reviewer's suggestion, we carried out the PEC oxidation of alkene in the presence of 3 mM BrO3 -(approximate saturated solubility). As shown in Fig. R9a, the corresponding selectivity and FE of epoxide were 64% and 41%, respectively, indicating that the addition of BrO3species did not obviously influence the epoxidation activity (in TBABF4 system, selectivity 43±5%, FE 41±3%). In addition, 20 mM NaBr was used to substitute TBABr to conduct PEC epoxidation reaction (Fig. R9b). An excellent selectivity (> 95%) of epoxide was achieved. These results confirmed that Brserves as a key mediator to effectively perform alkene epoxidation.
In the revised version, the corresponding description (Lines 127-129, Page 3) of "When TBABr was replaced by NaBr ( Fig. S8), an excellent selectivity (> 95%) of epoxide was also achieved." have been added in the main text. More detailed description is provided in the revised supplementary information (under the Fig. S8). 6). Similarly, the possibility of I -/IOx -, or Cl -/ClOis also to be tested to further extend the mediator scope. If NOT, the unique role of Br -/BrOshould be well discussed.
Response: According to the suggestion of the reviewer, we performed the PEC oxidation of alkene by using TBACl and TBAI as electrolyte to examine the possibility of Cl-and I-mediated epoxidation. Contrary to Br --mediator, epoxidation reactions by using both Cland Ias the mediator exhibit the poor activity and selectivity (Fig. R10). For the TBACl system, the epoxide is detected with poor performance (Fig. R10a), while in the TBAI system no oxidation of alkene is observed (Fig. R10b). The LSV experiments show that, in the system using TBACl as electrolyte, the negative shift of onset potential and the photocurrent increase relative to the case of TBABF4 are not very significant ( Fig. R11), which suggests that the oxidation of Clis not very competitive to the water oxidation on α-Fe2O3 photoanode under our experimental conditions. Therefore, poor epoxidation performance of the TBACl systems should originate from the unfavorable oxidation of Clto ClO -. For the TBAI system, the large negative shift of onset potential indicates that Iis easy to be oxidized on the α-Fe2O3 photoanode. The low epoxidation activity may be attributed to the low oxidation ability and the poor stability of IOin the TBAI system. For TBABr system, however, the Bris facile to be oxygenated to BrOspecies on α-Fe2O3 under our PEC conditions, and the formed BrOspecies is active enough to transfer its oxygen atom to the alkenes. Therefore, the Br -/BrOcycling plays an unique role in mediating the epoxidation.
In the revised version, we have added the above results and discussion in

To Reviewer #2:
In this work, the authors developed an approach to oxidize alkenes to epoxides using Br -/BrOas the mediator for electron transfer. This report is well-written and the authors provided sufficient evidence to show that alkene oxidation is by BrOrather than Br2. Isotope experiments further demonstrated that oxygen is from water oxidation.
Response: Thanks for the reviewer's helpful suggestions and comments on our manuscript. We have revised our manuscript according to the reviewer's suggestions.
1. However, a significant drawback of this paper is that this alkene epoxidation only can reach a good faradaic efficiency when alkene contains phenyl functional groups. This severely limits the application of this approach and should be mentioned and explained in the manuscript.

Response:
The epoxidation of aromatic alkenes represents an important chemical reaction, and the corresponding aromatic epoxide serves as a versatile intermediator for the application of pharmaceuticals. Thus, the synthesis of aromatic epoxide can gain wide application in many fields.
Actually, the selective epoxidation of aliphatic alkenes is quite challenging in most of the developing techniques for epoxidation. The selectivity and FE of direct PEC epoxidation of cyclooctene on bare α-Fe2O3 are only 9% and 3%, respectively. The selectivity and FE for the epoxidation of other aliphatic alkenes are even poor in these systems. For the Br-mediated EC systems, the FE values for epoxidation of aliphatic alkenes are in the range of 33-39% (details in SI, entries 6-7, Table S1). In our Br-mediated PEC systems, the selectivity and FE for the epoxidation of cyclooctene are 75% and 41%, respectively, which are much higher than those on the bare α-Fe2O3.
Moreover, the performance of the Br-mediated epoxidation of aliphatic alkenes is better than or comparable to that on MnOx and RuO2 anodes or in the Brmediated EC and PEC systems.
The relatively mediocre selectivity and FE values for epoxidation of aliphatic alkenes may stem from the relative inertness of C=C bond of aliphatic alkenes toward epoxidation, the reactivity toward other active radical species, and/or the chemical lability of aliphatic epoxide. In our systems, several by-products, including ketones, bromine-substituted products (as shown by GC-MS analysis,   2. Another problem that I found is that the authors cited ref 5 as an example of indirect electrochemical epoxidations. However, in the original paper, it was assumed that the reaction was through direct epoxidation. Thus, this should be corrected.

Response:
Ref. 5 reports that the presence of Clis able to switch off the combustion pathway of ethene to CO2 on ruthenium-based oxide electrode, and promote the epoxidation reaction channel. In this paper, such a switching effect of Clis attributed to the surface reactive sites blocking by adsorbing of Clon the surface of the RuO2 electrode. As pointed out by the reviewer, the epoxidation reaction still occurs through direct epoxidation, but not via a mediated process.
In the revised manuscript, this reference is not cited here.
Overall, this is a good paper, and I support its publication on Nat. Commun.

Response:
Thanks for the reviewer's helpful suggestions and comments on our manuscript.

To Reviewer #3:
The authors report a photoelectrochemical route through which epoxidation is mediated by Br, driving ultimate oxygen atom transfer from water to olefins.

Response:
We thank the reviewer for reviewing our manuscript, and appreciate your insightful and previous comments and suggestions. We have revised our manuscript according to the suggestions.

Can the authors articulate specifically why introducing light is important?
Response: In the electrochemical method, the halogen-mediators is oxidized directly by the external bias potential. Accordingly, a high voltage has to be applied for halogen-mediated reaction, which would be energy-extensive consuming. By contrast, in the photoelectrochemical system, the external bias potential is only used to drive the photoinduced conduction band electron to the cathode. The halogen-mediator is oxidized by the photogenerated hole left in the valence band of photoanode (Fe IV =O in the case of α-Fe2O3). Low voltage is enough to achieve the halogen-mediated epoxidation reaction. Therefore, much less electric energy is needed. In other word, in the photoelectrochemical reaction, the oxidation reaction is driven by the light, and the electric bias is only employed to enhance the separation of photoinduced carriers in the photoanode. Thus, the photoelectrochemical reaction provides a promising way to use directly the light energy from solar.
Specifically, on the α-Fe2O3 photoanode, the applied bias in the dark is 1.05 V vs. Fc/Fc + to achieve 0.6 mA cm -2 , as shown in Figs. R13 and S4. By contrast, only 0.15 V vs. Fc/Fc + of bias is needed to obtain the same current under irradiation. In addition, the electrochemical onset potential of bromide oxidation on the Pt electrode (the most common electrode) is ~ 500 mV higher than that on α-Fe2O3 with illumination (Fig. 3).
According to the reviewer's suggestion, the role of introducing light has been added and reorganized in the introduction part of revised manuscript (Lines 46-48, Page 2) as "Photoelectrochemical (PEC) techniques, which can utilize photogenerated holes/electrons to achieve chemical conversion and reduce greatly the consumption of electric energy compared with the pure electrochemical methods". Energy balance (Eb) = Energy Input (Ei) -Energy Output (Eo) In our system, energy output refers to the energy consumption from alkene (1) to epoxide (2). Therefore, the values of energy output should be a constant (1 mol of 2), regardless of electrochemical or photoelectrochemical systems. In photoelectrochemical system, the energy input is composed with electric input (Eie) and light input (Eis). In these systems, the cost of energy input is determined by the electric consumption, because the cost of light input could be from the solar energy, which is cheap and "green". The electric consumption (Eie) can be estimated by:

Eie= n × m × F × U / ηFE
Where n is the numbers of electrons (2); m is the mole of epoxide (mol); F is faradaic constant (96485 C/mol); U is applied bias (V); ηFE is the Faradic efficiency (82%) The energy difference between the EC and PEC systems is determined by difference in the applied bias (ΔU): ΔEie= n × m × F × ΔU / ηFE = 211.8 kJ/mol At 0.6 mA/cm 2 of current, the difference in the applied bias between EC and PEC systems is ΔU = 0.90 V. The energy difference is 211.8 kJ/mol, which should be supplied by the light energy.
In the revised version, the role of introducing light has been added and reorganized in the introduction part (Lines 46-48, Page 2) as "Photoelectrochemical (PEC) techniques, which can utilize photogenerated holes/electrons to achieve chemical conversion and reduce greatly the consumption of electric energy compared with the pure electrochemical methods".
That being said, the implementation of a photoelectrochemical scheme is of fundamental interest and will inspire other work in the field.
Response: Thank the reviewer for his/her precious comments and suggestions.

To Reviewer #4:
The authors present the photoelectrochemical (PEC) epoxidation of mostly aryl and diaryl substituted alkenes using alpha-Fe2O3 as photo anode and Pt as cathode and Bras redox mediator. Overall, the synthetic results for aryl and diaryl substituted alkenes are very good in terms of selectivity and yield with a significant variation in faradaic efficiency, Table 1. The authors claim that hypobromite, BrO -, is the oxidising species. I believe this research deserves publication in an organic chemistry oriented journal such as the Journal of

Organic Chemistry or Organic Letters but not in Nature Communications
Response: Thank the reviewer for the precious comments and suggestions.
The related experiment data and discussions have been added in the revised version. The responses are presented point-by-point as follows.
1. While the synthetic results for aryl and diaryl substituted alkenes are very good, the results for aliphatic cyclic and acyclic alkenes are not with mediocre selectivity (the byproducts were not given) and poor faradaic efficiency.

Response:
The epoxidation of aryl and diaryl substituted alkenes represents an important chemical reaction, and the corresponding aromatic epoxide serves as a versatile intermediator for the application of pharmaceuticals. Thus, the synthesis of aromatic epoxide can gain a wide application in many fields.
Actually, the selective epoxidation of aliphatic alkenes is quite challenging in most of the developing techniques for epoxidation. alkenes are in the range of 33-39% (details in SI, entries 6-7, Table S1). In our Br-mediated PEC systems, the selectivity and FE for the epoxidation of cyclooctene are 75% and 41%, respectively, which are much higher than those on the bare α-Fe2O3. Moreover, the performance of the Br-mediated epoxidation of aliphatic alkenes is better than or comparable to that on MnOx and RuO2 anodes or in the Br-mediated EC and PEC systems.
The relatively mediocre selectivity and FE values for epoxidation of aliphatic alkenes may stem from the relative inertness of C=C bond of aliphatic alkenes toward epoxidation, the reactivity toward other active radical species, and/or the chemical lability of aliphatic epoxide. In our systems, several by-products, including ketones, bromine-substituted products (as shown by GC-MS analysis, in the revised manuscript and supplementary information (Fig. S43).  Response: As pointed out by the reviewer, the mechanistic studies on the reactive species are challenging, due to the extreme short lifetime and low concentration of the reactive species. In our study, although the direct detection of active species is very difficult, we provide solid experimental evidence that the hypobromite species is the most possible reaction active species. Such a mechanistic assignment is based on the following experimental observations: (1) Much larger onset potential shift (between in TBABr and TBABF4 systems) on α-Fe2O3 than on the radical-involved TiO2 or Pt electrode was observed (Fig.   3), supporting the oxygen transfer pathway but excluding the radical-based mechanism for the Br --mediated epoxidation on α-Fe2O3; (2) The participation In the revised manuscript, we have added the above discussion (Lines 282-288, Page 7) as "Notably, the photocurrent in the Br-mediated epoxidation system is nearly unchanged by addition of the alkene substrate (Fig. 2a)  (b) How is hypobromite stable at pH -3, as the results in Table S3 suggests when the pKa of HOBr is 7.5.
Response: In our system, the PEC reaction was carried out in CH3CN solution with typically only 5 vol% H2O as oxygen source. Considering that the term "pH" is not applicable in the non-aqueous solution, the pH value listed in Table S4 refers to the pH of the added water to reflect relative acidity/basicity of solution, photoanode, due to its bandgap is about 2.1 eV (Fig. R1b, UV-vis diffuse spectra), the wavelengths of ~600 nm are relevant which can be used to excite the α-Fe2O3 photoanode to obtain the photo-generated holes/electrons.
The IPCE measurement (Fig. R8) showed that the visible-light activity started at ~600 nm and increased with the shortened wavelength. The efficiency exhibits a maximum value at a wavelength of 400 nm, and increased with increasing of applied biases, which reached ~21% at 0.55 V vs. Fc/Fc + .
To address the reviewer's concern, we have added the above result (Lines   have been added to support our assumption in the revised manuscript and supplementary information (Fig. S48). Brwere only 43±5% and 41±3%, which are much lower than that in Brmediated system. This epoxidation of alkenes in the absence of Brproceeds by the direct photoelectrochemical oxidation. As shown in our previous study R1 ( Fig. R15), the epoxidation can be achieved through the oxygen transfer from the high-valence iron-oxo (Fe IV =O) species to the alkenes. However, the poor compatibility between the hydrophilicity of hematite surface and the low polarity of C=C bond disfavors their interaction, and results in poor selectivity and FE.

Fig. R15
The proposed direct PEC epoxidation mechanism in TBABF4 system.
6. In some of the GC plots it looks like there are two products that are poorly separated. Why is that?
Response: It is clear that the GC plots with asymmetric peak come from the product oxide, and the asymmetry should result from the high polarity of these products. To confirm that the GC peak with broad width represents only one product, the corresponding GC plots are obtained under the identical GC conditions by using the standard sample at different concentrations. As depicted in Fig. R16, the widening of GC peaks is also observed at all sample concentrations, and good linearity between the peak area and sample concentration is obtained, confirming that the broad peak only represents one product rather than two poorly separated products.
To address the reviewer's concern, we have added the corresponding description as "The widening of GC peaks is also observed at all sample concentrations, and good linearity between the peak area and sample concentration is obtained, confirming that the broad peak only represents one product rather than two poorly separated products. Such a widening of GC peak may stem from the high polarity of product oxide." in the revised supplementary information (under the Fig. S37b). 7. The authors should also disclude the possibility of formation of benzylic cations or radical cations, followed by nucleophilic attack. CV's of some