Thermocatalytic epoxidation by cobalt sulfide inspired by the material's electrocatalytic activity for oxygen evolution reaction

New discoveries in catalysis by earth-abundant materials can be guided by leveraging knowledge across two sub-disciplines of heterogeneous catalysis: electrocatalysis and thermocatalysis. Cobalt sulfide has been reported to be a highly active electrocatalyst for the oxygen evolution reaction (OER). Under these oxidative conditions, cobalt sulfide forms oxidized surfaces that outperform directly prepared cobalt oxide in OER catalysis. We postulated that the catalytic activity of oxidized cobalt sulfide for OER could reflect a more general ability to catalyze O-transfer reactions. Herein, we show that cobalt sulfide (CoSx) indeed catalyzes the epoxidation of cyclooctene, a thermal O-transfer reaction. Similarly to OER, the surface-oxidized CoSx formed under reaction conditions outperformed the directly prepared cobalt oxide, hydroxide, and oxyhydroxide for epoxidation catalysis. Another notable phenomenological parallel to OER was revealed by the electron paramagnetic resonance (EPR) analysis of all spent Co-based catalysts that showed significant structural changes and the formation of paramagnetic Co(ii) and Co(iv) species. Mechanistic investigations suggest that a higher density of Co(ii) and/or an easier formation of high-valent Co species in the case of surface-oxidized cobalt sulfide is responsible for its high activity as an epoxidation catalyst. Our results provide important insight into the surface chemistry of Co-based catalysts and show the potential of oxidized CoSx as an earth-abundant catalyst for O-transfer reactivity beyond OER. This work highlights the utility of bridging electrocatalysis and thermocatalysis for the development of more sustainable chemical processes.


Catalytic epoxidation of cyclooctene by PhIO
To compare the catalytic activity of different materials the amount of product produced was normalized to the total amount of cobalt atoms (in mmol) in the catalytic material, as shown in Fig. 3 in the main text.In Fig. S1 we show the absolute amount of product formed to compare the reaction in presence of the different catalytic materials to the control reaction in the absence of a material.Fig. S2 shows the amount of benzaldehyde and PhI formed in presence of the different catalytic materials.Fig. S3 provides estimated TONs.Fig. S4 shows the catalytic activity of cobalt sulfides at RT compared to their catalytic activity at 80 °C.Fig. S5 shows the amount of product formed normalized with initial and post-catalytic surface areas for the different materials. .Estimated TON after 5 h reaction time for different Co-based materials from the amount of cyclooctene oxide formed and a normalization to a roughly estimated amount of cobalt exposed at the surface.The amount of surface cobalt was estimated via the measured BET surface area for the different materials and assuming uniform particle sizes, spherical particle shapes and exposure of certain crystal facets (see SI section S16 for further details).
Due to the inherent assumptions to these estimates the thereby obtained TON likely provide correct orders of magnitude but should not be taken as accurate absolute values.

PhIO2 formation
Thermal decomposition of PhIO during the epoxidation reaction led to the formation of PhI and PhIO2.This decomposition was accelerated in presence of (was catalyzed by) the herein examined Co-based materials.We recovered a white solid (mixtures with Co-based materials look grey) from the epoxidation reactions in presence and absence of Co-based materials and identified the solid as PhIO2 by pXRD and NMR spectroscopy (Fig. S6).Please also see the hazard warning for PhIO2 in the experimental section of the main text.

Successive catalytic cycles
Catalyst performance in successive cycles is difficult to evaluate for the catalytic systems used herein, because PhIO and PhIO2 cannot be separated from the spent catalyst.For some insight on the activity of the Co-based materials after exposure to catalytic conditions, we added more PhIO (220 mg, 1.00 mmol together with 0.5 mL of a solution of 3.3 mM cyclooctene and 0.02 mM 1,3,5-trimethoxybenzene in 1.5 mL toluene) to the reaction mixtures with each Co-based material after 1 h reaction time in one experiment.Fig. S7 shows the increase in product formation after addition of more PhIO at t = 1h compared to the product formation versus time under our standard catalytic conditions.To infer on each Co-based material's activity after exposure to catalytic conditions, we compared the product formation observed during the first hour of reaction (t = 0 to t = 1h) to the product formation observed during the first hour after addition of more PhIO (t = 1h to t = 2h).Based on this analysis, the catalytic activity of CoSx for epoxidation has decreased by ~25 %.This activity decrease is roughly consistent with or a bit less than what would be expected based on the surface area decrease of CoSx upon exposure to oxidative conditions of ~34% (see Fig. S15).CoSx-ox showed a similar decrease in product formation after further addition of PhIO by ~30 %.By a similar analysis Co3O4 showed increased product formation of ~18 % after addition of more PhIO, which is roughly consistent or a bit higher than what is expected based on a relatively constant surface area (Fig. S15).CoOOH showed an increased product formation by ~43 %, which is roughly consistent or a bit smaller than what is expected based on the surface area increase by 53% (Fig. S15).Co(OH)2, on the other hand, showed a decreased product formation after addition of more PhIO by 67%, despite an increase of surface area under oxidative conditions by a factor of two (Fig. S15).

Influence of a radical scavenger on the epoxidation of cyclooctene
While for t BuOOH the observed inhibition of catalysis is consistent with a free radical mechanism, the radical scavenger can also directly react with PhIO, 2 and the observed inhibition of catalysis is likely not indicative of a radical mechanism in case of PhIO.

Probing the isomerization of cis-2-octene
The commercially available cis-2-octene contained trace amounts of trans-2-octene, but this amount stayed constant after 5 h and no increase or decrease in the ratio of cis:trans olefin (as would be expected in case of isomerization catalysis) could be observed with any of the tested catalysts nor in absence of a Co-based material.Hence, thermal isomerization of cis-2-octene is not catalyzed by the Co-based materials and does not occur.Isomerization of cis-2-octene can therefore not explain the formation of E-2-methyl-3-pentyloxirane observed in the epoxidation of cis-2-octene.We also note that the trace amount of trans-2-octene present in commercially available cis-2-octene does not change during epoxidation catalysis and is likely not the source of the observed E-2-methyl-3-pentyloxirane.

S11
Cobalt sulfide materials after exposure to oxidative conditions that are similar to those of epoxidation reactions Due to the insolubility of PhIO in most solvents the catalytic materials are difficult to separate from the epoxidation reaction mixtures.To analyze the materials after exposure to oxidative conditions that are similar to those of the epoxidation catalysis, CoSx was treated with a soluble alternative of PhIO, namely 1-(tert-butylsulfonyl)2-iodosylbenzene, to give CoSx-ox-PhIO (see main text, experimental section).The XPS data of CoSx-ox-PhIO showed similar features as the XPS spectra of CoSx-ox (see main text, Figs. 2 and 4) indicating that the CoSx surface was oxidized when treated with 1-(tert-butylsulfonyl)2iodosylbenzene.Please note that the chemical nature of the CoSx-ox-PhIO surface after reaction with 1-(tert-butylsulfonyl)2-iodosylbenzene can only provide an approximation of what happens under actual epoxidation reaction conditions with PhIO, i.e.CoSx-ox-PhIO is perhaps not identical to the materials after epoxidation catalysis using PhIO.For instance, in contrast to reactions of alkenes with PhIO in the presence of the cobalt sulfides, similar reactions using 1-(tert-butylsulfonyl)2-iodosylbenzene as the oxidant did not lead to epoxide product formation.

Assessing a potential leaching of cobalt into the reaction mixture by UV-Vis spectroscopy
After completion of the catalytic epoxidation reactions with PhIO using the different Co-based materials, each reaction mixture was filtered and the solvent was removed.Then, 3 mL of an aqueous solution of disodium 3-hydroxy-4-nitrosonaphthalene-2,7-disulfonate (Nitroso-R-salt) (0.5 mM, in MQ water) was added to each reaction mixture and the resulting solutions were diluted by a factor of 10 with MQ water.UV-Vis spectra were then recorded of each mixture and compared to a 0.05 mM solution of the Nitroso-R-salt in MQ water and to a 0.05 mM solution of Nitroso-R-Salt in MQ water also containing 0.005 mM Co(OAc)2•4H2O as a representative sample of a solution containing Co(II)-ions (Fig. S12).The Nitroso-R-salt forms a complex with cobalt which has an absorption maximum at 395 nm and can be best identified (distinguished from the absorption of the free Nitroso-R-salt) by the absorption at 500 nm. 3 Co(II) can be identified in the last solution, in which Co(OAc)2•4H2O has been added on purpose.None of the reaction mixtures from the epoxidation catalysis with different Co-based materials showed an absorption characteristic of cobalt suggesting that no cobalt leaches from the catalyst or that leaching occurs only in quantities below the detection limit.Based on our control with Co(OAc) 2 •4H 2 O we would be able to detect the leaching of 0.1 % of the cobalt content in the catalysts used (typically 0.11 mmol cobalt) and probably much less.S3.This table evaluates the rough radius particles of CoSx, CoSx-ox, Co3O4, Co(OH)2, and CoOOH would need to have in order explain the measured BET surface area assuming uniform, spherical and smooth particles.We find that the particles would have to be much smaller than what we found by SEM (Fig. S14) with radii on the order of roughly 10-50 nm for the different Co-based materials.These particle radii are hence inconsistent with our results and suggest that the particles have significant surface roughness (see Table S4).

Estimated fraction of cobalt atoms at the surface
Determining the fraction of cobalt atoms at the surface of the different CoSx, CoSx-ox, Co3O4, Co(OH)2 and CoOOH is challenging due to the amorphous nature and large size and shape distribution of the corresponding particles.We derive a first, rough estimate of the fraction of cobalt atoms at the surface for the different materials from the measured BET surface area (Fig. S15) and assuming uniform, smooth, and spherical particles with one specific crystal facet mainly exposed.Physical properties and crystallographic data were retrieved from the Inorganic Crystal Structure Database (ICSD).This is shown for CoSx in Fig. S16 and Table S5, and similar considerations were applied for the other Cobased materials.

S21
Discussion of the contribution of carbon tape to XPS spectra.Carbon tape was used as a support for XPS analysis of the different materials.XPS analysis of this carbon tape showed that we should expect contributions to both C1s and O1s spectra from the carbon tape with a C/O ratio of roughly 1:0.4 (Table S6 and Fig. S18).Therefore, any O1s signal observed in the analysis of our materials will be convoluted with oxygen present on the carbon paper support.By determining the ratio of carbon to oxygen on the surface of our materials, we tried to account for the contribution of carbon paper to the observed O1s signal for the different materials.For example, XPS analysis of CoSx suggested a C:O ratio of ~1:0.9.Hence, to a first approximation, up to roughly half of the observed O1s signal may be due to the carbon tape support and only roughly half may be due to oxygen on the surface of CoSx.However, such considerations can only give a rough idea of the contribution of the carbon tape to the O1s signals measured for the Co-based materials, because the oxygen content on carbon paper can vary even for the same type of carbon paper, and because the XPS data may also be influenced by the specific amount and distribution of deposited sample on the carbon support.Therefore, O1s data of all materials examined herein should be interpreted with care.The fits shown in the main text of the O1s XPS spectra include a signal from carbon tape at ~531.6 eV, where the O1s maximum was found for the bare carbon tape (Fig. S18), and thereby account for the contribution from the carbon tape.NMR spectroscopy data from the epoxidation of cyclooctene by PhIO Fig. S20. 1 H-NMR spectra (400 MHz, MeCN-d3) of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by PhIO using CoSx-ox.Samples were taken at t = 0, t = 5 min, t = 15 min, t = 30 min, t = 1 h, t = 3 h and t = 5 h (from bottom to top).The signal at 2.84 ppm was used to quantify the amount of cyclooctene oxide formed and corresponds to the two protons of the oxirane group.The signal at ~7.7 ppm was used to quantify the amount of PhI, corresponding the two aromatic protons at 2 and 6 position.The signal at 7.4 ppm corresponds to one aromatic proton at the 4 position.

Fig. S21
. 1 H-NMR spectra (400 MHz, MeCN-d3) of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by PhIO using CoSx.Samples were taken at t = 0, t = 5 min, t = 15 min, t = 30 min, t = 1 h, t = 3 h and t = 5 h (from bottom to top).The signal at 2.84 ppm was used to quantify the amount of cyclooctene oxide formed and corresponds to the two protons of the oxirane group.The signal at ~7.7 ppm was used to quantify the amount of PhI, corresponding the two aromatic protons at 2 and 6 position.The signal at 7.4 ppm corresponds to one aromatic proton at the 4 position.

S24
Fig. S22. 1 H-NMR spectra (400 MHz, MeCN-d3) of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by PhIO using Co3O4.Samples were taken at t = 0, t = 5 min, t = 1 h and t = 5 h (from bottom to top).The signal at 2.84 ppm was used to quantify the amount of cyclooctene oxide formed and corresponds to the two protons of the oxirane group.The signal at ~7.7 ppm was used to quantify the amount of PhI, corresponding the two aromatic protons at 2 and 6 position.The signal at 7.4 ppm corresponds to one aromatic proton at the 4 position.Fig. S23. 1 H-NMR spectra (400 MHz, MeCN-d3) of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by PhIO using CoOOH.Samples were taken at t = 0, t = 5 min, t = 15 min, t = 30 min, t = 1 h, t = 3 h and t = 5 h (from bottom to top).The signal at 2.84 ppm was used to quantify the amount of cyclooctene oxide formed and corresponds to the two protons of the oxirane group.The signal at ~7.7 ppm was used to quantify the amount of PhI, corresponding the two aromatic protons at 2 and 6 position.The signal at 7.4 ppm corresponds to one aromatic proton at the 4 position.

S25
Fig. S24. 1 H-NMR spectra (400 MHz, MeCN-d3) of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by PhIO using Co(OH)2.Samples were taken at t = 0, t = 5 min, t = 15 min, t = 30min, t = 1 h, t = 3 h and t = 5 h (from bottom to top).The signal at 2.84 ppm was used to quantify the amount of cyclooctene oxide formed and corresponds to the two protons of the oxirane group.The signal at ~7.7 ppm was used to quantify the amount of PhI, corresponding the two aromatic protons at 2 and 6 position.The signal at 7.4 ppm corresponds to one aromatic proton at the 4 position.Fig. S25. 1 H-NMR (400 MHz, MeCN-d3) spectra of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by PhIO in the absence of a catalyst.Samples were taken at t = 0, t = 5 min, t = 15 min, t = 30 min, t = 1 h, t = 3 h and t = 5 h (from bottom to top).The signal at 2.84 ppm was used to quantify the amount of cyclooctene oxide formed and corresponds to the two protons of the oxirane group.The signal at ~7.7 ppm was used to quantify the amount of PhI, corresponding the two aromatic protons at 2 and 6 position.The signal at 7.4 ppm corresponds to one aromatic proton at the 4 position.S26. 1 H-NMR (500 MHz, MeCN-d3) spectra of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by t BuOOH using CoS x .Samples were taken at t = 0, t = 5 min and t = 5 h (from bottom to top).NMR signals marked with colored dots have been used to quantify the different products formed.The signal at 2.87 ppm was used to quantify the amount of cyclooctene oxide (red dot) formed and corresponds to the two protons of the oxirane group.The signal at 2.62 ppm was used to quantify the amount of cyclooct-2-en-1one (green dot) and corresponds to two methylene protons in alpha position to the C=O group.The singlet signals at 1.25 and 1.26 ppm correspond to tert-butyl groups of 3-(tert-butylperoxy)cyclooct-1-ene (blue dot) and ((tertbutylperoxy)methyl)benzene (yellow dot), respectively, and were used for their quantification.The singlet signal at 10 ppm was used to quantify the amount of benzaldehyde (purple dot) and the signal at 4.6 ppm was used to quantify the amount of benzyl alcohol (grey dot).Due to the complexity of the obtained NMR spectra we also used GC-MS to assist the identification of the formed products.

S27
Fig. S27. 1 H-NMR (500 MHz, MeCN-d3) spectra of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by t BuOOH using CoSx-ox.Samples were taken at t = 0, t = 5 min and t = 5 h (from bottom to top).NMR signals marked with colored dots have been used to quantify the different products formed.The signal at 2.87 ppm was used to quantify the amount of cyclooctene oxide (red dot) formed and corresponds to the two protons of the oxirane group.The signal at 2.62 ppm was used to quantify the amount of cyclooct-2-en-1-one (green dot) and corresponds to two methylene protons in alpha position to the C=O group.The singlet signals at 1.25 and 1.26 ppm correspond to tert-butyl groups of 3-(tert-butylperoxy)cyclooct-1-ene (blue dot) and ((tert-butylperoxy)methyl)benzene (yellow dot), respectively, and were used for their quantification.The singlet signal at 10 ppm was used to quantify the amount of benzaldehyde (purple dot) and the signal at 4.6 ppm was used to quantify the amount of benzyl alcohol (grey dot).Due to the complexity of the obtained NMR spectra we also used GC-MS to assist the identification of the formed products.

S28
Fig. S28. 1 H-NMR (500 MHz, MeCN-d3) spectra of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by t BuOOH using Co3O4.Samples were taken at t = 0, t = 5 min and t = 5 h (from bottom to top).NMR signals marked with colored dots have been used to quantify the different products formed.The signal at 2.87 ppm was used to quantify the amount of cyclooctene oxide (red dot) formed and corresponds to the two protons of the oxirane group.The signal at 2.62 ppm was used to quantify the amount of cyclooct-2-en-1-one (green dot) and corresponds to two methylene protons in alpha position to the C=O group.The singlet signals at 1.25 and 1.26 ppm correspond to tert-butyl groups of 3-(tert-butylperoxy)cyclooct-1-ene (blue dot) and ((tert-butylperoxy)methyl)benzene (yellow dot), respectively, and were used for their quantification.The singlet signal at 10 ppm was used to quantify the amount of benzaldehyde (purple dot) and the signal at 4.6 ppm was used to quantify the amount of benzyl alcohol (grey dot).Due to the complexity of the obtained NMR spectra we also used GC-MS to assist the identification of the formed products.

S29
Fig. S29. 1 H-NMR (500 MHz, MeCN-d3) spectra of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by t BuOOH using CoOOH.Samples were taken at t = 0, t = 5 min and t = 5 h (from bottom to top).NMR signals marked with colored dots have been used to quantify the different products formed.The signal at 2.87 ppm was used to quantify the amount of cyclooctene oxide (red dot) formed and corresponds to the two protons of the oxirane group.The singlet signals at 1.25 and 1.26 ppm correspond to tertbutyl groups of 3-(tert-butylperoxy)cyclooct-1-ene (blue dot) and ((tert-butylperoxy)methyl)benzene (yellow dot), respectively, and were used for their quantification.The singlet signal at 10 ppm was used to quantify the amount of benzaldehyde (purple dot).Due to the complexity of the obtained NMR spectra we also used GC-MS to assist the identification of the formed products.S30. 1 H-NMR (500 MHz, MeCN-d3) spectra of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by t BuOOH using Co(OH)2.Samples were taken at t = 0, t = 5 min and t = 5 h (from bottom to top).NMR signals marked with colored dots have been used to quantify the different products formed.The signal at 2.87 ppm was used to quantify the amount of cyclooctene oxide (red dot) formed and corresponds to the two protons of the oxirane group.The signal at 2.62 ppm was used to quantify the amount of cyclooct-2-en-1-one (green dot) and corresponds to two methylene protons in alpha position to the C=O group.The singlet signals at 1.25 and 1.26 ppm correspond to tert-butyl groups of 3-(tert-butylperoxy)cyclooct-1-ene (blue dot) and ((tert-butylperoxy)methyl)benzene (yellow dot), respectively, and were used for their quantification.The singlet signal at 10 ppm was used to quantify the amount of benzaldehyde (purple dot) and the signal at 4.6 ppm was used to quantify the amount of benzyl alcohol (grey dot).Due to the complexity of the obtained NMR spectra we also used GC-MS to assist the identification of the formed products.

S31
Fig. S31. 1 H-NMR (500 MHz, MeCN-d3) spectra of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by t BuOOH in absence of a material.Samples were taken at t = 0, t = 5 min and t = 5 h (from bottom to top).NMR signals marked with colored dots have been used to quantify the different products formed.The signal at 2.87 ppm was used to quantify the amount of cyclooctene oxide (red dot) formed and corresponds to the two protons of the oxirane group.The signal at 2.62 ppm was used to quantify the amount of cyclooct-2-en-1-one (green dot) and corresponds to two methylene protons in alpha position to the C=O group.The singlet signals at 1.25 and 1.26 ppm correspond to tert-butyl groups of 3-(tert-butylperoxy)cyclooct-1-ene (blue dot) and ((tert-butylperoxy)methyl)benzene (yellow dot), respectively, and were used for their quantification.Due to the complexity of the obtained NMR spectra we also used GC-MS to assist the identification of the formed products., and a control in the absence of a catalyst after 5 h reaction time (from top to second from the bottom).The red spectrum on the very bottom provides a representative spectrum at the beginning of the epoxidation catalysis, here using Co(OH)2 (with t = 1 min).The quartet-doublet at 2.7 ppm was used to quantify the amount of E-2-methyl-3-pentyloxirane formed and corresponds to the one proton of the oxirane group at the 2 position.The triplet-doublet at 2.6 ppm corresponds to the one proton of the oxirane group at the 3 position.A baseline correction was conducted prior to integration.The signal at ~7.7 ppm was used to quantify the amount of PhI, corresponding the two aromatic protons at 2 and 6 position.The signal at 7.4 ppm corresponds to one aromatic proton at the 4 position., and a control in the absence of a catalyst after 5 h reaction time (from top to the second from the bottom).The red spectrum on the very bottom provides a representative spectrum at the beginning of the epoxidation catalysis, here using Co(OH)2 (with t = 1 min).The signal at 3.0 ppm was used to quantify the amount of Z-2-methyl-3-pentyloxirane formed and corresponds to the one proton of the oxirane group at the 2 position.The signal at 2.84 ppm corresponds to the one proton of the oxirane group at the 3 position.The doublet signal at 1.23 ppm corresponds to the methyl group at the 1 position.The signal at 2.7 ppm was used to quantify the amount of E-2-methyl-3-pentyloxirane formed and corresponds to the one proton of the oxirane group at the 2 position.A baseline correction was conducted prior to integration.The signal at ~7.7 ppm was used to quantify the amount of PhI, corresponding the two aromatic protons at 2 and 6 position.The signal at 7.4 ppm corresponds to one aromatic proton at the 4 position.

Plausible mechanism of the epoxidation of cyclooctene by t BuOOH
5][6] It has previously been suggested that either the O-O or the Co-O bond can be cleaved in the Co-OO t Bu, where the latter is thought more likely in presence of a rigid coordination chemistry and steric shielding at the Co center. 4e t BuOO and t BuO radicals formed via the Haber-Weiss mechanism likely lead predominantly to the observed allylic oxidation products (and the benzylic oxidation products from the solvent toluene), and also to the formation of cyclooctene oxide (Scheme S2). 4 Substrate oxidation by the freely diffusing peroxy and oxy radicals is consistent with the observed inhibition of (ep)oxidation catalysis in presence of a radical scavenger (Fig. S10).Scheme S1.Plausible activation pathway of t BuOOH by cobalt-based catalysts via the Haber-Weiss mechanism. 2,4,7heme S2.Plausible mechanism for the (ep)oxidation of cyclooctene by freely diffusing t BuOO (and/or t BuO) radicals. 2,4,7

Fig. S2 .
Fig. S1.a) Absolute amount of the formed cyclooctene oxide using CoSx, CoSx-ox, Co3O4, Co(OH)2, or CoOOH, or in the absence of a catalyst after 5 h reaction time.Here we show absolute amount of product formed to compare catalytic performance of the different materials versus the reaction in absence of a material.In the main text we show a similar figure normalized to the total Co content in the reaction mixture to compare different catalysts with each other (Fig. 3 in the main text).b) Yield of cyclooctene oxide in % based on the initial concentration of the oxidant PhIO added, plotted versus reaction time

Fig. S5 .
Fig. S5.Amount of cyclooctene oxide formed in the epoxidation of cyclooctene with PhIO using different Cobased materials normalized to the surface area of the as-prepared materials (Fig. S15) after (a) 5 min and (b) 5 h reaction time; and amount of cyclooctene oxide formed normalized by the surface area obtained after exposure to oxidative reaction conditions (Fig. S15) after (c) 5 min and (d) 5 h reaction time.These data show that similar qualitative conclusions are obtained by normalization to surface area as by normalization to total Co content, as discussed in the main text: CoSx and CoSx-ox outperform Co3O4, CoOOH, and Co(OH)2 in the epoxidation of cyclooctene using PhIO.
Fig.S6.a) pXRD patterns of the recovered PhIO2 (mixed with CoSx-ox) from the epoxidation of cyclooctene with PhIO (purple), the as-synthesized PhIO used for catalyis (red), and the computed patterns of alpha PhIO2 (blue) and beta PhIO2 (green) retrieved from the Cambridge Crystallographic Data Centre (CCDC) Crystallographic Database accessed on 07.03.2024.The comparison shows that the recovered solid formed during the reaction is distinct from the used PhIO and can be identified as predominantly alpha PhIO2.b)1 H NMR (500 MHz, DMSO-d6) spectrum of the recovered PhIO2 showing δ/ppm: 8.01 -7.92 (m, 2H), 7.62 -7.51 (m, 3H).Comparison to literature spectra is consistent with the identity of the white solid being PhIO2.1

Catalytic epoxidation of cyclooctene by t BuOOH Fig. S8 .
Fig. S8.Absolute amount of cyclooctene oxide in mmol formed in the epoxidation of cyclooctene using t BuOOH after (a) 5 min and (b) 5 h reaction time for each Co-based material and in the absence of a catalyst.The amount of cyclooctene oxide formed normalized to the total cobalt content in each material after (c) 5 min and (d) 5 h reaction time.

Fig. S10 .
Fig. S10.Cyclooctene oxide formation during the epoxidation of cyclooctene normalized to the total amount of cobalt in the catalyst material (black) and with addition of radical inhibitor 4-tert-butylphenol at t = 30 min (red) using (a) PhIO and (b) t BuOOH as the oxidant.

Fig. S11 .
Fig. S11.Thermal epoxidation of cis-2-octene (a,b) or trans-2-octene (c,d) with PhIO using CoSx, CoSx-ox, Co3O4, CoOOH, or Co(OH)2 as catalyst.(a,c) show the absolute amount of products formed (Z-2-methyl-3-pentyloxirane (orange) and E-2-methyl-3-pentyloxirane (green)) after 5 h reaction time in order to compare the catalytic performance of the different materials to the uncatalyzed reaction in the absence of a material.In the main text we show a similar figure normalized to the total Co content in the reaction mixture to compare different catalysts with each other (Fig. 5 in the main text).(b,d) show the selectivity with which the E-or Z-epoxide are formed for different materials without first subtracting the amount of products formed in the uncatalyzed reaction.Similar plots with such a subtraction are shown in Fig. 5 in the main text.

Fig. S12 .
Fig. S12.UV-Vis spectra of the filtered and dried reaction mixtures after epoxidation catalysis using different Cobased materials in a 0.05 mM solution of Nitroso-R-Salt in MQ water and comparison to a pure 0.05 mM solution of Nitroso-R-Salt in MQ water, and a 0.05 mM solution of Nitroso-R-Salt in MQ water also containing 0.005 mM Co(OAc)2•4H2O.

S15
Energy-dispersive X-ray spectroscopy (EDX) dataTableS2.EDX data for CoSx-ox, CoSx, CoSx-ox-PhIO, Co3O4, Co(OH)2, and CoOOH.All EDX samples were prepared in air for the measurements.Hence, the oxygen content detected by EDX, especially for cobalt sulfidebased samples (CoSx-ox, CoSx, CoSx-ox-PhIO), may have resulted from oxidation in air.Since the Co-based materials were otherwise handled under inert atmosphere in this work, the oxygen content measured by EDX for CoSx-ox, CoSx, CoSx-ox-PhIO should be interpreted with care.EDX measurements also show carbon and sometimes Al signals that probably stem from the carbon tape and Al-holder used to support the samples.The carbon and Al contents account for the missing percentages to give a total of 100 %.The carbon and Al contents have not been included in this table.

Fig. S14 .N2Fig. S15 .
Fig. S14.SEM images of a) CoS x , b) CoS x -ox, c) Co 3 O 4 , d) Co(OH) 2 , e) CoOOH and f) CoS x -ox-PhIO.The SEM images show that all examined materials have a large particle size distribution and contain large particles with sizes ranging roughly between few and 500 µm.In the main text we show SEM images of the Co-based materials with larger magnification to highlight surface morphology.

Fig. S16 .Thermocatalytic
Fig. S16.Crystal structure of the unit cell of Co9S8.The computed crystal plane (311) (red) cuts through one single cobalt atom.Computations were conducted using the program Mercury.

Thermocatalytic
Fig. S26.1 H-NMR (500 MHz, MeCN-d3) spectra of the reaction solution from a representative epoxidation of cyclooctene to cyclooctene oxide by t BuOOH using CoS x .Samples were taken at t = 0, t = 5 min and t = 5 h (from bottom to top).NMR signals marked with colored dots have been used to quantify the different products formed.The signal at 2.87 ppm was used to quantify the amount of cyclooctene oxide (red dot) formed and corresponds to the two protons of the oxirane group.The signal at 2.62 ppm was used to quantify the amount of cyclooct-2-en-1one (green dot) and corresponds to two methylene protons in alpha position to the C=O group.The singlet signals at 1.25 and 1.26 ppm correspond to tert-butyl groups of 3-(tert-butylperoxy)cyclooct-1-ene (blue dot) and ((tertbutylperoxy)methyl)benzene (yellow dot), respectively, and were used for their quantification.The singlet signal at 10 ppm was used to quantify the amount of benzaldehyde (purple dot) and the signal at 4.6 ppm was used to quantify the amount of benzyl alcohol (grey dot).Due to the complexity of the obtained NMR spectra we also used GC-MS to assist the identification of the formed products.

Thermocatalytic
Fig. S32.1 H-NMR (500 MHz, MeCN-d3) spectra of the reaction solution from a representative epoxidation of trans-2-octene by CoS x , CoS x -ox, Co 3 O 4 , CoOOH, Co(OH) 2 , and a control in the absence of a catalyst after 5 h reaction time (from top to second from the bottom).The red spectrum on the very bottom provides a representative spectrum at the beginning of the epoxidation catalysis, here using Co(OH)2 (with t = 1 min).The quartet-doublet at 2.7 ppm was used to quantify the amount of E-2-methyl-3-pentyloxirane formed and corresponds to the one proton of the oxirane group at the 2 position.The triplet-doublet at 2.6 ppm corresponds to the one proton of the oxirane group at the 3 position.A baseline correction was conducted prior to integration.The signal at ~7.7 ppm was used to quantify the amount of PhI, corresponding the two aromatic protons at 2 and 6 position.The signal at 7.4 ppm corresponds to one aromatic proton at the 4 position.

Thermocatalytic
Fig. S33.1 H-NMR (500 MHz, MeCN-d3) spectra of the reaction solution from a representative epoxidation of cis-2-octene by CoSx, CoSx-ox, Co3O4, CoOOH, Co(OH)2, and a control in the absence of a catalyst after 5 h reaction time (from top to the second from the bottom).The red spectrum on the very bottom provides a representative spectrum at the beginning of the epoxidation catalysis, here using Co(OH)2 (with t = 1 min).The signal at 3.0 ppm was used to quantify the amount of Z-2-methyl-3-pentyloxirane formed and corresponds to the one proton of the oxirane group at the 2 position.The signal at 2.84 ppm corresponds to the one proton of the oxirane group at the 3 position.The doublet signal at 1.23 ppm corresponds to the methyl group at the 1 position.The signal at 2.7 ppm was used to quantify the amount of E-2-methyl-3-pentyloxirane formed and corresponds to the one proton of the oxirane group at the 2 position.A baseline correction was conducted prior to integration.The signal at ~7.7 ppm was used to quantify the amount of PhI, corresponding the two aromatic protons at 2 and 6 position.The signal at 7.4 ppm corresponds to one aromatic proton at the 4 position.

ThermocatalyticThermocatalytic
Fig. S34.EPR spectra of as-prepared Co3O4 (red), CoOOH (purple), and Co(OH)2 (pink).In the main text (Fig.1c) we show the same data, but with the EPR spectra of as-prepared CoOOH and Co(OH)2 vertically scaled to show minor signals.
Amount of cyclooctene formed with respect to total amount of Co in the material versus time from at least three replicate experiments (solid lines) is compared to experiments, where after 1 h reaction time more PhIO has been added (dotted lines).The data of the solid lines at 5 h reaction time is shown in the main text, Fig.3.
Thermocatalytic Epoxidation by Cobalt Sulfide Inspired by the Wyss, Delley et al.Material's Electrocatalytic Activity for Oxygen Evolution Reaction

Table S4 .
Estimated roughness factor for the surfaces of CoSx, CoSx-ox, Co3O4, Co(OH)2, and CoOOH assuming the majority of particles have radii of 10 m or larger (based on SEM data, Fig.S14) and assuming spherical particles.We find roughness factors on the order of at least 100-1000 are needed to explain the measured surface areas.Due to the applied assumptions that do not perfectly represent the complex size and shape distribution of particles found experimentally, the derived roughness factors given in the table should only be taken as rough estimates and not as accurate values.Thermocatalytic Epoxidation by Cobalt Sulfide Inspired by the Wyss, Delley et al.Material's Electrocatalytic Activity for Oxygen Evolution Reaction

Table S6 .
Ratio of the integrated C1s and O1s XPS signals for carbon paper, CoSx, CoSx-ox, Co3O4, Co(OH)2, and CoOOH.Note that XPS data of CoSx-ox was measured using a different type of carbon paper support that was stored under air and that may contain a higher surface oxygen content.