Application of Prussian Blue Analogue‐Derived Mn−Co Catalysts in the Hydrogenation of CO to Higher Alcohols

MnxCo3‐x[Co(CN)6]2 Prussian blue analogues were synthesized with Mn : Co ratios in the range from 7 : 10 to 1 : 11 and pyrolyzed at 600 °C to obtain a highly nitrogen‐ and oxygen‐functionalized carbon matrix with embedded reduced Mn and Co species. These catalyst precursors were applied in the CO hydrogenation to higher alcohols at 260 °C and a pressure of 60 bar using a H2/CO ratio of 1. Metallic Co0 formed during pyrolysis was partially transformed into Co2C under reaction conditions. With increasing Co content CO conversion increased up to 10.6 % reaching a total alcohol selectivity of 19 %. Gas chromatograms revealed the expected formation of primary short‐chain alcohols, but also of secondary alcohols, acetic acid and propionaldehyde indicating olefin hydration, carbonylation and hydroformylation as reaction pathways, respectively. The obtained hydrocarbon fractions had a very high olefinicity, which is beneficial for both olefin hydration to secondary alcohols catalyzed by adsorbed carboxylic acids and for hydroformylation. Whereas the carbide‐based reaction pathway and the reductive hydroformylation are assumed to occur at the Co2C/Co0 interface, carbonylation is presumably catalyzed by an additional Co‐based active site. Thus, a unique class of multifunctional catalysts was obtained with highly promising properties bridging the gap between heterogeneous and homogeneous catalysis.


Introduction
The heterogeneously catalyzed hydrogenation of carbon monoxide to higher alcohols (HAS) is a promising process to synthesize alcohols and olefines as chemical feedstock and fuel additives. [1,2] This direct synthesis of alcohols over transition metal-based catalysts has the potential to decrease the worldwide CO 2 emissions due to the production of CO 2neutral fuels, provided that CO and H 2 originate from renewable sources. [3] Several suitable CO hydrogenation catalysts have been studied in the past, with a special focus on Co-Cu-based [4][5][6][7] systems, which combine methanol (MeOH) synthesis (Equation 1) with Fischer-Tropsch synthesis (FTS, Equations 2 and 3), and also on Rh-based [8][9][10] systems. Combining these pathways leads to the HAS (Equation 4). [11,12] CO þ 2H 2 ! CH 3 OH (1) The key step of FTS and MeOH synthesis is the adsorption of CO: by adsorbing CO dissociatively, the growth of the surface intermediates occurs via surface polymerization. [13,14] These intermediates are C x H y species, which undergo different reaction pathways, either resulting in the formation of hydrocarbons or oxygenates. In addition to dissociative CO adsorption, CO needs to be adsorbed associatively as well. CO ads is assumed to be inserted into the growing hydrocarbon chain, resulting in the corresponding long-chain alcohols after hydrogenation. This carbide-based reaction pathway is well established for hydrotalcite-based Co-Cu catalysts with a Co : Cu ratio of 2 : 1 (2CoCu) as confirmed by steady-state kinetic investigations. [4,5,15,16] This reaction pathway is based on the same growth intermediates for the formation of oxygenates and hydrocarbons. By further investigation of the product distribution using the Anderson-Schulz-Flory (ASF) distribution, essentially the same chain growth probabilities were obtained, leading to the conclusion that other possible reaction pathways can be neglected for the HAS using the 2CoCu catalyst. [5] Still, reactions like the water-gas shift reaction (WGSR, Equation 5), the carbidization of bulk metal carbides (Equation 6) and the Boudouard equilibrium (Equation 7) must be taken into account. [11,17,18] CO þ H 2 O Ð CO 2 þ H 2 (5) 2CO Ð CO 2 þ C In contrast to the 2CoCu-based catalyst, Rh-based catalysts show a more complicated reaction network. Several studies demonstrated that the application of Rh-based catalysts leads to a significant selectivity towards acetic acid and aldehydes. These intermediates originate from the carbonylation of methanol and the hydroformylation of olefines, respectively. [10,19,20] The key aspects of these mechanisms will be summarized briefly, starting with the hydroformylation, which comprises three steps: [21,22] The first step is the addition of an olefin to the metal carbonyl hydride, defining whether the linear or branched product is obtained. Then, a CO ligand is migrating from the metal complex to the hydrocarbon chain in the second step followed by the hydrogenolysis of the M-alkyl bond. [21] To achieve a high activity, the polarity of the M-H bond is important. Complexes with an acidic character lead to an easier hydrogenolysis in the third step while also favoring the first step. Due to this observed benefit, the HCo(CO) 4 complex is one of the most important hydroformylation catalysts because of its strong acidic character. [23] Depending on the reaction conditions, the resulting aldehyde can be further hydrogenated leading to the formation of the corresponding branched or linear alcohol. Due to the fast hydrogenation over Co-based catalysts, this consecutive reaction is especially important for these systems. [24] A successful combination of the carbide and hydroformylation pathways has already been achieved by Kruse and coworkers, [25] who investigated alkali-promoted Co-Mn-based catalysts derived from oxalate precursors. Based on several studies focusing on the olefin selectivity of similar systems, [26][27][28] the group discovered that these Co-Mn-based catalysts are not only FTS catalysts, but can also be applied in HAS. An oxygenate selectivity of up to 25 % was achieved with the unpromoted catalyst, and the addition of potassium led to an increasing selectivity as well as CO conversion. Adding potassium led to the neutralization of acidic site, which resulted in an increased oxygenate selectivity. [18] Furthermore, alkali metals in general suppress the formation of methane due to a higher carbon chain growth probability. [29,30] Not only potassium is an important HAS promoter, but also manganese, which is beneficial for the formation of Co 2 C that has been identified as a key species for the formation of higher alcohols due to the preferred insertion of CO into the carbon chain at the Co 2 C/Co 0 interface. [31][32][33] H 2 adsorption is assumed to be suppressed by Mn, which favors the dissociative adsorption of CO, facilitating carbidization as well as the formation of olefins. [31] The remaining well-known mechanism is the carbonylation of methanol leading to higher alcohols. This reaction is usually homogenously catalyzed using a square-planar nucleophilic Rh(I) anion. [34] To transfer this process to a heterogeneously catalyzed reaction, supported Rh [35][36][37] , Ir [38][39][40] , Au [41,42] or Ni carbonyl complexes [43][44][45] were investigated in the past due to their similarity to the in-depth analyzed homogenous systems. One of the most efficient ways to improve the carbonylation activity of the different complexes has been the enrichment of electron density of the applied transition metal. [46][47][48] Combining these reaction pathways could lead to a catalyst which is able to catalyze the HAS on an industrial scale. This work describes HAS using Prussian blue analogue (PBA) precursors, which have not yet been applied in the HAS to the best of our knowledge. The Mn-Co-based PBAs have been studied extensively as catalyst precursors for electrochemical reactions. [49][50][51] PBAs are derived from the historical Prussian blue (Fe 4 ½Fe CN ð Þ 6 � 3 ), which has been deemed the first developed coordination polymer. [52] PBAs are obtained by replacing Fe with Co, Ni, Mn, Cu, V or Cr, which is not affecting the crystal structure of the original PB shown in Figure 1. [49,50,53,54] PBAs have a face-centered cubic structure with the space group Fm3 m. [55] The transition metals are placed at the carbon-coordinated R site (green) and the nitrogen-coordinated P site (dark-blue). Thus, the metal ions are coordinated by these cyanide groups resulting in a stoichiometric ratio of C : N = 1 : 1. The lattice parameters depend strongly on these transition metals, especially on the transition metal placed at the R site. [56] Furthermore, the stability of PBAs is influenced by the defect sites, with the vacancy sites inside of the crystal structure being the most important ones. This vacancy can be filled by alkali cations with the goal to influence the electron density of the unit cell resulting in its contraction. [49,57] Exceeding a certain amount of inserted alkali cations results in a distortion of the crystal structure to a less symmetric rhombohedral geometry. [49] In addition to alkali cations, which can coordinate exclusively at the A site, water is also an important structural component. [58,59] Water can either be present in the form of zeolitic water, which is competing for the A site with the alkali metal, or as coordinated water, which is coordinated to the vacancies at the P site. [60] A rather straightforward approach to synthesize these PBAs is based on co-precipitation using aqueous Co and Mn acetate and K 3 [Co(CN) 6 ] solutions. While this synthesis has been performed with water as the solvent in many studies, it is known in literature that by using a mixture of EtOH and water the morphology and structural properties of the PBAs can be influenced. [52,61] By applying the microemulsion approach, Hu and coworkers [61] observed the formation of Mn-Co-based nanocubes. Switching to water as the solvent resulted in irregularly shaped microframes with an increase in particle size. The authors concluded that by applying an EtOHcontaining solution, the nucleation process is slowed down due to the decreased mobility of the M 2 + ions. Therefore, the growth process of the nanocubes is retarded resulting in a controlled formation rate. This effect was further enhanced by the addition of polyvinylpyrrolidone (PVP) acting as a surfactant. As with the influence of the solvent, the authors observed that without the addition of the surfactant, the size of the nanocubes increased and irregularly shaped particles were obtained. The capping agent PVP is assumed to be weakly coordinated to the M 2 + ions resulting in a steric stabilization and allowing the formation of nanocubes protected by PVP. [61] While the structure in Figure 1 is resulting from an A x P y ½R K CN ð Þ 6 � 1À y compound, recent studies show that it is possible to exceed the limit of two transition metals by developing PBAs with concentration gradients of two different P transition metals. [62,63] For example, Jeon et. al. [63] investigated PBAs with several different transition metals at the P-site.

Characterization of the precursors
The structure of the dried catalyst precursors before as well as after pyrolysis was investigated by XRD. Figure 2 shows the obtained patterns of the different investigated states of the catalysts, which are labelled according to the molar Mn : Co ratios determined by elemental analysis. The applied references are shown in Table S1. Figure 2a shows that the coprecipitation route resulted in phase-pure precursors. Very distinct reflections can be observed at 2θ = 17.1, 24.2, 34.5 and 38.7°originating from the mixed Mn x Co 3-x [Co(CN) 6 ] 2 PBA precursor. By adding Co acetate to the synthesis, Mn-Co-PBA precursors with increasing Co content were obtained as indicated by a shift of the reflections. Figure 2b shows this shift for the reflection at 2θ = 24.2°. The addition of Co(Ac) 2 led to a shift towards higher 2θ values, resulting in a reflection at 2θ = 24.7°for the Co 3 [Co(CN) 6 ] 2 precursor. Figure 2b shows clearly that phase-pure PBA precursors were obtained. The XRD pattern of the pyrolyzed Mn 10 Co 6 sample shown in Figure 2c displays strong reflections originating from Mn-based phases. Reflections of MnO 2 and Mn 2 Co 2 C were observed at 2θ = 35.1, 40.7, 58.8, 69.7 and 73.9°a nd at 2θ = 41.0, 47.8 and 70.2°, respectively. With the addition of Co(Ac) 2 , these reflections decreased in intensity and were not observed when the molar amount of Co(Ac) 2 was twice the amount of Mn(Ac) 2 in the synthesis.
All pyrolyzed samples have strong reflections originating from Co fcc at 2θ = 44.3, 51.6 and 75.9°. Furthermore, by adding Co(Ac) 2 to the synthesis, reflections originating from Co hcp were observed with the most intense reflection at 2θ = 47.5°. While the Mn 7 Co 10 sample only had a slight shoulder at the overlapping Mn 2 Co 2 C reflection at 2θ = 47.8°, lower molar Mn : Co ratios led to an increase in intensity. A clear reflection originating from Co 2 C at 2θ = 42.7°was only observed for Mn 10 Co 6 . After addition of Co(Ac) 2 , no clear reflections were observed, although it cannot be excluded that the Co fcc reflection at 2θ = 44.3°was overlapping with a Co 2 C reflection at 2θ = 42.7°. The observed reflections and assigned phases for the pyrolyzed catalysts are summarized in Table S2. Figure 2d shows the most intense reflections in more detail.
The phases observed in each pyrolyzed catalyst are shown in Table 1, in which the phases with the highest intensity are highlighted. The size of the crystallites of the different precursors in their dried and pyrolyzed states and of the spent catalysts were derived by using the Scherrer equation (Table 2). Table 1 shows that by addition of Co acetate, the resulting mixed PBAs had a smaller crystallite size in their dried state in the range from 18.0 to 24.0 nm. After pyrolysis, the Co crystallite sizes were found to be smaller, whereas the Co 2 C crystallite sizes observed after reaction were rather similar amounting to about 18 nm. To further investigate the texture of the mixed PBAs before and after pyrolysis, N 2 physisorption was performed.
The resulting isotherms shown in Figure S1 were analyzed using the BET and BJH methods. The obtained results are summarized in Table 3. Furthermore, Figure S2a shows the adsorption and desorption isotherms of the Mn 10 Co 6 and Mn 1 Co 11 precursors. While the Mn 10 Co 6 precursor led to a type I(a) adsorption isotherm, by adding Co(Ac) 2 the observed isotherms were no longer resembling type I(a), but rather type II. Furthermore, the hysteresis changed as well from type H4 to H3. Table 3 shows that the dried precursors were strongly influenced by the addition of Co(Ac) 2 during the synthesis. The Mn 7 Co 10 precursor had a decreased specific surface area (A s,BET ) in contrast to the Mn 10 Co 6 precursor, and the pore volume (V p ) as well as the pore radii distribution (r p,max ) were shifting to higher values due to the addition of Co(Ac) 2 . For lower Mn : Co ratios, the pore volume increased further, while the pore radii distribution shifted to slightly lower values.
The change of the molar Mn : Co ratio during synthesis led to no significant changes of the specific surface areas, with the exception of the Mn 1 Co 11 precursor, which has a slightly reduced surface area. After pyrolyzing the dried precursors for 3 h at 600°C, a significant decrease of the specific surface areas was observed. The Mn 10 Co 6 PBA had the lowest surface area after pyrolysis, but also the highest maximum of the pore radius distribution. These structural properties changed due to the addition of Co(Ac) 2 . The decrease of the surface area due to pyrolysis is found to be less strong with increasing Co amount, resulting in an average value of about 100 m 2 g À 1 . The  pyrolysis of the PBAs was further investigated using thermogravimetry coupled with mass spectrometry (TG-MS). Figure 3 shows the obtained TG and DTG curves for the Mn 1 Co 11 precursor revealing two significant steps. The first mass loss of about 15 % occurred at a temperature of 95°C, and an increase of the m/z ratios 17 and 18 was observed indicating the release of water. The vacancies in the PBA crystal structure are known to be occupied by water to a certain extent. [60] Furthermore, a second very distinct mass loss was observed starting at a temperature of 300°C, which has a prolonged tailing resulting in a mass loss of 31.7 % up to a temperature of 505°C. The DTG curve shows that the maximum of this mass loss is centered at around 375°C. The QMS data showed that m/z ratios of 13,14,15,26,27,28,30,44,52, and 53 increased in intensity. The intensities of the different observed mass fragments are shown in Figure S3, which were assigned to the formation of HCN, NO, CO x , CH x and (CN) 2 . Obviously, the linking CN groups decompose during this step, resulting in the formation of a carbon matrix. Furthermore, it can be assumed that the Mn and Co ions, which had been linked by the CN groups, are reduced and embedded in this growing carbon matrix. It is known in literature that potassium influences the thermal stability of PBAs, suggesting that the observed tailing of the decomposition to higher temperatures is the result of potassium being present in the vacancies.
In addition to these significant mass losses, an additional mass loss of~3.5 % was observed around a temperature of 555°C. The QMS data showed an increase of the mass fragments 28 and 14, but no increase of m/z = 27. Therefore, it can be assumed that this mass loss is not resulting from HCN formation, but from N 2 leaving the sample. Above the chosen pyrolysis temperature of 600°C no further mass loss took place.
Elemental analysis was also performed using the synthesized PBAs and the pyrolyzed precursors. The results are summarized in Table 4, which shows that the desired nominal Mn : Co ratios were successfully achieved. When lowering the molar Mn : Co ratio, the Mn content reached a value of 3.1 % for the pyrolyzed Mn 1 Co 11 sample. The amount of Mn was similar to the amount of K in this sample, which reached values between 2.4 and 3.1 wt% in all the pyrolyzed samples. The amount of K shows that the desired in situ deposition of K was also successful and should therefore lead to the desired promoting effect. Overall, the pyrolyzed samples mostly consisted of Co and C, while C was the element with the highest content in the dried precursors. After pyrolysis, all samples had a carbon content of 27.5-30.6 wt %, which was higher than in the dried state. Furthermore, it can be assumed that this carbon matrix is nitrogen-functionalized due to the observed nitrogen content in all samples. Table 4 also shows that by further increasing the molar amount of Co, the N content decreased. Furthermore, the carbon matrix of every investigated sample is also oxygen-functionalized as shown by the high residual content of O. Therefore, a unique O-and Nfunctionalized carbon matrix was obtained with embedded Kand Mn-promoted metallic Co nanoparticles with an average size of 18 nm.

Induction period
Due to the carefully chosen slow heating over 24 h TOS, the catalysts reached the desired reaction temperature without a negative influence of the high exothermicity of the previously mentioned reactions. The results after the induction period of 72 h TOS are shown in Figure 4. By increasing the molar amount of Co, CO conversion increased almost linearly up to 10.6 % for Mn 1 Co 11 . As the Co 17 catalyst demonstrates, completely removing Mn from the synthesis led to a strong decrease of CO conversion. The addition of Co(Ac) 2 first led to an increased selectivity of the secondary alcohols, but when exceeding a certain molar amount resulting in the Mn 3 Co 13 catalyst, their selectivity decreased again. This trend continued for the catalysts with an even higher molar amount of Co. In contrast, the selectivity of the primary alcohols decreased   Figure 4 shows that by increasing the molar amount of Co, the C 2 + HC selectivity behaved similarly to the selectivity of the secondary alcohols. At first, a strong increase of the selectivity was observed, which decreased after surpassing a certain molar amount of Co. Figure 4 shows that the CO 2 selectivity had no clear trend in contrast to the other investigated products. By synthesizing Cu-free catalysts, it was assumed that the influence of the WGSR producing CO 2 should decrease. However, it is known in literature that Co 2 C is an active phase for the WGSR, especially when it is promoted with K. [64] Therefore, it is reasonable to assume that K-promoted Co 2 C is the reason for this rather high CO 2 selectivity. Still, in addition to the continuous formation of Co 2 C (Equation 8) under HAS conditions, the metastable nature of Co 2 C must be considered. It is known that Co 2 C easily decomposes in the presence of H 2 O (Equation 9) resulting in Co and CO 2 . [65] 2 Co þ 2 CO ! Co 2 C þ CO 2 (8) The formed carbon deposited on the surface can be converted by H 2 O to syngas (Equation 10). In addition, solidstate reactions like the Boudouard equilibrium (Equation 6) and the steam gasification of carbon (Equation 10) must be considered. [11,17,66] C þ H 2 O ! CO þ H 2 (10) Due to pyrolysis, Mn and Co are homogenously distributed in the carbon matrix resulting in nanoparticles in close vicinity. This is beneficial for carbide formation resulting in a consumption of the carbon matrix, which is presumably replenished by the Boudouard equilibrium.
Furthermore, every catalyst exhibited a strong tendency to form hydrocarbons. While methane was the hydrocarbon with the highest selectivity reaching a value of up to 26 % for the Mn 3 Co 13 catalyst, the C 2 À C 8 hydrocarbon fractions reached a summed-up selectivity of 23 % for the same catalyst. The Co 17 catalyst exhibited the lowest C 2 À C 8 hydrocarbon selectivity of 21 %. Obviously, metallic Co 0 is finely embedded inside the functionalized carbon matrix resulting in a minimum of exposed active Co 0 sites for the FTS.
To gain deeper insight into the HAS mechanism over the PBA-based catalysts, the distribution of the alcohols and hydrocarbons shown in Figure 4 was further investigated. As Figure 5a and 5b show, the distribution of the formed alcohols was strongly influenced by the addition of Co(Ac) 2 . The Mn 10 Co 6 catalyst had a total alcohol selectivity of 16 %. MeOH, EtOH, 2-BuOH as well as 1-PeOH and 1-HexOH had a similar selectivity, while the other primary alcohols had a more than halved value. This distribution is rather interesting compared with other well-investigated catalysts, for which the alcohol distribution was described by the ASF distribution. [5] This was mainly attributed to the close vicinity of Co 0 and Co 2 C, resulting in CO insertion competing with the chain propagation on the surface of the catalyst according to the carbide mechanism. [4,5,16] Since these two phases are also present in the investigated catalysts, it must be assumed that an additional type of active site for HAS enabling alcohol carbonylation had been created, which leads to this different distribution.
This hypothesis was further investigated by analyzing the alcohol distribution of the other PBA-based catalysts. When adding Co(Ac) 2 during the synthesis, the selectivities of the longer-chain primary alcohols decreased, while the MeOH selectivity increased. Furthermore, the selectivities of the C 3 + secondary alcohols reached significantly higher values as well. The selectivities of the primary alcohols increased even further for higher molar amounts of Co, while the selectivities of the secondary alcohols decreased. Interestingly, the EtOH selectivity increased again when reaching a Mn : Co ratio of 3 : 13. By further decreasing the Mn : Co ratio, the longer-chain primary alcohols reached higher selectivites. Therefore, it can be assumed that the carbonylation of the primary alcohols occurs in addition resulting in longer-chain alcohols.
It is difficult to rationalize how the secondary alcohols are formed during HAS over the PBA-derived catalysts. One possible mechanism would be the aldol addition of ketones, which results in these secondary alcohols. This mechanism cannot be fully excluded, but by analyzing the collected liquid from the cooling trap in a GC-MS, no ketones were observed. The C 2 À C 5 aldehydes were observed instead as shown in Figure S4.
Another possible mechanism is olefin hydration, [67][68][69] which is well described in literature and has been investigated for the HAS especially over HAS systems supported on zeolites. [67,[69][70][71] Usually, an acidic catalyst is necessary for this reaction, but NH 3 temperature-programmed desorption experiments (not shown) demonstrated that the pyrolyzed catalysts had no acidic sites. However, it can be hypothesized that the acidic environment is resulting from the carbonylation of the primary alcohols yielding carboxylic acids as intermediates. In addition to acetic acid, which was calibrated prior to the experiments, two other peaks with a similar distinct shape were observed for the catalysts exceeding a molar ratio of Mn : Co of 3 : 13. The observed peaks are shown in Figure S5, providing further evidence that the carbonylation of alcohols is also taking place to a certain extent yielding acids. Adsorbed carboxylic acids may form the required acid sites on the surface of the catalyst during CO hydrogenation at 60 bar and 260°C, which are able to protonate olefins forming secondary carbenium ions followed by hydration to the secondary alcohol.
In addition to the presence of acetic acid, Figure S6 also shows that propionaldehyde was formed under HAS conditions. Due to the large area of this peak compared with the other investigated compounds, it seems highly likely that propionaldehyde is formed by the hydroformylation of ethene. Thus, the hydroformylation-hydrogenation tandem reaction resulting in primary alcohols may also influence the observed selectivities of the primary alcohols. It is worth noting that the hydroformylation of ethene to propionaldehyde followed by its hydrogenation to 1-propanol has also been observed in high-pressure pulse experiments for the 2CoCu catalyst when a threshold molar fraction of pulsed ethene of about 1 vol % was surpassed pointing to the presence of highly coordinatively unsaturated Co sites. [72] As shown in Figure 5c, methane was the hydrocarbon fraction with the highest selectivity for every investigated catalyst. While maintaining a rather constant selectivity for the catalysts derived with a higher Mn(Ac) 2 content, reaching an equal or higher amount of Co(Ac) 2 led to a strong increase from~17 % to 26.3 % for the Mn 3 Co 13 catalyst. A further increase of the molar amount of Co led to a lower selectivity around 21 %. The hydrocarbon fraction with the second highest selectivity is the C 3 fraction. By increasing the molar amount of Co, the resulting catalyst had a decreasing selectivity of the C 3 + HC fractions, but an increasing C 2 fraction. Furthermore, the chain growth probabilities of the alkanes, olefins and the total hydrocarbon fraction were compared with the primary alcohols. Figure S7 shows that alcohols and hydrocarbons have clearly different chain growth probabilities, indicating a complex reaction network favoring long-chain alcohols. Figure 5c suggests that the hydrocarbon fraction can still be described using the ASF distribution. Actually, this only holds for alkanes resulting in values for olefins and alkanes of 0.46 and 0.59, respectively. Olefins are further consumed by hydration and hydroformylation, which is not considered by the ASF distribution.
To further study the effect of the higher amount of Co in the catalyst, Figure 5d compares the olefinicity of the C 2 À C 6 fractions. By adding Co(Ac) 2 to the synthesis, the olefinicity of all fractions increased significantly. The strongest increase from 49.8 up to 70.4 % was observed for the C 2 fraction. For the higher Co content, the olefinicity of the hydrocarbon fractions decreased again. The only exception of this phenomenon is the C 3 fraction, which remains almost constant around a value of 72 % when Co(Ac) 2 is applied in the synthesis. Interestingly, while the C 2 olefinicity decreased at first more strongly than the olefinicity of the C 4 À C 6 fractions, removing Mn(Ac) 2 completely from the synthesis resulted in the reverse case. The olefinicity of the C 2 fraction reached a value of 68.0 %, while the C 4 À C 6 fractions reached an olefinicity similar to that of the Co 17 catalyst. This observation shows that the addition of Co(Ac) 2 to the synthesis was beneficial for the formation of olefines, but when reaching the Mn : Co ratio of 6 : 12, the hydrogenation of the olefins to the alkenes as consecutive reaction seems to be no longer retarded. This was observed especially for the longer-chain products.
The investigation of the induction period showed clearly that a carefully chosen molar amount of Mn introduced in the synthesis is important. The promoting effect of Mn is essential to achieve high conversion in HAS but exceeding this molar amount of Mn results in a decrease of CO conversion as well as a shift towards hydrocarbons. Nonetheless, completely removing Mn from the synthesis was proven to be detrimental, resulting in a CO conversion of only 5.2 % as well as a high CO 2 selectivity of 36 %.

Temperature variation
The catalysts derived from the Mn-Co-based PBAs were also investigated at 280°C to gain further insight into the interplay of the different reaction mechanisms. The normalized summarized selectivities are shown in Figure 6. The increase in temperature led to a higher CO conversion for every catalyst. The Mn 7 Co 10 and Mn 6 Co 12 catalysts reached values of 12-13 %, which corresponds to an almost doubled CO conversion compared with the values at 260°C. When further increasing the molar amount of Co, this increase became even stronger. The Mn 3 Co 13 catalyst reached a CO conversion of 30.2 %, while the Mn 2 Co 12 catalyst led to a value of 41.0 %. The Mn 2 Co 12 catalyst was investigated at 285°C and not at 280°C to avoid the formation of a hot spot due to the strong exothermicity of the involved reactions. The Mn 1 Co 11 catalyst reached a CO conversion of 24.9 % at 280°C. By comparing the selectivities determined at 280°C (Figure 6) with the previously discussed data obtained at 260°C (Figure 4) it becomes clear that the most significant change is the increasing CO 2 selectivity. Each catalyst had a selectivity of above 40 %, while the catalysts with a Mn : Co ratio of 3 : 13 or lower led to a higher CO 2 selectivity. As mentioned before, Co 2 C is a known catalyst for the WGSR especially when this phase is promoted by K. Obviously, the higher temperature increases the rates of the involved solidstate reactions (Equations 8 and 9) resulting in a higher rate of the WGSR.
The total selectivity of the primary and secondary alcohols ( Figure 6) follows similar trends as observed for the catalysts at 260°C. With increasing Co content, the selectivity of the secondary alcohols decreased, while the selectivity of the primary alcohols increased. Still, compared with the obtained data at 260°C, the total alcohol selectivity is lower at 280°C, which fits very well to the previously in-depth analyzed hydrotalcite-based 2CoCu-based catalyst. [4,5] To gain further insight into the product distribution, Figure 7 shows the selectivities as well as the olefinicities of the different catalysts at 280°C.
MeOH was the alcohol with the highest selectivity for the Mn 7 Co 10 and Mn 6 Co 12 catalysts. The overall distribution of the alcohols was similar to the one shown in Figure 5a. When the Co content increased further, the distribution of the alcohols at 280°C differed strongly from the previously observed selectivities at 260°C. MeOH was no longer the alcohol with the highest selectivity and even decreased in selectivity. Instead, the EtOH selectivity reached significantly higher values. This trend continues for an even higher Co content in the catalyst. Similarly, the 1-PrOH and 1-BuOH selectivities increased as well for the Mn 3 Co 13 and Mn 2 Co 12 catalysts, while the Mn 1 Co 11 catalyst achieved even a slightly higher 1-PeOH and 1-HexOH selectivity. The strong increase of the EtOH selectivity followed by a similar behavior of the higher primary alcohols indicates that the rate of carbonylation of the primary alcohols is increasing at higher temperature and with a higher Co content in the catalyst, which was already discussed when investigating the catalysts at 260°C. It has to be pointed out that the catalyzed carbonylation of MeOH is usually performed at lower temperatures both in homogenous and heterogenous processes. [48,[73][74][75] Furthermore, the carbonylation of MeOH is a process which is usually performed in the presence of methyl iodide, [48] which is not required for the PBA-based catalysts applied at higher temperatures. By comparing the selectivities of the secondary alcohols at 280°C (Figure 7b) and 260°C (Figure 5b), it becomes obvious that the higher temperature resulted in a decreasing selectivity. It is known that olefin hydration, which is leading to secondary alcohols, can be performed at temperatures between 250-270°C. [67,70] By taking into consideration the previously observed increase of the selectivities of the primary long-chain alcohols, it can be assumed that both reactions are competing. Due to the higher rate of hydrogenation of the acids formed via the carbonylation reaction pathway, the amount of carboxylic acids is decreasing, which should lead to a lower acidity of the catalysts at higher temperatures. Therefore, by combining the insight from Figure 7a and 7b, it can be assumed that carbonylation is the dominant reaction pathway. The distribution of the primary alcohols shows clearly that the C 2 + alcohols selectivity increased, while the MeOH selectivity decreased. Especially the change of the EtOH selectivity must be pointed out here: while MeOH was the alcohol with the highest selectivity at 260°C, this was no longer the case at 280°C and a lower molar Mn : Co ratio. The favored carbonylation of primary alcohols also led the previously discussed lowered selectivity of secondary alcohols. Figure 7c shows that the selectivities of the hydrocarbon fractions changed slightly due to the increase to 280°C. The most significant changes were observed in the C 1 and C 3 fraction. While the methane selectivity increased for the Mn 7 Co 10 and Mn 6 Co 12 catalysts by about 2 %, the CH 4 selectivity of the catalysts with a lower Mn : Co ratio decreased, as clearly seen for the Mn 3 Co 13 catalyst, which reached a selectivity of 26.4 % at 260°C. By increasing the temperature to 280°C, the methane selectivity decreased to 20.6 %. A similar value was reached for the Mn 1 Co 11 catalyst. In contrast to the methane selectivity, the C 3 + HC selectivity decreased at 280°C. This was not only observed for the Mn 7 Co 10 and Mn 6 Co 12 catalysts, but also for the catalysts a lower Mn : Co ratio. This is rather surprising when comparing the PBA-based catalysts to previously investigated Co-based systems. For the 2CoCu catalyst, an increase of the selectivities of not only methane, but also of the higher hydrocarbon fractions was observed. [5] Still, a similarity between the hydrotalcite-based 2CoCu and the PBA-based catalysts can be found by analyzing the olefinicity of the different hydrocarbon fractions. Figure 7d shows that the olefinicity of the hydrocarbon fractions are all higher at 280°C, apart from the C 2 hydrocarbon fraction. Similar to the 2CoCu catalyst, the higher temperature led to a lower olefinicity of the C 2 hydrocarbon fraction. [5] Still, the selectivity of the C 2 hydrocarbon fraction did not deviate for the different PBA-based catalysts, which means that the rate of ethylene hydrogenation was heavily dependent on the molar amount of Co. While a high Co amount led to a faster hydrogenation to ethane, increasing the molar amount of Mn stabilized the olefinicity of the C 2 fraction.
Based on this gained insight into this new class of catalysts for the HAS, this study proposes a reaction network with an interplay of the different reaction mechanism for the HAS consisting of the carbonylation of primary alcohols, the reductive hydroformylation of olefines, and the olefin hydration reaction pathway in addition to the well investigated interplay of MeOH and FT synthesis. Figure 8 visualizes this interplay of different reaction pathways. Interestingly, a solid cobalt FTS catalyst with high selectivity to higher olefins has been recently integrated in tandem with molecular cobalt reductive hydroformylation catalysts to achieve a slurry-phase direct conversion of syngas to higher alcohols at 200°C and 120 bar. [76] Very high selectivities to C 2 + alcohols were achieved in this three-phase system at high CO conversion, similar to the PBA-derived catalysts, which combine FT synthesis of olefins with subsequent reductive hydroformylation in a single solid phase.

Characterization after reaction
After subjecting the catalysts derived from the Mn-Co-based PBAs to the temperature variation, the spent samples were collected and investigated again by XRD. The diffractograms are shown in Figure 9. The observed phases for each investigated catalyst after HAS are summarized in Table 5, and  the phases with the highest reflections intensities are highlighted. Similar to Figure 2, the diffractograms show clearly that the variation of the Mn : Co ratio led to significant structural changes under HAS conditions. The Mn 7 Co 10 and Mn 6 Co 12 catalysts after reaction had very distinct reflections at 2θ = 24.4, 31.5, 38.1, 51.9 and 60.5°originating from MnCO 3 . Furthermore, previously observed reflections originating from Mn 2 Co 2 C, Co fcc and Co hcp are visible in the diffractograms, and clearly detectable Co 2 C reflections are also present at 2θ = 37.2, 42.7, 45.9, 56.8 and 71.5°.
Similar to the pyrolyzed samples, a decreasing Mn : Co ratio lowered the intensity of the reflections originating from the Mn-based phases. While the diffractogram of the Mn 3 Co 13 catalyst showed significant MnCO 3 reflections, this was not the case for the samples with a higher Co content. The formation of this carbonate phases seems to be closely linked to the presence of MnO 2 . By comparing the diffractograms before and after the HAS, it becomes obvious that the MnO 2 reflections are absent after performing HAS. Therefore, it can be assumed that the formation of MnCO 3 may be similar to carbonate formation on Pd/Al 2 O 3 [77] because of the observed consumption of MnO 2 (Equation 11).
In addition to this transformation, the structural transformation of Co 0 to Co 2 C was also confirmed by the XRD analysis of the spent samples. Obviously, the embedding of the Co nanoparticles in the carbon matrix led to a successful carbidization. In contrast to literature, we were able to successfully transform metallic Co into Co 2 C in the absence of Mn in the catalyst as shown by the diffractogram of the spent Co 17 catalyst in Figure S8. So far, this transformation has been linked to the presence of MnO x phases, and their absence resulted in inhibited carbidization. [31,[78][79][80][81] In addition to the observed Co 2 C reflections, reflections originating from the Co hcp phase were observed in the catalyst without Mn. Thus, the highly functionalized carbon matrix favors carbidization, and the presence of Mn promotes this transformation in the PBA-based catalysts even further. The Mn 1 Co 11 catalyst was also analyzed by EDX mapping shown in Figure S9 before and after the HAS to gain deeper insight into the structural changes. As Figure S9a shows, Co nanoparticles were formed during pyrolysis, which were homogeneously dispersed in the carbon matrix. In contrast, Mn was only located at the Co nanoparticles. After performing HAS, the distribution of the Co nanoparticles changed significantly. As shown in Figure S9b, the Co nanoparticles agglomerated while still being embedded in the carbon matrix.
Furthermore, TEM images of the pyrolyzed and spent Mn 1 Co 11 catalyst were obtained (Figure S10) to investigate the particle size distribution ( Figure S11). The pyrolyzed sample had a maximum of 12-13 nm in good agreement with the XRD-derived Co fcc crystallite size of 13 nm (Table 2), with most Co particles having a size between 10 and 16 nm. After reaction, the mean Co 2 C particle size was found to be higher with most particles being in the range from 14 nm to 20 nm again in good agreement with 18 nm derived by applying the Scherrer equation.
To gain further insight into the transformations occurring during the synthesis and under reaction conditions, XPS measurements were performed subsequent to transfer in air. Figure 10 shows the obtained spectra in the Co 2p and C 1s regions, while the Mn 2p, O 1s, N 1s and K 1s regions are shown in Figure S12. Calibration of the binding energy scale was accomplished using the adventitious carbon signal at 284.5 eV. [82] The dried precursors revealed the presence of a very distinct peak at 779.7 eV in the Co 2p region reflecting the Co II and Co III oxidation states. [83] The presence of CoO can be excluded due to the missing satellites, but not that of Co(OH) 2 . [84,85] It is reasonable to assume that the surface of the precursors suffered from hydrolysis due to contact with air to some extent. After pyrolysis, the Co 2p signal indicates the presence of metallic Co 0 at 778.2 eV and of CoO at 779.7 eV, [84,85] which is clearly identified by the satellites at higher binding energies. Metallic Co was also confirmed by the XRD results shown in Figure 2. After the induction period and temperature variation, the spent catalyst was investigated again by XPS. The corresponding Co 2p spectra revealed a significantly decreased intensity of the signals.
The maximum of the Co 2p 3/2 peak at approximately 780 eV is difficult to detect. An even worse signal-to-noise ratio was observed in the Mn 2p spectra of the spent sample (Figure S12a), where no peak at all was found after HAS. Thus, it must be assumed that the low intensity of the Co 2p and Mn 2p XP spectra of the spent sample is due to the formation of a thick carbon layer covering the surface of the catalyst.
The obtained C1 s spectra (Figure 10 b) show a very intense peak at 284.5 eV in every catalyst state, which is the typical binding energy of adventitious carbon. [82] In the dried and pyrolyzed states, a shoulder was observed at higher binding energies, which results from carbon-oxygen species. This multitude of different species demonstrates that the carbon matrix of the catalyst is significantly oxygen-functionalized. The C 1s spectrum of the spent sample after HAS only contains the intense peak at 284.5 eV indicating pronounced coking. In addition to the recorded spectra, the relative surface compositions of the investigated components are summarized in Table 6. In combination with the ICP-MS analysis shown in Table 4, it becomes clear that the outer surface of the pyrolyzed catalysts mostly consists of the oxygen-and nitrogen-functionalized carbon matrix with low amounts of Co, Mn and K, whereas essentially no information about the state of the catalyst surface after reaction is obtained due to heavy coking.
Finally, the question has to be addressed which active sites catalyze the observed alcohol carbonylation and reductive olefin hydroformylation. Due to the formation of a strongly functionalized carbon matrix during the PBA pyrolysis, it is highly likely the Co particle size is bimodal. In addition to the 18 nm Co 2 C nanoparticles detected by XRD and confirmed by TEM, which catalyze the carbide mechanism and reductive hydroformylation at the Co 2 C/Co 0 interface, we assume that also Co species are present, which are efficiently anchored by the oxygen-and nitrogen-functionalized carbon matrix and have similar catalytic alcohol carbonylation properties as the Co carbonyl hydride complexes applied in homogeneous catalysis. This hypothesis is supported by a recent study, where an unmodified Co carbonyl catalyst was applied in hydroformylation. [86] The authors conducted kinetic and infrared spectroscopy measurements with the conclusion that this unmodified Co catalyst was transformed into the desired Co carbonyl hydride species at 140°C and 30 bar. Furthermore, in contrast to so far known literature, this complex was also stable under these low syngas pressures. [86] Therefore, the presence of additional active sites resembling Co carbonyl hydride complexes with respect to their catalytic properties at 260°C and 60 bar seems rather likely. Further structural studies are needed to identify these additional active sites in the highly Co-loaded carbon matrix in the presence of the 18 nm Co 2 C nanoparticles. In addition, kinetic steady-state co-feeding and transient pulse experiments are in progress to further unravel the complex reaction network catalyzed by K-and Mnpromoted Co species embedded in the functionalized carbon matrix. Recently, chemical transient kinetics (CTK) was applied as a powerful tool to gain deeper insight into the chainlengthening mechanism over Co-Cu-and Co-Mn-based catalysts. The authors concluded that the formation of formate-/ carboxylate-derived intermediates is the key step. [87] Whether this pathway applies for the PBA-based catalysts as well is the focus of ongoing kinetic studies.
While the presented PBA-derived catalysts still suffer from a moderate oxygenate selectivity, a high CO conversion of up to 10.6 % at a GHSV of 12.000 cm 3 g À 1 h À 1 was achieved using the Mn 1 Co 11 catalyst. [4,5] A wide variety of mechanisms was shown to be involved in the oxygenate formation, and in comparison to other Co-Mn-based systems, an interplay between the formation of primary and secondary alcohols was found. Furthermore, the transformation of Co 0 to Co 2 C was observed even without the presence of Mn. Ongoing studies with more efficiently promoted PBA-derived catalysts include catalytic testing at lower temperatures, which are more favorable for the stability of the in situ formed Co 2 C and a higher oxygenate selectivity. [88] Conclusions Catalysts derived from PBA-based precursors by pyrolysis were found to have a high potential for the synthesis of higher alcohols. It was shown that Mn-Co-based catalysts promoted with K can be obtained by a microemulsion synthesis followed by pyrolysis at 600°C, which decomposed the Mn x Co 3À x ½Co CN ð Þ 6 � 2 �nH 2 O precursors resulting in metallic 13-28 nm Mn-promoted Co nanoparticles embedded in an oxygen-and nitrogen-functionalized carbon matrix. The structure of the catalyst precursor changed significantly under HAS Figure 10. Recorded XP spectra in the a) Co 2p and b) C 1s region for the Mn 1 Co 11 catalyst in its dried and pyrolyzed states and after reaction subsequent to transfer in air. reaction conditions at 260°C, 60 bar, a H 2 /CO ratio of 1 and a gas hourly space velocity of 12000 cm 3 g cat À 1 h À 1 resulting in the formation of 18 nm Co 2 C nanoparticles. The catalysts were not only active in the carbide-based reaction pathway to higher alcohols and olefins, but also in the carbonylation of primary alcohols, reductive hydroformylation and olefin hydration. Accordingly, acetic acid and propionaldehyde were detected as intermediates. These reaction pathways resulted in a significant CO conversion of 10.6 % and a broad product distribution of primary and secondary alcohols yielding a total alcohol selectivity of 19 % for the Mn 1 Co 11 catalyst. Comparing the chain growth probabilities and the distribution of the primary alcohols and hydrocarbons showed that the values differ strongly, further demonstrating the presence of other reaction mechanisms. Furthermore, a CO 2 selectivity between 27-36 % was observed at 260°C after 72 TOS, although a Cufree catalyst was investigated, which agrees with results in literature showing that Co 2 C promoted with K is an active catalyst for the WGSR.
The catalysts were also tested at 280°C. Decreasing the amount of Mn to 3.1 wt % in the Mn 1 Co 11 catalyst resulted in the carbonylation of primary alcohols to be the favored reaction pathway. Accordingly, the EtOH selectivity increased significantly at 280°C, while the MeOH selectivity decreased, which was also observed for the long-chain alcohols. Due to the resulting higher consumption of the formed carboxylic acids, the number of generated acidic sites is assumed to decrease resulting in the lower selectivity of secondary alcohols.
EDX mapping of the Mn 1 Co 11 catalyst after catalytic testing at 280°C showed that Mn is mostly present at the boundary between the Co 2 C NPs and the carbon matrix favoring carbide formation. The carbide mechanism and reductive hydroformylation are assumed to occur at the Co 2 C/Co 0 interface. Additional Co species seem to be present, which are efficiently anchored by the oxygen-and nitrogen-functionalized carbon matrix and have similar catalytic alcohol carbonylation properties as the Co carbonyl hydride complexes applied in homogeneous catalysis.

Experimental Section
The applied PBA-based catalysts were synthesized using the microemulsion method with specific Mn : Co ratios aiming at Mn x Co 3À x Co CN ð Þ 6 ½ � 2 � nH 2 O precursors. The microemulsion was obtained at room temperature with a EtOH/ H 2 O ratio of 1. The metal salts Mn(Ac) 2 * 4 H 2 O (Aldrich Chemistry, 99 %) and Co(Ac) 2 * 4 H 2 O (Sigma-Aldrich, 98 %) with the desired molar ratio were dissolved in 150 mL EtOH and 50 mL demineralized water together with an excess of 11.12 g polyvinylpyrrolidone (Sigma-Aldrich, PVP-K30). K 3 ½Co CN ð Þ 6 � �nH 2 O (Acros Organics, 95 %), which was dissolved in 100 mL demineralized water, was added dropwise under rigorous stirring. Afterwards, the suspension was further aged for 24 h followed by centrifugation for 10 min at 11000 rpm. After freeze-drying the precursor for 24 h at À 50°C, the dried precursor was pyrolyzed at 600°C in a constant flow of N 2 (100 cm 3 min À 1 ) using a heating rate of 2 K min À 1 for 3 h.
Kinetic investigations were conducted in a flow setup equipped with six gas lines. All instruments were calibrated to normal conditions (25°C, 1013 kPa). The high-pressure reaction unit of the setup contained a 1 = 4 " fixed-bed reactor equipped with an axial thermocouple in the middle of the catalyst bed. The reproducible placement of the catalyst bed in the isothermal zone was ensured by insertion of a quartz stick to retain the same position. The reactor tube with an inner diameter of 4.5 mm was made of stainless steel and additionally coated with a thin film of SilcoNert2000 ® to ensure inert conditions. The reactor was heated by a single-zone furnace. 100 mg of a sieve fraction of 250-355 μm of the pyrolyzed catalyst were filled into the reactor. After the pressure had been ramped up to 60 bar with 1 bar min À 1 , the desired flow of 20 cm 3 min À 1 with the syngas atmosphere of H : CO = 1 was established leading to a gas hourly space velocity of 12000 cm 3 g cat À 1 h À 1 . After at least 5 h, the catalyst was heated with 5 K min À 1 to 160°C and slowly with 0.06 K min À 1 to 260°C. Each catalyst was kept at 260°C for 72 h before starting a temperature variation, during which the temperature was increased from 260°C to 280°C in steps of 5°C.
Data evaluation was performed by using 2-D online GC. The customized GC application was optimized for the separation of alcohols and hydrocarbons as well as the separation of paraffins and olefins within the hydrocarbon distribution leading to further insight into the HAS. Volume contraction during HAS, which results in changed molar fractions of products in the effluent gas stream, was compensated by an inert standard. To calculate exact mass balances, information regarding the products of the reaction is necessary. The C-balance (Equation 12) is defined as C À balance ¼ _ n C; out _ n C; in ¼ P i y i ð ðCÞ i;out z C;i Þ y CO; in (12) whereas the factor z i is the number of atoms for the component i. Furthermore, _ n i; out (Equation 13) can be defined by using the molar volume V m , the volumetric flow _ V and the ideal gas constant R as The degree of conversion X CO (Equation 14) of CO was derived using X CO ¼ _ n CO; in À _ n CO; out _ n CO; in (14) Additionally, the selectivity S i (Equation 15) of product i was obtained relative to the total amount of products formed.
Catalyst mass-based reaction rates (Equation 16) were calculated, where the mass of the catalyst m cat was used as reference. To obtain this rate, the conversion X CO and the molar fraction y CO;in were taken into consideration together with the total molar flow entering the reactor P _ n i; in .
r CO ¼ X CO y CO;in P _ n i; in m cat ¼ _ n CO; consumed m cat (16) X-ray diffraction patterns (XRD) were recorded using a Bruker D8 Discover with a theta-theta geometry equipped with Cu K α radiation (λ = 0.15406 nm, 40 kV, 40 mA) and a Lynxeye XEÀ T detector, where fluorescence was automatically suppressed. The patterns were recorded in a 2θ range from 10 to 80°with a step size of 0.02°and a time of 2 s per step. Measurements were conducted at room temperature at atmospheric pressure. Additionally, an airscatter aperture above the sample and a motorized aperture were applied in front of the tube in the primary beam path. Evaluation of the recorded diffraction patterns was conducted using the Diffrac.EVA Software equipped with access to the Powder Diffraction File 2 (PDF-2) database provided by the International Centre for Diffraction Data (ICDD). The Scherrer equation was applied to derive crystallite sizes using the 2θ = 24.6°reflection of dried PBA, the 2θ = 75.9°reflection of Co fcc after pyrolysis and the 2θ = 56.7°reflection of Co 2 C after reaction. N 2 physisorption measurements were performed with sieve fraction of the precursors. Prior to the measurement, the samples were pretreated at 200°C for 2 h and analyzed afterwards at a constant temperature of 77 K using the Belsorp mini from BEL Japan. A Cahn TG 2131 thermobalance coupled with an Omnistar MS (Balzers Instruments) was used for thermal analysis. The investigated sample was placed in a continuous flow of He with a rate of 100 NmL min À 1 and heated to 800°C with a heating rate of 2 K min À 1 . Elemental analysis was performed at MikroLab Kolbe in Oberhausen, which is associated to the Fraunhofer institute UMSICHT.
Transmission electron microscopy (TEM) and energy dispersive Xray spectroscopy (EDX) images of the pyrolyzed and spent sample were obtained using a C S -corrected JEOL JEM-2200FS microscope operated at 200 kV accelerating voltage. X-ray photoelectron spectroscopy (XPS) measurements were conducted subsequent to transfer in air using a UHV setup with Al Kα radiation (1486.3 eV, 14.5 kV, 45 mA), a monochromator in the primary beam path, and a Gammadata Scienta SES 2002 analyzer for which the pass energy was set to 200 eV. The base pressure in the measurement chamber was 6 * 10 À 10 mbar. The obtained spectra were analyzed using the CasaXPS software with Shirley-type background subtraction and symmetric and asymmetric Gaussian-Lorentzian line shapes.