Oxidation of Benzylic Alcohols and Lignin Model Compounds with Layered Double Hydroxide Catalysts

Alcohol oxidation to carbonyl compounds is one of the most commonly used reactions in synthetic chemistry. Herein, we report the use of base metal layered double hydroxide (LDH) catalysts for the oxidation of benzylic alcohols in polar solvents. These catalysts are ideal reagents for alcohol oxidations due to their ease of synthesis, tunability, and ease of separation from the reaction medium. LDHs synthesized in this study were fully characterized by means of X-ray diffraction, NH3-temperature programmed desorption (TPD), pulsed CO2 chemisorption, N2 physisorption, electron microscopy, and elemental analysis. LDHs were found to effectively oxidize benzylic alcohols to their corresponding carbonyl compounds in diphenyl ether, using O2 as the terminal oxidant. LDH catalysts were also applied to the oxidation of lignin β-O-4 model compounds. Typically, for all catalysts, only trace amounts of the ketone formed from benzylic alcohol oxidation were observed, the main products comprising benzoic acids and phenols arising from β-aryl ether cleavage. This observation is consistent with the higher reactivity of the ketones, resulting from weakening of the Cβ–O4 bond that was shown to be aerobically cleaved at 180 ◦C in the absence of a catalyst.

Our interest in these catalysts was piqued by a report by Choudary et al. [21], who found that Ni-Al-LDH catalysts were effective for alcohol oxidations. This was rather unexpected since Ni is rarely reported as an oxidation catalyst. Further exploration in this field showed that there are a variety The ability to function in polar solvents is particularly useful for the application of LDHs to the oxidation and depolymerization of lignin. Other researchers have applied LDHs to the oxidative depolymerization of lignin model compounds with a limited degree of success; however, the full potential of LDHs for this application has yet to be explored. Indeed, Corma and Bolm used a vanadate-containing Cu-hydrotalcite (Cu-HT) for the oxidation of lignin and lignin models [28]. While the catalyst was highly active for the oxidation of the lignin β-O-4 model, the active species was found to be homogenous in nature. Additionally, Baguc et al. studied the use of Ru/HT (i.e., Ru supported on HT) for the oxidation of the lignin model monomer veratryl alcohol with O 2 , finding the catalyst to yield the corresponding aldehyde in near quantitative yield [29]. Beckham et al. used Ni/HT for the depolymerization of a lignin model β-O-4 compound under thermolytic conditions, resulting in complete conversion in 1 h at 270 • C [30]. Against this background, we set out to develop a catalyst system that not only functioned well in polar solvents, but also had the potential to act as a lignin depolymerization catalyst. To this end, the synthesis, characterization, and evaluation of a variety of LDH catalysts for the oxidation of benzylic alcohols, as well as their application to models of the bio-polymer lignin, was studied.

Catalyst Characterization
Generally, LDHs are expected to form under basic conditions as long as the metal cations (M 2+ and M 3+ ) have an ionic radius similar to Mg 2+ and the trivalent metal ratio (χ) is between 0.2 and 0.4 as noted above [26]. In this study all catalysts were synthesized with the theoretical χ value within the aforementioned limits. The metal ratios of prepared catalysts and their physical properties are summarized in Table 1. Elemental analysis revealed that the catalysts returned a similar molar metal ratio to that of the solutions used for LDH synthesis. Scanning electron microscopy (SEM) images of Ni-Al-LDH-1 and Ni-Cr-LDH are shown in Figures S2 and S3, and illustrate typical LDH platelet morphology. FT-IR analysis ( Figure S4) revealed that all catalysts displayed bending and stretching bands characteristic of LDH structures. Specifically, bands corresponding to the bending mode of interlayer water and asymmetric carbonate stretching were observed at ca. 1634 cm −1 and ca. 1346 cm −1 , respectively. Cavani et al. [27] noted that in order to incorporate copper (II) into LDH structures, it must be present with another bivalent metal in a Cu 2+ /M 2+ ratio of less than or equal to 1. This empirical rule is attributed to the tendency of Cu 2+ compounds to undergo Jahn-Teller distortions causing elongation of the octahedral coordination sphere. Therefore, Cu 2+ ions must be accompanied by another M 2+ metal such that the coordination sphere is undistorted in the brucite-like LDH structure. Whereas the Cu 2+ /M 2+ rule was followed for the synthesis of Cu-Ni-Cr-LDH, the rule was disregarded in the synthesis of Cu-Cr-LDH. Unsurprisingly, in the Cu-Cr-LDH, a minor amount of a crystalline phase, which was identified as the mineral malachite, was apparent in addition to the LDH phase ( Figure 1). ratio to that of the solutions used for LDH synthesis. Scanning electron microscopy (SEM) images of Ni-Al-LDH-1 and Ni-Cr-LDH are shown in Figures S2 and S3, and illustrate typical LDH platelet morphology. FT-IR analysis ( Figure S4) revealed that all catalysts displayed bending and stretching bands characteristic of LDH structures. Specifically, bands corresponding to the bending mode of interlayer water and asymmetric carbonate stretching were observed at ca. 1634 cm −1 and ca. 1346 cm −1 , respectively. Cavani et al. [27] noted that in order to incorporate copper (II) into LDH structures, it must be present with another bivalent metal in a Cu 2+ /M 2+ ratio of less than or equal to 1. This empirical rule is attributed to the tendency of Cu 2+ compounds to undergo Jahn-Teller distortions causing elongation of the octahedral coordination sphere. Therefore, Cu 2+ ions must be accompanied by another M 2+ metal such that the coordination sphere is undistorted in the brucite-like LDH structure. Whereas the Cu 2+ /M 2+ rule was followed for the synthesis of Cu-Ni-Cr-LDH, the rule was disregarded in the synthesis of Cu-Cr-LDH. Unsurprisingly, in the Cu-Cr-LDH, a minor amount of a crystalline phase, which was identified as the mineral malachite, was apparent in addition to the LDH phase ( Figure 1). In addition, acidity and basicity measurements were conducted on all catalysts (Table 1 and Figures S5-S9). Notably, the Ni-Al-LDHs had the highest number of base sites on a weight basis, adsorbing 157.4 μmol CO2•g −1 and 79.2 μmol CO2•g −1 for Ni-Al-LDH-2 and Ni-Al-LDH-1, respectively. According to NH3-TPD experiments, Ni-Cr-LDH possessed the highest number of acid sites of the samples analyzed. While Mg-Al-LDH-1 had relatively few acid sites, it contained a higher relative proportion of strong acid sites (NH3 desorbed >450 °C, Figure S7). While both Ni-Al-LDH-1 and Ni-Al-LDH-2 had a similar number of total acid sites, Ni-Al-LDH-2 had a higher proportion of medium In addition, acidity and basicity measurements were conducted on all catalysts (Table 1 and Figures S5-S9). Notably, the Ni-Al-LDHs had the highest number of base sites on a weight basis, adsorbing 157.4 µmol CO 2 ·g −1 and 79.2 µmol CO 2 ·g −1 for Ni-Al-LDH-2 and Ni-Al-LDH-1, respectively. According to NH 3 -TPD experiments, Ni-Cr-LDH possessed the highest number of acid sites of the samples analyzed. While Mg-Al-LDH-1 had relatively few acid sites, it contained a higher relative proportion of strong acid sites (NH 3 desorbed >450 • C, Figure S7). While both Ni-Al-LDH-1 and Ni-Al-LDH-2 had a similar number of total acid sites, Ni-Al-LDH-2 had a higher proportion of medium  Figures S5 and S6). Indeed, strong and medium acid sites were virtually absent in Ni-Al-LDH-1.

Solvent Screening
Inspired by the work of Choudary et al. [21], we endeavored to find a solvent system that would be suitable for the oxidation of lignin model compounds. Given the polar nature of alcohols, including the lignin macromolecule, it is crucial to find a polar solvent that is effective for alcohol oxidations using LDH catalysts. Therefore, utilizing Ni-Al-LDH-1, which has a similar composition to the Ni-Al-LDH used by Choudary et al. [21], a series of solvents varying in polarity were screened for the oxidation of 1-phenyl ethanol, 1. As shown in Table 2, only limited conversions of 1 were obtained in most polar solvents ( Table 2, entries 4-9). Acetophenone, 1a, was obtained in near quantitative yield in toluene ( Table 2, entry 3), while reaction in 1,4-dioxane (Table 2, entry 7) yielded only 8% of 1a. In an attempt to amalgamate the polar ether properties of 1,4-dioxane with the electron-rich, aromatic character of toluene, phenyl ether ( Table 2, entry 1) was trialed, resulting in 49% conversion of 1 to 1a. Having identified a suitable solvent for benzylic alcohol oxidation, LDH catalysts containing metals that are traditionally used in oxidation chemistry (copper and chromium) were synthesized (Section 3.1) and screened for activity in the oxidation of 1. As shown in Table 3, Ni-Cr-LDH (entry 6), Ni-Al-LDH-1 (thermally pretreated at 175 • C for 3 h, entry 4), and Ni-Al-LDH-2 (entry 5) gave high yields of 1a (>90%). Exploratory experiments (not shown) with Ni-Al-LDH-1 showed that light thermal pretreatment (175 • C for 3 h) appears to be optimal for catalytic performance. This is presumably since the temperature is high enough to remove H 2 O and/or CO 2 coordinated to otherwise coordinatively unsaturated Ni 2+ species that are proposed active sites (see Section 2.2.7 below). Unfortunately, thermal pretreatment with catalysts other than Ni-Al-LDH-1 did not follow the same trend. Notably, Mg-Al-LDH-1 and Mg-Al-LDH-2 showed relatively little conversion of 1, indicating that activity is not the result of catalyst basicity. Likewise, conversion of model compound 1 did not trend with acidity. Indeed, Ni-Cr-LDH, which had the highest number of acid sites, did not show the highest conversion of 1 to 1a. Moreover, while Cu-Cr-LDH had neither the most acid or base sites, it demonstrated the highest conversion of 1 to 1a, suggesting a reaction mechanism not related simply to the acidity or basicity of the catalyst.

Catalyst Loading Study
In order to elucidate the optimal amount of catalyst needed, a catalyst loading study was performed using Ni-Cr-LDH. The amount of catalyst was incrementally increased while keeping the amount of starting material constant at 2 mmol. As can be seen from Table 4, using 0.5 g of Ni-Cr-LDH for every 2 mmol of starting material proved to be optimal (entry 4). While this is a large amount of catalyst, it is not uncommon in the literature [25,29]. The need for a large amount of catalyst relative to starting material suggests that the active site corresponds to defect sites that are present in low concentration on the catalyst surface.

Leaching Study
In order to determine whether conversion was the result of leached metal in the solution, a hot filtration experiment was performed in which the Ni-Cr-LDH catalyst was hot filtered from the reaction mixture after 1 h. A sample was taken, after which the filtrate was transferred to a fresh flask and allowed to react for an additional 23 h at 150 °C. Analysis of the reaction mixture indicated a 39% conversion at 1 h with no additional conversion post-filtration, suggesting that catalysis occurred on the LDH surface and not via free metal species in solution. Elemental analysis of the reaction mixture at 24 h post-filtration did not reveal significant amounts of metal leached into solution (<1 ppm Cr and 3 ppm Ni).

Catalyst Loading Study
In order to elucidate the optimal amount of catalyst needed, a catalyst loading study was performed using Ni-Cr-LDH. The amount of catalyst was incrementally increased while keeping the amount of starting material constant at 2 mmol. As can be seen from Table 4, using 0.5 g of Ni-Cr-LDH for every 2 mmol of starting material proved to be optimal (entry 4). While this is a large amount of catalyst, it is not uncommon in the literature [25,29]. The need for a large amount of catalyst relative to starting material suggests that the active site corresponds to defect sites that are present in low concentration on the catalyst surface.

Catalyst Loading Study
In order to elucidate the optimal amount of catalyst needed, a catalyst loading study was performed using Ni-Cr-LDH. The amount of catalyst was incrementally increased while keeping the amount of starting material constant at 2 mmol. As can be seen from Table 4, using 0.5 g of Ni-Cr-LDH for every 2 mmol of starting material proved to be optimal (entry 4). While this is a large amount of catalyst, it is not uncommon in the literature [25,29]. The need for a large amount of catalyst relative to starting material suggests that the active site corresponds to defect sites that are present in low concentration on the catalyst surface.

Leaching Study
In order to determine whether conversion was the result of leached metal in the solution, a hot filtration experiment was performed in which the Ni-Cr-LDH catalyst was hot filtered from the reaction mixture after 1 h. A sample was taken, after which the filtrate was transferred to a fresh flask and allowed to react for an additional 23 h at 150 °C. Analysis of the reaction mixture indicated a 39% conversion at 1 h with no additional conversion post-filtration, suggesting that catalysis occurred on the LDH surface and not via free metal species in solution. Elemental analysis of the reaction mixture at 24 h post-filtration did not reveal significant amounts of metal leached into solution (<1 ppm Cr and 3 ppm Ni).

Leaching Study
In order to determine whether conversion was the result of leached metal in the solution, a hot filtration experiment was performed in which the Ni-Cr-LDH catalyst was hot filtered from the reaction mixture after 1 h. A sample was taken, after which the filtrate was transferred to a fresh flask and allowed to react for an additional 23 h at 150 • C. Analysis of the reaction mixture indicated a 39% conversion at 1 h with no additional conversion post-filtration, suggesting that catalysis occurred on the LDH surface and not via free metal species in solution. Elemental analysis of the reaction mixture at 24 h post-filtration did not reveal significant amounts of metal leached into solution (<1 ppm Cr and 3 ppm Ni).

Oxidation of Compounds 2 and 3 over LDH Catalysts
In order to determine the efficacy of these catalysts with electron-rich substrates, model compound 2 was used as a starting material. Catalysts that returned a yield of >80% in the conversion of 1 to 1a were screened for catalytic activity in the oxidation of 2. Additionally, Ni-Al-LDH-1 was also used in the reaction for comparison purposes [21]. Oxidation of 2 with the aforementioned catalysts afforded two products, the expected ketone, 2a, as well as the alcohol elimination product, 2b. As can be seen from Table 5, catalysts containing nickel yielded higher amounts of the dehydration product 2a.
Dehydration was found to be most prominent for Ni-Al-LDH-2. On the other hand, copper-containing catalysts tended to be more selective towards the ketone product, 2a. In an attempt to elucidate the relationship between catalyst functionality and alcohol dehydration, acidity and basicity measurements were compared to catalyst performance. No absolute trend between acidity or basicity and alkene yield was elucidated, although the highest yields of 2b and 3b (see Tables 5 and 6) were obtained for the LDH catalysts possessing the highest numbers of acid and base sites (Ni-Al-LDH-1, Ni-Al-LDH-2, and Ni-Cr-LDH). Mechanistically this may occur via an E2-type mechanism in which the metal alkoxide is formed on the catalyst surface, followed by deprotonation of the β-carbon and subsequent elimination of the metal oxide to form the alkene. Although both Ni-Al-LDH catalysts both performed well, overall, oxidation with Ni-Cr-LDH resulted in the highest yield of 2a. In the case of Cu-Cr-LDH, near quantitative conversion of 2 was observed. However, the selectivity to 2a (ca. 50%), which was the only identifiable compound by GC/MS, was significantly lower than other LDH catalysts (entries 2-6, Table 5), presumably due to the conversion of 2 to unidentifiable products, evidenced by the darkening of the reaction mixture.

Oxidation of Compounds 2 and 3 over LDH Catalysts
In order to determine the efficacy of these catalysts with electron-rich substrates, model compound 2 was used as a starting material. Catalysts that returned a yield of >80% in the conversion of 1 to 1a were screened for catalytic activity in the oxidation of 2. Additionally, Ni-Al-LDH-1 was also used in the reaction for comparison purposes [21]. Oxidation of 2 with the aforementioned catalysts afforded two products, the expected ketone, 2a, as well as the alcohol elimination product, 2b. As can be seen from Table 5, catalysts containing nickel yielded higher amounts of the dehydration product 2a. Dehydration was found to be most prominent for Ni-Al-LDH-2. On the other hand, copper-containing catalysts tended to be more selective towards the ketone product, 2a.
In an attempt to elucidate the relationship between catalyst functionality and alcohol dehydration, acidity and basicity measurements were compared to catalyst performance. No absolute trend between acidity or basicity and alkene yield was elucidated, although the highest yields of 2b and 3b (see Tables 5 and 6) were obtained for the LDH catalysts possessing the highest numbers of acid and base sites (Ni-Al-LDH-1, Ni-Al-LDH-2, and Ni-Cr-LDH). Mechanistically this may occur via an E2type mechanism in which the metal alkoxide is formed on the catalyst surface, followed by deprotonation of the β-carbon and subsequent elimination of the metal oxide to form the alkene. Although both Ni-Al-LDH catalysts both performed well, overall, oxidation with Ni-Cr-LDH resulted in the highest yield of 2a. In the case of Cu-Cr-LDH, near quantitative conversion of 2 was observed. However, the selectivity to 2a (ca. 50%), which was the only identifiable compound by GC/MS, was significantly lower than other LDH catalysts (entries 2-6, Table 5), presumably due to the conversion of 2 to unidentifiable products, evidenced by the darkening of the reaction mixture. In order to further increase lignin-like functionality on the substrate and explore functional group sensitivity, the phenolic model compound 3 was chosen. As shown in Table 6, the use of Nicontaining catalysts for the oxidation of 3 favored the formation of the dehydration product 3b. Unfortunately, poor mass balances were obtained for the oxidation of compound 3 due to suspected polymerization (chromatographically immobile material). This suggests that phenolic groups may need to be protected, e.g., by benzylation, prior to benzylic oxidation. Unexpectedly, a small amount of benzaldehyde 3c was also formed during the oxidation of 3 as a result of Cα-Cβ bond cleavage. Aldehyde formation was most prevalent when the Ni-Cu-Cr-LDH and Ni-Al-LDH-1 were used as catalysts. Moreover, Cu-Cr-LDH was active in the oxidation of 3 but did not yield identifiable products. The production of 3b likely results in phenolic or styrenic coupling reactions leading to high molecular weight polymers. In order to further increase lignin-like functionality on the substrate and explore functional group sensitivity, the phenolic model compound 3 was chosen. As shown in Table 6, the use of Ni-containing catalysts for the oxidation of 3 favored the formation of the dehydration product 3b. Unfortunately, poor mass balances were obtained for the oxidation of compound 3 due to suspected polymerization (chromatographically immobile material). This suggests that phenolic groups may need to be protected, e.g., by benzylation, prior to benzylic oxidation. Unexpectedly, a small amount of benzaldehyde 3c was also formed during the oxidation of 3 as a result of C α -C β bond cleavage. Aldehyde formation was most prevalent when the Ni-Cu-Cr-LDH and Ni-Al-LDH-1 were used as catalysts. Moreover, Cu-Cr-LDH was active in the oxidation of 3 but did not yield identifiable products. The production of 3b likely results in phenolic or styrenic coupling reactions leading to high molecular weight polymers.

Catalyst Reusability Study
Catalyst reusability was studied using Ni-Al-LDH-1 and Ni-Cr-LDH in the oxidation of 1 ( Table  7). After the reaction, the catalysts were filtered and washed with THF and hexanes, and then dried in a vacuum oven prior to re-use. Re-usability tests for Ni-Al-LDH-1 and Ni-Cr-LDH demonstrated a significant decrease in activity upon successive use. The X-ray diffractogram of the spent Ni-Cr-LDH ( Figure S10) displayed similar peaks to the fresh catalyst with the exception of a new highly crystalline peak corresponding to chromium (III) oxide, while N2 physisorption analysis revealed a significant decrease in surface area (27.8 m 2 •g −1 post reaction), which is believed to be the result of phase segregation in the LDH, in addition to adsorbed organics blocking pores. Similarly, Ni-Al-LDH-1 displayed the characteristic LDH diffraction pattern but also a highly crystalline peak corresponding to Al(OH)3 ( Figure S11). From these observations, it is evident that the LDH structure of the catalysts was largely retained after use, although limited segregation of the M(OH)3 phase occurs (which decomposes, in the case of Cr, to Cr2O3). Similar results were obtained for LDHs tested with other substrates (see, for example, Figure S12). Other researchers have reported that full activity of LDH catalysts for the oxidation of benzyl alcohol is regained upon washing LDH catalysts with aqueous sodium carbonate [29,30]. After washing Ni-Cr-LDH with Na2CO3, a small amount of activity was regained. We believe that inability to completely regenerate the Ni-Cr-LDH catalyst may be related to the phase segregation observed by X-ray diffraction ( Figure S10). The effect of carbonate washing was more pronounced in the case of Ni-Al-LDH-1. Indeed, Ni-Al-LDH-1 showed no activity for the oxidation of compound 1 after the first use. However, after washing with carbonate solution, activity was completely regained. In fact, conversion increased from 72% to 100%, possibly due to an increase in the number of defect sites after reconstitution with Na2CO3.

Catalyst Reusability Study
Catalyst reusability was studied using Ni-Al-LDH-1 and Ni-Cr-LDH in the oxidation of 1 (Table 7). After the reaction, the catalysts were filtered and washed with THF and hexanes, and then dried in a vacuum oven prior to re-use. Re-usability tests for Ni-Al-LDH-1 and Ni-Cr-LDH demonstrated a significant decrease in activity upon successive use. The X-ray diffractogram of the spent Ni-Cr-LDH ( Figure S10) displayed similar peaks to the fresh catalyst with the exception of a new highly crystalline peak corresponding to chromium (III) oxide, while N 2 physisorption analysis revealed a significant decrease in surface area (27.8 m 2 ·g −1 post reaction), which is believed to be the result of phase segregation in the LDH, in addition to adsorbed organics blocking pores. Similarly, Ni-Al-LDH-1 displayed the characteristic LDH diffraction pattern but also a highly crystalline peak corresponding to Al(OH) 3 ( Figure S11). From these observations, it is evident that the LDH structure of the catalysts was largely retained after use, although limited segregation of the M(OH) 3 phase occurs (which decomposes, in the case of Cr, to Cr 2 O 3 ). Similar results were obtained for LDHs tested with other substrates (see, for example, Figure S12). Other researchers have reported that full activity of LDH catalysts for the oxidation of benzyl alcohol is regained upon washing LDH catalysts with aqueous sodium carbonate [29,30]. After washing Ni-Cr-LDH with Na 2 CO 3 , a small amount of activity was regained. We believe that inability to completely regenerate the Ni-Cr-LDH catalyst may be related to the phase segregation observed by X-ray diffraction ( Figure S10). The effect of carbonate washing was more pronounced in the case of Ni-Al-LDH-1. Indeed, Ni-Al-LDH-1 showed no activity for the oxidation of compound 1 after the first use. However, after washing with carbonate solution, activity was completely regained. In fact, conversion increased from 72% to 100%, possibly due to an increase in the number of defect sites after reconstitution with Na 2 CO 3 . Other workers have reported that LDH anions may play an integral role in alcohol oxidation as evidenced by the reduced catalyst activity when anions are absent or substituted by another anion in the LDH [21,25]. In order to ascertain whether carbonate acts as a stoichiometric base, the amount of CO 3 2− present in each LDH was calculated based on the idealized LDH formula where χ is the trivalent metal ratio, A is the anionic species, and n is the charge of the anionic species. The water content was purposefully ignored as this can vary between LDHs [26]. As can be seen in Table 8, CO 3 2− was present in a sub-stoichiometric quantity compared to the substrate 1; hence, carbonate did not act as a stoichiometric base in these oxidation reactions.

Mechanistic Considerations
In order to determine whether oxidation proceeds via a two-electron or radical pathway, 1-cyclopropy-1-phenylcarbinol was used as a probe molecule. If oxidation proceeds via a benzylic radical then the highly strained cyclopropyl ring would open, yielding a linear propyl chain. On the other hand, if the reaction proceeds through a hydride shift (as shown in Figure 2) the cyclopropyl ring would remain after benzylic oxidation. Analysis of the reaction mixture post-oxidation revealed the presence of cyclopropyl phenyl ketone in 64% yield, with no evidence of the ring-opening product (Scheme 1). This suggests that the oxidation of benzylic alcohols to ketones proceeds through a two-electron pathway. Other workers have reported that LDH anions may play an integral role in alcohol oxidation as evidenced by the reduced catalyst activity when anions are absent or substituted by another anion in the LDH [21,25]. In order to ascertain whether carbonate acts as a stoichiometric base, the amount of CO3 2− present in each LDH was calculated based on the idealized LDH formula [M 2+ 1−χM 3+ χ(OH)2] χ+ (A n− )χ/n, where χ is the trivalent metal ratio, A is the anionic species, and n is the charge of the anionic species. The water content was purposefully ignored as this can vary between LDHs [26]. As can be seen in Table 8, CO3 2− was present in a sub-stoichiometric quantity compared to the substrate 1; hence, carbonate did not act as a stoichiometric base in these oxidation reactions.

Mechanistic Considerations
In order to determine whether oxidation proceeds via a two-electron or radical pathway, 1cyclopropy-1-phenylcarbinol was used as a probe molecule. If oxidation proceeds via a benzylic radical then the highly strained cyclopropyl ring would open, yielding a linear propyl chain. On the other hand, if the reaction proceeds through a hydride shift (as shown in Figure 2) the cyclopropyl ring would remain after benzylic oxidation. Analysis of the reaction mixture post-oxidation revealed the presence of cyclopropyl phenyl ketone in 64% yield, with no evidence of the ring-opening product (Scheme 1). This suggests that the oxidation of benzylic alcohols to ketones proceeds through a twoelectron pathway. Given that catalytic activity is regained and even enhanced after washing with sodium carbonate, it follows that catalysis likely occurs on the catalyst edge sites (110 plane) in LDHs or an equivalent site where interlamellar carbonate anions are exposed to the reactants. Figure 2 shows a plausible mechanism, which is a modified version of that proposed by Tang et al. [30], in which the alcohol is first deprotonated by carbonate to form an alkoxide, which coordinates to an unsaturated metal site. Hydroperoxide oxidation of the metal alkoxide with a concomitant hydride shift from the alkoxide to the hydroperoxide results in net alcohol oxidation and the regeneration of metal hydroxide. Deprotonation of bicarbonate by the metal hydroxide forms water and regenerates a coordinatively unsaturated metal site.

Oxidation of Lignin Model Dimer Compounds over LDH Catalysts
Lignin is an amorphous biopolymer synthesized in planta in the secondary cell walls via oxidative radical condensation of three monolignols (sinapyl, coniferyl, and p-coumaryl alcohol) [31]. As such, it is composed of a variety of linkages, the most abundant of which is the β-O-4 linkage, which can compose as much as 60% of the linkages in hardwood species [31]. Moreover, several Given that catalytic activity is regained and even enhanced after washing with sodium carbonate, it follows that catalysis likely occurs on the catalyst edge sites (110 plane) in LDHs or an equivalent site where interlamellar carbonate anions are exposed to the reactants. Figure 2 shows a plausible mechanism, which is a modified version of that proposed by Tang et al. [30], in which the alcohol is first deprotonated by carbonate to form an alkoxide, which coordinates to an unsaturated metal site. Hydroperoxide oxidation of the metal alkoxide with a concomitant hydride shift from the alkoxide to the hydroperoxide results in net alcohol oxidation and the regeneration of metal hydroxide. Deprotonation of bicarbonate by the metal hydroxide forms water and regenerates a coordinatively unsaturated metal site.

Oxidation of Lignin Model Dimer Compounds over LDH Catalysts
Lignin is an amorphous biopolymer synthesized in planta in the secondary cell walls via oxidative radical condensation of three monolignols (sinapyl, coniferyl, and p-coumaryl alcohol) [31]. As such, it is composed of a variety of linkages, the most abundant of which is the β-O-4 linkage, which can compose as much as 60% of the linkages in hardwood species [31]. Moreover, several recent reports on the depolymerization of lignin have focused on the benzylic oxidation of the β-O-4 linkage, followed by a secondary cleavage step [14,16,17,32,33].
Inorganics 2018, 6, x FOR PEER REVIEW 9 of 18 recent reports on the depolymerization of lignin have focused on the benzylic oxidation of the β-O-4 linkage, followed by a secondary cleavage step [14,16,17,32,33]. While promising results were obtained when benchtop reactions were performed on nonphenolic compounds 1 and 2, only the dehydration product (4c, 100% yield) was observed when lignin model dimer 4 was subjected to optimized reaction conditions (100% O2, 0.5 g Ni-Al-LDH-1, 150 °C, in DPE). Thus, in order to determine if higher temperatures would enhance the rate of oxidation rather than dehydration, a pressurized reaction system was used. Indeed, when lignin model dimer compounds were reacted under slightly elevated temperatures (i.e., 180 °C) using 8% O2/N2 (50 bar) significant conversion was observed (Tables 9-11). It should be noted that the use of pressurized oxygen significantly increases safety concerns for reactions in organic media. Indeed, Stahl and coworkers [34] recently reported limiting oxygen concentrations (LOC) for nine organic solvents, finding that ca. 8% O2 counter-balanced with inert gas was generally non-combustible. Thus, in this study 8% oxygen counter-balanced with nitrogen was used, which provides a nearly stoichiometric amount of oxygen (ca. 3 equiv.) for the oxidation of lignin model dimers. In addition to addressing safety concerns, the use of near stoichiometric amounts of oxygen limits over-oxidation to dicarboxylic acids, which are common products of aromatic ring over-oxidation [35]. While promising results were obtained when benchtop reactions were performed on non-phenolic compounds 1 and 2, only the dehydration product (4c, 100% yield) was observed when lignin model dimer 4 was subjected to optimized reaction conditions (100% O 2 , 0.5 g Ni-Al-LDH-1, 150 • C, in DPE). Thus, in order to determine if higher temperatures would enhance the rate of oxidation rather than dehydration, a pressurized reaction system was used. Indeed, when lignin model dimer compounds were reacted under slightly elevated temperatures (i.e., 180 • C) using 8% O 2 /N 2 (50 bar) significant conversion was observed (Tables 9-11). It should be noted that the use of pressurized oxygen significantly increases safety concerns for reactions in organic media. Indeed, Stahl and coworkers [34] recently reported limiting oxygen concentrations (LOC) for nine organic solvents, finding that ca. 8% O 2 counter-balanced with inert gas was generally non-combustible. Thus, in this study 8% oxygen counter-balanced with nitrogen was used, which provides a nearly stoichiometric amount of oxygen (ca. 3 equiv.) for the oxidation of lignin model dimers. In addition to addressing safety concerns, the use of near stoichiometric amounts of oxygen limits over-oxidation to dicarboxylic acids, which are common products of aromatic ring over-oxidation [35].  Although modest amounts of the ketone resulting from benzylic oxidation were detected, small molecules resulting from cleavage of the model linkages were observed in more substantial yields (Table 9). Wang et al. [36] recently reported that lignin β-O-4 models oxidized at the benzylic position are more easily fragmented than the benzylic alcohol analogue, due to the weakening of the Cβ-O4 bond by approximately 87 kJ/mol. Thus, it follows that the modest yields of 4d are explained by the ready cleavage of the Cβ-O4 bond, as indicated by the observation of the phenol 4b. Product 4a, which likely also results from oxidative cleavage of 4d, was generally present in higher yield than 4b presumably due to phenol polymerization; while the mechanism for oxidative cleavage of 4d is unclear, it is presumed to undergo a similar route as that observed by Mottweiler et al. [37]. Indeed, Mottweiler et al. noted that the use of a Cu-V-LDH in the presence of O2 resulted in a large amount of the A-ring acid and aldehyde. In addition, enol ether product 4c was observed, resulting from benzylic alcohol dehydration. Product 4c is particularly interesting because enol ethers are known to undergo hydrolysis under acidic conditions [38]. This production of enol ethers in lignin would result in an easily hydrolysable linkage. Additionally, product 4e, that likely results from non-oxidative cleavage of 4d, was observed as a minor co-product. Products such as 4e are commonly observed in heterogeneous oxidation reactions of lignin model compounds [39][40][41][42]. Given the tendency of enol ethers to undergo metal-catalyzed cleavage reactions under oxidizing conditions, one can speculate that benzoic acid 4a is produced from the enol ether (4c and 5c) [43]. Indeed, when 4c was used as the feedstock in a control experiment (using Cu-Cr-LDH as catalyst), 4a was produced in low yield (7%) after 24 h. Surprisingly, the Ni-Cr-LDH catalyst, which successfully oxidized compounds 1-3, produced modest conversions in the cases of dimer models 4-6. This may be due, in part at least, to the catalyst's small average pore diameter (2.9 nm). Unlike compounds 1-3, which are relatively small, lignin dimer model compounds 4-6 (ca. 1.5 nm) [44] approach the pore diameter of Ni-Cr-LDH (Table 1). Moreover, other catalysts with larger pore diameters showed increased conversion of compounds 4-5. Indeed, Ni-Al-LDH-1, with a pore diameter of 7.3 nm, was found to be the most active catalyst for conversion of the lignin model compounds used in this study, resulting in >99% conversion of models 4 and 5 (Tables 9 and 10). Unfortunately, yields of individual products were low (<20%).
In these experiments phenols resulting from β-aryl ether cleavage were recovered in low yields. As commonly reported for oxidation of lignin model compounds, phenols are often converted into unidentifiable products [16,17,36]. In order to investigate the stability of phenols under the reaction conditions, guaiacol (4b) was subjected to the same conditions in the absence of catalyst. After 16 h at 180 °C under 8% O2/N2 (720 psi), guaiacol was converted (76%) to unidentifiable products and Although modest amounts of the ketone resulting from benzylic oxidation were detected, small molecules resulting from cleavage of the model linkages were observed in more substantial yields (Table 9). Wang et al. [36] recently reported that lignin β-O-4 models oxidized at the benzylic position are more easily fragmented than the benzylic alcohol analogue, due to the weakening of the C β -O 4 bond by approximately 87 kJ/mol. Thus, it follows that the modest yields of 4d are explained by the ready cleavage of the C β -O 4 bond, as indicated by the observation of the phenol 4b. Product 4a, which likely also results from oxidative cleavage of 4d, was generally present in higher yield than 4b presumably due to phenol polymerization; while the mechanism for oxidative cleavage of 4d is unclear, it is presumed to undergo a similar route as that observed by Mottweiler et al. [37]. Indeed, Mottweiler et al. noted that the use of a Cu-V-LDH in the presence of O 2 resulted in a large amount of the A-ring acid and aldehyde. In addition, enol ether product 4c was observed, resulting from benzylic alcohol dehydration. Product 4c is particularly interesting because enol ethers are known to undergo hydrolysis under acidic conditions [38]. This production of enol ethers in lignin would result in an easily hydrolysable linkage. Additionally, product 4e, that likely results from non-oxidative cleavage of 4d, was observed as a minor co-product. Products such as 4e are commonly observed in heterogeneous oxidation reactions of lignin model compounds [39][40][41][42]. Given the tendency of enol ethers to undergo metal-catalyzed cleavage reactions under oxidizing conditions, one can speculate that benzoic acid 4a is produced from the enol ether (4c and 5c) [43]. Indeed, when 4c was used as the feedstock in a control experiment (using Cu-Cr-LDH as catalyst), 4a was produced in low yield (7%) after 24 h. Surprisingly, the Ni-Cr-LDH catalyst, which successfully oxidized compounds 1-3, produced modest conversions in the cases of dimer models 4-6. This may be due, in part at least, to the catalyst's small average pore diameter (2.9 nm). Unlike compounds 1-3, which are relatively small, lignin dimer model compounds 4-6 (ca. 1.5 nm) [44] approach the pore diameter of Ni-Cr-LDH (Table 1). Moreover, other catalysts with larger pore diameters showed increased conversion of compounds 4-5. Indeed, Ni-Al-LDH-1, with a pore diameter of 7.3 nm, was found to be the most active catalyst for conversion of the lignin model compounds used in this study, resulting in >99% conversion of models 4 and 5 (Tables 9 and 10). Unfortunately, yields of individual products were low (<20%).
In these experiments phenols resulting from β-aryl ether cleavage were recovered in low yields. As commonly reported for oxidation of lignin model compounds, phenols are often converted into unidentifiable products [16,17,36]. In order to investigate the stability of phenols under the reaction conditions, guaiacol (4b) was subjected to the same conditions in the absence of catalyst. After 16 h at 180 • C under 8% O 2 /N 2 (720 psi), guaiacol was converted (76%) to unidentifiable products and evident darkening of the reaction solution was observed. This leads us to the conclusion that polymerization of phenols in the presence of oxygen was responsible for their low yields. evident darkening of the reaction solution was observed. This leads us to the conclusion that polymerization of phenols in the presence of oxygen was responsible for their low yields. While models 4 and 5 serve as sufficiently complex models to establish reactivity trends, they do not accurately represent lignin linkages. Consequently, to better reflect native and technical lignins, model complexity was increased by the addition of a γ-carbinol group (compound 6; Table  11). The addition of a γ-carbinol provides another alcohol that can be oxidized. Once oxidized at the α or γ-position, 6 can undergo retro-aldol reactions further complicating the product mixture. Indeed, a retro-aldol reaction at the oxidized γ-position produces 4 via loss of formaldehyde, while oxidation at the α-position produces a ketone that can also undergo further oxidation.
The oxidative fragmentation of model 6 was investigated using Ni-Al-LDH-1, Cu-Cr-LDH, and Ni-Cr-LDH. Unexpectedly, Ni-Al-LDH-1 catalyzed oxidation resulted in only 34% conversion (Table  11), whereas Cu-Cr-LDH afforded similar conversion as for models 4 and 5 (80-90%). In contrast, Ni-Cr-LDH, which showed similar reactivity trends for models 4 and 5, showed relatively lower conversion for 6 (23%). The benzoic acid (4a) resulting from cleavage of the Cα-Cβ bond was observed in low yield (<7%) for all three catalysts. Furthermore, compounds similar to those produced after the oxidation of 4 and 5 (e.g., 6c) were not observed when 6 was subjected to the same reaction conditions.
The difference in the reactivity of 6 compared to 4 and 5 may have been due to diverging reaction pathways. It is hypothesized that reaction intermediates included both benzylic ketones formed via oxidation and enol ethers via dehydration. Basic sites were likely responsible for the deprotonation of the benzylic carbinol for catalysts such as Ni-Al-LDH-1. After the substrate was coordinated to the catalyst surface, base sites likely deprotonated the β-carbon. Thus, it stands to reason that substitution at the β-carbon by a γ-carbinol likely made deprotonation of the β-carbon more difficult, as suggested by the absence of 6c in the product mixture. While models 4 and 5 serve as sufficiently complex models to establish reactivity trends, they do not accurately represent lignin linkages. Consequently, to better reflect native and technical lignins, model complexity was increased by the addition of a γ-carbinol group (compound 6; Table 11). The addition of a γ-carbinol provides another alcohol that can be oxidized. Once oxidized at the α or γ-position, 6 can undergo retro-aldol reactions further complicating the product mixture. Indeed, a retro-aldol reaction at the oxidized γ-position produces 4 via loss of formaldehyde, while oxidation at the α-position produces a ketone that can also undergo further oxidation.
The oxidative fragmentation of model 6 was investigated using Ni-Al-LDH-1, Cu-Cr-LDH, and Ni-Cr-LDH. Unexpectedly, Ni-Al-LDH-1 catalyzed oxidation resulted in only 34% conversion (Table 11), whereas Cu-Cr-LDH afforded similar conversion as for models 4 and 5 (80-90%). In contrast, Ni-Cr-LDH, which showed similar reactivity trends for models 4 and 5, showed relatively lower conversion for 6 (23%). The benzoic acid (4a) resulting from cleavage of the C α -C β bond was observed in low yield (<7%) for all three catalysts. Furthermore, compounds similar to those produced after the oxidation of 4 and 5 (e.g., 6c) were not observed when 6 was subjected to the same reaction conditions.
The difference in the reactivity of 6 compared to 4 and 5 may have been due to diverging reaction pathways. It is hypothesized that reaction intermediates included both benzylic ketones formed via oxidation and enol ethers via dehydration. Basic sites were likely responsible for the deprotonation of the benzylic carbinol for catalysts such as Ni-Al-LDH-1. After the substrate was coordinated to the catalyst surface, base sites likely deprotonated the β-carbon. Thus, it stands to reason that substitution at the β-carbon by a γ-carbinol likely made deprotonation of the β-carbon more difficult, as suggested by the absence of 6c in the product mixture.  To investigate the reason for the low yields of the ketone resulting from benzylic oxidation, models 4d-6d were subjected to the same reaction conditions as 4-6 (Tables 9-11) in the absence of a catalyst. As shown in Table 12, oxidized lignin model dimers 4d-6d were converted near quantitatively; this indicates that lignin models, once oxidized, were easily cleaved under oxidative conditions at 180 °C. Moreover, 4a was obtained in good yields from 4d-6d, indicating that the primary pathway for the production of 4a from 4-6 began with benzylic oxidation. The hypothesis that 4d-6d were converted to 4a as a result of oxidation by molecular oxygen was confirmed by the finding that in the absence of O2, 5d and 6d were not converted to identifiable products. Evidently, oxygen was a strong enough oxidant at elevated temperatures to cleave the Cβ-O bond in the ketone form of β-O-4 model compounds, indicating that oxidation of the benzylic alcohol group in β-O-4 models is the more challenging step in terms of the kinetics. To investigate the reason for the low yields of the ketone resulting from benzylic oxidation, models 4d-6d were subjected to the same reaction conditions as 4-6 (Tables 9-11) in the absence of a catalyst. As shown in Table 12, oxidized lignin model dimers 4d-6d were converted near quantitatively; this indicates that lignin models, once oxidized, were easily cleaved under oxidative conditions at 180 • C. Moreover, 4a was obtained in good yields from 4d-6d, indicating that the primary pathway for the production of 4a from 4-6 began with benzylic oxidation. The hypothesis that 4d-6d were converted to 4a as a result of oxidation by molecular oxygen was confirmed by the finding that in the absence of O 2 , 5d and 6d were not converted to identifiable products. Evidently, oxygen was a strong enough oxidant at elevated temperatures to cleave the C β -O bond in the ketone form of β-O-4 model compounds, indicating that oxidation of the benzylic alcohol group in β-O-4 models is the more challenging step in terms of the kinetics.  To investigate the reason for the low yields of the ketone resulting from benzylic oxidation, models 4d-6d were subjected to the same reaction conditions as 4-6 (Tables 9-11) in the absence of a catalyst. As shown in Table 12, oxidized lignin model dimers 4d-6d were converted near quantitatively; this indicates that lignin models, once oxidized, were easily cleaved under oxidative conditions at 180 °C. Moreover, 4a was obtained in good yields from 4d-6d, indicating that the primary pathway for the production of 4a from 4-6 began with benzylic oxidation. The hypothesis that 4d-6d were converted to 4a as a result of oxidation by molecular oxygen was confirmed by the finding that in the absence of O2, 5d and 6d were not converted to identifiable products. Evidently, oxygen was a strong enough oxidant at elevated temperatures to cleave the Cβ-O bond in the ketone form of β-O-4 model compounds, indicating that oxidation of the benzylic alcohol group in β-O-4 models is the more challenging step in terms of the kinetics.

Catalyst Preparation
Catalysts were prepared by co-precipitation under conditions of low supersaturation. In general, two solutions, one containing metal nitrates and the other containing a mixture of NaOH and Na 2 CO 3 , were added simultaneously and stirred while maintaining a constant pH (generally 9-10, with the exception of Cu-containing LDHs which were precipitated at lower pH). The concentration of the metal nitrate solution used was typically ca. 1.5 M (total metals), while the base solution contained Na 2 CO 3 (ca. 1.0 M) and the calculated amount of NaOH (ca. 3 M) required for complete reaction with the divalent and trivalent metal ions. The solutions were mixed at room temperature at an addition rate of ca. 3 mL·min −1 , with vigorous mechanical stirring. Unless otherwise stated, the precipitate was aged in the synthesis solution overnight at 70 • C and isolated by a cycle of centrifuging/decanting/washing with deionized water until the washings reached a neutral pH. The resulting solid was dried at 60 • C in vacuo. Additional synthetic details can be found in the supporting information. All catalysts were stored under atmospheric conditions. Unless otherwise specified, catalysts were used without further pretreatment.

Catalyst Characterization
Surface area, average pore diameter, and pore volume were determined using a Tristar 3000 porosity system (Micromeritics, Norcross, GA, USA) or a Gemini VII analyzer (Micromeritics, Norcross, GA, USA) using the Brunauer-Emmett-Teller (BET) method by N 2 adsorption at −196 • C. Samples were outgassed under vacuum for at least 6 h at 160 • C prior to measurement. Note that powder X-ray diffraction (XRD) measurements confirmed that the LDH structure was retained after this pre-treatment (see Figure S1). XRD measurements were performed on a X'Pert system (PANanalytical, Almelo, The Netherlands) using Cu Kα radiation (λ = 1.5406 Å) and a step size of 0.02 • . Elemental analysis was performed on a 720-ES inductively coupled plasma-optical emission spectrometer (Varian/Agilent, Santa Clara, CA, USA). Scanning electron microscopy (SEM) was performed on a S-2700 instrument equipped with a PGT EDS analyzer (Hitachi, Dallas, TX, USA) with a thin window detector and a LaB 6 electron gun. FT-IR spectroscopy was performed on a Nicolet 6700 FT-IR instrument (ThermoFisher Scientific, Waltham, MA, USA) equipped with a smart iTR diamond attenuated total reflection (ATR) attachment. In all cases 32 scans were taken with a resolution of 4 cm −1 . Details of the pulsed CO 2 chemisorption and NH 3 -TPD measurements are given in the Supporting Information.

General Procedure for Oxidation of 1-Phenyl Ethanol Derivatives
In a typical reaction, the alcohol compound (2 mmol), solvent (10 mL), and catalyst (0.5 g) were added to a three-neck flask equipped with an oxygen bubbler, a reflux condenser, and a glass stopper. The reaction mixture was stirred at 150 • C for 24 h, after which it was cooled to room temperature and dichloromethane (ca. 10 mL) was added. The reaction mixture was then filtered through Whatman 1 filter paper. The catalyst was washed with dichloromethane or tetrahydrofuran and the washings added to the filtrate. When compound 3 was used as the substrate, 1,4-dimethoxybenzene (0.25 g, 1.8 mmol) was added to the reaction mixture prior to reaction as an internal standard. Conversion, selectivity, and yield were determined using GC (for details see Supplementary Materials).

Conclusions
LDH materials containing a variety of first row transition metal ions were found to be active catalysts for the oxidation of benzylic alcohols and lignin model dimer compounds using phenyl ether as solvent and O 2 as the terminal oxidant. Upon repeated use, catalyst activity declined, although washing the spent catalyst (i.e., Ni-Al-LDH-1 and Ni-Cr-LDH) with aqueous Na 2 CO 3 was found to restore activity in the oxidation of 1; this suggests that carbonate species play an essential role in the oxidation reaction. In the conversion of 2 and 3, Ni-containing LDH catalysts were found to show activity for alcohol dehydration, in parallel to alcohol oxidation. Moreover, in the case of phenyl ethanol derivative 3, the formation of significant amounts of unidentifiable products suggests that phenol protection is a necessity in order to prevent the occurrence of polymerization reactions. Typically, for all catalysts only trace amounts of the ketone resulting from benzylic alcohol oxidation were observed for the β-O-4 model compounds. Rather, monomeric products arising from β-aryl ether cleavage were formed. This observation is consistent with the higher reactivity of the ketones, resulting from weakening of the C β -O 4 bond that was shown to be aerobically cleaved at 180 • C in the absence of catalyst.