Effects of Surface Oxygen-Containing Groups of the Flowerlike Carbon Nanosheets on Palladium Dispersion, Catalytic Activity and Stability in Hydrogenolytic Debenzylation of Tetraacetyldibenzylhexaazaisowurtzitane

Surface Oxygen-Containing Flowerlike Abstract: The inﬂuence of the surface chemical properties of the carbon support on the Pd dispersion, activity and stability of Pd(OH) 2 /C catalyst for the hydrogenolytic debenzylation of tetraacetyldibenzylhexaazaisowurtzitane (TADB) was studied in detail. The ﬂowerlike nanosheet carbon material (NSC) was chosen as the pristine support, meanwhile chemical oxidation with nitric acid and physical calcination at 600 ◦ C treatments were used to modify its surface properties, which were denoted as NSC ox-2 (treated with 20 wt% HNO 3 ) and NSC-600, respectively. The three carbon supports and the corresponding catalysts of Pd/NSC, Pd/NSC-600, and Pd/NSC ox-2 were characterized by scanning electron microscope (SEM), transmission electron microscopy (TEM), nitrogen sorption isotherm measurement (BET), powder X-ray diffraction (XRD), Raman spectra, X-ray photoelectron spectra (XPS), temperature-programmed desorption (TPD), temperature-programmed reduction (H 2 -TPR), thermogravimetric analysis (TG), and element analysis. The debenzylation activities of Pd/NSC, Pd/NSC-600, and Pd/NSC ox-2 , as well as the three catalysts after pre-reduction treatment were also evaluated. It was found that the activity and stability of the Pd(OH) 2 /C catalysts in the debenzylation reaction highly depended on the content of surface oxygen-containing groups of the carbon support.


Introduction
Carbon materials have been widely used as support to fabricate the heterogeneous catalysts due to its high stability, large tunable specific surface area, chemical inertness, easy tailorable surface chemical properties [1][2][3], especially for the precious metal catalysts. Palladium on carbon is an important class of catalyst, which can be used in several industrial chemical processes including hydrogenation, dehydrogenation, hydrogenolysis, hydro-dechlorination, nitroarenes reduction, and C-C coupling [4][5][6]. It is well known that the activity and selectivity of the Pd-based catalysts supported on carbons, either Pd/C or Pd(OH) 2 /C, strongly depend on the preparation methods, the physicochemical properties of Pd particles, such as size, morphology, and dispersion of Pd species, and the nature and structure of carbon supports, involving porous structures and surface properties [2,[7][8][9]. There are two main general routes for the preparation of Pd-based catalysts on carbons: the ion exchange and the deposition-precipitation methods. In the case of ion exchange, both anionic and cationic palladium precursors can be used [10][11][12], resulting well dispersed Pd-based catalysts. In this process, the metal loading and dispersion aazaisowurtzitane (TAIW), which can be nitrated easily to give CL-20 (Scheme 1) [39,40]. However, the transformation of TADB to TAIW needed a relative intensive condition, including the addition of acidic cocatalyst and a high dose of Pd catalyst, so TADB was a proper substrate to distinguish the performance of the debenzylation catalysts.
As an outstanding representative of high energy density materials (HEDM), hexanitrohexaazaisowurtzitane (CL-20) has attracted much attention in the past decades. Hexabenzylhexaazaisowurtzitane (HBIW) has been the main precursor for the fabrication of CL-20 [38], but it cannot be nitrated directly to afford CL-20 due to its low stability of the cage framework. So the debenzylation and transformation of HBIW became the key step for the synthesis of CL-20. Usually, HBIW is firstly converted to TADB in the presence of acylation agent, and subsequently TADB can be further transformed to tetraacetylhexaazaisowurtzitane (TAIW), which can be nitrated easily to give CL-20 (Scheme 1) [39,40]. However, the transformation of TADB to TAIW needed a relative intensive condition, including the addition of acidic cocatalyst and a high dose of Pd catalyst, so TADB was a proper substrate to distinguish the performance of the debenzylation catalysts. Scheme 1. Route for the synthesis of CL-20.
In our previous work [41], a flowerlike nanosheet carbon (NSC) material was proved to be a suitable palladium catalyst support, which showed an evidently hierarchical porous structure. Herein, the chemical treatment, oxidation using various concentrations of HNO3, and a physical treatment, calcination at 600 °C under nitrogen atmosphere, were used to adjust the surface chemical properties of the NSC samples. It was found that the activity and stability of the corresponding Pd catalyst in the debenzylation reaction of tetraacetyldibenzylhexaazaisowurtzitane (TADB) highly depended on the content of surface oxygen-containing groups of carbon support.

Activities and Stabilities of the Catalysts in Hydrogenolytic Debenzylation of TADB
The catalytic activities of Pd/NSC, Pd/NSC-600, and Pd/NSCox-2 were evaluated by using the hydrogenolytic debenzylation of TADB as a probe reaction. Considering the solubility of TADB and the promotional effect of acidic medium to the debenzylation reaction, acetic acid was chosen as a solvent. The hydrogen consumption amounts as a function of reaction times are shown in Figure 1. It could be seen that the hydrogen consumption rates over Pd/NSCox-2 and Pd/NSC catalysts were significantly faster than that of Pd/NSC-600 in the first 2 h, indicating that the Pd/NSCox-2 and Pd/NSC had higher activities than Pd/NSC-600. After that the hydrogen consumption rates of the three catalysts decreased gradually. The total hydrogen consumption amount of Pd/NSCox-2 within 10 h was similar to that of Pd/NSC, which was higher than the data of Pd/NSC-600. The activities of the three catalysts was in order of Pd/NSCox-2 > Pd/NSC > Pd/NSC-600 in the whole process, revealing that the pretreatments of carbon support with nitric acid and calcination at high temperature under N2 flow could improve or inhibit the activities of the corresponding catalysts, respectively. These results were consistent with the product yield and the substrate conversion (Table S1). It can be found that the activity of Pd/NSCox-2 is much better than that of the reported catalysts. In our previous work [41], a flowerlike nanosheet carbon (NSC) material was proved to be a suitable palladium catalyst support, which showed an evidently hierarchical porous structure. Herein, the chemical treatment, oxidation using various concentrations of HNO 3 , and a physical treatment, calcination at 600 • C under nitrogen atmosphere, were used to adjust the surface chemical properties of the NSC samples. It was found that the activity and stability of the corresponding Pd catalyst in the debenzylation reaction of tetraacetyldibenzylhexaazaisowurtzitane (TADB) highly depended on the content of surface oxygen-containing groups of carbon support.

Activities and Stabilities of the Catalysts in Hydrogenolytic Debenzylation of TADB
The catalytic activities of Pd/NSC, Pd/NSC-600, and Pd/NSC ox-2 were evaluated by using the hydrogenolytic debenzylation of TADB as a probe reaction. Considering the solubility of TADB and the promotional effect of acidic medium to the debenzylation reaction, acetic acid was chosen as a solvent. The hydrogen consumption amounts as a function of reaction times are shown in Figure 1. It could be seen that the hydrogen consumption rates over Pd/NSC ox-2 and Pd/NSC catalysts were significantly faster than that of Pd/NSC-600 in the first 2 h, indicating that the Pd/NSC ox-2 and Pd/NSC had higher activities than Pd/NSC-600. After that the hydrogen consumption rates of the three catalysts decreased gradually. The total hydrogen consumption amount of Pd/NSC ox-2 within 10 h was similar to that of Pd/NSC, which was higher than the data of Pd/NSC-600. The activities of the three catalysts was in order of Pd/NSC ox-2 > Pd/NSC > Pd/NSC-600 in the whole process, revealing that the pretreatments of carbon support with nitric acid and calcination at high temperature under N 2 flow could improve or inhibit the activities of the corresponding catalysts, respectively. These results were consistent with the product yield and the substrate conversion (Table S1). It can be found that the activity of Pd/NSC ox-2 is much better than that of the reported catalysts. It was found that the Pd species on the carbon surface were characterized by a high mobility. Pre-reduction might lead to the aggregation of Pd particles on carbon with a consequent decrease in the Pd dispersion and the catalytic activity in the debenzylation reaction [24,42]. In order to understand the contribution of the surface oxygen-containing groups on carbon to the stability of Pd/C catalyst, the pre-reduction was performed on the three catalysts. The activities data of the resulted samples, It was found that the Pd species on the carbon surface were characterized by a high mobility. Pre-reduction might lead to the aggregation of Pd particles on carbon with a Catalysts 2021, 11, 441 4 of 17 consequent decrease in the Pd dispersion and the catalytic activity in the debenzylation reaction [24,42]. In order to understand the contribution of the surface oxygen-containing groups on carbon to the stability of Pd/C catalyst, the pre-reduction was performed on the three catalysts. The activities data of the resulted samples, Pd/NSC ox-2 -R, Pd/NSC-R, and Pd/NSC-600-R, are also given in Figure 1a. It could be seen that the hydrogen consumption amounts at a desired time over Pd/NSC ox-2 -R, Pd/NSC-R, and Pd/NSC-600-R appeared different degrees of decline comparing to the corresponding unreduced catalysts. However, the decrease in the activity of Pd/NSC ox-2 -R was much lower comparing to the other two, implying the high stability of Pd/NSC ox-2 in the pre-reduction process. In addition, the stability of Pd/NSC ox-2 , Pd/NSC, and Pd/NSC-600 was also evaluated by the recycling experiment in the debenzylation reaction of TADB under a similar condition. As shown in Figure 1b, the hydrogen consumption amounts of the re-used Pd/NSC ox-2 and Pd/NSC catalysts were larger than that of the re-used Pd/NSC-600. Moreover, the reaction rate over Pd/NSC ox-2 in the whole process was faster than over the other two. The total hydrogen consumption amount of the re-used Pd/NSC ox-2 within 10 h was about 3.0 L, which was much higher than that of the re-used Pd/NSC-600 (0.9 L). In addition, the yield of product and the conversion of TADB over the re-used Pd/NSC ox-2 were 75 and 83%, respectively (Table S1), which were significantly higher than the data of the re-used Pd/NSC (60 and 70%) and Pd/NSC-600 (10 and 15%), revealing the best stability of Pd/NSC ox-2 in the hydrogenolytic debenzylation reaction. All these results indicated that the chemical nature of the surface groups exerted a considerable influence on the interaction of the Pd species with the carbon carrier [3]. In particular, the presence of relative high amounts of oxygen-containing groups enabled higher activities and stabilities of the catalysts.
In addition, the activity declined trends of the three catalysts under three conditions: fresh one, after pre-reduction and the recovered one, were displayed in Figure 2. It could be seen that the pre-reduction treatment might cause the activity decreases of Pd/NSC and Pd/NSC-600 in some extent possibly due to the sintering of Pd particles and consequently the decrease in Pd dispersion in the reduction process. The activities of the re-used Pd/NSC and Pd/NSC-600 decreased further in the recycling experiments, particular for the Pd/NSC-600 catalyst, implying that there was one more factor leading to the catalyst deactivation, excepting the reduction effect of the hydrogenolytic debenzylation reaction itself. For Pd/NSC ox-2 , the pre-reduction treatment had little effect on its activity, however it decreased in the recycling experiment, further revealing the existence of other reason that caused the catalyst deactivation, although it was not clearly now.

Characterization of the Carbon Supports
The morphology of the NSC, NSC-600, and NSCox-2 samples is showed in Figure 3, and similar flowerlike nanosheet structures were observed, indicating that the macrostructure of carbon supports did not change obviously during the chemical or physical treatment.

Characterization of the Carbon Supports
The morphology of the NSC, NSC-600, and NSC ox-2 samples is showed in Figure 3, and similar flowerlike nanosheet structures were observed, indicating that the macrostructure of carbon supports did not change obviously during the chemical or physical treatment. Figure 2. Yields of the product (columns) and conversions (points) of TADB in the hydrogenolysis reaction over various catalysts. Pd/NSC-1, Pd/NSC-600-1, and Pd/NSCox-2-1 presented the re-used catalysts.

Characterization of the Carbon Supports
The morphology of the NSC, NSC-600, and NSCox-2 samples is showed in Figure 3, and similar flowerlike nanosheet structures were observed, indicating that the macrostructure of carbon supports did not change obviously during the chemical or physical treatment. The N2 adsorption-desorption isotherms of the three carbon supports are shown in Figure 4. The specific surface area, and pore volume of these supports are listed in Table  1. The isotherms of NSC, NSC-600, and NSCox-2 samples showed a combination of type I and type IV curves with a type H3 hysteresis loop revealing the existence of the interparticle mesoporosity with the volume 0.9, 0.68, and 0.81 cm 3 /g, respectively. In addition, its pore size distribution (PSD) curve ( Figure S1) showed several peaks in the range of 0.5 to 100 nm, further confirming the existence of mesoporous and macroporous structures from the accumulation of carbon nanosheets. The hysteresis loop shapes and the pore size distributions ( Figure S1) of NSC-600 and NSCox-2 were similar to that of NSC, implying the similar textural properties of the carbon supports, so the effects of the textural properties of support on the performance of the corresponding catalysts could be diminish greatly. The surface area of NSC-600 was also similar to that of NSC (Table  1), while the data of NSCox-2 (510 m 2 /g) decreased in certain extent possible due to the destruction of a few flowerlike microspheres during the oxidation. The surface area order of the three carbon samples, NSC-600 ≈ NSC > NSCox-2, was different from the activity order of the corresponding catalysts, indicating that surface area of the carbon support was not the main reason affecting the Pd dispersion and the activity of the Pd-based catalyst on carbon. The N 2 adsorption-desorption isotherms of the three carbon supports are shown in Figure 4. The specific surface area, and pore volume of these supports are listed in Table 1. The isotherms of NSC, NSC-600, and NSC ox-2 samples showed a combination of type I and type IV curves with a type H3 hysteresis loop revealing the existence of the interparticle mesoporosity with the volume 0.9, 0.68, and 0.81 cm 3 /g, respectively. In addition, its pore size distribution (PSD) curve ( Figure S1) showed several peaks in the range of 0.5 to 100 nm, further confirming the existence of mesoporous and macroporous structures from the accumulation of carbon nanosheets. The hysteresis loop shapes and the pore size distributions ( Figure S1) of NSC-600 and NSC ox-2 were similar to that of NSC, implying the similar textural properties of the carbon supports, so the effects of the textural properties of support on the performance of the corresponding catalysts could be diminish greatly. The surface area of NSC-600 was also similar to that of NSC (Table 1), while the data of NSC ox-2 (510 m 2 /g) decreased in certain extent possible due to the destruction of a few flowerlike microspheres during the oxidation. The surface area order of the three carbon samples, NSC-600 ≈ NSC > NSC ox-2 , was different from the activity order of the corresponding catalysts, indicating that surface area of the carbon support was not the main reason affecting the Pd dispersion and the activity of the Pd-based catalyst on carbon.     In order to insight the phase structures of the carbon samples, XRD measurements were performed ( Figure S2). The diffraction patterns of NSC, NSC-600, and NSC ox-2 were similar. The broad peak at 23 • and the weak peak at 44 • , corresponding to the (002) and the (100) diffractions of the graphitic domains, respectively, revealed the similar partial graphitic amorphous structure of the NSC, NSC-600, and NSC ox-2 . The Raman spectra of NSC, NSC-600, and NSC ox-2 clearly displayed the characteristic D bands and G bands at 1340 and 1589 cm −1 ( Figure S3), corresponding to the defects or disordered portions, and the ordered sp 2 carbon in the samples, respectively. The I D /I G ratio is often applied to measure the defects property of carbon materials [43]. It could be seen that the I D /I G ratios of the three carbon samples were comparable.
The changes of surface functional groups after chemical and physical treatments were proved by the elemental analysis data of the samples of NSC, NSC-600 and NSC ox-2 (Table S2). It was found that the hydrogen contents in the three samples were around 2%, and no nitrogen atoms were detected in NSC and NSC-600, but there was a little bit of nitrogen (1.3%) in NSC ox-2 , due to the interaction of nitrate ions or nitrogen oxides with the carbon matrix, which might give a positive effect on the Pd dispersion [25]. The oxygen content in NSC ox-2 was much higher than that of the NSC sample due to the further oxidation of carbon surface by nitric acid, while the oxygen content in NSC-600 was lower than that of NSC, revealing the elimination of a few oxygen-containing groups in the calcination process, particular for the carboxylic group and anhydride group. The wetting angles of the carbon supports were also detected ( Figure S4), which were in order of NSC ox-2 < NSC < NSC-600, showing a significant difference in hydrophilicity of the carbon supports due to the variety of concentrations of the surface oxygen-containing groups. The high hydrophilicity of NSC ox-2 may arise preferentially adsorption and nucleation of the Pd species on its surface and be profitable to improve the Pd dispersion. In addition, TG measurement was conducted to study the decomposition of oxygen-containing groups of NSC, NSC-600, and NSC ox-2 under N 2 atmosphere. The TG data showed that the pyrolysis of the three samples was a continuous process ( Figure S5). The mass losses of NSC ox-2 , NSC, and NSC-600 from 100 • C up to 750 • C were 15.2, 6.2, and 3.5%, respectively. The weight losses of the three carbon supports were only slightly lower comparing to their total oxygen contents (Table S2), revealing that the mass losses could be mainly assigned to the decomposition of the oxygen-containing functional groups.
TPD-technique was a method to characterize precisely the nature of oxygen-containing groups present on carbon material surface, and their amounts and thermal stability through the quantification of CO and CO 2 emissions during temperature increase in inert atmosphere. So, the TPD measurement was performed to further detect the surface structures of the three carbon samples. The deconvolution method (using multiple Gaussian functions) was applied to the CO 2 and CO profiles obtained for the three carbon samples, and the corresponding deconvolutions are shown in Figure 5. Based on the literature results [44], the CO 2 spectra of the NSC and NSC ox-2 samples were decomposed into three peaks as follows. The two peaks below 300 • C could be attributed to two kinds of carboxylic acids, and the peak centered at relative high temperature range (400-550 • C) was assigned to the carboxylic anhydrides. For the NSC-600 sample, only two weak peaks at 733 and 892 • C were detected, which were assigned to the decomposition of thermal stable lactone groups. The CO spectra of the NSC and NSC ox-2 were also decomposed into three peaks, corresponding to the decomposition of carboxylic anhydride (300-600 • C), phenols (510-750 • C) and carbonyl-quinones (600-900 • C), respectively [44]. For NSC-600, only carbonyl-quinones signals were observed. The amounts of oxygen-containing groups calculated from the surface areas of the deconvoluted CO 2 and CO peaks are given in Tables 2 and 3, respectively. It clearly displayed that the 20% HNO 3 oxidation led to significantly increases of the concentrations of oxygen-containing groups, while the calcination treatment at 600 • C resulted in the disappearance of most carboxylic acids, carboxylic anhydrides and phenol groups except the stable lactone and carbonyl-quinone groups.   Table 3. Results of the deconvolution of CO-TPD spectra using a multiple Gaussian function.    Table 3. Results of the deconvolution of CO-TPD spectra using a multiple Gaussian function.

Characterization of the Catalysts
TEM images of Pd/NSC, Pd/NSC-600, Pd/NSC ox-2 as well as the corresponding samples after pre-reduction treatment are given in Figure 6. As Figure 6a,i show, for the Pd/NSC and Pd/NSC ox-2 catalysts, it was hardly to distinguish the most Pd nanoparticles in the low magnified TEM images, representing that most of the Pd species were uniformly dispersed on NSC and NSC ox-2 . On the contrary, for the Pd/NSC-600 sample, a few large Pd particles could be observed obviously (Figure 6e), implying the low Pd dispersion. The statistical particle size distributions revealed that the ratio of Pd particles less than 3 nm on Pd/NSC ox-2 was about 75% (Figure 6k), which was higher than the data of Pd/NSC (65%) and Pd/NSC-600 (55%). The average particle sizes of Pd/NSC, Pd/NSC-600, and Pd/NSC ox-2 were 2.8, 5.0, and 2.0 nm, respectively (Figure 6), further demonstrating the highest Pd dispersion on NSC ox-2 . In addition, the Pd dispersity of Pd/NSC (42%), Pd/NSC-600 (23%), and Pd/NSC ox-2 (58%), was also determined by CO chemisorption method (Table S3), which was in accordance with the TEM data. These results indicated that the chemical oxidative treatment with nitric acid was beneficial to the high dispersion of palladium on carbon due to the introduction of more oxygen-containing groups, which was profitable to enhance the interaction between Pd species and carbon support, but the physical calcination exhibited a negative effect on the metal dispersity due to the decomposition of unstable carboxylic acid and anhydride groups.
Catalysts 2021, 11, x FOR PEER REVIEW 9 of 18 ticles in the low magnified TEM images, representing that most of the Pd species were uniformly dispersed on NSC and NSCox-2. On the contrary, for the Pd/NSC-600 sample, a few large Pd particles could be observed obviously (Figure 6e), implying the low Pd dispersion. The statistical particle size distributions revealed that the ratio of Pd particles less than 3 nm on Pd/NSCox-2 was about 75% (Figure 6k), which was higher than the data of Pd/NSC (65%) and Pd/NSC-600 (55%). The average particle sizes of Pd/NSC, Pd/NSC-600, and Pd/NSCox-2 were 2.8, 5.0, and 2.0 nm, respectively ( Figure 6), further demonstrating the highest Pd dispersion on NSCox-2. In addition, the Pd dispersity of Pd/NSC (42%), Pd/NSC-600 (23%), and Pd/NSCox-2 (58%), was also determined by CO chemisorption method (Table S3), which was in accordance with the TEM data. These results indicated that the chemical oxidative treatment with nitric acid was beneficial to the high dispersion of palladium on carbon due to the introduction of more oxygen-containing groups, which was profitable to enhance the interaction between Pd species and carbon support, but the physical calcination exhibited a negative effect on the metal dispersity due to the decomposition of unstable carboxylic acid and anhydride groups. Walker [45,46] and Derbyshire [47] emphasized that the surface chemical property of carbon had an important influence on metal dispersion. For precisely understanding the effects of various oxygen-containing groups on the Pd dispersions, we checked the possibility of a direct correlation between the amounts of various oxygen-containing groups and the Pd dispersions of Pd/NSC, Pd/NSC-600, and Pd/NSCox-2. To reduce the  Walker [45,46] and Derbyshire [47] emphasized that the surface chemical property of carbon had an important influence on metal dispersion. For precisely understanding the effects of various oxygen-containing groups on the Pd dispersions, we checked the possibility of a direct correlation between the amounts of various oxygen-containing groups and the Pd dispersions of Pd/NSC, Pd/NSC-600, and Pd/NSC ox-2 . To reduce the occasionality of only three samples, the Pd/NSC ox-1 and Pd/NSC ox-3 samples were also checked ( Figures S6 and S7, Tables S4 and S5). It was found that there were obvious correlations between the amounts of carboxyl group and anhydride group, as well as phenol group and the Pd dispersions of the catalysts ( Figure S8), while no correlation was found between the amount of carbonyl-quinone and the Pd dispersions. The results indicated that the concentrations of carboxyl groups and anhydride groups, as well as phenol groups instead of carbonyl-quinone groups on the carbon surface had significant effect on the Pd dispersions and consequently on the activity of the catalysts. However, NSC ox-3 had the highest concentration of oxygen-containing groups, but the Pd dispersity of Pd/NSC ox-3 (45%) was similar to Pd/NSC ox-1 (46%), and much lower than that of Pd/NSC ox-2 (58%), implying that there were other factors that influenced the Pd dispersion. In other words, a proper concentration of oxygen-containing groups for an ideal carbon support was desirable.
The TEM data of Pd/NSC-R, Pd/NSC-600-R, and Pd/NSC ox-2 -R showed that the prereduction treatment led to the growth of Pd particles in a certain extent ( Figure 6, Table S6), but the variations of Pd particle size of the three samples were much different. For Pd/NSC ox-2 , the change of Pd particle size was inconspicuous, and there were still about 70% of Pd particles maintaining below 3 nm after reduction treatment (Figure 6l). On the contrary, for Pd/NSC-R and Pd/NSC-600-R catalysts, the Pd particle size and distribution changed obviously comparing to the fresh ones (Figure 6b,d, or Figure 6f,h). The stability of the three catalysts in a reduction condition was in order of Pd/NSC-600 < Pd/NSC < Pd/NSC ox-2 . In order to further investigate the change of Pd particle size and dispersion in the reduction condition, the recovered Pd/NSC ox-2 , Pd/NSC, and Pd/NSC-600 catalysts were also characterized by TEM ( Figure S9). For the recovered Pd/NSC ox-2 catalyst, highly dispersed Pd particles with an average size of 3.2 nm still could be seen clearly, but obvious agglomerations of Pd particles were observed for the recovered Pd/NSC and Pd/NSC-600 catalysts, and their average sizes of Pd particles increased to about 4.5 and 10.2 nm, respectively (Table S6). The TEM figures of the recovered Pd/NSC ox-2 -R, Pd/NSC-R, and Pd/NSC-600-R samples ( Figure S9) and the Pd/NSC ox-2 , Pd/NSC, and Pd/NSC-600 samples after the cycling experiment ( Figure S10) indicated the further aggregation of Pd particles during the debenzylation reaction (Table S6). It further showed that Pd/NSC ox-2 had a higher stability comparing to the two others. The results indicated that the high concentration of surface oxygen-containing groups of the carbon support was beneficial to improve the stability of the corresponding palladium-carbon catalyst.
It has been known that the Pd catalysts with a smaller average particle size generally result in better catalytic performance [48]. Combining the above activity data, one can deduce that the Pd particle size and dispersion is a main factor influencing the debenzylation activity. Further analysis showed that there was a direct correlation between the Pd particle sizes and the debenzylation activity of the corresponding catalysts ( Figure 7). However, the result does not mean that this relationship is a strictly linear. It has been known that the catalytic activity of the noble metal nanoparticles can be understood by assuming the defective and strained nature of their surface structure. The physical study revealed that there was a greater degree of disorder in the atomic arrangement for smaller nanoparticles [49], so the detailed surface microstructure of the novel metal particles also had significant effects on its catalytic activity, particular for small nanoparticles or clusters. In addition, the influence of the nature of carrier also need pay attention as presented in above.
has been known that the catalytic activity of the noble metal nanoparticles can be understood by assuming the defective and strained nature of their surface structure. The physical study revealed that there was a greater degree of disorder in the atomic arrangement for smaller nanoparticles [49], so the detailed surface microstructure of the novel metal particles also had significant effects on its catalytic activity, particular for small nanoparticles or clusters. In addition, the influence of the nature of carrier also need pay attention as presented in above.

Figure 7.
Correlation between the average Pd particle sizes of the three fresh, pre-reduction treated and recovered catalysts with the product yields. The data were from Tables S1 and S6.
Comparing the XRD patterns of Pd/NSC, Pd/NSC-600 and Pd/NSC ox-2 ( Figure S11), three sharp peaks at 2θ of~40, 46, and 68 • for the Pd/NSC-600 catalyst could be found, which were assigned to the (111), (200), and (220) diffractions of Pd 0 , respectively, demonstrating the present of large Pd 0 particles on it [50]. The corresponding diffraction peaks of Pd/NSC were much weaker than that of Pd/NSC-600. No obvious signal of Pd 0 could be detected, and only two weak and wide peaks at around 34 and 43 • corresponding to the carbon support were observed for Pd/NSC ox-2 , indicating the high Pd dispersion and the existence of PdO species. This might be one more reason why the Pd/NSC ox-2 catalyst exhibited high activity. It had been reported that the unreduced Pd/C showed a much higher activity than the corresponding reduced Pd/C [37,51]. The nitrogen adsorption/desorption isotherms for Pd/NSC, Pd/NSC-600, and Pd/NSC ox-2 are shown in Figure 4. The curves of the three catalysts were similar to their corresponding support, revealing the similar textural property, which was not the key factor influencing the debenzylation activity of the catalysts.
The high catalytic activity of the catalyst is generally related to its reducibility. The TPR profiles of the catalysts as a function of reduction temperature are presented in Figure 8. It is a known fact that bulk Pd species can be reduced by H 2 at low temperature to form β-PdHx species, which shows a negative signal due to the hydrogen spillover [52]. An obviously negative peak was observed at low temperature range of 60 to 70 • C for Pd/NSC, Pd/NSC-600, and Pd/NSC ox-2 . Besides, a distinct positive peak centered at around 140 • C was obtained for Pd/NSC ox-2 , suggesting the reduction of Pd 2+ →Pd 0 and the good reducibility. For Pd/NSC and Pd/NSC-600, the reduction peaks of Pd 2+ were inconspicuous. The peak in the high temperature range (>500 • C) was assigned to the reduction of surface oxygen-functional groups (Table S7). Correlation between the average Pd particle sizes of the three fresh, pre-reduction treated and recovered catalysts with the product yields. The data were from Tables S1 and S6.
Comparing the XRD patterns of Pd/NSC, Pd/NSC-600 and Pd/NSCox-2 ( Figure S11), three sharp peaks at 2θ of ~40, 46, and 68° for the Pd/NSC-600 catalyst could be found, which were assigned to the (111), (200), and (220) diffractions of Pd 0 , respectively, demonstrating the present of large Pd 0 particles on it [50]. The corresponding diffraction peaks of Pd/NSC were much weaker than that of Pd/NSC-600. No obvious signal of Pd 0 could be detected, and only two weak and wide peaks at around 34 and 43° corresponding to the carbon support were observed for Pd/NSCox-2, indicating the high Pd dispersion and the existence of PdO species. This might be one more reason why the Pd/NSCox-2 catalyst exhibited high activity. It had been reported that the unreduced Pd/C showed a much higher activity than the corresponding reduced Pd/C [37,51]. The nitrogen adsorption/desorption isotherms for Pd/NSC, Pd/NSC-600, and Pd/NSCox-2 are shown in Figure 4. The curves of the three catalysts were similar to their corresponding support, revealing the similar textural property, which was not the key factor influencing the debenzylation activity of the catalysts.
The high catalytic activity of the catalyst is generally related to its reducibility. The TPR profiles of the catalysts as a function of reduction temperature are presented in Figure 8. It is a known fact that bulk Pd species can be reduced by H2 at low temperature to form β-PdHx species, which shows a negative signal due to the hydrogen spillover [52]. An obviously negative peak was observed at low temperature range of 60 to 70 °C for Pd/NSC, Pd/NSC-600, and Pd/NSCox-2. Besides, a distinct positive peak centered at around 140 °C was obtained for Pd/NSCox-2, suggesting the reduction of Pd 2+ →Pd 0 and the good reducibility. For Pd/NSC and Pd/NSC-600, the reduction peaks of Pd 2+ were inconspicuous. The peak in the high temperature range (>500 °C ) was assigned to the reduction of surface oxygen-functional groups (Table S7). The XPS analysis has often been used to determine the surface chemical composition and the bonding environment of various carbon materials. So XPS studies of Pd/NSC, Pd/NSC-600, and Pd/NSCox-2 were performed ( Figure S12, Table S8). The C 1s spectra of the three catalysts are showed in Figure 9a. The signals were deconvoluted The XPS analysis has often been used to determine the surface chemical composition and the bonding environment of various carbon materials. So XPS studies of Pd/NSC, Pd/NSC-600, and Pd/NSC ox-2 were performed ( Figure S12, Table S8). The C 1s spectra of the three catalysts are showed in Figure 9a. The signals were deconvoluted into four peaks referred to the aliphatic/aromatic carbon group (CHx, C-C/C=C) at 284.7 eV, carbons coupled with hydroxyl groups (C-OH) at 286.0 eV, carbonyl groups (C=O) at 287.0 eV and carboxylic groups, esters or lactones (-COOR) (288.8 eV) [53], respectively, demonstrating the existence of various oxygen-containing groups. The O 1s signal of the three catalysts could be fitted to four peaks centered at 531.0 eV (oxygen in quinone C=O), 532.9 eV (oxygen in OH, ether, ester, anhydride, and carboxyl), 534.5 eV (ether oxygen in ester and anhydride, as well as hydroxyl in carbonyl group (COOH)) and 536.5 eV (adsorbed water) [54]. The surface atomic concentrations of C and O within different functional groups were determined using the Peak-Fit program ( Table 4). It can be seen that the concentrations of C and O coupled with the carboxylic groups were both in order of Pd/NSC-600 < Pd/NSC < Pd/NSC ox-2 , which were consistent with the order of their Pd dispersion, demonstrating the relationship between the concentration of carboxylic group and the Pd dispersion. This was consistent with the above analysis of TPD data. The total surface oxygen contents and the O/C atomic ratios calculated by XPS data were similar to the corresponding data determined by elemental analysis (Table S9). The XPS spectra for the doublet Pd 3d5/2 and 3d3/2 of Pd/NSC, Pd/NSC-600 and Pd/NSC ox-2 are shown in Figure 9c. The Pd (3d) signal could be fitted to two pairs of double peaks. The Pd 3d3/2 peak at 340.1 eV and Pd 3d5/2 peak at 335.9 eV were assigned to Pd 0 species, and the Pd 3d3/2 peak at 342.8 eV and Pd 3d5/2 peak at 337.8 eV were assigned to Pd 2+ species, which were similar to the literature data [55]. The ratios of Pd 2+ /Pd 0 for the three catalysts were summarized in Table S8. The Pd/NSC ox-2 sample presented the highest content of Pd 2+ species than the other two catalysts, possibly due to the interaction of the Pd atoms with more surface oxygen-containing groups of carbon.
Catalysts 2021, 11, x FOR PEER REVIEW 12 of 18 536.5 eV (adsorbed water) [54]. The surface atomic concentrations of C and O within different functional groups were determined using the Peak-Fit program ( Table 4). It can be seen that the concentrations of C and O coupled with the carboxylic groups were both in order of Pd/NSC-600 < Pd/NSC < Pd/NSCox-2, which were consistent with the order of their Pd dispersion, demonstrating the relationship between the concentration of carboxylic group and the Pd dispersion. This was consistent with the above analysis of TPD data. The total surface oxygen contents and the O/C atomic ratios calculated by XPS data were similar to the corresponding data determined by elemental analysis (Table S9). The XPS spectra for the doublet Pd 3d5/2 and 3d3/2 of Pd/NSC, Pd/NSC-600 and Pd/NSCox-2 are shown in Figure 9c. The Pd (3d) signal could be fitted to two pairs of double peaks. The Pd 3d3/2 peak at 340.1 eV and Pd 3d5/2 peak at 335.9 eV were assigned to Pd 0 species, and the Pd 3d3/2 peak at 342.8 eV and Pd 3d5/2 peak at 337.8 eV were assigned to Pd 2+ species, which were similar to the literature data [55]. The ratios of Pd 2+ /Pd 0 for the three catalysts were summarized in Table S8. The Pd/NSCox-2 sample presented the highest content of Pd 2+ species than the other two catalysts, possibly due to the interaction of the Pd atoms with more surface oxygen-containing groups of carbon.

Preparation of the Carbon Support and the Reaction Substrate
TADB was synthesized in our own laboratory. Briefly, 100 mL of DMF, 50 mL of Ac 2 O, 0.9 mL of PhBr, 50 g of HBIW, which was prepared using the Nielsen's method [38], and 1.68 g of 9% Pd(OH) 2 /C catalysts were placed into the reactor. Then, the system was exchanged with pure hydrogen. The reaction mixture was stirred for 15 h at 25 • C. The sample was received by filtering and washing with anhydrous ethanol. The flowerlike nanosheet carbon material (NSC) was prepared according to our previous work [41]. Briefly, glucose (1.8 g), zinc acetate (8.4 g) and SiO 2 ( Figure S13) (Zn/Si mass ratio = 4:1) were added into 25 mL of deionized water. Then, the mixture was put Teflon-lined stainless steel autoclave and heated at 180 • C for 24 h. After cooled to room temperature, the solid product was recovered by filtration, washed with deionized water and dried in an oven at 120 • C for 4 h. Subsequently, the above solids were calcinated under N 2 flow at 750 • C for 2 h. The obtained black solids were added to 20 wt% HF solution under stirred for about 6 h and washed with water. Lastly, the porous carbon materials were collected and dried at 100 • C.

Treatment of the Carbon Supports
The NSC sample was initially dispersed in nitric acid solution. Then, the mixture was heated at 50 • C for 2 h with magnetic agitation. After that, the NSC mixture was filtered and washed with abundant deionized water and dried in an oven at 100 • C. The received sample was denoted as NSC ox-1 , NSC ox-2 , and NSC ox-3 , respectively, when 10, 20, or 30 wt% nitric acid was used. NSC was directly calcinated under N 2 flow of 100 mL/min at 600 • C for 2 h with the heating rate of 10 • C/min in a quartz tube furnace. The collected sample was denoted as NSC-600.

Catalyst Preparation
The 5 wt% Pd(OH) 2 /C catalysts were prepared by deposition-precipitation method. Briefly, 4 g of as-prepared NSC support was dispersed in deionized water (40 mL) under magnetic stirring at room temperature. Then the PdCl 2 (0.34 g) solution in diluted HCl was added. After stirred for another 1 h, the NaOH solution (5 wt%) was dropwise added into the mixture until the pH value of the suspension up to 10. The suspension was stirred for an additional 6 h. Then the mixture was filtered and washed with plenty of deionized H 2 O. The filter cake was dried in oven to give Pd/NSC catalyst. Pd/NSC-600, Pd/NSC ox-1 , Pd/NSC ox-2 , and Pd/NSC ox-3 catalysts were prepared in the same manner using NSC-600, NSC ox-1 NSC ox-2 , and NSC ox-3 as the support instead of NSC, respectively. In addition, the above Pd/NSC, Pd/NSC-600, and Pd/NSC ox-2 catalysts were pre-reduced under 10% H 2 /Ar flow at 200 • C for 2 h, which were denoted as Pd/NSC-R, Pd/NSC-600-R, and Pd/NSC ox-2 -R, respectively.

Catalytic Activity Tests
The activities of the catalysts were investigated in the hydrogenolytic debenzylation of TADB. Firstly, 30 g of TADB, 1.26 g of catalyst, 120 mL of CH 3 COOH and 30 mL of H 2 O were put into a vessel, which was immediately purged three times with hydrogen. The reaction was gone under an atmospheric pressure of H 2 at 45 • C for 10 h with vigorously stirred. After that, the filtrate was concentrated under vacuum, and the resulted white solid was washed with ethanol. Finally, the filter cake was dried in vacuum oven at 65 • C for 6 h.

Carbon Supports and Catalysts Characterization
Physical adsorption was measured at 77 K on a Micrometrics, ASAP 2460 instrument (Micrometrics, Norcross, GA, USA). Prior to the measurements, the sample (about 100 mg) was degassed under vacuum at 150 • C for 6 h. The X-ray diffraction (XRD) patterns were recorded with an X-ray diffractometer (Bruker/AXS D8 Advance, Ettlingen, Germany) operated at 40 kV and 30 mA using Cu Kα radiation (λ = 0.154 nm). The samples were scanned from 10 • to 80 • (2θ) with a step size of 0.02 • and a scanning rate of 3.5 • /min. Raman spectra was collected at room temperature on a Renishaw 2000 spectrograph with spectral resolution of 2 cm −1 . A 532 nm laser was used as the excitation source and the power output was about 40 mW. The water contact angles of the samples were measured using theta optical tensiometer (Dataphysics-TP50, Filderstadt, Germany) and electro-optics with a closed-circuit television camera connected to a computer (Attension Theta software). The sample was placed on a sample stage and a droplet of distilled water (about 2 µL) was deposited on the surface of the sample. Each sample was measured thrice, and the average value was recorded. The surface chemical composition of the samples was analyzed by XPS (Thermo Fisher, ESCALAB 250 Xi, Waltham, MA, USA), using monochromatic Al Kradiation (1486.6 eV) operating at an accelerating power of 15 kW. Transmission electron microscopy (TEM) images were recorded over a JEM 2100 (JEOL, Tokyo, Japan) microscope, operated at an acceleration voltage of 200 KV and electric current of 20 mA. Scanning electron microscopy (SEM) images were measured using a JEOL JSM-35C instrument (JEOL, Tokyo, Japan). Before the test, the samples were sprayed with gold. Temperature-programmed desorption (TPD) was performed at a U-shaped tubular micro-reactor with 50 mg sample that was heated (10 • C/min) from room temperature to 950 • C under a helium stream (40 cm 3 /min) using Quantachrome ChemBet 3000 (Boynton Beach, FL, USA). The type and amounts of oxygen-containing groups were determined by decomposition temperatures and the amounts of the decomposition products (CO and CO 2 ) measured by on-line mass spectrometry. The temperature programmed reduction with H 2 experiment (H 2 -TPR) was performed on a chemisorption analyzer (Micromeritics, AutoChem II 2920, Norcross, GA, USA). Firstly, 0.05 g of the catalyst was pretreated under O 2 at 200 • C for 30 min. After cooled to room temperature, the sample was purged by He until the baseline remained unchanged. The TPR profile was obtained by heating the sample from 30 to 750 • C in a flow of 10% H 2 /He (30 mL/min) with a heating rate of 10 • C/min. The H 2 consumption was monitored by a thermal conductivity detector. Thermogravimetric analysis (TG) was carried out on a thermal analyzer TA Q600 (New Castle, DE, USA) with a heating rate of 5 • C/min in a nitrogen flow (100 mL/min). The products were analyzed using an Agilent Technologies 6100 (Waldbronn, Germany) high performance liquid chromatography (HPLC). The sample was prepared by dissolving the product (4 mg) in acetonitrile (5 mL), followed by filtration to remove particulates. The injection volume was 20 uL, and the mobile phase was 75% methanol in water at a 0.8 mL/min flow rate; a UV/vis detector operating at 230 nm was employed.

Conclusions
In summary, the effects of surface chemical properties of the flowerlike nanosheet carbon materials (NSC) on the Pd dispersion and the activities and stabilities of the corresponding Pd(OH) 2 /C catalysts had been investigated. It was found that the Pd particle size and dispersion was a main factor influencing the debenzylation activity. The proper concentration of carboxyl group and anhydride, as well as phenol groups on the carbon surface could contribute to improve the hydrophilicity of the carbon support and enhanced the interaction between Pd species and the support, which consequently increased the Pd dispersion, activity and stability of the corresponding catalyst for the hydrogenolytic debenzylation reaction. This work may offer an efficient strategy to improve the activity and stability of the Pd-based catalyst on carbon.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/catal11040441/s1: Figure S1: Pore size distributions (PSD) of the three carbon supports and the corresponding catalysts; Figure S2: XRD patterns of the three carbon supports; Figure S3: Raman spectra of the carbon supports; Figure S4: Contact angles of the carbon supports; Figure S5: TG results for the carbon supports, which were measured under a nitrogen atmosphere with a heating rate of 5 • C/min; Figure S6: Results of deconvolutions on TPD profiles using a multiple Gaussian function; Figure S7: TEM images of the Pd/NSC ox-1 (a), and Pd/NSC ox-3 (c) samples, and the corresponding histograms of particle size distribution; Figure S8: Correlations between the Pd dispersion (determined by CO chemisorption method) and: (a) the amount of carboxyl group, and (b) the amount of anhydride group determined from the data of CO 2 -TPD spectra; (c) the amount of anhydride group, (d) the amount of phenol group, and (e) the amount of carbonylquinone group determined from the data of CO-TPD spectra; Figure S9: TEM images and the corresponding histograms of particle size distribution of Pd/NSC (a,b), Pd/NSC-R (c,d), Pd/NSC-600 (e,f), Pd/NSC-600-R (g,h), Pd/NSC ox-2 (i,j), and Pd/NSC ox-2 -R (k,l) samples after reaction of the hydrogenolysis debenzylation of TADB; Figure S10: TEM images and the corresponding histograms of particle size distribution of the Pd/NSC (a,b), Pd/NSC ox-2 (c,d), and Pd/NSC-600 (e,f) catalysts after cycling experiment; Figure S11: X-ray diffraction patterns of the various catalysts; Figure S12: XPS survey spectrum of the Pd/NSC-600, Pd/NSC, and Pd/NSC ox-2 catalysts; Figure S13: Nitrogen adsorption/desorption isotherm (a) and pore size distribution (b) of the SiO2 sample at 77K. Table S1: Catalytic performance of the catalysts in the hydrogenolysis reaction of TADB; Table S2: Total elementary composition and TG weight loss; Table S3: Dispersion of palladium on different carbon supports; Table S4: Results of the deconvolution of CO 2 -TPD spectra using a multiple Gaussian function; Table S5: Results of the deconvolution of CO-TPD spectra using a multiple Gaussian function; Table S6: The particle size of Pd nanoparticles calculated by TEM; Table S7: H 2 consumptions of the various catalysts determined from the H 2 -TPR curves; Table S8: XPS binding energies (eV), surface atomic concentrations and atomic ratios of the Pd/NSC, Pd/NSC-600, and Pd/NSC ox-2 catalysts; Table S9: Surface atomic concentrations and atomic ratios of the Pd/NSC-600, Pd/NSC, Pd/NSC ox-1 , Pd/NSC ox-2 , and Pd/NSC ox-3 catalysts.