Highly Dispersed Co Nanoparticles Prepared by an Improved Method for Plasma-Driven NH3 Decomposition to Produce H2

1 College of Environmental Science and Engineering, Dalian Maritime University, Dalian 116026, China; duxiaomin1202@163.com (X.D.); binzhu@dlmu.edu.cn (B.Z.) 2 State Key Laboratory of Fine Chemicals, School of Chemical Engineering, Dalian University of Technology, Dalian 116024, China; yiyanhui@dlut.edu.cn (Y.Y.); hongchenguo@163.com (H.G.) * Correspondence: liwang@dlmu.edu.cn (L.W.); ntp@dlmu.edu.cn (Y.Z.); Tel.: +86-411-84724357 (L.W.)


Introduction
NH 3 decomposition has been considered to be an attractive route to supply CO x -free H 2 for proton exchange membrane fuel cell (PEMFC) vehicles [1][2][3]. Until now, the noble metal Ru, due to its high turnover frequency (TOF), is still the most active component for NH 3 decomposition, and the formation rate of H 2 reached as high as 4.0 mol/(h·g cat ) −1 using K-Ru/MgO-CNTs catalyst with complete conversion of ammonia at 450 • C, but the scarcity and high price of Ru limits its use on a large scale [4][5][6]. Whereas, cheap metal catalysts show low activity towards NH 3 decomposition due to the strong adsorption of N atoms onto the surface of cheap metal catalysts [1,[7][8][9][10]. As far as we know, the highest formation rate of H 2 was 2.0 mol/(h·g cat ) −1 using CeO 2 -doped Ni/Al 2 O 3 catalyst with 98.3% ammonia conversion at 550 • C [7]. Recently, the combination of non-thermal plasma with cheap metal catalyst displayed a powerful ability in enhancing NH 3 decomposition [11][12][13]; 99.9% conversion of NH 3 was achieved in combination mode, but only 7.4% and 7.8% was obtained for Fe-based catalyst alone and plasma alone, respectively, which experienced an unexpected strong

Characterization
The physicochemical properties of as-prepared Co catalysts were examined using various characterization techniques, including X-ray diffraction (XRD), X-ray fluorescence (XRF), transmission electron microcopy (TEM), H 2 temperature-programmed reduction (H 2 -TPR), and NH 3 temperature-programmed desorption (NH 3 -TPD). In this study, the fumed SiO 2 used as a support for Co catalyst was an amorphous material with a Brunauer-Emmett-Teller (BET) surface area of 297.8 m 2 ·g −1 . The theoretical Co loading was designed to be 30 wt %, but the actual Co loading through XRF analysis was 27.7 wt % and 27.4 wt % for the improved prepared catalyst and the conventional prepared catalyst, respectively (see Tables S1 and S2 in Supporting Information). Figure 1 shows the XRD patterns of as-prepared fumed SiO 2 -supported Co catalysts using conventional and improved preparation methods, respectively. Besides, pure fumed SiO 2 was analyzed as a reference in Figure 1 (a). Clearly, the same diffraction peaks were observed at 2θ of 31.1, 36.7, 44.6, 59.2, and 65.2 as shown in Figure 1 [37,38]. Namely, the difference in preparation approach did not influence the phase structure of Co catalysts, and they both finally existed in the form of Co 3 O 4 over fumed SiO 2 support. However, by contrast, the intensity of diffraction peaks of Co catalyst prepared with improved method was weaker than that with conventional preparation method, suggesting that the average particle size of the former is smaller than that of the latter according to the Debye-Scherrer formula [39]; this observation is also supported by the results of TEM as follows.

Characterization
The physicochemical properties of as-prepared Co catalysts were examined using various characterization techniques, including X-ray diffraction (XRD), X-ray fluorescence (XRF), transmission electron microcopy (TEM), H2 temperature-programmed reduction (H2-TPR), and NH3 temperature-programmed desorption (NH3-TPD). In this study, the fumed SiO2 used as a support for Co catalyst was an amorphous material with a Brunauer-Emmett-Teller (BET) surface area of 297.8 m 2 ·g −1 . The theoretical Co loading was designed to be 30 wt %, but the actual Co loading through XRF analysis was 27.7 wt % and 27.4 wt % for the improved prepared catalyst and the conventional prepared catalyst, respectively (see Table S1 and S2 in Supporting Information). Figure 1 shows the XRD patterns of as-prepared fumed SiO2-supported Co catalysts using conventional and improved preparation methods, respectively. Besides, pure fumed SiO2 was analyzed as a reference in Figure 1 (a). Clearly, the same diffraction peaks were observed at 2θ of 31.1, 36.7, 44.6, 59.2, and 65.2 as shown in Figure 1 (b) and (c), which matched well with the characteristic structure of Co3O4 (JCPDS file No: 43-1003), and those diffraction peaks represented the (220), (311), (400), (511), and (440) planes of Co3O4, respectively [37,38]. Namely, the difference in preparation approach did not influence the phase structure of Co catalysts, and they both finally existed in the form of Co3O4 over fumed SiO2 support. However, by contrast, the intensity of diffraction peaks of Co catalyst prepared with improved method was weaker than that with conventional preparation method, suggesting that the average particle size of the former is smaller than that of the latter according to the Debye-Scherrer formula [39]; this observation is also supported by the results of TEM as follows. TEM images of as-prepared Co catalyst supported on fumed SiO2 using different approaches were shown in Figure 2. Clearly, a very poor dispersion of Co catalyst was observed on fumed SiO2 using the conventional preparation method, and the particle size of Co was much larger than 5 nm; some particle sizes were around 50 nm, as shown in Figure 2a,b. However, the use of combining vacuum-freeze drying and plasma calcination techniques in the process of catalyst preparation enabled the Co particles to disperse highly and homogeneously onto the fumed SiO2 support, and the average Co particle size was less than 5 nm, mostly around 2-3 nm in Figure 2c,d. Actually, it is difficult to obtain such smaller nanoparticles with a high metal loading of about 27 wt % using the conventional preparation method. TEM images of as-prepared Co catalyst supported on fumed SiO 2 using different approaches were shown in Figure 2. Clearly, a very poor dispersion of Co catalyst was observed on fumed SiO 2 using the conventional preparation method, and the particle size of Co was much larger than 5 nm; some particle sizes were around 50 nm, as shown in Figure 2a,b. However, the use of combining vacuum-freeze drying and plasma calcination techniques in the process of catalyst preparation enabled the Co particles to disperse highly and homogeneously onto the fumed SiO 2 support, and the average Co particle size was less than 5 nm, mostly around 2-3 nm in Figure 2c,d. Actually, it is difficult to Using NH3 as probe molecule, the influence of preparation approach on the chemical properties of catalyst was evaluated through NH3-TPD, as displayed in Figure 3. Clearly, two major desorption peaks were observed, one at the low temperatures of 150-220 °C corresponded to the weak adsorption of NH3 on the catalyst, and the other at the high temperatures of 220-350 °C was attributed to the strong adsorption of NH3. It is worth noting that the desorption amount of NH3 over Co catalyst prepared with the improved method was much higher than that with the conventional preparation method, revealing that the improved method leads to an increase in the number of active sites for NH3 adsorption; this finding can be ascribed to the high dispersion of Co nanoparticles, as evidenced by the results of TEM in Figure 2. In addition, the desorption temperature of adsorbed NH3 on the catalyst with improved preparation method shifted towards higher temperature, reflecting that the binding ability of NH3 with the catalyst was stronger than that with the catalyst prepared using conventional preparation method. This inferred that the acidity of catalyst was strengthened by the improved preparation method as well and, more importantly, the increase in active site number and acid strength both facilitated the adsorption of NH3 on the catalyst, finally promoting the dissociation of NH3 on the catalyst. Using NH 3 as probe molecule, the influence of preparation approach on the chemical properties of catalyst was evaluated through NH 3 -TPD, as displayed in Figure 3. Clearly, two major desorption peaks were observed, one at the low temperatures of 150-220 • C corresponded to the weak adsorption of NH 3 on the catalyst, and the other at the high temperatures of 220-350 • C was attributed to the strong adsorption of NH 3 . It is worth noting that the desorption amount of NH 3 over Co catalyst prepared with the improved method was much higher than that with the conventional preparation method, revealing that the improved method leads to an increase in the number of active sites for NH 3 adsorption; this finding can be ascribed to the high dispersion of Co nanoparticles, as evidenced by the results of TEM in Figure 2. In addition, the desorption temperature of adsorbed NH 3 on the catalyst with improved preparation method shifted towards higher temperature, reflecting that the binding ability of NH 3 with the catalyst was stronger than that with the catalyst prepared using conventional preparation method. This inferred that the acidity of catalyst was strengthened by the improved preparation method as well and, more importantly, the increase in active site number and acid strength both facilitated the adsorption of NH 3 on the catalyst, finally promoting the dissociation of NH 3 on the catalyst.  H2-TPR technique was used to evaluate the reduction behavior of Co3O4/fumed SiO2 prepared with different methods, and the resulting profiles are displayed in Figure 4. Clearly, the reduction of Co3O4 on fumed SiO2 support occurred in the temperature range of 275-550 °C. Two groups of reduction peaks were observed, i.e., the low temperature reduction peaks (α) consisted of α1 and α2 in the range of 275-400 °C, and the high temperature reduction peaks (β) with a consecutive-broad peak consisted of β1 and β2 in the range of 370-550 °C. More importantly, by contrast, the reduction temperature of catalyst with improved preparation method shifted towards higher temperature, representing that the improved method strengthened the interaction of Co with fumed SiO2 support. This difference in metal-support interaction can be explained by the difference in particle sizes of Co catalyst prepared by different methods (Figure 2). Actually, the reduction process of as-prepared catalyst was very complicated, since these peaks obtained were heavily overlapped. Therefore, the analysis of each peak area using peak fit function (Gaussian) of Origin software was employed to understand the H2-TPR profiles obtained (see Figure S1 in Supporting Information), the area ratio of β1/β2 was found to be 1/3, which is quantitatively consistent with the theoretical value (1/3) of area ratio of Co3O4 reduction peaks [40,41]. This indicates that β1 and β2 corresponded to the two-step reduction of Co 3+ → Co 2+ → Co 0 of Co3O4, as do α1 and α2 based on 5/16 (≈ 1/3) area ratio of α1/α2. Besides, the result of XRD in Figure 1 also supported the assignment of α and β to Co3O4. According to the reduction temperature of Co3O4, the low temperature reduction peaks (α1 and α2) could be due to the reduction of bulk Co3O4, whereas the high temperature reduction peaks (β1 and β2) were attributed to the reduction of Co3O4 that interacted with fumed SiO2 [42,43].
Interestingly, the above results reveal that the application of vacuum-freeze drying and plasma calcination techniques in the preparation process of catalyst not only results in highly dispersed metal nanoparticles along with the increase of active site number, but also strengthens the acidity of catalyst and the metal-support interaction. Thus, it is feasible and crucial to manipulate the properties of catalysts through exploiting novel preparation techniques. H 2 -TPR technique was used to evaluate the reduction behavior of Co 3 O 4 /fumed SiO 2 prepared with different methods, and the resulting profiles are displayed in Figure 4. Clearly, the reduction of Co 3 O 4 on fumed SiO 2 support occurred in the temperature range of 275-550 • C. Two groups of reduction peaks were observed, i.e., the low temperature reduction peaks (α) consisted of α 1 and α 2 in the range of 275-400 • C, and the high temperature reduction peaks (β) with a consecutive-broad peak consisted of β 1 and β 2 in the range of 370-550 • C. More importantly, by contrast, the reduction temperature of catalyst with improved preparation method shifted towards higher temperature, representing that the improved method strengthened the interaction of Co with fumed SiO 2 support. This difference in metal-support interaction can be explained by the difference in particle sizes of Co catalyst prepared by different methods (Figure 2). Actually, the reduction process of as-prepared catalyst was very complicated, since these peaks obtained were heavily overlapped. Therefore, the analysis of each peak area using peak fit function (Gaussian) of Origin software was employed to understand the H 2 -TPR profiles obtained (see Figure S1 in Supporting Information), the area ratio of β 1 /β 2 was found to be 1/3, which is quantitatively consistent with the theoretical value (1/3) of area ratio of Co 3 O 4 reduction peaks [40,41]. This indicates that β 1 and β 2 corresponded to the two-step reduction of Co 3+ → Co 2+ → Co 0 of Co 3 O 4 , as do α 1 and α 2 based on 5/16 (≈ 1/3) area ratio of α 1 /α 2 . Besides, the result of XRD in Figure 1 also supported the assignment of α and β to Co 3 O 4 . According to the reduction temperature of Co 3 O 4 , the low temperature reduction peaks (α 1 and α 2 ) could be due to the reduction of bulk Co 3 O 4 , whereas the high temperature reduction peaks (β 1 and β 2 ) were attributed to the reduction of Co 3 O 4 that interacted with fumed SiO 2 [42,43].
Interestingly, the above results reveal that the application of vacuum-freeze drying and plasma calcination techniques in the preparation process of catalyst not only results in highly dispersed metal nanoparticles along with the increase of active site number, but also strengthens the acidity of catalyst and the metal-support interaction. Thus, it is feasible and crucial to manipulate the properties of catalysts through exploiting novel preparation techniques.  . H2-TPR profiles of as-prepared Co/fumed SiO2 catalysts using different approaches.

Performance of Prepared Catalyst in Plasma-Catalytic NH3 Decomposition
Our previous studies showed that Co-based catalyst exhibited the best activity towards NH3 decomposition to H2 in the presence of DBD plasma [12]. Here, the influence of catalyst preparation method on the performance of plasma-catalytic NH3 decomposition was investigated, as shown in Figure 5. Compared to the conventional preparation method, Co/fumed SiO2 catalyst prepared with the improved method greatly promoted the reaction performance, and the conversion of NH3 increased from 25.8 to 72.7% at the reaction temperature of 400 °C in Figure 5a, increased by a factor of almost 3 and, correspondingly, the energy efficiency of H2 formation increased from 2.3 to 5.7 mol(kW·h) −1 in Figure 5b. In addition, changing the reaction temperature from 300 °C to 450 °C through increasing DBD energy input resulted in a significant increase of NH3 conversion by 80.8% (from 16.1 to 96.9%) in the case of catalyst prepared by the improved method whereas, at the same conditions, the NH3 conversion only increased by 47.3% (from 4% to 51.3%) over catalyst using the conventional preparation method. Note that the reaction temperature required for complete conversion of NH3 in the case of using improved preparation method shifted towards lower temperature, at least 50 °C lower in comparison with that using conventional preparation method in Figure 5a. Figure 5. Plasma-catalytic NH3 decomposition over Co/fumed SiO2 catalyst with different preparation methods: (a) the conversion of NH3; (b) the energy efficiency of H2 generation (NH3 feed rate 40 mL/min −1 , supported catalyst 0.88 g, discharge gap 3 mm, discharge frequency 12 kHz; The reaction temperature originated from electric heat released by discharge, and was determined using an IR camera and thermocouple tightly attached to the outer wall of the reactor [12]).

Performance of Prepared Catalyst in Plasma-Catalytic NH 3 Decomposition
Our previous studies showed that Co-based catalyst exhibited the best activity towards NH 3 decomposition to H 2 in the presence of DBD plasma [12]. Here, the influence of catalyst preparation method on the performance of plasma-catalytic NH 3 decomposition was investigated, as shown in Figure 5. Compared to the conventional preparation method, Co/fumed SiO 2 catalyst prepared with the improved method greatly promoted the reaction performance, and the conversion of NH 3 increased from 25.8 to 72.7% at the reaction temperature of 400 • C in Figure 5a, increased by a factor of almost 3 and, correspondingly, the energy efficiency of H 2 formation increased from 2.3 to 5.7 mol(kW·h) −1 in Figure 5b. In addition, changing the reaction temperature from 300 • C to 450 • C through increasing DBD energy input resulted in a significant increase of NH 3 conversion by 80.8% (from 16.1 to 96.9%) in the case of catalyst prepared by the improved method whereas, at the same conditions, the NH 3 conversion only increased by 47.3% (from 4% to 51.3%) over catalyst using the conventional preparation method. Note that the reaction temperature required for complete conversion of NH 3 in the case of using improved preparation method shifted towards lower temperature, at least 50 • C lower in comparison with that using conventional preparation method in Figure 5a.

Performance of Prepared Catalyst in Plasma-Catalytic NH3 Decomposition
Our previous studies showed that Co-based catalyst exhibited the best activity towards NH3 decomposition to H2 in the presence of DBD plasma [12]. Here, the influence of catalyst preparation method on the performance of plasma-catalytic NH3 decomposition was investigated, as shown in Figure 5. Compared to the conventional preparation method, Co/fumed SiO2 catalyst prepared with the improved method greatly promoted the reaction performance, and the conversion of NH3 increased from 25.8 to 72.7% at the reaction temperature of 400 °C in Figure 5a, increased by a factor of almost 3 and, correspondingly, the energy efficiency of H2 formation increased from 2.3 to 5.7 mol(kW·h) −1 in Figure 5b. In addition, changing the reaction temperature from 300 °C to 450 °C through increasing DBD energy input resulted in a significant increase of NH3 conversion by 80.8% (from 16.1 to 96.9%) in the case of catalyst prepared by the improved method whereas, at the same conditions, the NH3 conversion only increased by 47.3% (from 4% to 51.3%) over catalyst using the conventional preparation method. Note that the reaction temperature required for complete conversion of NH3 in the case of using improved preparation method shifted towards lower temperature, at least 50 °C lower in comparison with that using conventional preparation method in Figure 5a. Figure 5. Plasma-catalytic NH3 decomposition over Co/fumed SiO2 catalyst with different preparation methods: (a) the conversion of NH3; (b) the energy efficiency of H2 generation (NH3 feed rate 40 mL/min −1 , supported catalyst 0.88 g, discharge gap 3 mm, discharge frequency 12 kHz; The reaction temperature originated from electric heat released by discharge, and was determined using an IR camera and thermocouple tightly attached to the outer wall of the reactor [12]). Figure 5. Plasma-catalytic NH 3 decomposition over Co/fumed SiO 2 catalyst with different preparation methods: (a) the conversion of NH 3 ; (b) the energy efficiency of H 2 generation (NH 3 feed rate 40 mL/min −1 , supported catalyst 0.88 g, discharge gap 3 mm, discharge frequency 12 kHz; The reaction temperature originated from electric heat released by discharge, and was determined using an IR camera and thermocouple tightly attached to the outer wall of the reactor [12]).
Combining the results of characterizations in Figure 1 to 4, the improved preparation method did not affect the phase composition of catalyst (Figure 1), but significantly increased the dispersion of catalyst with a narrow particle size of 2-3 nm (Figure 2), which actually creates much more active sites for NH 3 decomposition, enhancing the specific reactivity of catalyst, and this is also directly evidenced by the result of NH 3 -probe experiments presented in Figure 3. Notably, the adsorption amount of NH 3 over the catalyst with improved preparation method is much larger than that with conventional preparation method (Figure 3), this directly points to the fact that enhancing the adsorption step of NH 3 decomposition is one of the reasons for the high activity of catalyst with improved preparation method. Recently, CoPt/TiO 2 with Co particle size of~1 nm displayed a much higher Fischer-Tropsch reaction rate, which was also found to be due to increasing the amount of active site caused by using plasma-assisted preparation [44]. More importantly, in this study, the improved preparation method increased the acid strength of catalyst as well (Figure 3), as demonstrated by the increase in adsorption strength of NH 3 over catalyst, which can promote the dissociation step of NH 3 ; this is another crucial reason that explains the high activity of catalyst with the improved preparation method. Besides, the improved preparation method strengthened the interaction of Co with fumed SiO 2 (Figure 4), indicating the difference in electronic structure of catalyst with different preparation methods, and this could influence the activity of catalyst as well.
In addition, using Co/fumed SiO 2 catalyst prepared by the improved method, the influence of the combining mode of plasma and catalyst was investigated on the performance of plasma-catalytic NH 3 decomposition, as shown in Scheme 1 and Figure 6. About 3 g Co/fumed SiO 2 was packed in the reactor with a packing volume of about 3.1 mL, and the combining mode of plasma and catalyst changed through changing discharge volume "V", but the packed catalyst was fixed. Namely, changing "V" from 3.3 to 0.4 mL enabled the catalyst to be partly packed in the field of plasma, as shown in Scheme 1. Combining the results of characterizations in Figure 1 to 4, the improved preparation method did not affect the phase composition of catalyst (Figure 1), but significantly increased the dispersion of catalyst with a narrow particle size of 2-3 nm (Figure 2), which actually creates much more active sites for NH3 decomposition, enhancing the specific reactivity of catalyst, and this is also directly evidenced by the result of NH3-probe experiments presented in Figure 3. Notably, the adsorption amount of NH3 over the catalyst with improved preparation method is much larger than that with conventional preparation method (Figure 3), this directly points to the fact that enhancing the adsorption step of NH3 decomposition is one of the reasons for the high activity of catalyst with improved preparation method. Recently, CoPt/TiO2 with Co particle size of ~1 nm displayed a much higher Fischer-Tropsch reaction rate, which was also found to be due to increasing the amount of active site caused by using plasma-assisted preparation [44]. More importantly, in this study, the improved preparation method increased the acid strength of catalyst as well (Figure 3), as demonstrated by the increase in adsorption strength of NH3 over catalyst, which can promote the dissociation step of NH3; this is another crucial reason that explains the high activity of catalyst with the improved preparation method. Besides, the improved preparation method strengthened the interaction of Co with fumed SiO2 (Figure 4), indicating the difference in electronic structure of catalyst with different preparation methods, and this could influence the activity of catalyst as well.
In addition, using Co/fumed SiO2 catalyst prepared by the improved method, the influence of the combining mode of plasma and catalyst was investigated on the performance of plasma-catalytic NH3 decomposition, as shown in Scheme 1 and Figure 6. About 3 g Co/fumed SiO2 was packed in the reactor with a packing volume of about 3.1 mL, and the combining mode of plasma and catalyst changed through changing discharge volume "V", but the packed catalyst was fixed. Namely, changing "V" from 3.3 to 0.4 mL enabled the catalyst to be partly packed in the field of plasma, as shown in Scheme 1. Scheme 1. Scheme of combining mode of plasma and Co/fumed SiO2 catalyst (note: HV denotes high voltage; catalyst was fixed at about 3 g , but the discharge volume changes with the shortening of the length of the HV electrode, which results in the catalyst being partly packed in the field of plasma by changing the discharge volume "V" from 3.3 to 0.4 mL). Scheme 1. Scheme of combining mode of plasma and Co/fumed SiO 2 catalyst (note: HV denotes high voltage; catalyst was fixed at about 3 g , but the discharge volume changes with the shortening of the length of the HV electrode, which results in the catalyst being partly packed in the field of plasma by changing the discharge volume "V" from 3.3 to 0.4 mL).
In Figure 6a, interestingly, the conversion of ammonia was greatly enhanced with discharge volume decrease, and partly packing catalyst into the discharge area was found to be better than that of full-packing mode. Among the cases studied, the discharge volume with 0.4 mL showed the best activity towards NH 3 decomposition, in this case, the reaction temperature with 98.0% NH 3 conversion was only 380 • C, which was 140 • C lower than that in the case of catalyst alone. At the reaction temperature of 380 • C, the conversion of NH 3 over Co/fumed SiO 2 is only 6.2% without plasma whereas, at the same conditions, the use of DBD plasma significantly enhanced the reaction performance, and the conversion of NH 3 increased by a factor of 16 (from 6.1% to 98.0%) with decreasing discharge volume from 3.3 to 0.4 mL. Correspondingly, the energy efficiency of H 2 formation increased from 11.9 to 15.9 mol(kW·h) −1 ; this is the highest H 2 formation rate obtained in ammonia decomposition so far, as shown in Figure 6b. In addition, Figure 6c displayed that the specific energy input (SEI) significantly increased with decreasing discharge volume, which might be the reason for the high performance shown in Figure 6a increasing on the reaction performance, the reaction temperatures with different discharge volumes were all controlled at around 350 • C by adjusting energy input, then the relationship of ammonia conversion and SEI was presented in Figure 6d. Clearly, the conversion of ammonia increased with SEI increasing, demonstrating that the high performance resulting from high SEI was not due to heating of the catalyst. Furthermore, our previous studies revealed that increasing energy input of discharge can significantly facilitate the desorption of the strong-adsorbed N from catalyst surface (rate-limiting step in ammonia decomposition) [11], thus, the nature of the contribution of high SEI was to accelerate the rate-limiting step of ammonia decomposition.
changing "V" from 3.3 to 0.4 mL enabled the catalyst to be partly packed in the field of plasma, as shown in Scheme 1. Scheme 1. Scheme of combining mode of plasma and Co/fumed SiO2 catalyst (note: HV denotes high voltage; catalyst was fixed at about 3 g , but the discharge volume changes with the shortening of the length of the HV electrode, which results in the catalyst being partly packed in the field of plasma by changing the discharge volume "V" from 3.3 to 0.4 mL).
Catalysts 2018, 8, x FOR PEER REVIEW 8 of 13 Figure 6. Influence of discharge volume on (a) ammonia conversion, (b) energy efficiency of H2 formation, (c) specific energy input (SEI) and (d) relationship of ammonia conversion with SEI (NH3 feed rate 40 mL/min, discharge gap 3 mm, discharge frequency 12 kHz, and the packing amount and packing volume of catalyst in the reactor was fixed at about 3 g and 3.1 mL, respectively. Changing the discharge volume "V" from 3.3 to 0.4 mL enabled the catalyst to be partly packed in the discharge area, as shown in Scheme 1, and the catalyst was fully packed in the field of plasma only when the discharge volume was over 3.1 mL; The reaction temperature originated from electric heat released by discharge, and was determined using an IR camera and thermocouple tightly attached to the outer wall of the reactor [12]).
In Figure 6a, interestingly, the conversion of ammonia was greatly enhanced with discharge volume decrease, and partly packing catalyst into the discharge area was found to be better than that of full-packing mode. Among the cases studied, the discharge volume with 0.4 mL showed the best activity towards NH3 decomposition, in this case, the reaction temperature with 98.0% NH3 conversion was only 380 °C, which was 140 °C lower than that in the case of catalyst alone. At the reaction temperature of 380 °C, the conversion of NH3 over Co/fumed SiO2 is only 6.2% without plasma whereas, at the same conditions, the use of DBD plasma significantly enhanced the reaction performance, and the conversion of NH3 increased by a factor of 16 (from 6.1% to 98.0%) with decreasing discharge volume from 3.3 to 0.4 mL. Correspondingly, the energy efficiency of H2 formation increased from 11.9 to 15.9 mol(kW·h) −1 ; this is the highest H2 formation rate obtained in ammonia decomposition so far, as shown in Figure 6b. In addition, Figure 6c displayed that the specific energy input (SEI) significantly increased with decreasing discharge volume, which might be Figure 6. Influence of discharge volume on (a) ammonia conversion, (b) energy efficiency of H 2 formation, (c) specific energy input (SEI) and (d) relationship of ammonia conversion with SEI (NH 3 feed rate 40 mL/min, discharge gap 3 mm, discharge frequency 12 kHz, and the packing amount and packing volume of catalyst in the reactor was fixed at about 3 g and 3.1 mL, respectively. Changing the discharge volume "V" from 3.3 to 0.4 mL enabled the catalyst to be partly packed in the discharge area, as shown in Scheme 1, and the catalyst was fully packed in the field of plasma only when the discharge volume was over 3.1 mL; The reaction temperature originated from electric heat released by discharge, and was determined using an IR camera and thermocouple tightly attached to the outer wall of the reactor [12]).

DBD Plasma-Catalytic Reactor
NH 3 decomposition for H 2 generation was carried out in a DBD reactor with a catalyst bed in the discharge area at atmospheric pressure (Scheme 2). The DBD reactor was a typical cylindrical reactor using a stainless-steel rod (2 mm o.d.) as a high-voltage electrode placed along the axis of a quartz tube Catalysts 2019, 9, 107 9 of 13 (10 mm o.d. × 8 mm i.d.) which was used as a discharge dielectric. An aluminum foil sheet tightly covered the outside of the quartz cylinder and served as a ground electrode. A 3 mm of discharge gap was used, and catalyst was fully packed in the discharge area unless otherwise noted. The DBD reactor was connected to an AC high voltage power supply with a peak voltage of up to 30 kV and a variable frequency of 5-20 kHz. In this study, the discharge frequency was fixed at 12 kHz, and NH 3 with a purity of 99.999% was fed into the DBD reactor at a total flow rate of 40 ml/min. The products of NH 3 decomposition were analyzed on-line using a gas chromatograph (Shimadzu GC-2014) equipped with a thermal conductivity detector (TCD). The input power driving the reaction was determined from the product of the apparent voltage and current of AC power supply, and the discharge power was measured using a four-channel digital oscilloscope (Tektronix DPO 3012, high-voltage probe Tektronix P6015A, Tektronix Tech. Corp., Beaverton, OR, USA, current probe Pearson 6585, Pearson Electronics, Inc., San Jose, CA, USA). sheet tightly covered the outside of the quartz cylinder and served as a ground electrode. A 3 mm of discharge gap was used, and catalyst was fully packed in the discharge area unless otherwise noted. The DBD reactor was connected to an AC high voltage power supply with a peak voltage of up to 30 kV and a variable frequency of 5-20 kHz. In this study, the discharge frequency was fixed at 12 kHz, and NH3 with a purity of 99.999% was fed into the DBD reactor at a total flow rate of 40 ml/min. The products of NH3 decomposition were analyzed on-line using a gas chromatograph (Shimadzu GC-2014) equipped with a thermal conductivity detector (TCD). The input power driving the reaction was determined from the product of the apparent voltage and current of AC power supply, and the discharge power was measured using a four-channel digital oscilloscope ( To evaluate the reaction performance of plasma-catalytic NH3 decomposition to produce H2, the conversion of NH3 was calculated using Equation (1). The energy efficiency of H2 formation (mol(kW·h) −1 ), defined as the number of moles of H2 produced per kilowatt hour, was calculated using Equation (2). The specific energy input (SEI), defined as the energy input per discharge volume, was calculated using Equation (3) To evaluate the reaction performance of plasma-catalytic NH 3 decomposition to produce H 2 , the conversion of NH 3 was calculated using Equation (1). The energy efficiency of H 2 formation (mol(kW·h) −1 ), defined as the number of moles of H 2 produced per kilowatt hour, was calculated using Equation (2). The specific energy input (SEI), defined as the energy input per discharge volume, was calculated using Equation (3)

Catalyst Preparation
Cobalt nitrate was provided by the Tianjin Kermel Chemical Reagent Co., Ltd (Tianjin, China). Fumed SiO 2 was purchased from the Dalian Luming Nanometer Material Co., Ltd (Dalian, China). Catalysts were synthesized either using the conventional preparation method and improved preparation method. Incipient wetness impregnation was used in this study. Briefly, cobalt nitrate (the theoretical metal loading was 30 wt %) was dissolved in deionized water. The support of fumed SiO 2 was calcined, in advance, at 400 • C for 5 h to remove impurities, such as H 2 O, before impregnation, and then the pretreated support was added to the cobalt nitrate solution and stirred until it was thoroughly mixed. For "conventional preparation method", the resulting mixture was kept at room temperature for 3 h and dried in air overnight at 110 • C. The dried sample was finally calcined in air at 540 • C for 5 h. Different from the conventional preparation method, for the "improved preparation method", the resulting mixture was kept at room temperature for 3 h, followed by vacuum-freeze drying overnight at −50 • C before dried in air at 120 • C for 5 h, then the dried sample was calcined in a He-DBD plasma environment at 400 • C for 3 h to obtain the as-prepared catalyst. In addition, all the as-prepared catalysts were treated in NH 3 -DBD plasma at 400 • C for 0.3-1.0 h to reduce catalysts before evaluating their activity in NH 3 decomposition.

Catalyst Characterization
X-ray diffraction (XRD) patterns of as-prepared catalysts were recorded using a Rigaku D-Max 2400 X ray diffractometer with Cu K α radiation. Transmission electron microcopy (TEM) was used to characterize metal particles formed on the support surface (FEI Tecnai G2 F30 microscope, point resolution 0.2 nm, operated at 300 kV, Utrecht, Netherlands).
The reduction behavior of as-prepared catalyst was evaluated by H 2 temperature-programmed reduction (H 2 -TPR) using a Chemisorption instrument (ChemBET 3000, Quantachrome, Boynton Beach, FL, USA). The sample (100 mg) was pretreated at 500 • C for 1 h under He flow (20 mL/min), and then cooled to 50 • C. The pretreated sample was exposed to a H 2 /He mixture (10 vol% H 2 ) and was heated from 150 to 800 • C at a constant heating rate of 14 • C/min to get a H 2 -TPR profile. The acid-base properties of the as-prepared catalyst were tested by NH 3 temperature-programmed desorption (NH 3 -TPD) using the same Chemisorption instrument with operating H 2 -TPR. The sample (140 mg) was pretreated at 500 • C for 1 h under He flow (20 mL·min −1 ), and then cooled to 150 • C. The pretreated sample was saturated with NH 3 for 30 min, and then purged with He flow for 1 h at 150 • C. The TPD profile was recorded while the sample was heated from 150 to 600 • C at a constant heating rate of 14 • C·min −1 under He flow.
The specific surface area (S g ) of fumed SiO 2 support was tested by N 2 physisorption at −196 • C (Micrometrics ASAP 2020, Norcross, GA, USA). Prior to the N 2 physisorption measurement, fumed SiO 2 was degassed at 350 • C for 3 h, and S g was calculated using the Brunauer−Emmett−Teller (BET) equation.
The metal loading of fumed SiO 2 supported catalyst with different preparation methods was determined using X-ray fluorescence (XRF, SRS-3400, Bruker, Germany).

Conclusions
CO x -free H 2 generation from plasma-catalytic NH 3 decomposition has been significantly promoted over Co/fumed SiO 2 catalyst prepared with an improved preparation method, which featured the use of vacuum-freeze drying and DBD plasma calcination techniques during catalyst preparation. Compared with the activity of the catalyst prepared by the conventional preparation method, the conversion of NH 3 increased by 47% on Co/fumed SiO 2 catalyst prepared by improved method and, correspondingly, the energy efficiency of H 2 production increased from 2.3 to 5.7 mol(kW·h) −1 . The enhanced activity was mainly attributed to the high dispersion of Co particles on fumed SiO 2 with a narrow particle size distribution (2-3 nm), which brought more active sites, stronger acidity, and a strong metal-support interaction. In addition, the reaction performance was significantly improved with the increase of specific energy input. At 380 • C, the highest energy efficiency of H 2 formation achieved, so far, was 15.9 mol(kW·h) −1 over improved prepared Co/fumed SiO 2 catalyst with 98.0% ammonia conversion at the optimal conditions. Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4344/9/2/107/s1. Figure S1. Peak analysis of H 2 -TPR profile obtained over Co 3 O 4 /fumed SiO 2 catalyst; Table S1. XRF analysis of Co/fumed SiO 2 with improved preparation method; Table S2. XRF analysis of Co/fumed SiO 2 with conventional preparation method.