Activity, Selectivity, and Durability of Ruthenium Nanoparticle

Subscriber access provided by Kaohsiung Medical University

Energy, Environmental, and Catalysis Applications

Activity, Selectivity, and Durability of Ruthenium Nanoparticle Catalysts for Ammonia Synthesis by Reactive Molecular Dynamics Simulation: Size Effect Sung-Yup Kim, Hong Woo Lee, Sung Jin Pai, and Sang Soo Han ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.8b05070 • Publication Date (Web): 18 Jul 2018 Downloaded from http://pubs.acs.org on July 24, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Activity, Selectivity, and Durability of Ruthenium Nanoparticle Catalysts for Ammonia Synthesis by Reactive Molecular Dynamics Simulation: Size Effect Sung-Yup Kim, Hong Woo Lee, Sung Jin Pai, and Sang Soo Han* Computational Science Research Center, Korea Institute of Science and Technology (KIST), 5 Hwarangno 14-gil, Seongbuk-gu, Seoul 02792, Republic of Korea

AUTHOR INFORMATION Corresponding Author *[email protected]. Tel.: +82 2 958 5441. Fax: +82 2 958 5451.

Keywords Ruthenium nanoparticle, catalyst, ammonia synthesis, activity, selectivity, durability, molecular dynamics

1

ACS Paragon Plus Environment

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABSTRACT We report a molecular dynamics (MD) simulation employing the reactive force field (ReaxFF), developed from various first-principles calculations in this study, on ammonia (NH3) synthesis from nitrogen (N2) and hydrogen (H2) gases over Ru nanoparticle (NP) catalysts. Using ReaxFF-MD simulations, we predict not only the activities and selectivities but also the durabilities of the nano-catalysts and discuss the size effect and process conditions (temperature and pressure). Among the NPs (diameter = 3, 4, 5, and 10 nm) considered in this study, the 4 nm NPs show the highest activity, in contrast to our intuition that the smallest NP should provide the highest activity, as it has the highest surface area. In addition, the best selectivity is observed with the 10 nm NPs. The activity and selectivity are mainly determined by the hcp, fcc, and top sites on the Ru NP surface, which depend on the NP size. Moreover, the selectivity can be improved more significantly by increasing the H2 pressure than by increasing the N2 pressure. The durability of the NPs can be determined by the mean stress and the stress concentration, and these two factors have a trade-off relationship with the NP size. In other words, as the NP size increases, its mean stress decreases, while the stress concentration simultaneously increases. Due to these two effects, the best durability is found with the 5 nm NPs, which is also in contrast to our intuition that larger NPs should show better durability. We expect that ReaxFF-MD simulations, along with first-principles calculations, could be a useful tool in developing novel catalysts and understanding catalytic reactions.

2


Page 2 of 28

Page 3 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1. INTRODUCTION Ammonia (NH3) synthesis from nitrogen and hydrogen is still challenging despite technological developments over the past decades, largely because the nitrogen molecule (N2) is very stable due to its triple bond. The Haber-Bosch process, a representative method for NH3 synthesis, is highly energy intensive and requires high operating temperatures (400 ~ 500 ˚C) and pressures (150 ~ 300 bar), and a Fe-based catalyst [(e.g., magnetite (Fe3O4))] is often used. According to both theoretical and experimental studies performed over nearly a hundred years1, ruthenium (Ru), osmium (Os), and iron (Fe) are the best pure metal catalysts for NH3 synthesis, although magnetite has been commonly used because of its cost and availability. In particular, according to a volcano plot2 obtained from first-principles calculations, Ru shows superior activity over Fe and Os for NH3 synthesis. It was also experimentally reported that a Ru-based catalyst can produce NH3 with a higher yield at lower pressure and temperature than the conventional Fe-based catalyst3. These reports indicate that Ru is a promising catalyst for NH3 synthesis. To improve the catalytic activity, the nano-size effect and shape control of catalyst particles are conventionally considered, as the catalytic activity is very sensitive to the atomic structure and geometry of a catalyst. Van Hardeveld and Van Montfoort4 reported an active atomic site for N2 dissociation on nickel (Ni), palladium (Pd), and platinum (Pt) called the B5 site, involving an ensemble of five metal atoms that were mainly found at steps on metal surfaces. Afterward, through a joint experimental and theoretical study, Dahl et al.5 showed that the B5 site is most active for N2 dissociation and NH3 synthesis on ruthenium. Jacobsen et al.6 reported that the fraction of the number of B5 sites over the number of total surface atoms in Ru nanoparticles (NPs) is maximized when the NP diameter is in the range of 1.8~2.5 nm. Raróg-Pilecka et al.7 compared the fraction of B5-type sites with the 3



Page 4 of 28

experimental turnover frequency (TOF) of NH3 synthesis over Ru catalysts using a barium (Ba) promoter. For Ru NPs with d < 2 nm, the fraction of B5 sites and TOF showed similar trends; however, as the NP size increased (d > 2 nm), so did the discrepancy. As already mentioned, B5 sites are mainly found at steps on the NP surface. However, as the Ru NP size increases, the fraction of terraces over steps on the Ru surface increases, which can lead to discrepancies between theory and experiment, although the step site is basically more reactive than the terrace site.8 For

a

practical

catalyst,

not

only

activity

but

also

selectivity

and

durability/recyclability must be considered, and the selectivity and durability/recyclability depend on the atomic structure and size of the catalyst, as observed in the case of activity. It is usually expected that a smaller NP catalyst will show higher activity but lower stability. Recently, theoretical approaches, such as first-principles calculations, have been widely used to examine catalytic reaction mechanisms and design novel catalysts. For example, Honkala et al.9 calculated the rate of NH3 synthesis over a Ru catalyst, compared the calculated results with experimental results, and found that the calculated rate was within a factor 3 to 20 of the experimental rate, depending on the size of the Ru NPs. Additionally, Hellman et al.10-12 reported a micro-kinetic model based on a first-principles calculation to predict the activities of Ru catalysts for NH3 synthesis and showed a quantitative agreement between theory and experiment. Although first-principles calculations are undoubtedly valuable in developing catalysts, they mainly provide information regarding the catalyst activity and selectivity. The substantial computational cost of the calculation framework limits the number of atoms (system size) and the timescale (atom dynamics) for simulations of nano-sized structures, which leads to difficulty in predicting the durability/recyclability of nano-sized catalysts. In this regard, molecular dynamics (MD) simulations employing a reactive force field 4


Page 5 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(ReaxFF)13-15 can be a useful protocol to predict the activity, selectivity, and durability of nano-sized catalysts, because ReaxFF allows prediction of the chemical reactions between atoms (bond formation and dissociation) in nano-sized systems and the MD simulation allows modeling of the dynamic behaviors of atoms at given temperature or pressure conditions. Indeed, ReaxFF-MD simulations have been applied to various metallic16,17 and metal-oxide1824

catalysts, although these works focused mainly on the activity of the catalysts. In this work, we investigate the activity, selectivity, and durability/recyclability of Ru

NP catalysts for NH3 synthesis by ReaxFF-MD simulations; herein, the ReaxFF parameters for the Ru-N-H ternary system were developed from first-principles calculations. Then, we find the optimum size to maximize the activity, selectivity and durability of Ru NPs and discuss the reasons for this optimum size. In addition, we investigate the effects of H2 and N2 pressure on NH3 synthesis by the Ru NPs.

2. SIMULATION DETAILS To simulate NH3 generation from N2 and H2 gases over Ru NP catalysts, we performed MD simulations with ReaxFF, and thus, the ReaxFF parameters for ternary Ru-NH were required for this simulation. Based on the reported ReaxFF parameters for the binary N-H system,25 we developed ReaxFF by considering the atom parameters for Ru; bond parameters for Ru-Ru, Ru-N, and Ru-H; and off-diagonal parameters for Ru-N and Ru-H. The developed ReaxFF for the Ru-N-H system is summarized in Table S4 of the supporting information (SI). Using a successive one-parameter search technique,26 the ReaxFF parameters were optimized against various first-principles training data, in which we considered equations of states of various Ru crystals, surface formation energies for various Ru surfaces, adsorption energies of N and H atoms on Ru surfaces, various reaction pathways 5



(N2 dissociation, H diffusion, and NH3 formation) on Ru surfaces, and Ru-N and Ru-H bond dissociations in several non-periodic systems. Details of the ReaxFF development can be found in the SI. With the developed ReaxFF, we performed MD simulations using the LAMMPS software package27,28. To integrate Newton’s equations of motion of the atoms during the MD simulations, the Verlet algorithm29 was used with a time step of 0.5 fs (femtoseconds). All calculations were performed in the canonical NVT ensemble at 1,500 K to accelerate the chemical reactions between the Ru NPs and N2/H2 gases, and the temperature was maintained using a Nosé-Hoover thermostat with a damping parameter of 0.01 fs-1. To accelerate the chemical reactions between the solid catalyst and gas phases, we used a temperature of 1,500 K, which is higher than that used in the conventional Haber-Bosh process (673~773 K). In addition, various pressures of N2 and H2 gases were considered in the range of 300~1,000 atm, which is similar to the range used in experiments.3,30 To justify the use of 1,500 K, we calculated rate constants for N2 dissociation over the Ru NP (diameter = 4 nm) at various temperatures (700, 900, 1100, 1300, and 1500 K) and then obtained an energy barrier for the reaction. According to the previous DFT calculation,8 energy barriers for N2 dissociation over Ru(0001) surfaces are 43.8 kcal/mol for the terrace surface and 9.2 kcal/mol for the stepped surface. The developed ReaxFF reproduces the energy barriers (40.6 kcal/mol for the terrace and 11.0 kcal/mol for the step) as shown in Figures S4 and S5. Because the 4 nm Ru NP considered in our MD simulation has both of the terrace and steps, it can be expected that the energy barrier for N2 dissociation over the 4 nm NP is between 9.2 (step surface) and 43.8 kcal/mol (terrace). Indeed, our MD simulation performed at various temperatures reveal the energy barrier of 17.1 kcal/mol, which can justify the use of 1,500 K in this work. The details are found in Figures S12 and S13. 6


Page 6 of 28

Page 7 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


In the simulation model, we considered spherical Ru NP catalysts on a Ru slab, where the slab was assumed to be a support material, as shown in Figure 1. Typically, metal oxides or carbon materials are used as supports for metal NP catalysts. Considering these supports requires the development of additional ReaxFF parameters between the metallic catalyst and the support, which remains for future work. Instead, in this work, the Ru slab was considered to be the support, with the assumption that the Ru slab did not affect chemical reactions between the Ru and N2/H2 gases. We considered the support to investigate effects of interactions between Ru NP active materials and supports on their catalysis properties, in particular, the durability. And, a “flat” surface structure of the support can affect the durability of Ru NPs. The Ru atoms in the support were fixed during all MD simulations. In addition, several Ru NP sizes (diameter = 3, 4, 5, and 10 nm) were considered to investigate the size effects on the activity, selectivity, and durability of the catalyst, and the number of atoms and surface area are summarized in Table S5. If N2 and H2 gases coexist near Ru NP catalysts, H2 molecules are preferentially adsorbed on the Ru surface over N2, and then, the dissociated H atoms block the sites for N2 dissociation, which is known as the ‘H2 poisoning effect’.31 To avoid such ‘H2 poisoning effect’ and facilitate ammonia generation, our MD simulations proceeded in two steps. In the first step during the MD simulations, the Ru NP catalyst was surrounded by only gaseous N2 molecules at a given pressure (Figure 1a), and then, N2 dissociation reactions occurred on the Ru surface. In the second step, all surrounding N2 molecules that were not dissociated in the first step were removed, and the dissociated N atoms on the Ru support were also removed because they could contribute to NH3 generation. Then, H2 molecules were filled in the system, where the number of H2 molecules was determined based on the pressure. During the ReaxFF-MD simulations, these H2 molecules dissociated on the Ru NP surface and interacted with N atoms dissociated on the Ru surface 7



(Figure 1b), leading to the generation of NH3 molecules plus other byproducts (mainly N2H2)32. We termed one completion of the two steps as one cycle. To investigate the durability or recyclability of the Ru NP catalysts, the MD simulations were performed up to 3 cycles. Before the MD simulations, the simulation models were optimized until an energy tolerance of 10-5 eV was reached. Shapes of metallic NPs might change with their size. However, in this work, we focused on size effects of the Ru NPs on their catalytic performances with an assumption that their shape in the “free-standing” condition is always spherical irrespective of their size. According to an experimental report,33 the Ru NPs with an average size of 2.2 nm had a spherical shape. Because we considered Ru NPs with sizes of 3 to 10 nm bigger than the experimental case, the spherical shape of the Ru NPs in the present study is reasonable. Moreover, we fully optimized the NPs using the developed ReaxFF before MD simulations for prediction of their catalytic properties and found that their spherical shapes were still retained.

3. RESULTS AND DISCUSSION 3.1 Activity and selectivity of Ru NP catalysts for NH3 synthesis Among the three major properties for efficient catalysis, we first investigated the activity and selectivity of Ru NP catalysts. As already mentioned, to simulate NH3 synthesis in this work, we considered two processes: 1) N2 dissociation on bare Ru NP surfaces and 2) H2 dissociation on N-dissociated Ru surfaces, which leads to NH3 generation. Figure 2a shows the effects of Ru NP size on N2 dissociation. It is usually expected that smaller NPs will lead to higher activity due to their larger surface area. In Table S5, the smallest Ru NP indeed has the largest surface area. In terms of the N2 dissociation behavior (Figure 2a), it is 8


Page 8 of 28

Page 9 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


clear that the Ru NP with a diameter of 10 nm has a lower activity than the smaller NPs. However, although the Ru NPs with diameters of 3, 4, and 5 nm do not show a significant difference, the average values in the MD timescale of 100~250 ps reveal that the 4 nm Ru NP has a slightly higher N2 dissociation activity than the other NPs, i.e., the number of dissociated N atoms per surface area is 5.09×1018 m-2 for the 3 nm Ru NP, 5.29×1018 m-2 for the 4 nm Ru NP, and 4.99×1018 m-2 for the 5 nm Ru NP, which is in contrast to our expectation based on the surface area. According to a previous DFT calculation,8,34-37 the preferential adsorption sites of N, H, and NH3 on a Ru surface are hcp, fcc, and top, respectively, which is also in agreement with our DFT calculation shown in Figures S2~S7. This implies that N2 dissociation depends on the number of hcp sites on the Ru surface. Thus, we counted the number of hcp, fcc, and top sites of the Ru NPs as a function of their size, as shown in Figure S10. Indeed, the number of hcp sites is highest for the 4 nm Ru NP, followed by the 3, 5, and 10 nm Ru NPs. This trend is well correlated with the N2 dissociation activity predicted by the ReaxFF-MD simulation, as shown in Figure 2a. After the N2 dissociation process, the H2 dissociation process was performed, which leads to NH3 generation (Figure 2b). As in the N2 dissociation case, the largest NH3 generation is also found with the 4 nm Ru NP. However, interestingly, the second best NH3 generation is observed with the 10 nm Ru NP, which has the lowest N2 dissociation activity. This is associated with the number of the fcc sites, which are the preferential H2 dissociation sites. In Figure S10, the 10 nm Ru NP has the most fcc sites, indicating that on this NP more H2 molecules are dissociated into H atoms than on other NPs. Thus, as more dissociated H atoms interact with dissociated N atoms, more NH3 is generated. Additionally, the 10 nm NP has more top sites, which are the preferential sites for NH3, than the 3 and 4 nm NPs, which 9



can provide more active NH3 generation. On the other hand, during NH3 formation, the N2H2 byproduct is also generated, as shown in Figure 2b. Based on the generation numbers of NH3 and N2H2, the selectivities were calculated with the equation, Selectivity = No. of NH3/(Total No. of NH3 and N2H2), and the results are shown in Figure 2c. The highest selectivity is found with the 10 nm NP, while the lowest is found with the 3 nm NP. This is because the 10 nm Ru NP generates the least N2H2 and the 3 nm NP generates the most N2H2, as shown in Figure 2b. In the case of N2H2 formation, a smaller Ru NP size leads to more N2H2 generation. This trend is oppositely correlated with the number of fcc sites, which implies that the selectivity of the Ru NPs can be controlled by changing of the H2 pressure, as the fcc site is the preferential adsorption site of H atoms. The details are discussed in the next paragraph. To investigate effects of N2 and H2 pressure on the activity and selectivity of Ru NPs for NH3 synthesis, we performed ReaxFF-MD simulations at various pressure conditions and then counted the number of generated NH3 and N2H2, as shown in Figure 3. During these simulations, the 3 nm Ru NP, the smallest NP in this study, was considered because of the computational cost. The simulated activity and selectivity of the NPs when the N2 pressure is changed under fixed H2 pressure are shown in Figures 3a and b. As the N2 pressure increases, the number of generated NH3 also increases because the higher pressure facilitates more N2 dissociation. However, the increase in N2 pressure can also promote N2H2 generation. Thus, in terms of selectivity, increasing the N2 pressure is not necessarily beneficial. As seen in Figure 3b, the selectivities at PN2 = 300 and 1,000 atm are similar, while an extremely high pressure, such as 4,000 atm, can provide higher selectivity. We also investigated the activity and selectivity as functions of H2 pressure under fixed N2 pressure (Figures 3c and d). As with the N2 pressure, increasing the H2 pressure promotes NH3 production. Moreover, higher 10


Page 10 of 28

Page 11 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


H2 pressure provides higher selectivity. To form the byproduct (N2H2) on the Ru NP surface, the N atoms dissociated on the Ru surface must diffuse to form a chemical bond (N—N) between two N atoms. However, as the H2 pressure increases, more H atoms are passivated on the Ru surface, which can prevent the diffusion of N atoms. Additionally, according to our DFT calculations (Figure S6), the diffusion barrier of H atoms on the Ru surface is lower than that of N atoms, and furthermore, the developed ReaxFF reasonably describes the diffusion behavior predicted by the DFT calculation. Due to these effects, an increase in H2 pressure can improve the selectivity of Ru NPs. According to our simulation, among the NPs (3, 4, 5, and 10 nm) considered in the present work, the 4 nm NP, not the 3 nm, show the highest activity, and the best selectivity is observed with the 10 nm NP. Raróg-Pilecka et al.7 experimentally investigated the activity of Ru NPs with 1 to 4 nm, where they synthesized Ru NPs with various sizes by controlling Ru loading (e.g., 1 nm for 1 wt% Ru to 4 nm for 32 wt% Ru). And they found that the 4 nm Ru NP showed the best activity, which can support our simulation result. Moreover, according to a reported experiment,38 selectivity of Ru NPs increased with increasing Ru particle sizes from 1.7 to 12 nm, which supports our simulation result. To further support the size effects of Ru NPs on activity and selectivity, we additionally performed MD simulations for a system wherein it has both 3 nm and 4 nm NPs, and then compared its activity and selectivity with those for a system including two 3 nm NPs. According to the simulations, the system including the 3 and 4 nm Ru NPs shows higher activity and selectivity than the system including two 3 nm NPs, which is shown in Figure S15. This result clearly supports our claim that the 4 nm Ru NP has better properties in activity and selectivity than the 3 nm NP. 11



3.2 Durability or recyclability of Ru NP catalysts To investigate the durability/recyclability of the Ru NP catalysts, the N2/H2 purging simulations onto Ru NPs were performed up to 3 cycles. The initial simulation model for the 2nd cycle simulation was obtained by removing all chemical species, such as NH3, N2H2 and surrounding H2, except for N or H attached to the Ru NP surface, from the simulation model obtained after the 1st cycle simulation. The initial simulation model for the 3rd cycle was similarly obtained. Figure 4 shows the number of NH3 molecules generated by various Ru nano-catalysts with diameters of 3, 4, 5, and 10 nm as a function of a cycle number. For the 4, 5, and 10 nm NPs, the number of produced NH3 molecules decreases with the cycle number, although the difference between the 2nd and 3rd cycle is not significant. Interestingly, for the 3 nm NP, the number of NH3 molecules is the highest in the 2nd cycle. Moreover, as a measure of durability, we investigated the difference between the maximum and minimum number of NH3 molecules at 400 ps during cycling. As the NP size increases to 5 nm, the difference decreases, while above 5 nm, the difference increases. This indicates that the 5 nm Ru NP shows the best durability for NH3 synthesis. The decrease in NH3 generation with cycle number results from the mechanical degradation of the Ru NPs. The mechanical degradation is associated with the stresses acting upon the particles and the stress distribution. Thus, to discuss the durability of the Ru NPs, we calculated the von Mises stresses (σv) exerted on the NPs from one to three cycles with the following equation:

12


Page 12 of 28

Page 13 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(1)

where the

,

, and

are the normal stresses, and the

,

, and

are the

shear stresses in an x-y-z coordinate system. Figure 5a shows the mean von Mises stress on the NPs over time. As expected, the smaller particle shows a higher mean stress. The higher stress can lead to more structural degradation, which decreases NH3 generation. In Figure 4, the difference between the maximum and minimum number of NH3 during cycling shows the trend of 3 nm > 4 nm > 5 nm, which can be explained by the mean stress. However, the 10 nm Ru NP shows a greater difference between the maximum and minimum number of NH3 than the 5 nm Ru NP. This phenomenon is hard to explain with only the mean stress. Therefore, we also investigated von Mises stress distributions in the Ru NPs. Figure 5b shows the von Mises stress distributions on the surfaces and cross sections of the Ru NPs. It is found that as the NP size decreases, the stresses are more evenly distributed between the surface region and the core. Conversely, as the NP size increases, the stress concentration is more pronounced near the NP surface. To support this, we then analyzed the von Mises stress distribution inside the 3 and 10 nm Ru NPs along the radial direction, as shown in Figure 5c. For the 3 nm Ru NP, the stresses gradually increase from the core to the surface. On the other hand, in the 10 nm Ru NP, the stress abruptly increases near the surface, corresponding to the stress concentration phenomenon near the surface. Owing to the stress concentration, the 10 nm Ru NP has a higher stress value (19.5 GPa) near the surface than the 3 nm Ru NP (18.7 GPa), although the 3 nm NP has a higher mean stress than the 10 nm NP, as shown in Figure 5a. As such, the 13



stress concentration, along with the mean stress value, can aggravate the durability the Ru nano-catalyst. Figure 5d shows a schematic diagram of the contributions of the mean stress and the stress concentration to the durability, in which a trade-off relationship between the two factors and the NP size is observed. In addition, the best NP size to minimize contributions of the two factors is found to be 5 nm. To further investigate the size dependency of Ru NPs on durability, we also calculated von Mises stress for a system wherein it has both 3 nm and 10 nm NPs after the N2 dissociation (Figure S16). For the system, the 10 nm Ru NP shows the stress concentration phenomenon near the surface higher than in the 3 nm Ru NP, which is similar to our previous simulations for the isolated 3 and 10 nm NPs. As already mentioned, for the 3 nm Ru NP, the NH3 generation is the highest in the 2nd cycle, while the other Ru NPs (4, 5, and 10 nm) show the maximum generation in the 1st cycle (Figure 4). The unique property observed in the 3 nm Ru NP results from structural deformation during cycling. In Figure 6, the initial shape of both the 3 and 10 nm Ru NPs is spherical; however, after the 2nd cycle, the 3 nm NP shows more structural deformation (more facetted) than the 10 nm NP. We calculated the root-mean-square displacements (RMSDs) between the initial structure and that after the 2nd cycle and found that the 3 nm Ru NP indeed shows a higher RMSD (0.034 vs. 0.006), which clearly indicates that the smaller NP experiences more deformation during cycling. Such a severe deformation in the smaller size leads to a lowering of the mean stress value. In Figure S11, the 3 nm Ru NP obtained after the 2nd cycle has a lower mean stress than those obtained after the 1st and 3rd cycles, revealing that during the 2nd cycle, NH3 generation reaches a maximum. The severe structural deformation in the 3 nm NP is caused by two effects. Among the NPs considered in this work, 14


Page 14 of 28

Page 15 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


it has the smallest size, providing the highest surface area. A large number of dangling atoms on the surface can lead to surface relaxation followed by structural deformation. Additionally, as the NP size decreases, effects of the support on the structural deformation of the NP can increase. In other words, a flat surface of the support can deform shapes of the NPs. Indeed, as seen in Figure 6, after the 2nd cycle, the surface of the 3 nm NP in contact with the support is very flat, and the degree of contact is more significant than in the 10 nm NP. This result clearly shows that the support can also affect the durability of the NP catalysts.

4. CONCLUSIONS Using MD simulations with ReaxFF, we predicted the activity, selectivity, and durability/recyclability of Ru NP catalysts for NH3 generation from N2 and H2 gases, and effects of the NP size (3, 4, 5, and 10 nm) and the temperature/pressure were also investigated. Among the NPs considered in this study, the NP with a diameter of 4 nm shows the highest activity, in contrast to our intuition that the smallest NP should provide the highest activity, owing to it having the highest surface area. Another interesting point is also found in the selectivity, where the best selectivity is observed with the 10 nm NP. The activity and selectivity are mainly determined by the hcp, fcc, and top sites on the Ru NP surface, which rely on the NP size. Moreover, the selectivity can be improved more significantly by increasing the H2 pressure than by increasing the N2 pressure. On the other hand, the durability of the NPs can be determined by the mean stress and the stress concentration, and these two factors have a trade-off relationship with the NP size. In other words, as the NP size increases, its mean stress decreases, while the stress concentration

15



simultaneously increases. Due to these two effects, the best durability is found with the 5 nm NP, which is also in contrast to our intuition that larger NPs should show better durability. Using ReaxFF-MD simulations, one can predict not only the activities and selectivities but also the durabilities of the NP catalysts on the nano-scale. Here, we need to mention that predicting the durability of NP catalysts with first-principles calculations is currently challenging, which implies another novelty of this work. Moreover, the ReaxFFMD simulation can be used to perform a process simulation to maximize NH3 synthesis over NP catalysts by controlling the temperature or pressure. Therefore, we expect that ReaxFFMD simulations, along with first-principles calculations, could be a useful tool in developing novel catalysts and understanding catalytic reactions.

ASSOCIATED CONTENT Supporting Information The Supporting Information is available free of charge on the ACS Publication website at DOI: Details on development of ReaxFF parameters for the Ru-N-H system and additional ReaxFF-MD simulation results.

AUTHOR INFORMATION Corresponding Author *E-mail: [email protected]

16


Page 16 of 28

Page 17 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ACKNOWLEDGMENT This work was supported by the Creative Materials Discovery Program through National Research Foundation of Korea (NRF-2016M3D1A1021140) and KIST institutional project (2E28000).

17



References (1) Liu, H. Z., Ammonia Synthesis Catalyst 100 Years: Practice, Enlightenment and Challenge. Chin. J. Catal. 2014, 35 (10), 1619-1640. (2) Norskov, J. K., Electronic Factors in Catalysis. Prog. Surf. Sci. 1991, 38 (2), 103-144. (3) Saadatjou, N.; Jafari, A.; Sahebdelfar, S., Ruthenium Nanocatalysts for Ammonia Synthesis: A Review. Chem. Eng. Commun. 2015, 202 (4), 420-448. (4) van Hardeveld, R.; van Montfoort, A., The Influence of Crystallite Size on the Adsorption of Molecular Nitrogen on Nickel, Palladium and Platinum. Surface Science 1966, 4 (4), 396-430. (5) Dahl, S.; Törnqvist, E.; Chorkendorff, I., Dissociative Adsorption of N2 on Ru(0001): A Surface Reaction Totally Dominated by Steps. Journal of Catalysis 2000, 192 (2), 381-390. (6) Jacobsen, C. J. H.; Dahl, S.; Hansen, P. L.; Tornqvist, E.; Jensen, L.; Topsoe, H.; Prip, D. V.; Moenshaug, P. B.; Chorkendorff, I., Structure Sensitivity of Supported Ruthenium Catalysts for Ammonia Synthesis. J. Mol. Catal. A-Chem. 2000, 163 (1-2), 19-26. (7) Raróg-Pilecka, W.; Miskiewicz, E.; Szmigiel, D.; Kowalczyk, Z., Structure Sensitivity of Ammonia Synthesis over Promoted Ruthenium Catalysts Supported on Graphitised Carbon. Journal of Catalysis 2005, 231 (1), 11-19. (8) Logadottir, A.; Norskov, J. K., Ammonia Synthesis over a Ru(0001) Surface Studied by Density Functional Calculations. Journal of Catalysis 2003, 220 (2), 273-279. (9) Honkala, K.; Hellman, A.; Remediakis, I. N.; Logadottir, A.; Carlsson, A.; Dahl, S.; Christensen, C. H.; Norskov, J. K., Ammonia Synthesis from First-Principles Calculations. Science 2005, 307 (5709), 555-558. (10) Hellman, A.; Baerends, E. J.; Biczysko, M.; Bligaard, T.; Christensen, C. H.; Clary, D. C.; Dahl, S.; van Harrevelt, R.; Honkala, K.; Jonsson, H.; Kroes, G. J.; Luppi, M.; Manthe, U.; Norskov, J. K.; Olsen, R. A.; Rossmeisl, J.; Skulason, E.; Tautermann, C. S.; Varandas, A. J. C.; Vincent, J. K., Predicting Catalysis: Understanding Ammonia Synthesis from FirstPrinciples Calculations. Journal of Physical Chemistry B 2006, 110 (36), 17719-17735. (11) Hellman, A.; Honkala, K.; Remediakis, I. N.; Logadottir, A.; Carlsson, A.; Dahl, S.; Christensen, C. H.; Norskov, J. K., Insights into Ammonia Synthesis from First-Principles. Surface Science 2006, 600 (18), 4264-4268. (12) Hellman, A.; Honkala, K.; Remediakis, I. N.; Logadottir, A.; Carlsson, A.; Dahl, S.; Christensen, C. H.; Norskov, J. K., Ammonia Synthesis and Decomposition on a Ru-Based Catalyst Modeled by First-Principles. Surface Science 2009, 603 (10-12), 1731-1739. (13) van Duin, A. C. T.; Dasgupta, S.; Lorant, F.; Goddard, W. A., ReaxFF: A Reactive Force Field for Hydrocarbons. J. Phys. Chem. A 2001, 105 (41), 9396-9409. (14) van Duin, A. C. T.; Larter, S. R., Molecular Dynamics Investigation into the Adsorption of Organic Compounds on Kaolinite Surfaces. Organic Geochemistry 2001, 32 (1), 143-150. (15) Chenoweth, K.; van Duin, A. C. T.; Goddard, W. A., ReaxFF Reactive Force Field for Molecular Dynamics Simulations of Hydrocarbon Oxidation. J. Phys. Chem. A 2008, 112 (5), 1040-1053. (16) Zhang, C.; van Duin, A. C. T.; Seo, J. W.; Seveno, D., Weakening Effect of Nickel Catalyst Particles on the Mechanical Strength of the Carbon Nanotube/Carbon Fiber Junction. Carbon 2017, 115, 589-599. (17) Gai, L. L.; Shin, Y. K.; Raju, M.; van Duin, A. C. T.; Raman, S., Atomistic 18


Page 18 of 28

Page 19 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Adsorption of Oxygen and Hydrogen on Platinum Catalysts by Hybrid Grand Canonical Monte Carlo/Reactive Molecular Dynamics. J. Phys. Chem. C 2016, 120 (18), 9780-9793. (18) Chenoweth, K.; van Duin, A. C. T.; Persson, P.; Cheng, M. J.; Oxgaard, J.; Goddard, W. A., Development and Application of a ReaxFF Reactive Force Field for Oxidative Dehydrogenation on Vanadium Oxide Catalysts. J. Phys. Chem. C 2008, 112 (37), 1464514654. (19) Shin, Y. K.; Kwak, H.; Vasenkov, A. V.; Sengupta, D.; van Duin, A. C. T., Development of a ReaxFF Reactive Force Field for Fe/Cr/O/S and Application to Oxidation of Butane over a Pyrite-Covered Cr2O3 Catalyst. ACS Catal. 2015, 5 (12), 7226-7236. (20) Senftle, T. P.; Meyer, R. J.; Janik, M. J.; van Duin, A. C. T., Development of a ReaxFF Potential for Pd/O and Application to Palladium Oxide Formation. Journal of Chemical Physics 2013, 139 (4), 15. (21) Goddard, W. A.; Mueller, J. E.; Chenoweth, K.; van Duin, A. C. T., ReaxFF Monte Carlo Reactive Dynamics Application to Resolving the Partial Occupations of the M1 Phase of the MoVNbTeO Catalyst. Catalysis Today 2010, 157 (1-4), 71-76. (22) Chenoweth, K.; van Duin, A. C. T.; Goddard, W. A., The ReaxFF Monte Carlo Reactive Dynamics Method for Predicting Atomistic Structures of Disordered Ceramics: Application to the Mo3VOx Catalyst. Angew. Chem.-Int. Edit. 2009, 48 (41), 7630-7634. (23) Goddard, W. A.; Chenoweth, K.; Pudar, S.; van Duin, A. C. T.; Cheng, M. J., Structures, Mechanisms, and Kinetics of Selective Ammoxidation and Oxidation of Propane over Multi-metal Oxide Catalysts. Top. Catal. 2008, 50 (1-4), 2-18. (24) Goddard, W. A.; van Duin, A.; Chenoweth, K.; Cheng, M. J.; Pudar, S.; Oxgaard, J.; Merinov, B.; Jang, Y. H.; Persson, P., Development of the ReaxFF Reactive Force Field for Mechanistic Studies of Catalytic Selective Oxidation Processes on BiMoOx. Top. Catal. 2006, 38 (1-3), 93-103. (25) Kamat, A. M.; van Duin, A. C. T.; Yakovlev, A., Molecular Dynamics Simulations of Laser-Induced Incandescence of Soot Using an Extended ReaxFF Reactive Force Field. J. Phys. Chem. A 2010, 114 (48), 12561-12572. (26) Vanduin, A. C. T.; Baas, J. M. A.; Vandegraaf, B., Delft Molecular Mechanics - A New Approach To Hydrocarbon Force Fields Inclusion Of A Geometry-Dependent Charge Calculation. J. Chem. Soc.-Faraday Trans. 1994, 90 (19), 2881-2895. (27) Aktulga; Fogarty; Pandit; Grama, Parallel Computing 2012, 38, 245-249. (28) Plimpton, S., Fast Parallel Algotithms For Short-Range Molecular-Dynamics. Journal of Computational Physics 1995, 117 (1), 1-19. (29) Spreiter, Q.; Walter, M., Classical Molecular Dynamics Simulation with the Velocity Verlet Algorithm at Strong External Magnetic Fields. Journal of Computational Physics 1999, 152 (1), 102-119. (30) The Claude Process for Ammonia Synthesis. Nature 1921, 107, 765. (31) Aslan, M. Y., Supported Ru Based Ammonia Synthesis Catalysts. MSc thesis, Middle East Technical University 2012. (32) Ozaki, A., Sir Hugh Taylor, F.R.S., Boudart, M, Kinetics and Mechanism of the Ammonia Synthesis. Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences 1960, 258 (1292), 47-62. (33) Gang L.; Nagasawa. H.; Kanezashi, M.; Yoshioka, T.; Tsuru, T.; Graphene Nanosheets Supporting Ru Nanoparticles with Controlled Nanoarchitectures Form a HighPerformance Catalyst for COx-Free Hydrogen Production from Ammonia. Journal of Material Chemistry A 2014, 2, 9185-9192. 19



(34) Jacobi, K.; Dietrich, H.; Ertl, G., Nitrogen Chemistry on Ruthenium Single-Crystal Surfaces. Applied Surface Science 1997, 121, 558-561. (35) Schwegmann, S.; Seitsonen, A. P.; Dietrich, H.; Bludau, H.; Over, H.; Jacobi, K.; Ertl, G., The Adsorption of Atomic Nitrogen on Ru(0001): Geometry and Energetics. Chemical Physics Letters 1997, 264 (6), 680-686. (36) Shi, H.; Jacobi, K., Hydrogen Vibrations on the Ru(001) Surface Revisited. Surface Science 1994, 313 (3), 289-294. (37) Lindroos, M.; Pfnur, H.; Feulner, P.; Menzel, D., A Study of the Adsorptioin Sites of Hydrogen on Ru(001) at Saturation Coverage by Electron Reflection. Surface Science 1987, 180 (1), 237-251. (38) Eslava, J.; Sun, X.; Gascon, J.; Kapteijn, F.; Rodriguez-Ramos, I. Ruthenium Particle Sizeand Cesium Promotion Effects in Fischer-Tropsch Synthesis over High-Surface-Area Graphite Supported Catalysts. Catalysis Science & Technology 2017, 7, 1235-1244.

20


Page 20 of 28

Page 21 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1. Simulation models. (a) Initial atomic configuration for the N2 dissociation process, (b) initial atomic configuration for the H2 dissociation process, and (c) MD snapshot after H2 purging, leading to NH3 generation.

21



Figure 2. Number (No.) of (a) dissociated N and (b) generated NH3 (thick line) and N2H2 (thin line) on various Ru NPs and (c) NH3/(N2H2+NH3) (selectivity) of the Ru NPs. Color codes are pink = 3 nm, green = 4 nm, blue = 5 nm, and gray = 10 nm.

22


Page 22 of 28

Page 23 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3. Effects of N2 and H2 pressure on the activity and selectivity of the 3 nm Ru NP for NH3 synthesis. (a) Generated number (No.) of NH3 (thick line) and N2H2 (thin line) with time at fixed H2 pressure (1,000 atm), (b) NH3/(N2H2+NH3) (selectivity) with time at fixed H2 pressure (1,000 atm), (c) generated No. of NH3 (thick line) and N2H2 (thin line) with time at fixed N2 pressure (1,000 atm), and (d) NH3/(N2H2+NH3) (selectivity) with time at fixed N2 pressure (1,000 atm). Color codes are purple = 300 atm, blue = 1,000 atm, and orange = 4,000 atm.

23



Figure 4. Effects of the cycle number on the number of NH3 produced on the Ru NP surfaces during the MD simulation: (a) 3 nm, (b) 4 nm, (c) 5 nm, and (d) 10 nm. The orange, navy and green lines correspond to the 1st, 2nd and 3rd cycle, respectively.

24


Page 24 of 28

Page 25 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(a)

(b) 25



(c)

Page 26 of 28

(d)

Figure 5. (a) The mean von Mises stresses for the 3 nm (orange), 4 nm (navy), 5 nm (pink), and 10 nm (black) Ru NPs. The stresses were calculated during the 1st NH3 generation cycle. (b) Distributions of the von Mises stresses on the surfaces and cross sections of the NPs obtained after the N2 dissociation process, where the same N coverage (19%) was considered. (c) The von Mises stress distribution inside the NPs along the radial direction. The stresses were analyzed as a function of the normalized radius (r/R), where r is the distance from the core of each NP along the radial direction and R is the radius of each NP. In addition, the same N coverage was considered. (d) Schematic diagram showing the effects of the mean stress and the stress concentration with NP diameter on the durability of the NP catalysts for NH3 synthesis.

26


Page 27 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6. Effects of the cycle number on the von Mises stress distribution for the (a-b) 3 nm and (c-d) 10 nm Ru NPs. Here, (a) and (c) are for the initial structures, and (c) and (d) are for the structures obtained after the 2nd cycle of NH3 synthesis.

27



TOC GRAPHICS

NH3 Synthesis

28


Page 28 of 28

Activity, Selectivity, and Durability of Ruthenium Nanoparticle

Recommend Documents