Rhenium containing binary catalysts for ammonia synthesis

Applied Catalysis A: General 209 (2001) 317–325

Rhenium containing binary catalysts for ammonia synthesis Ryoichi Kojima, Ken-ichi Aika∗ Department of Environmental Chemistry and Engineering, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, 4259 Nagatsuta, Midori-ku, Yokohama 226-8502, Japan Received 25 April 2000; received in revised form 18 August 2000; accepted 21 August 2000

Abstract The rhenium nitride was prepared and applied for ammonia synthesis. Under the synthesis conditions (623 K, 0.1 MPa), the nitride was partly decomposed (Re3 N + Re) but the activity was stable and higher (179 ␮mol h−1 g−1 at 623 K under 0.1 MPa) than the other transition metal nitrides (VN, NbN, Mo2 N, and W2 N), although the BET surface area was immeasurably low. The active phase was considered to be Re3 N. The effect of transition metal addition to the rhenium catalyst for ammonia synthesis was investigated. Cobalt and iron were effective, but nickel, chromium, and copper were not effective. The best combination was Co-Re (1:4) (492 ␮mol h−1 g−1 at 623 K under 0.1 MPa, expressed as Co-Re4 catalyst). NH3 treatment of Co-Re4 at 973 K gave a new XRD pattern (probably some kind of rhenium nitride phase), which was suggested to be responsible to the high activity of Co-Re4 catalyst. The ammonia synthesis rate over Co-Re4 catalyst was increased under high pressure (2372 ␮mol h−1 g−1 at 623 K under 3.1 MPa), however, the rate overshot above 1.1 MPa. The kinetic analysis indicated that Re surface was strongly covered either by hydrogen or by nitrogen (ammonia). By the addition of cobalt, the degree of nitrogen (ammonia) poisoning was weakened, but at the same time, the degree of hydrogen poisoning was strengthened. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Ammonia synthesis; Rhenium catalysts; Nitride; Addition of cobalt

1. Introduction Rhenium-based catalysts have not been studied extensively, since this element was not discovered until 1925 and was not readily available until many years later; the crustal abundance of this element was very low. However, rhenium sometimes attracts researchers because of its special properties (high melting point, hardness, catalytic ability of hydrogenation, and so on). It is well known that rhenium is an effective promoter of Pt/Al2 O3 for reforming reaction [1]. Recently, the synthesis method of rhenium (group ∗ Corresponding author. Tel.: +81-45-924-5416; fax: +81-45-924-5441. E-mail address: [email protected] (K.-i. Aika).

7) nitride and its application in hydrodesulfurization and hydrodenitrogenation reaction have been also reported by Clark et al. [2]. Re metal itself has received attention due to its intrinsic activity of ammonia synthesis. Somorjai and co-workers [3,4] have showed that the single crystal and polycrystalline surface are quite active for ammonia synthesis. Polycrystalline rhenium is more active than the most active crystal plane of iron (Fe(1 1 1)), and exposure to water or presulfidation has little effect on its catalytic activity (these behaviors are in marked contrast to those of iron). The reaction rate is remarkably sensitive to the catalyst surface structure. The reactivity ratio of Re(0 0 0 1), Re(1 0 1 0), Re(1 1 2 0), and Re(1 1 2 1) crystal faces are to be 1:94:920:2820, respectively. Asscher and co-workers [5,6] have stud-

0926-860X/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 0 0 ) 0 0 7 6 4 - X

318

R. Kojima, K.-i. Aika / Applied Catalysis A: General 209 (2001) 317–325

ied the adsorption of dinitrogen on rhenium single crystals, and concluded that activation of dinitrogen on rhenium was structure-sensitive, which agreed well with the result of ammonia synthesis on the same single crystals. Ozaki et al. also have found that ammonia synthesis activity of Re/AC is remarkably promoted with the addition of potassium metal [7]. In the series of these works, the authors first became interested in whether nitride formation is beneficial or not for ammonia synthesis. Although transition metal carbides and nitrides of groups 4–6 metals are thermodynamically stable and have been studied for the application in catalysis of hydrogenation [8–10], those of group 7 are relatively unstable and have received far less attention. Clark et al. [2] have reported that rhenium nitride is formed either from NH4 ReO4 or ReCl3 through ammonia reduction at temperatures between 573 and 623 K, and it decomposes to the metal above 645 K. It cannot be prepared directly from the elements at elevated temperature because of its instability, nor can it be formed from the oxides because their reduction temperatures are above the decomposition temperature of the oxide. Thus, the application of rhenium nitride for the catalysis must be studied below 645 K. The development of an active catalyst based on rhenium was carried out using methods similar to those for one based on the molybdenum, which resulted in finding Co3 Mo3 N, the active catalyst [11]. Co3 Mo3 N was prepared with sophisticated procedures, but was found to be active and stable. The purpose of this work is to apply rhenium nitride for ammonia synthesis, and to investigate the effect of the other transition metal additions. Before focusing on rhenium, the nitrides of other metals were also studied.

temperature programmed reaction of the precursors under flowing ammonia following the procedure of Clark et al. [2] and Choi et al. [12]. For Re3 N, the precursor (ammonium perrhenate) was heated under flowing ammonia gas (160 ml min–1 ) at 5.0 K min−1 to 623 K (Re3 N). For Mo2 N, the oxide precursor was heated under flowing ammonia gas (60 ml min−1 ) at 10.0 K min−1 to 623 K, 0.6 K min−1 to 723 K, and 3.0 K min−1 to 973 K. The other nitride catalysts, the precursors were heated under ammonia gas (60 ml min−1 ) at 5.0 K min−1 to 973 K (for VN) or 1073 K (for W2 N and NbN). The catalysts were held at 623 K for 2 h (for Re3 N) or at 973 K for 1 h (for Mo2 N and VN) or 1073 K for 1 h (for W2 N and NbN) and then quenched to room temperature. Ammonium perrhenate (NH4 ReO4 ) was used also as a precursor of rhenium “metal” catalyst. For the mixed metal catalysts, the precursors were prepared by mixing the metal (Cr, Fe, Co, Ni, and Cu) nitrate and ammonium perrhenate. The samples were heated under flowing ammonia gas (160 ml min–1 ) at 5.0 K min−1 to 973 K and held at 973 K for 2 h, and then quenched to room temperature. The high temperature (973 K) was necessary to reduce the second metal precursor. For Cr-Re and Cu-Re systems, the ratios of Cr (or Cu) and Re were 1:4 (mol:mol, Re contents were 80.0 mol%), and the samples were expressed as Cr-Re4 and Cu-Re4 , respectively. For Co-Re, Fe-Re, and Ni-Re systems, the ratios of Co (or Fe, Ni) and Re were 1:3, 1:4, 1:6, and 1:9 (mol:mol, Re contents were 75.0, 80.0, 85.7, and 90.0 mol%, respectively), and the samples were expressed as Co-Re3 , Co-Re4 , Co-Re6 , Co-Re9 , Fe-Re3 , Fe-Re4 , Fe-Re6 , Fe-Re9 , Ni-Re3 , Ni-Re4 , Ni-Re6 , and Ni-Re9 , respectively. 2.2. Ammonia synthesis

2. Experimental 2.1. Catalyst preparations For rhenium, ammonium perrhenate (NH4 ReO4 ) was used as the precursor of rhenium nitride catalyst (Re3 N) or rhenium metal. For the other transition metal, metal oxides (MoO3 , WO3 , V2 O5 , and Nb2 O5 ) were used as the precursor of “nitride” catalysts (Mo2 N, W2 N, VN, and NbN). These transition metal “nitride” catalysts were synthesized by the

After catalyst preparations, ammonia synthesis was performed in the same apparatus. The standard reaction condition was at 623 K under 0.1 MPa with a flow rate of 60 ml min−1 of (N2 + 3H2 ). The catalyst weight was 0.4 g. The reaction apparatus and the detection method of ammonia has been reported elsewhere [13]. Ammonia synthesis reaction under 0.1 MPa was performed in a glass tube flow system. The ammonia synthesis rate (activity) was measured after the catalyst was stabilized at the reaction temper-

R. Kojima, K.-i. Aika / Applied Catalysis A: General 209 (2001) 317–325

ature (for 30 min). The ammonia yield was calculated by detecting the decrease in conductivity of protons in a diluted sulfuric acid solution (0.00216N) with respect to the time, where the vent gas from the reactor was made to flow through the sulfuric acid solution. A linear decrease in conductivity versus time was observed during the measurement. Since high pressure ammonia synthesis was done in the other apparatus, the catalyst must be moved in the air. Thus, the catalyst was once stored in a glass bottle (in air), then, reactivated with the reactant gas (60 ml min−1 ) at 623 K. Ammonia synthesis reaction after the reactivation was performed in a computer-controlled stainless steel reactor of a flow system. The method of measuring the ammonia synthesis rate (activity) was the same with the case of before the passivation. High pressure activity measurement were performed under 1.1, 2.1, and 3.1 MPa at 623 K after the reactivation with the reactant gas for Co-Re4 catalyst, Ni-Re4 catalyst, and Re3 N (after 5 h reaction, Re-Re3 N). 2.3. Kinetic analysis The kinetic analyses on rhenium catalysts are mainly based on a model described elsewhere [14]. The principle is repeated as follows: a r = k1 PNn2 PHh2 PNH 3

(1)

An empirical rate expression of ammonia synthesis is adopted, where r is the rate of ammonia synthesis (lNH3 (STP) g−1 h−1 ), q the gas flow (l h−1 ), w the weight of the catalyst, and y0 is the concentration (mole fraction) of NH3 in the gas mixture leaving the catalyst bed. According to the conservation of mass, the following can be applied at a point in a reactor: qy + r dw = q(y + dy)     1 1 dw = dy q r

319

limited range with a constant ratio of H/N. Thus, the changes in ammonia concentration at the exit with the change of the flow rate could give the reaction rate r0 =

(1/w) dy0 d(1/q)

(5)

The study in [14] demonstrated that an empirical linear relationship of m log y0 = log C − log q

(6)

was applicable to the runs for every catalyst. From Eq. (6), one can obtain   1 (m−1) dy0 = C d (7) my0 q Inserting this in Eq. (5), then the rate expression can be given as    1 C (1−m) (8) y r0 = w m 0 where C and m are constants within the limited range of q. The reaction order with respect to ammonia (a) is therefore obtained as (1 − m). The order of H2 (h) now can be determined by Eq. (6) through changing the partial pressure of hydrogen while keeping a constant partial pressure of nitrogen or reversibly to obtain the order of N2 (n). For the term C, which is independent of ammonia concentration, is a function of hydrogen and nitrogen pressure. Therefore, the orders of n, h, and a in Eq. (1) can be determined experimentally. C = k2 PHh2 PNn2

(9)

C = k20 PHh2

(10)

log C = log k20 + h log PH2

(11)

(2)

C = k200 PNn2

(12)

(3)

log C = log k200 + n log PN2

(13)

If the gas flow is kept constant, then the integration of Eq. (3) (from the inlet to the outlet) gives Z y0   1 w = dy (4) q r 0 In practical measuring, the weight of catalyst is kept unchanged; instead, the gas flow rate is changed in a

The constituent gases of the reactant (N2 , H2 , He (ml min−1 )) were as follows: (5, 15, 0), (10, 30, 0), (15, 45, 0), and (20, 60, 0) for NH3 order; (5, 24, 31), (5, 30, 25), (5, 42, 13), and (5, 54, 1) for H2 order, and (6, 24, 30), (10, 24, 26), (15, 24, 21), and (22, 24, 14) for N2 order. These analyses were performed for Co-Re4 catalyst, Ni-Re4 catalyst, and Re3 N (after 5 h reaction, Re-Re3 N) at 623 K under 3.1 MPa.

320

R. Kojima, K.-i. Aika / Applied Catalysis A: General 209 (2001) 317–325

2.4. BET and XRD measurements BET surface areas were measured at 77 K using 100% N2 . The measurements were performed before the ammonia synthesis runs for all the catalysts. The rhenium catalysts were analyzed by X-ray diffraction (XRD) for phase identification, using a Rigaku Geigerflex RAD-1VB with a Cu K␣ monochromatized radiation source. The measurements were performed before (Re3 N, Re, Co-Re4 , and Ni-Re4 ) and after 5 h ammonia synthesis runs (only Re3 N).

3. Results and discussion 3.1. Comparison of the synthesis rates over transition metal nitrides and their surface areas Table 1 shows the synthesis conditions, ammonia synthesis rates, and surface areas of rhenium nitride (Re3 N), rhenium metal, and the other transition metal nitrides. The surface areas of transition metal nitrides (Mo2 N, W2 N, VN, and NbN) were measurable (20–190 m2 g−1 ), on the other hand, that of Re3 N was extremely low (0.5 m2 g−1 ). That of Re metal was unable to be measured. Thus, the activity was compared with gram-base in Table 1. In spite of the low surface area, the rate over Re3 N was very high, the initial activity was 430 ␮mol h−1 g−1 . (Re3 N was stable below 645 K, thus the reaction temperature was decided at 623 K.) On the other hand, the rates over other (high surface area) transition metal nitrides were much lower; 35, 24, 36, and 60 ␮mol h−1 g−1 for Mo2 N, W2 N, VN, and NbN, respectively, even

Fig. 1. Time course of ammonia synthesis rate at 623 K under 0.1 MPa over 0.4 g of rhenium catalysts: (䊐) Re3 N; (䉫) Re metal.

at 673 K. Rhenium metal itself was more active than these nitrides. The activity was 219 ␮mol h−1 g−1 at 673 K and 108 ␮mol h−1 g−1 at 623 K, of which value was lower than that of Re3 N. This is surprising if we consider the low surface area. 3.2. Rhenium nitride catalyst Fig. 1 shows the time course of ammonia synthesis rates at 623 K under 0.1 MPa over Re3 N and Re metal. The catalytic feature of Re3 N was drastic deactivation. But it stopped after 2 h run. On the other hand, the activity of rhenium metal was lower than that of nitride, but stable.

Table 1 Synthesis conditions, ammonia synthesis rates and surface areas (Sg ) of the nitridesa Catalyst

Precursor

Tn (K)

Rate (␮mol h−1 g−1 )

Tr (K)

Sg (m2 g−1 )

Re3 N Re metal Re metal Mo2 N Mo2 N W2 N VN NbN

NH4 ReO4 NH4 ReO4 NH4 ReO4 MoO3 MoO3 WO3 V2 O5 Nb2 O5

623 973 973 973 973 1073 973 1073

430 108 219 20 35 24 36 60

623 623 673 623 673 673 673 673

0.5 – – 186 186 43 18 32

a

Tn is the maximum (last) temperature at nitridation; Tr is the reaction temperature.

R. Kojima, K.-i. Aika / Applied Catalysis A: General 209 (2001) 317–325

321

Table 2 Ammonia synthesis rates at 623 K under 0.1 MPa over transition metal added rhenium catalysts Catalyst

Rate (␮mol h−1 g−1 )

Cr-Re4 Fe-Re4 Co-Re4 Ni-Re4 Cu-Re4

119 275 492 143 60

The partial decomposition of the nitride seems to be corresponding to the catalytic deactivation. The initial activity might be over-estimated by counting the decomposed ammonia from nitride, Re3 N. But after the system stabilized (after 2 h reaction), the activity was still higher than that of Re metal, suggesting more effectiveness of the nitride surface. This catalyst (Re3 N after 5 h reaction) is expressed as Re-Re3 N. 3.3. The effect of the other transition metal additions to rhenium metal

Fig. 2. XRD patterns of rhenium nitride and metal: (a) Re3 N (before the reaction); (b) Re3 N (after the reaction); (c) Re metal.

XRD measurement was performed before and after the reaction to examine the cause of the deactivation. Fig. 2 shows XRD patterns of Re3 N (fresh), Re3 N (after 5 h reaction), and Re metal. During 5 h ammonia synthesis run, Re3 N was decomposed to mixture of the nitride (a short and wide peak) and the metal (sharp peaks). Since the activity after 2 h ammonia synthesis was stable, there seems to be some equilibrium between Re3 N and Re metal under this condition.

The nitride surface must be the active state, but it is unstable. The second metal addition may stabilize Re3 N (such as Co3 Mo3 N). All the sample were treated at 973 K under ammonia stream. Table 2 shows the ammonia synthesis rates at 623 K under 0.1 MPa over transition metal (Cr, Fe, Co, Ni, and Cu) added rhenium catalysts. The ratio of the metal and Re were 1:4 (Re contents were 80 mol%) for all catalysts. Although the activity of cobalt and rhenium metals were very low (3 and 108 ␮mol h−1 g−1 ), the addition of cobalt to rhenium increased the activity (492 ␮mol h−1 g−1 for Co-Re4 catalyst). The addition of iron was also effective, however, the rate over Fe-Re4 catalyst (275 ␮mol h−1 g−1 ) was lower than that over Co-Re4 catalyst. Interestingly, the addition of chromium and nickel did not improve the activity much (119 and 143 ␮mol h−1 g−1 for Cr-Re4 and Ni-Re4 , respectively). Only the first two exceeded the activity of Re-Re3 N (Re3 N after 5 h reaction). The addition of copper (60 ␮mol h−1 g−1 ) was even less active than Re metal (108 ␮mol h−1 g−1 ), Cu seems to behave as a simple diluent. Then the effects of the second metal amount were investigated. Fig. 3 shows ammonia synthesis rates on the three mixed Re catalysts (Co-Re, Fe-Re, and Ni-Re) as a function of Re content. For Co-Re system,

322

R. Kojima, K.-i. Aika / Applied Catalysis A: General 209 (2001) 317–325

Fig. 3. Ammonia synthesis rates at 623 K under 0.1 MPa over Re catalysts as a function of the amount of rhenium in binary systems: (䉫) Re; (䊐) Re-Re3 N; (䉬) Co-Re; (䉱) Fe-Re; (䊉) Ni-Re.

the maximum activity was obtained when Re content was 80% (Co-Re4 catalyst). A similar trend was found for Fe-Re system. Contrary to these two systems, the activity of Ni-Re systems did not change between 75 and 90% Re. Apparent activation energies over these catalysts were about 14 kcal mol−1 , these value were on the same level with Fe-K2 O-Al2 O3 or Co3 Mo3 N catalysts [11]. Fig. 4 shows XRD patterns of Co-Re4 and Ni-Re4 catalysts. The patterns did not change after 5 h ammo-

nia synthesis run. For Co-Re4 catalyst, rhenium and cobalt metal were observed, and no bimetallic phase consisting of these two metals was observed. However, the pattern of Co-Re4 catalyst was similar (not exactly same) to that of Re3 N after 5 h ammonia synthesis, which was a mixture of nitride and metal (see Fig. 2(b)). Under the catalyst preparation condition at 973 K, nitride could not be formed (see Fig. 2(c)). By the addition of cobalt, some kind of rhenium nitride phase may be formed partly under NH3 flow even at 973 K. The formation of rhenium “nitride” may be concerned with the high activity of Co-Re4 catalyst. On the other hand, for Ni-Re4 catalyst, only rhenium was observed, and nickel and the bimetallic phase were not observed. This pattern was similar to that of “Re + Re3 N”, but the wide peak of Re3 N was smaller than in the case of Co-Re4 catalyst. By the addition of nickel, formation of nitride phase seems to be more difficult than in the case of cobalt addition under NH3 flow at 973 K. In spite of the high activity of Co-Re or Fe-Re catalysts, it is to be noted that the surface areas of these catalysts were too low to be measured by BET method (measurement limit is about 0.5 m2 g−1 for the used apparatus). H2 or CO chemisorption measurement was even impossible, then, TOF cannot be calculated. Ammonia synthesis over Re (or Fe, Mo) is known as surface structure-sensitive reaction [4]. From the studies using Re single crystals reported by Somorjai et al., Re(1 1 2 1) is the most active surface. However, the surface structure of the practical catalysts cannot be observed. Thus, even if the authors are successful to measure the surface area and H2 chemisorption, the real TOF cannot be calculated. However, the authors found that XRD results gave important information for our new catalysts. Thus, the authors focussed on XRD for the identification of product phase during the preparation method of new active catalyst. An interesting effect of cobalt addition is to stabilize rhenium nitride formation which is concerned with the activity increase. 3.4. High pressure ammonia synthesis and kinetic analysis

Fig. 4. XRD patterns of Co-Re4 and Ni-Re4 catalysts: (a) Co-Re4 ; (b) Ni-Re4 .

Fig. 5 shows ammonia synthesis rates at 623 K as a function of the total pressure over 0.4 g of Re metal, Re-Re3 N (Re3 N after 5 h reaction, thus, it was

R. Kojima, K.-i. Aika / Applied Catalysis A: General 209 (2001) 317–325

323

Fig. 5. Ammonia synthesis rates at 623 K over Re catalysts as a function of total pressure: (䉫) Re; (䊐) Re-Re3 N; (䉬) Co-Re4 ; (䊉) Ni-Re4 .

Fig. 6. Relation between NH3 concentration at the outlet of the reactor (y0 , mol%) and total flow rate (q, l h−1 ) at 623 K under 3.1 MPa of N2 +3H2 on rhenium catalysts: (䉫) Re; (䊐) Re-Re3 N; (䉬) Co-Re4 ; (䊉) Ni-Re4 .

a mixture of metal and nitride), Co-Re4 catalyst, and Ni-Re4 catalyst. The catalyst was once passivated (treated by dilute oxygen), then reactivated with the reaction gas at 623 K for 2 h. The rates (120, 182, 600, and 143 ␮mol h−1 g−1 over Re, Re-Re3 N, Co-Re4 , and Ni-Re4 , respectively) did not decrease after the reactivation, and also XRD pattern did not change. The rate was increased with increasing the reaction pressure. Especially, the rate over Co-Re4 catalyst at 623 K under 3.1 MPa was 2372 ␮mol h−1 g−1 , this value was higher than those over Fe-K2 O-Al2 O3 and Co3 Mo3 N (2010 and 2113 ␮mol h−1 g−1 at 623 K under 3.1 MPa, respectively) [11]. The activity of Co-Re4 catalyst was surprisingly high considering the low surface area (below 0.5 m2 g−1 ) of this catalyst (BET surface areas of Fe-K2 O-Al2 O3 and Co3 Mo3 N were about 14−20 m2 g−1 ). The rates over Re (480 ␮mol h−1 g−1 ) and Re-Re3 N (526 ␮mol h−1 g−1 ) were on the same level, and the rate over Ni-Re4 (415 ␮mol h−1 g−1 ) was lower than those over Re and Re-Re3 N at 623 K under 3.1 MPa. However, above 1.1 MPa, the increase of the rate was somewhat smaller, especially for Re, Re-Re3 N, and Ni-Re4 . It is guessed that this catalytic behavior is related with the active site retardation. Thus, the kinetic analysis was performed over these catalysts.

The results of the kinetic analyses are shown in Figs. 6–8 (for NH3 , H2 , N2 orders, respectively). The values of the reaction order for ammonia synthesis on Re, Re-Re3 N, Co-Re4 , and Ni-Re4 at 623 K under 3.1 MPa are shown in Table 3. Judging from the

Fig. 7. Dependence of reaction rate (C) on hydrogen pressure PH2 (atm) at 623 K under 3.1 MPa and total flow rate of 60 ml min−1 on rhenium catalysts: (䉫) Re; (䊐) Re-Re3 N; (䉬) Co-Re4 ; (䊉) Ni-Re4 .

324

R. Kojima, K.-i. Aika / Applied Catalysis A: General 209 (2001) 317–325

Fig. 8. Dependence of reaction rate (C) on nitrogen pressure PN2 (atm) at 623 K under 3.1 MPa and total flow rate of 60 ml min−1 on rhenium catalysts: (䉫) Re; (䊐) Re-Re3 N; (䉬) Co-Re4 ; (䊉) Ni-Re4 .

almost linear plots of Figs. 6–8, we conclude that these analyses were proper. The values of NH3 order were calculated from 1 − m (m is the inverse of the slope of plots of Fig. 6). The values of H2 order and N2 order were calculated from the slope of plots of Figs. 7 and 8, respectively. The reaction orders on Re and Re-Re3 N were compared, these values were similar to each other. Under high pressure operation with high ammonia production, Re metal must turn more to the nitride, which may be the reason why the reaction order was not different in the two samples. For Re metal, the N2 order (0.92) was almost 1; thus the rate-determining step on Re was the activation of dinitrogen, this feature was the same for Fe, Ru, and Mo catalysts. The NH3 order (−0.82) was negative, which was similar to Fe and Mo catalysts [15]. The surface of Re catalyst was retarded by the product ammonia. Table 3 Reaction order for ammonia synthesis on Re, Re-Re3 N, Co-Re4 , and Ni-Re4 at 623 K under 3.1 MPa Catalyst

NH3 order

H2 order

N2 order

Re Re + Re3 N Co-Re4 Ni-Re4

−0.82 −0.90 −0.54 −1.23

−0.23 −0.24 −0.70 −0.28

0.92 0.97 0.91 0.91

The H2 orders for Fe and Mo catalysts were generally positive; however, the H2 order for Re (−0.23) was slightly negative. It is interesting to find that the surfaces of Re catalyst was retarded by the reactant hydrogen, but weakly. It is well known that the surface of Ru catalysts were strongly retarded by hydrogen. Thus, the Re catalyst was has intermediate nature between molybdenum and ruthenium catalysts (like the periodicity) as far as hydrogen retardation is concerned. Generally, Ru catalyst does not give high productivity under high pressure because of the strong hydrogen poisoning. However, the relieving of hydrogen poisoning was possible if one uses CeO2 as a supporter [16]. Then, the rate can be increased under high pressure because of the less hydrogen poisoning. The catalytic behavior of Re was thus very similar to Ru/CeO2 catalyst. For Co-Re4 catalyst, the N2 order (0.91) was also almost 1; thus the rate-determining step on this catalyst was the activation of dinitrogen. The NH3 order (−0.54) was less negative and the H2 order (−0.70) was more negative than in the case of Re. The surface of Co-Re4 catalyst is poisoned by both the product ammonia and the reactant hydrogen. By the addition of cobalt, the degree of ammonia poisoning was weakened, but at the same time the degree of hydrogen poisoning was strengthened. However, for Ni-Re4 catalyst, the NH3 order (−1.23) was more negative than Re, and H2 order and N2 order were almost the same as those of Re. Addition of nickel was not effective from the viewpoint of kinetics.

4. Conclusions Rhenium nitride was found to be quite active for ammonia synthesis in spite of its low surface area, but it was deactivated because of decomposition of nitride to the mixture of metal and nitride. The activity of this mixture was still higher than the metal and was stable at 623 K, suggesting more effectiveness of nitride surface. Then, rhenium “metal” was used as basic material to develop new catalysts. Addition of cobalt was very effective, and the optimized ratio of Co and Re was 1:4 (Re 80 mol%). This catalyst consisted of a mixture of rhenium nitride, rhenium metal and cobalt metal, although nitride cannot be formed under this

R. Kojima, K.-i. Aika / Applied Catalysis A: General 209 (2001) 317–325

catalyst preparation condition without the addition of cobalt. The ammonia synthesis rate on the catalyst was increased with increasing the reaction pressure (2372 ␮mol h−1 g−1 at 623 K under 3.1 MPa), although the surface area was too low to be measured by BET method. Kinetic analysis revealed that the rhenium containing catalyst has an intermediate nature between Mo and Ru catalysts as far as the hydrogen and nitrogen (ammonia) retardation are concerned. Rhenium-containing catalyst must be developed more, because the activity is quite high, although the surface areas are extremely low. References [1] S.M. Augustine, W.M.H. Sachtler, J. Catal. 116 (1989) 184. [2] P. Clark, B. Dhandapani, S.T. Oyama, Appl. Catal. A 184 (1999) L175.

325

[3] N.D. Spencer, G.A. Somorjai, J. Catal. 78 (1982) 142. [4] M. Asscher, J. Carrazza, M.M. Khan, K.B. Lewis, G.A. Somorjai, J. Catal. 98 (1986) 277. [5] M. Asscher, G.A. Somorjai, Surf. Sci. 143 (1984) L389. [6] G. Haase, M. Asscher, Surf. Sci. 191 (1987) 75. [7] A. Ozaki, K. Aika, H. Hori, Bull. Chem. Soc. Jpn. 44 (1971) 3216. [8] K. Aika, A. Ozaki, J. Catal. 14 (1969) 311. [9] S.T. Oyama, Catal. Today 15 (1992) 179. [10] S. Ramanathan, C.C. Yu, S.T. Oyama, J. Catal. 173 (1998) 10. [11] R. Kojima, K. Aika, Chem. Lett. 5 (2000) 514. [12] J.G. Choi, R.L. Curl, L.T. Thompson, J. Catal. 146 (1994) 218. [13] Z. Zhong, K. Aika, J. Catal. 173 (1998) 535. [14] K. Aika, A. Ohya, A. Ozaki, Y. Inoue, I. Yasumori, J. Catal. 92 (1985) 305. [15] K. Aika, K. Tamaru, in: A. Nielsen (Ed.), Ammonia, Catalysis and Manufacture, Springer, Berlin, 1995, p. 103. [16] K. Aika, Y. Niwa, Science and Technology in Catalysis, Vol. 49, Kodansha, Tokyo, 1998, p. 327.