Ind. Eng. Chem. Prod. Res. Dev. 1904, 2 3 , 41-44
41
Liquid-Phase Hydrogenation with Nickel Supported on Modified Alumina-Aluminum Phosphate George Marcelln,' Roger
F. Vogel, and Harold E. SwHl
Gulf Research & Development Company, Pksburgh, Pennsylvanla 15230
The llquid-phase hydrogenation of ck -24hyC2hexenal was performed with a number of nickel catalysts supported on modified alumina-aluminum phosphate (AAP). The AAP was modified by inclusion of a second cation during
preparation, which resulted in supports with easily controllable surface properties. The addition of even small amounts of Mg, Ca, or Zn was found to affect the surface area and pore size distribution of the support greatly. When these supports were compounded with nickel they yielded catalysts with high surface areas and large pores. Their activities for the reaction of interest were far superior to that obtained with a reference commercial catalyst.
Introduction Catalytic hydrogenation of liquid feedstreams is a process of considerable importance to the chemical industry. These reactions are typically carried out using supported metal catalysts with nickel being widely used as the hydrogenating agent. Because these reactions are conducted in the liquid phase, and liquid-liquid diffusion coefficients are small, the activity of most Catalysts is partly governed by the diffusivities of the reactants into the catalyst pores. In order to overcome the diffusion limitations, most catalysts designed for liquid-phase reactions use large-pore supports. Since most commercial large-pore supports are prepared by "destructive methods", such as selective sintering of the small pores, typical catalysts have large average pore sizes but reduced surface area (Newsome et al., 1960). For diffusion-limited reactions, it is the accessible surface area which is important for activity. Thus, it is important to maximize both the surface area and the average pore size in order to achieve the highest possible catalytic activity. Previous reports have shown the utility of the composite alumina-aluminum phosphate (AAP) as a support for nickel in liquid-phase hydrogenations (Campelo et al., 1982a,b; Marcelin et al., 1983b). AAP is easily prepared and may be considered similar to silica-alumina, since aluminum phosphate and silica are very similar structurally. In aluminum orthophosphate, APO,, the PO, groups share electrons with the aluminum atom to form a threedimensionally bonded network in which the coordination of the aluminum atom is tetrahedral like that of the phosphorus atom. This AP04network is thus analogous to a Si(Si0,) network and shows polymorphism analogous to SiOz,with high- and low-quartz, tridymite, cristobalite, and amorphous forms. AAP also exhibits variable pore size and surface area properties depending on its stoichiometry. Within certain stoichiometric compositions, AAP can exhibit both high surface area and large porosity. Recent reports have shown that modifications of AAP by the inclusion of additional cations, such as Mg, during preparation can lead to supports with even higher surface area and larger pores (Marcelin et al., 1982,1983a; Vogel and Marcelin, 1983). Due to their high surface areas and porosities, modified AAP supports should be ideally suited to applications involving reactions operating in the diffusion-controlled regime. This report details our research into the preparation and evaluation of new family of catalysts based on modified alumina-aluminum phosphates. The catalysts evaluated consisted of ternary modifications of AAP as a Ol96-432lf 84f l223-OO4l$Ol.5O/O
support for nickel in the liquid phase hydrogenation of 2-ethyl-2-hexenal. This work was limited to evaluating modifiers in the periodic group 11, specifically magnesium, calcium, and zinc. Experimental Section Catalyst Preparation. The preparation of these catalysts has been described previously (Marcelin et al., 1982; 1983a). Nitrate salts of the metals and phosphoric acid were dissolved in a common aqueous solution. This solution was slowly added, along with a 14% aqueous solution of ammonia, to a well-stirred container at constant pH, typically 9. The resulting slurry was filtered and washed with water. The wet filter cake was then thoroughly mix-mulled with nickel carbonate. If extrudates were desired, methyl cellulose was admixed at a level of 1 wt % of the solids present on a calcined basis and the mixture extruded through a 1/16-in.die using a Loomis ram extruder. All catalysts were dried at 120 "C and calcined at 350 "C prior to evaluation. The preparation of the silica-supported catalyst followed a similar procedure, by use of a commercially prepared silica gel (Davison Grade 59) sized to -100 mesh. All catalysts contained 20 wt % nickel after hydrogen reduction. The modified alumina-aluminum phosphate supports me written in stoichiometric shorthand. For example, the notation Mg8AAP is used to denote a support with the composition Mg0~8AlzO3*APO4 and 4Ca13AlOAP represents 4Ca0~13Alz03~10AP0,. Notations such as MgAAP denote generic magnesia-alumina-aluminum phosphate, i.e., an entire family of supports. Characterization. The surface area and pore characteristics of the unreduced catalysts were obtained by nitrogen sorption at -196 "C on an Aminco Adsorptomat, after pretreatment a t 200 "C in vacuum for 2 h. Singlepoint surface areas were obtained with a Micromeritics surface area analyzer. A Phillips 3100 automatic powder diffractometer and a DuPont 1090 thermal analyzer were used to obtain X-ray diffraction and differential thermal analysis data. Catalytic Activity. Finished catalysts were evaluated for the liquid-phase hydrogenation of 2-ethyl-2-hexenalby use of a fixed-bed continuous flow reactor consisting of a 13 mm i.d. stainless steel tube into which 5 mL of 20-40 mesh granular catalyst was loaded. Following hydrogen reduction a t 400 "C for 18 h, the catalytic activity was measured by using a feed consisting of 15 wt % cis-2ethyl-2-hexenalin 2-ethyl-1-hexanol. This dilution helped to prevent temperature changes due to reaction exothermicity. Feed rates were 33 g/h and 4 L/h of the liquid 0 1984 American Chemical Soclety
42
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984
Table I. Physical Properties and Activity of 20% Nickel Catalysts Supported on MgAAP catalyst support stoichiometry MgO A1,0, AlPO, median pore radius, A pore volume, cm3/g BET surface area, m Z / g pore radius distribution, 70 vol 200-300 ‘4 100-200 50-100 40-50 30-40 20-30 < 20 conversion, 70 nickel-based conversion, 7c
-
A
B
C
D
E
F
G
1 1 1 120 0.64 220
2 7 4 140 0.74 220
87 0.72 240
130 0.43 160
88 0.58 240
130 0.61 210
130 0.58 220
16 41 22 4 4 7 6 63 143
21 47 18 3 3 4
10 33 29 7 9 10 2 47 10 6
27 31 14 4 6 9 9 40 93
17 28 20 0 19 10 6 28 62
26 37 16 3 4 6 8 69 165
23 37 17 4 4 7 8 75 178
1 60 130
feed and hydrogen, respectively, at 100 psig. Measurements were obtained for a period of 5 h at 50 “C. Temperatures were monitored with a series of three iron-constantan thermocouples embedded at three points in the catalyst bed and were observed not to vary significantly during the course of the reaction. The product stream was sampled hourly and analyzed in an off-line gas chromatograph. No appreciable change in activity was observed over the reaction period for any of the catalysts studied. Long-Term Activity. Aging studies were performed using 1/16-in.extrudates at more rigorous conditions than the standard activity tests in order to closely approximate realistic process conditions. All reaction conditions were the same as for the screening runs, except the temperature was maintained at 90 OC and the reaction was allowed to proceed for 100-120 h with samples taken hourly during the first 4 h and every 4 h thereafter.
Results and Discussion A number of catalysts were prepared using magnesia as a modifier for AAP (MgAAP). The support stoichiometry was varied and all catalysts contained 20% metal after reduction. Table I lists the physical properties and activity of the MgAAP-supported catalysts. The activity results are presented in two formats: the conventional percent conversion, which is based on a constant volume of catalyst loaded and the term “nickel-basedconversion”,which takes into account the nickel loading and catalyst density and represents the percent conversion per gram of nickel. The single value of activity reported in the table represents the measured activity after 5 h on-stream. Catalysts based on MgAAP supports all exhibit reasonably high surface area and porosity, showing little change in the surface area, except for catalysts with high phosphate content (catalyst D). The lower surface area of the high phosphate catalyst is not surprising, since it follows the pattern which was observed for other AAPbased materials (Marcelin et al., 1983a,b). What is surprising is the lack of change in surface area when the support stoichiometry is changed in the other examples. It was previously observed that the MgAAP supports exhibit a variable surface area as the stoichiometry is changed and highest surface areas were obtained for supports high in alumina. In contrast, when nickel is included in the catalyst preparation, only minor changes in the BET surface areas are observed with increasing alumina content. For example, compare the surface areas of catalysts C and E. They have the same surface area even though the alumina content varies considerably. When nickel car-
Table 11. Physical Properties of Nickel Oxide Prepared by Calcination of Nickel Carbonate median pore radius, A pore volume, cm3/g BET surface area, m’/g pore radius distribution, 5% vol 200-300 A 100-200 50-1 00 40-50 30-40 20-30 < 20
31 0.17 106 2.7 4.5 8.8 9.7 28.0 34.2 12.1
Table 111. Physical Properties and Activity of 20% Nickel Catalysts Supported on CaAAP catalyst support stoichiometry CaO Al,O, AlPO, median pore radius, .4 pore volume, cm3/g BET surface area, m’/g pore radius distribution, % vol 200-300 A 100-200 50-100 40-50 30-40 20-30 < 20 conversion, % nickel-based conversion, 7%
H
I
J
1 1 1 160 0.73 190
1 1 8 170 0.62 170
4 13 10 160 0.80 220
27 49 14 2 2 3 3 66 141
39 34 12 3 4 6 2 32 81
31 44 13 2 2 4 4 59 156
bonate is calcined at 350 OC, the resulting nickel oxide exhibits a significant surface area and porosity, as seen in Table 11. It is possible that this leveling effect on the surface properties of Ni/MgAAF’ is caused by the inclusion of nickel in the form of carbonate as a precursor. Apart from the high surface area, these catalysts also exhibit large median pore radii and reasonably large pore volumes. In all cases, these materials have a high concentration of large pores. Tables I11 and IV show data for a series of similar catalysts prepared using Ca- and Zn-modified AAP. Again the large-pore, high surface area characteristic of these materials is observed. As expected, the high phosphate version of these catalysts showed the lowest surface area. It is interesting to compare the surface properties of the modified AAP catalysts with those of catalysts which are commonly used in liquid-phase reactions. Table V gives comparable data for three such catalysts: Ni-3266, a
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984 43 Table IV. Physical Properties and Activity of 20% Nickel Catalysts Supported on ZnAAP catalyst support stoichiometry ZnO A1203 AlPO, median pore radius, A pore volume, cm3/g BET surface area, m’/g pore radius distribution, % vol 200-300 A 100-200 50-100 40-50 3 0-4 0 20-30 < 20 conversion, % nickel-based conversion, %
K
L
M
1 1 1 120 0.65 200
1 1 8 100 0.39 170
4 13 10 180 0.67 180
22 39 20 4 4 7 4 52 110
22 28 16 5 6 12 11 26 65
43 34 9 3 2 4 5 57 149
ZnA8AP
Table V. Physical Properties and Activity of Other “Liquid-Phase’’ Catalysts catalyst N 0 P Ni-3266 NiIA2AP Ni/SiO, nickel content,
50
20
20
55
100
100
0.3
0.64
0.85
125
225
260
9 18 27 12 13 14 6 50 32
16 35 23 6 7 9 4 40 100
3 53 35 2 2 2 4 29 72
%
median pore radius, A pore volume, cm’lg surface area, m’/g pore radius distribution, % vol 200-300 A 100-200 50-100 40-50 30-40 20-30 Ca > Mg > Zn. This appears to be independent of stoichiometry as evidenced by the two MgAAP samples. The surface area values of various supports heated to different temperatures are shown in Table VI. Calcination to temperatures higher than the observed DTA exotherm resulted in a sudden loss of surface area in all cases, and XRD patterns obtained for the calcined materials showed the decrease in surface area to be accompanied by a cor-
Ind. Eng. Chem. Rod. Res. Dev. 1984, 23,44-50
44
Table VI. Thermal Stability of Modified AAP Supports. Single-Point Surface Area (m'/g) after Calcination at T ("C) 500 "C
800 "C
1000
compn 4Mg13AlOAP MgA8AP Mg8AAP A2AP 4Ca13AlOAP ZnA8AP
161 120 340 97 226 109
153 100 225 96 170
0 0
0
-
"C
1100 "C
-
-
95 91
-
-
0
-
mercial catalyst. Because this catalyst was supplied in the form of 1/16-in.extrudates, the test catalyst was formed to the same dimensions. Figure 2 shows the performance of these two catalysts over the course of the reaction. After the initial lining-out period, both catalysts showed little evidence of deactivation. In terms of average conversion per reactor load, the MgAAP-supported catalyst proved somewhat better, yielding an average conversion of 63% as compared with 43% for Ni-3266. Although the length of the test was not sufficient to make predictions as to the expected long-term life of the catalysts, it served as an indication that the experimental catalyst may be commercially viable. Conclusions Modifications of AAP by inclusion of the additional cations Mg, Ca, and Zn lead to supports with high prosities and high surface areas. These supports in combination with nickel result in highly efficient liquid-phase hydrogenation catalysts.
Y+3266 -- AR-43'
1
20
40
80
BO
Acknowledgment
120
100
Time , HI
Figure 2. Comparison of 20% Ni on 4Mg13AlOAP catalyst with commercial reference material. Condition detailed in text.
responding increase in crystallinity. From this, it is clear that the nature of the DTA exotherms is almost certainly the onset of phase separation, leading to crystallization and pore collapse with loss of surface area Studies of the effect of calcination temperature on the surface area of MgAAP have shown that high levels of modifier typically result in less thermally stable materials (Marcelin et al., 1983a). It should be pointed out, however, that although for the materials evaluated, the surface area and porosity properties were not modifier dependent; all the modifiers evaluated belonged to group 11. Those from other groups may indeed have different effecta on the surface properties. The performance of these materials for longer term activity was compared with that of the reference com-
The authors thank J. A. Tabacek, N. A. White, and R. H. Hazlett for their work in the preparation of the catalysts, R. L. Slagle, W. R. Grinder, N. C. George, and V. S. Sikora for performing the activity runs, and D. M. Regent for thermal analysis. Registry No. Magnesium, 7439-95-4; alumina, 1344-28-1; aluminum phosphate, 7784-30-7; nickel, 7440-02-0; calcium, 7440-70-2; zinc, 7440-66-6; cis-2-ethyl-2-hexena1,88288-45-3.
Literature Cited Campelo, J. M.; Garcia, A.; Luna, D.; Marinas, J. M. Appl. Catel. 1982a, 3,
315-325. Campelo, J. M.; Garcia, A.; Luna, D.; Marinas, J. M. Gazz. Chim. Its/. 1982b, 772, 221-225. Marcelin, G.;Vogei, R. F.; Swift, H. E. US. Patent 4365095, 1982. Marcelin, Q.; Vogel, R. F.; Kehi, W. L. "Preparation of Catalysts. 111". Elsevbr Science Publishers 6. V.; Amsterdam, l983a; pp 169-179. Marcelin, Q.; Vogei, R. F.; Swift, H. E. J . Catal. lB83b, 83, 42-49. Newme, J. W.; Heiser, H. W.; Ruseell, A. S.; Stumpf, H. C. "Alumina Properties"; Aiumlnum Company of America: Pittsburgh, PA, 1960. Vogel, R. F.;Marcelin G. US. Patent 4376067, 1983.
Received for review May 25, 1983 Accepted September 9, 1983
Reduction of Nitrobenzene to Aniline Jaime Wknlak' and Miriam Klein oepertment of Chemlcel E
w
~B , " Unhrwelty of the Negev, Beer-Sheva, Israel
Nitrobenzene was hydrogenated to aniline in the liquid phase, using Raney nickel, ruthenium on carbon, rhodium on carbon, rhodium on alumina, and nickel on Inert carrier catalysts. Raney nickel catalysis is a complex process that goes through azoxybenzene and arobenrene lntermedlates. All other catalysts yield aniline directly. Within the experimental range the reaction stopped at the aniline step; no cyclohexylamine was formed. Energy of actlvation for Raney nickel was 14.1 kcal/mol, and for palladium on carbon It was 9.7 kcal/mol. No plausible mechanism was found for reduction with Raney nickel.
Introduction Aniline is an important chemical in the m a n d a d w e of dyes and d-imtion accelerators and It is usu&y manufactured by fiereductionof nitrobenzene CGH5N02 + 3H2
+
CeH5NH2 + 2H20
0196-4321/84/1223-0044$01.50/0
From the data on AGfo of Dean (1973) it is possible to estimate the value of the equilibrium constant of the reaction a~ 1.53 X 1084 at 298 K indicating that the reaction may be considered irreversible for all practical purposes. Gharda and Sliepcevich (1960) studied the hydrogenation of nitrobenzene in the vapor phase at 400 OC over a 0 1984 American Chemical Society