Hydrogenolysis of Phenol over Supported

Eun-Jae Shin† and Mark A. Keane*. Department of Chemical Engineering, The University of Leeds, Leeds LS2 9JT, United Kingdom. The gas-phase ...
4 downloads 0 Views 114KB Size
Ind. Eng. Chem. Res. 2000, 39, 883-892

883

Gas-Phase Hydrogenation/Hydrogenolysis of Phenol over Supported Nickel Catalysts Eun-Jae Shin† and Mark A. Keane* Department of Chemical Engineering, The University of Leeds, Leeds LS2 9JT, United Kingdom

The gas-phase hydrogenation/hydrogenolysis of alcoholic solutions of phenol between 423 and 573 K has been studied using a Y zeolite-supported nickel catalyst (2.2% w/w Ni) and Ni/SiO2 catalysts (1.5-20.3% w/w Ni). This is a viable means of treating concentrated phenol streams to generate recyclable raw material. Phenol hydrogenation proceeded in a stepwise fashion with cyclohexanone as a reactive intermediate while a combination of hydrogenolysis and hydrogenation yielded cyclohexane. Hydrogenolysis to benzene is favored by high nickel loadings and elevated temperatures. A catalytic hydrogen treatment of cyclohexanone and cyclohexanol helped to establish the overall reaction network/mechanism. The possible role of thermodynamic limitations is considered and structure sensitivity is addressed; reaction data are subjected to a pseudo-first-order kinetic treatment. Hydrogen temperature-programmed desorption (H2-TPD) has revealed the existence of different forms of surface hydrogen. Selectivity is interpreted on the basis of the H2-TPD profiles and the possible phenol/catalyst interactions. The zeolite sample only catalyzed (via the surface Bro¨nsted acidity) anisole formation in the presence of methanol, but this was suppressed when hexanol was used; the zeolite then promoted hydrogenolysis. The zeolite, however, deactivated and this was not reversed by heating in hydrogen. The results of the hydrogen treatment of aqueous rather than alcoholic phenol solutions are presented, where a switch from methanol to water was accompanied by a move from highly selective hydrogenolysis to highly selective hydrogenation. Introduction The commercial significance of the reaction of phenol with hydrogen is 2-fold, that is, as a synthesis route and as a pollution abatement methodology. On one hand, the selective hydrogenation of phenol yields cyclohexanone, a key raw material in the production of both caprolactam for nylon 6 and adipic acid for nylon 66.1 Phenol is, on the other hand, an established environmental toxin2 and phenolic waste originates from a variety of industrial sources including oil refineries, petrochemical units, polymeric resin manufacturing, and plastic units.3 Thermal and/or catalytic oxidations are at present the widely favored pollution abatement techniques for treatment/disposal of such organic waste.4 However, effective incineration is energetically demanding and an incomplete combustion can generate highly toxic products.5 To minimize any negative environmental effects and conserve resources, the first abatement option that should be considered is the recovery and reuse of raw material. Catalytic hydrogenation/hydrogenolysis is now emerging as a viable alternative to destructive incineration whereby the hazardous substances are transformed into useful products.6-8 This approach has relatively low-energy requirements and minimal toxic emissions, in short, a “best practicable environmental option”. The hydrogenation of phenol in the gas9-16 and liquid phase17-19 has been reported for a range of palladium-9,11,12,14,17-21 and platinum-12,13,16 based catalyst systems where the emphasis was placed firmly on * To whom correspondence should be addressed: Tel.: +44 113 2332428. Fax: +44 113 2332405. E-mail: chemaak@ leeds.ac.uk. † Present address: National Renewable Energy Laboratory, Golden, CO 80401-3393.

achieving high cyclohexanone selectivities. The use of nickel catalysts, by comparison, has received far less attention.10,22,23 In this paper, the effects of nickel loading (supported on silica) on gas-phase activity/ selectivity are reported and compared with the action of a nickel-exchanged Y zeolite. We focus on the potential of catalytic hydrogen treatment as a means of pollution abatement/waste minimization and examine the conversion of methanol, water, and mixed watermethanol-based phenol solutions. The treatment of a purely aqueous phenol feedstock is of particular significance because phenol, as a pollutant, typically arises in high concentrations in aqueous media.24,25 Experimental Section Catalyst Preparation, Activation, and Characterization. Five silica-supported (Cab-O-Sil 5 M, 194 m2 g-1) nickel catalysts with metal loadings in the range 1.5-20.3% w/w Ni were prepared by homogeneous precipitation/deposition,26 and the catalytic action of each was compared with a 2.2% w/w Ni/Na-Y (Linde, Na58(AlO2)58(SiO2)134(H2O)260) zeolite prepared by ion exchange.27 The nickel content (accurate to within (2%) was measured by atomic absorption spectrophotometry and the water content by thermogravimetry as described elsewhere.28 The hydrated samples, sieved in the 150-200-µm mesh range, were activated by direct heating in a 100 cm3 min-1 stream of dry hydrogen at a fixed rate of 5 K min-1 to a final temperature of 673 K which was maintained for 18 h; the reactor temperature was constant to within (1 K. Temperature-programmed reduction (TPR) experiments were performed ex situ from the catalytic reactor. A known weight (≈200 mg) of the hydrated precursor was activated by heating at 5 K min-1 in a 100 cm3 min-1 flow of 6% v/v H2 in N2 to 1073 K, and hydrogen

10.1021/ie990643r CCC: $19.00 © 2000 American Chemical Society Published on Web 03/17/2000

884

Ind. Eng. Chem. Res., Vol. 39, No. 4, 2000

consumption was measured using a thermal conductivity detector (TCD). A secondary ion mass spectrometric analysis (SIMS, VG ESCALAB) of each activated catalyst (pressed into indium foil) revealed only the presence of nickel and silica; there was no evidence of even trace amounts of impurities (or potential promoters) on the surface. Surface nickel dispersion/size/morphology was studied using a combination of CO chemisorption, X-ray diffraction (XRD) line broadening, and transmission electron miscroscopy (TEM) as described in previous reports:26,29 the average nickel particle diameter inferred from each technique was in reasonable agreement and did not diverge by more than (15%. Chemisorption measurements were made directly after TPR by pulsing 50-µL aliquots of CO in a 40 cm3 min-1 stream of dry helium, monitoring the uptake at 273 K by TCD. XRD line broadening (JEOL JDX-85 diffractometer) analysis focused on the nickel line at a 2θ angle of 52.2°. TEM analysis of the freshly activated and used catalysts was performed using a Philips CM20 transmission electron microscope operated at an accelerating voltage of 200 keV; specimen preparation involved ultrasound dispersion in isobutyl alcohol with deposition on a holey carbon grid. The H2-TPD of the activated catalysts (≈0.1 g) was performed in a dual stainless steel ultrahigh vacuum chamber: full details of the equipment and procedure are available elsewhere.30 In each case the samples were evacuated (10-6 Torr) for at least 10 h and then heated to 1073 K at 5 K min-1; the temperature and mass signals (PGA100, Leybold) were continually monitored. A preadsorption of hydrogen prior to TPD, while resulting in a greater concentration of hydrogen desorbed, had no effect on the temperature characteristics of hydrogen loss. The surface Bro¨nsted acidity generated upon reduction of the divalent nickel ions in the zeolite sample was characterized by IR as outlined elsewhere;27 the band at 395 cm-1 was used as an index to probe changes in zeolite crystallinity.31 Catalytic Procedure. All the catalytic reactions were carried out under atmospheric pressure, in situ immediately after the activation step, in a fixed-bed glass reactor (ratio of catalyst particle to reactor diameter ) 90) over the temperature range 423 e T e 573 K. The catalytic reactor approximated the plug flow pattern32 and was fully described previously23 but some pertinent details are given below. A Merck-Hitachi LC6000A pump was used to deliver the organic feed at a fixed rate and the vapor was carried through the catalyst bed in a stream of purified hydrogen. Methanolic solutions (where the methanol/reactant mole ratio ) 4) of phenol, cyclohexanone, and cyclohexanol and hexanolic (in the case of Ni/Na-Y), mixed aqueous/ methanolic and purely aqueous solutions of phenol were used as feedstock. The reactant molar feed rate was varied from 4.9 × 10-3 to 2.6 × 10-2 h-1 to test adherence to pseudo-first-order kinetic behavior where hydrogen was maintained in at least 6-fold excess relative to stoichiometric quantities; the overall gas space velocity was maintained at 2400 h-1 and the W/F (ratio of catalyst weight to reactant molar flow rate) was in the range 20-103 g mol-1 h. The catalytic system was found to operate with negligible diffusion retardation of the reaction rate or heat-transport effects; effectiveness factor (η) > 0.99 at 473 K. In a series of blank experiments, the passage of methanolic or aqueous solutions of phenol in a stream of hydrogen through the empty reactor (under standard reaction conditions)

Table 1. Physical Characteristics of the Activated Ni/SiO2 Catalysts sample

% Ni w/w

% H2O w/w

H2/Nia

d,b nm

Ni/SiO2-A Ni/SiO2-B Ni/SiO2-C Ni/SiO2-D Ni/SiO2-E

1.5 6.1 11.9 15.2 20.3

0.6 1.2 2.0 3.6 4.5

1.22 1.42 1.19 1.29 1.25

1.4 1.9 2.5 3.1 3.7

a Mole ratio: hydrogen consumed (during TPR) to nickel content. b Average Ni particle diameter.

did not result in any detectable conversion. At the end of each catalytic run the catalyst was heated in flowing dry hydrogen at 5 K min-1 to 673 K and maintained at this temperature for at least 12 h. Catalyst deactivation was probed by ascending and subsequent descending reaction temperature sequences over the entire temperature interval that was considered. In addition, one standard set of process conditions (phenol feed, T ) 473 K, W/F ) 69 g mol-1 h) was routinely repeated to ensure that the catalyst had not suffered any long-term deactivation. The reactor effluent was frozen in a liquid nitrogen trap for subsequent analysis which was made using an AI Cambridge GC94 chromatograph equipped with a flame ionization detector and employing a DB-1 50 m × 0.20 mm i.d., 0.33 µm capillary column (J&W Scientific); data acquisition and analysis were performed using the JCL 6000 (for Windows) chromatography package. Catalytic action is discussed in this paper in terms of fractional conversion (x), percentage selectivity (S), and percentage yield (Y) where, for the production of cyclohexanone from phenol

xC6H5OH ) SC6H10O(%) )

[C6H5OH]in - [C6H5OH]out [C6H5OH]in [C6H10O]out

[C6H5OH]in - [C6H5OH]out

YC6H10O(%) )

[C6H10O]out [C6H5OH]in

× 100

× 100

(1)

(2)

(3)

and [ ]in and [ ]out denote the concentrations entering and exiting the reactor, respectively. All the reactants were AnalaR grade and were used without further purification. Results and Discussion Conversion of Methanolic Solutions of Phenol over Ni/SiO2. The pertinent physical characteristics of the activated Ni/SiO2 catalysts are given in Table 1 where the samples are labeled according to the nickel content. Nickel/silica prepared by controlled deposition/ precipitation is characterized by a smaller and narrower size distribution than samples prepared by impregnation;26,33,34 the metal particle sizes given in Table 1 are in good agreement with the literature.34 The total hydrogen consumed during TPR exceeded that required for a complete reduction of the nickel content which suggests the presence of spillover hydrogen. Hydrogen spillover is now well-established for supported transition metal catalysts where molecular hydrogen dissociates on the metal into atomic hydrogen which then spills over onto the (typically) oxide support.35 The gas-phase hydrogenation of phenol, cyclohexanone, and cyclohexanol generated a range of products that are listed in

Ind. Eng. Chem. Res., Vol. 39, No. 4, 2000 885 Table 2. Reaction Products Generated during the Hydrogen Treatment of Methanolic Solutions of Phenol, Cyclohexanone, and Cyclohexanol over Ni/SiO2-A and Ni/SiO2-E: 423 e T e 573 K; t Denotes a Product Formed in Trace Quantities, Where S < 2% products feedstock

Ni/SiO2-A

Ni/SiO2-E

phenol cyclohexanone cyclohexanol

cyclohexanone, cyclohexanol, benzene cyclohexanol, phenol, benzene cyclohexanone, phenol, benzenet

cyclohexanone, cyclohexanol, benzene, cyclohexenet, cyclohexane cyclohexanol, phenolt, benzene, cyclohexenet, cyclohexane benzene, cyclohexane

Table 3. Free Energies at 498 K for the Possible Reaction Steps Observed in the Hydrogen Treatment of Phenol reaction

∆G, kJ mol-1

C6H5OH + 2H2 h C6H10O C6H5OH + 3H2 h C6H11OH C6H5OH + H2 h C6H6 + H2O C6H5OH + 3H2 h C6H10 + H2O C6H5OH + 4H2 h C6H12 + H2O C6H10O + H2 h C6H11OH C6H10O h C6H5OH + 2H2 C6H10O h C6H6 + H2O + H2 C6H10O + H2 h C6H10 + H2O C6H10O + 2H2 h C6H12 + H2O C6H11OH h C6H10O + H2 C6H11OH h C6H5OH + 3H2 C6H11OH h C6H6 + H2O + 2H2 C6H11OH h C6H10 + H2O C6H11OH + H2 h C6H12 + H2O C6H6 + 2H2 h C6H10 C6H6 + 3H2 h C6H12 C6H10 + H2 h C6H12

-1.1 -2.0 -29.7 -20.1 -40.8 -0.8 0.9 -28.6 -18.9 -39.8 0.9 1.5 -27.9 -18.4 -38.9 22.7 -26.1 -48.0

Table 2. The lower nickel-loaded silica (Ni/SiO2-A) promoted predominantly hydrogenation and dehydrogenation steps while Ni/SiO2-E exhibited appreciable hydrogenolysis activity. The hydrogenation of phenol can occur in a stepwise fashion yielding cyclohexanone as the partially hydrogenated product and cyclohexanol as the fully hydrogenated product. The generation of cyclohexane and the trace amounts of cyclohexene results from combined hydrodehydroxylation and hydrogenation steps and was only observed for the higher metal-loaded catalysts. The possibility of thermodynamic limitations was considered by estimating the equilibrium constants using the NIST database,36 and the calculated free energies (taking 498 K as a representative temperature) for all the possible reaction steps in the conversion of phenol are given in Table 3. The thermodynamical calculations indicate that the standard free energies of reaction are negative for each hydrogenolysis/ hydrogenation step with the exception of the partial hydrogenation of benzene to cyclohexene. The hydrogenation of benzene over Ni/SiO2 has been reported elsewhere37 to yield cyclohexane as the sole product with no detectable cyclohexene formation. The observed products and thermodynamically favored steps are illustrated in the reaction network given in Figure 1. All hydrogenolysis and the benzene to cyclohexane steps can be considered to be irreversible while the stepwise hydrogenation of phenol may be governed by equilibria. Cyclohexanone can also be generated via hydrogen addition to form cyclohexene-1-ol followed by a tautomerism, but as the authors could not find any thermochemical data for this reaction, it has been omitted from the reaction scheme. The product distribution, in the catalytic step, was essentially time-invariant over each nickel catalyst and at every temperature that was considered; typical time-on-stream profiles in terms of product yield are given in the inset to Figure 2; the

Figure 1. Reaction network (showing only the thermodynamically favorable steps) in the catalytic hydrogen treatment of phenol over Ni/SiO2.

activities and selectivities quoted in this paper represent steady-state values. A prolonged use of each catalyst (up to 500 h-on-stream) did not result in any appreciable loss of activity or modification to the selectivity trends. In marked contrast, a short-term catalyst deactivation has been noted elsewhere9,11,14,15 in the case of palladium-based catalysts. Analysis of the used nickel catalysts did not reveal any significant modification to the nickel particle morphology or size distribution. Variations in reaction temperature had a considerable effect on both catalytic activity and selectivity. The temperature dependence of phenol conversion (at a fixed inlet molar feed rate) is shown in Figure 2a where the fractional conversion is shown to decrease at elevated temperatures. A drop in phenol conversion with increasing temperature has been reported previously11,12,14,15,22 and was attributed to thermodynamic limitations11,22 and/or a temperature-induced desorption of phenol.15 The higher nickel-loaded catalysts served to extend conversion at elevated temperatures, notably in excess of 523 K. Increased reactivity with increasing metal loading has been remarked upon in previous studies involving supported metal catalysts.9,10 The reaction kinetics can be approximated by a pseudo-first-order treatment which can be tested by integrating the design equation for a plug flow reactor. A pseudo-first-order dependence will yield a straight line plot relating fractional conversion (x) to W/F where

ln

(1 -1 x) ) kWF

(4)

and three representative plots are given in Figure 3. A least-squares fit, forced to go through the origin, was

886

Ind. Eng. Chem. Res., Vol. 39, No. 4, 2000

Figure 2. Variation with reaction temperature of the fractional conversion (x) of (a) phenol, (b) cyclohexanone, and (c) cyclohexanol over Ni/SiO2-A (9), Ni/SiO2-C (b), and Ni/SiO2-D (2); W/F ) 69 g mol-1 h. Inset: Product yield from a phenol feedstock as a function of time-on-stream in terms of the formation of cyclohexanol over Ni/SiO2-A at 523 K (9), benzene over Ni/SiO2-C at 523 K (b), and cyclohexane over Ni/SiO2-D at 423 K (2).

Figure 3. Pseudo-first-order relationships for the hydrogen treatment of phenol at 573 K over Ni/SiO2-A (9), Ni/SiO2-C (b), and Ni/SiO2-D (2). Inset: Variation of the rate constant (per unit nickel area) with nickel crystallite diameter.

used to determine the rate constant which increased in magnitude with increasing nickel content. This trend is suggestive of structure sensitivity and this was probed by relating the experimentally determined pseudo-firstorder rate constant to catalyst structure, extracting a specific rate that takes account of the exposed nickel surface area. The variation with nickel particle diameter of the rate constant (per unit nickel area) is illustrated in the inset to Figure 3; an average area of 0.0633 nm2 was taken per surface nickel atom.38 There is a clear increase in rate upon increasing the nickel particle size from ≈1 to 4 nm. The nickel particle sizes consid-

Figure 4. Variation with temperature of reaction selectivity for the conversion of phenol to (a) cyclohexanone, (b) cyclohexanol, and (c) benzene over Ni/SiO2-A (9), Ni/SiO2-C (b), and Ni/SiO2-D (2); W/F ) 69 g mol-1 h. Inset: Selectivity in terms of cyclohexane production over Ni/SiO2-C ([) and Ni/SiO2-D (1).

ered in this study largely fall within the so-called mitohedrical39 region wherein catalytic reactivity can show a critical dependence on morphology. Each catalyst is characterized by a narrow particle size distribution and strong metal/support interactions; the relationship shown in Figure 3 supports the involvement of an ensemble effect where the conversion of phenol is clearly structuresensitive. Selectivity Trends: Conversion of Cyclohexanone and Cyclohexanol. Cyclohexanone and cyclohexanol were used as feed to probe the overall reaction network and examine the (possible) generic effect(s) of varying nickel content on activity/selectivity. The conversion of both oxygenates is illustrated as a function of temperature in Figure 2b,c where the process conditions were identical to those employed for the treatment of phenol (Figure 2a). The fractional conversion of cyclohexanone declined with increasing temperature to pass through a minimum at ≈538 K. At higher temperatures the dehydrogenation reaction predominated, as will become obvious from the selectivity trends, and conversion was elevated once more. The higher nickelloaded samples again consistently delivered higher conversions and the dip in conversion with increasing temperature was barely perceptible in the case of Ni/ SiO2-D and Ni/SiO2-E. The conversion of cyclohexanol, while a more demanding reaction, was again promoted to a greater degree with increasing nickel content. The effect of temperature on product selectivity is illustrated in Figure 4. Taking the lowest nickel loading (Ni/SiO2A), cyclohexanol was by far the predominant product at T e 473 K, but with an increase in temperature cyclohexanone formation was increasingly promoted and was the preferred product at the highest temperatures (>523 K) that were considered. Hydrogenolysis of the -OH substituent to yield benzene was only initiated at

Ind. Eng. Chem. Res., Vol. 39, No. 4, 2000 887 Table 4. Product Selectivity in the Hydrogen Treatment of Phenol at Three Representative Temperatures over Each Ni/SiO2 Sample (See Table 1): W/F ) 69 g h mol-1 S% product

Ni/SiO2-A Ni/SiO2-B Ni/SiO2-C NiSiO2-D Ni/SiO2-E

cyclohexanone cyclohexanol benzene cyclohexene cyclohexane

2 98 0 0 0

T ) 423 K 4 2 96 95 0 0 0 0 0 3

26 64 523 K. Selectivity for the hexanolic feed at representative temperatures is summarized in Table 6. The Ni/Na-Y catalyst also showed strong hydrogenolytic characteristics and did not generate any measurable quantities of cyclohexanol or cyclohexanone. The formation of cyclohexene and cyclohexane is more likely the result of partial and full hydrogenation of benzene immediately following hydrodeoxygenation. Both selectivity (Figure 9b) and conversion (Figure 9c) were strongly time-dependent. As the reaction proceeded, cyclohexene was favored at the direct expense of cyclohexane and benzene; as the catalyst deactivated, the accumulation of coke deposits can be considered to inhibit both desorption of benzene after hydrogenolysis and complete hydrogenation to cyclohexane. The H2-TPD profile for the zeolite sample, shown in Figure 5, is characterized by a sharp peak at 803 K, a broad and less intense peak at 453 K, and illdefined temperature maxima at 573 and 643 K. The only H2-TPD profiles for Ni/Na-Y that the authors could unearth in the literature reveal the presence of both high-50,51 (>900 K) and low-50 (≈700 K) temperature peaks. Both cited reports are lacking in experimental detail and interpretative input which does not allow a direct comparison with the TPD profile generated in this study and an explicit assignment of the peaks to distinct surface hydrogen species is not feasible. However, it is known that hydroxyl groups on a zeolite support can stabilize spillover hydrogen,35 and the Bro¨nsted acidity may well have a bearing on the interplay between hydrogenation/hydrogenolysis. The extent of deactivation, quantified by the parameter R recorded in Table 6, was markedly greater at higher temperatures and exceeded that observed for the formation of anisole. Indeed, the degree of coking/deactivation was so great that activity declined at higher reaction temperatures. Moreover, the initial conversions dropped to less than 0.05 after three reaction cycles, regardless of temperature. Deactivation was largely irreversible and a regeneration of the zeolite catalyst in hydrogen at 673 K did not fully remove the occluded coke and was only partially effective in raising catalytic activity; cyclohexene was the sole product over the regenerated zeolite. Conversion of Aqueous Solutions of Phenol over Ni/SiO2. In practice, phenol as a pollutant arises in high concentrations in aqueous media, and to properly assess the viability of catalytic hydrogen treatment as an abatement methodology, it is necessary to apply it to a “wastewater” mimic. The more active Ni/SiO2-D and Ni/ SiO2-E samples were chosen, in this instance, as model

Figure 10. Effect of water content (as a mole fraction, X) in the feed on reaction selectivity for the hydrogen treatment of phenol over Ni/SiO2-D at 473 K: (2) cyclohexanol; (9) cyclohexanone; (b) benzene; ([) cyclohexane. Table 7. Reaction Selectivity and Hydrodeoxygenationto-Hydrogenation Selectivity Ratio at Representative Temperatures for the Hydrogen Treatment of Purely Methanolic and Purely Aqueous Solutions of Phenol over Ni/SiO2-E S% solvent

T, K

C6H10O

C6H11OH

C6H6

C6H12

SHDO/SH

water methanol water methanol water methanol water methanol water methanol

423 423 473 473 523 523 553 553 573 573

3 23 38 20 28 11 24 3 16 1

97 59 53 27 52 6 34 2 7 0

0 2 9 44 20 80 42 94 77 99

0 16 0 9 0 3 0 1 0 0

0 0.2