study of hdo carbonyl, carboxylic and gua, over sulfided como and nimo

applied catalysis A

Applied Catalysis A, 109 (1994) 77-96 ELSEVlER

Study of the hydrodeoxygenation of carbonyl, carboxyl.ic and guaiacyl groups over sulfided CoMo/ y-A1203 and NiMo / 3/-A1203 catalysts.

I. Catalytic reaction schemes

Etienne Laurent, Bernard Delmon* Unite’de Catalyse et Chimie des Mat.z%aw Divisb UniversitP Catholique de Louvain,

Place Croix du Sud 2 Boite 17, B-1348 Louvain-la-Neuve. Belgium

(Received 8 Februari 1993, revised manuscript received 6 November 1993)

Abstract

The elimination of specific oxygenated groups of biomass-derived pyrolysis oils (bio-oils) is necessary for improving their stability. These are mainly unsaturated groups like alkene, carbonyl and carboxylic functions, as well as guaiacyl groups. For practical applications, it is desirable that the reactions are performed selectively in order to avoid excessive hydrogen consumption. The reactions must be done at relatively low temperature in order to limit competitive thermal condensation reactions. In this study, model oxygenated compounds were used, namely 4-methylacetophenone, di- ethyldecanedioate and guaiacol. They were tested simultaneously in one reaction test in the presence of sulfided cobalt-molybdenum and nickel-molybdenum supported on y-alumina catalysts in a batch system. Their reactivity and conversion scheme were determined. The ketonic group is easily and selectively hydrogenated into a methylene group at temperatures higher than 200°C. Carboxylic groups are also hydrogenated to methyl groups, but a parallel decarboxylation occurs at comparable rates. A temperature around 300°C is required for the conversion of carboxylic groups as well as for the conversion of the guaiacyl groups. The main reaction scheme of guaiacol is its transformation in hydroxyphenol which is subsequently converted to phenol. But in batch reactor conditions, guaiacol gives a high proportion of heavy products. CoMo and NiMo catalysts have comparable activities and selectivities. However, the NiMo catalyst has a higher decarboxylating activity than CoMo and also leads to a higher proportion of heavy products during the conversion of guaiacol.

Key words: hydrodeoxygenation, sultided-nickel-molybdenum, sulfided cobalt-molybdenum, ketone, carboxylic ester, guaiacol, reaction scheme

*Corresponding author. Fax. ( + 32-10)473649.

0926-860X/94/$07.00 0 1994 Elsevier Science B.V. All rights reserved

SSDIO926-860X (93) E0229-6

78 E. Laurent, B. Delmon /Applied Catalysis A IO9 (1994) 77-96

1. Introduction

Most studies about hydrodeoxygenation reactions focused on phenolic and furanic groups, because these are the main oxygenated groups in petroleum, coal and biomass- derived liquids or shale oils and because they are highly refractory against deoxygenation [ 11. However, all these liquids may contain all other kinds of oxygenated chemical groups. This is especially the case for oils produced by rapid pyrolysis of lignocellulosic biomass in which ketones and aldehydes, carboxylic acids and esters, aliphatic and aromatic alcohols and ethers have been detected in significant quantities [ 241. The hydrodeoxygenation of all these oxygenated groups must be performed if the aim is full hydrorefining of pyrolysis oils. It is thus interesting to discern the performances of conventional hydrotreatingcatalysts

for the deoxygenation of these oxygenated chemical groups. However, the study of the hydrodeoxygenation of certain chemical groups may find an

additional interest in other reasons than the full refining of bio-oils, namely the improvement of unwanted properties and characteristics of these oils. In particular, the high oxygen content and chemical complexity of biomass pyrolysis oils account for the instability against heat and as a function of time. Condensation reactions induce the formation of heavier compounds during prolonged storage [ 51. These polymerization reactions may quickly lead to the formation of a solid coke-like product during distillation or catalytic treatment of the oils [6,7]. A stabilization step is required for storing these oils without quality degradation and is mandatory before their catalytic hydroprocessing at high temperature (300-400”C). Several authors showed that a low-temperature hydrotreatment using metal or sulfided catalysts enables the further full hydrorefining of pyrolysis oils to oxygen-free hydrocarbon products [ 7-91. The stabilizing effect has generally been attributed to the elimination of unsaturated chemical groups such as alkenes, aldehydes, ketones, carboxylic acids and guaiacol-type molecules [9-l 11. Indeed, alkenes, carbonyls and carboxylic groups are known to be reactive chemical groups. To simplify, condensation reactions in pyrolysis oils may be compared to the phenol-form01 polymerization reactions leading to bakelite. On the other hand, it is also known that pure polyhydroxyphenols or methoxyphenols thermally form gums and char upon heating at typical hydrotreatment temperatures 1121. The hydrodeoxygenation reactions of these two categories of chemical groups are the key reactions for the stabilization of pyrolysis oils. The operation should be performed quickly and at low temperature in order to avoid, as much as possible, competitive condensation reactions. If these groups can be removed selectively, a stable oil that could then be used directly for energy purposes could hopefully be produced. This should be advantageous since the cost of the corresponding refining would be limited because of the relatively low hydrogen consumption.

In the present paper, simple model compounds containing a ketonic, a carboxylic ester and a guaiacol-type chemical group were selected on the basis of a detailed GC-MS analysis of a typical rapid pyrolysis oil sample [ 41. They were reacted together in competition. This was done more for empirically taking account of the real composition of pyrolysis oils, rather than for studying the influence of the presence of one compound on the reactivity of the others. Conventional sulfided hydrotreatment catalysts were used for the study of their

E. Laurent, B. Delmon/Applied Catalysis A 109 (1994) 77-96 79

hydrodeoxygenation because, until now, these catalysts are the best candidates for hydroprocessing complex and highly oxygenated oils derived from biomass. Compared to hydrotreatment tests with whole pyrolysis oil, the use of model compounds is advantageous because it gives more insight in the reaction mechanisms and pathways as well as in the

way the catalyst works. Another advantage is that catalytic reactions are, in principle, independent of competitive thermal polymerization reactions. It also permits to save much of the time and effort that would be required by the difficult analysis of pyrolytic oils.

The literature concerning the hydrodeoxygenation of the chemical groups we have selected for study with sulfided catalysts is very scarce. The reduction of carbonyls is a reaction that has been intensively studied in the field of organic synthesis. Numerous recipes exist for a more of less selective transformation of ketones or aldehydes into corresponding alcohols over platinum group metal catalysts [ 131. The reactions can be done under very mild conditions: atmospheric pressure and a temperature of 1OO’C are typical. On the other hand, Maier studied the hydrogenation of ketonic groups over metallic nickel supported over y-alumina [ 141. The reactions were done at 250°C. The ketonic group is transformed exclusively to a methylene group with high selectivity. Weisser and Landa [ 151 reported that single metal sulfides may be used for the hydrogenation of ketones. With pure MO& or WS2 catalysts, the hydrogenation of the ketonic group into an alcoholic group is relatively easy. Typically, this can be achieved at a temperature around 200°C and a relatively low hydrogen pressure. However, if deoxygenation is contemplated, higher temperature and pressure are needed (300°C and 100 atm) because the intermediary alcohol is less easily converted than the ketonic group with these pure sulfide phases. Concerning conventional bi-metallic sulfided hydrotreatment catalysts, only Durand et al. [ 161 reported on the efficiency of a sulfided nickel-molybdenum catalyst supported on y-alumina for the hydrogenation of ketones at 250°C. Their NiMo catalyst had a slightly better activity than Maier’s supported nickel catalyst. The ketonic group was transformed directly into a methylene group.

Concerning carboxylic groups, Weisser and Landa [ 1.51 reported that single metal sulfides perform the hydrogenation. Nevertheless, more severe conditions are required than

for the hydrogenation of carbonyl groups (more than 250°C). Alcohols are produced intermediately but cannot be recovered in appreciable amounts because dehydration reactions are important under the conditions required for the hydrogenation. The reaction leads preferentially to the transformation of the carboxylic group into a methyl group, although they are also partly converted through a side reaction which is the decarboxylation. This reaction may occur with a high selectivity when nickel or cobalt sulfides [ 151 or metallic nickel, cobalt or palladium catalysts are used [ 17,181. Recent researches deal with the hydrotreating of vegetable oils [ 18,191. Gusmao et al. [ 191 reported that triglycerides may be hydrotreated in presence of sulfided nickel-molybdenum catalysts supported on y- alumina. The reaction starts at a temperature around 300°C. Normal alkanes are almost exclusively produced..

The hydroprocessing of guaiacol type molecules over CoMo and NiMo catalysts has been studied by sever,al groups these last ten years [ 20-241. Bredenberg et al. [ 20,211 and Kallury et al. [ 221 used unsulfided or poorly sulfided catalysts in their studies. The reaction conditions or catalyst working state were not optimum and the results probably not easily reproducible. It is thus very difficult to draw unequivocal conclusions from their studies. In

80 li Luurent, B. Delmon /Applied Catalysis A 109 (1994) 77-96

contrast, the studies of Klein and co-workers [23,24] give some valuable indications concerning the reactivity of anisole and guaiacol type molecules. These molecules were

fully converted at temperatures between 250°C and 300°C over a sulfided CoMo catalyst. Guaiacol appeared to be converted faster than anisole. The authors proposed preliminary reaction schemes. The initial reaction would be the rupture of the G-methyl group to give a phenolic group and methane. Subsequently, the reaction would proceed through hydrodeoxygenation of phenols to benzene and cyclohexane. They reported that, in their batch experiments, anisole is almost quantitatively converted, but that guaiacol and catechol that contain two oxygens form heavier compounds or coke. This was deduced from the impossibility to close mass balances.

The present work was undertaken with the aim to shed light on the reactivity and reaction schemes of carbonyl, carboxylic and guaiacol type oxygenated groups over conventional hydrotreatment catalysts. The originality of this work is that, for the first time, these hydrodeoxygenation reactions are compared under identical reaction conditions and that CoMo and NiMo catalysts are compared. In a companion paper [ 251, we report poisoning experiments with oxygen, sulfur and nitrogen containing compounds which give some informa- tion about the catalytic functions required for the various reactions and their resistance to poisoning.

2. Experimental

2.1. Reactor

A closed batch reactor with a volume of 0.57 dm3 was used in all the reaction tests. It was provided with an externally operated magnetic stirrer and a flat-blade turbine, a thermo- well and a line ended by a stainless steel filter immersed in the liquid phase. The reactant solution occupied one third of the total volume of the reactor. It was checked that at the operating stirring speed the catalyst particles are homogeneously suspended in the liquid phase. Preliminary experiments and the results reported in the present paper indicated that the observed reaction rates are not limited by physical processes.

2.2. Catalysts

The two industrial catalysts used in this study were cobalt-molybdenum and nickel- molybdenum supported on -y-alumina catalysts from Procatalyse (HR 306 and HR 346). These catalysts were selected because they had been used in numerous earlier studies in our laboratory and by other groups. Their general characteristics and properties are well esta- blished and they may be considered as reference catalysts. The catalysts contained 14% molybdenum oxide and 3% either nickel oxide or cobalt oxide. In one experiment, the alumina support alone (Rhone-Poulenc) was tested. It had a surface area of 240 m’/g. The BET surface area of the catalysts was 180 m*/g for the NiMo and 210 m*/g for the CoMo. The extrudates were crushed and sieved to a 0.3 15-0.5 mm fraction.

The catalysts (including the pure alumina support) were activated by a standard reduction-sulfidation procedure. It consisted in placing 1.5 g of dry catalyst in a glass reactor.

II Laurent, B. Delmon /Applied Catalysis A 109 (1994) 77-96 81

Under a flow of 100 ml/min of argon, the temperature was raised up to 120°C at a rate of

lO”C/min and held for 1 h. Then, the gas was switched to the mixture 15 vol.-% H,S in hydrogen ( 100 ml/min) and the temperature raised to 400°C in half an hour. The sulfidation procedure continued for 3 h after which the catalyst was cooled to room temperature and the gas was switched back to argon. The catalyst was kept protected from oxygen until it was poured within a fraction of a second in the solution of reactants.

2.3. Solution of model compounds

The three model compounds selected were 4-methylacetophenone (4MA), di-ethyldecanedioate (also called di-ethyl sebacate) (DES) and guaiacol (GUA). The structures of these molecules are presented in Fig. 1. They contain a ketonic group (4MA), two carboxylic ester groups (DES) and a methoxy group (GUA), respectively. One criterion for the selection of these compounds was the possibility to detect their products in one single

chromatographic analysis. These model compounds were dissolved in hexadecane. A known quantity of pentadecane

was added to the reactant solution and served as internal standard (IS.) in the chromatographic analysis. A sulfur containing compound (CS,) was added in order to generate a hydrogen sulfide partial pressure under reaction test conditions (approximately 0.5 bar,

H2S/H2 = 7 - 10-3). The composition of the standard reaction mixture is reported in Table 1.

A series of experirnents was performed with ethyldecanoate and decanoic acid as single reactant dissolved in hexadecane. The initial concentration of these compounds was 0.179 mol/l.

The reactants and gases were all commercially available from Aldrich, Merck, Janssen and 1’Air Liquide and used without any further purification.

OH

4-methyl acetophenone (4&$.& guaiacol (GUB)

d&ethyl decanedioate (di-ethyl sebacate m

Fig. 1. Structures of the oxygenated model compounds: 4-methylacetophenone (4MA). guaiacol (GUA) and di-

ethyldecanedioate (DES).

82 E’. Laurent, B. Delmon /Applied Catalysis A 109 (1994) 77-96

Table 1

Composition of the solution of reactants

Compound Quantity (g) Concentration

(mol/l)

Hexadecane (solvent) 114.5 3.03 Pentadecane (I.S.) 2.50 0.07

4-Methylacetophenone (4MA) 5.20 0.232

di-Ethyldecanedioate (DES) 5.02 0.116

Guaiacol (WA) 4.81 0.232

CS2 0.32 0.025

2.4. Reaction procedure

The batch reactor containing the mixture of reactants and the freshly sulfided catalyst was sealed. Air was evacuated and a pressure of 1 MPa of hydrogen was fixed. The temperature was increased at a rate of 4.5Wmin under mild agitation. A liquid sample was withdrawn, for the determination of the initial concentration of the reactants, upon reaching the final temperature (between 260 and 300°C). The total pressure was then increased to 7 MPa by adding hydrogen and the agitation was set at its maximum value. That moment was

considered as the zero time of reaction. The temperature of reaction stabilized in a few minutes and then could be controlled at f 1°C. Typically, during a 150 min run, ten volumes (0.5 ml) of the liquid reacting mixture were sampled. The liquid sampling line was always flushed with 3 ml of reacting mixture before taking a sample. No sampling of the gas phase was done. Hydrogen was added during the tests in order to keep the total pressure constant at 7 MPa.

A blank test experiment without catalyst was performed at 300°C for 150 min in order to check the thermal stability of the reactants in the more severe conditions used for the catalytic tests.

2.5. Analysis

The liquid samples were analyzed by gas chromatography on a Packard, model 428, equipped with a split injector and a flame ionization detector (FID), both working at 300°C. The reactants and products were separated on a 25 meter long DB-5 capillary column. Before the injection, the liquid samples were diluted with an equal volume of benzene, because produced catechol and phenol are very poorly soluble in the aliphatic hexadecane solvent. Using an initial temperature of 35’C held for 2 min and a further temperature programming of lO”C/min, all the reactants and products could be satisfactorily separated. The electronic response of the detector was integrated by a Hewlett Packard model 3308A integrator. The chromatographic peaks were attributed by GC-MS analysis and by comparison of the retention times with those of known compounds. Response factors of the reactants and produc:ts were determined experimentally using pure compounds and they served as a basis for the determination of the molar balances and selectivities.

NGOCNHA

Highlight

NGOCNHA

Highlight

NGOCNHA

Highlight

E. Laurent, B. Delmon /Applied Catalysis A 109 (1994) 77-96 83

2.6. Expression of the results

In order to compare the activity of the CoMo and NiMo catalysts, empirical reaction rate constants were calculated from the conversion data. The data were therefore fitted to a simple first order kinetic equation ( 1) :

- lnX, = kWt (1)

where Xi is the ratio of the concentration of the reactant in sample i ( Ci) to the concentration of the reactant in the initial sample (C,), k the pseudo first order rate constant (min- ’ g cata-‘), W the weight of catalyst (g) and t the time (min).

In most cases, the experimental data did not follow a pseudo first order over the whole conversion range. For that reason, only first points were used for the determination of the rate constants except for the conversion of guaiacol for which initial or middle conversion points were used. Also, the rate of appearance of the products of guaiacol conversion were calculated. For the conversion of the carboxylic ester group, as two parallel reaction pathways exist, the selectivity is expressed as a ratio of the final products.

In the simple kinetic Eq. ( 1) , the increase of the catalyst concentration due to liquid sampling is not taken into account. The correction is very low since most of the time only initial points were used in the determination of the rate constants. It has been checked that a corrected value of the rate constant of the conversion of 4MA and DES is never more than 4% from the value of the constant obtained without correction. No corrections were done for the determination of the rate constants of the conversion of guaiacol using middle conversion data so that the comparison with the conversion rate of the other reactants is possible.

3. Results

3.1. Thermal reactivity

The conversion of the three reactant molecules (% conv.) and the percentage of identified products (relative to the initial concentration of the corresponding reactant) (% prod.) after 150 min of reaction test at 300°C without catalyst are reported in the first two columns of Table 2. All the model compounds are slightly converted. A total of 18% of the initial 4MA concentration disappeared during the run. The decrease in concentration is conjugated with the production of ethyl methyl benzene in almost identical molar amount. 20% of the di- ester DES is converted; the derived products identified represent 16% of the initial DES concentration. These products are the same as those resulting from the catalytic conversion

of DES (see later). Less than 10% of GUA is converted. Small quantities of catechol and phenol were observed, which account for 60% of the converted GUA. These low conversions may be attributed to thermal reactions and to a residual catalytic activity (equipment surface and catalyst dust that could not be cleaned). The relatively low deviation of the molar balances permits to conclude that thermal side reactions leading to heavy products are very limited in our reaction conditions.

84 ,E. Laurent, B. Delmon /Applied Catalysis A IO9 (1994) 77-96

3.2. Activity of the alumina support

The conversions obtained with the alumina support alone at 280°C are reported in columns 3 and 4 of Table 2. The conversion of the ketonic group (4MA) is limited. Ethyl methyl benzene can account for nearly all the converted 4MA indicating that heavy products are not formed. The DES and GUA molecules interact more strongly with the alumina support

than 4MA. Approximately 30% of DES is transformed. 2 1% of the initial molar concentration of DES was identified as volatile products in the chromatographic analysis. Mainly, these products are the mono acid-mono ester and the di-acid corresponding to the di-ester DES plus two other non-identified products. Typical products obtained with the sulfided

catalysts (see later) were present in very low quantity (less than 2%). GUA is converted up to 40%, catechol being the only significant product observed and accounting for only 37% of the conversion. The evolution of the Ci/Co ratios of DES and GUA as a function of the reaction time are shown in Fig. 2. The conversion of guaiacol is relatively rapid at the beginning of the reaction but levels off after a certain time. At the end of the test, the conversion rate of guaiacol was nearly zero. The conversion rate of DES also tends to

Table 2

Percentages of conversion of 4-methylacetophenone (4MA). di-ethyldecanedioate (DES) and guaiacol (GUA)

in the standard reaction mixture and molar percentage (Ci/Co) of the products identified by GC after runs with

no catalyst (300”(I), sutfided alumina support alone (28O”C), sulfided CoMo (280°C) and sulfided NiMo

(280°C); catalyst weight: 1.5 g; reaction time: 150 mitt

No catalyst Alumina support CoMo catalyst NiMo catalyst

Conv. (%) Prod. (%) Conv. (%) Prod. (%) Conv. (%) Prod. (%) Conv. (%) Prod. (%)

4MA 18 18 18 17 100 100 100 100 DES 20 16 30 21 60 50 82 68 GUA 10 6 40 15 57 42 65 38

!!!f*,*; .,.,.,.,.,* 1 ’ 0 20 40 60 80 100 120 140 160

Reaction time (min)

Fig. 2. Evolution of the molar fraction Ci/Co (Xi) of di-ethyldecanedioate (DES, 0) and guaiacol (GUA, 0) as a function of the reaction time in presence of the suhided alumina support.


decrease as a function of the time of reaction, indicating a decrease of the catalytic activity

of the alumina.

3.3. Catalytic HDO conversion of the ketonic group (4MA) with CoMo and NiMo

4MA is rapidly converted in the presence of CoMo and NiMo catalysts. The conversion reaches 100% in less than 120 min at 260°C. The catalytic action of these active phases is

clear when comparing these results with the conversion obtained in the presence of the alumina support alone in identical conditions (Table 2). Ethyl methyl benzene is the only

compound produced. Even at the lowest temperature used (26O“C), methyl benzyl alcohol or methyl styrene were not observed in the chromatographic analysis. The evolution of the

0 25 50 75 100 125 150

Reaction time (min) Fig. 3. Evolution of the molar fraction of 4-methylacetophenone (0) and ethyl methyl benzene (0) as a function

of the reaction time. Catalyst: NiMo; reaction temperature: 260°C.

3 -

-0 20 40 60 SO 100 120 140 160

Reaction time (min)

Fig. 4. Pseudo first order kinetic plot of the conversion of 4-methylacetophenone at 260°C over CoMo (0) and

NiMo (0) catalysts.

86 E. Laurent, B. Delmon /Applied Catalysis A 109 (1994) 77-96

Table 3

Pseudo first order rate constants for the HDO conversion of 4-methylacetophenone over sulfided CoMo and NiMo

at 260,280 and 300°C

Temperature (“C)

260

280

300

CoMo NiMo

(min-’ g-‘) (mm’ gg’)

39.4. 1o-3 26.5*10-3

57.8.1O-3 54.8.10-3

84.9. 1O-3 83.1*10-’

molar fraction Ci/Co of 4MA and ethyl methyl benzene as a function of the reaction time in a typical test is presented in Fig. 3.

The first order ki.netic plot is presented in Fig. 4. The relation is linear provided that points at conversion higher than 90% are not considered. Beyond this value, the points fall below the first order straight line. The value of the slope of the linear regression between

the first points corresponds to the rate constant. Rate constants were measured at three different temperatures for the CoMo and NiMo

catalysts and are presented in Table 3. The activities of the two catalysts are very similar. Activation energies are 12 kcal/mol and 17.5 kcal/mol with the CoMo and NiMo catalyst, respectively.

3.4. Catalytic HDO conversion of carboxylic ester groups with CoMo and NiMo

The conversion of the carboxylic di-ester DES in the standard mixture leads to a series of products identified in the GC analysis. These were the corresponding mono acid-mono ester, the corresponding di-acid, C,e ethyl ester, Cl0 acid, Cg ethyl ester, Cg acid, CIOHZ2, C,H,, and CsH,, hydrocarbons. The molar balance is generally between 80 to 90% when the concentrations of these compounds are added (Fig. 5 and Table 2). Nevertheless, other

090 0 50 100 150 200 250

Reaction time (min) Fig. 5. Evolution of the molar percentages of DES (0). the sum of its carboxylated products (A), the sum of the

aliphatic hydrocarbons products (0) and the sum of reactant and all products (k), Catalyst: NiMo; temperature 300°C.

E. L.aurent. B. Delmon /Applied Catalysis A 109 (1994) 77-96 87

peaks were observed in the chromatographic analysis which closely surround the main peaks. They may explain the deviation from 100%. Indeed, in an other run in which ethyl decanoate was used alone, the balance was almost 100% when all the peaks bordering the

two main products were considered. The evolution of the molar percentages of the reactant, the sum of the carboxylated

products and the sum of the hydrocarbons as a function of the reaction time is shown in Fig. 5. Clearly, it is observed that the products of the reaction of DES containing a carboxylic group behave like intermediate products and that the aliphatic hydrocarbons are final products.

The observed products may be theoretically seen as the result of the rupture of different bonds. Acids are the re:sult of a de-esterification reaction. Cl0 intermediate compounds result from the full reduction of one carboxylic group by hydrogenation to a CH, group. A reaction of decarboxylation accounts for the production of C, carboxylated intermediate compounds. Decane is the final product obtained when the hydrogenation reaction occurs on both

carboxylic carbons of the molecule. When the two carboxylic groups of the molecule are eliminated, octane is produced. Nonane is the result of a combination of the reactions of hydrogenation and decarboxylation. It is supposed that the small peaks surrounding the main products are isomers that may be the result of a rearrangement of the aliphatic chain.

The number of products of DES conversion makes a precise determination of the reaction scheme difficult. This is why we performed some reaction tests with ethyl decanoate and decano’ic acid. The objective was to know if the acid form is a necessary intermediate product or, at least, if it affects the conversion by giving one kind of product preferentially. The evolution of the concentration of ethyl decanoate and its products as a function of time is shown in Fig. 6.

The carboxylic acid behaves as an intermediate product and does not accumulate in the reaction medium. Its maximum concentration is approximately 15% of the initial concentration of the ester. Decane and nonane are the final products. When decano’ic acid is reacted

instead of the ester, decane and nonane are also produced.

190

0 25 50 75 100 125 150

Reaction time (min)

Fig. 6. Evolution of the molar fraction of ethyl decanoate (0) and its products decano’ic acid (0). nonane (A)

and decane (A) as a function of the reaction time. Catalyst: NiMo, temperature: 280°C.


The pseudo first order kinetic equation ( 1) has been applied to the data of the conversion

of ethyldecanoate and decanoic acid. In the logarithmic plot, the data follow relatively well

the pseudo first order relation except at high conversion. The corresponding points are situated above the straight line passing through the initial points. This indicates that the catalytic activity increases when the concentration of the reactants diminishes and, consequently, that the reactions are inhibited more by the reactants than by the products. The rate constants calculated by linear regression through the first points are reported in Table 4 for the NiMo catalyst. ‘The carboxylic ester group is converted faster than the carboxylic acid group. The pseudo first order conversion rate constant is approximately double that of the

conversion of the carboxylic acid. The selectivity of the conversion of ethyldecanoate and decanoic acid may be expressed

as the ratio of the molar fraction of the two final products, nonane and decane. The relation between these two products is plotted in Fig. 7. The relation is not perfectly linear. The ratio nonanefdecane tends to decrease when the conversion advances. This behavior is observed with the ester and the acid. The ratios nonane/decane calculated by linear regression through the first points, as shown by the straight lines in Fig. 7, are reported in Table 4. When the carboxylic acid is reacted, a slightly higher yield of the decarboxylated product is obtained than with the carboxylic ester.

The influence of the temperature of reaction on the conversion of DES in the standard reactant mixture was studied. Overall conversion rate constants and decarboxylation per-

Table 4

Activity and selectivity of the catalyst NiMo for the HDO conversion of ethyldecanoate and decano’ic acid;

temperature: 280°C

Reactant Activity

(min-’ g-l)

Selectivity

nonane/decane

Ethyldecanoate 9.1.10-3 1.1 Decanok acid 4.6. 1O-3 1.5

Xi decane

Fig. 7. Evolution of the molar fraction of nonane as a function of the molar fraction of decane when ethyl decanoate

(0) and decanok acid (0) are reacted. Catalyst: NiMo, temperature: 28OT.


Table 5

Comparison of the activity and selectivity of the NiMo and CoMo catalysts for the HDO conversion of di-

ethyldecanedioate (DES) in the standard reactant mixture at 260,280,3OO”C

Temperature

(“C)

CoMo

Activity

(min-’ g-l)

Decarboxylation

rate (o/o)

NiMo

Activity

(min-’ g-l)

Decarboxylation

rate (%)

260 2.8*10-3 36.2 3.3. 1o-3 48.3

280 6.1*10-3 42.2 7.6. 1O-3 57.4

300 15.2.10-3 49.2 17.0*10-3 66.5

centages (determined on the basis of the final products octane, nonane and decane) obtained at three temperatures with the CoMo and NiMo catalysts are reported in Table 5. It is observed that the NiMo catalyst has a higher activity than the CoMo catalyst and also leads to a higher decarboxylation rate. Activation energies are 26 kcal/mol and 25 kcal/mol for

the CoMo and NiMo catalyst, respectively. The yield of decarboxylated products increases by 20% over the 40°C temperature range

with the NiMo catalyst and by 13% with the CoMo catalyst. Reaction rate constants for the individual reaction paths (hydrogenation and decarboxylation) were calculated assuming a simple parallel conversion scheme of DES. The results give an activation energy of 31 and 30 kcal/mol for the decarboxylation reaction and 22.5 and 18 kcal/mol for the hydrogenation reaction with the CoMo and NiMo catalyst, respectively.

3.5. Catalytic HDO conversion of guaiacol with CoMo and NiMo

The products of the conversion of guaiacol are catechol (2-hydroxyphenol) and phenol. Traces of benzene and cyclohexane are also detected at long reaction times. Methylated products do not appe,ar in measurable quantities. The molar balance is about 80 to 90% at a moderate guaiacol conversion (between 50 to 70%) with the CoMo catalyst. The balance is always poorer with the NiMo catalyst (65 to 75%) at a similar guaiacol conversion. When compared to thie conversion obtained with the alumina support alone (Table 2), the sulfided bimetallic CoMo and NiMo catalysts have a higher catalytic activity. Nevertheless, the effect is less mark:ed than with 4MA and DES.

The evolution of the molar fraction of guaiacol and its products in a typical test (CoMo catalyst) is plotted in Fig. 8. Catechol is produced in appreciable quantity at the very first, indicating that catechol is a primary product. Nevertheless, when the initial rate of conversion of guaiacol is compared to the initial rate of production of catechol (straight lines in Fig. 8), it is observed that catechol does not account for the whole guaiacol conversion. Other products, not detected in the chromatographic analysis, are produced. Phenol is not present in the first samples but is produced almost continuously as a function of the time of reaction. This suggest that phenol is not a primary product or, at least, that it is produced in very low proportion directly from guaiacol.

The rate constants for the conversion of guaiacol were determined by linear regression in a first order kinetic logarithmic plot (Fig. 9). The conversion data do not follow pseudo first order kinetics over the whole conversion range. Typically, the logarithmic plot may be


170

.g 0,s

2 & 0,6

2 5 E

0,4

k- 0,2

030 0 50 100 150 200 250 300

Reaction time (min)

Fig. 8. Evolution of the molar fraction of guaiacol ( l ), catechol (A), phenol (0) and their sum (It) as a function of the reaction time. Catalyst: CoMo, temperature: 280°C.

0 40 80 120 160

Reaction time Fig. 9. Typical logarithmic plot of the guaiacol conversion data.

divided in two regions: initial conversion and middle conversion. The rate constants calculated from the initial conversion data are always higher than those obtained with the middle conversion data. In Fig. 8, the rate constant at initial conversion was used for drawing the negative exponential curve tentatively fitting the guaiacol conversion data. It may be observed that this curve fits the first points but deviates quickly from the conversion data. As apparent first order kinetic is generally valid for the HDO of model oxygenated compounds at moderate conversion, the observed deviation must be due to a diminution of the catalytic activity.

Reaction rate constants were determined from the guaiacol conversion (initial conversion (tic& and middle conversion (km,,,)) and from the catechol+ phenol production (kPRo) . They are reported in Table 6.

The NiMo catalyst. has a better activity for the conversion of guaiacol than the CoMo catalyst no matter halw the activity is expressed. Nevertheless, the CoMo catalyst has a

E. Laurent, B. Delmon /Applied Catalysis A 109 (I 994) 77-96 91

Table 6

Pseudo first order rate constants of the reaction of guaiacol calculated from the initial conversion data (kio&.

from the middle conversion data (km ouA) and from the production of catechol + phenol (k,,,) at 260,280 and

300°C with the sulfided CoMo and NiMo catalysts

Temperature CoMo NiMo

&XL4 km FUA km kbun km G”A ho (min-’ gg’), (mm’ g-‘) (min-’ g-l) (min-’ g-‘) (en-’ g-‘) (tin-’ g-‘)

260 4.5. 1o-3 1.3. 1o-3 1.3. 1o-3 7.7*10-3 2.0. 1o-3 1.7. 1o-3

280 7.5*10-3 3.3*10-3 2.7. lo-’ 12.6~10-~ 4.2. lo-’ 3.7. 1o-3

300 14.0.10-3 7.7*10-3 7.0*10-3 19.1*10-3 11.5*10-3 9.0. 1o-3

slightly better selectivity (ratio kpRo IkouJ forn the production of catechol + phenol than

the NiMo catalyst. This means that side reactions are more important with the NiMo catalyst.

The apparent activation energies that were calculated from these rate constants are 17 kcal/

mol, 26.9 and 25.1 kcal/mol for k&o,, kmGuA and kpRo of the CoMo catalyst, respectively.

The activation energi’es with the NiMo catalyst are 14.0,26.5 and 25.0 kcal/mol for k&u,, km oUA and kPRO, respectively.

4. Discussion

4.1. Ketonic group

The hydrogenation of the carbonyl group of 4-methylacetophenone is easily catalyzed by sulfided CoMo and NiMo catalysts at temperatures higher than 200°C. Its conversion leads quantitatively to ethyl methyl benzene with no intermediate products observed. This confirms the results reported by Weisser and Landa [ 151, who mentioned that the selectivity of the transformation of ketone to alcohol was lower with sulfides than with noble metal catalysts because deh:ydration reactions occur and become important at temperatures higher than 250°C. In our experiments, alcohol or alkene were never observed. This may be partly due to the higher temperature and partly to the presence of the alumina support. Indeed, it is well known that -,-alumina catalyzes dehydration reactions. Under hydrotreating conditions, aliphatic alcohols are quickly transformed in presence of y-alumina, mainly into alkenes [ 261. Maier et al. [ 271 observed that y-alumina alone may perform the hydrogenation of tertiary alcohols into aliphatic hydrocarbons at temperatures as low as 180°C. An alternative explanation for the absence of intermediate alcohol in the products is that benzyl alcohol is known to b’e poorly stable [ 151. The absence of intermediate alkene agrees with the observations of Durand et al. [ 161 that a temperature as low as 175°C is necessary in order to detect alkenes during the hydrodeoxygenation of ketonic groups with a sulfided NiMo catalyst. The rate limiting step is thus the first hydrogenation of the C=O double bond. The low activation energies that were measured ( 12 and 17.5 kcal/mol) are typical of double bond hydrogenation reactions [ 131.

92 E. Laurent, B. Delmon /Applied Catalysis A IO9 (1994) 77-96

4.2. Carboqlic group

Carboxylic groups are more refractory to deoxygenation than carbonyl groups. Our results indicate that a carboxylated molecule reacts at temperatures near 300°C in presence of CoMo and NiMo catalysts. Their conversion into hydrocarbons is almost quantitative with little side degradation reactions.

Two main reaction pathways exist: one is the carboxylic group hydrogenation and the second is the decarboxylation reaction. A third reaction accounts for the production of carboxylic acid as intermediate product during the conversion of carboxylic esters (de- esterification) . Using pure carboxylic acid (decandic acid), we have shown that it is less reactive than the corresponding carboxylic ester (ethyl decanoate) and that it gives a slightly higher yield of decarboxylated products (Table 4). The measured activity for the conversion of decandic acid may be compared to a calculated value assuming a consecutive reaction scheme for the conversion of ethyl decanoate. The calculated value of the conversion of the supposed intermediate acid is 56.6. 10e3 and is one order of magnitude higher than the value measured by reacting the acid alone (Table 4). This indicates that the scheme we have assumed is not correct. Consequently, we may deduce that the reaction proceeds for a large part directly from the ester to the hydrocarbon products and that the production of the acid is a limited side: reaction of the ester conversion scheme. The following simple reaction scheme summarizes these findings (Fig. 10).

Weisser and Landa [ 151 interpreted the hydrogenation of the carboxylic group to a methyl group as proceeding through the hydrogenation of the C=O double bond to an intermediate hemiacetal, the latter being transformed successively in an aldehyde, an alcohol, an alkene to give finally the hydrocarbon product. None of these intermediate compounds were identified in our CC analysis. Nevertheless, the observation that n-decane, the product of the hydrogenation of ethyl decanoate or decanoic acid, was bordered by small peaks, that we may presume to be isomers, is an indication that the reaction proceeds by a carbocation enabling a rearrangement of the carbon skeleton. The relatively low activation energy of this reaction pathway (20 kcal/mol) agrees with the fact that the hydrogenation of the C=O double bond is rate limiting in the conversion process. Our value is comparable to values reported for the hydrogenation of carboxylic esters to alcohols ( 15-18 kcal/mol) with metallic catalysts [ 281.

The literature concerning decarboxylation is relatively scarce. Weisser and Landa [ 151 presented this reaction as a side reaction of the hydrogenation of carboxylic acids. On the other hand, Maier et al. [ 171 studied this reaction with the aim to develop a catalytic organic

R-CH,

R-~-(,-H

R-H L t

Fig. 10. Simple reaction scheme of the conversion of carboxylic ester group with sulfided CoMo and NiMo catalysts.

E. Lmrent, B. Delmon /Applied Catalysis A IO9 (1994) 77-96 93

synthesis route. They tested several carboxylic acids with a Nily-alumina and a Pd/silica catalyst. The reactions were very effective when “promoted” carboxylic groups (e.g. benzoic acid) were reacted with the nickel catalyst. The reactions were less selective when aliphatic acids were used. With the palladium catalyst, the yields of decarboxylated products were generally around 95%. They proposed a reaction mechanism based on the adsorption of the carboxylic group through the LY and p carbons. In our experiments, the hydrocarbon produced by the decarboxylation reaction was always surrounded by multiple peaks. This may be an indication that an alkene or a carbocation are intermediately produced in this conversion route, enabling the rearrangement of the aliphatic chain. A non-concerted mechanism may thus be at stake with sulfided catalysts.

The estimated activation energies for this reaction path (30 kcal/mol) are typical of carbon-carbon bond breaking reactions.

With the CoMo and NiMo catalysts, the decarboxylation yields are between 30 and 70% at temperatures between 260 and 300°C. Weisser and Landa [ 151 reported that, with single phase MO& or WS2 catalysts, the decarboxylation rate is no more than 10% when heat stable carboxylic acids are reacted. It is not easy to explain the origin of the higher decarboxylating activity of promoted molybdenum catalysts on the basis of our results. Many parameters may influence (presence or not of a support, dispersion state, presence of promoter sulfide phase, etc.). Weisser and Landa [ 151 also reported that cobalt and nickel sulfide phases are much more decarboxylating than molybdenum sulfide. Further investi- gations concerning this point are necessary.

Our results indicate that a NiMo catalyst has a higher decarboxylating activity than a CoMo catalyst. The origin of this difference could tentatively be searched in a difference of acidity and ensuing cracking activity of the two solids. In that respect, the results could be compared to the higher acidic cracking activity of a NiMo catalyst compared to a CoMo catalyst reported by L,edoux and Djellouli [ 291 in the hydrodenitrogenation of pyridine. However, this interpretation requires further comparative characterization of the acido-basic properties of these catalysts.

We also observed that the selectivity for the decarboxylation reaction decreases slightly with the conversion of the carboxylated reactant (Fig. 7). This may find its origin in the fact that the inhibition of the hydrogenation pathway is higher than that of the decarboxy-

lation pathway, due to competitive adsorption of the carboxylated reactant on corresponding active sites or to a prolmotion of the decarboxylation reaction by the reactant molecule. The first possibility seems likely, as the total activity is also inhibited by the carboxylated reactant.

4.3. Guaiacyl group

The thermal conversion of guaiacol at 300°C is low. Only 10% is transformed and small quantities of catechol and phenol are produced. This result is in agreement with the literature dealing with the thermal conversion of guaiacyl groups under inert atmosphere [ 30-331. For example, Bredenberg and Ceylan [ 3 1 ] reported a conversion of guaiacol of 40% after 2 h pyrolysis in the presence of tetraline at 345°C. The products were mainly catechol and phenol and accounted for less than 50% of the conversion. In our experiments, in presence of sullided y-alumina alone at 28O”C, the conversion reached 40%. It was around 60% with


the CoMo and NiMo catalysts. There is thus a substantial difference when a catalyst is present.

The initial reaction in the conversion of guaiacol is the rupture of the O-methyl bond of the methoxy group leading to the production of catechol. This is also the first reaction in the conversion scheme of anisole. The same observations were done by Hurff and Klein [ 231. The catechol is subsequently transformed by elimination of one hydroxyl group into phenol. At that stage, the further conversion leads to cyclohexane and benzene. A simple reaction scheme is presented in Fig. 11.

Under our reaction conditions, the conversion of guaiacol was limited to the production

of catechol and phenol. The quantity of these products could never account for the quantity of converted guaiacol. The default of the molar balance was between 15 to 30% for a guaiacol conversion of about 60%. Hurff and Klein [ 231 reported an average 80% material balance closure at moderate guaiacol conversion with a CoMo catalyst at 250°C. Petrocelli and Klein [ 241 obtained a 90% closure at 300°C and 50% conversion of guaiacol with the same CoMo catalyst. These results are in close agreement with ours. The impossibility to close the material balance can be attributed to the formation of heavy products or coke. This is deduced by analogy with the fact that methoxyphenols or hydroxyphenols form a high proportion of char during their pyrolysis [ 30-331.

The apparent activation energies obtained for the initial overall conversion of guaiacol (14-17 kcal/mol) are much lower than could be expected for a carbon+oxygen bond rupture. Hurff and Klein [23] reported an activation energy of 29.7 kcal/mol for the conversion of anisole. The low activation energies that we observed are indicative that polymerization reactions occur at the beginning of our reaction tests. Indeed, condensation reactions are known to have a low activation energy and rapid coking of fresh hydrotreating catalysts is a well known phenomenon [ 341. This is why, in continuous flow experiments, a stabilization period is required before measurement of a stationary activity of the catalysts. Nevertheless, it is likely that the initial coking rate of one compound is indicative of its coke formation tendency [ 341. In this respect, guaiacol has a high coking tendency when compared to the other model compounds of the present work or to the phenols of a previous work [351 for which the mass balances were always near 100% under our batch reactor conditions. The activation energies calculated from the rate constants at middle conversion or from the production of catechol and phenol are higher (25-26 kcal/mol) than the activation energies calculated from the initial guaiacol conversion rate constants. Neverthe- less, they are still lower than what could be expected. This may be explained by the fact that the product catechol has also a high tendency to react through condensation reactions to high molecular weight compounds [ 241.

Fig. 11. Simple reaction scheme of the conversion of guaiacol with sulfided CoMo and NiMo catalysts.


The conversion rate of guaiacol appears to decrease relatively quickly when the alumina support was used alone. With the sulfided catalysts, a decrease of the catalytic activity for

the conversion of guaiacol was also observed (compare kioo, and km,“,). This loss of activity may have two origins. As the condensation reactions lead to the formation of coke or heavy products deposited on the catalyst, the strongest active sites may be quickly blocked. But the deactivation may also be attributed to the production of catechol. Indeed, catechol has been observed by infrared spectroscopy to adsorb strongly on y-alumina [ 361. Guaiacol had a lower interaction strength. The active sites for the decomposition of the methoxy group may thus be blocked by competitive adsorption of catechol.

5. Conclusions

Ketonic, carboxylic and guaiacyl groups may be hydrodeoxygenated with sulfided CoMo and NiMo hydrotreating catalysts. A temperature above 200°C is required for the hydrogenation of the ketonic group. Carboxylic and guaiacyl groups need temperatures around 300°C. Ketones are quantitatively hydrogenated in the corresponding hydrocarbons. The same reaction occurs with carboxylic groups but corresponds only to one pathway, the other pathway being decarboxylation. Carboxylic acids and esters react at a similar rate and with a similar selectivity. The selectivity of decarboxylation is more important with the NiMo catalyst than with the CoMo catalyst and increases with temperature. The initial reaction of the conversion of guaiacol is the rupture of the O-methyl bond of the methoxy group giving hydroxyphenol which may be further deoxygenated. Under our batch reaction conditions, the conversion of guaiacol in light products is not quantitative indicating that undetected heavy products form. The selectivity in light products was lower with the NiMo catalyst.

Acknowledgement

The work was performed under the frame of CEC contract Joub-0055.

References

[l] E. Furimsky, Catal. Rev.-Sci. Eng., 25 ( 1983) 421.

[2] H. Pakdel, H.G. Zhang and C. Roy, in A.V. Bridgwater (Editor), Advances in Thermochemical Biomass

Conversion, Interlaken, May 1992, in press.

[3] J. Piskorz, D. St. A.G. Radlein, D. Scott and S. Czemik, in A.V. Bridgwater and J.L. Kuester (Editors),

Research in Thermochemical Biomass Conversion, Phoenix, April 1988, Elsevier Applied Science, London

and New York, p. 557.

[4] R. Maggi and B. Delmon, Fuel, in press.

]5] M.M. Essayegh, Thesis, Pierre et Marie Curie University, Paris, 1988.

161 D.C. Elliott, Analysis and Comparison of Biomass Pyrolysis/Gasification Condensates, Final Report, PNL 5943 UC-61 D, 1986.

[7] D.C. Elliott and E.G. 13aker, in D.L. Klass (Editor), Energy from Biomass and Wastes X, IGT, Chicago, 1987, p. 765.


[S] E. Churin, R. Maggi, P. Grange and B. Delmon, in A.V. Bridgwater and J.L. Kuester (Editors), Research

in Thermochemical Biomass Conversion, Phoenix, April, 1988, Elsevier Applied Science, London and New

York, 1988, p. 896.

[9] J. Gagnon and S. Kaliaguine, Ind. Eng. Chem. Res., 27 (1988) 1783. [lo] E. Laurent, P. Grange and B. Delmon, in G. Grassi, A. Collina and H. Zibetta (Editors), 6th European

Conference on Biomass for Energy, Industry and Environment, Athens, April 1991, Elsevier Applied

Sciences, London and New York, 1991, p. 667.

[ 111 D.C. Elliott and E.G. Baker, US Patent 4 795 841 (1989).

[ 121 J.B. Bredenberg and R. Ceylan, Fuel, 62 ( 1983) 343.

[ 131 P.N. Rylander, Catalytic Hydrogenation over Platinum Metals, Academic Press, New York and London,

1967, p. 229.

[ 141 W.F. Maier, K. Gergman, W. Bleicher and P. v. R. Schleyer, Tetrahedron Lett., 22 ( 1981) 4227.

[ 151 0. Weisser and S. Landa, Sulphide Catalysts, their Properties and Applications, Pergamon, Oxford, 1973,

p. 168-175.

[ 161 R. Durand, P. Geneste, C. Moreau and J.L. Pirat, J. Catal., 90 (1984) 147.

[ 171 W.F. Maier, W. R’oth, I. Thies and P. v. Rag& Schleyer, Chem. Ber., 115 (1982) 808.

[ 181 G. Cecchi and A. Bonfand, Etudes et Recherches No. 2.9 (1987) 397.

[ 191 J. Gusmo, D. Brodzdi, G. DjBga-Mariadassou and R. Frety, Catal. Today, 5 (1989) 533.

[20] J.B. Bredenberg, M. Huuska, J. R;ity and M. Korpio, J. Catal., 77 (1982) 242.

1211 J.B. Bredenberg, M. Huuska and P. Toropainen, 1. Catal., 120 (1989) 401.

[22] R.K.M.R. Kallury, W.M. Restivo, T.T. Tidwell, D.G.B. Boocock, A. Crimi and J. Douglas, J. Catal., 96

(1985) 535.

[23] S.J. Hurff and M.T. Klein, Ind. Eng. Chem. Fundam., 22 (1983) 426.

[24] F.P. Petrocelli and M.T. Klein, Fuel Sci. Technol. Int., 5 (1987) 63.

[25] E. Laurent and B. Delmon, Appl. Catal., 000 (1994) 000.

[26] B.S. Gevert, J.-E. Otterstedt and F.E. Massoth, Appl. Catal., 31 (1987) 119.

[27] W.F. Maier, I. Thies and P. v. R. Schleyer, Z. Naturforsch., 37b (1982) 392.

[28] C. Travers, Ph. D. Thesis, Ecole Nationale Sup&ieure du P&role et des Moteurs, Paris, 1982.

[29] M.J. Ledoux and El. Djellouli, Appl. Catal., 67 (1990) 81.

1301 R. Ceylan and J.B. Bredenberg, Fuel, 61 ( 1982) 377.

[31] J.B. Bredenberg and R. Ceylan, Fuel, 62 (1983) 343.

[32] J.R. Lawson and M.T. Klein, Ind. Eng. Chem. Fundam., 24 (1985) 203.

[33] A. Vuori and J.B. ;Bredenberg, Holzforschung, 3 (1984) 133.

[ 34 I P. Wiwel, P. Zeuthen and A.C. Jacobsen, in C.H. Bartholomew and J.B. Butt (Editors), Catalyst Deactivation

1991 (Studies in Surface Science and Catalysis, Vol. 68), Elsevier, Amsterdam, 1991, p. 257.

[35 I E. Laurent and B. IDelmon, Ind. Eng. Chem. Res., 32 (1993) 2516.

[36 I J.B. Bredenberg and Z. Sarbak, J. Chem. Tech. Biotechnol., 42 (1988) 221.

study of hdo carbonyl, carboxylic and gua, over sulfided como and nimo

Documents