Hydrogenation of 2-ethylanthraquinone on alumina-supported palladium catalysts

Applied Catalysis A: General 263 (2004) 71–82

Hydrogenation of 2-ethylanthraquinone onalumina-supported palladium catalysts

The effect of support modification with Na2SiO3

A. Drelinkiewicza,∗, R. Kangasb, R. Laitinenc, A. Pukkinenc, J. Pursiainenb

a Faculty of Chemistry, Jagiellonian University, Ingardena 3, 30-060 Krakow, Polandb Department of Chemistry, University of Oulu, P.O. Box 3000, 90014 Oulu, Finland

c Oulu Research Centre, Kemira Chemicals, P.O. Box 171, 90101 Oulu, Finland

Received in revised form 26 November 2003; accepted 1 December 2003

Abstract

Hydrogenation of 2-ethylanthraquinone (eAQ) in the fixed-bed reactor was studied in the “Anthra” (AQ) and “All-Tetra” systems onegg-shell palladium profile catalysts (0.7–2 wt.% Pd). Alumina modified with Na2SiO3 solution was used as a support. Solution of activequinones (eAQ and 2-ethyltetrahydroanthraquinone H4eAQ) was saturated with hydrogen and circulated through the catalyst bed at thetemperature 50◦C and the pressure 5 bar. The concentrations of eAQ, H4eAQ and degradation product were analysed by the GC method.Modification of the starting support with Na2SiO3 was performed with three Sh, Lg and Vc procedures differing in the method of supportdrying and the time of support contacting with the Na2SiO3 solution. Efficiency of modification was studied by specific surface area, porosityand Pd concentration profile. Procedure of support modification did not influence appreciably the rate of the main reaction eAQ to eAQH2

reduction, while evidently affected the course of side reactions consuming desirable eAQH2 and producing H4eAQ and degradation products.Performance of the catalysts in terms of their ability towards hydrogenolytic reactions producing degradation products and decrease in theactivity during the hydrogenation run have been correlated with the efficiency of support modification with Si species. The better this efficiency,as in Vc procedure, the better maintenance of catalysts activity during the hydrogenation experiment.© 2003 Elsevier B.V. All rights reserved.

Keywords:2-Ethylanthraquinone; Egg-shell catalysts; Deactivation

1. Introduction

Catalytic hydrogenation of 2-ethylanthraquinone (eAQ)is the main stage in the anthraquinone process of H2O2production [1,2]. In the cyclic process, eAQ is hydro-genated (Scheme 1) to yield 2-ethylantrhahydroquinone(eAQH2). Oxidation of the latter eAQH2 resulted in H2O2and regeneration of the starting eAQ. However, eAQH2,the primary and desirable product, is partly hydrogenatedto different side products (Scheme 2), among whichonly 2-ethyltetrahydroanthrahydroquinone (H4eAQH2) isdesired, because it can participate in the oxidation re-action producing H2O2. eAQ and H4eAQ are thereforetermed “active quinones”, while all other hydrogenation

∗ Corresponding author.E-mail address:[email protected] (A. Drelinkiewicz).

products formed (2-ethyloctahydroanthrahydroquinone,2-ethylanthrone, 2-ethyloxanthrone, dimers[1–5]) not capa-ble of H2O2 formation are termed “degradation products”.From technological point of view, the low content of degra-dation products is acceptable. The anthraquinone processcan be realised as the “Anthra” (AQ) or the “All-Tetra”system[1]. When concentration of H4eAQ is kept low andeAQ is the quinone taking predominant part in cyclic pro-cess (Scheme 1), the system corresponds to “Anthra”. In“All-Tetra” system, the solution normally contains 70% ofH4eAQ and 30% of eAQ. Pd-supported catalysts are com-monly used in the anthraquinone hydrogenation carried outas fluid or fixed-bed operations. Several papers have alreadyreported the kinetics of fluid-bed operation[3–5,11] whilethe available literature devoted to the fixed-bed operationconsists mostly of patents[6–10].

The catalysts applied in a fixed-bed operation must fulfila number of requirements, but the most important is a long

0926-860X/$ – see front matter © 2003 Elsevier B.V. All rights reserved.doi:10.1016/j.apcata.2003.12.010

72 A. Drelinkiewicz et al. / Applied Catalysis A: General 263 (2004) 71–82

Scheme 1.

working life. As a crucial factor for catalysts performancein terms of activity and life time, the thickness of Pd shellhas been considered. Spherically shaped alumina-supportedegg-shell palladium profile catalysts with Pd penetration nodeeper than 40–50�m [8] have been reported as the pre-ferred ones. However, the fixed-bed hydrogenation on a se-ries of Pd catalysts with various supports (�- and�-Al2O3,MgO, silica-alumina, calcium carbonate, SiO2, silicon car-bide, magnesium carbonate) showed the best performance ofPd/SiC catalysts[6]. No loss of quinones was then observedthus indicating that formation of degradation products wasstrongly (or completely) prevented. On the other hand, rel-atively high content of degradation products was found on�-Al2O3-supported Pd. In a number of papers, an opinionis presented[8,10,11] that alumina-supported Pd catalystsexhibited rather short life time in the anthraquinone hydro-genation. We have reported in the previous paper[12] thatdeactivation of egg-shell Pd catalysts with alumina and silicasupports in the fixed-bed operation quite well correlated withthe course of reactions producing degradation products. Thetype of support: alumina or silica was found to exhibit anessential role. Alumina-supported catalyst with sufficientlydeep penetration of Pd (∼200�m) exhibited especially hightendency to the formation of degradation products. Such acatalyst was irreversibly deactivated during the hydrogena-tion run. Various alkaline reagents have been used in orderto modify the properties of the starting�-Al2O3 and in con-sequence to restrict the formation of degradation products.

Scheme 2.

A remarkable improvement in properties has been reachedonly if silica species have been introduced onto�-Al2O3,e.g. support has been pre-impregnated with Na2SiO3 solu-tion. Such a reagent, considered to be most effective[12] istherefore used in the present work. The objective is to exam-ine the role of silica species as well as the thickness of Pdshell in the alumina-supported catalysts in the course of an-thraquinone hydrogenation carried out as the fixed-bed op-eration. In particular, the effects of procedure of the startingsupport pre-impregnation with Na2SiO3 and Pd loading inthe catalysts on their performance in terms of activity, abilitytowards side reactions producing degradation products anddeactivation during the hydrogenation run are presented.

2. Experimental

2.1. Materials

The reagents for hydrogenation experiments (eAQ,H4eAQ, organic solvents) as well as materials for the cata-lysts preparation (support, PdCl2) were supplied by KemiraChemicals.

Industrially manufactured alumina (�-Al2O3) contain-ing 10 wt.% of SiO2 (BET surface area 229 m2/g, porosity0.54 cm3/g, density 0.58 g/cm3) in the form of sphericallyshaped grains of the very uniform diameter of 1.6–1.7 mmin size was used as a support. The catalysts of various Pdloading (0.7–2 wt.%) with an egg-shell Pd concentrationprofile were prepared by precipitation method, consisted ofthe following processing steps.

2.1.1. Pre-impregnation of support with Na2SiO3 solutionThe aim of this preliminary treatment was partial blockage

of the pores (especially near to the outer surface of grains) inorder to prevent penetration of Pd2+ ions into the bulk of the

A. Drelinkiewicz et al. / Applied Catalysis A: General 263 (2004) 71–82 73

Table 1The conditions of the starting support pre-impregnation with Na2SiO3 solution, procedure Sh, Lg and Vc

Procedure Treatment of support prior topre-impregnation with Na2SiO3

Concentration of Na2SiO3

solution (mol/dm3)Time ofimpregnation (h)

Sh Drying 16 h at 120◦C 0.007 2Lg Drying 16 h at 120◦C 0.007 16Vc Evacuation at room temperature (0.1 Tr) for 2 h 0.007 2

grains and simultaneously facilitate the precipitation of pal-ladium hydroxide owing to modification of the alkalinity ofthe starting support. Three procedures Sh, Lg and Vc of thesupport pre-impregnation with Na2SiO3 solution were used(Table 1). They differ in method of support drying (16 h at120◦C in Sh and Lg procedures, evacuation at room temper-ature in Vc procedure) prior to the incorporation of Na2SiO3solution and the time period of pre-impregnation withNa2SiO3 (2 h in Sh and Vc methods, 16 h in procedure Lg).

Pre-impregnation was carried out as follows: support(10 g) was introduced into an excess of Na2SiO3 solution(400 cm3) and slowly mixed at room temperature. The liq-uid was then decanted off, the support washed with water(up to pH∼7) and dried for 4 h at 120◦C. In Vc procedure(Table 1), solution of Na2SiO3 was introduced directly tothe vacuum vessel to avoid the contact of the evacuatedsupport with air.

2.1.2. Sorption of Pd2+ aquachlorocomplexesSorption of Pd2+ aquachlorocomplexes on the modified

support was carried out using PdCl2 solution of the com-position: 2.65 g PdCl2/dm3, NaCl:PdCl2 molar ratio 2.5,pH 3.5–4. The support pre-impregnated with Na2SiO3 wasadded to the excess of PdCl2 solution (180 cm3 of solutionper 10 g of support) and the suspension was gently and con-tinuously agitated for an appropriate time period. Impreg-nation time was varied to get different final Pd loading incatalysts (0.7–2 wt.% Pd). The catalyst grains were removedfrom the PdCl2 solution before total sorption of Pd2+ hadoccurred. The time of impregnation was chosen from an ex-perimentally determined relationship between the content ofPd in the catalysts and the time of support contacting withPdCl2 solution. Such relationships were separately deter-mined for supports modified by Na2SiO3 with Sh, Lg andVc procedures.

2.1.3. Hydrolysing of the sorbed Pd2+ compoundsHydrolysing of the sorbed Pd2+ compounds to palladium

hydroxide (or hydrated palladium oxide) was carried outwith 0.4 M Na2CO3 solution at 70◦C. After hydrolysis thegrains of catalysts were washed with water until eliminationof Cl− ions and dried for 4 h at 120◦C.

The content of palladium in the catalysts was calculatedfrom the difference between PdCl2 concentration deter-mined before and after incorporation of Pd onto the support.The concentration of Pd2+ was analysed with AAS method(Perkin-Elmer 4100 atomic absorption spectrometer). In

selected samples, the content of Pd was also colorimetri-cally determined[13].

The catalysts were reduced immediately before the hy-drogenation test in the fixed-bed reactor. Gaseous hydrogenpassed for 10 min at 50◦C through the catalyst bed wettedwith eAQ solution.

2.2. Characterisation methods of catalysts

BET surface area, pore volume and size distribution ofpores were evaluated by nitrogen adsorption and desorptionisotherm analysis (Sorptomatic 1990).

Depth profile analysis of elements Al, Na, Si and Pd inthe supports modified with Na2SiO3 as well as those ofcatalysts containing various Pd loading was investigated bymeans of scanning electron microscope (Philips XL-30). Themicroscope was equipped with an X-ray microprobe, whichmade it possible to carry out distribution studies. In orderto expose the cross-section of sample for the analysis, thegrains were resin-bounded, cut and polished.

2.2.1. Hydrogenation experimentsHydrogenation experiments were carried out using the

same apparatus and procedure as described earlier[12]. Incatalytic tests, two solutions were used in which the concen-tration of quinones was equal to 60 g/dm3. The first solutioncontained only eAQ while the composition of the secondone corresponded to that applied in the “All-Tetra” system:42 g/dm3 H4eAQ and 18 g/dm3 eAQ (70% of H4eAQ).

The quinone solution was saturated with hydrogen beforeintroducing into fixed-bed reactor. The saturation was car-ried out at the same pressure as in the subsequent hydrogena-tion test (5 bar). The fixed-bed hydrogenator was chargedwith 25 cm3 of a catalyst and a total 4 kg of quinone solutionwas circulated through the catalyst bed (cylinder: diame-ter 54.4 mm, height 10.8 mm, cross-section area 23.23 cm2)with the flow rate of 170 kg/h. Solution was introduced ontothe top of the catalyst bed. During the catalytic run, solu-tion was continuously saturated with hydrogen in order tomaintain the constant concentration of hydrogen (equilib-rium concentration of H2 at pressure of hydrogen over theliquid equal to 5 bar).

Each catalytic run comprised three subsequent stagesAQ-I, AQ-II and Tetra-III.

2.2.1.1. Stage AQ-I. In the first stage, a fresh catalyst and afresh eAQ solution (60 g/dm3) were used. The hydrogenation


was carried out at the temperature 50◦C and the pressure5 bar. Although after 5 h of the test, the reduction of eAQto eAQH2 was almost complete (%R ∼90) the hydrogena-tion was continued for 18 h. The samples of solution takenfrom the reactor after 1, 2, 3.5, 5 and 23 h of reaction wereoxidised with O2 and analysed. The concentration of eAQand H4eAQ was determined with GC method in conditions:column DB-1701 30 m×0.32 mm×0.25�m, FID detector,isothermal run 230◦C, flow rate of He 1.3 cm3/min.

The products other than active quinones (eAQ, H4eAQ)produced in small quantities were only qualitatively anal-ysed with GC–MS method. Most of such products, althoughwere detected by the more sensitive thin layer chromatorga-phy (TLC) were not detected by GC or HPLC[14]. Most ofthem probably remain already on the top of chromatographiccolumns. Thus, cumulative content of these compounds de-signed as the degradation products was calculated from themass balance similarly as it has been done by other authors[3].

The content of H2O2 produced in oxidation of the samplesof anthraquinone solution taken from the reactor was anal-ysed by titration with KMnO4 solution. The value (AQ-I,1 h) presents a number of H2O2 moles formed during the firsthour of reaction, (AQ-I, 2 h) corresponds to the second hourof hydrogenation. From the number of H2O2 moles deter-mined in this stage, the conversion (%R) of quinone (eAQ)to hydroquinone (eAQH2) in the main reaction (Scheme 1)was calculated.

The data obtained in AQ-I stage enabled to describe thecourse of eAQ hydrogenation and in particular to estimatethe catalyst activity, formation of H4eAQ and degradationproducts.

2.2.1.2. Stage AQ-II. After 23 h of hydrogenation in AQ-Istage, the new eAQ solution (4 kg) was introduced intoreactor without contacting the catalyst with air. The hydro-genation was carried out for 2 h at 50◦C and 5 bar withcirculation of the eAQ solution at the rate 170 kg/h. Thesamples of liquid were taken in every hour and the numberof H2O2 moles (AQ-II, 1 h) and (AQ-II, 2 h) was determined.

2.2.1.3. Stage Tetra-III. The solution containing H4eAQ(70%) and eAQ (30%) of total quinones concentration60 g/dm3 was introduced into the apparatus. The hydro-genation was carried out for 1 h at the same conditions asin AQ-II stage. In the solution taken after 1 h of reaction,the number of H2O2 moles (Tetra-III, 1 h) was determined.

Performance of the catalysts was characterised by theirinitial activity and deactivation occurring during the hydro-genation run. As a measure of initial activity, the numberof H2O2 moles (AQ-I, 1 h) formed during the first hourof the initial AQ-I stage has been assumed. As a measureof catalysts deactivation, the decline in the activity duringthe initial AQ-I and second AQ-II stages have been as-sumed. The former is given by Deact-1 and Deact-2, whereasDeact-3 presents decline in the activity during both AQ-I and

subsequent AQ-II stages. From the number of H2O2 molesformed during the first and second hours of reaction in AQ-Iand AQ-II stages, the decrease in the activity characterisedby Deact-1 and Deact-2, respectively were calculated as fol-lows:

Deact-1= (AQ-II , 1 h)

(AQ-I, 1 h)

Deact-2= (AQ-II , 2 h)

(AQ-I, 2 h)

Deact-3 presenting a decline in the activity during both theAQ-I and AQ-II stages was calculated from the number ofH2O2 moles determined after the first hour of reaction inthe Tetra-III and in the initial AQ-I stage:

Deact-3= (Tetra-III, 1 h)

(AQ-I, 1 h)

3. Results and discussion

3.1. Modification of alumina support with Na2SiO3 solution

Industrially produced�-Al2O3 containing 10 wt.% SiO2was used as the support in the present work. Homogeneousdistribution of Si, Al and Na in bulk of grains was deter-mined by X-ray microprobe analysis in SEM microscope(Fig. 1). Practically, the same values of Si/Al, Si/Na andAl/Na ratios determined over the cross-section of supportgrains in the core and near to the outer surface of the grainsare seen inTable 2. An additional amount of silica was intro-duced under pre-impregnation of the starting support withNa2SiO3 solution carried out with three Sh, Lg and Vc pro-cedures (Table 1). Treatment with Na2SiO3 resulted in everyprocedure in the increase in the support mass with simulta-neous decrease in the specific surface area and pore volume(Table 3). Partial blockage of the pores is evidenced by thepore distribution curves inFig. 2. The detailed analysis of

Table 2X-ray microprobe analysis data for the starting support and samplespre-impregnated by Na2SiO3, procedures Sh, Lg and Vc

Sample Si/Al Si/Na Al/Na

Starting supportOutside 0.109 24.5 226Core 0.129 27.5 214

ShOutside 0.209 1.97 9.4Core 0.085 2.47 29.6

LgOutside 0.23 3.26 14.07Core 0.08 2.43 30.14

VcOutside 0.25 2.6 10.2Core 0.11 3.13 28.3


Fig. 1. X-ray profile analysis. Distribution of Si, Al and Na over thecross-section of the starting support.

Table 3Influence of the type of pre-impregnation procedure (Sh, Lg and Vc) onthe properties of supports

Procedure Increaseof themass (%)

Specificsurface area(m2/g)

Pore volume(cm3/g)

Mean porediameter(Å)

Startingsupport

– 229 0.54 94

Sh 10 177 0.43 100Lg 11 180 0.43 95Vc 9.6 194 0.45 92

Fig. 2. Pore distribution in the starting support and supports pre-impre-gnated with Na2SiO3, procedure Sh, Lg and Vc.


their shape shows that besides the partial blockage of poresalso some new were created probably as a result of partialdissolution of alumina under alkaline Na2SiO3 solution (pH∼10–11). Especially distinct was the dissolution of Lg sam-ple, contacted 16 h with Na2SiO3 solution. The best block-age of pores was attained in the Vc sample, evacuated priorto the contacting with Na2SiO3 solution.

The penetration of silica into the bulk of support grainswas investigated by X-ray microprobe analysis over thecross-section of the grains. InTable 2, the results of quan-titative point analysis of the particular elements Si, Al andNa near to the outer surface of grains and in their core arepresented. As already described, all these elements werehomogeneously distributed through the bulk of the startingsample (Table 2). In all three Sh, Lg and Vc samples, anevident increase in the Si/Al ratio near to the outer surfaceis seen. Similar Si/Al are reached, but the highest one be-ing in the Vc sample. The difference between particularpre-impregnation procedures manifested also in the thick-ness of Si shell in the modified Sh, Lg and Vc supports(Fig. 3). The depth of Si penetration was found to dependon the method of support drying prior to the contacting itwith Na2SiO3 solution. In the Sh and Lg samples driedonly at 120◦C, the penetration of Si was rather limitedand the distinct layer of Si near to the outer surface wasformed (Fig. 3). The broader distribution of Si throughoutthe grains was formed in the Vc sample, evacuated prior toNa2SiO3 treatment. Hence, the latter Vc procedure resultedin the best efficiency of modification with silica species.

3.1.1. Characterisation of palladium catalystsEffective modification of the pore structure near the outer

surface allowed to prepare catalysts with egg-shell Pd con-centration profile in which penetration of Pd was stronglylimited. Incorporation of palladium resulted in a small in-crease in the specific surface area and porosity compared tothose prior to the insertion of Pd while the mean pore di-ameter slightly decreased (Table 4). It is not excluded thatthese effects were due to the alkalisation of catalysts samplesduring the hydrolysis of Pd2+ compound (see preparationof catalysts). Alkaline properties of Na2CO3 solution (pH∼10) may induce some changes in the pore structure (for-mation of new pores) in the same way as Na2SiO3 solution.

X-ray microprobe analysis (Table 4) of Pd concentrationprofile showed that the depth of Pd2+ penetration did notdepend of the type of procedure used in the modificationwith Na2SiO3 solution. Practically, the same penetration of

Table 4Influence of the type of pre-impregnation procedure (Sh, Lg and Vc) on the properties of palladium catalysts

Catalyst Specific surfacearea (m2/g)

Pore volume(cm3/g)

Mean porediameter (Å)

Depth of Sipenetration (�m)

Depth of Pdpenetration (�m)

1.04% Pd (Sh) 200 0.46 93 50 ∼401.05% Pd (Lg) 201 0.44 88 100 ∼401.09% Pd (Vc) 208 0.45 85 Wide, indistinct layer ∼40

Pd equal to only∼40�m (Fig. 3) can be seen at the samePd loading (∼1 wt.% Pd) in the catalysts with supportspre-impregnated with Sh, Lg and Vc procedures. Hence, theproperties of support have no significant influence on thedeposition profiles of Pd during the catalysts preparation.

3.2. Hydrogenation experiments

In Scheme 2, the plausible “reaction pattern” of eAQhydrogenation has been proposed taking into account theresults of our previous investigations[4] as well as infor-mation presented by other authors in scientific[1,3,5,11,15]and patent literature[10,16]. This “reaction pattern” issomewhat simplified in order to make easier discussion ofthe results obtained in the present work.Scheme 2showsthat eAQH2, the desirable product is in fact an intermediatebecause it is produced in eAQ reduction but consumed invarious side reactions. The course of eAQH2 consuming re-actions needs some explanation since degradation productsare then formed, compounds leading to the loss of startingeAQ. As already described, only the low content of degrada-tion products is acceptable by technology of anthraquinoneprocess.

In the consumption of eAQH2, two main reaction routes(i) and (ii) are distinguished in the present work. In the first(i), the consecutive process of aromatic rings hydrogenationeAQH2 → H4eAQH2 → H8eAQH2 proceeds. Saturationof the first phenyl ring produced H4eAQH2-active quinone.Hydrogenation of the second phenyl ring (H4eAQH2 →H8eAQH2) usually occurs only after eAQ had been com-pletely reduced to eAQH2 [3] (Scheme 1). Hence, eAQ hasa strong inhibitory effect on second phenyl ring saturation.Since in the present study eAQ was not fully reduced toeAQH2, only the first stage of this consecutive reactions cor-responding to the formation of H4eAQH2 could be takeninto account.

The second reaction pathway (ii) comprises the hy-drogenolysis reactions of C–O bonds and the tautomerisa-tion of eAQH2 to oxanthrone. All the compounds producedthen (2-ethylanthrone, 2-ethylanthracene, dianthrones) rep-resent degradation products (loss of quinone). As alreadyobserved [10], the changes in 2-ethylanthrone contentquite well correlated with the total degradation of quinoneon 0.3% Pd/�-Al2O3. Degradation products obtained viatautomerisation as dianthrone are not well defined or notdetectable by GC[3]. In the hydrogenation on Pd/Al2O3,the tautomerisation process was responsible for the loss of


Fig. 3. X-ray depth profile analysis of Si and Pd in the catalysts 1.04% Pd (Sh), 1.05% Pd (Lg) and 1.09% Pd (Vc).

about 5% of quinone[3]. Taking into account such effects,it may be assumed in the present work that the degradationproducts formed before complete hydrogenation of eAQ toeAQH2 are predominantly the products of reaction pathway(ii).

The mechanism of the irreversible slow deactivation ofPd/Al2O3 catalysts observed during fluid and fixed-bedhydrogenation has not yet been explained[10,11]. Partialblockage of Pd centres by bulky molecules produced onquinone molecules condensation has been considered asa probable effect. AsScheme 2shows, the formation ofdimers is most probable phenomenon along reaction path-way (ii) because dimerisation is a very characteristic featureof anthrone compounds[17]. Therefore, the course of par-allel reactions (i) and (ii) consuming eAQH2 is especiallytaken under consideration in the present work. The discus-sion is limited to the initial AQ-I stage, i.e. to the reaction

on a fresh catalyst. Comparison of such initial activity withthose in the subsequent AQ-II and Tetra-III stages provideddata about catalysts deactivation.

3.2.1. The effect of support modificationThree catalysts of similar Pd loading with differently mod-

ified supports 1.04% Pd (Sh), 0.81% Pd (Lg) and 0.70% Pd(Vc) were used in the hydrogenation experiments (Table 5).

Table 5Activity and deactivation data for Pd catalysts with supports pre-impregnated by Sh, Lg and Vc methods

Catalyst AQ-I, 1 h Tetra-III, 1 h Deact-1 Deact-2 Deact-3

1.04% Pd (Sh) 0.318 0.294 0.81 0.80 0.920.81% Pd (Lg) 0.235 0.153 0.45 0.46 0.650.70% Pd (Vc) 0.298 0.342 0.67 0.71 1.15


Their initial activities expressed as the number of H2O2moles determined after first hour of the initial AQ-I stage(AQ-I,1 h) are listed inTable 5. They are not very much dif-ferent thus showing similar rate of the main reaction: eAQto eAQH2 reduction (Scheme 1). However, together withthe latter eAQ reduction also eAQH2 consuming side reac-tions proceeded from the very beginning of the AQ-I stage.In Fig. 4a, dependence of the yield (Y) of eAQH2 as the

Fig. 4. The effect of pre-impregnation procedure. Hydrogenation in thepresence of 1.04% Pd (Sh), 0.81% Pd (Lg) and 0.70% Pd (Vc) catalysts:(a) the yield of eAQH2, (b) the content of H4eAQH2, and (c) the contentof degradation products as a function of eAQ to eAQH2 conversion (%R).

function of eAQ to eAQH2 conversion (%R) is presented.Since eAQH2 is an intermediate product (Scheme 2), itsyield has passed over the maximum valueYmax. The valueof Ymax as well as %R at which this value is reached, bothdepend on the relation between the rates of this intermediate(eAQH2) formation and consumption. The maximum yieldof eAQH2 is as high as 70–80% thus showing much higherrate of eAQH2 formation (in eAQ reduction,Scheme 1)compared to the rate of its consumption (parallel processesin Scheme 2). It is seen that all three curves inFig. 8 areclose to each other showing that over all three catalysts theprogress of earlier discussed reactions as well as their rel-ative rates were similar. However, if we compare the con-tent of H4eAQH2-product of reaction route (i) (Fig. 4b) andthat of degradation products produced in route (ii) (Fig. 4c),the difference in the performance of particular catalysts isclearly visible. Such a difference is manifested by variouscontribution of routes (i) and (ii) in the whole process ofeAQH2 consumption. H4eAQH2 was formed from the verybeginning of the AQ-I stage on all three catalysts. However,the content of H4eAQH2 was the highest on 0.7% Pd (Vc)catalysts. The latter catalyst produced degradation productsonly when more than 35% of eAQ was reacted to eAQH2(it corresponds toR = 35%) (Fig. 4c). Within the wholereaction the content of degradation products was smaller on0.7% Pd (Vc) than those produced on both 1.04% Pd (Sh)and 0.81% Pd (Lg). High ability of both latter catalysts to-wards rote (ii) manifested also in the formation of degra-dation products from the very beginning of the AQ-I stage(Fig. 4c).

In order to discuss difference in performance of individ-ual catalysts, it is convenient to introduce theNS coefficientdefined as the ratio of the number of H4eAQH2 moles{N(H4eAQH2)} to the number of moles of degradation prod-ucts{N (D)}.

NS = {N (H4eAQH2)}{N (D)}

For all three catalysts, theNS coefficient is higher than1 (Fig. 5b) thus showing predomination of route (i) phenylring hydrogenation over the hydrogenolytic processes de-scribed as route (ii). The value ofNS as high as∼17 isreached for the catalyst with support Vc, while for bothcatalysts with Sh and Lg supports, the smallerNS valuesequal to 2–4 are visible. Thus, on 0.7% Pd (Vc) catalysteAQH2 is mainly consumed owing to the phenyl ring hy-drogenation producing H4eAQH2-active quinone. As a con-sequence, very high content of the active quinones ACQ(where ACQ = eAQ + H4eAQ) practically equal to theinitial content of eAQ was reached only on the latter cat-alyst. (Fig. 5a). On the other side, in the presence of both1.04% Pd (Sh) and 0.81% Pd (Lg) catalysts the content ofactive quinones slowly decreased from the very beginningof the AQ-I stage owing to relatively high contribution ofhydrogenolytic reaction route (ii) in the eAQH2 consumingprocesses.


Fig. 5. The effect of pre-impregnation procedure. Hydrogenation in thepresence of 1.04% Pd (Sh), 0.81% Pd (Lg) and 0.70% Pd (Vc) catalysts:(a) the content of active quinones, and (b)NS coefficient as a functionof eAQ to eAQH2 conversion (%R).

Activity of all three catalysts decreased after the initialAQ-I stage of the run (Table 5). As a measure of catalystdeactivation, the ratio of the activity in the AQ-II stage tothe initial one in the AQ-I stage characteristic for the freshcatalyst was assumed. Decline in the activity was calculatedafter the first and the second hour (Deact-1 and Deact-2,respectively) of the experiment. Different decrease in theactivity of individual catalysts given by Deact-1 is seen,changing from 0.81 to 0.46. Because of similarity in theDeact-1 and Deac-2 no further evident decrease in activityoccurred during AQ-II stage.

In last Tetra-III stage of the run, solution of the composi-tion corresponding to the All-Tetra system was used (70%H4eAQ, 30% eAQ). It was determined in separate seriesof hydrogenation experiments that activity of fresh cata-lysts in the “All-Tetra” system was 1.1–1.2 times higherthan the one determined in the fresh eAQ solution. Thus, ifcatalyst activity did not permanently decrease during bothAQ-I and AQ-II stages, activity determined in Tetra-IIIstage should be higher than the initial one and the valueof Deact-3 higher than 1 should be reached. As the data inTable 5show, Deact-3 > 1 wasattained only for 0.7% Pd(Vc).

Hence, only the catalyst with Vc support offered re-versible deactivation under the present hydrogenation con-ditions. Although activity of the latter somewhat decreasedduring the initial AQ-I stage, it was almost fully restored inthe Tetra-III stage. In contrast, activity of both 1.04% Pd(Sh) and 0.81% Pd (Lg) decreased permanently during thewhole hydrogenation run thus showing more irreversibledeactivation of both latter samples. Deact-3 below 1 con-firms that their activities were not restored in the Tetra-IIIstage as was the activity of 0.7% Pd (Vc).

In all three catalysts, the thickness of palladium shell wasvery similar (Fig. 3) while the main difference was the effi-ciency of support modification with silica species. This effi-ciency was the best in the 0.7% Pd (Vc) catalyst. The highestSi/Al ratio near the outer surface was reached in this sam-ple (Table 2) as well as Si penetrated more deeply insidethe grains of Vc support. Therefore, it may be supposed thatthe difference in the performance of all three catalysts canbe related with the efficiency of support modification withSi species. Such an efficiency did not influence the activityin the main eAQ to eAQH2 reduction (Fig. 4a) but it ex-hibited an essential role on the course of eAQH2 consump-tion along parallel reactions (i) and (ii). The latter route (ii)hydrogenolytic reaction are then restricted (Fig. 4c). As aconsequence, among three catalysts samples of similar Pdprofile but various Si penetrations, the reversible deactiva-tion offered only 0.7% Pd (Vc) catalyst. Procedure Vc wastherefore used to prepare the catalysts of various Pd loading(0.7–2 wt.% Pd).

3.2.2. The effect of Pd content in the catalystPd concentration profiles in the catalysts of various Pd

loading prepared using support pre-impregnated with Vcprocedure are presented inFig. 6. The thickness of Pd shellevidently depends on the Pd loading. It slightly grows whenthe content of Pd does increase (Fig. 6), from ca. 35�m atthe lowest Pd content (0.7%) up to ca. 65�m at the highestPd loading equal to 2%. In all the catalysts, a broad andindistinct Si shell appeared indicating deep penetration ofsilica species.

All the catalysts of various Pd loading exhibited practi-cally the same initial activity. As the data inTable 6show,almost the same number of H2O2 moles (AQ-I, 1 h) was de-termined after the first hour of the AQ-I stage. Thus, mostlypalladium located on the very external surface of catalyst’sgrains takes part in the hydrogenation. Consequently, the

Table 6Activity and deactivation data for catalysts of various Pd loading, supportpre-impregnated with Vc procedure

Pd (%) AQ-I, 1 h Tetra-III, 1 h Deact-1 Deact-2 Deact-3

0.70 0.298 0.342 0.67 0.71 1.151.09 0.294 0.68 0.761.20 0.330 0.353 0.78 0.79 1.071.60 0.291 0.342 0.81 0.82 1.172.00 0.281 0.377 0.68 0.55 1.34


Fig. 6. Depth profile analysis of Pd in the catalysts with support pre-impregnated by Na2SiO3, procedure Vc.

internal surface of catalyst (Pd located inside of grains) isnot efficiently utilised in the main eAQ to eAQH2 reaction.This effect shows that this main reaction is in the presentconditions controlled by the diffusion processes. This con-clusion is in good agreement with literature data[4,5,11,18]showing that Pd-supported catalysts are extremely active inthe anthraquinone hydrogenation and the reaction is alwayscontrolled by mass transfer processes.

Fig. 7a shows that when Pd loading in the catalyst in-creases, the maximum yield of eAQH2 (Ymax) slowly de-creases from 78% for catalyst with 0.7% Pd to the 65% forthe sample with 2% Pd. As already described, the maximumyield Ymax of intermediate eAQH2 depends on the relationbetween the rates of its formation and that of its consump-tion. Since the rate of eAQH2 formation is practically thesame (very similar values of (AQ-I, 1 h) inTable 6), higherPd loading resulted in the increased rate of eAQH2 consum-ing reactions. It can be seen inFig. 7b and cthat high Pdloading especially promotes route (ii) of eAQH2 consump-tion leading to the degradation products. The content ofdegradation products distinctly grow with the increase in Pdloading whereas the content of H4eAQ was practically thesame. Consequently, theNS coefficient as high as 17 at 0.7%Pd decreased to 1–1.5 for catalysts with 1.6–2% Pd (Fig. 8b).In the former catalyst, phenyl ring hydrogenation (route (i))evidently predominated over the hydrogenolysis reactions,while in the latter such a predomination is distinctly smaller.The increased contribution of route (ii) resulted in a slow de-crease in the content of active quinones from the very begin-ning of reaction on 1.6–2% Pd-containing catalysts (Fig. 8a).

Thus, high Pd loading promotes processes consumingeAQH2 and especially the hydrogenolytic reactions along

route (ii) are accelerated. It was observed that when theloading of Pd increased the depth of palladium penetrationalso increased (Fig. 6). Hence, the formation of degradationproducts proceeded also inside the catalyst pores and moreinternal Pd was used for converting eAQH2 to degradationproducts.

Activity of all catalysts decreased during the first AQ-Istage thus showing partial deactivation, very similar for cat-alysts of various Pd loading (Table 6). However, they offeredsomewhat higher activity in last “Tetra-III” stage than theinitial ones (Deact-3 > 1). Hence, all the catalysts with Vccarrier, independently of Pd loading only reversibly deacti-vated during the hydrogenation run.

In conclusion, performance of the present catalysts de-pended on the efficiency of alumina modification with Sispecies as well as the thickness of Pd shell in the grainsof catalysts. Among three Sh, Lg and Vc procedures, thebest efficiency with silica modification has been reached inthe Vc support evacuated prior to incorporation of Na2SiO3solution. It manifested in the high surface concentration ofsilica and in the broad silica shell inside the grains. Verysimilar thickness of Pd shell (ca. 40�m, Fig. 3) in the cat-alysts of similar Pd content but with various Sh, Lg andVc supports suggests that the properties of support have nosignificant influence on the deposition profiles of Pd duringcatalysts preparation. Penetration of palladium was found todepend on Pd loading in the catalysts. At higher Pd contentsuch a penetration was deeper.

The depth of Pd penetration is of primary importancebecause it promotes the formation of degradation products.Deep penetration of palladium at high Pd loading (1.6–2%Pd) in the catalysts with Vc support has no essential


Fig. 7. The effect of Pd loading in catalysts on the course of eAQhydrogenation: (a) the yield of eAQH2, (b) the content of H4eAQH2, and(c) the content of degradation products as a function of eAQ to eAQH2

conversion (%R).

influence on the rate of the main eAQ to eAQH2 reduction,whereas it accelerates route (ii) of eAQH2 consumptionresulting in the degradation products. Although degradationproducts were formed at relatively high content, activityof all catalysts with Vc support decreased only reversiblyduring the hydrogenation run. On the other hand, despitevery thin Pd shell in the catalysts with Sh and Lg carriers,their activities decreased irreversibly during the whole cat-alytic experiment. Better activity maintenance offered bythe catalysts with Vc support revealed that efficiency with

Fig. 8. The effect of Pd loading in catalysts on the course of eAQhydrogenation: (a) the content of active quinones, and (b)NS coefficientas a function of eAQ to eAQH2 conversion (%R).

silica modification is indeed an important variable affectingperformance of the studied catalysts.

The observed phenomenon showing a profitable role ofsilica species may be interpreted by modification of the sur-face properties. It has already been reported[19] that be-tween anthraquinone molecules and not fully coordinatedAl3+ ions in alumina the charge transfer coordination com-pounds are formed leading to the strong adsorption of an-thraquinone. On the other side, in the case of silica onlyweak, physical adsorption of anthraquinone occurred[19].It is not excluded that pre-impregnation of the starting alu-mina with Na2SiO3 resulted in modification of adsorptionproperties of the surface towards anthraquinone molecules.Although such an effect seems very probable, further in-vestigations especially concerning the surface properties ofcatalysts are needed.

References

[1] Ulmans Encyclopedia of Industrial Chemistry, vol. A13, VCH, Wein-heim, 1989, p. 443.

[2] C.S. Cronan, Chem. Eng. 6 (1959) 118.[3] E. Santacesaria, M. di Serio, R. Velotti, U. Leone, J. Mol. Catal. 94

(1994) 37.


[4] A. Drelinkiewicz, J. Mol. Catal. 101 (1995) 61;A. Drelinkiewicz, J. Mol. Catal. 75 (1992) 321.

[5] T. Berglin, N.H. Shon, Ind. Eng. Chem. Process, Des. Dev. 22 (1983)150.

[6] UK Patent 876 459 (1961).[7] UK Patent 858 035 (1961).[8] US Patent 3 615 207 (1971).[9] US Patent 4 061 598 (1977).

[10] US Patent 4 800 075 (1989).[11] E. Santacesaria, M. di Serio, R. Velotti, U. Leone, Ind. Eng. Chem.

Res. 33 (1994) 277.[12] A. Drelinkiewicz, A.Pukkinen, R. Kangas, R. Laitinen, Catal. Lett.,

submitted.

[13] J.G. Fraser, F.E. Beamish, W.A. McBryde, Anal. Chem. 26 (1954)495.

[14] A. Drelinkiewicz, Bull. Pol. Acad. Chem. 43 (1995) 51.[15] E. Santacesaria, P. Wilkinson, P. Babini, S. Carra, Ind. Eng. Chem.

Res. 27 (1988) 780.[16] US Patent 789 115 (1974).[17] J. Houben, das Anthracene und die Anthrachinone, G. Thieme Verlag,

Leipzig, 1929.[18] S.W. Sjawcillo, W.I. Sawuszkina, E.M. Zhernovskaya, Zh. Obsh.

Khim. 28 (1959) 1752.[19] W.W. Lunin, G.L. Markarjan, O.W. Cetina, DAN SSSR 264 (1982)

884.

Hydrogenation of 2-ethylanthraquinone on alumina-supported palladium catalysts

Documents