Journal of Catalysis 361 (2018) 51–61

Contents lists available at ScienceDirect

Journal of Catalysis

journal homepage: www.elsevier .com/locate / jcat

The role of water in the reusability of aminated silica catalysts for aldolreactions

https://doi.org/10.1016/j.jcat.2018.02.0160021-9517/� 2018 The Authors. Published by Elsevier Inc.This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

⇑ Corresponding author.E-mail address: Joris.Thybaut@UGent.be (J.W. Thybaut).

Anton De Vylder a, Jeroen Lauwaert b, Dolores Esquivel c, Dirk Poelman d, Jeriffa De Clercq b,Pascal Van Der Voort e, Joris W. Thybaut a,⇑a Laboratory for Chemical Technology (LCT), Department of Materials, Textiles, and Chemical Engineering, Ghent University, Technologiepark 914, 9052 Ghent, Belgiumb Industrial Catalysis and Adsorption Technology (INCAT), Department of Materials, Textiles, and Chemical Engineering, Ghent University, Valentin Vaerwyckweg 1, 9000Ghent, BelgiumcDepartamento de Química Orgánica, Instituto Universitario de Investigación en Química Fina y Nanoquímica (IUIQFN), Facultad de Ciencias, Universidad de Córdoba, Campusde Rabanales, Edificio Marie Curie, E-14071 Córdoba, Spaind LumiLab, Department of Solid State Sciences, Ghent University, Krijgslaan 281-S1, 9000 Ghent, BelgiumeCenter for Ordered Materials, Organometallics and Catalysis (COMOC), Department of Chemistry, Ghent University, Krijgslaan 281-S3, 9000 Ghent, Belgium

a r t i c l e i n f o

Article history:Received 20 December 2017Revised 13 February 2018Accepted 14 February 2018

Keywords:Cooperative catalysisAminated silicaReusabilityAldol reactionDeactivationEnamine

a b s t r a c t

The reusability of methylaminopropyl active sites grafted on mesoporous amorphous silica, either withcooperative silanol groups or trimethylsilated, was assessed in the aldol reaction of acetone with4-nitrobenzaldehyde. Raman, 13C NMR, and UV–Vis spectroscopy demonstrated the presence of stableenamines on the spent catalysts. These enamines are produced as a side product from iminium interme-diates in the catalytic cycle. Water co-feeding enhances the desorption of the iminium intermediates and,hence, suppresses the formation of these stable enamine species. The reusability of the cooperativecatalyst increased to 70% with co-feeding 0.69 wt% water, while an almost complete reusability wasachieved for the trimethylsilated catalyst. Continuous-flow experimentation showed that the cooperativeeffect of the silanol groups was lost during the first 7 h on-stream, yet activity losses continued, mostlikely due to silica hydrolysis. Activity losses persisted on the more hydrophobic trimethylsilated catalyst,but were significantly less pronounced.� 2018 The Authors. Published by Elsevier Inc. This is an open access article under theCCBY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Aldol reactions constitute an important class of reactions toproduce new carbon-carbon bonds in a wide range of processes,such as in the production of chalcones in the pharmaceuticalindustry [1] and 2-ethylhexanol for PVC plasticizers [2]. A potentialnovel application in the bio-based industry is the production of liq-uid fuels from lignocellulosic biomass by upgrading smaller furanicmolecules, such as 2-furaldehyde (furfural) and 5-(hydroxymethyl)furfural (HMF), to heavier hydrocarbons [3]. At present, commer-cial scale aldol reactions are catalyzed by a homogeneous, strongbase catalyst, such as NaOH, which permits to perform the reactioneven at room temperature [4]. However, these catalysts also exhi-bit several important disadvantages, such as a short lifetime, anenvironmental risk, equipment corrosion and an energy intensiveseparation from the reaction products [5,6]. Hence, the pursuit ofa reusable, heterogeneous catalyst could lead to a more sustainable

alternative production route. In this sense, either an acidic or abasic heterogeneous catalyst can be employed.

Heterogeneous acid catalysts, such as the widely employedzeolite-type materials, have already been investigated for liquid-phase aldol reactions [7–9]. However, unlike base catalysts, theBrønsted acid function in zeolites promotes undesired side reac-tions such as isomerization and cyclodehydration [9]. Moreover,the formation of large condensation products inside the microp-ores causes rapid pore-blocking and subsequent deactivation[10]. Development of zeolites with larger mesopores reduces thecatalyst’s susceptibility to coking, but also decreases its acidstrength [8]. Hence, acidic zeolite-type materials appear not idealas reusable, sustainable heterogeneous catalysts for liquid-phasealdol reactions.

Candidate heterogeneous base catalysts for liquid-phase aldolreactions comprise mixed oxide materials such as hydrotalcitebased catalysts [11–17], ion-exchange resins [18], and cooperativeacid-base silica catalysts [19–37]. The mixed oxide and hydrotal-cite materials typically suffer from fast deactivation by strongadsorption of reagents and ambient CO2 during storage. Complete

52 A. De Vylder et al. / Journal of Catalysis 361 (2018) 51–61

regeneration of its basic character is difficult [14,15]. Additionally,side-reactions such as the Cannizzaro reaction, typically occur to asignificant extent on such catalysts [38]. Anion-exchange resins arecomposed of quaternary ammonium ions interacting with freeOH� ions, which are the actual active sites. These catalysts are alsosubject to deactivation, caused by neutralization of the OH� ionsand pore-blocking [39,40]. Aminated mesoporous silica materials,either ordered or amorphous, possess large mesopores and havebasic amine sites [41]. Despite a typically lower activity, thesematerials are less prone to pore-blocking by large condensedproducts.

The idea for using amines in aldol reaction reactions originatesfrom natural aldolase enzymes, which catalyze aldol reactions viaan enamine-type mechanism [42]. List et al. [43] employed theamino acid L-proline as a homogeneous catalyst in the aldol reac-tion of 4-nitrobenzaldehyde with acetone in dimethylsulfoxide(DMSO). They proposed a cooperative effect between the acidiccarboxylate and the basic amine: the acid function promotes thenucleophilic attack of the basic amine on this carbonyl group,and assists in proton-transfers [43]. A heterogeneous amine cata-lyst can be synthesized by functionalizing aminosilanes on silica[19–35,44]. The weakly acidic silanol groups, which are intrinsi-cally present at the silica surface, act as the cooperative partnerin the reaction mechanism, equivalent to the cooperative effectof the carboxyl group in L-proline [28,37]. Secondary amines arepreferred, since primary amines allow the formation of a stableinhibiting imine species, and tertiary amines do not allow the for-mation of the crucial enamine intermediate [22,23]. However, careshould be taken that the substituent on the secondary amine is nottoo large, in order to minimize steric hindrance when approachingthe nitrogen lone electron pair [24]. It has been shown that amethyl substituted secondary amine with a propane linker on amesoporous amorphous silica is an efficient catalyst for the aldolreaction when silanol groups are neighboring the amine [25–27].Previously it was found that, when the amines are grafted ran-domly on the surface, at least 1.7 of these silanol groups arerequired per amine function to assure full promotion [29]. Increas-ing the acid strength of the promoting acid function has beenobserved to decrease the catalytic activity [30–33]. This can beexplained by an unfavorable shift in the equilibrium of the freeacid base couple to the neutralized one [30]. Hence, currently themost active aminated mesoporous amorphous silica catalyst foraldol reactions is a secondary amine, with a small substituent suchas a methyl group, promoted by an excess of silanol groups.

Not only the activity, but also the catalyst stability is an impor-tant factor to take into account when assessing the commercialpotential of a catalyst [45]. This means that a catalyst should exhi-bit the same level of activity throughout its lifetime. However, dueto the nature of a batch reactor, deactivation cannot be observedseparately from the normal time-evolution in the course of a singleexperiment. Hence, recycling of the spent catalyst in a subsequentexperiment should be performed. Opposed to a batch-type reactor,a continuous-flow reactor can be used to observe catalyst deactiva-tion directly by measuring the catalytic activity with time-on-stream. This allows for more detailed insights into the specificdeactivation behavior and corresponding mechanism. For aldolreactions using homogeneous L-proline, Zovota et al. [46] identi-fied off-cycle species such as oxazolidinones and oxapyrrolizidines,using H-NMR, when insufficient water is present. We have alsorecently reported that an ethylene-bridged ordered periodic meso-porous organosilica material, functionalized with cysteine, exhibitsa lower turnover frequency (TOF) in a recycle experiment [35].Shylesh et al. [23] have performed gas-phase self-aldol reactionof butanal in a continuous-flow reactor on a cooperative acid-base mesoporous amorphous silica catalyst, but reported no infor-mation about catalyst deactivation with time-on-stream. Clearly

no detailed information about the deactivation behavior of ami-nated silica materials has been reported yet for aldol reactions.

In the present work, our aim is to systematically investigate thephenomena potentially involved in catalyst deactivation in hetero-geneous amine catalyzed aldol reactions. Consecutive experimentsin a batch-type reactor have been performed as well as time-on-stream experiments in a continuous-flow reactor. Using a varietyof characterization techniques combined with mechanisticinsights, conditions have been identified whereby the recyclabilityof the aminated mesoporous amorphous silica catalysts for thealdol reaction approaches 100%.

2. Experimental procedures

2.1. Synthesis of the cooperative acid-base catalyst and themonofunctional base catalyst

The procedure for grafting secondary amine functional groupson mesoporous amorphous silica, and the endcapping of the silanolgroups with trimethylsilyl groups, was reported more elaboratelybefore [24] and is briefly summarized here. Silica gel 60 (grade7734, Sigma-Aldrich) was used as a catalyst support material. Forexperiments in the batch reactor, a pellet size range between 63mm and 210 mm was employed. For the continuous-flow reactor,the pellet size was increased to a range between 250 mm and500 mm to avoid a significant pressure drop over the catalyst bed.The support material was first heated at a rate of 2 �C/min and cal-cined in air at 700 �C for 6 h. After calcination, the material wasmaintained at a temperature of 150 �C to avoid adsorption of mois-ture. Next, 5 g of the hot support material was suspended in 30 mLof toluene (Sigma-Aldrich, anhydrous, 99.8%). The amine precursor,N-methylaminopropyltrimethoxysilane (MAPTMS, ABCR), wassubsequently added with a molar ratio of precursor to free silanolgroups of 0.25. The necessary volume of precursor is calculatedassuming that, because of the employed calcination procedure,the number of free silanol groups is equal to 1.1 OH/nm2 [21]. Toensure full promotion of each randomly grafted amine site by thesilanol groups, at least 1.7 silanol groups are required per aminesite [29]. Hence, in this work a ratio of 4 free silanol groups for eachmethylaminopropyl function was employed. After adding the pre-cursor, the mixture was refluxed for 24 h at 110 �C under an argonatmosphere. The obtained catalyst was then filtered and subse-quently stirred in 250 mL chloroform (>99.8%, <50 ppm H2O, CarlRoth) for 4 h. After vacuum drying for 24 h at room temperature,the cooperative acid–base catalyst was ready to be used in catalyticexperiments. A desiccator was used for storage of the catalysts inorder to avoid modifications due to the presence of atmosphericwater vapor.

To obtain the monofunctional base catalyst, the silanol groupsof the cooperative material were endcapped with trimethylsilylgroups [29,47]. For this, 5 g of the dried cooperative acid-base cat-alyst was completely covered with 100 mL of 1,1,1,3,3,3-hexamethyldisilazane (99%, HMDS, ABCR). The mixture was then stirredat room temperature for 4 h, subsequently filtered, and stirred in250 mL chloroform (>99.8%, <50 ppm H2O, Roth) for 4 h. After vac-uum drying for 24 h at room temperature, the monofunctionalbase catalyst was ready to be used in catalytic experiments. Thiscatalyst was also stored in a desiccator.

2.2. Characterization of the fresh and spent catalysts

Nitrogen adsorption-desorption measurements were carriedout at �196 �C, using a Micromeritics Tristar II surface area andporosity analyzer. The specific surface area and pore volume weredetermined using the Brunauer–Emmett–Teller (BET) method. The

A. De Vylder et al. / Journal of Catalysis 361 (2018) 51–61 53

average pore size and pore volume was obtained using the Barrett–Joyner–Halenda (BJH) method.

The concentration of amine groups was determined via elemen-tal (CHNS) analysis. These measurements were performed on aThermo Flash 2000 elemental analyzer using V2O5 as catalyst, inorder to ensure the total oxidation of the sample. The mass percentof nitrogen in the sample is obtained by referring the obtainedpeak area to a calibration curve of methionine (USP, 99%) thatwas obtained prior to the measurements. Conversion to the con-centration of active sites, in mol per gram, was then performedwith the molar mass of nitrogen.

The presence of organic species on the catalyst surface, bothbefore and after reaction, was determined with Diffuse ReflectanceInfrared Fourier Transform (DRIFT) spectroscopy, as well as Ramanspectroscopy. The DRIFT spectra were recorded using a Nicolet6700 by Thermo Scientific with a liquid nitrogen cooled MCT-Adetector, using a Graseby Specac diffuse reactant cell, operatingin vacuum at 140 �C. The Raman spectroscopy was performed ona NXR FT-Raman module of Thermo Scientific, with a laser of1064 nm and an InGaAs detector. A resolution of 4 cm�1 was usedon a total of 512 scans per sample at a laser intensity of 0.31 W.

The 13C CP/MAS NMR spectra were recorded at 100.61 MHz on aBruker Avance III HD 400 WB NMR spectrometer at room temper-ature. An overall of 1000 free induction decays were accumulated.The excitation pulse and recycle time were 6 ms and 2 s. Chemicalshifts were measured relative to tetramethylsilane standard. Allsamples were dried at 120 �C for 24 h prior to analysis.

UV–Vis diffuse reflection measurements were performed usinga Varian Cary 500 UV–Vis-NIR spectrophotometer equipped withan internal BaSO4-coated integrating sphere.

2.3. Determination of the water content in stock organic solvents

To quantify the amount of water in the organic solventdimethylsulfoxide (DMSO, �99%, FG, Sigma-Aldrich) and acetone(99.6%, Acros), Karl-Fischer titrations were performed on a MKC-501 coulometric cell with diaphragm (Kyoto Electronics). ForDMSO, anolyte solution Hydranal Coulomat AG (Sigma-Aldrich)and catholyte solution Hydranal Coulomat CG (Sigma-Aldrich)were used. For acetone, methanol-free anolyte solution HydranalCoulomat AK (Sigma-Aldrich) and methanol-free catholyte solu-tion Hydranal Coulomat CG-K (Sigma-Aldrich) were used.

Karl-Fischer titrations indicated an average water content instock acetone of 0.31 ± 0.05 wt% and in DMSO of 0.09 ± 0.03 wt%.Both acetone and DMSO were stored under an argon atmospherein order to minimize any absorption of ambient water. It was con-firmed that storage under these circumstances during the course ofthe experimentation did not significantly change the amount ofwater in both DMSO as well as in acetone.

2.4. Drying of stock acetone

In order to further decrease the amount of water in the reactor,some experiments were performed with dried acetone. To removewater, 100 mL acetone was poured on 10 g freshly regenerated 3 Åmolecular sieves (Sigma-Aldrich) and thoroughly mixed for onehour. This resulted in a 3-fold decrease in the water content to0.09 ± 0.02 wt%.

2.5. Catalytic performance evaluation of the fresh and spent catalystsin the batch reactor

The catalyst performance was initially evaluated in a batch-typereactor (Parr 4560 mini, 300 mL). The detailed procedure has beenreported previously [24] and is also briefly summarized here. Theconsidered benchmark reaction is the aldol reaction of acetone

(99.6%, Acros) with 4-nitrobenzaldehyde (99%, Acros), resultingin 4-hydroxy-4-(4-nitrophenyl)butan-2-one as aldol adduct and4-(4-nitrophenyl)-3-buten-2-one as condensate, see Scheme 1.

To start an experiment, the reactor was first loaded with anamount of catalyst equivalent with 0.1 mmol active amine sites,51.7 to 55.0 g DMSO as solvent, 0.5 g to 3 g water added(double deionized, Milli-Q�), and 0.25 g methyl 4-nitrobenzoate(>99%, Sigma-Aldrich) as internal standard. This mixture was thenheated to 55 �C, under constant stirring. Acetone (45 g) wasseparately heated to 55 �C and was used to dissolve 0.45 g4-nitrobenzaldehyde before injecting in the reactor. Acetone wasadded in excess, in order to suppress the direct interaction of4-nitrobenzaldehyde with the amine site. The amount of solventwas adapted according to the amount of water that is added tothe reactor in order to avoid effects of a concentration change onthe measured kinetics. The temperature in the reactor was main-tained at 55 �C using a PID-controller (CAL 9500P), with a thermo-couple inside the reactor vessel. This PID controller steered both anelectrical heating jacket as well as a cooling unit for an adequatetemperature control. The reaction mixture was mechanically stir-red at 600 rpm. The time of injection was taken as the start ofthe reaction (t = 0). The reaction was monitored for 4 h by takinga sample (0.3 mL) of the reaction mixture every 15 min duringthe first hour, and subsequently every 30 min for the remaining3 h. For each experiment, the total decrease of reaction volumedue to sampling was less than 5% whereby the effect of samplingon the kinetic data is considered to be negligible.

After the reaction, the spent catalyst was removed from thereactor, filtered, and stored in a desiccator, in order to avoidatmospheric water vapor from interacting with the catalyst. Allrecycling experiments were performed by repeating the proce-dures described above using the recovered spent catalyst withinmaximum 24 h after the previous run. In order to characterizethe surface intermediates on the spent catalysts with Raman spec-troscopy and 13C NMR, the catalysts were additionally stirred in100 mL chloroform (>99.8%, <50 ppm H2O, Carl Roth) for 4 h,filtered, and vacuum dried for 24 h.

2.6. Catalytic performance testing in a continuous-flow liquid-phasereactor

Stability tests were performed in a liquid-phase packed-bedtubular reactor, which was developed in-house. A continuouslygently stirred liquid feed mixture was fed at the top of the reactorwith a HPLC pump (Eldex Laboratories model 2SMP). The reactortube itself is made of stainless steel 316 with an inner diameterof 80 mm and length of 300 mm. At 100 mm from the bottom, afine stainless steel mesh is installed to support the catalyst bed.The reactor is electrically heated, with its temperature being con-trolled at 55 �C by an outer thermocouple with an Omron E5CKPID controller. The temperature in the catalyst bed was measuredwith an internal thermocouple of 1 mm and allowed confirmingthe absence of radial temperature gradients. The backpressure isgenerated in the effluent section using a 1.8 barg check valve.Manometers at the entrance and outlet of the reactor allowedverifying the absence of a significant pressure drop over thereactor. A three-way valve after the backpressure valve allowedto periodically draw samples from the reactor effluent. The reactorwas operated in the intrinsic kinetics regime, as has been verifiedby the proper correlations [48], and is summarized in the Support-ing Information.

The feed mixture composition was equivalent to the batch reac-tor feed composition, consisting of 0.45 wt% 4-nitrobenzaldehyde,44.6 wt% acetone, 52.5 wt% DMSO, 0.25 wt% methyl4-nitrobenzoate (internal standard) and 2.19 wt% water. The activ-ity of the cooperative acid-base catalyst and the monofunctional

Scheme 1. Aldol reaction of acetone and 4-nitrobenzaldehyde towards the aldol adduct 4-hydroxy-4-(4-nitrophenyl)butan-2-one, and dehydration to the condensationproduct 4-(4-nitrophenyl)-3-buten-2-one.

Table 1Catalyst properties, for both the cooperative acid-base catalyst as well as themonofunctional base catalyst, fresh and spent in the aldol reaction of acetone with 4-nitrobenzaldehyde. BET surface area, pore size and pore volume are determined vianitrogen adsorption–desorption measurements.

BET surfacearea ±30 (m2/g)

Pore size±0.3 (nm)

Pore volume±0.04 (cm3/g)

Silica gel 60 451 5.7 0.70Cooperative acid-base catalystFresh 314 5.6 0.52Spent 296 5.6 0.48

Monofunctional base catalystFresh 272 5.4 0.45Spent 280 5.2 0.44

54 A. De Vylder et al. / Journal of Catalysis 361 (2018) 51–61

base catalyst were compared using the conversion at the same sitetime. Site time is defined as the space time W

F04�NB, multiplied with

the active site concentration Csite. With W ðkgcatÞ the catalyst mass

and F�4-NBmol4�NB

s

� �the molar inlet feed of 4-nitrobenzaldehyde and

Csitemolsitekgcat

� �determined with CHNS analysis, site time has units

molsitesmol4�NB

� �.

2.7. Sample analysis, quantification of effluent composition, anddefining the catalyst activity

The reaction samples were analyzed using a reversed-phasehigh-performance liquid chromatograph (RP-HPLC), from Agilent(1100 series) using the same procedure as reported before [29].The HPLC Eclipse XDB-C18 column was operated at 30 �C using agradient method with water (0.1% trifluoroacetic acid, Acros) and30 v% to 62 v% acetonitrile (HPLC grade, Acros) as solvents. Thismethod separates all the components in a period of 14 min. Thecomponents were identified using a UV-detector with a variablewavelength that has been programmed for an optimal absorptionfor each component. Quantification of the different componentsin the reaction mixture was performed by relating the surfaceareas of the component peaks to the amount of internal standardadded to the reactor. The catalyst activity in the batch reactorwas quantified by the turnover frequency, which was determinedfrom the slope of the initial linear part of the conversion of 4-nitrobenzaldehyde as a function of reaction time, the concentra-tion of active amine sites and the initial concentration of 4-nitrobenzaldehyde. The catalytic activity in the continuous-flowreactor was quantified by comparing conversions at the same sitetime. Repeat experiments have indicated that the experimentalerror in both reactors, calculated as the 95% confidence interval,was below 5%. With the rule of error propagation, it was deter-mined that ratios of the TOF values in the batch reactor have anexperimental error of about 10%.

3. Results and discussion

3.1. Catalyst characterization and validation of the synthesisprocedure

The specific BET surface area, average pore size, and total porevolume of the silica gel support and the synthesized catalysts weredetermined from the nitrogen adsorption–desorption isothermsand are summarized in Table 1. The functionalization of the pris-tine support material, and further endcapping of the silanol groupswith trimethylsilyl groups, caused a small decrease in surface areaand pore volume. This decrease can be directly related to adecrease in free volume and an increase in catalyst mass, whichis typically observed after the functionalization of these materials[49]. The catalyst, however, retained its mesoporous structureand average pore size of 5 nm after functionalization and endcap-ping of the silanol groups.

The presence of methylaminopropyl groups on the surface wasverified with Diffuse Reflectance Infrared Fourier Transform(DRIFT) spectroscopy, as can be seen in Fig. 1S in Supporting

Information. In the pristine silica gel support material, a pro-nounced narrow peak at 3745 cm�1 indicated the presence of freesilanol groups on the surface. After reaction with N-methylaminopropyltrimethoxysilane (MAPTMS), several bands appeared in theregion between 2800 cm�1 and 3000 cm�1, characteristic of theCAH stretching vibrations of the methylaminopropyl groupattached to the surface [29]. At 3305 cm�1 a small, broad peak indi-cates the presence of N-H vibrations. The peak that remains at3745 cm�1 confirms that free silanol groups were still present aftergrafting. When the catalyst was treated with HMDS to endcap theresidual silanol groups, the peak at 3745 cm�1 effectively disap-peared, as is shown in Fig. 1S. The endcapping of the silanol groupsresulted in an increased intensity of the CAH stretching vibrationsat 2960 cm�1 from the trimethylsilyl groups that replaced the sila-nol groups. Using CHNS analysis, the amine loading was deter-mined to be 0.17 mmol/g. Using HMDS as a rough silanol titrator,the amount of silanol groups was estimated at 0.69 mmol/g. Thisindicates that there are at least 4 silanol groups available for eachamine site in the cooperative acid-base catalyst, as was targetedduring the catalyst synthesis.

3.2. Catalytic performance in the aldol reaction of acetone with 4-nitrobenzaldehyde

The catalytic reaction performed in this work is the enaminecatalyzed aldol reaction of acetone with 4-nitrobenzaldehydetowards the primary aldol adduct 4-hydroxy-4-(4-nitrophenyl)butan-2-one, followed by dehydration to the secondary condensationproduct 4-(4-nitrophenyl)-3-buten-2-one (Scheme 1). The pro-posed reaction mechanism using a methyl substituted secondaryamine promoted by surface silanol groups is given in Scheme 2.The catalytic cycle starts with the formation of a hydrogen bondbetween the free silanol group (1) and the oxygen of the carbonylgroup of acetone (2). Next, the free electron pair of the amineattacks the activated carbonyl group of acetone, and a carbino-lamine (3) is formed via a proton-transfer, assisted by the surfacesilanol group. This carbinolamine dehydrates, yielding an enamineintermediate (4), which then attacks the carbonyl group of an acti-vated 4-nitrobenzaldehyde molecule and forms a new carbon-carbon bond. The formed iminium intermediate (5) is hydrolyzedtowards the product carbinolamine (6) and, subsequently, desorbsfrom the active site as the aldol adduct. When the silanol groups

(1)

(2) (3)

(4)

(5)(6)

Scheme 2. Proposed enamine catalytic reaction cycle of the aldol reaction of acetone with 4-nitrobenzaldehyde, catalyzed by a secondary cooperative aminated silica catalyst[22,23,29].

A. De Vylder et al. / Journal of Catalysis 361 (2018) 51–61 55

are endcapped with trimethylsilyl groups, the initial activation byhydrogen-bonding, as well as the cooperativity in proton transferreactions, are absent which results in a lower turnover frequency.

A potential side-reaction for both the cooperative and mono-functional catalyst is the reaction of the amine with the carbonylgroup of 4-nitrobenzaldehyde, which yields an iminium ion. Thisiminium can then react with acetone, via a Mannich type mecha-nism, and yield the condensate as a primary product [50,51].However, the direct interaction of 4-nitrobenzaldehyde with theamine is suppressed by using a 260 molar ratio excess of acetone.

Preliminary experiments were performed at 55 �C using 45 gacetone in 55 g DMSO, with 0.1 mmol active amine sites, and0.45 g 4-nitrobenzaldehyde. DMSO is chosen as solvent becauseof its full miscibility with water, allowing to investigate the effectof different amounts of water on the catalyst stability and activity.To adequately describe these effects, the initial water content inthe organic solvents should also be known. Hence, using coulomet-ric Karl-Fischer titration, the initial water content in the reactor,using stock acetone and DMSO, was determined at 0.19 wt%. Underthese conditions, the cooperative acid-base catalyst exhibited aturnover frequency (TOF) of 7.04 ± 0.40 10�4 s�1 in the aldol reac-tion. While, at the same reaction conditions, the monofunctionalbase catalyst exhibited a lower TOF of 2.56 ± 0.13 10�4 s�1, asexpected from previous research [29]. The adduct and condensateselectivity amounted to 73% and 27% respectively, for the cooper-ative acid-base catalyst and 77% and 23% for the monofunctionalbase catalyst, and remained constant throughout the whole exper-iment. When reusing the spent catalyst in a second experiment,only 38% of the initial TOF was obtained for the cooperative acid-base catalyst while 77% of the initial TOF was retained for themonofunctional base catalyst.

To obtain insights in the cause of this reduction in initial TOF,characterization of the spent catalysts was performed. The specificBET surface areas, average pore sizes, and total pore volumes of thespent catalysts are also listed in Table 1. The spent catalysts exhib-ited a slight decrease in surface area and pore volume, but retainedtheir average pore size of 5.0 nm. However, the CHNS elementalanalysis indicated a nitrogen content increase of 60% for the spent

cooperative acid-base catalyst and a 15% increase for the spentmonofunctional base catalyst. Moreover, as will be discussed inmore detail below, the originally white catalyst changed to darkyellow and retained this color even after additional stirring in chlo-roform and drying. This color change, together with the highernitrogen content, indicates that the changes in catalytic behaviorupon recycling may be caused by surface species that remain onthe catalyst.

Raman spectroscopy of these spent catalysts was performed toidentify the organic groups in the species remaining on the catalystsurface. The Raman spectrum is displayed in Fig. 1a for the spentcooperative acid-base catalyst, and Fig. 1b for the spent monofunc-tional base catalyst. The spectra for both catalysts exhibit sharppeaks at 1630 cm�1, 1590 cm�1, 1340 cm�1 and 1110 cm�1. Thesecorrespond to, respectively, a C@C stretching band conjugated withan aromatic group, an aromatic C@C stretching, a symmetric NO2

stretch and a weak band due to aromatic in-plane CAH deforma-tion vibrations [52]. The bands below 300 cm�1 are characteristicof the lattice vibrations of the SiAOASi network.

Further characterization of the spent catalyst was performedwith 13C cross-polarization/magic-angle-spinning (CP/MAS) NMRspectroscopy. The spectrum is displayed in Fig. 2 for the spent coop-erative acid-base catalyst. The peak at 50 ppm can be associatedwith residual methoxy groups from the amine precursor that didnot react with the surface silanol groups. The three peaks at 8, 22and 49 ppm are associated with the sp3 carbon atoms of the propylchain and the peak at 33 ppm is associatedwith themethyl functionof the secondary amine. Additional bands are observed at 147 ppm,originating from an aromatic carbon attached to a nitro group and127 ppm and 123 ppm from aromatic carbons. The shoulder peakat 150 ppm can be assigned to the aromatic carbon in para-position of the nitro group when bounded to a sp3 carbon. Whenbounded to a sp2 carbon, the corresponding peak appears around144 ppm and is indistinguishable from the large peak at 147 ppm.The small broad peak at 70 ppm can be assigned to an sp3 carbonattached to both the aromatic ring and an alcohol function, whilethe small peak at 38 ppm can be assigned to an sp3 carbon atomattached to the carbon atom at 70 ppm and an sp2 carbon atom.

020040060080010001200140016001800

Ram

an in

tens

ity (a

.u.)

Raman shift (cm-1)

C=C-CaromNO2

Carom-H

b)

a)

C=C-CaromNO2

Carom-H

Si-O-Si

Si-O-Si

Fig. 1. Raman spectrum of the spent cooperative acid-base catalyst (a) and spentmonofunctional base catalyst (b).

(6) (6’)

(5) (5a) (5b)

H2OH2O

H2O

Scheme 3. Possible species on the surface of a spent catalyst corresponding withthe characterization results and their proposed formation pathways. The iminiumintermediate species (5) from the catalytic cycle in Scheme 2, or from the aldoladduct readsorption via (6), can further dehydrate into a conjugated species (5b) viaits enamine intermediate (5a). Condensate readsorption via (60) also yields theconjugated species (5b).

56 A. De Vylder et al. / Journal of Catalysis 361 (2018) 51–61

The Raman and 13C NMR spectra indicate that the specieson the surface of the spent catalysts are derivatives of4-nitrobenzaldehyde. Due to the absence of peaks in the 13C NMRbetween 170 ppm and 180 ppm and in the Raman spectrum at1700 cm�1, corresponding to carbonyl groups from aldehydes orketones, physisorbed 4-nitrobenzaldehyde, or physisorbed prod-ucts, can be excluded. The direct addition of 4-nitrobenzaldehydewith the secondary aminedid not lead to the observedRamanpeaks.This is shown in Fig. 2S (Supporting Information), which is a Ramanspectrum of a catalyst that was exposed to 4-nitrobenzaldehyde inDMSO, without acetone, under reaction conditions.

Several possible surface species, stemming from the iminiumintermediate (5 in Scheme 2), are proposed in Scheme 3. This imi-nium intermediate is formed after the attack of the enamine on thecarbonyl group of 4-nitrobenzaldehyde, or from the aldol adductreadsorption (6) and subsequent dehydration. It can subsequentlyrearrange towards the enamine form (5a). Upon dehydration of(5a), the enamine form of the condensation product (5b) is formed.This latter species can also be formed from the dehydration of thecarbinolamine (60). It can, however, reasonably be expected thatthe pathway via product readsorption on a free amine site onlyoccurs to a limited extent due to the large excess of acetone used.

-50050100150200250

Inte

nsity

(a.u

.)

ppm

a

bf

e gh

i

c

e

f

g

hi

d

d

ac

b

Fig. 2. 13C CP/MAS NMR spectrum of the spent cooperative acid-base catalyst. Thegrafted secondary amine can be identified in the spectrum, as well as the carbonatoms originating from the aromatic ring with a nitro substituent.

Because charged species are unlikely to exist on a dried catalyst,the iminium intermediate is expected to only be stable underliquid-phase conditions. Any iminium intermediates that are onthe surface of the catalyst after removal of the material from thereactor by filtration, are expected to convert to (5a), or directlydehydrate to (5b). The 4 different candidate species to be presenton the surface of a spent catalyst correlating with the Raman spec-troscopy and 13C NMR characterization results, are (5a), (5b), (6)and (60) in Scheme 3. Recent computational research on the relativestability of enamines [53] has shown that conjugated enamines,such as dienamines with phenyl groups, are thermodynamicallystable due to their conjugation effects. Possibly, the enamine spe-cies (5b) thus blocks access to the catalytic site when recyclingthe catalyst in a subsequent experiment.

Water was co-fed in the reactor to enhance the hydrolysis of theiminium intermediate (5) in the catalytic cycle and, hence, preventthe formation of the site blocking species as indicated in Scheme 3.In this work, the polar solvent DMSO was chosen because of itscomplete miscibility with water in a broad range of conditions.Even though this solvent does not yield the highest turnover fre-quencies (TOF) for amine catalyzed aldol reactions, it allows to per-form a mechanistic investigation of the deactivation phenomenaand the effect of co-feeding water.

3.3. Effect of water on the reusability of aminated silica catalysts foraldol reactions in the batch reactor

Several catalytic experiments were performed with a varyingamount of water added to the starting mixture of the batch reactor.Care was taken that the amount of solvent, i.e., DMSO, was adaptedto compensate for the volume of the water added to the system.Experiments were also performed with dried acetone, resultingin a total water content of 0.09 wt% in the reactor. After 4 h of reac-tion, the spent catalysts were removed from the reactor, dried, andreused in a second experiment under identical operating condi-tions, or stored in a desiccator for further characterization.

A. De Vylder et al. / Journal of Catalysis 361 (2018) 51–61 57

The effect of water on the activity of the cooperative acid-baseand the monofunctional base catalysts, as indicated by the TOF, isdisplayed in Fig. 3a and b respectively. When 0.69 wt% water waspresent in the reactor with the cooperative acid-base catalyst, theTOF of the first run amounted to 9.38 ± 0.47 10�4 s�1, compared to7.04 ± 0.35 10�4 s�1 without water addition. For the monofunc-tional base catalyst, an increase from 2.56 ± 0.13 10�4 s�1 to 3.66± 0.18 10�4 s�1 is observed. Such an enhancement in the reactionrate has also previously been observed when a small amount ofwater is co-fed when using homogeneous amino-acid catalysts[54]. This can be related to the increased hydration of the reactiveiminium species on the catalytic surface, which leads to anincreased product desorption rate. When the water amount wasfurther increased up to 3.19 wt%, the TOF lowered again to 7.06± 0.35 10�4 s�1 for the cooperative acid-base catalyst. This can beattributed to (i) an unfavorable shift in the carbinolamine-enamine equilibrium [46,50], (ii) protonation of amine groupsand (iii) hydrogen bonding of amine sites with surface silanolgroups, assisted by water molecules [55]. The decrease in TOF with3.19 wt% water also occurs for the monofunctional base catalyst,albeit to a lesser extent. This indicates that endcapping the silanolgroups, and thereby increasing the hydrophobicity of the support,decreases its susceptibility to the negative effects of water.

During a single experiment, the selectivity rapidly converged toa constant value. Additionally, when more water was included inthe starting mixture in the reactor, a pronounced shift in selectivitytowards the aldol adduct was observed for both catalysts (Fig. 3S inSupporting Information). Hence, it can be concluded that the aldoladduct and condensate are in equilibrium at the investigated con-ditions, and the addition of water shifts this equilibrium to the sideof the aldol adduct.

The catalyst reusability ratios are calculated as the ratio of TOF inthe second run to that in the first run and are also displayed in Fig. 3.The reusability of the cooperative acid-base catalyst increased from28%with 0.09 wt%water to amaximumof 70%with 0.69 wt%water.Correspondingly, the reusability of the monofunctional base cata-lyst increased from 58% with 0.09 wt% water to 92% with 0.69 wt%water, and remained constant for higher water amounts. The pres-ence of silanol groups in the cooperative acid-base catalyst thereforeseems to limit the reusability that can be achieved. Yet, almost fullreusability of themonofunctional base catalyst canbe achievedwithat least 0.69 wt% water. This demonstrates how deactivation of theamine active sites can be minimized by the addition of water. Thecatalyst characterization data in Tables 1S and 2S (Supporting Infor-mation) shows that, at the time scale of these batch experiments,

0

2

4

6

8

10

Ratio (-)

TOF

(10-4

s-1)

Water content (wt%)

0.0

0.2

0.4

0.6

0.8

1.0

0.09 0.19 0.69 1.19 2.19 3.19

-4-1

)a

Fig. 3. Turnover frequency (TOF – s�1) for the aldol reaction of acetone with 4-nitrobenzcatalyst (b, blue) as a function of the amount of water in the reactor. Full bars represent ththe TOF observed in the second and the first run. Lines are a guide for the eye. Absolutebars indicate a 95% confidence interval. (T = 55 �C, 0.45 g 4-nitrobenzaldehyde, 45 g acetowt% water in stock acetone, 0.09 wt% water in dried acetone, 0.09 wt% water in stock D

structural degradation of the porous silica network is not a pro-nounced deactivation phenomenon for both types of catalyst.

It should further be noted that when a spent monofunctionalbase catalyst, after use under dry reaction conditions, was recycledfor a second experiment with 2.19 wt%, only 70% reusability wasachieved. This shows that, while such an amount of water can pre-vent the deactivation of the amine active site by favoring thehydrolysis of the iminium intermediate during the first run, it can-not fully regenerate the already deactivated sites after a first run inwhich insufficient water was used.

Raman spectra were also recorded for each spent catalyst, andare plotted as a function of the water amount in the reactor inFig. 4. It can be seen that the peaks at 1630 cm�1, 1600 cm�1,1340 cm�1 and 1110 cm�1, associated with the aromatic speciesbearing a nitro group, gradually disappear when more water is pre-sent in the reactor.

Additionally, a 13CNMR spectrumwas recorded for a cooperativeacid-base catalyst that was spent in a reaction with 1.19 wt% waterin the reactor and is displayed in Fig. 5. Also here, the characteristicpeaks of the nitro-containing aromatic species decreased consider-ably due to the addition of water.

As mentioned above, a distinct change in color of the spent cat-alyst was observed after reaction, which persisted after stirring thecatalyst in chloroform. By co-feeding water, it appears that the col-orization of the spent catalyst becomes less pronounced as a func-tion of the amount of water, as is shown in Fig. 6. Removinginherently present water from the stock acetone solution by dryingcaused the cooperative acid-base catalyst to appear dark red afterbeing spent. The color variations of the spent monofunctional basecatalyst seem to be shifted slightly towards the colors of the coop-erative acid-base catalysts spent after having used a higher amountof water. Hence, the same trends are expected for the monofunc-tional base catalysts.

UV–Vis reflection characterization of the spent cooperativeacid-base catalysts was performed and is displayed in Fig. 7. Whenmore water was added to the reactor, the peak intensity of thespent catalyst is decreased. This indicates that less chromophoricsurface species are present on a spent catalyst when more wateris added to the reactor.

The peak maximum for the catalysts spent with 0.09 wt% ofwater is at 520 nm, while the peak maximum for the catalystsspent with at least 1.19 wt% of water is located at 432 nm. Thisshift to shorter wavelengths is a typical phenomenon that can beseen when the degree of conjugation in an organic compounddecreases. Since this shift correlates with the increased reusability,

0.09 0.19 0.69 1.19 2.19 3.190

2

4

6

8

10

Ratio (-)

TOF

(10

s)

Water content (wt%)

0.0

0.2

0.4

0.6

0.8

1.0)b

aldehyde, using a cooperative acid-base catalyst (a, red) and a monofunctional basee first run, empty bars the second run, and the right axis indicates the ratio betweenwater content in the reactor has been determined with Karl-Fischer titrations. Errorne, 52–55 g DMSO, 0.20 g methyl 4-nitrobenzoate, 0.5 g to 3.0 g water added, 0.31MSO).

100012001400160018002000

Ram

an in

tens

ity (a

.u.)

Wavenumber (cm-1)

(a)

(b)

(d)

(e)

(f)

(c)

100012001400160018002000

Ram

an in

tens

ity (a

.u.)

Wavenumber (cm-1)

(a)

(b)

(d)

(e)

(f)

(c)

Fig. 4. Raman spectrum of the spent cooperative acid-base catalyst (left) and the spent monofunctional base catalyst (right) (a: 0.09 wt% water, b: 0.19 wt% water, c: 0.69 wt%water, d: 1.19 wt% water, e: 2.19 wt% water and f: 3.19 wt% water).

-50050100150200250

Inte

nsity

(a.u

.)

ppm

d

b

c

ae

ed

c b

a

Fig. 5. 13C NMR result of a cooperative acid-base catalyst, spent in the aldolreaction of 4-nitrobenzaldehyde with acetone, using 1.19 wt% water in the reactor.

0.09 wt% 0.19 wt% 0.69 wt% 1.19 wt% 2.19 wt% 3.19 wt%water:

a)

b)

Fig. 6. Spent catalyst color as a function of the water amount in the reactor. (a)Cooperative acid-base, (b) monofunctional base catalyst.

0

5

10

15

20

25

30

35

40

300 400 500 600 700 800

Rel

ativ

e re

flect

ion

(%)

Wavelength (nm)

f

e

dc

ba

Fig. 7. Percentage UV–Vis reflection as a function of wavelength of the spentcooperative acid-base catalysts, relative to a fresh catalyst (a: 0.09 wt% water, b:0.19 wt% water, c: 0.69 wt% water, d: 1.19 wt% water, e: 2.19 wt% water and f: 3.19wt% water).

58 A. De Vylder et al. / Journal of Catalysis 361 (2018) 51–61

a more highly conjugated species such as 5b (Scheme 3) is believedto be responsible for the deactivation of the catalyst. The possiblechromophoric species at 432 nm, present on the spent catalystsafter use with at least 1.19 wt% water, could then correspond tothe more hydrated species 5a, 6 and 60 (Scheme 3).

As mentioned above, the nitrogen content on the spent catalystwas higher than that on the fresh catalyst. Hence, the catalystsspent using different amounts of water in the reactor were subjectto CHNS elemental analysis. The observed amount of nitrogen onthe spent catalyst as a function of the amount of water used inthe experiment is expressed relative to the total amount of activesites on a fresh catalyst, and is defined as ‘excess N’ in Fig. 4S in

Supporting Information. Since possible leaching effects are maskedby the increase in nitrogen content, it has also been verified thatthere was no active site leaching by performing batch experimentswith 4-chlorobenzaldehyde as a substrate, and ethyl-4-methylbenzoate as internal standard, under the same reaction con-ditions. These results are also included in the Supporting Informa-tion, Fig. 4S, and prove that leaching is not a significantdeactivation factor at the investigated batch reactor conditions.The increase in the nitrogen content of the spent catalyst comparedto the fresh catalyst can be used to quantify the fraction of sitesthat are covered with a nitrogen containing species, such as theones proposed in Scheme 3. From the results in Fig. 4S it is clearthat, when using 0.09 wt% water, also denoted as ‘dry’ conditions,the fraction of covered sites is significantly larger (90%) on thespent cooperative acid-base catalyst than on the spent monofunc-tional base catalyst (40%). This could be the result of the relativelylow amount of water with respect to active sites under these ‘dry’conditions, i.e., with 0.09 wt% of water, only 5 mmol of water isavailable for 0.1 mmol amines and 0.4 mmol silanol groups. Hence,adsorption of several water molecules on the hydrophilic silanolgroups can already significantly lower the amount of water mole-cules available for iminium ion hydrolysis. When no silanol groupsare present for water adsorption, as is the case in the monofunc-tional base catalyst, more water is available for iminium ionhydrolysis. When a water excess is present, i.e. 0.69 wt% or more,this discrepancy between both catalysts disappears.

It can be concluded from the Raman spectra, the UV–Vis reflec-tion spectra, the elemental CHNS analysis and the 13C NMR, thatwater reduces the amount of surface species that remain on aspent catalyst and increases the catalyst’s reusability. It howeverappears that, despite having no remaining surface species on the

A. De Vylder et al. / Journal of Catalysis 361 (2018) 51–61 59

amine site, the activity of the cooperative acid-base catalyst canonly be regained for 70% by adding water.

3.4. Differences in activity and stability of the cooperative acid-baseand the monofunctional base catalysts

It was shown in the previous section that sufficient water isrequired to minimize the deactivation of the amine sites. Yet, onlya maximum reusability of 70% can be achieved for the cooperativeacid-base catalyst, while the monofunctional base catalyst isalmost entirely reusable. Additionally, it is seen that adding morewater even seems to decrease the reusability of the cooperativeacid-base catalyst. To investigate the behavior of the cooperativeacid-base catalyst with high water contents, and to confirm thestability of the monofunctional base catalyst without having toremove and recycle the catalyst in numerous consecutive batchexperiments, continuous-flow experimentation was performedunder the same reaction conditions. Water is added to the feedamounting to 2.19 wt%. This large amount has been selected toreduce the deactivation of the amine sites to a minimum by com-plete iminium ion hydrolysis, and to understand the potential neg-ative effects of a large amount of water on the catalyst stability.Both catalysts were compared at the same site time of 165.1molsite s mol�1

4-nitrobenzaldehyde, which indicates that the average timeavailable, per site, for a molecule to react is identical. Fig. 8 showsthe conversion for both catalysts with time-on-stream.

Till 7 h on stream, the cooperative acid-base catalyst exhibited ahigher activity than the monofunctional base catalyst. This agreeswell with the observations made in the batch reactor, where thecooperative acid-base catalyst exhibited a higher activity thanthe monofunctional base catalyst at the same conditions. However,after 7 h on-stream, the activity of the cooperative catalyst hasdropped to the same level as the monofunctional base catalyst,which has only slightly deactivated after such a time on stream.This first, and most rapid, deactivation of the cooperative acid-base catalyst is attributed to the loss of the cooperativity betweenthe silanol groups and the amine active sites. Indeed, when thecooperative ability has been completely lost after 7 h on stream,only the activity of the isolated amine site remains, whereby theactivity of the cooperative acid-base catalyst has been reduced tothat of the monofunctional base catalyst. The silanol groups, thatare responsible for the cooperative effect, seem to lose theirhydrogen-bonding ability towards the carbonyl substrates. Thiscould be due to water deprotonating the silanol groups, or formingan extensive multilayer coverage on the surface of the silica gel[56], which limits access to the silanol groups for the carbonyl

0

5

10

15

20

0 5 10 15 20 25 30

Con

vers

ion

(%)

Time-on-stream (h)

Fig. 8. Stability of the cooperative acid-base catalyst (d, blue) and the monofunc-tional base catalyst (j, red) with time-on-stream for 28 h. (T = 55 �C, Ptot = 1.8 bar,site time = 165.1 molsite s mol4-nitrobenzaldehyde�1 , 0.45 wt% 4-nitrobenzaldehyde, 44.6wt% acetone, 52.5 wt% DMSO, 0.25 wt% methyl 4-nitrobenzoate, 2.19 wt% water).Lines are a guide to the eye.

substrates. Dehydration of the silica support material, and com-plete regeneration of the silanol group’s cooperative ability, is onlypossible at elevated temperatures which would destroy the organicmoieties on our catalyst. This silanol group deactivation alsoexplains why these cooperative acid-base catalysts are not reusa-ble in the batch reactor, and the addition of larger amounts ofwater seems to even decrease the reusability ratio, even thoughthe amine site itself can be made fully reusable.

After 8 h on-stream, the slope of the evolution of the conversionwith time-on-stream for the cooperative acid-base catalyst changes.The deactivation occurs at a reduced rate, yet, it remains faster thanthe decrease in conversion with the time-on-stream exhibited bythemonofunctional base catalyst. In this regime of prolonged expo-sure to the aqueous feed, the deactivation is most likely caused byhydrolysis reactions of the silica material that cause aminosilaneleaching, structural degradation of the support material and silicadissolution. This is suggested by the drop in surface area and porevolume for the cooperative acid-base catalyst, as is indicated inTable 3S (Supporting Information). The slower deactivation rate ofthe monofunctional base catalyst, and the more stable surface areaand pore volume after catalysis, could then be the result of theend-capping of the silanol groups with trimethylsilyl groups, whichintroduced a more hydrophobic surface environment. Thisincreased support stability by trimethylsilation has already beenobserved for other mesoporous silica materials such as MCM-48[57]. It should, however, be noted that endcapping with HMDS isnever complete due to steric hinderance, and that water can stillinteract with the silica surface under the umbrellas of thetrimethylsilyl groups [58], causing hydrolysis and subsequent deac-tivationwith long time-on-stream. Hence, themonofunctional basecatalyst is notunconditionally stable. Amore stable catalyst couldbedeveloped in the form of a periodic mesoporous organosilica mate-rial with hydrophobic organic groups in the support material [35].

4. Conclusions

For the first time, the reusability of aminated mesoporousamorphous silica catalysts was thoroughly investigated for thealdol reaction of 4-nitrobenzaldehyde with acetone in DMSO. It isfound that, under ‘dry’ reaction conditions, both the cooperativeacid-base catalyst, as well as the monofunctional base catalyst,exhibit a significantly poorer performance in consecutive batchexperiments. Spent catalyst characterization using Raman spec-troscopy, 13C NMR, and UV–Vis characterization, allowed identify-ing site blocking species originating from the iminiumintermediate in the catalytic cycle. To increase the hydration rateof this iminium intermediate and, hence, reduce their conversioninto site blocking species, water should be co-fed to the batch reac-tor. This results in an increased reusability up to 70% for the coop-erative acid-base catalyst and an almost complete reusability forthe monofunctional base catalyst.

Using continuous-flow experimentation at operating conditionswhere the formation of site blocking species is minimized, the dif-ference in deactivation behavior between the cooperative andmonofunction catalyst has been assessed further. In the firstinstance, a relatively rapid loss of the cooperative ability of the cat-alyst is observed, most likely due to the local buildup of water, orthe deprotonation of the silanol groups by water. With a longertime-on-stream, the activity of both catalysts further decreasesmore moderately, which can be related to hydrolysis reactions ofthe silica network that lead to aminosilane leaching, support degra-dation, and silica dissolution. This deactivation being even less pro-nounced for the monofunctional base catalyst than for thecooperative one, is attributed to its more pronounced hydrophobic-ity owing to the trimethylsilyl groups used for endcapping.

60 A. De Vylder et al. / Journal of Catalysis 361 (2018) 51–61

Acknowledgment

The authors acknowledge financial support from the Fund forScientific Research Flanders (FWO) through Grant Number3G006813 and the European Research Council under the EuropeanUnion’s Seventh Framework Programme (FP7/2007-2013) / ERCgrant agreement n�615456. J.L. is a postdoctoral fellow of theResearch Foundation - Flanders (12Z2218N).

Author Contributions

The manuscript was written through contributions of allauthors. All authors have given approval to the final version ofthe manuscript.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at https://doi.org/10.1016/j.jcat.2018.02.016.

References

[1] H. Suwito, Jumina Mustofa, A.N. Kristanti, N.N.T. Puspaningsih, Chalcones:synthesis, structure diversity and pharmacological aspects, J. Chem. Pharm.Res. 6 (2014) 1076–1088.

[2] G. Kelly, F. King, M. Kett, Waste elimination in condensation reactions ofindustrial importance, Green Chem. 4 (2002) 392–399.

[3] Q. Deng, J. Xu, P. Han, L. Pan, L. Wang, X. Zhang, J.-J. Zou, Efficient synthesis ofhigh-density aviation biofuel via solvent-free aldol condensation of cyclicketones and furanic aldehydes, Fuel Process. Technol. 148 (2016) 361–366.

[4] R.M. West, Z.Y. Liu, M. Peter, C.A. Gärtner, J.A. Dumesic, Carbon–carbon bondformation for biomass-derived furfurals and ketones by aldol condensation ina biphasic system, J. Mol. Catal. A: Chem. 296 (2008) 18–27.

[5] I. Fechete, Y. Wang, J.C. Védrine, The past, present and future of heterogeneouscatalysis, Catalysis Today 189 (2012) 2–27.

[6] P.M. Price, J.H. Clark, D.J. Macquarrie, Modified silicas for clean technology, J.Chem. Soc., Dalton Trans. (2000) 101–110.

[7] E. Dumitriu, V. Hulea, I. Fechete, A. Auroux, J.-F. Lacaze, C. Guimon, The aldolcondensation of lower aldehydes over MFI zeolites with different acidicproperties, Microporous Mesoporous Mater. 43 (2001) 341–359.

[8] A. Ungureanu, S. Royer, T.V. Hoang, D. Trong On, E. Dumitriu, S. Kaliaguine,Aldol condensation of aldehydes over semicrystalline zeolitic-mesoporous UL-ZSM-5, Microporous Mesoporous Mater. 84 (2005) 283–296.

[9] Y.-C. Chang, A.-N. Ko, Vapor phase reactions of acetaldehyde over type Xzeolites, Appl. Catal. A 190 (2000) 149–155.

[10] O. Kikhtyanin, V. Kelbichová, D. Vitvarová, M. Kubu, D. Kubicka, Aldolcondensation of furfural and acetone on zeolites, Catal. Today 227 (2014)154–162.

[11] L. Hora, V. Kelbichová, O. Kikhtyanin, O. Bortnovskiy, D. Kubicka, Aldolcondensation of furfural and acetone over Mg Al layered double hydroxidesand mixed oxides, Catal. Today 223 (2014) 138–147.

[12] I. Sádaba, M. Ojeda, R. Mariscal, R. Richards, M.L. Granados, Mg–Zr mixedoxides for aqueous aldol condensation of furfural with acetone: effect ofpreparation method and activation temperature, Catal. Today 167 (2011) 77–83.

[13] L. Faba, E. Díaz, S. Ordóñez, Aqueous-phase furfural-acetone aldolcondensation over basic mixed oxides, Appl. Catal. B 113–114 (2012) 201–211.

[14] S. Abelló, D. Vijaya-Shankar, J. Pérez-Ramírez, Stability, reutilization, andscalability of activated hydrotalcites in aldol condensation, Appl. Catal. A 342(2008) 119–125.

[15] J. Cueto, L. Faba, E. Díaz, S. Ordóñez, Performance of basic mixed oxides foraqueous-phase 5-hydroxymethylfurfural-acetone aldol condensation, Appl.Catal. B 201 (2017) 221–231.

[16] O. Kikhtyanin, Z. Tišler, R. Velvarská, D. Kubicka, Reconstructed Mg-Alhydrotalcites prepared by using different rehydration and drying time:physico-chemical properties and catalytic performance in aldolcondensation, Appl. Catal. A 536 (2017) 85–96.

[17] L. Smoláková, K. Frolich, J. Kocík, O. Kikhtyanin, L. Capek, Surface properties ofhydrotalcite based Zn(Mg)Al oxides and their catalytic activity in aldolcondensation of furfural with acetone, Ind. Eng. Chem. Res. 56 (2017) 4638–4648.

[18] J. Schütz, W. Bonrath, Y. Pressel, E. Ferfecki, K.-D. Topp, Aldol condensationscatalysed by basic anion exchange resins, ChemCatChem 8 (2016) 3584–3591.

[19] S. Cheng, X. Wang, S.-Y. Chen, Applications of amine-functionalizedmesoporous silica in fine chemical synthesis, Top. Catal. 52 (2009) 681–687.

[20] A.M. Chong, X. Zhao, A.T. Kustedjo, S. Qiao, Functionalization of large-poremesoporous silicas with organosilanes by direct synthesis, MicroporousMesoporous Mater. 72 (2004) 33–42.

[21] P.V. Der Voort, E. Vansant, Silylation of the silica surface a review, J. LiquidChromatogr. Related Technol. 19 (1996) 2723–2752.

[22] K. Kandel, S.M. Althaus, C. Peeraphatdit, T. Kobayashi, B.G. Trewyn, M. Pruski, I.I. Slowing, Substrate inhibition in the heterogeneous catalyzed aldolcondensation: A mechanistic study of supported organocatalysts, J. Catal.291 (2012) 63–68.

[23] S. Shylesh, D. Hanna, J. Gomes, S. Krishna, C.G. Canlas, M. Head-Gordon, A.T.Bell, Tailoring the cooperative acid-base effects in silica-supported aminecatalysts: applications in the continuous gas-phase self-condensation of n-butanal, ChemCatChem 6 (2014) 1283–1290.

[24] J. Lauwaert, E. De Canck, D. Esquivel, P. Van Der Voort, J.W. Thybaut, G.B.Marin, Effects of amine structure and base strength on acid–base cooperativealdol condensation, Catal. Today 246 (2015) 35–45.

[25] R.K. Zeidan, M.E. Davis, The effect of acid–base pairing on catalysis: an efficientacid–base functionalized catalyst for aldol condensation, J. Catal. 247 (2007)379–382.

[26] S.L. Hruby, B.H. Shanks, Acid–base cooperativity in condensation reactionswith functionalized mesoporous silica catalysts, J. Catal. 263 (2009) 181–188.

[27] K.C. Kim, E.G. Moschetta, C.W. Jones, S.S. Jang, Molecular dynamics simulationsof aldol condensation catalyzed by alkylamine-functionalized crystalline silicasurfaces, J. Am. Chem. Soc. 138 (2016) 7664–7672.

[28] N.A. Brunelli, S.A. Didas, K. Venkatasubbaiah, C.W. Jones, Tuning cooperativityby controlling the linker length of silica-supported amines in catalysis and CO2

capture, J. Am. Chem. Soc. 134 (2012) 13950–13953.[29] J. Lauwaert, E. De Canck, D. Esquivel, J.W. Thybaut, P. Van Der Voort, G.B.

Marin, Silanol-assisted aldol condensation on aminated silica: understandingthe arrangement of functional groups, ChemCatChem 6 (2014) 255–264.

[30] N.A. Brunelli, K. Venkatasubbaiah, C.W. Jones, Cooperative catalysis with acid–base bifunctional mesoporous silica: impact of grafting and co-condensationsynthesis methods on material structure and catalytic properties, Chem.Mater. 24 (2012) 2433–2442.

[31] N.A. Brunelli, C.W. Jones, Tuning acid–base cooperativity to create nextgeneration silica-supported organocatalysts, J. Catal. 308 (2013) 60–72.

[32] R.K. Zeidan, S.J. Hwang, M.E. Davis, Multifunctional heterogeneous catalysts:SBA-15-containing primary amines and sulfonic acids, Angew. Chem. Int. Ed.45 (2006) 6332–6335.

[33] J. Lauwaert, E.G. Moschetta, P. Van Der Voort, J.W. Thybaut, C.W. Jones, G.B.Marin, Spatial arrangement and acid strength effects on acid–basecooperatively catalyzed aldol condensation on aminosilica materials, J. Catal.325 (2015) 19–25.

[34] A. Massi, A. Cavazzini, L. Del Zoppo, O. Pandoli, V. Costa, L. Pasti, P.P.Giovannini, Toward the optimization of continuous-flow aldol and a-amination reactions by means of proline-functionalized silicon packed-bedmicroreactors, Tetrahedron Lett. 52 (2011) 619–622.

[35] J. Ouwehand, J. Lauwaert, D. Esquivel, K. Hendrickx, V. Van Speybroeck, J.W.Thybaut, P. Van Der Voort, Facile synthesis of cooperative acid-base catalystsby clicking cysteine and cysteamine on an ethylene-bridged periodicmesoporous organosilica, Eur. J. Inorg. Chem. (2016).

[36] W. Huybrechts, J. Lauwaert, A. De Vylder, M. Mertens, G. Mali, J.W. Thybaut, P.Van Der Voort, P. Cool, Synthesis of L-serine modified benzene bridgedperiodic mesoporous organosilica and its catalytic performance towards aldolcondensations, Microporous Mesoporous Mater. 251 (2017) 1–8.

[37] Y. Kubota, H. Yamaguchi, T. Yamada, S. Inagaki, Y. Sugi, T. Tatsumi, Furtherinvestigations on the promoting effect of mesoporous silica on base-catalyzedaldol reaction, Top. Catal. 53 (2010) 492–499.

[38] O. Kikhtyanin, E. Lesnik, D. Kubicka, The occurrence of Cannizzaro reactionover Mg-Al hydrotalcites, Appl. Catal. A 525 (2016) 215–225.

[39] J. Schütz, W. Bonrath, Y. Pressel, E. Ferfecki, K.-D. Topp, Aldol condensationscatalysed by basic anion exchange resins, ChemCatChem (2016).

[40] G.G. Podrebarac, F.T.T. Ng, G.L. Rempel, A kinetic study of the aldolcondensation of acetone using an anion exchange resin catalyst, Chem. Eng.Sci. 52 (1997) 2991–3002.

[41] E.L. Margelefsky, R.K. Zeidan, M.E. Davis, Cooperative catalysis by silica-supported organic functional groups, Chem. Soc. Rev. 37 (2008) 1118–1126.

[42] B. List, L. Hoang, H.J. Martin, Newmechanistic studies on the proline-catalyzedaldol reaction, PNAS 101 (2004) 5839–5842.

[43] B. List, R.A. Lerner, C.F. Barbas, Proline-catalyzed direct asymmetric aldolreactions, J. Am. Chem. Soc. 122 (2000) 2395–2396.

[44] W. Huybrechts, J. Lauwaert, A. De Vylder, M. Mertens, G. Mali, J.W. Thybaut, P.Van Der Voort, P. Cool, Synthesis of L-serine modified benzene bridgedperiodic mesoporous organosilica and its catalytic performance towards aldolcondensations, Microporous Mesoporous Mater. (2017).

[45] C. Hammond, Intensification studies of heterogeneous catalysts: probing andovercoming catalyst deactivation during liquid phase operation, Green Chem.19 (2017) 2711–2728.

[46] N. Zotova, A. Franzke, A. Armstrong, D.G. Blackmond, Clarification of the role ofwater in proline-mediated aldol reactions, J. Am. Chem. Soc. 129 (2007)15100–15101.

[47] P. Van Der Voort, S. Vercauteren, K. Peeters, E.F. Vansant, Some precautionswhen determining the silanol number, using chemical modification withmethylchlorosilanes, J. Colloid Interface Sci. 157 (1993) 518–519.

[48] R.J. Berger, E.H. Stitt, G.B. Marin, F. Kapteijn, J.A. Moulijn, Eurokin, Chem. React.Kinetics Pract., Cattech 5 (2001) 36–60.

[49] J. Lauwaert, J. Ouwehand, J. De Clercq, P. Cool, P. Van Der Voort, J.W. Thybaut,Tuning component enrichment in amino acid functionalized (organo)silicas,Catal. Commun. 88 (2017) 85–89.

A. De Vylder et al. / Journal of Catalysis 361 (2018) 51–61 61

[50] K. Kandel, S.M. Althaus, C. Peeraphatdit, T. Kobayashi, B.G. Trewyn, M. Pruski, I.I. Slowing, Solvent-induced reversal of activities between two closely relatedheterogeneous catalysts in the aldol reaction, ACS Catal. 3 (2013) 265–271.

[51] M. Nakadai, S. Saito, H. Yamamoto, Diversity-based strategy for discovery ofenvironmentally benign organocatalyst: diamine–protonic acid catalysts forasymmetric direct aldol reaction, Tetrahedron 58 (2002) 8167–8177.

[52] G. Socrates, Infrared and Raman characteristic group frequencies: tables andcharts, John Wiley & Sons, 2004.

[53] A. Castro-Alvarez, H. Carneros, A.M. Costa, J. Vilarrasa, Computer-Aided Insightinto the Relative Stability of Enamines, Synthesis (2017).

[54] M. Amedjkouh, Primary amine catalyzed direct asymmetric aldol reactionassisted by water, Tetrahedron Asymmetry 16 (2005) 1411–1414.

[55] G. Paul, G.E. Musso, E. Bottinelli, M. Cossi, L. Marchese, G. Berlier, Investigatingthe Interaction of Water Vapour with Aminopropyl Groups on the Surface ofMesoporous Silica Nanoparticles, ChemPhysChem 18 (2017) 839–849.

[56] R.P. Scott, P. Kucera, Solute-solvent interactions on the surface of silica gel, J.Chromatogr. A 149 (1978) 93–110.

[57] K. Koyano, T. Tatsumi, Y. Tanaka, S. Nakata, Stabilization of mesoporousmolecular sieves by trimethylsilylation, J. Phys. Chem. B 101 (1997) 9436–9440.

[58] T. Takei, A. Yamazaki, T. Watanabe, M. Chikazawa, Water adsorptionproperties on porous silica glass surface modified by trimethylsilyl groups, J.Colloid Interface Sci. 188 (1997) 409–414.