Impact of Hydrophobic Organohybrid Silicas on the Stability of Ni2P Catalyst Phase in the Hydrodeoxygenation of Biophenols

ChemCatChem

10.1002/cctc.201702001

On the impact of hydrophobic organohybrid silicas on the stability

of Ni P catalyst phase in the HDO of biophenols

2

a

b

a

d

Michael Dierks, Zhengwen Cao, Jinesh C. Manayil, Jeganathan Akilavasan, Karen Wilson,

a,*

c,*

Ferdi Schüth and Roberto Rinaldi

a.

b.

c.

Max-Planck-Institut für Kohlenforschung, Kaiser-Wilhelm-Platz 1, 45470 Mülheim-an-

der-Ruhr, Germany.

European Bioenergy Research Institute, School of Engineering & Applied Science, Aston

University, Birmingham B4 7ET, UK

Department of Chemical Engineering, South Kensington Campus, Imperial College

London, London, SW7 2AZ, UK

School of Science, RMIT University, 124 La Trobe St, Melbourne VIC 3000, Australia

d

^*corresponding authors: Ferdi Schüth, e-mail: schueth@kofo.mpg.de and Roberto Rinaldi,

e-mail: rrinaldi@imperial.ac.uk.

1

ChemCatChem

10.1002/cctc.201702001

Abstract

Hydrodeoxygenation (HDO) of lignocellulose-derived pyrolysis pyrolysis oils offers a process

option to produce fuel substitutes. However, catalyst deactivation and stability remain a significant

issue. Herein, the dependence of stability and activity of Ni

2

P/SiO HDO catalysts on the support

2

surface polarity is addressed in detail. The support surface polarity was adjusted by copolymerizing

tetraethyl orthosilicate (TEOS) with different types and amounts of organosilanes via a sol-gel

process in the presence of nickel nitrate and citric acid. After thermal treatment under an inert

atmosphere, Ni/SiO

2

precursors were formed. They were converted into Ni

2

P/SiO catalysts by

2

using NaH PO as a PH

2

3

source. The catalyst surface polarity was characterized by inverse gas

chromatography measurements of the adsorption free energy of methanol, and specific and

dispersive surface energies from polar and nonpolar probe molecule adsorption. The correlation

between catalyst performance and support surface polarity indicates that, to prevent deactivation

of the catalyst by water under reaction conditions, the affinity of the support towards polar

substances must be decreased below a threshold value.

2

ChemCatChem

10.1002/cctc.201702001

Introduction

High-energy density biofuels produced from the hydrodeoxygenation (HDO) of lignocellulose-

derived pyrolysis pyrolysis are considered as renewable and CO -neutral alternatives to petroleum-

2

derived fuels (e.g. gasoline, kerosene and diesel).¹^,²However, the high-oxygen content of pyrolysis

oils results in poor combustion properties, low stability upon storage and immiscibility with fossil

3

crude oils. These properties hamper the drop-in utilization of pyrolysis oil as a fuel. Accordingly,

a hydrodeoxygenation step is required, in which the oxygenated functional groups are removed by

reductive processes, generating water as a byproduct. Moreover, pyrolysis oils may also contain

up to 50 wt% water. Hence, the inevitable presence of water in the reaction medium usually leads

to deleterious changes in the catalysts under the severe HDO conditions.⁴^-⁸

Catalyst deactivation during hydrotreating of pyrolysis oils may occur through several

pathways. One of the major pathways for catalyst deactivation is coke formation, which can be

9

suppressed by using neutral supports (e.g. SiO

2

) instead of highly acidic materials. Also, in the

case of metal carbides, sulfides, nitrides or phosphides active phase oxidation can present an issue,

1

0

due to their strong affinity towards oxygen. Traditional metal sulfide catalysts are known to suffer

from deactivation during HDO in the presence of water by partial oxidation of the nickel sulfide

1

phase. To circumvent the problem of partial oxidation of active sites, it was demonstrated that

water-sensitive active sites could be protected against deactivation during HDO by adjustment of

the catalyst’s surface polarity.¹²^,¹³

Tailoring the surface polarity has been demonstrated to exert a significant impact on the

1

4

catalytic performance for a variety of reactions. Notably, the adjustment of surface polarity can

have a crucial influence on activity and selectivity for oxidation reactions of organic compounds

using peroxides.¹⁵^-¹⁷Moreover, sulfonic acid groups anchored on silicas, with different polarity,

showed an improved activity and stability for esterification reactions using hydrophobic

3

ChemCatChem

10.1002/cctc.201702001

catalysts.¹⁸^,¹⁹Also in metal-catalyzed reactions, e.g. dehydrogenative coupling of organosilanes

and hydrogenation of hydrophobic substrates, catalysts show increased activities when the catalyst

surface hydrophobicity was modified.²⁰^-²²Also, in the field of biomass conversion, several

examples have been reported on how the control of catalyst surface polarity can be beneficial for

the effective valorization of biomass. In this context, quantitative conversion of fructose to HMF,

without any side products, was achieved when a superhydrophobic solid acid was used as a catalyst.

Such a superior performance was not achieved by using sulfuric acid or Amberlyst-15 as the acid

catalyst.²³

Catalysts stability during biomass valorization can also be improved by adjustment of the

catalyst surface polarity. This especially holds true for the catalytic upgrading of pyrolysis bio-oil

where, as already mentioned, the catalyst stability constitutes a central problem due to the harsh

reaction conditions and presence of water, which is inherent to the bio-oil or is formed during the

reaction. Zapata et al. showed that HY zeolites modified with hydrophobic organosilanes showed

improved stability in hot water (T>150°C) for cresol alkylation and alcohol dehydration and can

24

be used for refining of biomass-derived pyrolysis oil. Moreover, the water tolerance of propyl

sulfonic acid SBA-15 catalysts, employed in esterification of pyrolysis bio-oil as a pretreatment

step, was improved by co-functionalization with octyl groups.²⁵

In this contribution, we present a systematic study of the impact support polarity on the

activity and stability of the Ni

2

P catalyst phase in the HDO of model compounds for pyrolysis oils.

Ni P supported on silica was chosen as a model system, since it is known to show good HDO

2

activity, yet due to the hydrophilic character of silica, these catalysts show strong deactivation by

oxidation processes involving water.²⁶^,²⁷Model catalysts were prepared using a modified sol-gel

approach, in which the polarity of the final catalyst can be adjusted by copolymerization of

tetraethyl orthosilicate (TEOS) with different types and amounts of organosilanes, in the presence

4

ChemCatChem

10.1002/cctc.201702001

of nickel nitrate and citric acid. After thermal treatment under an inert atmosphere, the obtained

Ni/SiO catalysts were converted into the corresponding Ni P phase via phosphidation with PH

generated by thermal decomposition of NaH PO . Recycling experiments performed using

2

3

2

guaiacol as a substrate revealed that it is possible to stabilize the catalyst against water deactivation

by copolymerizing TEOS with 50 mol% methyltriethoxysilane (MTES). The improved catalyst

stability correlates well with its surface polarity properties. In fact, the best catalyst exhibited the

lowest free energy of methanol adsorption, as determined by inverse gas chromatography (IGC).

Results and discussion

Characterization of the catalysts

To explore the influence of the catalyst hydrophobicity on the HDO performance of Ni

2

P/SiO

2

catalysts, a series of different model catalysts was prepared by acid-catalyzed co-condensation of

TEOS with different types and various quantities of organic functions (denoted as ‘R’, Table 1).

Ni supported on organohybrid silica supports were prepared by modifying a sol-gel synthesis

2

8

procedure reported by Takahashi et al. For the preparation of the unmodified Ni/SiO catalyst,

2

TEOS was employed as the silica source, nickel nitrate hexahydrate as the Ni source, and citric

acid monohydrate as an organic additive. In the synthesis, citric acid (CA) is responsible for the

formation of mesopores. Furthermore, CA helps to increase the dispersion of the Ni particles. To

-1

initiate the hydrolysis and condensation of the silica precursors a 0.25 mol L nitric acid solution

was added to the system.²⁸^,²⁹Employing methyltriethoxysilane (MTES), a transparent green

colored gel with an R:Si ratio of up to 0.5 could be prepared. However, when ethyltriethoxysilane

(ETES) or octyltriethoxysilane (OTES) were used, no clear (green) gel was obtained at a R:Si ratio

of 0.5. In these cases, transparent green gels were obtained at R:Si ratios of 0.2 (for ETES) and

0

.05 (for OTES). Utilizing phenyltriethoxysilane (PTES) as the organosilane, a transparent green

5

ChemCatChem

10.1002/cctc.201702001

gel could be produced at an R:Si ratio of 0.1. Employing 1,4-bis(triethoxysilyl)benzene (BTSB),

the R:Si ratio could be further increased to an R:Si ratio of 0.5, while a transparent green colored

gel was obtained.

As will be discussed in a greater deal of detail later in this section, the gels were converted

into Ni/SiO

70 °C for 1.5 h and, sequentially, to 400 °C for 2 h using a heating ramp of 100 °C h . Ni/SiO

precursor prepared using TEOS presents a 33 wt% nominal Ni loading. All the Ni/SiO precursors

were then phosphided with PH3(g), generated by thermal decomposition of NaH PO

2 NaH HPO4(s)+ PH3(g)). For the reaction, NaH PO (0.8 g, located on the inlet Ar-

flow side) and Ni/SiO precursor (0.1 g, located on the outlet Ar-flow side) were placed separately

approximately a 1-cm distance between the powders) in a crucible, as schematically represented

2

precursors by thermal treatment under an Argon flow. First, the gels were heated to

-1

1

2

(

2

PO2(s)→ Na

2

(

in Scheme 1. The system was flushed with Ar for 0.5 h and then heated to 300 °C and kept at this

temperature for 1 h. The exhaust gases were bubbled through an aqueous solution of sodium

hypochlorite to oxidize remaining (highly toxic) PH

3

in the gas stream. Table 1 lists the textural

catalysts.

properties and C-content of the Ni P/organohybrid-SiO

2

6

ChemCatChem

10.1002/cctc.201702001

Scheme 1 Experimental setup for the phosphidation of Ni/SiO

thermal decomposition of NaH PO at 300 °C.

2 3

precursors under an atmosphere of PH generated by

2

7

ChemCatChem

10.1002/cctc.201702001

Table 1. Textural properties and C-content of Ni

TEOS with various organosilanes and phosphided with PH

C.

2

P/organohybrid-SiO

2

catalysts prepared by co-condensation of

PO at 300

3

generated by thermal decomposition of NaH

2

°

C-content(%)

Found (%)

Pore

volume

Average

pore size

(nm)

Material

Organosilane

Monomer

R:Si

S

BET

-1

2

(m g )

Expected (%)

Estimated by

subtraction

3

-1

(

cm g )

[a]

(%)

Ni

2

P/SiO

2

-

465

0.53

7.3

5.0

4.7

6.2

12.7

5.0

0

8.6±0.1

11.3±0.2

13.5±0.2

12.1±0.3

15.5±0.2

15.6±0.2

-

TEOS

Ni

2

P/SiO

0

.05

449

470

484

532

568

0.42

0.51

0.68

0.52

5

2

6

7

2.7

4.9

3.5

6.9

5

OTES

Ni

2

P/SiO

0

.2

.5

.1

0ETES

Ni

2 2

P/SiO

0

2

0MTES

Ni

2 2

P/SiO

0

5

0MTES

Ni

2

P/SiO

2

1

0PTES

Ni

2

P/SiO

2

-

0

.2

.5

563

523

0.47

0.38

6.2

4.7

12

16

15.7±0.5

25.3±0.4

7.1

2

0BTSB

Ni

2 2

P/SiO

16.7

5

0BTSB

[

a] Estimated by subtraction of the C-content of Ni

2

P/SiO

2

-TEOS from the other material samples

8

ChemCatChem

10.1002/cctc.201702001

Figure 1 a) N

alkyl groups; b) Pore size distribution for Ni

groups; c) N adsorption-desorption isotherms for Ni

2

adsorption-desorption isotherms for Ni

2

P/SiO

2

catalysts prepared using TEOS and organosilanes with

2

P/SiO

2

catalysts prepared using TEOS and organosilanes with alkyl

P/SiO catalysts prepared using TEOS and organosilanes with

2

phenyl groups; d) Pore size distribution for Ni

groups.

2

P/SiO

2

catalysts prepared using TEOS and organosilanes with phenyl

9

ChemCatChem

10.1002/cctc.201702001

In all cases, the isotherms are of type IV(H2), typical for mesoporous materials with

complex pore structure. Specific surface areas, calculated by using the BET method (Table 1),

2

-1

2

-1

varied between 449 m g (Ni

Ni P/SiO -TEOS catalysts showed a surface area of 465 m g . These results indicate that the

different types and amounts of organosilane monomers exert an effect on the resulting SBET of the

NiP /SiO catalysts. A more significant increase in SBET was obtained when copolymerizing

2

P/SiO

2

-5OTES) and 568 m g (Ni

2

P/SiO

2

-10PTES), while the

2

-1

2

-1

2

-1

2

-1

phenyl-functionalized silanes with TEOS (523 m g < SBET < 568 m g vs. 465 m g for TEOS-

derived material). However, the increase in SBET was much less pronounced for the materials

2

-1

prepared by copolymerization of alkyl-functionalized silanes and TEOS (449 m g < SBET < 484

2

-1

m g ). In these cases, a significant effect upon SBET was only observed for the Ni

2

P/SiO

2

-50MTES

2

-1

catalyst (SBET 532 m g ). The average pore diameter for the Ni

2

P/SiO

2

-TEOS catalyst was

3

-1

determined to be 7.3 nm with a pore volume of 0.53 cm g . When alkyl or phenyl-functionalized

silanes were copolymerized with TEOS, the average pore size of the catalysts varied between 4.7

nm (Ni

catalysts, the pore size decreased for all the catalysts supported on organohybrid silicas. The only

exception to this trend was the Ni P/SiO -50MTES catalyst. For this catalyst, the pore diameter

2

P/SiO

2

-ETES) and 12.7 nm (Ni

2

P/SiO

2

-50MTES). Compared to Ni

2

P/SiO

2

-TEOS

2

3

-1

(

12.7 nm) and pore volume (0.68 cm g ) were both higher than for Ni

2

P/SiO -TEOS catalyst. As

2

expected, these results imply that the incorporation of 50 mol% methyltriethoxysilane into the silica

network had a main impact on the general pore structure and volume. The decrease of the pore

volume and pore diameter for most of the catalysts is due to the incorporation of the organic groups

3

0

and has been observed before for organic silica hybrid materials. The increased pore volume and

diameter of the Ni P/SiO -50MTES catalyst, compared to the Ni P/SiO -TEOS and Ni P/SiO

0MTES catalyst, likely results from an alteration of the general network of the material due to the

2

-

2

1

0

ChemCatChem

10.1002/cctc.201702001

high-content of organosilane, which can only form three bonds with other silicate moieties, in

contrast to TEOS, which can be connected to other additional four silicon atoms via Si-O-Si bonds.

Table 1 reveals another important feature of the Ni

2

P/SiO materials. A direct quantitative

2

determination of the amount of incorporated organic functionalities using elemental analysis is not

fully feasible, due to the presence of remnant carbonaceous materials after the thermal treatment at

o

4

00 C under Ar atmosphere, as indicated by the C-content of 8.6±0.1% found in Ni

2

P/SiO -TEOS

2

(Table 1). Therefore, it was only possible for us to estimate the C-content attributable to the

organosilane incorporation in the organohybrid silicas. By subtraction of the C-content of

Ni

2

P/SiO

2

-TEOS from those content of the other materials, we could verify that, apart from

-20BTSB, there is a good agreement between the estimated and expected C-content in

the samples.

To examine further whether the organosilane groups are preserved after the thermal

treatments, ATR-IR spectra of the phosphided materials were recorded (Figure 2). For Ni P/SiO

0MTES and Ni P/SiO -50MTES, the incorporation of methyl groups into the silica network is

verified by the sharp band at around 1270 cm , assigned to the symmetric bending vibration of

2

-

2

-1

3

1

CH

number of incorporated methyl groups into the network (Figure 2d and 2e). For the Ni

0ETES catalyst (Figure 2c), the band relative to the symmetric bending vibration of the Si-CH

3

group in Si-CH (Figure 2d and 2e). Logically, the signal intensity also correlates with the

3

2

P/SiO

2

-

2

-1

group is only present as low-intensity shoulder (at around 1260 cm ), because the IR absorption

of the Si-CH decreases compared to the Si-CH group. For the Ni P/SiO -5OTES catalyst, this

band was not detected, which might be a consequence of the low amount of incorporated octyl

groups and a further decrease of the absorption of the Si-CH -bending vibration, with increasing

alkyl chain length. In turn, the ATR-IR spectra of the phenyl modified catalysts exhibit the typical

2

3

2

3

1

-1

absorption bands of phenyl groups between 1650-1350 cm (aromatic carbon stretching vibrations)

1

ChemCatChem

10.1002/cctc.201702001

-1

and the C-H bending vibrations at 740 cm and 695 cm , indicating the successful incorporation

3

1

of phenyl groups into the silica network. However, in these cases, the characteristic silicon-phenyl

-1

linkage vibration, which appears at about 1120 cm ¸ is only present as a shoulder. This is due to

overlap with the broad and intense Si-O-Si asymmetric stretch vibrations, which also appears

-1 31

between 1000 and 1111 cm . Again, the intensity of the absorption bands agrees with the quantity

of incorporated organic groups (i.e. the IR band intensity increased from the Ni P/SiO -10PTES to

the Ni P/SiO -50BTSB catalyst). Finally, the absence of the characteristic IR bands related to

carboxylic/carboxylate groups (e.g. for R-COOH species: an intense signal centered at about 1675

2

−1

cm ; for the carboxylate ion form (a broad band between 1320 and 1450 cm and a signal centered

3

−1

2

o

at about 1530 cm ) reveals that CA was decomposed at 400 C (forming a CO

2

and H O, and a

2

carbonaceous material, as revealed by elemental analysis, Table 1).

Figure 2 ATR-IR spectra of Ni

2

P supported on alkyl-modified silicas: a) Ni

P/SiO -20MTES, e) Ni P/SiO -50MTES; Ni P supported on phenyl-modified silicas: f)

P/SiO -10PTES, h) Ni P/SiO -20BTSB, i) Ni P/SiO -50BTSB

2 2 2 2

P/SiO -TEOS, b) Ni P/SiO -5OTES, c)

Ni

2

P/SiO

2

-20ETES, d) Ni

-TEOS, g) Ni

2

Ni

2

1

2

ChemCatChem

10.1002/cctc.201702001

XRD patterns of the Ni/SiO

2

catalyst precursor after thermal treatment at 400°C under Ar

atmosphere (Figure 3) displays the typical (111) and (200) reflections characteristic of metallic Ni

2 = 44° and 51°, respectively, Figure 3a). This observation indicates that the reduction of Ni(II)

(

sites in the precursor can take place without the need of a reducing atmosphere (H

2

). This

observation is surprising because of the stability of NiO: Δ

f

Gᵒ(NiO) = –337.64 kJ mol⁻¹.³³

However, it is not unique, as recently, Abu-Zied and Asiri reported a detailed study of the thermal

3

4

decomposition of nickel citrate into Ni/NiO composite particles. In that study, the conversion of

o

NiO to Ni and O

2

occurred at 510 C under a N

2

atmosphere. Furthermore, previous results by

Dong et al. also demonstrated the conversion NiO(s) ⇄ Ni(s) + ½O2(g) to be possible for sputtered

o

35

NiO films heated under vacuum to temperatures as low as 400 C. Although this system differs

considerably from ours, those results lend insight into possible mechanism(s) underlining the

conversion of NiO into Ni, which was reported to start at the outermost surface of the film and

3

5

attributed to excess Ni(II) vacancies formed during the sputtering step. Ni(II) vacancies cause

2

-

neighboring O ions to become incompletely bonded. As a result, the decomposition of NiO films

should require less stringent conditions. However, the decomposition of NiO was demonstrated to

3

5

be limited to a film thickness of 100 nm. From previous observations reported in the literature,

the formation of the metallic Ni phase in our materials seems to be related to the formation of small

NiO clusters upon thermal decomposition of Ni citrate/nitrate species in the gel. Importantly, it is

apparent from the high C-content found in the Ni

2

P/SiO

2

catalysts (Table 1) that reducing gases,

hypothetically generated by partial decomposition of alkyl or aryl groups of the organohybrid

3

6

silicas, play no major role in the chemical reduction of the Ni in the material. In fact, considering

that the nominal Ni loading is 33 wt%, the reduction of Ni(II) to Ni(0) would consume a substantial

amount of the alkyl or aryl groups incorporated in the organohybrid silica support. Nonetheless,

1

3

ChemCatChem

10.1002/cctc.201702001

this hypothesis was not verified by our results, as demonstrated by the high C-content in the

samples (Table 1).

After the phosphidation of Ni/SiO

211) reflections for Ni P at 2 values of 40°, 44°, 47° and 54°, respectively, were observed. This

result demonstrates the successful transformation of the Ni/SiO precursor into the Ni P/SiO

Figure 3b-i). Notably, XRD patterns of the various Ni P/SiO catalysts, supported on either pure

silica or organohybrid silica supports, display no significant differences in terms of peak width,

2

precursors, the typical (111), (201), (210) and (300) and

(

2

(

2

which indicates that all the Ni

determined by TEM analysis).

2

P crystallites on the support are similar in size (ca. 12-13 nm, as

Figure 3 XRD of the Ni/SiO

2 2 2

precursor using 100% TEOS and Ni P/SiO catalysts obtained from phosphidation of a

Ni/SiO precursor prepared by using different amounts and different types of organosilanes.

2

To provide an in-depth analysis of the Ni

catalyst) and the Ni P/SiO -50MTES (best catalyst, as will be presented later), HR-TEM images

of the Ni P/SiO -TEOS and the Ni P/SiO -50MTES catalysts were collected (Figure 4). By

2

P phase in the Ni

2

P/SiO

2

-TEOS (benchmark

2

1

4

ChemCatChem

10.1002/cctc.201702001

inspection of the HR-TEM images, no significant difference between the two catalysts was

observed, as indicated by the histograms in Figure 4a and 4d. The average particle size for the

Ni

each catalyst, were found to be 13.6 ± 4.4 nm and 12.4 ± 3.9 nm, further supporting the statement

that there is no significant difference in terms of Ni P particle size between the two catalysts.

2

P/SiO

2

-TEOS and Ni

2

P/SiO -50MTES catalysts, determined by evaluating 200 particles for

2

Figure 4 a, b correspond to TEM images of Ni

2 2 2 2

P/SiO -TEOS catalyst; d, e TEM images of Ni P/SiO -50MTES; c)

and f) display the particle size distribution determined by evaluating the size of 200 nanoparticles.

In this section, it was demonstrated that Ni

2

P/SiO materials were successfully synthesized.

2

Furthermore, HR-TEM images of selected samples show that there are no significant differences

in terms of the particles size and distribution of the particles on the support. Nitrogen sorption

isotherms show that incorporation of phenyl groups into the framework results in a slight increase

of the SBET, while for the alkyl-modified catalyst the surface area decreases slightly with the only

exception of the Ni

Ni P/SiO -TEOS catalyst. Also with respect to the pore volume and pore diameter, the Ni

0MTES catalyst does not fit the general trend. For all the organically modified catalysts the pore

2

P/SiO

2

-50MTES catalyst, for which SBET increases slightly, compared to the

2

P/SiO -

2

5

1

5

ChemCatChem

10.1002/cctc.201702001

volume and pore diameter decreased, compared to the Ni

2

P/SiO

2

-TEOS catalyst. In contrast to this,

3

-1

for the Ni

2

P/SiO

2

-50MTES, the pore volume increases from 0.53 cm g (Ni

P/SiO -50MTES), and the pore diameter increases from 7.3 nm (Ni

P/SiO -50MTES). These results suggest that for the Ni P/SiO -50MTES

2

P/SiO

2

-TEOS) to

3

-1

0

.68 cm g (Ni

2

P/SiO -

2

TEOS) to 12.7 nm (Ni

2

the number of organosilanes introduced into the framework changes the structure of the material

compared to the unmodified one. This is likely due to the different bonding possibilities of the

organosilanes, resulting in a more open structure. Most importantly, elemental analysis and ATR-

IR spectroscopy of the samples provide firm evidence that there is a successful incorporation of

different amounts of methyl and phenyl groups into the silica network, which remained in the

o

structured even after thermal treatment at 400 C under Ar flow.

Catalyst tests

For the catalyst test, we chose the HDO of guaiacol as the model reaction (at 295 °C and 5 MPa H

2

pressure for 3 h). Using the Ni P/SiO -TEOS catalyst (Table 2, entry 1, benchmarking catalyst), a

2

6

3

9% conversion was achieved. The main products of the reaction are cyclohexane formed with a

6% yield and benzene with a 17% yield. Cyclohexene was obtained only in minor quantities (<1%

yield). Besides fully deoxygenated products, also partially deoxygenated aromatic and aliphatic

products were found. As oxygenated aromatic compounds, anisole and phenol were observed with

9

% and 5% yields, respectively. As aliphatic oxygenates, cyclohexanol, cyclohexanone and 2-

methoxycyclohexanol were formed, but in minor quantities (yields lower than 1%).

Ni P catalysts supported on organohybrid silicas modified with alkyl groups (Table 2, entry

-5), showed a decrease in activity by approximately 20%, compared to the benchmarking catalyst

Ni P/SiO -TEOS. The main products of the reactions are still cyclohexane and benzene. Notably,

an increase (by a factor of 2.5) in the amount of methylsilane groups incorporated in the silica had

2

1

6

ChemCatChem

10.1002/cctc.201702001

no effect on the reaction outcome (Table 2, entry 4 and 5). Furthermore, substitution of methyl by

ethyl or octyl groups, while the R:TEOS ratio is kept at 0.2 or 0.05 (Table 2 entry 2 and 3), did not

alter the results. From this, it can be concluded that incorporation of alkyl functional groups into

the silica network decreased the activity for guaiacol HDO reactions, compared to Ni

TEOS. Strikingly, Ni P catalysts supported on organohybrid silicas modified with phenyl groups

Table 2 entry 6, 7, and 8) show an activity similar to that of Ni P/SiO -TEOS catalyst.

Interestingly, irrespective of the nominal quantity of incorporated phenyl groups, the conversion is

not further improved above 67%. In terms of product yields, Ni P/SiO -10PTES presents an

identical profile to the benchmarking catalyst. Therefore, considering the (similar) catalyst

performance, we chose to focus our study on Ni P/SiO -TEOS, Ni P/SiO -20BTSB and Ni P/SiO

0MTES catalyst, which presented the most distinguished features in the series.

2

P/SiO -

2

(

2

-

5

Table 2. Catalytic results for guaiacol HDO using Ni

2

P catalysts supported on pure silica or silica modified with

different types and amounts of organic functional groups.

Yield (%)

deoxygenated

aromatic

aliphatic

products

oxygenates

Conversion

entry

catalyst

(%)

69

43

40

49

1

2

3

4

9

8

5

6

5

6

7

8

1

2

3

4

5

Ni

2

P/SiO

2

-TEOS

36

16

15

22

23

17

13

11

12

<1

P/SiO

-5OTES

-20ETES

-20MTES

-50MTES

Ni

2

P/SiO

2

P/SiO

2

1

7

ChemCatChem

10.1002/cctc.201702001

6

7

8

Ni

2

P/SiO

2

-10PTES

-20BTSB

-50BTSB

67

69

66

35

43

42

16

13

12

<1

9

8

7

6

5

4

<1

Ni

Effects of support polarity on catalyst activity

To put the current results into a broader context, the general nature of pyrolysis oils must be

considered. While fossil crude oils are considered as nonpolar, due to their low oxygen content,

pyrolysis oils can be considered as polar and, therefore, hydrophilic. This is a direct consequence

of the high oxygen content in pyrolysis oils, present in the form of different acids, aldehydes, esters,

alcohols, ethers, ketones, phenolics, sugars, furans, and up to 50 wt% water.³⁷^,³⁸In this manner,

adjusting catalyst polarity should exert a direct effect on the interaction between catalyst and

substrate. To characterize the interactions of the prepared catalysts with the polar constituents of

푺푷

pyrolysis oils, the free energy of methanol adsorption (−휟푮_풂_풅_풔(푴풆푶푯)) and surface energies

were determined, using inverse gas chromatography (IGC) for Ni

2

P/SiO

2

-TEOS, Ni

2

P/SiO -

2

0BTSB and Ni P/SiO -50MTES catalysts (Table 3). IGC is a powerful technique to determine

2

the surface properties associated with catalytic materials (details given in experimental section)

and it gives information on physical (dispersive) and chemical (specific/polar) interactions through

2

5

the adsorption of alkanes and polar probe molecules, respectively. Dispersive surface energy

provides insight into the surface hydrophobicity, higher the value of dispersive surface energy,

higher the hydrophobic character.³⁹^,⁴⁰In turn, the surface polarity (hydrophilicity) can be directly

correlated with the free energy of adsorption of polar probe molecules and specific surface energy.

Owing to detection and analysis difficulties while working with water as a probe molecule in IGC

(i.e. flame ionization detector, FID, cannot detect water), methanol was taken as a substitute for

푫

푺

free energy calculations. The dispersive surface energy (휸 ) was determined as a measure of the

1

8

ChemCatChem

10.1002/cctc.201702001

organophilicity of the catalyst. Among the three catalysts, the Ni

2

P/SiO

2

-50MTES catalyst

푺푷

-1

displayed the lowest −휟푮_풂_풅_풔(푴풆푶푯) value of 10.5 kJ mol , compared to 15.0 kJ mol and 17.8

-1

kJ mol for the Ni

2

P/SiO

2

-TEOS and Ni

2

P/SiO

2

-20BTSB catalyst, respectively. Considering the

dispersive surface energy, the Ni

².

2

P/SiO

2

-20BTSB catalyst displayed the highest value of 80 mJ m^-

Interestingly, the dispersive surface energies for the Ni

2

P/SiO

2

-TEOS and Ni

2

P/SiO -50MTES

2

-2

catalysts, which were determined to be 47 mJ m and 50 mJ m , are very similar, in spite of the

presence of methyl groups, in the case of the Ni P/SiO -50MTES catalyst. With higher dispersive

surface energy, Ni P/SiO -20BTSB possesses the most hydrophobic character among the three

2

catalysts. Literature validates dispersive surface energy as a measure of hydrophobicity of

materials, with carbons samples with more hydrophobic character possessing high dispersive

surface energy >150 mJ m^-²,⁴¹^,⁴²while hydrophilic silica⁴¹^,⁴²and cellulose²⁵^,³⁹^,⁴¹^,⁴³^,⁴⁴exhibit

-

2

-2

lower values ~ 50 mJ m and ~ 40 mJ m , respectively. Hence, Ni

2

P/SiO

2

-20BTSB which exhibits

P/SiO -TEOS (47 mJ

-2

dispersive surface energy of 80 mJ m should be more hydrophobic than Ni

2

-2

P/SiO

-50MTES (50 mJ m^-²).

m ) and Ni

2

1

9

ChemCatChem

10.1002/cctc.201702001

푺푷

푫

푺

Table 3 Free energies of methanol adsorption (−휟푮_풂_풅_풔(푴풆푶푯)) and dispersive surface energies (휸 ) for the

Ni

2

P/SiO

2

-TEOS, Ni

2

P/SiO

2

-20BTSB, Ni

2

P/SiO

2

-50BTSB, Ni

2

P/SiO

2

-20MTES and Ni

2

P/SiO

2

-50MTES catalysts,

measured using inverse gas chromatography.

Inverse gas chromatography measurements

−ΔG (MeOH) [kJ mol^-¹]

SP

γ [mJ m ]

D

-2

Sample

ads

S

Ni

2

P/SiO

2

-TEOS

15.0

17.8

13.9

13.0

10.5

47

Ni

2

P/SiO

2

-20BTSB

-50BTSB

80

82

51

50

Ni

2

P/SiO

2

-20MTES

-50MTES

2

The interactions of the modified catalysts with a typical polar constituent of pyrolysis oils

were exemplified using guaiacol as an example Figure 5a. As already discussed above,

modification of the catalyst with methyl groups decreased the activity for guaiacol HDO from

6

5±5%, for the Ni

contrast, the conversion of the Ni

found to be 67±3%, and hence did not decrease compared to the Ni

a). This correlates with the free energies of methanol adsorption, which are higher for the

Ni P/SiO -TEOS and Ni P/SiO -20BTSB catalysts, compared to the Ni P/SiO -50MTES catalyst.

2

P/SiO

2

-TEOS catalyst, to 46±4% for the Ni

P/SiO -20BTSB catalyst, modified with phenyl groups, was

P/SiO -TEOS catalyst (Figure

2

P/SiO

2

-50MTES catalyst. In

2

5

2

The modification of the catalyst with alkyl groups reduced the catalyst affinity towards polar

constituents of pyrolysis oils (e.g. guaiacol), compared to unmodified or phenyl group modified

catalysts. However, despite the fact that phenyl groups are hydrophobic, they are capable of

establishing strong π-π interactions⁴⁵with the phenolic components present in pyrolysis oils.

Therefore, this type of interaction appears to be vital to sustaining the conversion at the levels as

high as that found for the hydrophilic Ni

2

P/SiO -TEOS catalyst.

2

0

ChemCatChem

10.1002/cctc.201702001

HDO of the nonpolar bio-oil model compound diphenyl ether was performed to confirm

the assumption that the affinity of the catalyst towards less polar bio-oil compounds can be

characterized by their dispersive surface energy. Evidently, the catalytic results correlate well with

the dispersive surface energies of the three catalysts (Table 3). Thus, the Ni

2

P/SiO -20BTSB

2

catalyst, which has the highest dispersive surface energy among the three catalysts, displays an

increased diphenyl ether conversion, compared to the other two catalysts. The other two catalysts

on the other hand, display similar diphenyl ether conversion, which is in good agreement with their

comparable dispersive surface energies. The diphenyl ether conversion using the Ni

0BTSB catalyst, which was found to be 84±4%, is significantly higher compared to the other two

catalysts, whereas in the case of Ni P/SiO -TEOS the diphenyl ether conversion was found to be

0±7%, and 49±1%, in the case of the Ni P/SiO -50MTES catalyst (Figure 5b). Considering the

product distributions, cyclohexane was the primary product of the reaction, when Ni P/SiO

0BTSB was used with a selectivity of 59±4% compared to a benzene selectivity of 40±4%. For

2

P/SiO -

2

4

2

-

2

the other two catalysts, benzene was found as the main product, with selectivity values of

approximately 50%, and cyclohexane with ca. 40% selectivity.

2

1

ChemCatChem

10.1002/cctc.201702001

Figure 5 a) HDO of guaiacol using Ni

conditions: catalyst (50 mg), guaiacol (5 mmol), n-octane (9.5 mL), 295 °C, 5 MPa H

using Ni P/SiO -TEOS, Ni P/SiO -50MTES and Ni P/SiO -20BTSB. Reaction conditions: catalyst (50 mg), diphenyl

ether (2.5 mmol), n-octane (2.5 mmol), 295 °C, 5 MPa H , 6 h. All reactions were performed as triplicates.

2

P/SiO

2

-TEOS, Ni

2

P/SiO

2

-50MTES and Ni

2

P/SiO

2

-20BTSB catalysts. Reaction

2

, 3 h; b) HDO of diphenyl ether

2

These results show how the modification of Ni

2

P/SiO catalysts with either alkyl or phenyl

2

functions influences the affinity of the catalyst towards either polar or nonpolar bio-oil parts.

Evidently, the affinity towards polar compounds, characterized by the free energy of methanol

adsorption, is lowered by incorporation of alkyl functional groups into Ni

2

P/SiO , while this was

2

not the case when phenyl groups were incorporated. Moreover, the presence of phenyl groups also

intensified the interactions with the less polar substrate diphenyl ether, shown by the higher

dispersive surface energy, resulting in a higher conversion for the Ni

compared to the other two catalysts.

2

P/SiO -20BTSB catalyst

2

ChemCatChem

10.1002/cctc.201702001

Implications of support polarity for catalyst stability

Besides conversion, catalyst stability is another crucial parameter to assess the performance of Ni

catalyst phase, because these are known to deactivate in the presence of water via oxidation of the

2

P

2

6

active phase. To explore the influence of modifying Ni

amounts of organic functional groups on the stability of the Ni

were performed. The results are presented in Figure 6, for the Ni

0MTES catalysts, the first run was reproduced four times, the second run three times and the third

run two times (only in the case of the Ni P/SiO -50MTES catalyst). The error bars in Figure 6

indicate that the spread of the conversion and yield is within the range 5-12%.

2

P/SiO

2

catalysts with different types and

P phase, recycling experiments

P/SiO -TEOS and Ni P/SiO -

2

5

2

Figure 6 Guaiacol recycling experiments using the Ni

Ni P/SiO -20MTES and Ni P/SiO -50MTES catalysts. Reaction conditions: catalyst (50 mg), guaiacol (5 mmol), n-

octane (9.5 mL), 295 °C, 5 MPa H , 3 h. For the Ni P/SiO -TEOS and Ni P/SiO -50MTES catalysts, the first run was

repeated four times, the second run three times and the third run two times.

2 2 2 2 2 2

P/SiO -TEOS, Ni P/SiO -20BTSB, Ni P/SiO -50BTSB,

2

3

ChemCatChem

10.1002/cctc.201702001

26

As expected from previous studies, Ni

2

P/SiO

2

-TEOS catalyst strongly deactivated upon

recycling. The conversion decreased from 59±12%, in the first run, to 18±3% in the second run

Figure 6). Strikingly, Ni P/SiO -20BSTB and Ni P/SiO -50BSTB, modified with phenyl groups,

and Ni P/SiO -20MTES catalyst, also suffered from substatial deactivation after the first run.

Consistently, Ni P/SiO -20BSTB, Ni P/SiO -50BSTB and Ni P/SiO -20MTES showed a

deactivation profile similar to the Ni P/SiO -TEOS catalyst, in terms of decreasing conversion and

decreasing yield of deoxygenated products (Figure 6). In stark contrast, the cyclohexane yield for

Ni P/SiO -50MTES catalyst remained around 20% throughout the three recycling runs.

Interestingly, the benzene yield also decreased for this catalyst from 8±3% down to 2.0±0.1% from

the first to the third run. Most importantly, among the five investigated catalysts, only Ni P/SiO

0MTES catalyst presented an improved stability. Indeed, only a slight decrease in the conversion

from 43±4%, in the first run, to 39±2% in the second run, and 34±4% in the third run was observed.

The improved stability of the Ni P/SiO -50MTES catalyst was further corroborated by HR-

TEM results. Figure 7 compares HR-TEM images of the fresh Ni P/SiO -TEOS (Figure 7a) and

-50MTES (Figure 7b) catalysts to the catalysts after one HDO reaction. In the case of

(

2

-

5

2

Ni

2

P/SiO

2

the Ni

2

P/SiO -TEOS catalyst, the catalyst after one reaction visually differs significantly from the

2

fresh catalyst. There are new support domains formed in the material, which seem to be amorphous

and were not present on the fresh catalyst (indicated by yellow arrows in Figure 7a). In comparison,

the TEM image of Ni

2

P/SiO

2

-50MTES catalyst after one reaction (Figure 7b) does not differ

P/SiO -50MTES catalyst.

substantially from the fresh Ni

2

4

ChemCatChem

10.1002/cctc.201702001

Figure 7 TEM images of the fresh and recycled Ni

2

P/SiO

2

-TEOS catalysts (a) and Ni

2

P/SiO

2

-50MTES (b). Yellow

arrows indicate the new domains formed on the catalyst during recycling.

2

5

ChemCatChem

10.1002/cctc.201702001

XRD analysis of the recycled catalysts after one reaction provides additional pieces of

evidence for the statement that only the Ni P/SiO -50MTES catalyst repels water strongly enough

to prevent catalyst deactivation (Figure 8). Clearly, comparing the XRD pattern of the fresh

Ni P/SiO -TEOS catalyst (Figure 8a) to the XRD pattern of the Ni P/SiO -TEOS catalyst after

one reaction (Figure 8b) revealed significant differences between the two. Besides the reflections

for Ni P (▲), also those for the Ni12 phase (●), a more P lean phase compared to Ni P, and for

the NiP (■) phase were found, which is richer in P compared to Ni P. Also the XRD patterns of

the Ni P/SiO -20BSTB (Figure 8c) and Ni P/SiO -50BSTB (Figure 8d) catalysts after one

reaction display new reflections which can be related to the Ni12 phase Moreover, in these cases,

the intensity of the reflections is much lower, which could indicate a decrease in the crystallinity.

In contrast, the XRD patterns of the Ni P/SiO -20MTES (Figure 8e) and Ni P/SiO -50MTES

Figure 8f) catalysts, after one run, still resembled the pattern of a fresh Ni P/SiO catalyst. XRD

results demonstrate that the MTES modified catalysts to be less affected by water during recycling,

compared to the other catalysts. Nonetheless, very low-intensity reflections for the Ni12 phase

are observed. Even though the XRD pattern of the spent Ni P/SiO -20MTES catalyst still

P phase, recycling experiments showed that this catalyst presents decreased

2

P

5

2

P

5

.

2

(

2

P

5

2

resembled the Ni

performance.

2

6

ChemCatChem

10.1002/cctc.201702001

Figure 8 a) XRD of the fresh Ni

0BTSB catalyst after the first run, d) the Ni

first run, f) the Ni P/SiO -50MTES after the first run.

2

P/SiO

2

-TEOS catalyst b) the Ni

2

P/SiO

2

-TEOS catalyst after first run, c) the Ni

2 2

P/SiO -

2

P/SiO -50BTSB catalyst after first run, e) the Ni

2

P/SiO -20MTES after

2

It has previously been reported that SiO

2

supported Ni P catalysts deactivate faster during

2

6

HDO than Ni P supported on other supports, due to the hydrophilic nature of silica. Through the

2

hydrophilic nature of the support, water adsorption and the hydration of the catalyst surface is

promoted and, therefore, the oxidation of the active phase by water is accelerated. To explain the

deactivation behavior of the different catalysts, the affinity of these towards water was

SP

S

characterized, using specific surface energies (γ ), derived from polar probe interactions and the

SP

free energies of methanol adsorption (−ΔG_a_d_s(MeOH)).

SP

Considering the γ_S(Figure 9), the Ni

2

P/SiO

2

-20BSTB displays the highest value of 257

2

mJ/m , which decreases for the Ni

2

P/SiO

2

-50BSTB catalyst down to 193 mJ/m , in line with the

SP

increased amount of phenyl groups. Changing the organic function from phenyl to methyl, the γ_S

2

decreases to 121 mJ/m for the Ni

2

P/SiO

2

-20MTES catalyst. For the Ni

2

P/SiO -50MTES catalyst,

2

7

ChemCatChem

10.1002/cctc.201702001

2

this value further decreases to 81 mJ/m , consistent with the increased number of methyl groups.

SP

2

The γ_Sfor Ni

Ni P/SiO -50MTES catalyst. Nonetheless, the free energy of methanol adsorption, which can be

interpreted as an indicator for the affinity of the catalyst towards water, is 15.0 kJ mol for the

2

P/SiO

2

-TEOS, however, was found to be 89 mJ/m and is comparable to the one of

2

-1

2

1

Ni

0MTES catalyst, which displays a free energy of methanol adsorption of only 10.7 kJ mol , which

is the lowest value among all the investigated catalysts (Figure 9). In line with the observed

deactivation of the Ni P/SiO -20MTES catalyst during recycling, also this catalyst displays a

higher free energy of methanol adsorption (13.1 kJ mol ) than the Ni

Consistently, the free energies of methanol adsorption for the phenyl modified catalysts, which

2

P/SiO

2

-TEOS catalyst. This is comparably high, in contrast to the value of the Ni

2

P/SiO -

2

-

1

5

2

-1

2

P/SiO -50MTES catalyst.

2

-1

also deactivated during recycling, were found to be 13.9 kJ mol and 17.8 kJ mol for the

Ni P/SiO -50BSTB and the Ni P/SiO -20BSTB catalyst, respectively. The higher polar character

2

of phenyl modified catalysts can be explained in terms of the ability of phenyl groups to interact

with polar compounds via π-π interactions as well as via weak hydrogen bonding interactions,

which is well known from gas and liquid chromatography.⁴⁵These results suggest that the

interactions between water present under reactions conditions and the catalyst can be correlated

SP

ads

with (−ΔG (MeOH)), and must be decreased below a certain threshold to prevent the catalyst

from immediate deactivation. In the case of methyl group modified catalysts, this threshold lies

somewhere between a R:TEOS ratio of 0.2 and 0.5. Overall, these results highlight how proper

adjustment of the catalyst polarity can help to alleviate the deactivation of Ni

2

P/SiO catalysts.

2

8

ChemCatChem

10.1002/cctc.201702001

푺푷

Figure 9 Specific surface energy (휸_푺, represented by the colored bars) and free energy of methanol adsorption (-

ꢃꢄ

ΔG_ꢀ_ꢁ_ꢂ(MeOH), indicated by squares in gray color) for the catalysts tested in the recycling experiments.

Conclusions

In this work, the influence of the catalyst polarity on the activity and stability of HDO catalysts

was investigated, using a Ni

2

P/SiO catalyst as a model catalyst. Among the differently modified

2

catalysts, only the catalyst modified with methyl functions, having an initial R:Si ratio of 0.5, did

not show immediate deactivation during guaiacol HDO recycling experiments. Consistently, this

catalyst also displayed the lowest free energy of methanol adsorption. Strikingly, phenyl groups

are not suitable to stabilize the catalyst against water deactivation. Apparently, these groups may

establish polar π-π interactions, uncovering the support surface against water. Our results

demonstrate the modification of silica with methylsilane groups to be more suitable for improving

the stability of HDO catalysts against the deleterious effect of water under reaction conditions.

Apparently, there is a threshold, which is reached with increasing the amount of alkylsilane

2

9

Article Doi

Impact of Hydrophobic Organohybrid Silicas on the Stability of Ni2P Catalyst Phase in the Hydrodeoxygenation of Biophenols

DOI: 10.1002/cctc.201702001

Source and publish data:

Authors:

Article abstract of DOI:10.1002/cctc.201702001

Full text of DOI:10.1002/cctc.201702001

Products guided by the article

R&D Labs maybe for 100-66-3

Relevant to this article

Hot Product