CeO2 Catalyst

Apr 5, 2016 - After this paper was published ASAP April 20, 2016, text was moved from the caption of Figure 5 to the caption of Figure 4. The correcte...
16 downloads 19 Views 2MB Size
Subscriber access provided by UNIV OF CALIFORNIA SAN DIEGO LIBRARIES

Article x

2

Performance, Structure and Mechanism of ReO–Pd/CeO Catalyst for Simultaneous Removal of Vicinal OH Groups with H

2

Nobuhiko Ota, Masazumi Tamura, Yoshinao Nakagawa, Kazu Okumura, and Keiichi Tomishige ACS Catal., Just Accepted Manuscript • DOI: 10.1021/acscatal.6b00491 • Publication Date (Web): 05 Apr 2016 Downloaded from http://pubs.acs.org on April 12, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Catalysis is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

Performance, Structure and Mechanism of ReOx–Pd/CeO2 Catalyst for Simultaneous Removal of Vicinal OH Groups with H2

Nobuhiko Ota,[a] Masazumi Tamura,[a, c] Yoshinao Nakagawa,*[a, c] Kazu Okumura,[b] Keiichi Tomishige*[a, c]

[a]

Department of Applied Chemistry, School of Engineering, Tohoku University, 6-6-07, Aoba, Aramaki, Aoba-ku, Sendai 980-8579, Japan

[b]

Department of Applied Chemistry, Faculty of Engineering, Kogakuin University, 2665-1 Nakano-machi, Hachioji, Tokyo 192-0015, Japan [c]

Research Center for Rare Metal and Green Innovation, Tohoku University, 468-1, Aoba, Aramaki, Aoba-ku, Sendai 980-0845, Japan

*Corresponding author: Keiichi Tomishige School of Engineering, Tohoku University, 6-6-07, Aoba, Aramaki, Aoba-ku, Sendai, 980-8579, Japan E-mail: [email protected]

*Corresponding author: Yoshinao Nakagawa School of Engineering, Tohoku University, 6-6-07, Aoba, Aramaki, Aoba-ku, Sendai, 980-8579, Japan E-mail: [email protected]

1 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

Abstract

2

Ceria-supported rhenium catalyst modified with palladium (ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re

3

= 0.25)) is still the best catalyst for simultaneous hydrodeoxygenation. Higher Re loading amount

4

decreased the activity. The simultaneous hydrodeoxygenation of cyclic vicinal diols occur with high

5

cis-stereoselectivity. ReOx–Pd/CeO2 catalysts were characterized by means of XRD, TEM, H2-TPR,

6

XAFS, XPS, Raman, and DFT calculations. The Re species on ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re =

7

0.25) catalyst after reduction and after stoichiometric reaction of 1,2-hexanediol to 1-hexene were

8

ReIV and ReVI, and the ReIV species converted to ReVI through the stoichiometric reaction. The Re

9

species on ReOx–Pd/CeO2 are proposed to be randomly located on CeO2 surface, and probably only

10

monomeric Re species have catalytic activity for simultaneous hydrodeoxygenation. This model can

11

explain the higher activity of Re = 2 wt% catalyst than those of higher Re loading catalysts. The

12

reaction is proposed to proceed by the tetra/hexavalent redox cycle of Re center in the catalysis

13

followed by hydrogenation.

14

15

Keywords

16

heterogeneous catalysis, rhenium oxide, palladium, hydrodeoxygenation, deoxydehydration.

17

18

Introduction

19

In recent years, because of the increasing need for energy and material sources, biomass-derived

20

feedstocks have attracted attention as alternative renewable resources for chemicals and fuels [1–7].

21

Based on this background, biomass-derived chemicals, such as sugars and sugar alcohols, are

22

attractive starting materials to substitute for unsustainable fossil resources. Since oxygen is rich in

23

these biomass-derived materials compared with most value-added chemicals, hydrodeoxygenation,

2 Environment ACS Paragon Plus

Page 2 of 56

Page 3 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

1

especially partial hydrodeoxygenation, is a key reaction in the conversion of biomass to value-added

2

chemicals. For substrates with a few OH groups such as glycerol, a number of partial

3

hydrodeoxygenation systems that selectively remove one OH group have been reported [8–12].

4

However, it is much more difficult to selectively remove OH groups with these catalysts from

5

substrates with four or more OH groups such as erythritol, xylitol, and sorbitol [13, 14]. To develop

6

the conversion processes of these polyols, more effective and selective catalytic systems for

7

hydrodeoxygenation may be necessary. One of such systems is deoxydehydration (DODH; or

8

didehydroxylation), which converts vicinal two OH groups to one C=C double bond [15, 16].

9

Previously, Re [17, 18], V [19], and Mo [20–22] homogeneous and heterogeneous catalysts,

10

especially Re, have been reported to be active in DODH. DODH systems typically use non-H2

11

reductants such as PPh3 [23, 24], sulfite [25, 26], metal [27], hydroaromatics [28, 29] and alcohols

12

[29–37], and lower yields of the DODH products are obtained when H2 is used as a reductant [24, 29,

13

35, 38–40]. Existing data and proposals from the prior stoichiometric or catalytic DODH studies

14

point to three basic reactions that are likely involved in the DODH catalytic cycle: (1) two-electron

15

reduction of high valent species such as ReVII, (2) the formation of diolate by the coordination of diol

16

with the reduced Re species, and (3) elimination of produced alkene from the lower-valent-Re diolate

17

and regeneration of the catalyst. Steps (1) and (2) may be interchangeable: (1)’ Formation of diolate

18

with high valent species and (2)’ two-electron reduction of Re diolate.

19

As a similar reaction to the DODH, we reported simultaneous hydrodeoxygenation over ReOx–

20

Pd/CeO2 catalyst (Scheme 1) [41]. In this simultaneous hydrodeoxygenation process, two vicinal OH

21

groups are transformed to H atoms, and it is particularly effective in the selective conversion of

22

polyols. This catalyst system has excellent stability: the catalyst can be reused without loss of

23

activity and selectivity when the very small amount of organic deposit on the used catalyst is

3 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

removed by calcination. Because the substrate, main product, and active element of simultaneous

2

hydrodeoxygenation are similar to those of DODH, our system was supposed to be a combination of

3

DODH + hydrogenation. It is characteristic that ReOx–Pd/CeO2 catalyst showed much higher

4

activity (TOF) than all previously reported homogeneous and heterogeneous catalysts for DODH

5

[41]. Therefore, even though the reaction mechanism of ReOx–Pd/CeO2 catalyst is similar to that of

6

homogeneous catalysts, there can be some differences such as the structure of active sites and the

7

activation mechanism of the reductant.

8

It is well known that Re species can have various valences (0~+7), and the structure and/or

9

valence of Re species is much influenced by conditions. For example, Re species on reduced

10

Ir−ReOx/SiO2, which is catalytically active in hydrogenolysis of C-O bonds in ethers and alcohols,

11

are three-dimensional ReOx sub-nanometer order clusters with average Re valence of +2~3, and Ir

12

metal particles covered with those ReOx clusters [42]. Reduced Re–Pd/SiO2 (14 wt% Re, 1 wt% Pd),

13

which is catalytically active in hydrogenation of carboxylic acids, contains Pd0, Re0(HCP), Re0(FCC)

14

and Ren+ species [43]. Re0(HCP) and Re0(FCC) are present as metal particles with nanometer size,

15

and they are suggested to be covered with Ren+ species. Thus, in this report, ReOx–Pd/CeO2 catalysts

16

were characterized by means of various characterization methods for each state in the catalytic cycle.

17

On the basis of the spectroscopic analyses, investigation of reactivity of various substrates, kinetic

18

studies, and DFT calculations, we confirmed that the reaction proceeded via DODH + hydrogenation.

19

The active site was proposed to be monomeric ReIV/ReVI species stabilized on the CeO2 support.

20

21

Results and discussion

22

Screening of catalyst

23

First, 1,4-anhydroerythritol (1,4-AHERY) was used as a model substrate. In the preliminary report

4 Environment ACS Paragon Plus

Page 4 of 56

Page 5 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

1

[41], we tested various combination of group 5–7 metal (MOx) + Pd and ReOx + noble metal using

2

silica support, and the other combinations than ReOx–Pd showed lower activity and/or lower

3

selectivity to tetrahydrofuran. Besides, screening of support using ReOx–Pd/Support was conducted,

4

and ceria support dramatically increases the catalytic activity. However, the most effective catalyst

5

might be changed by employing ceria support. Therefore, initially, various combination of group 5–7

6

metal (MOx) + Pd (Figure S1) and ReOx + additive metal (selected from the viewpoint of the

7

hydrogen activation and/or hydrogenation activity) (Figure S2) were tested again using ceria support.

8

Nevertheless, ReOx was the most effective as an active species for simultaneous hydrodeoxygenation,

9

and other metal oxides were almost inactive (little (99%). Over ReOx–Ru/CeO2 or ReOx–Pt/CeO2, a small amount of 1-butanol was

16

observed. The main product over ReOx–Ir/CeO2 was dihydrofuran which is a precursor of

17

tetrahydrofuran. The addition of base metals (Co, Ni and Cu) did not promote this reaction, and

18

decreased the selectivity to tetrahydrofuran significantly.

19

Figure 1 shows the dependence of Pd amount on the simultaneous hydrodeoxygenation of

20

1,4-AHERY over ReOx–Pd/CeO2. The loading amount of Re was set constant (Re = 2 wt%).

21

Although the catalytic activity hardly depended on the amount of Pd, the conversion was maximum

22

at Pd/Re=0.25 and the conversion became 38 times higher than that on ReOx/CeO2. In the range of

23

Pd/Re ≥ 0.5, in addition to the slight decrease in activity, isomerization of 1,4-AHERY to

5 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

1,4-anhydrothreitol (1,4-AHTHR), which might be Pd-catalyzed dehydrogenation-hydrogenation

2

process, increased gradually with increasing Pd amount.

3

The effect of loading amount of Re on the simultaneous hydrodeoxygenation of 1,4-AHERY was

4

also investigated (Figure 2). The Pd/Re molar ratio was set constant (Pd/Re = 0.25). The selectivity

5

to THF was very high (>99%) except in the case of very large loading amount of Re (10 wt%, 97%

6

selectivity). The catalytic activity increased up to 2 wt% of loading amount, and then decreased

7

monotonically. In spite of the much higher loading amount of Re, the activity of 10 wt% Re catalyst

8

was even less than half of that of the best catalyst (2 wt% Re), which represents less than 1/10 of

9

activity per Re. The relation between Re loading amount and the activity will be discussed in later

10

sections based on the structure model. After all, it was indicated again that the ReOx–Pd/CeO2 (Re =

11

2 wt%, Pd/Re = 0.25), which was chosen in our previous report [41], is the optimum catalyst for this

12

reaction.

13

14

Effect of reaction conditions for the reaction of 1,4-AHERY over ReOx–Pd/CeO2

15

Figure 3 shows solvent effects on the simultaneous hydrodeoxygenation of 1,4-AHERY over

16

ReOx–Pd/CeO2. The reaction rate was much lower in water solvent. Ether and alkane solvents such

17

as 1,4-dioxane, 1,2-dimethoxyethane and dodecane, were good solvents, and 1,4-dioxane was the

18

best solvent in terms of the activity and selectivity among the tested solvents. In the case of alcohol

19

solvents such as 1-pentanol and 3-pentanol, the reaction proceeds slightly slower than the case of

20

ether solvents. These lower reaction rates are probably due to the competing coordination of solvent

21

molecules with the substrate at the active species. The difference in the activity in the solvents such

22

as ether, alkanes and alcohols was not large except the case of water. It should be noted that the

23

simultaneous hydrodeoxygenation of 1,4-AHERY was catalyzed under the neat condition in high

6 Environment ACS Paragon Plus

Page 6 of 56

Page 7 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

ACS Catalysis

selectivity.

2

The effects of H2 pressure and 1,4-AHERY concentration in the simultaneous hydrodeoxygenation

3

of 1,4-AHERY over ReOx–Pd/CeO2 was investigated. The reaction order with respect to H2 pressure

4

around the standard conditions was estimated to be 0 (Figure 4, the detailed data are shown in Figure

5

S4), which suggests that the reaction steps associated with hydrogen are fast, including reduction of

6

the catalyst and hydrogenation. The reaction rate was saturated with increasing concentration of

7

1,4-AHERY, and the reaction order with respect to the concentration of 1,4-AHERY around 15–20

8

wt% in the solvent was also estimated to be 0 (Figure 5, the detailed data are shown in Figure S5).

9

This behavior indicates that the reaction rate can decrease with increasing the conversion. In the case

10

of the standard reaction conditions (1,4-AHERY 1 g, 1,4-dioxane 4 g, the concentration of

11

1,4-AHERY 20 wt%), the reaction rate can decrease with increasing the conversion and decreasing

12

the 1,4-AHERY concentration. According to the time course of 1,4-AHERY reaction in the previous

13

report [41], the conversion was increased by ~20% at first 4 h, and because of decrease of the

14

reaction rate at higher conversion, the conversion reached >95% at 48 h. This behavior can be

15

explained to some extent by the effect of the 1,4-AHERY concentration.

16

Because alcohols such as 3-pentanol have been used as a reductant in the DODH reaction over

17

homogeneous Re-catalysts [29–37], 3-pentanol was tested instead of H2 as a reductant in the reaction

18

of 1,4-AHERY over ReOx–Pd/CeO2 catalyst (Table S1). Here, the reaction was performed in similar

19

conditions to standard ones except using 3-pentanol as the reductant and Ar gas to pressurize. The

20

conversion rate in the reaction using 3-pentanol was much lower than that in the reaction using H2

21

(2% vs. 38% conversion). These results suggest that the Re species on CeO2 can be reduced with H2,

22

not with alcohols.

23

7 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

Page 8 of 56

Reactivities of various substrates

2

The ReOx–Pd/CeO2 catalyst was applied to various substrates, and the reactivities were compared

3

at low conversion level (Table 1). The reactivity of 1,4-AHTHR, 3-hydroxytetrahydrofuran, and THF

4

was much lower than that of 1,4-AHERY (Table 1, entries 1–4). cis-1,2-Cyclopentanediol and

5

cis-1,2-cyclohexanediol were converted to corresponding cycloalkane at moderate conversion and

6

high selectivity, however, the reactivity of trans-1,2-cyclopentanediol and trans-1,2-cyclohexanediol

7

was much lower than that of cis- isomers (Table 1, entries 5, 6, 8, 9). These low reactivities of

8

trans-diols and mono-ols were also reported in Re-catalyzed DODH systems [17, 18]. These results

9

supported

DODH

+

hydrogenation and

mechanism

of

trans-1,2-cyclohexanediol,

ReOx–Pd/CeO2 the

system.

10

trans-1,2-cyclopentanediol

11

(cyclopentane and cyclohexane) were slightly formed probably via Pd-catalyzed isomerization, i.e.,

12

via dehydrogenation–hydrogenation process, to cis-isomer. In the case of trans-1,2-cyclopentanediol,

13

the cis-isomer, which was formed via Pd-catalyzed isomerization, was not detected. However, in the

14

case of trans-1,2-cyclohexanediol, the cis-isomer was detected in high selectivity with respect to the

15

conversion. These results suggest the difference in the presence of the trans-isomer on the reaction of

16

cis-isomer. For example, it is expected that adsorption of cis-1,2-cyclopentanediol to the Re center

17

was preferred for that of trans-1,2-cyclopentanediol, but there is no clear difference in the adsorption

18

between cis- and trans-1,2-cyclohexanediols to the Re center. In order to discuss the effect of

19

coexistence of the cis-diol and the trans-diol, the reactions of mixture of cis- and trans-diols (c : t =

20

1 : 1) were conducted (Table 1, entries 7 and 10). Conversion of trans-diols in these mixtures was

21

negligible (lower than experimental error), and the cycloalkane could be regarded as the product

22

from cis-diols. The yield of cyclopentane from mixture of cis- and trans-1,2-cyclopentanediols, was

23

comparable to that from pure cis-1,2-cyclopentanediol (Table 1, entries 5 and 7). However, In the

8 Environment ACS Paragon Plus

corresponding

From

cycloalkanes

Page 9 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

1

case of the reaction using mixture of cis- and trans-1,2-cyclohexanediols, the reactivity of

2

cis-1,2-cyclohexanediol was decreased to less than half of that in the absence of trans-isomer (Table

3

1, entries 8 and 10), and the reaction of cis-1,2-cyclohexanediol is suppressed by the presence of

4

trans-1,2-cyclohexanediol. Considering that the dihedral angle (O–C–C–O) of the vicinal OH groups

5

of cis-1,2-cyclopentanediol (~40°; determined by DFT calculation) is smaller than that in

6

trans-1,2-cyclopentanediol (~70°), trans-1,2-cyclopentanediol probably cannot form diolate with Re

7

center. On the other hand, trans-1,2-cyclohexanediol, cis-1,2-cyclohexanediol and 1,2-alkanediols

8

have similar dihedral angle of the vicinal OH groups (~60°), and the presence of

9

trans-1,2-cyclohexanediol can decrease the reactivity of the coexistent cis-1,2-cyclohexanediol by

10

the competitive formation of diolates. The reactivity of 1,2-alkanediols such as 1,2-butanediol (Table

11

1, entry 11) and 1,2-hexanediol (Table 1, entry 12) was almost the same as that of 1,4-AHERY, and

12

simultaneous hydrodeoxygenation products such as n-butane (from 1,2-butanediol) and n-hexane

13

(from 1,2-hexanediol) were obtained in high selectivity. The reactivity of 2,3-butanediol (entry 13)

14

was about half as compared to that of 1,2-alkanediols (Table 1, entries 11, 12). From 1,3-butanediol

15

substrate, 2-butanol was mainly produced and small amount of ethanol was formed (Table 1, entry

16

14), probably through a dehydrogenation–dehydration–hydrogenation process (Scheme 2) [8, 44]

17

and retro-aldol reaction following dehydrogenation of primary OH group [8, 44–47], respectively.

18

1,4-Butanediol was almost inert (Table 1, entry 15). Erythritol was mainly converted to

19

1,2-butanediol with similar reactivity to 1,2-hexanediol (Table 1, entry 16). The order of the

20

reactivity of the substrates was cis-vicinal diols on five-membered ring > 1,2-alkanediol > cis-vicinal

21

diols on six-membered ring ≈ vicinal alkanediols which have two secondary OH groups. The higher

22

reactivity of cis-vicinal diols on five-membered ring is probably due to the smaller dihedral angle of

23

vicinal OH groups of cis-vicinal diols. The order of the reactivity of substrates except cis-vicinal

9 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

diols on five-membered ring is probably determined by the steric hindrance around the OH groups.

2

3

Characterizations and reaction mechanism

4

Because of the similarity of reactivity order of various substrates such as cis/trans-diols between

5

ReOx–Pd/CeO2 and homogeneous Re-catalyzed DODH systems [16, 17], the reaction pathway over

6

ReOx–Pd/CeO2 is probably DODH + hydrogenation. As shown in the Introduction section, the

7

reaction mechanism of DODH with homogeneous Re catalysts is believed to be a two-electron redox

8

cycle of high valent Re and there are following three elementary reaction steps (1), (2), and (3): (1)

9

reduction of Re species, (2) diol coordination with Re species to form a diolate (steps (1) and (2)

10

may be interchangeable), and (3) elimination of produced alkene and regeneration of the catalyst [16,

11

17]. In order to investigate the reaction mechanism in detail, characterization by various techniques

12

was conducted for the ReOx–Pd/CeO2 catalysts before and after each elementary reaction step.

13

14

XRD and TEM

15

In the previous report, XRD measurements of ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re = 0.25) after

16

calcination and catalytic use were conducted, and the results indicate that both Re and Pd species on

17

ReOx–Pd/CeO2 were highly dispersed and/or amorphous [41]. We further measured the XRD pattern

18

of ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re = 0.25) after reduction at 423 K and those of ReOx–Pd/CeO2

19

(Re = 10 wt%, Pd/Re = 0.25) after calcination, reduction at 423 K and catalytic use (Figure S5). All

20

samples showed similar patterns; only peaks of CeO2 support were observed, and these were no

21

peaks for ReOx or Pd species. The TEM image of ReOx–Pd/CeO2 (Re = 2 wt%) after calcination is

22

shown in Figure S6. No particle due to ReOx or Pd species was observed. These XRD and TEM

23

results indicate that both ReOx and Pd species are highly dispersed on both ReOx–Pd/CeO2 catalysts.

10 Environment ACS Paragon Plus

Page 10 of 56

Page 11 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

1

2

Effect of gas-phase reduction pre-treatment

3

The temperature programmed reduction (TPR) profiles of ReOx–Pd/CeO2 (Re = 2 or 10 wt%,

4

Pd/Re = 0.25) and ReOx/CeO2 (Re = 2 wt%) after the calcination at 773 K are shown in Figure 5. All

5

patterns have a broad signal above 900 K with almost constant area, suggesting that this broad signal

6

above 900 K is not due to reduction of Re species, and is probably due to reduction of CeO2 support.

7

In the case of ReOx/CeO2, a sharp peak at 710 K was observed with a total H2 consumption in the

8

range of 440–770 K of 4.8 equivalents of Re in addition to a broad signal above 900 K (Figure 5c).

9

The H2 consumption value in the range of 440–770 K was significantly larger than that for the

10

reduction of Re (H2/Re ≤ 3.5), indicating that some amount of Ce was also reduced in this

11

temperature range. In the case of ReOx–Pd/CeO2 (Re = 2 wt%, Figure 5a), a signal with two sharp

12

peaks at 490 and 530 K was observed in addition to the broad signal above 900 K. The presence of

13

Pd lowered the reduction temperature of Re. The range of these peaks did not overlap with

14

ReOx/CeO2 at all, indicating that all the Re species was affected by Pd in the reducibility. The H2

15

consumption amount of low temperature signal (H2/Re = 5.6; 450–560 K) was also larger than that

16

for the reduction of Re. As reported previously, Re species on ReOx–Pd/CeO2 (Re = 2 wt%) was

17

partially reduced at 773 K to the valence of +4 [41]. For Re = 10 wt% catalyst, the low-temperature

18

signal had one peak at 550 K and the ratio of H2 consumption amount to Re was 2.9 (Figure 5b). The

19

reduction degree of Ce can vary in different Re amount and temperature, and thus the average

20

valence of Re could not be determined from TPR measurement alone. The determination of Re

21

valence by XANES and XPS will be discussed in later sections.

22

Next, the effect of reduction pre-treatment was investigated and the effect was compared with TPR

23

results. ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re = 0.25) catalyst was reduced in the gas phase at various

11 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

temperatures and then it was applied to the activity test at 413 K (Figure 6). We have reported that

2

the catalyst after reduction at 773 K shows similar conversion and selectivity to those of fresh

3

catalyst; however the reduced catalyst was exposed to air after reduction in that case [41]. In this

4

study, the reduced catalysts were loaded to the reactor without exposure to air. The catalytic activity

5

was dramatically decreased by reducing at higher temperature. In the cases of reduction at 673 K and

6

773 K, which temperatures were higher than the TPR peaks (Figure 5a), the reaction hardly

7

proceeded. Even in the case of reduction at 423 K, which temperature was in the left shoulder of the

8

first TPR peak (Figure 5a), the activity was significantly lower than that of non-reduced catalyst.

9

Considering that reduction of Re is involved in the catalytic cycle of DODH, this decrease of

10

catalytic activity is probably caused by reduction of CeIV to CeIII. Exposure to air can rapidly oxidize

11

the CeIII to CeIV, and the catalytic activity was restored to similar level of fresh catalyst as described

12

in the previous report [41].

13

14

Stoichiometric reactions using gas-phase reduced catalysts

15

In order to understand the mechanism of this process, stoichiometric reactions of 1,2-hexanediol

16

using gas-phase pre-reduced catalysts were performed in a test tube under an Ar atmosphere (Table

17

2). In these tests, to suppress reduction of Re species by hydrogen species via dehydrogenation of

18

substrate, the reaction temperature was set at 353 K. ReOx–Pd/CeO2 (Re = 0.5, 2, 4, and 10 wt%,

19

Pd/Re = 0.25) catalysts were reduced in gas-phase at 423 K in order to sufficiently reduce Re species

20

with minimum reduction of CeIV. In fact, the catalytic reaction with the catalyst without

21

pre-reduction hardly proceeded at too low temperature (Table S2) such as 393 K, probably because

22

reduction of Re did not proceed as shown in the TPR result (Figure 6). 1-Hexene, the DODH product,

23

was formed by the reaction with reduced ReOx–Pd/CeO2. In the case of Re = 2 wt% catalyst,

12 Environment ACS Paragon Plus

Page 12 of 56

Page 13 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

1

turnover number (TON) per Re atom achieved about 0.4 at 2 h (Table 2, entry 2) and it was almost

2

constant even at 4 h (Table 2, entry 3). It is suggested that about 40% of total Re is catalytically

3

active. Thus, the active Re species can be characterized using various spectroscopic methods using

4

this 423 K-reduced catalyst. On the other hand, TON per Re atom was decreased with increasing Re

5

loading amount. In the case of Re = 10 wt%, TON per Re atom achieved only 0.02 at 2 h (Table 2,

6

entry 5), and therefore the amount of the active Re species was very small in Re = 10 wt% catalyst.

7

This dependence of these stoichiometric reactions on the Re loading amount is consistent with the

8

results of the catalytic use under the standard conditions (Figure 2). The amount ratio of catalytically

9

active Re to total Re will be discussed below based on further characterizations and the structure

10

model.

11

12

XPS and Re L3-edge XANES analyses

13

The Re 4f XPS data of ReOx–Pd/CeO2 (before and after stoichiometric reaction, and after catalytic

14

use) are shown in Figure 7 and Table 3. The peak positions of the functions with lower binding

15

energy were within the range of ReIV (42.3 and 43.6 eV [48–54]), and those with higher binding

16

energy were within the range of ReVI (44.3 and 44.9 eV [50–52, 55, 56]). Therefore the samples

17

before and after stoichiometric reaction and after catalytic use under the standard conditions

18

contained both ReIV and ReVI species. The Pd 3d XPS data of the same samples and summary of

19

XPS data are shown in Figure S7 and Table 3, and these results indicated that all samples contained

20

only Pd0. Both Re and Pd species were highly dispersed because Pd/Re molar ratio calculated from

21

XPS was nearly equal to the loaded Pd/Re ratio. The absence of XRD peaks for Pd and Re species

22

also supports the high dispersion of both species. Thus, all the Re and Pd species in the ReOx–

23

Pd/CeO2 catalyst can be detected by XPS measurement.

13 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

Average valences of Re calculated from XPS on ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re = 0.25)

2

before and after stoichiometric reaction were +4.2 and +5.3, respectively (Table 3). The ratio of ReIV

3

to (ReIV + ReVI) was determined to be 91% and 37% for the sample of before and after

4

stoichiometric reaction, respectively (Table 3). The decreasing amount of ReIV and increasing

5

amount of ReVI species through the stoichiometric reaction fairly agreed with that about half of the

6

Re species reacted with 1,2-diol (TON = 0.42, Table 2, entry 3). The reaction mechanism of this

7

reaction is suggested to be the tetra/hexavalent redox cycle of Re center. The XPS result after

8

catalytic use was similar to that after stoichiometric reaction. The stoichiometric reaction can

9

proceed at lower temperature (353 K) than the reduction of Re (≥393 K), and therefore the active

10

reduced Re species after the catalytic use can react with the remaining substrate during cooling. In

11

addition, in order to investigate the effect of CeO2 support, Re 4f XPS measurement of ReOx–

12

Pd/SiO2 after catalytic use under the standard conditions was conducted, and comparison was made

13

between ceria support and silica support (Figure S8). Some Re species on ReOx–Pd/SiO2 after

14

catalytic use was Re0, and the valence of Re of ReOx–Pd/SiO2 from XPS was lower than that of

15

ReOx–Pd/CeO2. The results indicated that CeO2 support suppressed the reduction of Re species.

16

Figure 8 shows the Re L3-XANES spectra of ReOx–Pd/CeO2 and reference compounds. The first

17

peak in the L3-edge XANES is called as a white line, and the white line area in the L3-edge XANES

18

is known to be an informative indication of the electronic state. The larger white line area is due to

19

greater electron vacancy in d-orbital. As reported previously, a relative electron deficiency and

20

valence can be determined on the basis of the white line area [57–60]. Regarding the reference

21

compounds, valence of Re species is almost proportional to the white line area. Therefore, the

22

average value of Re species can be estimated by examining the white line area in the XANES spectra

23

[61, 62]. The valence of Re species on ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re = 0.25) can be

14 Environment ACS Paragon Plus

Page 14 of 56

Page 15 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

1

determined by the relation between the valence and the white line area of the reference compounds

2

(Figure S9). The average valence of the Re species on ReOx–Pd/CeO2 before and after stoichiometric

3

reaction was determined to be +4.2 and +5.3, respectively (Table 3), which was almost the same as

4

those calculated from XPS. The consistent results support tetra/hexavalent redox cycle of Re center

5

in the catalysis. In the case of ReOx–Pd/CeO2 after catalytic use, the average valence was determined

6

to be +5.0, which was also consistent with XPS.

7

8

Re L3-edge EXAFS analyses

9

In order to characterize the structure of Re species precisely and the interaction of Re species with

10

Pd metal clusters, Re L3-edge and Pd K-edge EXAFS analyses were carried out. Figure 9 shows the

11

Re L3-edge EXAFS spectra of ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re = 0.25) after calcination, before

12

and after stoichiometric reaction, and after catalytic use under the standard conditions. Table 4

13

summarizes the curve fitting results of Re L3-edge EXAFS. In the case of the calcined ReOx–

14

Pd/CeO2, the Fourier transform of k3-weighted spectrum was similar to that of NH4ReO4 (Figures 9d

15

and e), and the curve fitting results indicate the presence of Re–O bond with coordination number

16

(CN) of 4.0±0.3 (Table 4). It is suggested that the Re species were present as ReO4–. The spectrum of

17

ReOx–Pd/CeO2 was drastically changed by the reduction and the reaction. The curve fitting analysis

18

indicates the presence of Re=O, Re–O and Re–Pd bonds. The absence of Re–Re bond indicates that

19

almost all the Re species were in oxidized state, and this is also supported by XPS (Figure 7).

20

The length of Re–O bond (0.193–0.194 nm) on ReOx–Pd/CeO2 before and after stoichiometric

21

reaction and after the catalytic use was clearly longer than that (0.173 nm) on NH4ReO4 and ReOx–

22

Pd/CeO2 after the calcination, and the longer Re–O bond can be assigned to a single bond. Although

23

the coordination number was low (0.4), the presence of Re–Pd bond demonstrates the direct

15 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

interaction between Re species and Pd metal clusters. The bond length of the Re–Pd bond (0.266–

2

0.269 nm) on the catalyst after the reduction and the reaction corresponded with the direct metal

3

bond (cf. Re–Re in Re powder: 0.275 nm, Pd–Pd in Pd foil: 0.275 nm). Assuming that Re atoms

4

were adsorbed at three-fold hollow site of Pd (1 1 1) surface [42, 63], the coordination number (CN)

5

of Re–Pd should be 3 per one Re atom on Pd (1 1 1). In this case, the obtained CN (0.4) suggests that

6

about 10% of Re species interacted with Pd metal surface and about 90% of Re did not interact with

7

Pd metal. If the Pd-bound Re atom is incorporated in the Pd particles rather than located on the

8

surface of Pd, the CN of Re–Pd per one Re should be higher than 3, and the amount of Pd-bound Re

9

species will be further smaller. The small amount of Pd-bound Re species can be connected to small

10

dependency of activity on the Pd amount (Figure 1), indicating that the main active site is not the

11

small amount of Pd-bound Re species. Rather, the small decrease in the activity with increasing the

12

Pd amount above Pd/Re = 0.25 suggests that the Pd-bound Re species is inactive. The CN of the Re–

13

O single bond was increased by the stoichiometric reaction and the catalytic use under the standard

14

conditions (3.0→3.5 and 4.2, respectively), while the CN of the Re=O double bond and Re–Pd bond

15

are maintained during the reactions.

16

The results of Pd K-edge EXAFS analysis of ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re = 0.25) after

17

the catalytic use are shown in Figure S10 and Table S3, and the Pd–Pd bond (CN = 3.1) and Pd–Re

18

bond (CN = 1.6) were detected. It was also verified from Pd K-edge EXAFS that Pd species was

19

present as highly dispersed clusters from the small CN, and some ReOx species were attached to Pd

20

clusters.

21

In our previous reports, the valence of Re can be estimated approximately from the CN of Re=O

22

and Re−O (valence of Re = 2 × CNRe=O + 1 × CNRe−O) by assuming that all bonds of Re and O

23

contribute to valence of Re [42, 43, 64]. However, the calculated values of the Re valence from

16 Environment ACS Paragon Plus

Page 16 of 56

Page 17 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

1

EXAFS analyses were clearly higher (mono–divalent) than those from Re L3-edge XANES and Re

2

4f XPS (Table 3 and Figure S9). In fact, in the case of ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re = 0.25)

3

before stoichiometric reaction, the valence of Re estimated approximately from the CN of Re=O and

4

Re−O was +6.0 (about divalent higher than that from XANES and XPS (Table 3)). This difference

5

suggests that some bonds existed which did not contribute to the valence of Re, such as Re←O

6

coordination bonds.

7

8

Density functional theory (DFT) calculations and proposed structure model

9

As discussed in the previous sections, the catalyst contains two types of Re sites in addition to

10

small amount of Pd-bound Re species: catalytically active ones and inactive ones. Further analysis of

11

Re L3-edge EXAFS requires additional information for the structure of each Re species. We used

12

DFT calculations for the determination of Re species on CeO2 surface. Most of CeO2 surface is

13

assumed to be occupied with the (1 1 1) surface [65–68]. The structures of monomeric Re species

14

were calculated with or without coordinated 1,4-AHERY molecule using Ce70O140 model cluster. The

15

structure of the model cluster is shown in Figure S11. The ReIVO2 species on CeO2 (1 1 1) surface

16

have two Re=O bonds and two Re←O coordination bonds (Figure 10a). The O→Re←O plane leans

17

out of the perpendicular, and the Re atom was almost located above an O atom in the 2nd layer. The

18

ReVIO3 species on CeO2 surface have three Re=O bonds and one Re←O bond (Figure 10b). The

19

Re←O bond also slightly leans out of the perpendicular, toward an O atom in the 2nd layer. The ReIV

20

and ReVI species after coordination of 1,4-AHERY have two Re–OC single bonds, less number of

21

Re=O bonds by one than that before the coordination, and the same number of Re←O bonds to that

22

before the coordination (Figures 10c and 10d). The structure of transition state (TS) of DODH is

23

shown in Figure 10e, and the energy diagram is shown in Figure S12. The two C–ORe bonds are

17 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

dissociated simultaneously, and the Re–OC single bonds are converted to double bonds, similarly to

2

the reported transition state for DODH [16, 17]. An important point is that only one Re←O bond is

3

present in this calculated TS structure, and the coordination mode is more similar to ReVIO3/CeO2

4

than to ReIVO2/CeO2. The difference in the Re-support bonds between initial and transition states

5

suggests that polymeric Re species with many rigid Re–O bonds have lower activity.

6

Not all Re species can be monomeric ones because of the limited surface area of CeO2. Here we

7

pay attention to the surface density of Re species. Figure 11 shows CeO2 (1 1 1) surface, and the

8

circles of dotted line in Figure 11 represent sites that can be occupied with ReIV species (Re site), i.e.

9

the sites above the O atoms in 2nd layer. Considering the surface area of CeO2 (87 m2 g-1), the total

10

number of possible Re sites (circles of dotted line) corresponded to about 20 wt% loading of Re. In

11

the hexagon in Figure 11, three Re atoms can be present at a maximum. To produce monomeric Re

12

species, only one Re atom can be present in the hexagon. If adjacent Re sites are occupied, i.e., more

13

than one Re atoms are present in the hexagon in Figure 11, Re–O–Re bonds are formed. Now, we

14

consider two structure models of ReOx–Pd/CeO2, and discuss the correlation between the amount of

15

isolated Re species and the results of activity tests in both reactions under the standard conditions

16

and stoichiometric reactions (Figure 12).

17

In the model (I), it is assumed that the Re species are located monomerically as much as possible,

18

i.e., one Re species is located at the center of the hexagon in Figure 11 and no other Re species are

19

present in the surrounding six sites (repulsively-dispersed model). In this case, amount of isolated Re

20

species should increase linearly up to about 7 wt% of loading amount, where all the surface is filled

21

with hexagon with one Re atom (Figure 12, solid line). In this repulsively-dispersed model, all the

22

Re sites are equivalent and active when Re loading amount is lower than 7 wt%. However, the

23

activity tests revealed that the number of active sites did not linearly increase and rather decreased at

18 Environment ACS Paragon Plus

Page 18 of 56

Page 19 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

ACS Catalysis

Re > 2 wt%.

2

In the model (II), it is assumed that Re species are randomly located among all the possible sites in

3

Figure 12 (randomly-located model). Here, the amount of isolated Re species is calculated based on

4

the following equation (1). The part in parentheses in the equation (1) represents the probability of

5

absence of Re species in one Re site, and the six power of the part in parentheses represents the

6

number of Re sites around one certain Re site.

7

  Re amount [mol] Amount of isolated Re species [mol] = 1 −  × Re amount [mol]  Total amount of Re sites [mol]  6

8

(1)

9

10

11

In this case, amount of isolated Re species is maximized at about 2 wt% of Re, and decrease

12

monotonically after that (Figure 12, broken curve). The curve well agrees with the curves of the

13

catalytic activities vs. Re loading amount. This agreement suggests that the model (II) is plausible

14

and only isolated Re species is catalytically active. When loading amount of Re is 2 wt%, about 50%

15

of total Re amount are isolated and will have catalytic activity. Figure 13 shows the detailed structure

16

model of Re species in ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re = 0.25) after reduction (before

17

stoichiometric reaction) and after stoichiometric reaction, based on randomly-located model and

18

DFT calculations. The amount of Pd-bound Re species (~10%) was neglected in this model. The

19

DFT calculation (Figure 10) showed that the isolated ReIV species have two Re←O and two Re=O

20

bonds in the absence of substrate, giving CN of 2 and 2 for Re–O and Re=O shells, respectively, in

21

the EXAFS analysis. After reaction, the presence of substrate produces substrate-coordinated ReVI

22

species which have one Re←O, two Re–O and two Re=O bonds, giving CN of 3 and 2 for Re–O and

19 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

Re=O shells, respectively. According to XPS results, the inactive Re species have ReIV valent and the

2

valence is not changed during the DODH. As discussed above, the inactive Re species are probably

3

polymeric ones. Considering the coordination mode of substrate-bound ReIV species, we assume the

4

coordination mode of polymeric ReIV species as two Re←O, two Re–O, and one Re=O bonds, which

5

give CN of 4 and 1 for Re–O and Re=O shells, respectively. The 1 : 1 molar ratio of isolated species

6

and polymeric species gives the overall CN of 3 and 1.5 for Re–O and Re=O shells, respectively, for

7

reduced ReOx–Pd/CeO2 (Re = 2 wt%). For the catalyst after stoichiometric reaction, isolated species

8

and polymeric species gives the overall CN of 3.5 and 1.5 for Re–O and Re=O shells, respectively.

9

These coordination numbers based on this randomly-located model (II) and DFT calculations well

10

agreed with the results of Re L3-edge EXAFS analyses (Table 4).

11

In order to detect the formation of polymeric Re species in the catalysts with high Re loading, we

12

carried out Raman measurements. Unfortunately, the Raman spectra could not be measured

13

successfully for reduced and used catalysts because of the gray color of the catalysts. The Raman

14

spectra of calcined catalysts are shown in Figure S13. The Re=O bond (960–1000 cm-1 [69–71]) was

15

observed for all samples containing Re. For ReOx–Pd/CeO2 (Re = 4, 10 wt%) samples, broad band

16

around 900 cm-1 was observed with much larger intensity than the 920 cm-1 band of Pd/CeO2. The

17

intense band can be assigned to Re–O–Re [69–71]. The presence of Re–O–Re in ReOx–Pd/CeO2 (Re

18

= 4 wt%) sample also supports the randomly-located model.

19

20

Reaction mechanism

21

The reaction mechanism proposed by this study is summarized in Scheme 3. As previously

22

indicated, the reaction mechanism of simultaneous hydrodeoxygenation can be regarded as DODH +

23

hydrogenation. This reaction mechanism has following elementary reaction steps (1), (2), (3), and

20 Environment ACS Paragon Plus

Page 20 of 56

Page 21 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

1

(4): (1) the isolated ReVI species is reduced by H2 as a reductant with promotion by Pd, and becomes

2

the active ReIV species; (2) the substrate with vicinal OH groups is coordinated to the reduced ReIV

3

species with the two OH groups as diolate; (3) alkene is eliminated with the regeneration of ReVI

4

center (IV→VI); (4) Pd-catalyzed hydrogenation of the produced alkene gives the simultaneous

5

hydrodeoxygenation product. Additionally, the substrate is coordinated to not only the reduced ReIV

6

species but also the oxidized ReVI species even at room temperature. As discussed in the previous

7

paper, the role of CeO2 support is probably to prevent the overreduction of Re species by the

8

interaction between Re species and the CeO2 surface [41]. In addition to this role, another important

9

role of CeO2 support is to stabilize the monomeric Re species. According to the literature, polymeric

10

Re species are easily formed over supported noble metal–Re catalysts by the reduction [42, 72].

11

Even in the homogeneous MeReO3 catalyst for DODH, formation of oligomeric ReV-glycolate

12

species has been reported by the reduction [26]. The reasons for the much higher activity of ReOx–

13

Pd/CeO2 than other DODH catalyst can include the high rate of reduction of Re species assisted by

14

Pd and large number of isolated Re sites, which are probably the active sites.

15

16

Conclusions

17

ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re = 0.25) is still the best catalyst for simultaneous

18

hydrodeoxygenation. The simultaneous hydrodeoxygenation over ReOx–Pd/CeO2 catalyst is favored

19

for cis- or syn-elimination, and 1,2-elimination rather than 2,3-elimination. This regio- and

20

stereo-selectivity are similar to DODH systems such as CH3ReO3-catalyzed one. Reduction of CeO2

21

support decreased the activity. The Re species on ReOx–Pd/CeO2 (Re = 2 wt%, Pd/Re=0.25) catalyst

22

after reduction was mainly ReIV, and that after stoichiometric reaction of 1,2-hexanediol to 1-hexene

23

were ReIV and ReVI. The amount of converted ReIV to ReVI corresponded with that of produced

21 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

1-hexene. There are different types of ReIV; active ones and inactive ones. The structure model in

2

which Re species are randomly located on CeO2 (1 1 1) surface is proposed, and only isolated Re

3

species are suggested to be catalytically active. The roles of Pd and CeO2 support are to promote the

4

reduction to active ReIV centers and to stabilize the isolated ReIV species under pressurized H2,

5

respectively. The reaction is proposed to proceed by the reduction of the coordinated substrate by the

6

ReIV center followed by hydrogenation.

7

8

Experimental section

9

General

10

11

Suppliers of the reagents are listed in Supporting Information (Table S4).

Catalyst preparation

12

The M1Ox/CeO2 (M1 = V, Cr, Mn, Nb, Mo, W, Re) catalysts were prepared by impregnating M1Ox

13

with an aqueous solution of NH4VO3, Cr(NO3)3·9H2O, Mn(NO3)2·6H2O, (NH4)6Mo7O24·4H2O,

14

(NH4)NbO(C2O4)2·xH2O, (NH4)6H2W12O40·nH2O or NH4ReO4. CeO2 (Daiichi Kigenso Co., Ltd., HS,

15

BET surface area: 84 m2 g-1, after calcination at 873 K for 3 h) was used. The M1Ox–Pd/CeO2

16

catalysts were prepared by impregnating M1Ox/CeO2 after drying at 383 K for 12 h with aqueous

17

solution of Pd(NO3)2. The ReOx–Pd/SiO2 catalyst was prepared in a similar way to the case of ReOx–

18

Pd/CeO2 except using SiO2 (Fuji Silysia Chemical Ltd., G-6, BET surface area 535 m2/g) as a

19

support.

20

The ReOx–M2/CeO2 (M2 =Co, Ni, Cu, Ru, Rh, Ir, Pt) catalysts were prepared by impregnating

21

ReOx/CeO2 after drying at 383 K for 12 h with aqueous solution of Co(NO3)2·6H2O, Ni(NO3)2·6H2O,

22

Cu(NO3)2·3H2O, Ru(NO)(NO3)3-x(OH)x, RhCl3·3H2O, H2IrCl6 or H2PtCl6·6H2O. The loading

23

amount of M1 was 2 wt%, and M2/M1 molar ratio was 0.25.

22 Environment ACS Paragon Plus

Page 22 of 56

Page 23 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

These were calcined in air at 773 K for 3 h. The standard loading amount of Re and Pd was 2 and

1

2

0.3 wt%, respectively (Pd/Re molar ratio was 0.25). All catalysts were used in powdery form.

3

4

Activity test

5

Activity tests were performed in a 190 mL stainless steel autoclave. 1,4-Anhydroerythritol

6

(1,4-AHERY) and 1,4-dioxane was placed into the autoclaves with a spinner and appropriate amount

7

of catalyst. After sealing the reactors, their air content was purged by flushing thrice with hydrogen.

8

Autoclaves were then heated to the reaction temperature. The temperature was monitored using a

9

thermocouple inserted in the autoclave. During the experiment, the stirring rate was fixed at 250 rpm

10

(magnetic stirring). After an appropriate reaction time, the reactors were cooled down. The gases

11

were collected in a gas bag. The autoclave contents were transformed to vials, and the catalysts were

12

separated by centrifugation and filtration. The standard conditions for the reaction of

13

1,4-anhydroerythritol were as follows: 1 g 1,4-anhydroerythritol, 4 g 1,4-dioxane, 413 K reaction

14

temperature, 8.0 MPa initial hydrogen pressure (at reaction temperature), 4 h reaction time, and 150

15

mg catalyst. The parameters were changed appropriately in order to investigate the effect of reaction

16

conditions. Details of the reaction conditions are described in each result. The hydrodeoxygenation

17

of

18

cis-1,2-cyclohexanediol,

19

tetrahydrofuran, 1,2-butanediol, 1,4-butanediol, 1,3-butanediol, 1,2-hexanediol and erythritol were

20

tested in a similar way to the case of 1,4-AHERY. In addition, the reaction tests using

21

1,2-dimethoxyethane, dodecane, 1-pentanol, 3-pentanol, and water as a solvent were also carried out

22

in a similar way to the case of 1,4-dioxane. Appropriate reaction conditions of these substrates are

23

described in each result.

1,4-anhydrothreitol

(1,4-AHTHR),

cis-1,2-cyclopentanediol,

trans-1,2-cyclohexanediol,

2,3-butanediol,

23 Environment ACS Paragon Plus

trans-1,2-cyclopentanediol, 3-hydroxytetrahydrofuran,

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

The experimental method of the activity tests of the reduced catalysts is described below. The

2

calcined catalysts were reduced in H2 flow (30 mL/min) at appropriate temperature (423–773 K) for

3

1 h (at a heating rate 10 K/min) by using a glass tube. The reduced catalysts were introduced to an

4

autoclave under high grade Ar atmosphere in order to avoid exposing the catalyst to air. For

5

comparison, we also evaluated the catalyst passivated after the reduction in H2 flow [31]. The

6

catalyst was reduced at 773 K for 1 h in H2 flow and then passivated 2% O2 diluted with N2 (10

7

ml/min) for 15 min before the catalyst transfer. The reaction conditions for the reaction using

8

reduced catalysts were as follows: 1 g 1,4-anhydroerythritol, 4 g 1,4-dioxane, 413 K reaction

9

temperature, 8.0 MPa initial hydrogen pressure (at reaction temperature), 4 h reaction time, and 150

10

mg catalyst.

11

The stoichiometric reactions were performed in a 20 mL glass test tube, and were carried out in

12

the following procedures (1), (2), (3) and (4). (1) 1,2-Hexanediol and 1,4-dioxane was placed in the

13

test tube. The mixture was frozen using acetone sherbet and degasified (freeze-pump-thaw cycles) to

14

remove dissolved oxygen. (2) The catalysts were externally reduced with H2 (30 cc/min) at 423 K for

15

1 h (at a heating rate 10 K/min) by using a glass tube. (3) The reduced catalyst and a spinner were

16

added to the test tube containing the degasified solution in a gas bag filled with high grade Ar, and

17

the test tubes were sealed using a rubber stopper in the gas bag. (4) The test tubes were then heated to

18

the reaction temperature (353 K). The temperature was monitored using a thermocouple. During the

19

experiment, the stirring rate was fixed at 250 rpm (magnetic stirring).

20

21

Analysis of products

22

The products in both gas and liquid phases were analyzed using a gas chromatograph (Shimadzu

23

GC-2025) equipped with flame ionization detector (FID). A TC-WAX capillary column (diameter

24 Environment ACS Paragon Plus

Page 24 of 56

Page 25 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

1

0.25 mmφ, 30 m) was used for the separation. Products were also identified using GC-MS (QP5050,

2

Shimadzu). In the case of the reaction of erythritol, in addition to the analysis using FID-GC, HPLC

3

(Shimadzu LC-10A) equipped with a refractive index detector was used for quantitative analysis of

4

erythritol. An Aminex HPX-87C column (diameter 7.8 mm, 300 mm) were used HPLC. The

5

conversion and the selectivity were calculated on the molar-basis. The mass balance was also

6

confirmed in each result and the difference in mass balance was always in the range of the

7

experimental error. The agreement in terms of the mass balance indicated that polymeric by-products

8

were not formed (±10%).

9

10

Catalyst characterization

11

The surface areas of the catalysts were measured using BET method (N2 adsorption) with an

12

apparatus (Micromeritics Gemini). X-ray diffraction (XRD) patterns were recorded by a

13

diffractometer (Rigaku MiniFlex600). Cu Kα radiation was used as an X-ray source. The samples

14

after the reaction or reduction were transferred to an atmosphere separator (Air sensitive sample

15

holder; Rigaku Corporation) using a gas bag filled with high grade Ar to avoid exposure to air.

16

Transmission electron microscope (TEM) images were measured by Hitachi HF-2000EDX. The

17

samples were dispersed in ethanol with supersonic waves and placed on Mo or Cu grids under air

18

atmosphere. Temperature-programmed reduction (H2-TPR) was carried out in a fixed-bed reactor

19

equipped with a thermal conductivity detector (TCD) using 5% H2 diluted with Ar (30 ml/min). The

20

amount of catalyst was 50 mg, and temperature was increased from room temperature to 1273 K at a

21

heating rate of 10 K/min. Raman spectra were obtained with JASCO NRS-5100 spectrometer. The

22

powder sample was sandwiched between a slide glass and a cover glass under air atmosphere. The

23

excitation was conducted using a 532 nm laser and ×20 objective lens. The spectra were normalized

25 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

with the CeO2-derived 464 cm-1 peak.

2

The extended X-ray absorption fine structure (EXAFS and XANES) spectra were measured at the

3

BL01B1 station at SPring-8 with the approval of the Japan Synchrotron Radiation Research Institute

4

(JASRI; Proposal No. 2014A1119, No. 2014B1248). The storage ring was operated at 8 GeV, and a

5

Si (1 1 1) single crystal were used to obtain a monochromatic X-ray beam. The detailed conditions

6

for measurement of (i) reference compounds (Re powder, ReO2, ReO3, NH4ReO4, Pd foil, and PdO),

7

(ii) catalyst samples for Re L3-edge measurements, and (iii) catalyst samples for Pd K-edge

8

measurements were as follows (i), (ii), and (iii). (i) Ion chambers for I0 were filled with 85% N2 +

9

15% Ar and 100% Ar, respectively, for Re L3-edge and Pd K-edge measurement. Ion chambers for I

10

were filled with 50% N2 + 50% Ar and 70% Ar + 30% Kr, respectively, for Re L3-edge and Pd

11

K-edge measurement. The EXAFS and XANES data were collected in a transmission mode. The

12

edge jumps were in the range of 0.6–1.5. (ii) We prepared the sample after calcination and after the

13

reduction by pressing catalyst powder to disk. In the case of sample after the reduction, the sample

14

disk was transferred to the in-situ cell and the sample was heated to 423 K in H2 flow (30 mL/min)

15

under atmospheric pressure. The samples after the reaction were prepared by transferring the used

16

catalyst powder to a measurement cell after reaction and cooling in a gas bag filled with high grade

17

Ar to avoid exposure to air. The EXAFS and XANES data were collected in a fluorescence mode

18

using a 19-element Ge solid-state detector (19-SSD), and in the case of sample after calcination a

19

Lytle detector was used. Two ion chambers for I0 and I were filled with 85% N2 + 15% Ar and 100%

20

Kr, respectively, for Re L3-edge measurement. The edge jumps were in the range of 0.0017–0.013,

21

and in the case of sample after calcination the edge jump was about 0.07. (iii) The same sample for

22

Re L3-edge measurement was used. Two ion chambers for I0 and I were filled with 100% Ar and

23

100% Kr, respectively for Pd K-edge measurement. The EXAFS and XANES data were collected in

26 Environment ACS Paragon Plus

Page 26 of 56

Page 27 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

ACS Catalysis

a fluorescence mode using a Lytle detector. The edge jump was about 0.3.

2

For EXAFS analysis, the oscillation was first extracted from the EXAFS data using a spline

3

smoothing method [73]. Fourier transformation of the k3-weighted EXAFS oscillation from the k

4

space to the r space was performed to obtain a radial distribution function. The inversely Fourier

5

filtered data were analyzed using a usual curve fitting method [74, 75]. For curve fitting analysis, the

6

empirical phase shift and amplitude functions for the Re–O, Re–Re and Pd–Pd bonds were extracted

7

from data for NH4ReO4, Re metal and Pd metal, respectively. Theoretical functions for the Pd–Re

8

bond were calculated using the FEFF8.2 program [76]. Analyses of EXAFS data were performed

9

using a computer program REX2000, ver. 2.6 (Rigaku Corp.). Error bars for each parameter were

10

estimated by stepping each parameter, while optimizing the others parameter, until the residual factor

11

becomes two times as its minimum value [77].

12

In the analysis of XANES spectra, the normalized spectra were obtained by subtracting the

13

pre-edge background from the raw data with a modified Victoreen equation and normalizing them by

14

the edge height [57–60].

15

X-ray photoelectron spectra (XPS) were measured using an AXIS-ULTRA DLD (Shimadzu Co.,

16

Ltd.) using a monochromatic Al Kα X-ray radiation (hν = 1486.6 eV) operated at 20 mA and 15 kV

17

at room temperature under 10−8 Pa. The spectra in energy were referenced to the C 1s peak at 284.6

18

eV binding energy (BE). The catalysts after the reaction or reduction were transported to the

19

measurement chamber in high grade Ar atmosphere to avoid any exposure to air. Peak fitting for the

20

Re 4f and Pd 3d spectra were done using the computer program CasaXPS, ver. 2.3.15 (Casa software

21

Ltd.) employing a peaks with a 30% Lorentzian and 70% Gaussian peak share. Re 4f7/2 and Re 4f5/2

22

are separated by 2.43 eV and have relative peak area with a fixed ratio of 4:3 and the same full width

23

half maximum values (FWHM). Pd 3d5/2 and Pd 3d3/2 are separated by 5.26 eV and have relative

27 Environment ACS Paragon Plus

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

peak area with a fixed ratio of 4:3 and the same FWHM.

2

3

DFT (density functional theory) calculation

4

The calculations were conducted with Materials Studio DMol3 software [78–80]. The used functional

5

was PW91, and the basis sets were built-in double numerical functions plus d-functions for

6

non-hydrogen atoms (DND3.5; comparable to 6-31G(d)). Core electrons were represented by DSPP

7

(DFT semi-core pseudopotentials) [81] built in the software. The orbitals were unrestricted. The

8

CeO2 support was modeled with Ce70O140 cluster which is sculpted from the 5×5×5 supercell of the

9

fluorite structure (Figure S11). In this model cluster, there is no oxygen atom bonded to only one

10

atom, and there is no cerium atom bonded to less than 5 atoms, which minimizes the distortion of

11

structure from infinite CeO2 crystal. The geometry optimization was conducted without any

12

constraint. For small systems (MTO catalyst), transition states were optimized using initial Hessian

13

obtained by numerical differentiation. Frequency calculation for the optimized structure confirmed

14

the existence of one imaginary frequency mode, in which the two ReO–C bonds are being

15

simultaneously broken. For large systems (Ce70O140 cluster system), because calculation of Hessian

16

by numerical differentiation is practically difficult, the transition state was searched by fixing and

17

stepping the lengths (by 0.01 Å) of ReO…C bonds which are most changed near the transition state.

18

The maximum gradient of energy to Cartesian coordinates in the located transition state was below

19