Homogeneously catalysed conversion of aqueous formaldehyde to H2 and carbonate

Abstract

Small organic molecules provide a promising solution for the requirement to store large amounts of hydrogen in a future hydrogen-based energy system. Herein, we report that diolefin-ruthenium complexes containing the chemically and redox non-innocent ligand trop₂ dad catalyse the production of H₂ from formaldehyde and water in the presence of a base. The process involves the catalytic conversion to carbonate salt using aqueous solutions and is the fastest reported for acceptorless formalin dehydrogenation to date. A mechanism supported by density functional theory calculations postulates protonation of a ruthenium hydride to form a low-valent active species, the reversible uptake of dihydrogen by the ligand and active participation of both the ligand and the metal in substrate activation and dihydrogen bond formation.

ARTICLE

Received 27 Sep 2016 | Accepted 20 Feb 2017 | Published 28 Apr 2017

Small organic molecules provide a promising solution for the requirement to store large amounts of hydrogen in a future hydrogen-based energy system. Herein, we report that diolenruthenium complexes containing the chemically and redox non-innocent ligand trop2dad catalyse the production of H2 from formaldehyde and water in the presence of a base. The process involves the catalytic conversion to carbonate salt using aqueous solutions and is the fastest reported for acceptorless formalin dehydrogenation to date. A mechanism supported by density functional theory calculations postulates protonation of a ruthenium hydride to form a low-valent active species, the reversible uptake of dihydrogen by the ligand and active participation of both the ligand and the metal in substrate activation and dihydrogen bond formation.

DOI: 10.1038/ncomms14990 OPEN

Homogeneously catalysed conversion of aqueous formaldehyde to H2 and carbonate

M. Trincado1, Vivek Sinha2, Rafael E. Rodriguez-Lugo3, Bruno Pribanic1, Bas de Bruin2 & Hansjrg Grtzmacher1

1 Department of Chemistry and Applied Biosciences ETH Zrich, Laboratory of Inorganic Chemistry, Wolfgang Pauli Str. 10, Zrich CH-8093, Switzerland.

2 Supramolecular and Homogeneous Catalysis Group, van t Hoff Institute for Molecular Sciences (HIMS), University of Amsterdam, Science Park 904, Amsterdam 1098 XH, The Netherlands. 3 Labotatorio de Qumica Bioinorgnica, Centro de Qumica, Instituto Venezolano de Investigaciones Cientcas (IVIC), Caracas 1020-A, Venezuela. Correspondence and requests for materials should be addressed to M.T. (email: mailto:[email protected]

Web End [email protected] )or to B.d.B. (email: mailto:[email protected]

Web End [email protected] ) or to H.G. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 8:14990 | DOI: 10.1038/ncomms14990 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14990

The use of water as oxygen transfer reagent is a highly promising approach to develop oxygenation reactions of organic substrates under mild and environmentally benign

conditions1. The investigation of such methods may also give fundamental insight into the splitting of water into O2 and

H2 (refs 2,3). While steam reforming that converts hydrocarbons and water into oxygenated products and hydrogen or the water gas shift reaction over heterogeneous catalysts is well developed4, limited progress has thus far been made with homogeneous catalysts5. These may not only operate under much milder conditions but usually also give easier access to valuable mechanistic information68.

The conversion of methanol/water mixtures into CO2 and hydrogen is an important process with respect to the use of liquid organic fuels (LOFs) according to LOF-H2 $ LOF H2

(LOF-H2 fuel; LOF spent fuel). This reaction can be

promoted under mild conditions (o100 C) by well-dened group 8 or 9 metal complexes A912, B13,14 1K15 and D16 bearing a functional cooperative ligand (Fig. 1). Catalyst A (M Ru)

and analogues catalyse also the back-reaction, that is, the hydrogenation of CO2 to methanol with turnover (TON)

of 42000 (refs 17,18). [Co(triphos)(OOCR)](BF4) also catalyses the hydrogenation of carboxylic acids to alcohols and water, including formic acid (TON4200)19. The complex 1K, with a redox and chemically non-innocent diolen-diazadiene ligand, trop2dad 1,4-bis(5H-dibenzo[a,d]cyclohepten-5-yl)-1,4-diazabuta

-1,3-diene20, converts selectively MeOH/water mixtures into hydrogen and CO2. Of the few homogeneous systems known, 1K does not contain a phosphane as ligand21 and is the only example that catalyses this reaction in the absence of any additives.

Apart from methanol, formaldehyde is an interesting hydrogen storage molecule (a 1:1 formaldehyde/water mixture contains8.3 wt% of hydrogen). The dihydrogen-releasing reaction with water (H2CO H2O $ CO2 2H2) is strongly exothermic

(DHr 35.8 kJ mol 1), thus providing a strong driving

force for H2 production. This contrasts with dihydrogen release from methanol/water mixtures, which is endothermic (DHr 53.3 kJ mol 1) and thus requires harsher conditions

(Supplementary Note 1). Formaldehyde has been intensively studied due to its importance in the atmosphere, interstellar space and combustion chemistry22 and plays a key role in the metabolism of living cells23, acting as a source of reduction equivalents and C1 carbon feedstock for carbohydrates24. Few reports describe the conversion of aqueous aldehyde solutions to hydrogen and oxygenated products. Maitlis and coworkers have reported the conversion of acetaldehyde to acetic acid promoted by half-sandwich complexes of Ru25. More recently, Prechtl et al.26 described the dehydrogenation of aqueous methanol to formaldehyde under the release of H2 under mild conditions (298368 K). In this process, an oxidase enzyme is combined with the p-cymene Ru complex C to achieve the transformation of methanol to formaldehyde hydrate (enzyme catalysed), which can be further converted to CO2 at room temperature in a process catalysed by C using a hydrogen acceptor. To date, complex C is the only catalyst that promotes the dehydrogenation of aqueous formaldehyde solutions to CO2 and H2 with acceptable TON and turnover frequencies (TOFs) at 95 C (pH 5.5)27,28. With more

diluted aqueous formaldehyde solutions (1.6 M) TONs of 700 and TOFs of 3142 h 1 could be achieved even without additives29.

The reaction of formalin to CO2/H2 is also catalysed by complex D under similar reaction conditions and with acceptable TONs (178) but with very low reaction rates. The water soluble iridium(III) hydroxo complex (E) is able to produce H2 and CO2 in a 2:1 ratio from paraformaldehyde. Although this reaction is associated with very low TONs (up to 24), it proceeds at room

temperature and the rate of H2 production increases with increasing pH30. In the reactions with C and E, metal hydrides were detected that led to the proposition of a classical mechanism in which the substrate is activated and converted at the metal centre. By contrast, the bipyridonate in D is proposed to act as a cooperative ligand, fullling the role of an internal Brnsted base/acid in key steps of the catalytic reaction (see Fig. 1).

Metal complexes with cooperative ligands are the active sites in many enzymatic reactions. The mechanisms underpinning these remarkable and new transformations (reactions (a) and (b) in Fig. 1) are still under debate (Fig. 1). But there is an increasing evidence and consensus that metalligand cooperativity is involved in the key steps of these reactions3133. A simplied representation of metalligand cooperativity is shown in Fig. 1c. The cooperative site in the ligand can be adjacent to the metal centre as the amide function in complex A34 or slightly remote as suggested for complex B. Previous calculations imply that catalysis with complex 1K may be a special case where the conversion of the substrate is mediated by the ligand exclusively35,36. In every case, the substrate binds in the second coordination sphere of the catalytically active complex, becomes dehydrogenated and is released. With methanol as substrate, rst formaldehyde is formed in a dehydrogenation reaction, which is subsequently hydrated to the geminal diol. The latter is dehydrogenated to give formic acid, which decomposes to CO2 and H2 in an exergonic catalytic follow-up reaction. Formally, water serves as the oxygen donor in the overall alcohol oxygenation to CO2, with release of H2 as the desired product (Fig. 1c)37.

Herein we report a new catalytic system, which allows the formaldehydewater shift reaction under mild conditions with good-to-excellent yields and high TOFs. A mechanism based on experimental observations and density functional theory (DFT) calculations is proposed to explain the dehydrogenation of methanediol to formate and nally CO2 in a basic aqueous solution.

ResultsStoichiometric experiments. The previously reported and easily accessible complex 1K, which is active as dehydrogenation catalyst of methanol/water mixtures, forms a tight ion pair between the [Ru(H)(trop2dad)] anion and the [K(dme)2] cation. The structural data clearly show that the trop2dad ligand coordinates in its reduced enediamide form (trop-N -CH CHN -trop)

to the Ru(II) centre [trop2dad (1,4-bis(5H-dibenzo[a,d]

cyclohepten-5-yl)-1,4-diazabuta-1,3-diene]15. This complex serves as starting material for a number of new Ru(II)-trop2dad derivatives as shown in Fig. 2.

In order to obtain ion separated complexes that contain the [RuH(trop2dad)] anion in non-coordinated form, 1K was reacted with dibenzo-18-crown-6 (db18-C-6) or quaternary ammonium salts, [R4N]Br, to give the deep orange 1KC or burgundy red complexes 1Aa and 1Ab as shown in Fig. 2, where Aa and Ab indicate the quaternary ammonium cation nBu4N or Me3(dodecyl)N , respectively. These compounds show slightly but signicantly more deshielded 1H NMR signals for the hydride ligand, Ru-H (1KC: d 9.00 p.p.m.; 1Aa:

d 9.37 p.p.m.; 1Ab: d 9.65 p.p.m.) when compared to 1K

(d 10.25 p.p.m.). As previously observed, 1K reacts with

water under the release of one equivalent H2 and KOH to give the dark brown neutral complex [Ru(trop2dad)] 2 when heated at 60 C for several hours. Complex 2 could not be isolated; however, it is unequivocally characterized by NMR. The electronic structure is best described as a mixture of two resonance forms with either Ru(0) and a neutral diazadiene ligand, [trop-N CHCH N-trop], or as Ru(II) complex of the

2 NATURE COMMUNICATIONS | 8:14990 | DOI: 10.1038/ncomms14990 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14990 ARTICLE

cat A, B or D

cat C or D, 95 C

cat 1K (without base)

cat E (without base), 25 C

T < 100 C

a

MeOH + H2O + 2 MOH 3 H2 + M2(CO3) or CO2

2 H2 + M2(CO3) or CO2

H2C=O + H2O+2 MOH

PR2

M

N

b

PR2

N

N

+2

Ru

N

+2 CO

N

Ru

Ru Ru Ir Ir

CO

H

H H

K(dme)2

PR2

M = Fe, Ru

A

C D E

B 1K

CI

CI

CI

CI

O

O

N N

N

N

CO2

H HR C C

R

c

O

R

O

OH OH

H H

H

+

coopL

M

coopL M

H

H2O

H2

RCH2OH

HO

coopL

M

C

O

H

H

H2

coopL

M

trop

RCOOH

Figure 1 | Homogeneously catalysed reforming reaction of MeOH and formaldehyde. Catalysed dehydrogenation of (a) methanol/water and(b) formaldehyde/water mixtures. The structures of representative catalysts are given by general formulas AE and 1K (b); arrows indicate cooperative active sites in the ligand backbone. (c) Simplied sketch of the catalytic cycle highlighting the addition of water to the aldehyde and dehydrogenation of the acetal involving metalligand cooperation (solvent effects are neglected). A drawing of the trop unit is given at the bottom right.

dianionic form [trop-N -CH CHN -trop]. This 16-electron

complex is a strong Lewis acid and immediately reacts with 2-electron donors (PPh3) to give cleanly the purple 18 electron complex 3a, which exhibits a square pyramidal structure15. In order to test whether the new complexes would be catalytically active as dehydrogenation catalysts of aqueous formaldehyde solutions, we rst performed a series of stoichiometric model reactions. These are displayed in Fig. 3.

In an initial experiment, 1Aa was reacted with water (Fig. 3, equation (a)) and the formation of the neutral complex [Ru(trop2dad)] (2) occurred in a notably faster reaction (1 h) at 60 C giving a higher yield (470%) when compared to the ion pair 1K (25%). In tetrahydrofuran (THF), complex 2 reacts with the quaternary ammonium formate [nBu4N][HCO2] to give the hydride complex 1Aa (Fig. 3, equation (b)). This reaction is not

only an alternative synthesis for complex 1Aa but also demonstrates that 2 is a potent reagent for the conversion of formate to CO2. Because CO is a possible reaction intermediate in dehydrogenation reactions of formaldehyde, 2 was reacted with carbon monoxide under anhydrous conditions in THF. Complex 2 is converted immediately to give orange crystals of 5 and the carbonyl complex 3b (Fig. 3, equation (c)). The fact that the dinuclear complex 5 with one CO ligand (nCO 1912 cm 1) is

formed in high yield is very likely a consequence of its insolubility. The structure of 5 was determined by X-ray diffraction methods (vide infra) while soluble 3b was characterized by NMR spectroscopy. When 1Aa is reacted with an eight-fold excess of formalin (formalin is a 37% aqueous solution of H2CO, which mainly contains the hydrate of H2CO in the form of short chain oligomers, HO(CH2O)nH and methanol as stabilizer, vide

NATURE COMMUNICATIONS | 8:14990 | DOI: 10.1038/ncomms14990 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14990

PPh3

N

N

N

H2O, THF, 65 C

H2, KOH

N N N

N

THF, r.t.

Ru

Ru

H

H

Ru Ru

3a

2

1K

PPh3

K(dme)2

db18-C-6 or [R4N]+ Br

THF, r.t.

[cation]+ 1KC : cation = [K(db18-C-6)]

1Aa : cation = nBu4N

1Ab : cation = Me3(dodecyl)N

N

O

O

O

O

O

O

db18-C-6

trop

Figure 2 | Synthesis of penta-coordinated hydride Ru complexes 1 and neutral complex 3. A drawing of the trop unit and dibenzo-18-crown-6 (db18-C-6) is given at the bottom.

infra), gas evolution is observed immediately and the colour of the reaction mixture changes from dark brown to orange. Analysis of the reaction mixture after a few minutes indicated the formation of formic acid (18%, ratio relative to formaldehyde), the yellow zero-valent Ru(0) complex 4 (ref. 15), which contains a hydrogenated tropNH-CH2-CH2-NHtrop ligand (41%) and the orange dimeric complex 5H2 (10%), which has a similar structure as 5 with one CO ligand (nCO 1,918 cm 1), the difference being

the dangling trop unit containing a hydrogenated (CH2CH2)trop unit (Fig. 3, equation (d)). Furthermore, at least three different hydride complexes showing signals in the range d 5.7 to 7 p.p.m. in the 1H NMR spectrum are observed as further minor components in the reaction mixture. Finally, the 18-electron complex 3a was reacted with formalin at room temperature leading to the blue carbonyl complex 6 (lmax 377,

566 nm; nCO 1,936 cm 1), which like 5H2 contains a non-

coordinated saturated (CH2CH2)trop unit (Fig. 3, equation (d)).

When 6 is reacted with H2CO and KOH in a water/THF mixture at 60 C for 12 h, a sluggish reaction occurs. In the aqueous phase, K2CO3 and formate K(HCO2) was detected.

Spectroscopic analysis of the organic phase of the reaction mixture indicated the presence of some unreacted carbonyl complex 6 (10%) and the formation of complex 3a as main product. Additionally, Ph3PO is detected (ca. 20%), the new

complex [Ru(PPh3)(trop2dae)] (7) (dae diaminoethane) (12%

isolated yield), small amounts of the complex 1K, and two other unidentied species. The Ru(0) complex 7 contains coordinated CCtrop units and a hydrogenated NHCH2CH2NH ligand

backbone and can be prepared in good yield by reacting [Ru(trop2dad)(PPh3)] with 24 atm of H2 at 65 C for about 12 h. Mixing a solution of the dinuclear complexes 5 and 5H2 in

THF with a solution of 10 equivalents KOH in a water/THF mixture at 60 C resulted in conversion of the CO ligand to potassium formate and carbonate and concomitant formation of a complex mixture of inseparable ruthenium hydride complexes.

The reactions in equations (a)(f) in Fig. 3 indicate that various ruthenium complexes may be involved in the dehydrogenation of aqueous formulations of formaldehyde. The reaction between the hydride 1Aa and formalin gives formic acid as dehydrogenation

product (equation (d)) while 2 is involved in the decomposition of formate (equation (b)). The isolation of the hydrogenated Ru(0) complex 4 indicates that both the metal and the ligand may be involved in the dehydrogenation, while the conversion of formate to CO2 and 1Aa could be a metal-centred process.

Carbonyl complexes may also be formed. Note that in these cases hydrogenolysis of the bonds between C Ctrop and the Ru centre

could occur to give non-coordinated aliphatic (CH2CH2)trop groups. Under harsher reaction conditions, the CO groups are converted with KOH to formate and/or carbonate (water gas shift reaction) while the (CH2CH2)trop groups are at least partially dehydrogenated back to coordinating C Ctrop groups.

The reversible hydrogenation/dehydrogenation of these olenic groups has been previously observed by us with Ir-trop-type complexes38,39.

The molecular structures of 1Aa, 1Ab, 5, 5H2, 6 and 7 were determined with X-ray diffraction methods using single crystals. Structure plots of 1Ab, 5, 6 and 7 are given in Fig. 4ad, the other structures are shown in Supplementary Figs 14 and 17. Selected bond parameters are given in Table 1 with additional previously reported data for comparison15. The structures of [nR4N][RuH(trop2dad)] 1Aa and 1Ab show no close contact between the anion and the cation as seen in 1K and we assume a similar behaviour in solution (Fig. 4a). Otherwise the structure of the [RuH(trop2dad)] anion in the ammonium salts is not signicantly different from 1K and is between a trigonal bipyramid and square pyramid with the trop2dad ligand in its enediamide form coordinated to Ru(II). Both solid-state structures of the dinuclear complexes 5 and 5H2 are relatively similar (Fig. 4b). In both dimers, the ruthenium centres reside in a slightly distorted trigonal bipyramidal coordination sphere. Each of the ligands binds via the lone pairs at the nitrogen centres to one Ru centre (s,s-coordination) and one pair of p electrons of the imine function coordinates to the second metal centre as bridging ligand (s2-N0, m2-N, Z2-C N coordination). These

Z2-C-N units show signicantly longer distances (1.425(10)1.446(10) ) with respect to the only s-N coordinated imine(1.268(10)1.321(10) ) (the C N distance in the free ligand is

1.264(2) ). In contrast to the anionic hydride [RuH(trop2dad)]

4 NATURE COMMUNICATIONS | 8:14990 | DOI: 10.1038/ncomms14990 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14990 ARTICLE

[nBu4N]+

N

Ru

N

H

1Aa

[nBu4N](HCO2)

THF, r.t. CO2

b

CO

c

a

N

THF, 60 C

N

CO, 1 atm. r.t.

N Ru

N

Ru

N

Ru

N

+

H2

[nBu4N](OH)

C O

N

Ru N

[nBu4N]+

H2O

2

N

3b

~10%

5

Ru

N

80%

CO

H

N Ru N

Ru N

1Aa

d

H2CO, H2O, THF

H

H H H H

H

+

N

Ru

N

N

HCOOH 4

5H2

18%

41%

10%

e f

CO

H2CO (10 equiv) KOH (10 equiv)

H2CO, H2O, THF

H

H H H H

H

+ Ph3PO + 3a + 1K

N

60 C, 3 h

r.t.

H2O, THF, 60 C

N

N

N

N

Ru

N

Ru

PPh3

Ru

PPh3

3a

6

PPh3 7

K2CO3/K(HCO2)

H2 (24 atm), 65 C, 12 h

+ other Ru hydrides

+ other Ru complexes

Figure 3 | Stoichiometric reactivity of complexes 1Aa and 3a. (ac) Synthesis and reactivity of complex 2 towards formate and carbon monoxide. (df) Model reactions of complexes 1Aa, 3a or 6 with formalin.

in 1K, 1Aa and 1Ab, which have a short C C bond in the

NC CN ligand backbone (B1.37 avg.), the corres

ponding C31C32 and C63C64 bonds in 5 and 5H2 are signicantly longer (B1.45 avg.), indicating the neutral diazadiene N CC N form of the ligand. The RuRu bond

distances are 3.292(1) and 3.293(1) and exclude a metalmetal bond. As seen in most trop-type complexes, the coordinating C Ctrop bond (C4C5, C19C20 and C36C37) are 0.09

(avg.) longer than in the free ligand (1.336(2) )15. These data indicate that 5 and 5H2 are both best described as Ru(0)

complexes, each with strong stabilization of the low-valent metal centres by the ligand through p-back donation into the C N

and C C units. All the RuN distances in the divalent

ruthenium complexes 1 are in the range 1.963(3)1.978(5) . In comparison, an elongation is observed for the Ru1N1/2 and Ru2N3/4 bonds (0.13 in average) in the zero-valent complexes 5 and 5H2. In complex 6 (Fig. 4c), the Ru centre resides again in a distorted trigonal bipyramidal coordination sphere and the bond parameters of the NCCN unit indicate that the electronic

structure is best described as a mixture of resonance structures with

NC CN and N CC N units and Ru in the oxidation

states of II and 0, respectively. That one of the C Ctrop units

in the ligand became hydrogenated to give a CH2CH2 group in 6 (equation (e), Fig. 3) is clearly indicated by the long C19C20 bond (1.522(5) ). Figure 4d shows the structure of the reduced complex [Ru(PPh3)(trop2dae)] 7 with a distorted trigonal bipyramidal structure. The PPh3, one C Ctrop unit, and N2

are located in the equatorial plane while N1 and the remaining C Ctrop unit occupy the axial positions. The long C31C32

bond (1.502(3) ) shows that the C C bond in the trop2dad

ligand became hydrogenated. The Ru1N1/2 bond distances(2.189(2) avg.) are the longest among the presented ruthenium complexes and are in the same range as those observed in the previously reported ruthenium(0)-dae complex 4. The Ru centre in 7 has a formal oxidation state of zero. In the 13C NMR spectra, large coordination shifts D(d)13C 470 p.p.m. indicate strongly bound olens to low-valent Ru centres (see Supplementary Table 1). The CO ligands in the carbonyl complexes 5, 5H2 and 6

NATURE COMMUNICATIONS | 8:14990 | DOI: 10.1038/ncomms14990 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14990

a

b

C31 C32

N1 N2

Ru1

C36 ct3

C37

Ru2 N3

C31 C32

C51

C65

O1

C52

ct4 C63

C64

C4

N4

N1

ct5

H1

C5

N2

C5

C19

C20

Ru1

ct1

C4

C20 ct2 C19

c

d

C31

C32

P1

C32

C31

C33

C4

N1

C20 N2

Ru1

P1

C19

N2

C5

C20

N1

C5

C19

O1

ct1

C4

Figure 4 | ORTEP plots of the complexes. ORTEP plots of complexes (a) 1Ab, (b) 5, (c) 6 and (d) 7 as determined by X-ray diffraction studies. Ellipsoids are shown at 50% probability. One molecule of co-crystallized [NBu4]Br and hydrogen atoms, with the exception of H1 in 1Ab are omitted for clarity.

show resonances in the 13C{1H} NMR spectra in a narrow window between d 201.8 and d 208.5 p.p.m. The CO

stretching frequencies seen in the IR spectra decrease by 4200 cm 1 compared to free CO. These spectroscopic and structural data show that the ligand trop2dad is electronically remarkably exible and accommodates Ru between the oxidation states 0 and II. Furthermore, the ligand undergoes reversible

hydrogen uptake and release that occurs at both the C Ctrop and

the NC CN units.

Catalytic experiments. Formaldehyde in water forms methanediol (HOCH2OH), which can polymerize to poly(oxomethyleneglycols). The kinetics are strongly dependent on the temperature and pH of the aqueous solution40,41. The dehydrogenation of aqueous formaldehyde solutions can proceed following different pathways: (i) The Cannizzaro reaction42, simplied as 2H2C O H2O-HCOOH H3COH, followed

by catalytic dehydrogenation of methanol and formic acid as previously reported9,15. (ii) The decarbonylation of formaldehyde according to H2CO-H2 CO followed by a watergas shift

reaction, CO H2O-CO2 H2. (iii) Direct dehydrogenation of

methanediol according to H2C(OH)2-HCOOH H2 followed

by decomposition of formic acid, HCOOH-CO2 H2. Our

previous studies15 showed that base is required to convert the hydrogenated diamine trop2dae ligand as seen in 4 and 7 back to the unsaturated diazadiene trop2dad ligand as observed in 1Aa or 3a. The presence of base will also drive the reactions (i)(iii) by forming potassium carbonate as product. THF solutions of complexes 1K, 1KC, 1Aa, 1Ab, 3a and 47 were tested in the catalytic decomposition of various formaldehyde/water mixtures at 60 C using a reux condenser under an inert atmosphere of argon. The progress of the catalytic conversion in this biphasic system was followed by real-time volumetric measurements of released H2 gas. If not noted otherwise, each reaction was repeated three times and averaged data are given in Table 2 (experimental error of 5%). The catalytic reforming of aqueous formalin (c0 0.47 M) was investigated under various conditions

(entries 113). The highest TON and TOF are achieved with the anionic Ru(II) hydride complexes of type 1 and the neutral Ru(0) complex 4. Catalyst loadings can be as low as 0.4 mol%. As expected, the activity correlates with the concentration of base

6 NATURE COMMUNICATIONS | 8:14990 | DOI: 10.1038/ncomms14990 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14990 ARTICLE

Table 1 | Comparison of selected bond distances () of complexes 1-7 with previously reported data.

Complex C C

backbone C N

backbone RuN1 RuN2

Ruct Ructbackbone C Ctrop

1K* 1.377(8) 1.348(6), 1.355(8) 1.963(3)1.978(5)

2.034(5)2.033(5)

1.439(7), 1.433(6)

1Aaw 1.371(4) 1.352(5), 1.361(5) 1.965(3)1.959(3)

2.027(4)2.025(3)

1.435(6), 1.443(6)

1Ab 1.371(6) 1.360(5), 1.364(5) 1.977(3)1.976(3)

2.031(4)2.047(4)

1.448(4), 1.428(6)

1.444(7), 1.410(7)

4* 1.522(3) 1.487(3) 2.121(2) 1.970(2) 1.453(3)

5z dad-A1.459(10) dad-B1.436(9)

dad-A1.293(12), 1.446(10) dad-B1.425(10), 1.280(11)

(A) 2.051(6) 2.107(7)(B) 2.095(7) 2.144(6)

(A) 2.108(6) 2.056(6)(B) 2.040(6)

3a* 1.381(7) 1.344(6), 1.351(6) 2.007(3)1.976(4)

2.049(4)2.099(5)

Ru1) 2.038(5) Ru2) 1.985(5)

(A) 1.418(12), 1.439(10)(B) 1.430(11), 1.358(13)

5H2z dad-A1.441(10) dad-B1.449(10)

dad-A1.430(9), 1.321(10) dad-B1.429(9), 1.268(10)

(A) 2.099(7)2.058(6)(B) 2.110(7) 2.175(6)

(A) 2.067(3) 2.109(4)(B) 2.046(3)

Ru1) 2.022(3) Ru2) 1.988(4)

(A) 1.438(9), 1.399(5)(B) 1.452(9), 1.528(8)

6 1.388(5) 1.320(5), 1.339(5) 2.038(3)2.059(3)

2.068(2) 1.445(5), 1.522(5)

7y 1.502(3) 1.494(3), 1.474(3) 2.175(2)2.202(2)

2.052(2)2.060(2)

1.422(3), 1.459(3)

*See ref. 15.wComplex 1Aa co-crystallises with one [NBu ]Br molecule.

zdad-A and dad-B refer to each diazadiene ligand in the dimeric complexes. B is the fragment containing the Ru centre coordinated to the CO ligand. One molecule of thf co-crystallises with complexes 5 and 5H .

yOne molecule of dme co-crystallizes with complex 7.

(entries 13, Table 2). Remarkably, catalytic TON is observed in the absence of base but the reaction solution becomes acidic and the conversion drops to 23%. Catalyst 1K cannot be recycled. The conversion decreases strongly from 86% H2 in the rst run to 12% in the second run and the concomitant formation of an insoluble red precipitate is observed (entry 4). The same observation was made with the dibenzo-18-crown-6 complex 1KC as the catalyst. But the complexes 1Aa and 1Ab with quaternary ammonium cations could be reused and these catalysts allowed for repeated substrate addition to the reaction mixture up to six times with little loss of catalytic activity (entries 57). No precipitation of the catalyst was observed and TON values of 41700 could be achieved. After the sixth loading, the activity started to level off and yields dropped to ca. 20%. Full conversion, that is the release of two equivalents of hydrogen, was never achieved. NMR spectroscopic analysis of the solution indicates that there are always small amounts of K(HCO2) present (57%), although in stoichiometric reactions formate is converted to CO2 (equation (2), Fig. 3). The gas phase was analysed by gas chromatography using a thermal conductivity detector, which showed no detectable traces of CO (for detailed instrumentation setup and gas chromatographic analysis, see Supplementary Fig. 13). With the fully hydrogenated Ru(0) complex 4 as catalyst (entry 8), equally high activities as with complexes 1 were achieved, which is consistent with our previous suggestion that the ligand cooperates with the metal and participates in CH activation steps whereby the substrate is dehydrogenated and H2 is transferred to the ligand backbone.

As shown in equation (d) in Fig. 3, complex 1Aa also decarbonylates formaldehyde and transfers H2 to one of the

C Ctrop units and binds CO to the Ru centre forming complex

5H2 under mild acidic conditions (pH 6). Under optimal

catalytic conditions (pH 412), complex 5H2 is able to catalyse the formaldehyde reforming reaction but with signicantly lower TON and TOF numbers (entry 9). Similar results were obtained with the dimeric carbonyl complex 5 which contains the decoordinated but unsaturated C Ctrop unit (entry 10). We

tested the penta-coordinated phosphane complexes 3a, 6 and 7 as catalysts for formalin reforming (entries 1113). The best results were obtained with the complexes [Ru(PPh3)(trop2dad)] 3a and the zero-valent 18 valence congured Ru complex 7; the latter can be converted to 3a in the presence of an excess of base. In this reaction the formation of 1K is also observed while the PPh3 ligand is oxygenated to Ph3PO which is detected by 31P NMR spectroscopy. As in equation (e) in Fig. 3, Ru(II) complex 3a decarbonylates formaldehyde in the absence of base and gives complex 6 as the only product. This complex reacts sluggishly under catalytic conditions and shows the lowest performance of all Ru complexes we tested in the formalin-reforming process. The most notable results were achieved in the conversion of aqueous solutions of paraformaldehyde (c0 0.47 M), which is a

mixture of polyoxymethylenes, HO(CH2O)nH with n 8100

repeating units. With catalyst 1Aa, high conversion (up to 90% H2) and fast gas ow (TOF50420,000 h 1) is observed under basic conditions. The catalyst can be recycled without signicant loss of efciency (entries 14,15). It is very remarkable that the conversion of aqueous paraformaldehyde can be performed

NATURE COMMUNICATIONS | 8:14990 | DOI: 10.1038/ncomms14990 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14990

Table 2 | Catalytic activity in the decomposition of formaldehyde/water mixtures by Ru complexes*.

HO(CH2O)nH

HO(CH2O)n-1H

CH2(OH)2

H2O

O

0.4mol% [Ru] KOH (x equiv)

H H + 60 C

Entry Catalyst KOH (equivalents) TOF50 (h 1)w Total yield H2 (%)z TONmax/durationy 1 1K 23 115/12 h2|| 1K 2 8,109 56 280/12 h3|| 1K 6 17,500 (rst load) 86 430/2 h4 1K (second load) 12 103/2 h5|| 1Aa 6 15,101 (rst load) 90 450/12 min6z 1Aa 6 12,000 (sixth load) 59 1,787/20 min 7|| 1Ab 6 13,520 81 405/15 min8|| 4 6 17,000 90 450/2 h9 5H2 6 6,000 68 340/4 h10|| 5 6 7,500 75 375/4 h11|| 3a 4 3,537 58 290/4 h12|| 6 4 750 69.5 347/4 h13|| 7 4 4,091 65 325/4 h14# 1Aa 6 29,764 (rst load) 90 450/15 min 15** 1Aa 6 22,000 (second load) 85 765/15 min 16ww 1Aa 6 805 92 460/2 h17zz 1Aa 6 o5 /12 h

*Reaction conditions: formaldehyde (1.0 mmol) c 0.47 M, 0.4 mol% [Ru] at 60 C in water/THF (10:1).

wTOF values after 50% conversion (1 equivalents H released per formaldehyde unit).

zYield considering 2 equivalents H /equivalent HCOH. yTON mmol H released per mmol [Ru].

||Values are an average of three catalytic runs.zFinal value after the sixth addition of HCOH aq. to the reaction mixture of entry 5.

#Paraformaldehyde (1.0 mmol) c 0.47 M, 0.4 mol% [Ru] at 60 C in water/THF (10:1).

**Final value after the second addition of HCOH to the reaction mixture of entry 14. Average of three runs. wwSame conditions as in entry 14 under CO atmosphere. Average of three runs.

zzSame conditions as in entry 16 under air. Average of two runs.

H2O K2CO3 + 2 H2

under an atmosphere of CO although with lower activity (entry16). Only when the catalytic reaction is performed in the presence of oxygen (that is, without deoxygenating the solvents and/or under air), no activity is observed (entry 17).

We considered the possibility that formaldehyde disproportionates in a Cannizzaro reaction as shown in (i). In that case, the complexes listed in Table 2 may merely catalyse the dehydrogenation of formic acid and methanol. When formalde-hyde (0.5 M) is heated with 6 equivalents of KOH in D2O at 60 C (without any Ru catalyst), which corresponds with the conditions used in the Ru-catalysed reactions, only 6% conversion is obtained after 15 min. The conversion increases to about 30% after 12 h. It is known that the Cannizzaro reaction proceeds with high efciency at higher concentration43 and indeed with a 5 M formaldehyde solution at pH 14 and 60 C, 490% conversion

to formate and MeOH is achieved. When the catalyst 1Aa is added to this mixture and heated to 60 C, only 22% of H2 is

evolved. This corresponds approximately to the expected amount of H2 from the decomposition of formic acid (25%). Analysis of the reaction mixture by NMR reveals traces of formate (o2%) and 97% of methanol. Furthermore, a rather rapid decomposition of the Ru complex to an insoluble red solid was observed. Hence, methanol is not converted under these conditions, which is in contrast to our previous report where complex 1K was found to convert methanol/water mixtures but at much lower base concentrations and higher temperatures15. Note also that generally the efciency of the catalysis decreases with increasing formaldehyde concentration 40.5 M in water (Supplementary

Table 2). The higher catalytic efciency using diluted formaldehyde solutions was also observed previously by Prechtl et al.26. Hence we conclude that the catalytic reactions proceed via direct dehydrogenation of formaldehyde/ methanediol, and the Cannizaro reaction plays only a minor role under the applied reaction conditions.

Mechanistic DFT study. The experimental stoichiometric reactions and catalytic studies clearly indicate that there are several ruthenium species formed under catalytically relevant reaction conditions, and likely several mechanisms are operative in the conversion of aqueous formaldehyde solutions to carbonate and hydrogen under basic conditions (vide supra). A complete survey of all possible reaction pathways is far beyond the scope of this study. We therefore focussed on the possibility that the NCCN backbone of the ligand participates in key steps of the dehydrogenation reaction and we assumed that the reaction actually proceeds via dehydrogenation of methanediol44. The possible decarbonylation of formaldehyde was not investigated here because the experimental data show that this reaction pathway is likely less favourable or perhaps even a catalyst deactivation pathway. Likewise, the possible disproportionation of formaldehyde to formic acid and methanol was not further investigated. This reaction may be a side reaction but is unlikely to be the major reaction pathway. The experimental data provide some insight that guided our mechanistic DFT studies. Complexes 1 and 4 give rise to much higher TOFs and TONs than obtained with any of the other complexes isolated, and as such, species 3b, 5, 5H2 and 6 are not likely to be catalytically relevant (Table 2). Quantitative and fast formation of neutral complex 2 and H2 from the anionic hydride precursors 1K and 1Aa upon reaction with water under catalytically relevant conditions is of key importance. As is the formation of the catalytically highly active neutral species 4 with a fully hydrogenated dad-backbone upon reaction of 1Aa with water and formaldehyde at room temperature (see Fig. 3). This, in combination with substrate binding being comparatively disfavoured for the saturated (18 valence electron) anionic hydride species 1K and 1Aa, makes it most plausible to propose that the neutral, unsaturated (16 valence electron) species 2 and 4 are involved in the TON. As such, we focussed

8 NATURE COMMUNICATIONS | 8:14990 | DOI: 10.1038/ncomms14990 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14990 ARTICLE

#

HO HO

H

HO HO

OH

+2

H

N

N

H

O O O

O O O

N

0

N N Ru

N

Ru

Ru

N

Ru Ru TS-1

TS-2

N

N

N

B

0.0

0.4 +15.4 +9.6

2m

H H H

#

H

H

O

O

O

H

H H

H

H H

H H

H H H

O O O

Ru

Ru

N

N

N

Ru Ru Ru

N N N

N N

N

C

1.0

+4.4 +14.5 +6.7

C

B

O O

O

#

O

O

#

H

HH H H

H

H

O

O

H

H

H

N

N

N

Ru Ru Ru

TS-3

N N N N

TS-4

+2.2

+1.8 9.4 +7.7

H2

D

A

E

#

H

H

N

H H

H O=C=O

H

Ru Ru

Ru Ru

H

N N N

N

NCO2

H

N N

TS-5

H2

18.0

4.1 19.4 16.0

G

N

F F

N

+15.4 G(kcal mol1)

TS-1

TS-2

+14.5

+9.6

+6.7

B

+4.4

B

C

+2.2 +1.8

+7.7

0.0 0.4 1.0

TS-3

TS-4

D

2m

+H2

A

C

4.1

9.4

E + H2

TS-5

+CO2

16.0

+H2

F + H2

F + CO2+H2 +H2

18.0

19.4

G+CO2

24.0

2m+ CO2+2H2

Figure 5 | Computed pathway for methanediol dehydrogenation catalysed by complex 2m. Calculated pathway (Turbomole, DFT-D3 (disp3), BP86, def2-TZVP) along with their relative free energies (DG298K in kcal mol 1). All energies (also the transition states) are relative to the starting materials (complex 2m methanediol).

our DFT study on these neutral species. We chose simplied models of complexes 2 and 4 (referred as 2m and 4m, respectively)

as catalysts that contain no annulated benzo groups in the trop moiety. DFT methods were used for the calculations and the Minimum Energy Reaction Pathway (MERP) with 2m as catalyst is shown in Fig. 5. The MERP with the fully hydrogenated

complex 4m (and its related isomers) is given in Supplementary Figs 23 and 25 and has a very similar energy prole.

The catalytic cycle starts with exergonic formation of methanediol adduct A, followed by proton transfer from the alcohol moiety of the substrate to one of the dad-backbone nitrogen atoms via transition state TS-1, thus producing

NATURE COMMUNICATIONS | 8:14990 | DOI: 10.1038/ncomms14990 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14990

complex B0. The overall deprotonation step is slightly uphill with a barrier of 15.4 kcal mol 1 (the barrier of this process may be

overestimated, as proton transfer in solution is most likely solvent assisted). Complex B0 then rearranges to hydrogen-bond-stabilized complex B, which by b-hydride elimination subsequently converts to complex C0 via TS-2 at a transition state energy of 14.5 kcal mol 1. This low barrier is in marked

contrast to a recent computational study, which claims that b-hydride elimination from a related metal-bound alcohol/

alcoholate adduct is not feasible35. Note that, in the pathway involving the fully hydrogenated complex 4m (Supplementary

Fig. 23), the computed barrier for hydride transfer from the substrate to the metal is even lower ( 3 kcal mol 1). In this

particular case, this process does not proceed via a classical b-hydride elimination but instead involves an ion-pair polarized transition state stabilized by two hydrogen bonds (4-TS-1; Supplementary Fig. 23). Subsequent rearrangement of C0 produces C, which is preorganized for protonation of the hydride by the coordinated formic acid moiety to produce H2.

Complex C is clearly stabilized by a dihydrogen bond (hydrideproton interaction). Formation of dihydrogen adduct D via TS-3 has a rather low barrier (o 5 kcal mol 1) and

subsequent release of H2 is essentially barrier-less thus producing formate complex E. Direct b-hydrogen elimination from the formate ligand in E via TS-4 produces ruthenium hydride complex F0. This process has the highest barrier ( 17 kcal mol 1) in the catalytic cycle and seems to be the

TOF-limiting step for catalyst 2m. The computed overall barrier for the reaction seems to be somewhat low for a reaction requiring heating in the experimental reactions. The apparently underestimated barrier might be due to the simplied ligand used in the computational studies (truncated trop moiety), unaccounted explicit solvation effects in the gas-phase DFT calculations and/or limitations of the functional used. However, addressing all these issues is beyond the scope of the present paper, which aims at providing a qualitative picture of the most likely pathways occurring at the ruthenium centre. The resulting formation of F0 is quite exergonic, and loss of CO2 from F0 to form F is further downhill on the energy landscape. Proton transfer from the ligand to the metal across the RuN bond of F via transition state TS-5 leads to formation of H2 complex G, which readily loses H2 to complete the catalytic cycle. The MERP computed with catalyst 4m (Supplementary Fig. 23) involves ion-pair polarized, hydrogen-bond-stabilized intermediates (and transition states) in many of the computed steps but has comparably low barriers for all individual reaction steps. As such, the computed pathways for methanediol dehydrogenation by both catalysts 1Aa, which rapidly converts to 2 under the experimental conditions, and 4 provide viable pathways for formaldehyde dehydrogenation. It is quite likely that both are used and contribute to the observed catalytic activity. In both pathways, hydride migration from the substrate to the metal are key steps in the catalytic cycle to produce H2, and in both mechanisms metalligand cooperativity plays an important role.

DiscussionThe ruthenium complexes of type 1 with the [RuH(trop2dad)]

anion, as well as the neutral Ru(0) complex 4, catalyse the conversion of alkaline aqueous solutions of formaldehyde (as formalin or paraformaldehyde) into hydrogen and carbonate at 60 C in a biphasic reaction system with unprecedented high efciency. We assume that methanediol is the substrate that is converted by the ruthenium complexes as catalysts. The faster rate of formation of H2 from aqueous paraformaldehyde might be due to the higher concentration of methanediol in formalin solutions, which consists of higher oligomeric mixtures of methyleneglycols.

The catalytic system contains no phosphanes as ligands, which signicantly improves the energy balance of a catalytic system that is designed to deliver hydrogen as energy carrier (the synthesis of phosphanes necessitates the reduction of phosphate rock, which is a highly energy-intensive process that is likely not counterbalanced by the catalyst during its lifetime). It is very likely that water, and not O2, serves as source of oxygen in the nal product, and in reactions of organic aldehydes with O18-labelled water, RCH O16

H2O18-RCO18/16O18H H2, this hypothesis was proven15,28. In

aqueous formaldehyde, water is already incorporated into the acetal molecule prone to be dehydrogenated to formic acid. The ruthenium centre switches its oxidation states between zero and

II and the ligand very likely participates in various ways in individual steps of the catalytic reaction. The ligand serves not only as a redox non-innocent entity but also participates chemically in the dehydrogenation of methanediol and its oligomers. DFT calculations indicate that both the metal and the ligand play important roles and the activation of the substrate occurs via addition across the polar RuN bond followed by intramolecular b-hydrogen elimination steps at the metal centre. These reaction steps are meanwhile well-established transformations in organometallic chemistry. The stoichiometric experiments show that hydrogen may not only be transferred to the metal centre but can also be stored in various sites of the ligand to give complexes with hydrogenated ligands such as in the Ru(0) amine complex [Ru(trop2dae)] 4, which stores two equivalents of H2. Likely via 1,2-hydrogen shifts from the ligand to the metal centre, the hydrogen content of the ligand is then decreased and H2 is released. The olenic-binding sites may be hydrogenated as well and displaced from the metal centre. But because the CH2-CH2 groups remain in proximity of the metal centre, these reactions are also reversible. Finally, CO complexes have been detected which indicate that the ruthenium amine/imine complexes may likewise decarbonylate (hydrated) formaldehyde and catalyse the conversion of CO to carbonate in course of a watergas shift reaction. This allows to run the formaldehydewatergas shift reaction even under an atmosphere of CO, although with diminished efciency. The catalysts reported here may be considered as catalytic chameleons in the sense that they adapt to their environment and thereby show a remarkable ability to adjust to various reaction conditions.

Methods

Synthesis of [NBu4][Ru(trop2dad)] (1Aa). H2O is added (20 equivalents) to a solution of complex 1K (1.0 equivalents, 11 mM) in THF and the mixture is heated for 6 h at 60 C under an argon stream, while the initial brown solution turns gradually to dark red. After ltration, all volatiles are removed. The obtained residue is washed with n-hexane and dried under vacuum to give pure complex 2 as dark red solid. [Bu4N][HCO2] (1.1 equivalents) is added to a solution of complex 2 (1.0 equivalents, 9.3 mM) in THF and the mixture is stirred at room temperature for 1 h, causing a colour change from red to violet. The solution is ltered through a syringe lter (50 mm porosity). The obtained clear ltrate is layered with n-hexane and cooled to 32 C. After 2 days, air-sensitive burgundy red crystals of complex 1Aa are

isolated by ltration and dried in a stream of argon (51% yield).

Dehydrogenation of aqueous formaldehyde and paraformaldehyde. A 25 ml two-neck round-bottom ask is connected to a reux condenser with argon inlet/outlet, which is coupled to a water-lled gas burette, while the second neck of the ask is capped with a septum. The reaction vessel is purged with argonvacuum cycles for20 min in order to remove air and moisture. A degassed solution of formaldehyde (75ml of a 37 wt% aqueous solution, 1.0 mmol, 1.0 equivalents) or suspension of paraformaldehyde (30.0 mg, 1.0 mmol, 1.0 equivalents) in H2O (2 ml) is added followed by the required amount of base and the mixture is heated at 60 C. After equilibration, the required amount of ruthenium complex in THF (200 ml) is added with a syringe. The mixture is stirred vigorously and the volume of liberated gas is recorded periodically until gas evolution ceases. Released hydrogen was quantied by recording its volume displacement in the eudiometer and correcting its volume for water content.

Data availability. Atomic coordinates and structure factors for the reported crystal structures have been deposited in the Cambridge Crystallographic Data

10 NATURE COMMUNICATIONS | 8:14990 | DOI: 10.1038/ncomms14990 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14990 ARTICLE

Centre under deposition number CCDC-1502942 (1Aa), 1502945 (1Ab), 1502948 (5), 1502951 (5H2), 1502952 (6) and 1502953 (7). These data can be obtained free of charge from the Cambridge Crystallographic Data Centre via http://www.ccdc.cam.ac.uk/data_request/cif

Web End =www.ccdc.cam.ac. http://www.ccdc.cam.ac.uk/data_request/cif

Web End =uk/data_request/cif . Detailed experimental procedures, characterization of compounds and the computational details can be found in Supplementary Figs 127, Supplementary Tables 19 and Supplementary Methods and are also available from the corresponding author upon reasonable request.

References

1. Woodward, J., Orr, M., Cordray, K. & Greenbaum, E. Biotechnology: enzymatic production of biohydrogen. Nature 405, 10141015 (2000).

2. Dau, H. et al. The mechanism of water oxidation: from electrolysis via homogeneous to biological catalysis. Chemcatchem. 2, 724761 (2010).

3. Miller, S. L. & Urey, H. Organic compound synthesis on the primitive Earth. Science 130, 245251 (1959).

4. Steinberg, M. & Cheng, H. C. Modern and prospective technologies for hydrogen-production from fossil-fuels. Int. J. Hydrog. Energ. 14, 797820 (1989).

5. Balasubramanian, R. & Rosenzweig, A. C. Oxidation of methane by a biological dicopper centre. Nature 465, 115119 (2010).

6. Arndsten, B. A. & Bergman, R. G. Unusually mild and selective hydrocarbon C-H bond activation with positively charged iridium(III) complexes. Science 270, 19701973 (1995).

7. Perez, P. J. Alkane CH Activation by Single-Site Metal Catalysis (Springer, 2012).8. Periana, R. A. et al. A mercury-catalyzed, high-yield system for the oxidation of methane to methanol. Science 259, 340343 (1993).

9. Nielsen, M. et al. Low-temperature aqueous-phase methanol dehydrogenation to hydrogen and carbon dioxide. Nature 495, 8589 (2013).

10. Monney, A. et al. Base-free hydrogen generation from methanol using a bi-catalytic system. Chem. Commun. 50, 707709 (2014).

11. Beller, M. et al. Efcient and selective hydrogen generation from bioethanol using ruthenium pincer-type complexes. ChemSusChem. 7, 24192422 (2014).

12. Beller, M. et al. Selective hydrogen production from methanol with a dened iron pincer catalyst under mild conditions. Angew. Chem. Int. Ed. 52, 1416214166 (2013).

13. Hu, P., Diskin-Posner, Y., Ben-David, Y. & Milstein, D. Reusable homogeneous catalytic system for hydrogen production from methanol and water. ACS Catal. 4, 26492652 (2014).

14. Balaraman, E., Khaskin, E., Leitus, G. & Milstein, D. Catalytic transformation of alcohols to carboxylic acid salts and H2 using water as the oxygen atom source.

Nat. Chem. 5, 122125 (2013).15. Rodrguez-Lugo, R. E. et al. A homogeneous transition metal complex for clean hydrogen production from methanol-water mixtures. Nat. Chem. 5, 342347 (2013).

16. Fujita, K., Kawahara, R., Aikawa, T. & Yamaguchi, R. Hydrogen production from a methanol-water solution catalyzed by an anionic iridium complex bearing a functional bipyridonate ligand under weakly basic conditions. Angew. Chem. Int. Ed. 54, 90579060 (2015).

17. Rezayee, N. M., Huff, C. A. & Sanford, M. S. Tandem amine and ruthenium-catalyzed hydrogenation of CO2 to methanol. J. Am. Chem. Soc.

137, 10281031 (2015).18. Kothandaraman, J., Goeppert, A., Czaun, M., Olah, G. A. & Surya Prakash, G. K. Conversion of CO2 from air into methanol using a polyamine and a homogeneous ruthenium catalyst. J. Am. Chem. Soc. 138, 778781 (2016).19. Korstanje, T. J., van der Vlugt, J. I., Elsevier, C. J. & de Bruin, B. Hydrogenation of carboxylic acids with a homogeneous cobalt catalyst. Science 350, 298302 (2015).

20. Breher, F. et al. TROPDAD: a new ligand for the synthesis of water-stable paramagnetic [16 1]-electron rhodium and iridium complexes. Chem. Eur. J.

9, 38593866 (2003).21. Campos, J., Sharninghausen, L. S., Manas, M. & Crabtree, R. H. Methanol dehydrogenation by iridium N-heterocyclic carbene complexes. Inorg. Chem. 54, 50795089 (2015).

22. Houston, P. L. & Moore, C. B. Formaldehyde photochemistry: Appearance rate, vibrational relaxation, and energy distribution of the CO product. J. Chem. Phys. 65, 757763 (1976).

23. Cereghino, J. L. & Cregg, J. M. Heterologous protein expression in the methylotrophic yeast Pichia pastoris. FEMS Microbiol. Rev. 24, 4566 (2000).

24. Sutherland, J. D. The origin of lifeout of the blue. Angew. Chem. Int. Ed. 55, 104121 (2016).

25. Cook, J., Hamlin, J. E., Nutton, A. & Maitlis, P. M. Homogeneously catalyzed disproportionation of acetaldehyde into ethanol and acetic acid. J. Chem. Soc. Chem. Commun. 144145 (1980).

26. Heim, L. E., Thiel, D., Gedig, C., Deska, J. & Prechtl, M. H. G. Bioinduced room-temperature methanol reforming. Angew. Chem. Int. Ed. 54, 1030810312 (2015).

27. Heim, L. E., Schloerer, N. E., Choi, J.-H. & Prechtl, M. H. G. Selective and mild hydrogen production using water and formaldehyde. Nat. Commun. 5, 36213625 (2014).

28. Waals, D. et al. Self-sufcient formaldehyde-to-methanol conversion by organometallic formaldehyde dismutase mimic. Chem. Eur. J. 22, 17 (2016).

29. Heim, L. H., Vallazza, S., van der Waals, D. & Prechtl, M. G. H. Water decontamination with hydrogen production using microwave-formed minute-made ruthenium catalysts. Green Chem. 18, 14691474 (2016).

30. Suenobu, T., Isaka, Y., Shibata, S. & Fukuzumi, S. Catalytic hydrogen production from paraformaldehyde and water using an organoiridium complex. Chem. Commun. 51, 16701672 (2015).

31. Fontecilla-Camps, J. C., Volbeda, A., Cavazza, C. & Nicolet, Y. Structure/function relationships of [NiFe]- and [FeFe]-hydrogenases. Chem. Rev. 107, 42734303 (2007).

32. Luca, O. R. & Crabtree, R. H. Redox-active ligands in catalysis. Chem. Soc. Rev.

42, 14401459 (2013).

33. Trincado, M. & Grtzmacher, H. in Cooperating Ligands in Catalysis, in Cooperative Catalysis: Designing Efcient Catalysts for Synthesis (ed. Peters, R.) (Wiley-VCH, 2015).

34. Lei, M., Pan, Y. & Ma, X. The nature of hydrogen production from aqueous-phase methanol dehydrogenation with ruthenium pincer complexes under mild conditions. Eur. J. Inorg. Chem. 2015, 794803 (2015).

35. Li, H. & Hall, M. B. Role of the chemically non-innocent ligand in the catalytic formation of hydrogen and carbon dioxide from methanol and water with the metal as the spectator. J. Am. Chem. Soc. 137, 1233012342 (2015).

36. Jing, Y., Chen, X. & Yang, X. Theoretical study of the mechanism of ruthenium catalyzed dehydrogenation of methanol-water mixture to H2 and CO2.

J. Organomet. Chem. 820, 5561 (2016).37. Trincado, M., Banerjee, D. & Grtzmacher, H. Molecular catalysts for hydrogen production from alcohols. Energy Environ. Sci. 7, 24642503 (2014).

38. Laporte, C. et al. Strain in organometallics II: controlling the properties of tetra-coordinated iridium complexes using diastereomers of a bis(tropp) ligand system. Inorg. Chim. Acta 357, 19311947 (2004).

39. Piras, E. et al. Chiral phosphane alkenes (PALs): simple synthesis, applications in catalysis, and functional hemilability. Chem. Eur. J. 12, 58495858 (2006).

40. Hahnenstein, I., Hasse, H. ,G., Kreiter, C. & Maurer, G. 1H- and 13C-NMR-spectroscopic study of chemical equilibria in solutions of formaldehyde in water, deuterium oxide, and methanol. Ind. Eng. Chem. Res. 33, 10221029 (1994).

41. Hasse, H. & Maurer, G. Kinetics of the poly(oxymethylene) glycol formation in aqueous formaldehyde solutions. Ind. Eng. Chem. Res. 30, 21952200 (1991).

42. Watanabe, M., Osada, M., Inomata, H., Arai, K. & Kruse, A. Acidity and basicity of metal oxide catalysts for formaldehyde reaction in supercritical water at 673 K. Appl. Catal. A 245, 333334 (2003).

43. Pfeil, E.ber den mechanismus der cannizzaroschen reaktion. Chem. Ber. 84, 229245 (1951).

44. Bielinski, E. A. et al. Base-free methanol dehydrogenation using a pincer-supported iron compound and Lewis acid co-catalys. ACS Catal. 5, 24042415 (2015).

Acknowledgements

This work was supported by the Schweizer Nationalfonds (SNF), Eidgenssische Hochschule Zrich, the FOM-NWO-Shell Computational sciences for energy research initiative (project 13CSER003) and the RPA Sustainable Chemistry of the University of Amsterdam.

Author contributions

H.G., B.d.B. and M.T. directed and conceived this project. M.T., R.E.R.-L. and B.P. conducted the experimental work. V.S. conducted the computational work, guided by B.d.B. All authors discussed the results and wrote the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Trincado, M. et al. Homogeneously catalysed conversion of aqueous formaldehyde to H2 and carbonate. Nat. Commun. 8, 14990doi: 10.1038/ncomms14990 (2017).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2017

NATURE COMMUNICATIONS | 8:14990 | DOI: 10.1038/ncomms14990 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11