Catalytic upgrading of biomass is of paramount importance in view of a sustainable production of the chemicals and fuels needed by the modern society. Hydrodeoxygenation is a promising strategy to lower the oxygen content of oxygen-rich biomass sources (e.g. cellulose or hemicellulose) in order to get biofuels. 5 -hydroxymethylfurfural (HMF) and furfural (FF), obtained from C6 and C5 monosaccharides, respectively, are valuable platform chemicals for the production of useful chemicals and fuels.. The stepwise reduction/deoxygenation of HMF to alkyl furans (e.g. 2,5- dimethylfuran (DMF)) is highly challenging due to the elevated

temperatures needed that may generate several undesired products, like ring opening or self-condensation humin compounds.

The production of 2,5-dimethylfuran (DMF) from biomass sources represents a key strategy in the utilization of renewables for biofuel production. Fructose was already used as starting material in a two-step process involving the dehydration to HMF and the subsequent

hydrogenolysis of HMF to DMF.W In a later report, glucose was also converted to DMF in two steps through HMF formation and using ionic liquids as solvent. Low selectivities to DMF were found, probably due to the low temperature (120 °C) and poor ¾ solubility in the ionic liquids. More examples of efficient DMF production are reported from HMF itself. A Ru/C catalyst gave up to 81% selectivity to DMF in isopropanol as solvent and hydrogen donor. ^[ ¾ The same catalyst afforded even better DMF yields (up to 94.7%) in THF with ¾ (20 bar) starting from HMF or several sugar sources M Also a RU/C03O4 catalyst was successfully applied in the HMF reduction to DMF with yield up to 93.4%. PtCo nanoparticles encapsulated in carbon nanosphere afforded 98% yield DMF from HMFJ ^6]

Selective conversion of HMF to DMF (2,5-dimethylfuran) is a

challenging chemical transformation, involving carbonyl reduction and subsequent hydrogenolysis of two strong C-0 bonds. Usually harsher reaction temperatures are required for the hydrogenolysis to take place, although a few examples reporting milder conditions are known. From a previous report of our research group,! ^7] the use of copper doped porous metal oxides afforded, after optimization of the experimental conditions, up to 81% DMF+DMTHF (DMTHF: 2,5-dimethyltetrahydrofuran) combined yield, both potential biofuels. However, competing side reactions, especially ring opening processes, could not be prevented. A still further drawback of this particular catalyst system is that stability is relatively poor, especially at harsh reaction conditions, and that catalyst recycling is therefore limited. For example, whereas CuPMO yields a product yield of 81-78% at full HMF conversion for the first 3 cycles, the product yield dropped dramatically to 10% after 3 cycles, while unconverted HMF and other side products were the remaining components of the mixture.

Thus, current methods for DMF production from HMF suffer from the problem that side reactions due to either furan ring opening or

overreduction of intermediates cannot be adequately suppressed. In addition, the current best catalytic systems for this reaction are based on expensive ruthenium, platinum and palladium catalysts. Furthermore, also in these highly active systems, deactivation by char formation is often observed.

The present inventors therefore set out to provide an improved process for the valorization of HMF, which is economically attractive. More in

particular, they aimed at improving the yield and selectivity to DMF and other useful hydrogenation products, such as 2,5-furandimethanol (FDM), while avoiding the use of a noble metal catalyst system. An additional aim was to provide a robust catalyst system that allows for more extensive recycling (e.g. at least 4 cycles) than what is possible with current catalysts, in particular with other non-noble catalyst systems such as those

comprising CuPMO.

It was surprisingly found that at least some of these goals can be met by the use of catalysts in the form of CuZn nanoalloys, which are highly active and relatively inexpensive. Interestingly, the hydrogenation reaction was found to be easily tunable towards either DMF or FDM by choosing the reaction conditions. For example, platform chemical 5-hydroxymethylfurfural (HMF) was reduced to either 2,5-furandimethanol or 2,5-dimethylfuran with excellent selectivity (up to 99% and 97%, respectively), depending on the reaction temperature. Importantly, using the CuZn nanoalloy the [DMF + DMTHF] biofuel product yields were found to remain 90-79% over 6 cycles, showing much better product selectivity and in addition superior catalyst stability as compared to copper porous metal oxides. Accordingly, the invention relates to a method of reducing

hydroxymethylfurfural (HMF) to form 2,5-dimethylfuran (DMF) and/or 2,5- furandimethanol (FDM), comprising the steps of:

- providing a starting material comprising HMF in a solvent into a reactor;

- providing ¾ into the reactor; and

- contacting the starting material with a catalyst, wherein said catalyst is copper-zinc alloy in a nanop articulate (nanopowder) form.

The novel process represents a significant potential reduction of the process cost for DMF and related hydrogenation products. First, the nanop articulate Cu/Zn catalyst replaces the expensive ruthenium, platinum and palladium catalysts. Second, the method of the invention produces less side products which reduces the cost of purification and raw material recovery. Third, the catalyst shows superior robustness and stability and allows for more extensive recycling. Fourth, the nan op articulate CuZn was found to be uniquely suited for one-pot conversion reactions that require the use of a solid acid catalyst, like fructose to FDM and furanic ethers derived thereof, whereas CuPMO was found incompatible with such reaction conditions.

The use of metal-based catalyst systems in the conversion of HMF is known in the art. Table 1 provides a summary of the prior art regarding the reduction of HMF to DMF.

Table 1. Relevant examples in literature about the conversion of HMF to DMF.

Reference Catalyst Conditions Conversion DMF Catalyst

yield stability

Dumesic CuRu(3: l) 220°C, 1- Full 79% For 10 wt% et al. ω /C BuOH, 17 HMF feed in bar, 19 vapour- mL/min H2 phase,

deactivation after 1.7 times the amount of catalyst

Bell et al. Pd/C 17% cat. 47% 32% No

[2] loading, indication

3wt% HMF,

120°C,

EMIMCl

(ionic liquid),

lh

Vlachos Ru(5%)/C 42% cat. Full 81% DMF yield et al. Pi loading, 1.2 decreases to wt% HMF 47% and

tetrahydrofuran-dimethanol) in the presence of hydrogen gas and hydrogenation catalysts in the form of nan op articles. Disclosed is a dual hydrogenation pathway, wherein HMF is first hydrogenated to THFDM, followed by hydrogenation of THFDM to 1,6-hexanediol. Among the catalysts for the first reaction, many metal-based (nanop articulate) catalysts are mentioned, among others copper chromite. As can be concluded from Table 1, CuCr gives very poor results and nickel catalysts are preferred for HMF hydrogenation. Whereas a Cu/Zn-based catalyst is used in the second hydrogenation reaction of THFDM to produce 1,6-hexanediol, this catalyst is not a nanoalloy. The present inventors observed that have experimental data, which demonstrate that it is essential to use very fine particles, so there is a clear technical effect of using nanoalloys instead of Cu/Zn catalysts having a larger particle size. WO2011/149339 therefore fails to teach or suggest the use of a nanop articulate CuZn catalyst for HMF hydrogenation.

Zhu et al. (New J. Chem., 2003, 27, 208-210) disclose the hydrogenation of furfural to 2-methylfuran using a Cu/Zn catalyst. The catalyst was obtained via the continuous precipitation method and crushed to 20-40 mesh. So, as in WO2011/149339, the catalyst is not in a

nanop articulate form. Furthermore, despite the structural resemblance between furfural and HMF, their chemistry is very different. In particular, the reactivity of the extra hydroxymethyl-moiety of HMF at increased temperatures causes a number of unwanted/competing side-reactions that are not possible with furfural. Hence, hydrogenation conditions for furfural cannot be extrapolated to HMF, especially when aiming for a highly selective hydrogenation.

Hence, the present invention is the first to demonstrate the unexpected advantage of using a copper-zinc alloy nanopowder in the selective hydrogenation of biomass derivatives.

Provided is a method of reducing hydroxymethylfurfural (HMF) to form 2,5- dimethylfuran (DMF) and/or 2,5-furandimethanol (FDM), comprising the steps of:

- providing a starting material comprising HMF in a solvent into a reactor;

- providing ¾ into the reactor; and

- contacting the starting material with a catalyst, wherein said catalyst is copper-zinc alloy in a nanop articulate form, wherein said contacting is conducted at a reactor temperature of at least 160°C to produce DMF as the major product or at a reactor temperature of less than or equal to 140°C to form FDM as the major product. In one aspect, the method of the invention comprises reducing at least a portion of the HMF to form 2,5-dimethylfuran (DMF). For example, provided is a method wherein the reducing process produces DMF as the major product, preferably wherein the selectivity to DMF is at least 80%, more preferably at least 90%. This is suitably achieved by contacting the starting material and catalyst at increased temperatures, e.g. at a reactor temperature of at least 160°C, preferably at least 180°C, more preferably at least 200°C, or even higher like at least 210°C or at least 220°C.

In another aspect, the method of the invention comprises reducing at least a portion of the HMF to form 2,5-furandimethanol (FDM). For example, provided is a method wherein the converting produces FDM as the major product, preferably wherein the selectivity to FDM is at least 80%, more preferably at least 90%. This can be achieved by conducting the reaction at relatively low temperatures, e.g. at a reactor temperature of less than or equal to 140°C, preferably less than or equal to 130°C, more preferably less than or equal to 120°C.

At a reaction temperature of above 140 to 150°C, a mixture of DMF and FDM is formed. It was surprisingly found that a nan op articulate CuZn catalyst is highly compatible with the reaction conditions needed for the one-pot conversion of fructose to FDM and furanic ethers in the presence of acid catalysts.

Accordingly, in one embodiment a method of the invention comprises the use of a (solid) acid catalyst, such as Amberlystl5 or Nafion, or a similar acidic ion-exchange resin. For example, the method involves a one-pot conversion reaction of fructose to FDM and furanic ethers derived thereof, involving the use of a solid acid catalyst and CuZn alloy nanopowder. In contrast, CuPMO was found incompatible with such reaction conditions.

As used herein, the term "copper-zinc alloy" refers to an alloy consisting mainly of copper (over 50%) and zinc, to which smaller amounts of other elements may be added. Typical standard Cu/Zn alloys include: a) Cu= 55%/Zn= 45%,

b) Cu=65/Zn= 35%,

c) Cu= 70%/Zn= 30%,

d) Cu= 80%/Zn= 20%,

e) Cu=85%/Zn= 15%,

f) Cu= 90%/Zn= 10% In one embodiment, the alloy for use as catalyst in a method of the invention contains 50-70% Cu and 30-50% Zn. In a specific aspect, the catalyst contains about 60% copper and about 40% zinc. Without wishing to be bound by theory, it seems that presence of zinc helps to stabilize the copper active sites, probably due to two main reasons: a) copper nanoparticles are known to grow in size at high temperatures and the presence of zinc, upon at least partial oxidation to zinc oxide might prevent aggregation

phenomena; b) pure copper nanopowders lead to the formation of many unidentified side-products, so a good copper dispersion seems necessary. Furthermore, zinc-oxide may provide more active Lewis acidic sites to aid the required selective deoxygenation processes. A copper-zinc alloy nanop articles for use in the present invention typically has an average particle size of about 25-500 nm, preferably 50-200 nm. In one aspect, the particle size is <150 nm (SEM). Very good results are obtained with an average particle size of about 100 nm.

Copper-zinc alloy nanop articles can be produced according to procedures known in the art. See for example US7,413,725 and references cited therein. They are also commercially available from various sources, for example from Sigma Aldrich. The reduction process can be performed in any type of suitable solvent or solvent mixtures. In one embodiment, the solvent is an alcohol. Preferably, the solvent is selected from ethanol, isopropanol and methyl isobutyl carbinol (MIBC). Most preferred is ethanol.

In another embodiment, the solvent is an ethereal solvent. For example, the solvent is selected from tetrahydrofuran (THF), 2,5-dimethyl

tetrahydrofuran, 2-methyl-tetrahydrofuran and cyclopentyl methyl ether (CPME). 2,5-dimethyl tetrahydrofuran and 2-methyl-tetrahydrofuran are especially preferred since these are biomass derived solvents. 2-methyl tetrahydrofuran can be derived from furfural and 2,5-dimethyl

tetrahydrofuran from HMF itself.

It was surprisingly observed that in CPME, HMF is converted at a faster rate compared to ethanol and the amount of ring opening or side products and intermediates is negligible. Here the two products of the product mixture are DMF and DMTHF. In this solvent, DMTHF (the corresponding aromatic ring reduction product of DMF) is formed in a greater extent than in ethanol. DMTHF is also a biofuel. The highest combined yield of DMF+DMTHF up to (97%) can be obtained in CPME at a faster rate, thus this solvent is preferred if a mixture of DMF and DMTHF is preferred. Individual selectivity of DMF (in CPME) ranging from 67 - 88 % were observed. Accordingly, individual selectivities of 33-12 % of DMTHF were observed. No or negligible side products, unreacted starting material, or reaction intermediates were seen. Accordingly, in a preferred

embodiment a method of the invention is carried out using a solvent comprising or consisting of CPME.

The method of reducing HMF according to the invention comprises providing hydrogen into the reactor to provide a reducing environment. The ¾ pressure is not very critical for the process. Typically, the ¾ pressure during HMF reduction is at least 20 bar and less than or equal to 120 bar. Preferably, the ¾ pressure in the reactor is between 20 and 70 bar. Good results were obtained with 35, 40, 50, 55 or 60 bar.

The HMF to catalyst ratio (w/w) in a method of the invention is usually chosen between 1: 1 and 50: 1; a preferred range is from 2: 1 to 10: 1. For example, very good results were obtained with a 2.5: 1 ratio, a 3: 1 ratio, a 4: 1 ratio, a 5: 1 ratio or a 7: 1 ratio.

As will be appreciated by a person skilled in the art, a method according to the invention advantageously uses a substrate that is derived from a renewable sources. Preferred renewable sources include cellulose, hemicellulose, starch, sucrose, fructose, saccharose, and/or other mono - and disaccharides or crude lignocellulose, preferentially non-edible

lignocellulosic materials such as woodchips, straw, switchgrass, nutshells, corn stower. In particular, a method of the invention uses a starting material produced from the depolymerisation of cellulosic renewables to C-6 sugars and subsequent acid-catalyzed dehydration of C-6 sugars (hexoses) in multiple steps or, preferably, in a one pot procedure. See for example WO2013/085999 disclosing a process to make HMF from glucose, glucose- containing oligomers, glucose-containing polymers, or combinations thereof, the process comprising reacting a feedstock solution comprising glucose, glucose-containing oligomers, glucose-containing polymers, or combinations thereof, in the presence of a homogeneous Bronsted acid catalyst and a homogeneous Lewis acid catalyst, in an aqueous reaction solution, to yield a product solution comprising HMF. The process may further comprise extracting the HMF into a substantially immiscible organic extraction solution. Hence, the step of providing a starting material comprising HMF also encompasses introducing a renewable source into a reactor and allowing for the in situ formation of HMF (e.g. by using a solid acid catalyst), which is then reduced into one or more valuable compounds by the CuZn alloy nanopowder. In a specific aspect, the method comprises a one- pot conversion of fructose into furanics.

The total duration of the reduction process of the present invention can vary, e.g. on type and/or amount of starting materials used, reaction conditions, desired product(s) and yield(s), and the like. Generally speaking, the reaction is conducted for a period of at least 1 hour, preferably at least 6 hours, up to about 18 hours or even longer. In view of the robustness and stability of the copper-zinc alloy nan op articulate catalyst, a method of the invention advantageously comprising recycling at least a portion of the catalyst. For example, the catalyst is recycled at least 4 cycles, preferably at least 5 cycles. In one aspect, catalyst recycling is performed for 3-6 cycles. This will allow for process development towards continuous flow operation.

LEGEND TO THE FIGURES

Figure 1: Recycling runs for DMF production showing that the

nan op articulate Cu/Zn alloy catalyst can be recycled. For details see

Example 3.

Figure 2: Recycling experiments using CuPMO (left panel) or CuZn (right panel) in the conversion of HMF to biofuel additives [DMF+DMTHF].

Reaction conditions: HMF (0.500 g, 4 mmol), catalyst (0.200 g), cyclopentyl- methyl- ether CPME (20 mL), 20 bar H2, 220 °C, 6h. HMF conversion was >95% except in a) 4th cycle with CuPMO catalyst (52%) and b) in the 7th cycle with CuZn catalyst (35%).

Figure 3: Recycling tests for FDM production using CuZn alloy nanopowder catalyst. Reaction conditions: HMF (0.500 g), CuZn (0.100 g), EtOH (20 mL), 50 bar H2, 100 °C, 6h.

Figure 4. Recycling tests for FDM production with CuZn alloy nanopowder catalyst. Reaction conditions: HMF (0.500 g), CuZn (0.100 g), iPrOH (20 mL), 50 bar H2, 120 °C, 6h.

Figure 5: Ή NMR (CD3OD, 400 MHz) of the crude reaction mixture for the one-pot conversion of fructose to furanics in the presence of CuZn and Amberlyst acid catalyst. Full spectral range. Figure 6. Ή NMR (CD3OD, 400 MHz) of the crude reaction mixture for the one-pot conversion of fructose to furanics in the presence of CuZn alloy nanopowder and Amberlyst. Showing aromatic region only. Figure 7. Ή NMR (CD ₃ OD, 400 MHz) of the crude reaction mixture for the one-pot conversion of fructose to furanics in the presence of CuZn alloy nanopowder and Amberlyst. Showing aliphatic region only. Figure 8. Ή NMR (CD3OD, 400 MHz) of the crude reaction mixture for the one-pot conversion of fructose to furanics in the presence of CuPMO and Amberlyst. Showing full spectral range.

Figure 9: Analysis of the total crude product mixture obtained in the one-pot conversion of fructose to furanics in a one-pot procedure using solid acid catalyst (Amberlyst) and CuZn alloy nanopowder. The chart clearly shows the preferential formation of furanics. Only a small portion of the sample is not detectable by GC-FID (non-volatile). Figure 10: Analysis of the total crude product mixture obtained in the one- pot conversion of fructose to furanics in a one-pot procedure using solid acid catalyst (amberlyst) and the prior art catalyst CuPMO. The chart clearly shows that the formation of furanics is minimal. A large part of the sample is not detectable with GC-MS (non- volatile), in accordance with NMR measurements these are fructose derivatives.

EXPERIMENTAL SECTION

Materials and methods: All chemicals and solvents were used as received. 5- methyl-2-furanmethanol (MFM, 97%) was purchased from Acros Organics. 2,5-dimethylfuran (DMF, 99%), 5-methylfurfural (MF, 99%), 5- (hydroxymethyl)furfural (HMF, >99%), 2,5-dimethyltetrahydrofuran

(DMTHF, 96%, mixture of cis and trans) and copper-zinc nanopowders were purchased from Sigma Aldrich. Hydrogenation reactions: A stainless steel autoclave (100 niL) equipped with a mechanical stirrer was charged with catalyst (0.100 or 0.200 g), HMF (0.500 g) and toluene or decane (0.250 mL, internal standard) in the appropriate solvent (20 mL). The reactor was sealed and pressurized with ¾ (in a range of 10-30 bars), heated to the desired temperature and stirred at 800 rpm. After reaction, the autoclave was cooled down to room

temperature, the content was transferred to a centrifuge tube, centrifuged, and the reaction mixture was separated from the solids via decantation. Samples of the filtered solutions were injected in a Hewlett Packard 5890 GC-MS-FID with a Restek RTX-1701 capillary column. Selectivity is defined as the ratio of the peak area of a given compound and the total area of all the compounds produced in the reaction. Yield values were calculated based on the use of internal standard and calibration curves.

EXAMPLE 1: Selection of CuZn nanoparticles

This example describes the selection of CuZn nanoparticles from among a set of commercially available copper-based nanopowders for the studied reaction in hand.

Screening for HMF reduction (see Scheme 1 and 2) was carried out in ethanol solvent at 220 °C and 30 bar ¾ pressure and the results are summarized in Table 2.

Table 2. Screening of different copper nanoparticles in the DMF+DMTHF production.

DMF+DMTHF DMF/DMTHF

EntryM Catalyst

yield ratio M

1 Cu 55 18/1

2 CuO 30 30/1

3 CuZn 75 14/1

4 CuFe ₂ 0 ₄ 45 10/1 5M CuZnFe ₂ 0 ₄ 73 9/1

[a] Reaction conditions: 0.500 g HMF, 0.100 g catalyst, 20 mL EtOH, 0.250 mL toluene (internal standard), 220 °C, 30 bar ¾, 6h.

[b] 95% HMF conversion.

[c] Determined by GC-FID.

DMF DMTHF

(major) (minor)

Scheme 1

D F DSVITHF

HMF major minor

up to 96% combined yield

Scheme 2

With all catalysts, complete conversion of HMF was achieved within 6h, except with CuZnFe2O ₄ that afforded 95% conversion. Cu nanopowder afforded a good 55% DMF+DMTHF yield with a high DMF/DMTHF ratio (table 2, entry 1). No large presence of intermediates was found in the reaction mixture, which rather contains 11% of the ethyl ether 2- (ethoxymethyl)-5-methylfuran (EMMF) and other unidentified products. This high loading of copper respect to the substrate (20 wt%) might favor not only the extensive deoxygenation of the intermediates FDM and MFM but also undergo structural variations accounting for the formation of undesired products, of which more than 20% remain unidentified. More importantly, the presence of Zn seems to play a crucial role in the

stabilization of the catalytic copper sites, as described later. The use of CuO afforded even lower DMF yield among the tested Cu nanopowders indicating that the in-situ reduction of Cu ²⁺ to Cu° is a prerequisite for the

hydrodeoxygenation to take place (table 2, entry 2). After 6h, more than 50% of the products are FDM and MFM while the ratio DMF/DMTHF is the most favorable among all the nanopowders. The CuZn alloy afforded 75%

DMF+DMTHF yield with a high DMF/DMTHF ratio, while the two main intermediates to DMF (FDM and MFM) represent still 13% of the product.

Mixed copper-iron oxide nanopowder led to a lower product yield (45%, table 2, entry 4) and large presence of intermediates and side- products, accounting for a lower activity respect to the simple copper nanopowder. It seems that the presence of zinc is essential for an optimal catalytic activity (table 2, entry 3 and 5), in fact, no remarkable difference between the CuZn and CuZnFe2O ₄ nanopowders was observed in terms of main product yield, although the DMF/DMTHF ratio is slightly lower with the latter and conversion with CuZn nanopowder results in a cleaner product distribution (only 3% unidentified product respect to 11% with the iron oxide containing nanopowder). Interestingly, the CuZn alloy in a 60 mesh powder form resulted in no activity and humin formation.

EXAMPLE 2: Optimization of DMF production

Due to its promising catalytic features, the CuZn nanopowder was selected for further DMF yield improvement. First, the reaction time was prolonged to ensure a full hydrodeoxygenation of FDM and MFM intermediates and after 18h DMF yield improved from 75% to 83% (table 2, entry 3 vs. table 3, entry 1). With double catalyst loading, the yield of desired products raises to 88% after only 6 hours while the DMF/DMTHF ratio decreased to 7/1 (table 2, entry 2). Indeed, no relevant effects on the yield were observed when lowering the hydrogen pressure from 30 to 20 bar (table 3, entry 3);

interestingly, a big excess of DMF over DMTHF was found. Lower DMF yield and a dark yellow solution is probably due to concomitant self- condensation processes of HMF, when further decreasing hydrogen pressure. Then, all experiments were carried out at 20 bar ¾ in different solvents with 0.1 g nanopowder for 18 h. In ethanol, a combined 88%

DMF+DMTHF yield was obtained with a better DMF/DMTHF ratio respect to the identical value previously discussed (table 3, entry 4 vs. Table 3, entry 2). The ethanol solvent also plays a crucial role during the progress of the reaction due to a minor etherification process (up to 5% selectivity to EMMF). The use of other more hindered alcohol solvents (iPrOH and methyl isobutyl carbinol, MIBC) did afford yield improvements, probably due to minor interactions between the solvent and the reaction intermediates. Also the unidentified compounds stay below the 3% respect to the values up to 6% in ethanol. It is noteworthy that the DMF to DMTHF ratio is reduced to 7/1 and 4/1 in iPrOH and MIBC solvents, respectively, which implies a role of the more reactive ethanol in preventing from ring overreduction. A 10.0 wt% HMF concentration was tested in MIBC (table 3, entry 7) with the aim of proving the catalyst robustness in conditions that usually favor humin formation. Results very similar to those of the corresponding run at lower concentration (3.2 wt%, table 3, entry 6) and no high boiling products were detected, as confirmed by thermogravimetric and elemental analysis on the spent catalyst. Also ethereal solvent CPME (cyclopentyl methyl ether) was tested giving 97% selectivity to DMF+DMTHF with a slightly lower

DMF/DMTHF ratio of 3/1 (table 3, entry 8). CPME has a remarkable positive effect in the selective hydrogenolysis process, being a high-boiling, less polar and not functionalized solvent, affording the highest overall DMF+DMTHF yield. CPME has been recently identified as alternative green solvent to other ethereal compounds, such as THF, diethyl ether, dioxane, dimethoxy ethane and MTBE.

Table 3. Reaction conditions screening for the optimization of DMF- DMTHF production. M

time Cat. P DMF/

Entry ^[a] solvent [h] [gr] [bar] DMF+

DMTHF DMTHF

yield [ ] ratio

1 EtOH 18 0.1 30 83 16/1

2 EtOH 6 0.2 30 88 7/1

3 EtOH 6 0.2 20 84 36/1

4 EtOH 18 0.1 20 88 14/1

5 iPrOH 18 0.1 20 91 7/1

6 MIBC 18 0.1 20 92 4/1

MIBC 18 0.3 20 90 5/1

8 CPME 18 0.1 20 97 3/1

[a] Reaction conditions: 0.500 g HMF, 20 mL solvent, 220 °C, 0.250 mL toluene or decane (internal standard). Full HMF conversion, [bl Determined by GC-FID. Based on the internal standard method, [c] l0wt HMF concentration (1.67 g in 20 mL solvent) was used.

Tables 4A and 4B show further embodiments of the invention.

Table 4A: reaction conditions used 0.1 g catalyst, 0.5 g HMF (4 mmol), 20 mL ethanol, 30 bar ¾, 800 rpm.

5

Table 4B: Conditions: 0.2 g catalyst, 4 mmol HMF, 20 mL ethanol, 30 bar H ₂ , 800 rpm.

[a] Reaction in isopropanol. [b] 20 bar ¾. [c] 0.1 g catalyst, 20 bar ¾. EXAMPLE 3: Recycling of nanoparticulate catalyst

In order to test the catalyst stability, the reaction was performed in MIBC with 0.5 g HMF and 0.2 g catalyst at 220 °C for 15h. The CuZn

nanoparticulate catalyst was recycled and reused for three additional cycles (Fig. 1). The DMF+DMTHF selectivity decreased from 89% to 40% (1st and 4th cycle, respectively) but the catalytic activity was fully recovered by calcination of the spent catalyst and the catalyst could be reused for another cycle (87% DMF+DMTHF selectivity).

Reaction conditions were further optimized in CPME. Reaction time could be reduced to 3 h at 200 °C with and excellent yield 97% yield for

(DMF+DMTHF; DMF:DMTHF = 5: 1). For results see Table 5.

Surprisingly, even concentrated (10 wt%) HMF solutions were cleanly converted into a mixture of DMF and DMTHF (94% yield) in CPME, while DMF yield was 90%. Similarly, a 10 wt% HMF solution was converted in MIBC at 220°C and longer reaction time, albeit with slightly lower DMF + DMTHF yield (88%). Interestingly, a 5 w% catalyst loading (3 w% Cu), typical for noble metal catalysts was adopted in CPME solvent with full substrate conversion, 89% combined fuel yield, and excellent DMF/DMTHF ratio of 35: 1. This holds much promise for future upscaling, and continuous operation of this system.

Tn le 5, DM I ^'■ DM ^' I'1 1 1· ^" product ion i n ( Ί 'Μ Ι-: solvent .

Entr W Time T DMF+DMTHF DMF/DMTHF

(h) (°C) yield [%]M ratio ^[b]

1 6 220 97 3/1

2 6 200 97 5/1

5

3 3 200 96 5/1

4W 6 200 94 18/1

5M 18 220 89 35/1

[a] Reaction conditions: 0.500 g HMF, 0.100 g CuZn, 20 mL CPME,

0.250 mL decane, 20 bar Full HMF conversion, [b] Determined by

GC-FID. [c] 10 wt% HMF concentration (1.72 g HMF in 20 mL CPME)

and 0.2 g catalyst were used (7% Cu/HMF ratio), [d] 2.00 g HMF in

mL CPME, 40bar H ₂ , and 0.1 g catalyst were used (3% Cu/HMF ratio).

EXAMPLE 4: Recycling experiments demonstrating superior stability of CuZn nanobrass over CuPMO As a comparative example, recycling tests were additionally performed with CuPMO (catalyst also used in document Kumalaputri et al. (ref. 7) for the production of biofuel additives dimethylfuran (DMF) and dimethyl tetrahydrofuran (DMTHF).

Conditions were: CuPMO (0.200 g), HMF (0.500 g, 4 mmol) in CPME (20 mL) at 220°C for 6h under 20 bar H ₂ . After the first cycle, the content of the reactor was transferred into a centrifuge tube, centrifuged, the liquid phase decanted and analysed by GC-FID-MS, the solids were washed, dried overnight and directly used in the next cycle without further treatment. The combined yields of DMF (2,5-dimethyl furan) + DMTHF (2,5-dimethyl- tetrahydrofuran), both biofuel additives, obtained in each cycle are shown on Figure 1, left. These results show that for the first 3 cycles a product yield of 81-78% at full HMF conversion was achieved, consistently with the best result previously reported with the same catalyst in iPrOH (81%, Kumalaputri et al., ChemSusChem 2014, 7, 2266-2275). However, after 3 cycles, the product yield dropped dramatically to 10% while unconverted HMF and other side products were the remaining components of the mixture.

In contrast, a comparison of the recycling experiments using the CuZn nanobrass catalyst under analogous conditions, the product yields remain 90-79% over 6 cycles, showing much better product selectivity and in addition superior stability of the CuZn alloy nanopowder catalyst (see Figure 2).

EXAMPLE 5: Recycling experiment using CuZn alloy nanopowder, demonstrating catalyst stability under various conditions

Further recycling tests were performed with CuZn nanopowder under various reaction conditions, showing sufficient stability of this catalyst for the production of 2,5-furandimethanol (FDM). For each run 0.1 g catalyst and 0.5 g HMF was used. After each run, the liquid phase was separated by centrifugation and decantation and the solids were washed and dried before reusing in the next cycle. First, reactions in ethanol were carried out at 100°C using HMF (0.500 g), CuZn (0.100 g), EtOH (20 mL), 50 bar H ₂ , 100 °C and each cycle was performed for 6h. The results are summarized on Figure 3. FDM yields were as follows: 87% in the 1st cycle, 87% in the 2 ^nd cycle, 88% in the 3 ^rd cycle, 55% in the 4 ^th cycle, 34% in the 5 ^th cycle, 11% in the 6 ^th cycle. The decrease in activity was due to a decrease in conversion. No other side products were observed in these runs. An even better performance was observed using isopropanol and 120°C, for 6 hours when the FDM yield values for the first three cycles displayed >95% product yield (Figure 4). A slightly lower yield (66%) was observed in the 4 ^th cycle, and this value decreased gradually in the 5 ^th and 6 ^th cycle. In conclusion, an appropriate choice of the solvent can provide full selectivity to FDM and good catalyst stability.

The liquid samples in isopropanol (1 ^st , 3 ^rd and 5 ^th cycle) were analyzed by ICP for detection of metal traces and no leaching of copper and zinc was found (<lmg/Kg, see Table 6).

Table 6. Results of the ICP analysis on the liquid samples obtained in the 1 ^st , 3 ^rd and 5 ^th cycle in iPrOH with the CuZn catalyst.

EXAMPLE 6: HMF reduction to FDM

Copper zinc nanopowder and some commercial catalysts were also tested the hydrogenation of HMF at mild temperature for the achievement of useful diol building blocks (suppliers are indicated in table 6). Table 7. Composition details (and suppliers) of the commercial catalysts used.

The vessel was pressurized with 70 bar H ₂ and subsequently heated to 120 C for 3 hours. In the adopted experimental conditions, all commercial catalysts showed good to excellent activity in the hydrogenation of HMF to FDM (2,5-furandimethanol) and THFDM (2,5-tetrahydrofurandimethanol), which account together for a combined selectivity of >80% (see Scheme 3).

Scheme 3. Commercial compositions, such as catalyst D and catalyst G, both based on copper and nickel supported on S1O2 and Si02 Zr02, respectively, showed good selectivity to FDM (Table 8, entry 1 and 2) but no complete conversion of HMF, especially with catalyst G, which is characterized by a lower content of Cu and Ni than catalyst D. Ni Raney alloy showed complete conversion of HMF and high selectivity to THFDM (94%, table 8, entry 3), while Ni supported on ceria and zirconia (table 8, entry 4) led to an approximately equimolar mixture of FDM and THFDM. The strong interaction between nickel and the support might cause a reduced hydrogenation activity. Also the simultaneous presence of copper and nickel centres may attenuate the hydrogenation activity of nickel, which is very active in the reduction of C=C bonds (Table 8, entry 1 and 2 vs. entry 3).

Interestingly, Cu-Zn alloy showed quantitative conversion of HMF and excellent selectivity to FDM (94%, table 8, entry 5), with no ring- hydrogenation. With the exception of Pt/C, which also selectively reduced HMF to FDM (table 8, entry 6), other hydrogenation catalysts based on noble metals preferably gave THFDM or a mixture of FDM and THFDM. Interestingly, Pd/A Oe and Pd/C showed high selectivity to THFDM by hydrogenation of both C=0 and C=C bonds (table 8, entry 7 and 8).

Table 8. Hydrogenation of HMF at mild temperature with commercial catalysts

FDM

EntryM Catalyst Conversion [%] THFDM ['

[%]

1 Cat. D 96 85 9 (8/1)

2 Cat. G 85 71 12 (7/1)

3 Ni Raney >99 4 94 (10/1)

4 Ni/Ce0 ₂ /Zr0 ₂ >99 44 54 (9/1)

5 Cu/Zn >99 94 -

6 Pt/C 93 85 3 (5/1)

7 Pd/Al ₂ 0 ₃ >99 <1 99 (9/1)

8 Pd/C >99 - 89 (5/1)

9 Ru/Al ₂ 0 ₃ >99 44 54 (9/1)

10 Ru/C >99 4 88 (10/1)

[a] Reaction conditions: 0.5 g HMF, 0.1 g catalyst, 20 mL ethanol, 120 °C, 70 bars ¾, 3h. [b] In parentheses, the ratio between the cis and trans isomer is reported.

Also Ru/C gave THFDM with high selectivity (table 8, entry 10), but by changing the support from carbon to alumina an equimolar mixture of FDM and THFDM was found (table 8, entry 9).

The reduction to THFDM preferentially gave the cis isomer in all the experiments, ranging in a 5/1 to 10/1 the ratio between cis and trans isomer. In conclusion, copper-zinc (brass) nanoalloy proved to be an ideal catalyst for highly selective conversion of HMF to value added products for several reasons:

• The selectivity can be modulated by variation of the reaction

temperature giving either fuels DMF+DMTHF or FDM as main product (>90% yield for both cases);

• the conversion is clean (no char formation) despite the lability of

HMF at high temperatures and up to 10 wt% HMF concentration can be used;

· catalyst can be easily recovered and reused in consecutive runs and after several cycles activity can be regenerated by simple calcination.

EXAMPLE 7: Importance of nanoparticulate form of copper zinc alloy

A sample of a commercial 60 mesh (250 μ) copper zinc alloy (supplied by Sigma Aldrich) was grinded in a mortar until a finer powder (<25 μ) was collected. This sample (0.1 g) was tested in a the hydrodeoxygenation of HMF at 220 °C with 20 bars H ₂ in methyl isobutyl carbinol (MIBC) for 18 h. Analysis of the liquid phase showed that catalyst activity is low with this particle size. Conversion was not complete (83%) and FDM was the major product (73% GC selectivity). Other minor products were reaction

intermediates such as 5-methylfuran-2-carbaldehyde (MF) and (5- methylfuran-2-yl)methanol (MFM), which account together for 5%, and some unidentified products (5%). Some visible humins are formed probably due to the low catalyst activity and the persistence of labile HMF and FDM compounds. This demonstrates that the nano-scale of particle size is a requirement for high hydrogenation/dehydroxygenation activity. EXAMPLE 8: One-step, one pot conversion of fructose to furanics

The conversion of fructose directly to furanics is highly desired, because of the more affordable price of this starting material. This one-pot reaction consists of a system comprising an acid catalyst as well as a hydrogenation catalyst. During this two-step process, first the dehydration of fructose results in the formation of HMF or its ethers and this is followed by immediate reduction of these compounds over CuZn catalyst to 2,5-furan- dimethanol (FDM) and other furanic ethers. The advantage of this procedure also is, that in this case, isolation of HMF is not needed.

The present example confirms, that the nanoparticulate CuZn catalyst is uniquely suited for the one-pot conversion of fructose to FDM and furanic ethers derived thereof, but also shows that CuPMO is not compatible with the reaction conditions that require the use of a solid acid catalyst.

The solid acid catalyst used is Amberlyst 15 and the hydrogenation catalyst is either nanoparticulate CuZn or CuPMO.

One run using CuZn alloy nanopowder, and one run using CuPMO was conducted, using otherwise identical reaction conditions. The reaction conditions used are as follows: fructose (0.50 g); iPrOH (20mL); catalyst (0.1 g), Amberlyst 15 (0.04 g), dioxane internal standard (ΙΟΟμΚ) at 120°C, 30 bar H2, 12 h. HPLC analysis revealed full conversion of fructose (no peak at 13.4 min) for both runs. Extensive product identification was carried out by GC-MS-FID and Ή NMR. NMR analysis of product mixtures obtained with either CuZn or CuPMO catalyst

First, Ή NMR analysis of the crude reaction mixtures was conducted in order to determine crucial differences between the reaction using CuZn alloy nanopowder, compared to CuPMO, the spectra are displayed in Figures 5- 8. 1H NMR analysis of the mixture obtained with the CuZn catalyst shows clear peaks due to the formation of furanics which are labelled in Figure 6 (aromatic region) and Figure 6 (aliphatic region).

In contrast to the run with CuZn alloy nanopowder, the Ή NMR analysis of the crude mixture obtained with CuPMO catalyst shows negligible amount of signals that can be ascribed to furanics. The aromatic region is completely lacking these signals. The remainder is represented by isopropyl levulinate signals, CH3 and CH signals of isopropyl groups plus a complex set of signals between 3.40 and 4.30 ppm that correspond to the region of CH and CH2 of sugar derivatives. These can be attributed to intermediate product of fructose dehydration as previously observed in other reports (Li et al., Green Chem., 2012, 14, 2752) and hint at a possible neutralization of the basic supported PMO catalyst with the acidic resin.

4 5

Scheme 4. One-pot and two-step strategies for the valorization of fructose in iPrOH. REFERENCES

[1] Y. Roman-Leshkov, C. J. Barrett, Z. Y. Liu, J. A. Dumesic, Nature 2007, 447, 982-986.

[2] M. Chidambaram, A. T. BeU, Green Chem. 2010, 12, 1253-1262.

[3] J. Jae, W. Zheng, R. F. Lobo, D. G. Vlachos, ChemSusChem 2013, 6, 1158-1162.

[4] L. Hu, X. Tang, J. Xu, Z. Wu, L. Lin, S. Liu, Ind. Eng. Chem. Res. 2014, 53, 3056-3064.

[5] Y. Zu, P. Yang, J. Wang, X. Liu, J. Ren, G. Lu, Y. Wang, Appl. Catal. B 2014, 146, 244-248.

[6] G.-H. Wang, J. Hilgert, F. H. Richter, F. Wang, H.-J. Bongard, B.

Spliethoff, C. Weidenthaler, F. Schiith, Nat. Mater. 2014, 13, 293-300.

[7] A. J. Kumalaputri, G. Bottari, P. M. Erne, H. J. Heeres, K. Barta, ChemSusChem 2014, 7, 2266-2275.