1 Introduction

Scientific innovation in various fields is surely necessary to achieve carbon neutrality. At present, most carbon atoms are originated from fossil resources (crude oil, coal, and natural gas), which are transformed to electricity, hydrocarbon fuels and a variety of chemicals, and upon their utilization the carbon atoms are emitted as carbon dioxide (CO2). For the realization of carbon neutrality, the carbon atoms in fuels and chemicals should be originated from CO2 and/or renewable resources such as biomass [1, 2]. The shift of the carbon sources from fossil to CO2 and biomass causes the drastic process alteration for the production of fuels and chemicals. Fossil resources mainly consist of carbon and hydrogen atoms and have generally very low oxygen content. Therefore, they are suitable for the production of hydrocarbon fuels. In addition, a variety of oxygenates (oxygen-containing chemicals) have been produced by the addition of oxygen atoms to petroleum-derived hydrocarbons via oxidation with O2 and hydration reactions. In stark contrast, for the utilization of carbon atoms from CO2 and biomass, reduction reactions are much more important because of the high oxygen content in CO2 and biomass. Another important point is that reductants should be obtained in renewable methods such as renewable electricity and molecular hydrogen (H2). In particular, H2 will be available by the water electrolysis using renewable electricity. Actually, the applicability of H2 is very wide as the reductant, and H2 has been used for a variety of important organic reactions such as hydrogenation, hydrocracking, hydrodeoxygenation, hydrodesulfurization and so on. Much attention has been recently paid to CO2 hydrogenation, hydrodeoxygenation of biomass-derived platform chemicals and so on.

In terms of utilization of biomass as a substitute of fossil resources, the concept of “biomass-derived platform” has been proposed, and a series of biomass-derived chemicals are listed as key intermediates [3, 4]. A variety of cyclic ethers such as furfural, 5-hydroxymethylfurfural, their hydrogenated compounds, etc. are included in such the biomass-derived chemicals [5,6,7]. There have been a lot of reports on the catalyst development for the reductive transformation of these cyclic ethers, and this review is focusing on recent progress on catalysts for the ring-opening C-O hydrogenolysis of cyclic ethers. Biomass-derived cyclic ethers mentioned in the review are summarized in Fig. 1.

Fig. 1
figure 1

Biomass-derived cyclic ethers mentioned in this review

Full size image

Furfural has been synthesized from hemicellulose by hydrolysis to xylose and subsequent dehydration, and this method has been established practically [8]. Furfuryl alcohol is produced by the hydrogenation of C = O in furfural, and its synthesis has been already industrialized. Tetrahydrofurfuryl alcohol can be synthesized by the hydrogenation of both furan ring and C = O in furfural. 2-Furancarboxylic acid (FCA) is synthesized by the oxidation of furfural [9, 10]. Hydrogenation of furan ring in FCA gives tetrahydrofuran-2-carboxylic acid [10]. The reaction of furfural with an ionic liquid hydroxylamine salt gives 2-furonitrile, which can be further transformed into furfurylamine by the hydrogenation of the nitrile group [11]. The hydrogenation of both nitrile group and furan ring of 2-furonitrile gives tetrahydrofurfurylamine [12]. Direct amination of furfural to furfurylamine is possible with an appropriate catalyst [13].

5-Hydroxymethylfurfural (HMF) is the dehydration product of hexoses [14]. Dehydration of fructose is the easiest method to obtain HMF; however, glucose or even cellulose can be the reactant of HMF synthesis by dehydration. The hydrogenation of C = O and furan ring gives tetrahydrofuran-2,5-dimethanol [5, 7]. Tetrahydropyran-2-methanol can be obtained by the dehydration of 1,2,6-hexanetriol which is derived from the C-O hydrogenolysis of tetrahydrofuran-2,5-dimethanol [15]. The oxidation of HMF can give 2,5-furandicarboxylic acid (FDCA) [16, 17]. 2,5-Bis(aminomethyl)furan is synthesized from HMF by two-step amination [13] or oxidation to 2,5-diformylfuran followed by double amination [18, 19].

Sugar alcohols are also regarded as biomass-derived platforms [3, 4]. Dehydration of erythritol, a C4 sugar alcohol, produces 1,4-anhydroerythritol [20]. Similarly, dehydration of xylitol, the most popular C5 sugar alcohol, produces 1,4-anhydroxylitol [21]. In the case of sorbitol, which is the most popular C6 sugar alcohol, the dehydration occurs twice to give isosorbide [22].

The biomass-derived cyclic ethers listed in Fig. 1 are further converted to useful chemicals. One of the important target families is straight-chain compounds with two function groups such as diols, hydroxy carboxylic acids, etc. These compounds will be utilized as monomers for the synthesis of polymers [23, 24]. One of the typical target products from biomass-derived cyclic ethers is α,ω-diols such as 1,5-pentanediol, which have been used as monomers for the synthesis of polyesters, polycarbonates, etc. The basics and recent progress on the use of 1,5-pentanediol are also explained in this review. Ring-opening of cyclic ethers typically proceeds by hydration, and the production of levulinic acid with functional groups at 1- and 4-positions from furfuryl alcohol or HMF is well known [25]. However, it is impossible to synthesize molecules with functional groups only at the two terminal positions such as α,ω-diols by hydration of substrates listed in Fig. 1. Removal of functional groups is necessary, and the ring-opening C-O hydrogenolysis, which is the target reaction in this review, is a direct method to obtain useful less-functionalized straight-chain products from biomass-derived cyclic ethers. This review focuses on catalysts for the production of oxygen-containing chemicals, i.e. partial C-O hydrogenolysis. Catalysts for total C-O hydrogenolysis of cyclic ethers to hydrocarbons suitable to biofuel or lubricants do not require regio-selectivity among various types of C-O bonds, and, therefore, the key points of the catalyst development are different between selective ring-opening C-O hydrogenolysis and total C-O hydrogenolysis. Reviews of total C-O hydrogenolysis are available elsewhere [26,27,28]. The reactions that do not cleave C-O bonds in the ring such as the syntheses of 1,4-pentanediol and 1,4-butanediol are also excluded from this review, and there are reviews focusing on these reactions [29, 30]. This review will first explain the ring-opening hydrogenolysis of unsaturated substrates (compounds with furan ring) such as furfural and furfuryl alcohol, in section 2. Next, the ring-opening hydrogenolysis of saturated substrates will be explained in section 3, since the two types of reactions are much different in catalysts and mechanism. The synthesis method of each target product will be compared with other methods in section 4. The most important utilization of the target products, namely as a monomer of plastics, is introduced in section 5. Section 6 is the summary and outlook.

2 Ring-opening of unsaturated cyclic ethers

2.1 Ring opening of furfural or furfuryl alcohol

First, we introduce the typical products from furfural, which is the most accessible unsaturated cyclic ether from biomass. The reductive conversion of furfural almost always produces furfuryl alcohol first, and when furfuryl alcohol is used as a substrate, the products are usually similar to those from furfural. The combination of dissociation of C-O bond in furan ring in furfural or furfuryl alcohol and hydrogenation of C = C (and C = O in furfural) gives 1,2- and 1,5-pentanediols. The ease of production of C5 monomers is one of unique points of biomass utilization. While C6 compounds have been key intermediates in the petrochemical industry because of the very high availability of benzene, biomass includes hemicellulose mainly consisting of C5 xylose units, which can be converted easily into C5 building blocks such as xylitol and furfural. Furthermore, the conversions of these C5 building blocks into C5 monomers such as 1,5-pentanediol are easier than those of biomass-derived C6 building blocks into C6 monomers such as sorbitol or HMF into 1,6-hexanediol.

The direct synthesis of 1,2- and 1,5-pentanediols from furfural has been a hot topic [31,32,33,34] since Xu et al. reported Li-modified Pt/Co2AlO4 catalyst. In their report, with the optimized catalyst and reaction conditions (ethanol solvent, 1.5 MPa H2, 413 K), the yield of 1,5-pentanediol reached 34.9%, and 1,2-pentanediol and tetrahydrofurfuryl alcohol were also formed (16.2% and 31.3% yield, respectively). Furfuryl alcohol was quickly formed under the reaction conditions, and it was then converted to 1,5-pentanediol, 1,2-pentanediol and tetrahydrofurfuryl alcohol as parallel reactions. Tetrahydrofurfuryl alcohol was not converted to pentanediols at all under the reaction conditions on this catalyst. Various related systems have been reported so far, and there are common features in most systems: (i) furfuryl alcohol is quickly formed when furfural is used as a substrate; (ii) 1,2-pentanediol, 1,5-pentanediol and tetrahydrofurfuryl alcohol are formed as parallel reactions; and (iii) tetrahydrofurfuryl alcohol is not converted to pentanediols. Most catalysts are reusable with negligible or small loss of performance. The selectivity ratio of pentanediols and tetrahydrofurfuryl alcohol depends on the catalysts and reaction conditions. Typical catalysts contain Pt or Cu as an active metal with high dispersion (a few nanometer of particle size) as well as a basic component. High dispersion is beneficial for exposing more metal atoms on a particle surface and increasing the number of metal-support interface sites. In addition, the interaction between metal particles with good dispersion and basic support is generally strong, which can give good stability. Here, recent reports (published in 2019 or later) showing high selectivity to either pentanediol are summarized. Reports of ring-opening of 5-hydroxymethylfurfural with similar systems are very few, and they are also explained in this section.

2.1.1 Selective formation of 1,2-pentanediol

Before 2019, there were four papers reporting > 60% yield of 1,2-pentanediol: Pt/hydrotalcite (80% yield from furfuryl alcohol; 73% from furfural) [35], Pt/CeO2 (65% from furfural [36] and 77% from furfuryl alcohol [37], and Pd/MMT-K10 (66% from furfural) [38]. The average particle sizes of these noble metal catalysts were in the range of 1–4 nm. The effectiveness of Pt/hydrotalcite catalyst and its related ones in view of both selectivity and stability has been also proved by other groups [39, 40]; this type of catalysts are currently the most promising ones. The Pd/MMT-K10 system used very high temperature (493 K) to suppress the total hydrogenation to tetrahydrofurfuryl alcohol, since Pd has generally high activity in C = C hydrogenation. Table 1 summarizes representative reports of 1,2-pentanediol production from furfural and furfuryl alcohol.

Table 1 Systems of opening of furan ring in furfural and furfuryl alcohol to 1,2-pentanediol (> 60% yield)
Full size table

Important recently-developed systems will be explained in detail. Zhu et al. [40] reported the improvement of Pt catalyst by carefully engineering the structure of catalytic Pt sites. The support was Mg- and Al-containing layered double hydroxides (LDHs) coating γ-Al2O3 (MgAl-LDHs@Al2O3) which has similar surface properties to hydrotalcite. The structure of Pt species was tailored by changing the loading amount and reduction methods: single Pt atoms in 0.08 wt% Pt/MgAl-LDHs@Al2O3 after calcination and reduction with H2, two-dimensional (2-D) Pt clusters in 0.08 wt% Pt/MgAl-LDHs@Al2O3 after gas-phase H2 reduction, three-dimensional (3-D) Pt clusters in 1.67 wt% Pt/MgAl-LDHs@Al2O3 after gas-phase H2 reduction, large Pt particles in 1.67 wt% Pt/MgAl-LDHs@Al2O3 after heating in N2 and reduction with H2. The highest selectivity to 1,2-pentanediol was obtained with the catalyst containing 3-D cluster (86% at > 99% conversion). Atomic Pt sites, 2-D Pt cluster and large Pt particles favors the C-O cleavage at the side chain (benzyl-like position), step-by-step ring hydrogenation and one-step ring hydrogenation, respectively (Fig. 2).

Fig. 2
figure 2

Proposed reaction paths for the hydrogenation/hydrogenolysis of furfuryl alcohol over engineered Pt sites [40]. Reprinted with permission. Copyright (2020) American Chemical Society

Full size image

There have been recent two reports using non-Pt noble metal catalysts with high yield of 1,2-pentanediol. Pisal and Yadav reported that Rh/OMS-2 (OMS-2: manganese-oxide-based cryptomelane-type octahedral molecular sieves; KMn8O16) gave 87% yield of 1,2-pentanediol from furfural [41]. Pt/OMS-2 was much less active in the conversion of furfuryl alcohol intermediate. Pd/OMS-2 was very active, and good selectivity to 1,2-pentanediol was obtained; however, the selectivity was lower than that of Rh-OMS-2 because of the formation of tetrahydrofurfuryl alcohol. This behavior can be explained by the difference in the activity of two metals for furan ring hydrogenation (Pd > > Rh). The support, OMS-2, has rather strong basic sites judging from CO2-TPD; however, the introduction of Rh on OMS-2 weaken the basicity of OMS-2. As a result, Rh/OMS-2 has weak base sites (CO2 desorption temperature: around 373 K) and it was maintained after the reaction. Solvent effect was very large in this catalyst system, and methanol solvent was by far the best solvent for 1,2-pentanediol production among alcohols, toluene and 1,4-dioxane. Manganese-based supports have been less investigated than Mg- and/or Al-based supports, which are also weakly basic, for this target reaction. Further development of manganese-based support is expected in future studies.

Upare et al. [42] reported that Ru-Sn alloy supported on ZnO is very effective catalyst for the conversion of furfural to 1,2-pentanediol (84% yield). Ru/ZnO without Sn showed rather high selectivity to 2-methyltetrahydrofuran, and Ru-Sn alloy supported on SiO2 and Al2O3 showed much lower activity of furfuryl alcohol conversion and lower selectivity to 1,2-pentanediol. This behavior suggests that the role of Sn is the suppression of the side reaction to 2-methyltetrahydrofuran on monometallic Ru surface, and ZnO can be connected to activity in furfuryl alcohol conversion to 1,2-pentanediol, which can be explained by the surface basicity of ZnO. Pt-Sn/ZnO catalyst was also effective in 1,2-pentanediol production (77% yield); however, the activity and selectivity were slightly lower than those of Ru-Sn/ZnO. The selectivity including the 1,2-pentanediol/1,5-pentanediol ratio was drastically changed by the solvent (Table 2). 2-Propanol was the best solvent for the production of 1,2-propanediol. On the other hand, primary alcohols and non-alcohol solvents tended to produce larger amount of 1,5-pentanediol, and 66% yield of 1,5-pentanediol was obtained in g-butyrolactone solvent. Although not highlighted in the cited paper, this 1,5-pentanediol yield was even higher than those in most reports targeting 1,5-pentanediol by ring-opening hydrogenolysis of furfural or furfuryl alcohol (see the next subsection).

Table 2 Solvent effect on the ring-opening hydrogenolysis of furfural over Ru-Sn/ZnO [42]
Full size table

The system by Upare et al. mentioned above is rather exceptional among Ru-based catalysts because monometallic Ru surface is usually active in hydrogenation of furfuryl alcohol to tetrahydrofurfuryl alcohol. Yamaguchi et al. [44] investigated monometallic Ru catalysts on various supports (Al2O3, ZrO2, MgO, CeO2, TiO2, graphite and H-ZSM-5), and the selectivity to tetrahydrofurfuryl alcohol was always higher than that to 1,2-pentanediol. The highest 1,2-pentanediol yield in this report was 42% over Ru/MgO catalyst with the co-production of tetrahydrofurfuryl alcohol in 51% yield.

Non-noble metal catalysts are preferable in view of the catalyst cost. Cu-based catalysts have been known to give 1,2-pentanediol from hydrogenolysis of furfural or furfuryl alcohol. The first paper was published over 90 years ago, using CuCr2O4 catalyst (40% yield of 1,2-pentanediol from furfuryl alcohol) [45]. Cu-based catalyst systems for 1,2-pentanediol production always have considerable selectivity to 1,5-pentanediol, and it is more difficult to obtain high selectivity to 1,2-pentanediol than noble metal catalysts. Among the reported non-noble-metal catalysts, Cu@MgO-La2O3 (Cu 16 wt%, Mg/La = 3) catalyst showed the highest yield (64%) of 1,2-pentanediol [43]. Cu was the best active metal on this support, and the catalysts with other active metals mainly produced tetrahydrofurfuryl alcohol (Fig. 3).

Fig. 3
figure 3

Ring-opening hydrogenolysis of furfuryl alcohol over M/MgO-La2O3 catalysts [43]. PeD: pentanediol; THFA: tetrahydrofurfuryl alcohol; 2-MF: 2-methyfuran. Reaction conditions: furfuryl alcohol 1.5 g, 2-propanol 28.5 g, catalyst 0.3 g, 6 MPa H2, 413 K, 8 h. Reprinted with permission. Copyright (2022) Elsevier

Full size image

Generally, the combination of Mg and Al is more common than that of Mg and La for catalyst support. Cu catalysts on Mg-Al mixed oxide have been used for furfural or furfuryl alcohol hydrogenolysis, and the highest 1,2-pentanediol yield in each report was similar (48–55%) [46,47,48].

This type of catalyst (Cu catalyst on Mg-Al mixed oxide) was also applied to the reaction of HMF [49]. Cu1.5Mg1.5Al oxide catalyst gave 40% yield of 1,2-hexanediol from HMF in 2-propanol solvent at 423 K. The possible reaction pathways are shown in Fig. 4. It was proposed that HMF was converted to 5-methylfurfural in the first step, and then the C = O bond was hydrogenated to give 5-methylfurfuryl alcohol. The ring-opening hydrogenolysis of 5-methylfurfuryl alcohol produces 1,2-hexanediol in a similar way to that of furfuryl alcohol to 1,2-pentanediol. The yield of 1,2-hexanediol was not so high, and by-products are not mentioned clearly, although the yields of 5-methylfurfuryl alcohol, tetrahydrofuran-2,5-dimethanol, and 5-methylfurfural were rather low. It seems to be characteristic that 1,2,6-hexanetriol was not formed. If the first step of the hydrogenolysis is the hydrogenation of C = O bond of 5-HMF to 2,5-furandimethanol, the subsequent ring-opening hydrogenolysis will give 1,2,6-hexanetriol. Surely C-O hydrogenolysis of one of the hydroxymethyl groups in 2,5-furandimethanol can give 5-methylfurfuryl alcohol which is a precursor of 1,2-hexanediol, and this reaction route may also be involved. The ring-opening hydrogenolysis of 2,5-furandimethanol should be slower than that of 5-methylfurfuryl alcohol, and that of 5-methylfurfuryl alcohol should be fast enough to suppress the formation of 2,5-dimethylfuran.

Fig. 4
figure 4

Possible reaction pathway of HMF hydrogenolysis to 1,2-hexanediol

Full size image

2.1.2 Selective formation of 1,5-pentanediol

Because 1,5-pentanediol is more attractive target than 1,2-pentanediol, selective synthesis of 1,5-pentanediol from furfural and furfuryl alcohol has been intensively carried out. In fact, the pioneering report [34] focused on the synthesis of 1,5-pentanediol. However, increasing the selectivity to 1,5-pentanediol far above 1,2-pentanediol is not easy for the catalytic direct ring-opening of furfural or furfuryl alcohol. High yield of 1,5-pentanediol can be obtained more easily by ring-opening of saturated tetrahydrofurfuryl alcohol, which will be discussed later.

Representative systems for 1,5-pentanediol production by ring-opening hydrogenolysis of furfural or furfuryl alcohol are summarized in Table 3. Catalysts for formation of 1,5-pentanediol typically use Co as one of active components, including the work in 2011 [34, 50, 51]. Recent works tend to use mixed oxides as precursors. A typical preparation method of the mixed oxides is the thermal decomposition of hydrotalcite-like compounds (HTlc) which are synthesized by deposition of aqueous metal nitrates with alkali hydroxide and alkali carbonate. This method is helpful to produce solid solution alloy whose surface is well contacted with basic metal oxides, giving good support-metal interaction and resistance to sintering.

Table 3 Catalytic systems of opening of furan ring in furfural or furfuryl alcohol to 1,5-pentanediol (> 30% yield with > 50% yield of total pentanediols)
Full size table

Recently-reported systems, grouped according to the catalyst components, will be introduced in detail. Ni1Co11Al (HTlc-derived; Ni:Co:Al = 1:11:3) showed good selectivity to 1,5-pentanediol in the reaction of furfuryl alcohol, and 43% yield of 1,5-pentanediol was achieved (the main by-product was tetrahydrofurfuryl alcohol (30% yield)) [52]. Monometallic Co12Al exhibited low activity in the reaction of furfuryl alcohol and monometallic Ni12Al was active in the hydrogenation of furfuryl alcohol to tetrahydrofurfuryl alcohol. On Ni1Co11Al, the formation of 1,5-pentanediol and THFA from furfuryl alcohol proceeds in parallel, which is similar to the systems of 1,2-pentanediol formation, and the formation of 1,5-pentanediol was more favorable at the reaction time of 433 K and 4 MPa H2 pressure. Due to the strong hydrogenation capability of Ni, hydrogen molecules are easier to be adsorbed on the surface of Ni than that of Co, and the adsorbed H2 can be dissociated into activated hydrogen atoms rapidly. The activated hydrogen atoms are then transferred from Ni to the adjacent Co surface via hydrogen spillover, and subsequently being conveyed to the furan ring for the C-O hydrogenolysis reaction. Ni-Co-Al with higher Ni content (Ni:Co:Al = 4.5:3.5:2.2) was reported by another group [53]. This catalyst gave 48% yield of 1,5-pentanediol with 25% yield of tetrahydrofurfuryl alcohol and 19% yield of 1,2-pentanediol in the reaction of furfural. The product distribution was very sensitive to the reduction temperature of the catalyst, and medium reduction temperatures such as 673 and 773 K were favorable to the formation of 1,5-pentanediol. Higher reduction temperatures such as 873 and 973 K led to the formation of tetrahydrofurfuryl alcohol. A more recent paper reported Ni-Co-Al catalyst with Ni:Co:Al = 7:27:18 and reduction temperature of 773 K showing 43% yield of 1,5-pentanediol from furfural [54]. Similar catalyst system (Ni-Co-Al mixed oxide catalysts) has been reported to be effective to the conversion of HMF to 1,2,6-hexanetriol [62]. The Ni-Co-Al (Ni:Co:Al = 0.5:2.5:1) showed high yield of 1,2,6-hexanetriol (65%) together with 20% yield of tetrahydrofuran-2,5-dimethanol at 393 K. To the synthesis of 1,2,6-hexanetriol from HMF or 2,5-furandimethanol, both types of reactions that produce 1,2-pentanediol and 1,5-pentanediol from furfural (or furfuryl alcohol) are applicable. The obtained yield of 1,2,6-hexanetriol from HMF was similar to those of the sum yield of 1,2- and 1,5-pentanediols from furfural with similar catalysts.

The combination of Cu, Co and Al for mixed oxide catalysts has been also reported. Cu-Co-Al (HTlc-derived; Cu:Co:Al = 0.2:5.8:2.2) gave 45% yield of 1,5-pentanediol with 22% yield of tetrahydrofurfuryl alcohol and 15% yield of 1,2-pentanediol in the reaction of furfuryl alcohol [55]. Monometallic Co-Al and Cu-Al catalysts showed much lower catalytic activity (lower conversion of furfuryl alcohol) than bimetallic Co-Co-Al catalysts. The same group also reported another type of Cu-Co-Al catalysts: precursors prepared by homogeneous deposition with urea instead of the combination of sodium hydroxide and sodium carbonate. The crystal structure of the precipitate was rather like malachite instead of hydrotalcite. Cu-Co-Al (malachite-like-compound-derived; Cu:Co:Al = 0.1:2.9:1) gave 41% yield of 1,5-pentanediol in the reaction of furfural, and another main product was tetrahydrofurfuryl alcohol with similar yield to 1,5-pentanediol [56]. The main substrate in this article was furfural, but the yield of 1,5-pentanediol from furfural was clearly lower than 40% on this catalyst. After all, both reports for Cu-Co-Al catalysts showed similar yield of 1,5-pentanediol to that over Ni-Co-Al catalysts introduced above (40+% from furfuryl alcohol).

The combination of Co, Mg and Al was also reported. With this combination, only Co can have the metallic state under the reaction conditions. Co-Mg-Al (HTlc-derived; Co:Mg:Al = 1.5:1.5:1) gave 52% yield of 1,5-pentanediol with 12% yield of tetrahydrofurfuryl alcohol, 9% yield of 1,2-pentanediol, and 9% yield of 1-pentanol in the reaction of furfuryl alcohol [57]. This is the highest yield of 1,5-pentanediol produced directly from furfuryl alcohol with Co-based catalysts. The similar level of the yield of 1,5-pentanediol (49%) was obtained in the reaction of furfural. Lower Co/Mg ratio than 1 decreased the activity of furfuryl alcohol conversion, and the undesirable side reaction of furfuryl alcohol decreased the selectivity to 1,5-pentanediol. Higher Co/Mg ratio than 1 gave high reactivity of furfuryl alcohol; however, probably because of the insufficient basicity, the selectivity to 1,5-pentanediol was lower. Further combination of Pt with HTlc-derived Co-Mg-Al was reported [58]. Pt@Co2+ (Co:Mg:Al = 1:3:1; 2 wt%Pt; impregnation of HTlc with Pt before calcination for decomposition of HTlc) gave 47% yield of 1,5-pentanediol with 27% yield of tetrahydrofurfuryl alcohol, and 19% yield of 1,2-pentanediol. Comparable results were obtained when furfural was used as the substrate. The effect of M2+ in mixed metal oxides derived from layered double hydrotalcite was investigated, and only Co was effective to the formation of 1,5-pentanediol (Fig. 5). Pt-Co-Mg-Al catalyst prepared by co-precipitation for all components including Pt showed almost no formation of 1,5-pentanediol, indicating the necessity of Pt species with appropriate positions/structures to activate H2 molecules.

Fig. 5
figure 5

Ring-opening hydrogenolysis of furfuryl alcohol over Pt@HTlc-derived mixed metal oxide catalysts (Pt 2 wt%, Bivalent metal:Mg:Al = 1:3:1) [58]. FAL: furfuryl alcohol; PDO: pentanediol; THFA: tetrahydrofurfuryl alcohol. Reaction conditions: substrate:catalyst = 1:1 (w/w), 2-propanol solvent, 423 K, 3 MPa H2 and 4 h. Reprinted with Creative Commons Attribution 4.0 International License https://creativecommons.org/licenses/by/4.0/

Full size image

All the reports introduced above in this section used basic catalysts. On the other hand, a series of works using Cu-zeolite catalysts which have acidic nature have been reported by Dai et al. [59,60,61]. The catalytic performance of Cu-zeolite for furfuryl alcohol reduction was evaluated in the continuous-flow fixed-bed reactor, and much higher 1,5-pentanediol/1,2-pentanediol selectivity ratio was obtained than those with Pt-, Cu- and Co-based catalysts in other reports. The highest yield of 1,5-pentanediol (84%; reaction at 453 K) was obtained with Cu-MFI (H-ZSM-5; SiO2/Al2O3 = 60) with hierarchical structure (Cu 5 wt%; ammonia evaporation and hydrothermal treatment after Cu loading). The catalyst prepared without hydrothermal treatment or that simply prepared by impregnation of commercial support showed lower selectivity to 1,5-pentanediol (Fig. 6), demonstrating the good performance of Cu species located inside the pore system. The Cu particle size in the developed catalysts was 2–3 nm. Although the size is larger than the micropore size of MFI, the confinement of Cu particles in the zeolite particle was confirmed by TEM. When silicalite-1 without acidity was used instead of H-ZSM-5, 1,5-pentanediol selectivity became much lower, and tetrahydrofurfuryl alcohol became one of major products. When tetrahydrofurfuryl alcohol was used as substrate over the Cu/MFI catalyst, pentanediols were rarely produced. This result indicated that tetrahydrofurfuryl alcohol is not an intermediate in the ring-opening hydrogenolysis, and this behavior is similar to other ring-opening hydrogenolysis using basic catalysts mentioned above.

Fig. 6
figure 6

Ring-opening hydrogenolysis of furfuryl alcohol over Cu-MFI catalysts prepared with different methods [59]. IE: simple impregnation; AE: impregnation followed by ammonia evaporation; AEH: impregnation followed by ammonia evaporation and hydrothermal treatment; FFA: furfuryl alcohol; PDO: pentanediol; THFA: tetrahydrofurfuryl alcohol. Reaction conditions: catalyst 1 g (5 wt% Cu; reduced at 573 K), 2 wt% furfuryl alcohol in ethanol 0.2 mL/min, 2.5 MPa H2, 433 K. Reprinted with permission. Copyright (2023) Elsevier

Full size image

2.1.3 Reaction mechanism of ring-opening hydrogenolysis of furfuryl alcohol assisted by base

As mentioned in the introductory paragraph of this subsection 2.1, the reaction pathways are mostly the same among systems: furfuryl alcohol is quickly formed when furfural is used as the substrate, and ring-opening of furfuryl alcohol occurs without formation of tetrahydrofurfuryl alcohol. The reactivity of tetrahydrofurfuryl alcohol is very low. The catalysts usually contain basic components, although there is an exception (Cu/zeolite catalysts). However, there is little consensus of the reaction mechanism of ring-opening of furfuryl alcohol. Many papers in this field have a section of “reaction mechanism”, and several mechanisms have been proposed using various data such as reactivity of related molecules, kinetics, spectroscopy of adsorbed species and density functional theory (DFT) calculations. However, the evidence level is generally not high; the use of another alcohol solvent in the conversion of alcohol (furfuryl alcohol) largely decreases the reliability of the adsorption studies using IR and DFT which are much affected by a protic solvent.

The proposed mechanisms are mainly divided into two categories: (i) direct dissociation of C-O bond in the furan ring adsorbed on the catalyst surface by the attack of active hydrogen species, and (ii) stepwise reaction involving partial hydrogenation of furan ring (to 2,3- or 4,5-dihydrofuran derivatives) as the first step (Fig. 7). More papers proposed the former type of reaction mechanism than those proposing the latter type ([34,35,36, 41,42,43, 46,47,48, 50, 52, 53, 57, 63, 64]). The role of base was frequently proposed as the adsorption site of -CH2OH group of furfuryl alcohol [35, 41, 43, 47]. Some papers proposed that the furan ring of furfuryl alcohol is adsorbed on metal surface [36, 47]. The hydrogen species dissociated on the metal surface attack either C-O bond and the subsequent keto-enol tautomerization produces 1-hydroxy-4-penten-2-one and 5-hydroxy-3-pentenal, and the hydrogenation gives 1,2-pentanediol and 1,5-pentanediol, respectively. For the type (ii) mechanism, the C(sp3)-O bond formed by the hydrogenation is dissociated by hydrogenolysis, and 2,3- and 4,5-dihydrofurfuryl alcohol is finally converted to 1,5-pentanediol and 1,2-pentanediol, respectively [38, 40, 65].

Fig. 7
figure 7

Two types of mechanisms proposed for 1,2-pentanediol or 1,5-pentanediol formation from furfural. Blue and red arrows are rate-determining steps and also determine the selectivities to 1,2-pentanediol and 1,5-pentanediol

Full size image

Both types of mechanisms have several weaknesses. The factor to determine the 1,2-pentanediol/1,5-pentanediol ratio is not clear for both mechanisms. For the type (i) mechanism, it is difficult to explain why the attack of hydrogen species to C-O bond is favored over C = C hydrogenation. It seems difficult to strongly bind -CH2OH group at the base site in alcohol solvents, especially in primary alcohols. Furan rings have some aromatic nature, and therefore the C-O bonds in furan rings are stronger than those in saturated compounds. For the type (ii) mechanism, the role of base is not clear. Based on this situation, here we summarize the reported results related to mechanistic discussion.

The dependence of reaction results on reaction conditions, especially reaction orders with respect to hydrogen pressure, is fundamental data for discussion of reaction mechanism of hydrogenolysis [66]. However, the reaction orders for the ring-opening hydrogenolysis of furfuryl alcohol have not been well investigated. It should be also noted that the reduction degree of catalysts may depend on hydrogen pressure and reaction temperature, which complicates the determination of reaction order. On the other hand, the dependence of selectivity on hydrogen pressure has been sometimes reported. Generally, high hydrogen pressure increases the selectivity to tetrahydrofurfuryl alcohol. Low hydrogen pressure decreases the reaction rate, and the selectivity to 2-methylfuran or 2-methyltetrahydrofuran becomes larger if the catalyst has activity in this hydrodeoxygenation reaction. The 1,5-pentanediol/1,2-pentanediol ratio does not largely depend on the hydrogen pressure. One example is shown in Fig. 8 [64]. This behavior of selectivity change suggests that the number of involved hydrogen atom in the rate-determining step of each reaction decreases in the following order: total hydrogenation to tetrahydrofurfuryl alcohol > ring-opening hydrogenolysis to pentanediols > side chain hydrodeoxygenation to 2-methylfuran.

Fig. 8
figure 8

Effect of H2 pressure on furfuryl alcohol hydrogenolysis over Cu(10 wt%)/Al2O3 [64]. Reaction conditions: 0.06–0.25 g Cu, 4 g furfuryl alcohol, 36 g ethanol, 413 K, 4 h, approximately 20% conversion. PeD: pentanediol; MF: methylfuran; THFA: tetrahydrofurfuryl alcohol. Reprinted with permission. Copyright (2016) Dalian Institute of Chemical Physics

Full size image

As described above, typical catalysts for ring-opening hydrogenolysis of furfuryl alcohol are basic. However, it is difficult to determine the true effect of base in the catalysis because changing the basicity usually alters other properties of catalysts such as metal dispersion and reducibility. One of limited examples for the systematic control of basicity was given by Li et al. [67]. They prepared Pt/MgAl2O4 catalysts modified with various amount of Li. The Li addition did not largely change the Pt particle size from STEM measurement. The reaction results of furfuryl alcohol over these catalysts are shown in Table 4. There were conflicting phenomena: for 4 wt% Pt catalysts, the addition of Li first much increased and then gradually decreased the activity. The catalyst with large amount of Li (10 wt%) still showed higher activity than the Li-free 4 wt% Pt catalyst. The selectivity to tetrahydrofurfuryl alcohol decreased and that to pentanediols increased by the Li addition. On the other hand, for 8 wt% Pt catalysts, the addition of Li rather decreased the activity, while the trend of selectivity change was similar to 4 wt% Pt catalysts. Another strange point is that the 4 wt% Pt/MgAl2O4 catalyst showed much lower activity than the 8 wt% Pt/MgAl2O4 catalyst even considering the difference of Pt amount. The low activity of 4 wt% Pt/MgAl2O4 catalyst may be due to a specific reason to this catalyst, for example, covering the Pt surface with MgAl2O4 support. If this 4 wt% Pt/MgAl2O4 is removed from this table, the addition effect of Li becomes simpler: activity and selectivity to tetrahydrofurfuryl alcohol were decreased, and selectivity to pentanediols was increased.

Table 4 Ring-opening hydrogenolysis of furfuryl alcohol over nPt/mLi/MgAl2O4 catalysts (n, m: loading amount in wt%) [67]
Full size table

Another investigation of the effect of base is reported by Li et al. [68], by using the concept of reversed catalyst (more popular name: “inverse catalyst” [69], where metal oxide is deposited on the surface of large metal particles. They compared the catalytic performances of Cu powder, BaO/Cu powder and Li2O/Cu powder in furfuryl alcohol hydrogenolysis (Table 5). Loading basic oxides on Cu powder significantly increased the activity and selectivity to 1,2-pentanediol. The main by-product over unmodified Cu powder catalyst was 2-methylfuran, and the selectivity to 2-methylfuran was decreased by loading basic oxides, although the formation rate of 2-methylfuran itself was also increased. They further prepared aluminum-hydroxide-modified Cu catalyst (Al(OH)3/Cu) by leaching Al in CuAl alloy with NaOH aq. This preparation method is like that of skeletal Ni (activated Raney Ni) catalysts, and the formed metal catalysts should have large metallic surface area [70]. The activity of Al(OH)3/Cu catalyst was much higher than those of basic oxide/Cu powder catalysts, probably mainly because of the larger metallic surface area. They further compared the apparent activation energies of furfuryl alcohol hydrogenolysis over Cu powder and Al(OH)3/Cu. While the apparent activation energy of 2-methylfuran formation was similar over both catalysts (125 kJ/mol), that of 1,2-pentanediol formation over Al(OH)3/Cu (75 kJ/mol) was much lower than that over Cu powder (123 kJ/mol). These results suggest that modification with base increased the activity in pentanediol formation, although the effect of the surface structure of Cu cannot be excluded.

Table 5 Ring-opening hydrogenolysis of furfuryl alcohol over Cu-based reversed catalysts [68]
Full size table

DFT calculation is a strong tool in the investigation of reaction mechanism. There are several studies using DFT for the reaction mechanism of ring-opening hydrogenolysis of furfuryl alcohol [36, 42, 53, 68, 71]. However, most of them did not consider the involvement of solvent molecules; in addition to the effect of permittivity, alcohol solvents can be adsorbed competitively on the catalyst surface, especially at base sites. Most studies only investigate the reaction route of ring-opening hydrogenolysis and not the hydrogenation reactions, while in the real systems the selectivity ratio of pentanediols/tetrahydrofurfuryl alcohol is very important. The calculation was usually carried out with a flat surface model under periodic boundary condition, which is difficult to include the support effect, while supports (usually basic) have large effects on the ring-opening hydrogenolysis. Some DFT studies proposed the mechanism involving C-O bond dissociation in furan ring (type (i) mechanism in Fig. 7) [36, 42, 53, 68], and another DFT study proposed the mechanism involving partial hydrogenation of furan ring followed by C-O hydrogenolysis (type (ii) mechanism) [71].

While more papers proposed the mechanism involving C-O bond dissociation in furan ring (type (i)), the mechanism involving partial hydrogenation (type (ii)) can more reasonably explain the experimental results, especially for the suppression of ring hydrogenation. In order to understand the inherent properties of possible intermediates, the calculated structure and internal energy of dihydrofurfuryl alcohols by DFT are summarized in Fig. 9. The addition of hydrogen atoms to C-O bonds (1,2- and 1,5-dihydrofurfuryl alcohols) much increased the internal energy, which suggests the difficulty of the direct attack of C-O bonds in furan ring. 2,3-Dihydrofurfuryl alcohol and 4,5-dihydrofurfuryl alcohol are similarly stable, and both of them can be formed by partial hydrogenation. An important point is that the C(sp3)-O bond becomes significantly longer than even tetrahydrofuran or tetrahydrofurfuryl alcohol. The long C-O bonds suggest the higher reactivity toward C-O dissociation reactions. Of course, the adsorption on the metal surface much changes the energy level of each species, but these inherent stability trends can affect the reactivity over any catalysts. On the other hand, as mentioned earlier, the mechanism involving C-O hydrogenolysis of dihydrofurfuryl alcohol does not include the effect of base. Then, we propose another mechanism which is a variant of type (ii) (Fig. 10). First, the furan ring is partially hydrogenated to 2,3-dihydrofurfuryl alcohol or 4,5-dihydrofurfuryl alcohol. Next, the C(sp3)-O bond in 2,3- and 4,5-dihydrofurfuryl alcohol is dissociated by the assistance of base via intramolecular E2 reaction to give 1,3-pentadien-1,5-diol and 2,4-pentadien-1,2-diol, respectively. The alcohol solvent stabilizes the products of the intramolecular elimination reaction. Promotion of the elimination step over the total hydrogenation of the dihydrofurfuryl alcohol intermediate increases the selectivity to pentanediols. This mechanism can explain not only the positive effect of base but also the selectivity changes with hydrogen pressure: The C-O dissociation of dihydrofurfuryl alcohol toward pentanediol does not involve hydrogen species, while further hydrogenation to tetrahydrofurfuryl alcohol involves addition of hydrogen species. Higher hydrogen pressure favors total hydrogenation to tetrahydrofurfuryl alcohol.

Fig. 9
figure 9

Calculated structures and internal energies of hydrogenation products of furfuryl alcohol. Calculation level: B3LYP/6-311 + + G(d,p), Gaussian09 program package. The relative energy (ΔE) to furfuryl alcohol + H2 is shown in parentheses in unit of kJ/mol. The lengths are shown in Å

Full size image
Fig. 10
figure 10

Another possible mechanism of ring-opening hydrogenolysis of furfuryl alcohol

Full size image

The selectivity control of 1,2-pentanediol and 1,5-pentanediol is further difficult to be explained. The selectivity trends cannot be simply explained by active metals or acidity/basicity. Generally, the 1,2-pentanediol/1,5-pentanediol selectivity ratio was not largely changed by the reaction temperature. This means that the activation energy (Ea) is similar for the formations of 1,2- and 1,5-pentanediols. DFT studies using energy diagram are not useful to explain such selectivity difference. The selectivity seems to be rather controlled by steric effect. If the selectivity is simply controlled by the steric hinderance, the attack of hydrogen species at 5-position of furfuryl alcohol prevails over that at 2-position where hydroxymethyl group is bonded. For both types of mechanisms above (direct attack or partial hydrogenation followed by C-O dissociation), the C-O bond at 1,5-positions is dissociated, and 1,2-pentanediol will be the main product. Therefore, increasing the steric hindrance around the catalytic active site can be one of the promising approaches to improving the selectivity to 1,2-pentanediol. The formation of 1,5-pentanediol requires the directing effect, for example, the furfuryl alcohol adsorption with the -CH2OH group makes the access of active hydrogen species on catalyst easier to the positions close to the -CH2OH. One interesting study for the selectivity control was reported by using Pt-embedded Al2O3 (Pt@Al2O3) catalyst prepared by calcination of Al-based metal-organic framework (MOF) ligating Pt [72]. When the Pt@Al2O3 catalyst was used for furfural hydrogenolysis with NaBH4 as the reducing agent, 1,5-pentanediol (75% yield) and tetrahydrofurfuryl alcohol (25%) were obtained, and 1,2-pentanediol was not produced at all. Furfural hydrogenation to furfuryl alcohol proceeds readily with NaBH4 in the absence of catalysts, and further conversion of furfuryl alcohol requires catalysts. When the same catalyst was used for the hydrogenolysis of furfuryl alcohol with H2, much higher temperature (423 K with H2 vs. 318 K with NaBH4) was necessary. The main product was 1,2-pentanediol (42% yield), and 1,5-pentanediol was not produced at all with Pt@Al2O3 + H2 system. When NaBO2, which is a by-product of reduction reaction with NaBH4, was added to the system for Pt@Al2O3 + H2, 21% yield of 1,5-pentanediol was obtained along with 18% yield of 1,2-pentanediol. The authors proposed that the Pt surface modified with NaBO2 more strongly binds the most functionalized carbon atom in the furan ring (C2 of furfuryl alcohol), and then the C2-O1 bond in the furan ring of furfuryl alcohol is dissociated. Extensive modification of metal surface and low temperature reaction may give higher 1,5-pentanediol selectivity. The catalysts containing Co tend to show higher selectivity to 1,5-pentanediol, and this trend may be due to the modification of metallic Co species with oxidized CoOx induced by lower reducibility of Co than other active metals (Cu and noble metals).

2.1.4 Reaction mechanism of ring-opening hydrogenolysis of furfuryl alcohol assisted by acid

Cu-zeolite catalysts with strong acidity have been reported to give good activity in 1,5-pentanediol formation from furfuryl alcohol with a flow reactor. The discoverer group first proposed the reaction mechanism as shown in Fig. 11 [60]. The carbon atoms bonded to the oxygen atom in the furan ring are first protonated, and then the C(sp3)-O bond is dissociated. If the proton is added to the 5-position of the furan ring (route (1) in Fig. 11), a carbocation species with the positive charge at the 2-position is formed, and the ring opening finally gives 1,2-pentanediol. If the proton is added to the 2-position of the furan ring (route (2)), a carbocation species with the positive charge at the 5-position is formed, which is finally converted to 1,5-pentanediol. However, the mechanism has a clear weakness which the authors admitted: Carbocations with more bonded carbon chains at the charged carbon atom are generally much more stable (cf. relative stability of tertiary/secondary alkyl carbocations), and route (1) which finally produces 1,2-pentanediol seems to be energetically easier. In this respect, it is difficult to explain why 1,5-pentanediol was the major product over Cu-zeolite catalysts. In their more recent paper, the mechanism involving direct C-O bond cleavage by the Cu cluster species inside the micropores of zeolite was proposed based on DFT calculation [61]. However, the calculated activation barrier was unacceptably high: 1.84 eV (177 kJ/mol) and 2.38 eV (229 kJ/mol) for formations of 1,5- and 1,2-pentanediol, respectively. After all, there are no satisfying data to determine the mechanism of ring-opening hydrogenolysis over Cu-zeolite catalysts.

Fig. 11
figure 11

Proposed mechanism of pentanediol formation from furfuryl alcohol over Cu-zeolite catalysts [60]. Reprinted with permission. Copyright (2022) Royal Society of Chemistry

Full size image

To discuss the inherent nature of furfuryl alcohol in acidic conditions, here, the protonated forms of free furfuryl alcohol are calculated by DFT, and the structures are summarized in Fig. 12. As expected, the protonation at the 5-position of the furan ring, which can lead to the formation of 1,2-pentanediol, is the most energetically feasible. The protonation at the 2-position, which is connected to the formation of 1,5-pentanediol, is calculated to be less stable by 28 kJ/mol. The trend is opposite to the experimentally observed 1,5-pentanediol/1,2-pentanediol selectivity ratio. Another important point is that the protonation at the -CH2OH group spontaneously breaks the C-O bond to release H2O, leaving carbocation species. The energy level of this protonation-dehydration is comparable to protonation at the 5-position. The carbocation species can be an intermediate of 2-methylfuran formation, rearrangement reaction (Piancatelli rearrangement) to cyclopentane ring, and polymerization reaction to humin [73, 74]. These are side reactions that should be suppressed; indeed, it is well known that furfuryl alcohol reduction in acidic media tends to involve humin formation to give low yields of target products [75]. The good yield of pentanediols over Cu-zeolite catalysts may be related to the utilization of gas-phase flow systems.

Fig. 12
figure 12

Calculated energy change in protonation of furfuryl alcohol and 2,3-dihydrofurfuryl alcohol. Calculation level: B3LYP/6-311 + + G(d,p), Gaussian09 program package. The relative energy (ΔE) is shown in kJ/mol

Full size image

The mechanism involving partial hydrogenation which is discussed in the previous section cannot be excluded. If the C2 = C3 double bond in furfuryl alcohol is hydrogenated, the O1-C2 single bond becomes reactive (Fig. 9). C-O hydrogenolysis reactions of saturated alcohols and ethers are possible by acid + metal bifunctional catalysis. The protonated forms of 2,3-dihydrofurfuryl alcohol are also calculated by DFT and summarized in Fig. 12. The protonation at the 4-position is by far the most energetically preferred, and even more than the protonation of furfuryl alcohol. The O1-C2 bond is much weakened by this protonation and will be cleaved by the addition of hydrogen species to finally give 1,5-pentanediol. Different from the case of furfuryl alcohol, the protonation at the -CH2OH group of 2,3-dihydrofurfuryl alcohol is not energetically feasible, and the proton is moved to the O1 (ether) atom. The protonation at the O1 atom is still much less energetically feasible (by 54 kJ/mol) than that at the C4 atom. Therefore, if the partial hydrogenation of furfuryl alcohol at the C2 = C3 double bond occurs, the acid-metal bifunctional catalysts give 1,5-pentanediol.

2.2 Ring opening of furancarboxylic acids

2-Furancarboxylic acid (FCA) and 2,5-furandicarboxylic acid (FDCA) can be produced by the oxidation of furfural and HMF, respectively. In particular, FDCA is a monomer of poly(ethylene 2,5-furandicarboxylate) resin, which is also called poly(ethylene furanoate) or PEF, and an alternative to poly(ethylene terephthalate) (PET). The production of FDCA has been very intensively investigated [16, 17]. The large-scale production of FDCA that is expected to be realized in near future makes FDCA as one of platform chemicals for further conversions [10]. For the reductive conversions of FCA or FDCA, it is important to keep the carboxylic groups in the substrates because the products of reduction of carboxylic groups can be directly produced from furfural or HMF. Ring-opening hydrogenolysis with the retention of the carboxylic acid moiety(ies) is an important chemical conversion of FCA and FDCA. The potential ring-opening products from FCA are 2-hydroxyvaleric acid, 5-hydroxyvaleric acid and valeric acid (Fig. 13). 5-Hydroxyvaleric acid (route (2)) can be present in the lactone form, δ-valerolactone. In alcohol solvents, the products can be converted to their corresponding esters. At present, there are no systems producing 2-hydroxyvaleic acid from FCA (route (1)) in good yield. The potential ring-opening products from FDCA are 2-hydroxyadipic acid, adipic acid and their lactone/ester derivatives.

Fig. 13
figure 13

Ring-opening hydrogenolysis products of furancarboxylic acids

Full size image

Ring-opening hydrogenolysis of FCA is discussed first. This reaction is typically catalyzed by Pt. Other noble metals as catalytic active centers promote ring hydrogenation instead. The reaction of FCA in methanol solvent using Pt/Al2O3 gave 55% yield of methyl 5-hydroxyvalerate together with 7% yield of δ-valerolactone and 20% of methyl tetrahydrofuran-2-carboxylate at 373 K [76]. Al2O3 was a better support than other conventional ones such as C, SiO2, TiO2 and ZrO2 in views of both activity and selectivity. Acetic acid is an effective solvent in the synthesis of δ-valerolactone [76, 77], although the reaction rate was lower than that in methanol. When Pt/SiO2 was used at lower temperature (313 K) in water solvent, 78% yield of 5-hydroxyvaleric acid was obtained together with 20% yield of tetrahydrofuran-2-carboxylic acid [78]. In this low-temperature condition, Pt/SiO2 showed lower reaction rate (TOFsurface) and higher ratio of ring opening to ring hydrogenation than Pt/TiO2 and ZrO2, although the difference was not large. When the same system (Pt/SiO2, 313 K, water solvent) was applied to furfuryl alcohol reduction, 1,2-pentanediol, 1,5-pentanediol and tetrahydrofurfuryl alcohol were formed in almost 1:1:1 ratio. Tetrahydrofuran-2-carboxylic acid is much less reactive over both Pt/Al2O3 and Pt/SiO2, which is a similar behavior to ring opening of furfuryl alcohol and tetrahydrofurfuryl alcohol. Reactivity of furan toward ring opening is comparable to FCA, while furan-3-carboxylic acid is much less reactive.

A series of Pt catalysts were also applied to ring-opening of FDCA [79]. Pt/Nb2O5⋅xH2O was effective to the catalytic conversion of FDCA to adipic acid; however, the highest yield of adipic acid was 38% with low carbon balance of 46%, which was obtained at 473 K in water. When Pt/H-ZSM-5 was used, the yield of adipic acid was 7%, and high yield of 2-hydroxyadipic acid (73%) was obtained with 91% carbon balance instead. Al2O3, TiO2, SiO2 and ZrO2-supported Pt catalysts showed similarly high yield of 2-hydroxyadipic acid (59–69%). The support effect is not so drastic, and the role of Nb2O5⋅xH2O seems to be the promotion of 2-hydroxyadipic acid hydrogenolysis to adipic acid as a solid acid that can work in water.

While Pt/Nb2O5⋅xH2O was active in stepwise C-O hydrogenolysis of FDCA to adipic acid, Pt-MoOx/TiO2 catalyst was reported to be active in one-step removal of oxygen atom of furan ring in FDCA or FCA [80, 81]. The activity strongly depended on the amount of Mo, which was optimized to be 0.5 wt% for wide range of Pt loadings (Pt 4–20 wt%). When the amount of Mo was fixed to be 0.5 wt%, the formation rate of valeric acid from FCA was increased with increasing the Pt loading amount, while the final valeric acid yield was almost independent of the Pt loading amount (around 60%). This behavior can be explained by the catalyst structure: Mo species are located on the surface of TiO2 support as sub-monolayer oxide. Some of Mo species are located between TiO2 and Pt metal surfaces, and the Pt-Mo bimetallic site works as the active site. The performance of Pt-MoOx/TiO2 catalyst for adipic acid synthesis from FDCA was not good, only 21–22% yield was obtained [79, 80]. Low solubility of FDCA in water solvent can affect the low performance. In addition, unlike simple catalysts such as Pt/SiO2, the stability of Pt-MoOx/TiO2 catalyst was not good, and deactivation was observed by reuses even when mild regeneration was applied.

For the reaction mechanism, we proposed a variant of the mechanism composed of partial hydrogenation, ring opening by intramolecular elimination and hydrogenation (Fig. 14) [81]. The C2 = C3 double bond is first hydrogenated, and the regioselectivity is directed by the strong adsorption of FCA with the carboxylic acid group. The C2-O1 bond is dissociated to restore the conjugation of C = C and C = O bonds. When Mo species is present, the oxophilic MoIV species extracts oxygen atom from the intermediate, and valeric acid is formed instead of 5-hydroxyvaleric acid. On the other hand, Sun et al. [78] proposed another mechanism where the first step is the addition of one hydrogen atom to the C4 atom of strongly bound FCA molecule on Pt surface with the furan ring. The intermediates have many C-Pt bonds, and the increase of the number of C-Pt bonds by dissociation of C-O bond is the driving force of the ring-opening. The reaction pathway of O1-C2 bond dissociation was compared with ring hydrogenation. However, the carboxylic acid group was not interacted with the Pt surface in this calculation result. It is difficult to explain the much different reactivity between FCA (2-furancarboxylic acid) and 3-furancarboxylic acid. It is also difficult to explain the comparable reactivity of FCA, furfuryl alcohol and furan but different regioselectivity between FCA and furfuryl alcohol.

Fig. 14
figure 14

Proposed mechanism of FCA reduction over Pt-MoOx/TiO2 catalyst [81]. Reprinted with permission. Copyright (2019) American Chemical Society

Full size image

3 Ring opening of saturated cyclic ethers

3.1 Tetrahydrofuran and tetrahydropyran derivatives with -CH2OH group: M-M’Ox type catalysts (M: noble metal; M’: group 5–7 metal)

One family of the biomass-derived saturated cyclic ethers is the derivatives of tetrahydrofuran and tetrahydropyran with -CH2OH group at the 2-position, such as tetrahydrofurfuryl alcohol, tetrahydrofuran-2,5-dimethanol and tetrahydropyran-2-methanol. Tetrahydrofurfuryl alcohol is the most popular compound in this category. Selective C-O hydrogenolysis of tetrahydrofurfuryl alcohol to 1,5-pentanediol was first reported in 2009 by using Rh-ReOx/SiO2 catalyst [82], which triggered the research on the catalyst development for C-O hydrogenolysis of cyclic ethers. The Rh-ReOx/SiO2 catalyst was also found to be active in C-O hydrogenolysis of glycerol to propanediols, which is also an important reaction in biomass upgrading, almost at the same time [83]. The C-O hydrogenolysis reactions of saturated cyclic ethers and glycerol are closely related; insights obtained in the catalyst development and mechanistic studies for hydrogenolysis of glycerol are useful for hydrogenolysis of saturated cyclic ethers [66, 84].

The insights obtained by earlier studies are summarized first. Typical catalysts for hydrogenolysis of saturated cyclic ethers are M-M’Ox (M = Rh. Ir, Pt; M’=Re, Mo, W, V) supported on relatively inert support such as SiO2 and C [66]. Weaker interaction between support and loaded species is advantageous to construction of interface site between different loaded species, M and M’Ox. The particle size of supported metal was typically 2–3 nm. Low surface concentration of loaded species (i.e. small ratio of loading amount to surface area) may not be necessary; rather, there are recent reports showing that high loadings rather lead to higher activity by promoting the interaction between M and M’Ox [85, 86]. Water solvent has been almost always used. In contrast to the direct ring opening of furfuryl alcohol, the hydrogenolysis of tetrahydrofurfuryl alcohol almost always produces 1,5-pentanediol, and the formation of 1,2-pentanediol is usually negligible. Earlier studies tended to use Rh as M. For example, the Rh-ReOx/C (Rh 4 wt%, Re/Rh = 0.25) catalyst reported by us gave high yield of 1,5-pentanediol (94%) and 1,6-hexanediol (84%) from tetrahydrofurfuryl alcohol and tetrahydropyran-2-methanol, respectively, at 373 K [87]. This 1,5-pentanediol yield is still the highest one among the reported systems of tetrahydrofurfuryl alcohol hydrogenolysis. Rh-MoOx/C catalyst was also reported to be effective in this kind of reaction [88, 89]. The structures of Rh-ReOx and Rh-MoOx catalysts have been well investigated, and the low-valent ReOx clusters or monomeric MoIVOx species are located on the surface of Rh metal via direct Rh-M’ bonds [90,91,92]. Rh-ReOx/SiO2 was also applied to the C-O hydrogenolysis of tetrahydrofuran-2,5-dimethanol to 1,6-hexanediol by the combination with Nafion SAC-13 to obtain 86% yield [15]. Other reported supports for Rh-M’Ox catalysts include TiO2 [93, 94], ZrO2 [95] and M’Ox itself [96, 97]. Recent works have tended to use Ir- or Pt-based catalysts because of the high cost of Rh and controllable structure of Ir-M’Ox and Pt-M’Ox species. These Ir- and Pt-based catalysts have been typically developed for C-O hydrogenolysis of glycerol where high selectivity is more difficult to be obtained. Some catalysts have been also utilized in derivatives of tetrahydrofuran and tetrahydropyran. The first report of Ir-based heterogeneous catalyst for C-O hydrogenolysis of tetrahydrofurfuryl alcohol was published by us [98]: Ir-ReOx/SiO2 catalyst, which was originally developed for glycerol hydrogenolysis to 1,3-propanediol [99], giving 82% yield of 1,5-pentanediol at 373 K. Although the activity based on the weight of active metal was lower, turnover frequency based on the number of active surface metal atoms determined by CO adsorption (TOFCO) was higher than Rh-ReOx/SiO2 under the same conditions. The structure of Ir-ReOx catalysts has been characterized well: low-valent Re cluster species with OH group(s) are formed on metallic Ir surface via direct Ir-Re bonds [100, 101]. Later, Wang et al. [102] reported the hydrogenolysis of tetrahydrofurfuryl alcohol with a fixed-bed reactor and Ir-MoOx/SiO2 catalyst, and the same group further reported similar system using Ir-VOx/SiO2 catalyst [103]. The obtained yields of 1,5-pentanediol were about 50%.

The fixed-bed reactor system is easily extended to consecutive reactors. It was demonstrated that double layered catalysts of Pd/SiO2 + Ir-ReOx/SiO2 in a fixed-bed reactor were effective to the conversion of HMF to 1,6-hexanediol [104]. Pd/SiO2 worked in the hydrogenation of HMF to tetrahydrofuran-2,5-dimethanol, and Ir-ReOx-/SiO2 was used for the subsequent C-O hydrogenolysis to 1,2,6-hexanetriol and furthermore to 1,6-hexanediol. Under optimal conditions of 373 K, 7 MPa H2 and mixture solvent of water + THF (40:60), 58% yield of 1,6-hexanediol was obtained together with 1,5-hexanediol (25%), hexanol (8%) and hexane (8%). The deactivation was observed at longer reaction time, which can be due to the leaching of Re species.

Wang et al. [105] reported a systematic survey of various bimetallic catalysts prepared by atomic layer deposition in tetrahydrofurfuryl alcohol hydrogenolysis. The results are shown in Figs. 15 (conversion) and 16 (selectivity). Pt-WOx/C, Pt-MoOx/C, Ir-WOx/C, and Ir-MoOx/C showed high selectivity to 1,5-pentanediol. Considering the activity in Fig. 15, Pt-WOx/C can be most promising. The combination of Pt and W has been very intensively investigated as catalysts for glycerol hydrogenolysis to 1,3-propanediol.

Fig. 15
figure 15

Conversion of tetrahydrofurfuryl alcohol over M1-M2Ox/C (M1: 10 wt%; M2/M1 = 1/2) [105]. a Pt-based catalysts and b other catalysts. “*” includes Ce, Cr, Mn, Y, Zn, Al and Co as M2. Reaction conditions: 1 wt% tetrahydrofurfuryl alcohol aqueous solution, space time 1.67 min g/mL, 3.6 MPa, 473 K. Reprinted with permission. Copyright (2018) Springer Nature

Full size image
Fig. 16
figure 16

Selectivity for tetrahydrofurfuryl alcohol hydrogenolysis in Fig. 15 [105]. PeD: pentanediol; Pathway I: 1,5-pentanediol, tetrahydropyran and δ-valerolactone; Pathway II: 2-methyltetrahydrofuran and 2-pentanol; Pathway III: tetrahydrofuran and 1-butanol; Pathway IV: 1,2-pentanediol. Reprinted with permission. Copyright (2018) Springer Nature

Full size image

There have been several reports of ring-opening C-O hydrogenolysis using Pt-WOx based catalysts. Pt/WOx/Al2O3 (Pt 1 wt%, WO3 7 wt%) catalyzed the ring-opening hydrogenolysis of tetrahydrofurfuryl alcohol and tetrahydropyran-2-methanol to 1,5-pentanediol and 1,6-hexanediol, respectively (Fig. 17) [106]. High selectivity was obtained, and high yields are expected although not demonstrated. The same work showed the reaction results of non-cyclic ethers (Fig. 18). It is suggested that the reaction proceeded parallelly in the two different routes (A: dissociation of C-O bond close to OH; B: dissociation of C-O bond at the opposite side of OH). This behavior on Pt/WO3/Al2O3 was rather different from that on Ir-ReOx/SiO2 [107] and Rh-ReOx/SiO2 [108]: they catalyzed the hydrogenolysis of 2-isopropoxyethanol to ethanol and 2-propanol with very high selectivity, meaning that only the route A proceeded. The difference between Pt/WOx/Al2O3 and M-ReOx/SiO2 (M = Ir, Rh) includes the reaction temperature and the molar ratio of metal oxide (M’Ox) to metal (M). The reaction rate of 2-isopropoxyethanol on Rh-ReOx/SiO2 was calculated to be 20 mmol g-cat−1 h−1 at 393 K, and that on Pt/WOx/Al2O3 was estimated to be 4 mmol g-cat−1 h−1 at 453 K. High catalytic activity of Rh-ReOx/SiO2 enabled low reaction temperature, which is connected to high selectivity. Another important point is that larger amount of metal oxide (WOx) gives larger amount of acid sites, which can be active sites for the side reactions.

Fig. 17
figure 17

Hydrogenolysis of cyclic ethers over Pt/WO3/Al2O3 (Pt 1 wt%, W 7 wt%) catalyst [106]. Red: 1,2-diol; blue: 1,n-diol; yellow: 1-ol; green : 2-ol. Conditions: Catalyst (50 mg), substrate (3 mmol), H2O (9 mL), H2 (5 MPa), 453 K, 5 h (a: 15 h). Reprinted with permission. Copyright (2020) Elsevier

Full size image
Fig. 18
figure 18

Hydrogenolysis of various ethers over Pt/WO3/Al2O3 catalyst [106]. Red: A1; blue: A2; yellow: B2; ×: A/B ratio. Conditions: Catalyst (50 mg), substrate (3 mmol), H2O (9 mL), H2 (5 MPa), 453 K, 5 h. Reprinted with permission. Copyright (2020) Elsevier

Full size image

Several Pt-W catalysts have been used in hydrogenolysis of tetrahydrofurfuryl alcohol [105, 109,110,111,112]. Among them, SiO2-supported catalysts showed good performance at lower temperature. We used Pt-WOx/SiO2 (Pt 4 wt%, W/Pt = 0.25) catalyst, which was originally developed for the hydrogenolysis of 1,4-anhydroerythritol and glycerol [113], in hydrogenolysis of tetrahydrofurfuryl alcohol (Fig. 19) [112]. High yield of 1,5-pentanediol (89%) was obtained, and the average reaction rate was estimated to be 1.4 mmol g-cat−1 h−1 at 413 K. The yield value was higher than those obtained by other Pt-W catalysts, and the reaction rate was also in the highest level in the reported Pt-W catalysts, considering the lower reaction temperature. An important point is that the optimum W amount in SiO2-supported Pt-W catalysts is lower than that in Al2O3-supported ones. Excess W species showing unnecessary acidity can be negligible in Pt-WOx/SiO2 catalyst. Ring-opening hydrogenolysis of tetrahydrofuran-2,5-dimethanol, an HMF-derived substrate related to tetrahydrofurfuryl alcohol, has been less investigated. He et al. [93] reported the ring-opening hydrogenolysis over Pt-WOx/TiO2 catalysts. 1,6-Hexanediol was obtained over Pt-WOx/TiO2 (Pt 10 wt%, W 10 wt%) catalyst with 72% yield and an average formation rate of 2.7 mmol g-cat−1 h−1 at 433 K. The yield did not reach that obtained by Rh-ReOx/SiO2 + Nafion SAC-13 (86%) described above [15], while the reaction rate was not so different (Rh-ReOx/SiO2 + Nafion SAC-13: average formation rate 1.4 mmol g-cat−1 h−1 at 393 K). Stability can be the advantage of the Pt-W-based catalyst system.

Fig. 19
figure 19

Hydrogenolysis tetrahydrofurfuryl alcohol over Pt-WOx/SiO2 (Pt 4 wt%, W/Pt = 0.25) catalyst [112]. PeD: pentanediol; PeOH: pentanol. Reaction conditions: tetrahydrofurfuryl alcohol 0.5 g, H2O 4 g, catalyst 0.2 g, H2 8 MPa, 413 K

Full size image

The Pt-WOx/SiO2 (Pt 4 wt%, W/Pt = 0.25) catalyst is an excellent one. Although the 1,5-pentanediol yield was slightly lower and the reaction temperature was higher than those in the optimal Rh-based catalysts, the Pt-WOx/SiO2 catalyst was stable [113] while Rh- and Ir-based catalysts are usually gradually deactivated. The Pt-WOx/SiO2 catalyst can be used for selective C-O hydrogenolysis reactions of many substrates including linear polyols and cyclic ethers containing OH groups such as 1,4-anhydroerythritol and 1,4-anhydroxylitol as well as tetrahydrofurfuryl alcohol. The hydrogenolysis of other cyclic ethers will be explained later. As mentioned above, the combination of Pt and W has been intensively investigated for the catalysts of glycerol hydrogenolysis of 1,3-propanediol. The performance of Pt-W catalysts (activity and 1,3-propanediol selectivity) has been much different among reports, even for catalysts with similar supports and compositions. Although the 1,3-propanediol yield over Pt-WOx/SiO2 catalyst from glycerol is not the highest (maximum yield 57% [113]); highest reported yield: 75% over Pt/Nb14W3O44 catalyst [114], this catalyst is simple and practical, using a commercial conventional SiO2 support and simple preparation method (sequential impregnation). Other catalysts showing good yield of 1,3-propanediol used specially designed supports prepared by researchers themselves. Here, we comment the reaction mechanism of C-O hydrogenolysis over our Pt-WOx/SiO2 catalyst. Many papers have claimed that the Brønsted acidity is strongly correlated to the production of 1,3-propanediol from glycerol over Pt-W catalysts, and the dissociation of C-O bond proceeds at the Brønsted acid sites [115,116,117]. However, the mechanism should not be so simple. It is strange that only W species is effective among various Brønsted acids. Pt is by far the most effective noble metal for the bimetallic catalysts containing W for selective hydrogenolysis, and this fact is difficult to be explained if W species is the active site for C-O bond dissociation.

The reaction mechanism of C-O hydrogenolysis over Pt-WOx/SiO2 catalyst was investigated from the kinetic viewpoint [112]. The kinetics depends on the substrate. The Pt-WOx/SiO2 catalyst preferably dissociates the C-O bond at the O-C(CH2OH) structure in linear polyols and tetrahydrofurfuryl alcohol (type-1). Other types of C-O bond can be dissociated if the substrate does not have the O-C(CH2OH) structure. The latter type of reaction (type-2) should be suppressed to obtain high yield of 1,5-pentanediol and 1,3-propanediol from tetrahydrofurfuryl alcohol and glycerol, respectively. The kinetics are different between type-1 and type-2. The reaction order with respect to substrate concentration (Fig. 20A) was almost zero for type-1 substrates (1,2-propanediol, glycerol, tetrahydrofurfuryl alcohol and 1,4-anhydroerythritol) and one for type-2 substrates (1,3-butanediol, 1,4-butanediol and 1,4-pentanediol). The reaction order with respect to hydrogen pressure (Fig. 20B) was little positive for type-1 substrates and almost zero for type-2 substrates. The difference in kinetics suggests different mechanisms between these two types of substrates. The zeroth reaction order with respect to hydrogen pressure is explained by the reaction mechanism of simple dehydration + hydrogenation catalyzed by acid and metal, respectively, where acid-catalyzed dehydration is the rate-determining step. The first-order kinetics on substrate concentration for type-2 substrates means that the adsorption of substrate on the acid site is weak. On the other hand, positive and zeroth reaction order with respect to hydrogen pressure and substrate, respectively, in the case of type 1 substrates is more reasonably explained by direct hydrogenolysis mechanism where hydride species attacks the strongly-bound substrate on the catalyst. The direct hydrogenolysis mechanism with hydride as active species has been proposed as shown in Fig. 21 for other M-M’Ox catalysts, especially those with M’=Re and M = Rh or Ir [66, 118]. The typical reaction order with respect to hydrogen pressure was one for direct hydrogenolysis mechanism over M-M’Ox catalysts. Lower reaction order in the case of Pt-WOx/SiO2 may be due to the involvement of dehydration + hydrogenation mechanism to small extent even in the type-1 substrates. The active site is the interface between noble metal (M) and oxide of M’.

Fig. 20
figure 20

Double logarithmic plots for the kinetics of the conversion of various aqueous alcohols over Pt-WOx/SiO2 (Pt 4 wt%, W/Pt = 0.25) at 413 K. A Conversion rate vs. substrate concentration; B conversion rate vs. hydrogen pressure [112]. PrD: propanediol; THFA: tetrahydrofurfuryl alcohol; AHERY: anhydroerythritol; BuD: butanediol; PeD: pentanediol; BuOH: butanol; PeOH: pentanol

Full size image
Fig. 21
figure 21

Direct C-O hydrogenolysis mechanism over M-M’Ox (M = Rh, Ir, Pt; M’ = Re, Mo, W) catalysts for type-1 substrates in Fig. 20

Full size image

There are several studies on the investigation of C-O hydrogenolysis mechanism of Pt-W catalysts based on reaction kinetics; however, they used less selective catalysts. The possible involvement of an unselective reaction route with different kinetics complicates the overall kinetics, hindering the precise discussion of the mechanism of the main reaction route. It is essential to use catalysts with high selectivity to clarify the reaction mechanism. We further note the role of acidity. The correlation between the catalyst acidity and activity in C-O hydrogenolysis has been well documented [71, 88, 95]; however, it does not always mean that the reaction mechanism is acid-mediated one. The active sites in direct C-O hydrogenolysis have acidity, which can lead to spurious correlation between activity and acidity.

3.1.1 Ring-opening of tetrahydrofurfuryl alcohol over Ni catalysts

Bimetallic catalysts containing noble metals have been used in the studies mentioned above for tetrahydrofurfuryl alcohol hydrogenolysis to 1,5-pentanediol. Non-noble metal catalysts for this reaction have been attempted to be developed, and typical catalysts are Ni-based ones. High yield of 1,5-pentanediol using Ni-based catalysts was first reported by Wijaya et al. [119]. The catalyst was Ni-La/HT (HT: hydrotalcite; Ni/Y = 2.5) prepared by co-precipitation, hydrothermal treatment and hydrogen treatment. The yield of 1,5-pentanediol reached 92% from tetrahydrofurfuryl alcohol in a 2-propanol solvent at 423 K. Use of furfural as a substrate is possible (56% yield of 1,5-pentanediol) because Ni catalyst has good activity in total hydrogenation of furfural to tetrahydrofurfuryl alcohol [120]. This is in contrast to the case of M-M’Ox catalysts where the activity in ring hydrogenation is low and two step reaction is necessary for the direct conversion of furfural to 1,5-pentanediol [104, 121, 122]. The same research group also reported Ni-Y2O3 and Ru-modified Ni-Y2O3 catalysts, although the 1,5-pentanediol yield was not as high as the case of Ni-La/HT catalyst [123, 124]. According to the structure-performance relationship for Ni-Y2O3 catalysts, the catalysts with low crystallinity (Ni0 crystallite size < 5 nm) showed the best activity [123].

It has been recently reported that Ni-La(OH)3 (Ni 40 wt%, La/Ni = 2.5) catalyst gave 73% yield of 1,5-pentanediol using a 2-propanol solvent without H2 feeding in the continuous flow fixed bed reactor at 413 K [125]. This report also included the solvent effect using the batch reactor at 423 K (Fig. 22). It is interesting that this Ni catalyst cannot catalyze the reaction of tetrahydrofurfuryl alcohol to 1,5-pentanediol in the water solvent. This is in contrast to the noble-metal-containing bimetallic catalysts, where water is a good solvent for the selective C-O hydrogenolysis. The ring-opening hydrogenolysis over the Ni-La(OH)3 catalyst without H2 occurred only in secondary alcohol solvents: neither primary alcohols nor 1,4-dioxane worked. On the other hand, under H2, the hydrogenolysis proceeded in secondary alcohols, primary alcohols and 1,4-dioxane. These results indicate that both secondary alcohols and H2 can supply active hydrogen species for the hydrogenolysis reaction. The reaction in the batch reactor under pressurized H2 can give about 90% yield of 1,5-pentanediol in 2-propanol solvent, by extending the reaction time to 24 h [125, 126]. Similar Ni-La(OH)3 catalysts were also investigated by another research group, and the 1,5-pentanediol yield reached 91.7% (2-propanol solvent, 4 MPa H2, 443 K) [127]. The combination of Ni with Pr instead of La has been also reported, although the obtained 1,5-pentanediol yield (70%) was lower than that over Ni-La catalysts [128].

Fig. 22
figure 22

Effect of solvent on hydrogenolysis (a) or transfer hydrogenolysis (b) of tetrahydrofurfuryl alcohol over Ni-La(OH)3 catalyst [125]. Reaction conditions: 2.5 wt% substrate solution 5 mL, catalyst (Ni 40 wt%, La/Ni = 2.5) 0.1 g, H2 (a) or Ar (b) 3 MPa, 423 K, 15 h. PrOH: propanol; BuOH: butanol; PeOH pentanol. Reprinted with permission. Copyright (2022) American Chemical Society

Full size image

The proposed mechanism for Ni-La(OH)3 catalyst is shown in Fig. 23 [125]. Hydride species formed on Ni from secondary alcohols or H2 are active species. The substrate molecule is adsorbed as alkoxide on the base sites. The C-O bond neighboring to the -CH2OH group is attacked by the hydride species. The attack by hydride species was also suggested by more recent study using D2 instead of H2 [126]. The mechanism is similar to that proposed for M-M’Ox catalysts [66]: substrate is adsorbed at the -CH2OH group as alkoxide on the M’Ox site, and H2 is dissociated to proton and hydride on M’Ox-modified M surface. The hydride attacks the C-O bond neighboring to the -CH2OH group, which has been also proposed for M’Ox-modified M surface (Fig. 21).

Fig. 23
figure 23

Proposed mechanism for hydrogenolysis of tetrahydrofurfuryl alcohol over Ni-base catalysts [125]. Reprinted with permission. Copyright (2022) American Chemical Society

Full size image

Other reaction mechanisms have been also proposed for the Ni-based catalysts. The solvent effect in the reaction of tetrahydrofurfuryl alcohol over Ni-WOx/SiO2 (Ni 10 wt%, W 30 wt%) is shown in Fig. 24 [129]. This catalyst contains large amount of W species and non-basic support (SiO2), and therefore, this catalyst is rather acidic. Significant amount of 1,2,5-pentanetriol was detected at lower conversion level when water was used as a solvent, which is connected to the reaction pathway from tetrahydrofurfuryl alcohol to 1,5-pentanediol via 1,2,5-pentanetriol (Fig. 25). In contrast, when 1,4-dioxane was used as a solvent, tetrahydropyran was formed, suggesting that tetrahydrofurfuryl alcohol dehydration to dihydropyran proceeded (Fig. 26). In these two cases, the reaction mechanism is different from the direct ring-opening C-O hydrogenolysis. A similar mechanism was also proposed for the reaction of tetrahydrofurfuryl alcohol to tetrahydropyran over zeolite-supported Ni phyllosilicate catalysts [130].

Fig. 24
figure 24

C-O bond hydrogenolysis of tetrahydrofurfuryl alcohol over Ni-WOx/SiO2 (Ni 10 wt%, W 30 wt%) catalysts [129]. a Effect of catalyst loading in water, b effect of reaction time in water, and c Effect of reaction time in 1,4-dioxane. PDO: pentanediol; PTO: pentanetriol; THP: tetrahydropyran. Reaction conditions: 523 K, 3.4 MPa H2, 5 wt % tetrahydrofurfuryl alcohol in water or 1,4-dioxane (60 g). Reprinted with permission. Copyright (2018) John Wiley and Sons

Full size image
Fig. 25
figure 25

Reaction pathway for C-O hydrogenolysis of tetrahydrofurfuryl alcohol over Ni-WOx/SiO2 under aqueous-phase conditions at 523 K [129]. Reprinted with permission. Copyright (2018) John Wiley and Sons

Full size image
Fig. 26
figure 26

Reaction pathway for the C-O hydrogenolysis of tetrahydrofurfuryl alcohol over Ni-WOx/SiO2 using 1,4-dioxane as solvent [129]. Reprinted with permission. Copyright (2018) John Wiley & Sons

Full size image

3.1.2 Tetrahydrofuran derivatives with functional groups other than -CH2OH

Biomass-derived tetrahydrofuran derivatives without -CH2OH group on the ring include 3-hydroxytetrahydrofuran, 1,4-anhydroerythritol, 1,4-anhydroxylitol and isosorbide. The reports on ring-opening C-O hydrogenolysis of these compounds are much more limited than that of tetrahydrofurfuryl alcohol. The recently reported results of ring-opening hydrogenolysis with good selectivity are summarized in Fig. 27, and details of each system are explained below.

Fig. 27
figure 27

Recently-reported selective ring-opening hydrogenolysis of tetrahydrofuran derivatives except tetrahydrofurfuryl alcohol

Full size image

Ring-opening C-O hydrogenolysis of 3-hydroxytetrahydrofuran was reported to be catalyzed by Ir-ReOx/SiO2 (Ir: 4 wt%, Re/Ir = 2). The main product was 1,3-butanediol with 79% selectivity at 24% conversion of 3-hydroxytetrahydrofuran [98]. Ring-opening C-O hydrogenolysis of 3-hydroxytetrahydrofuran on Pt-WOx/SiO2 (Pt: 4 wt%, W/Pt = 0.25) gave 1,3-butanediol with 90% selectivity at 40% conversion [113]. The reaction rate of ring-opening C-O hydrogenolysis of 3-hydroxytetrahydrofuran to 1,3-butanediol over Ir-ReOx/SiO2 was clearly higher than that over Pt-WOx/SiO2 even at lower reaction temperature. The reaction of 3-hydroxytetrahydrofuran over Rh-MoOx/SiO2 (Rh: 4 wt%, Mo/Rh = 0.13) was not selective, although the reaction rate based on total active metal weight was clearly higher than those over Ir-ReOx/SiO2 and Pt-WOx/SiO2 [131]. These trends of activity and selectivity (activity: Rh-MoOx/SiO2 > Ir-ReOx/SiO2 > Pt-WOx/SiO2; regioselectivity at low conversion level: Ir-ReOx/SiO2 ≈ Pt-WOx/SiO2 > Rh-MoOx/SiO2) are similar to those in glycerol hydrogenolysis.

1,4-Anhydroerythtiol (cis-3,4-dihydroxytetrahydrofuran) is easier to be obtained from biomass than 3-hydroxytetrahydrofuan. The ring-opening C-O hydrogenolysis of 1,4-anhydroerythritol catalyzed by Rh-MoOx/SiO2 (Rh: 4 wt%, Mo/Rh = 0.13) is not so selective as shown in Fig. 28(a) [131]. A variety of products were observed in the wide conversion level, meaning that it is not easy to achieve high yield of the specific target product. One of the products obtained in a high yield is 2-butanol (51%). M-M’Ox/SiO2 catalysts generally have low activity in C-O hydrogenolysis of mono-alcohols. This is because alcohols can be adsorbed on the M’Ox sites as alkoxides; however, the formation of alkoxide rather strengthens the C-O bond and the alkoxide itself is not reactive in C-O hydrogenolysis. Rh-MoOx/SiO2 catalyst has relatively low regioselectivity in C-O hydrogenolysis among M-M’Ox catalysts. Both the OH groups in 1,4-anhydroerythritol are secondary ones. Combining these features enabled the selective formation of secondary mono-alcohol (2-butanol) from 1,4-anhydroerythritol over Rh-MoOx/SiO2 catalyst.

Fig. 28
figure 28

Time courses of 1,4-anhydroerythritol hydrogenolysis. a Rh-MoOx/SiO2 (Rh 4 wt%, Mo/Rh = 0.13) catalyst [131]. Reaction conditions: 1,4-anhydroerythritol 1 g, water 4 g, catalyst 0.1 g, H2 8 MPa, 393 K. Reprinted with permission. Copyright (2016) John Wiley and Sons. b Pt-WOx/SiO2 (Pt 4 wt%, W/Pt = 0.25) catalyst [113]. Reaction conditions: 1,4-anhydroerythritol 0.5 g, water 4 g, catalyst 0.2 g, H2 8 MPa, 413 K. BuT: butanetriol; BuD: butanediol; BuOH: butanol; HTHF: hydroxytetrahydrofuran. Reprinted with permission. Copyright (2020) Royal Society of Chemistry

Full size image

A recent progress is the ring-opening hydrogenolysis of 1,4-anhydroerythritol over Pt-WOx/SiO2 (Pt 4 wt%, W/Pt = 0.25) giving 54% yield of 1,3-butanediol (Fig. 28(b)) [113]. The selectivity to 1,2,3-butanetriol is high at low conversion level, and then 1,2,3-butanetriol is converted to 1,3-butanediol in a similar manner to glycerol hydrogenolysis to 1,3-propanediol (Fig. 29). In contrast, Rh-MoOx/SiO2 catalyst showed lower selectivity to 1,2,3-butanetriol and higher selectivities to 2,3-butanediol and 2-butanol at low conversion level. The Rh-MoOx/SiO2 catalyst can dissociate multiple C-O bonds except the C-O(Mo) bond in adsorbed alkoxide at once, while Pt-WOx/SiO2 catalyst dissociates the C-O bond neighboring to the C-O(W) bond step-by-step. The selectivity trends were also observed in glycerol hydrogenolysis: high initial 1-propanol selectivity (ca. 50%) over Rh-MoOx/SiO2 [132] and high initial 1,3-propanediol selectivity (ca. 60%) over Pt-WOx/SiO2 [113]. When erythritol is used as a substrate, the Pt-WOx/SiO2 catalyst produces good yield of 1,4-butanediol (51% yield) via successive hydrogenolysis of C-O bonds neighboring -CH2OH groups [133]. Other M-M’Ox catalysts (M: noble metal; M’: Mo, W, Re) are less selective in erythritol hydrogenolysis [134, 135]; however, selective double hydrodeoxygenation of erythritol to 1,2-butanediol is possible with ReOx-M/CeO2 (M = Pd, Ni) catalysts (≤ 77% yield) [136, 137]. Recently, synthesis of 2,3-butanediol has been reported with Ir-FeOx catalysts without or with further Mo modification (Ir-Fe-Mo catalyst) (≤ 36% yield) [138,139,140]. Various butanediols can be produced by C-O hydrogenolysis from erythritol or 1,4-anhydroerythritol.

Fig. 29
figure 29

Reaction routes of 1,4-anhydroerythritol over Pt-WOx/SiO2 catalyst [113]. Reprinted with permission. Copyright (2020) Royal Society of Chemistry

Full size image

The Pt-WOx/SiO2 (Pt 4 wt%, W/Pt = 0.25) catalyst was also used for ring-opening of 1,4-anhydroxylitol [133]. 1,2,3,5-Pentanetetraol was mainly formed in the initial stage (67% selectivity at 82% conversion). This reaction is the ring-opening C-O hydrogenolysis at the O-C(CH2OH) structure, which is a similar reaction to tetrahydrofurfuryl alcohol hydrogenolysis. 1,2,3,5-Pentanetetraol can be also formed directly from xylitol over the same catalyst; however, the reaction of 1,4-anhydroxylitol is more selective. Extending the reaction time gave mixture of pentanetriols, pentanediols and pentanols depending on the degree of reaction progress. The selectivity pattern from xylitol and 1,4-anhydroxylitol became similar when pentanetriols become the main products.

Hydrogenolysis of isosorbide has been less investigated, and there have been no report giving good yield of sole oxygenate product. Isosorbide is sometimes observed as an intermediate in the biofuel production from carbohydrates or sorbitol by total hydrodeoxygenation with acid + metal bifunctional catalysts, and therefore, isosorbide hydrodeoxygenation has been tested during the investigation of the reaction pathways in some reports [141,142,143,144,145]. Harsh reaction conditions have been applied, and C-C bond dissociation also occurred. Hexane selectivity of 65% among hydrocarbon products at the reaction progress level of 31% oxygenates remaining was reported with Pt/SiO2-Al2O3 catalyst [143]. Another report showed 1-hexanol yield of 25% over Ru-MoOx/Mo2C catalyst [141]. However, higher yields of hexane or 1-hexanol can be obtained from sorbitol instead. The use of isosorbide as substrate has little merit. Very recently, Rh/SiO2-catalyzed hydrogenolysis of isosorbide has been reported, giving diols and triols with each 29% carbon-based yield [146]. As the main components of products, 2,3-hexanediol and 1,6-hexanediol were detected with 7% and 5% yield, respectively. As a variant of isosorbide hydrogenolysis, hydrogenolysis of isosorbide diacetate has been reported [147]. 1,6-diacetoxyhexane was obtained with 22% yield by using Ru/C + La(OTf)3 catalysts (Eq. (1)). The acetate moieties in isosorbide diacetate were removed by hydrogenolysis, and those in the product were derived from the solvent. The synthesis requires hydrogenolysis of secondary-alcohol-derived C-O bonds in ethers and esters, while primary-alcohol-derived C-O bonds should not react. This catalyst system still has some activity in dissociation of primary-alcohol-derived C-O bonds, and the yield of target product was decreased at higher conversion level.

(1)

Furfurylamine is a primary amine compound derived from furfural. Hydrogenation of furfurylamine can produce tetrahydrofurfurylamine. The ring-opening C-O hydrogenolysis of furfurylamine and tetrahydrofurfurylamine has been investigated. Zhang et al. [11] investigated the furfurylamine hydrogenolysis over various noble metal catalysts. The carbon-supported noble metal catalysts and monometallic Pt catalysts on various supports were first screened in tetrahydrofuran solvent, and Pt/CeO2, Pt/ZrO2 and Pt/Zr0.6Ce0.4O2 showed relatively good yield of aminopentanols (1-amino-2-pentanol and 5-amino-1-pentanol). The highest yield of 5-amino-1-pentanol (52%) was obtained with Pt/ZrO2 catalyst in a THF + water mixed solvent (20:1 in volume) at 423 K. The main by-product was tetrahydrofurfurylamine, which was stable in this reaction condition. Based on these selectivity trends, this reaction is similar to the ring-opening C-O hydrogenolysis of furfuryl alcohol explained in section 2.1. The same research group has very recently reported the reduction of 2,5-bis(aminomethyl)furan which is an amine compound derived from HMF [148]. The primary product of ring-opening hydrogenolysis, 2-hydroxyhexamethylenediamine, was not obtained in high yield because of the cyclization to 2-aminomethylpiperidine. 2-Aminomethylpiperidine was obtained with 72% yield over Pt/γ-Al2O3 catalyst in a water solvent. As a similar reaction to tetrahydrofurfuryl alcohol hydrogenolysis, Hong et al. [149] reported the tetrahydrofurfurylamine hydrogenolysis. The catalyst, Rh-ReOx/SiO2, for the reaction of tetrahydrofurfurylamine is similar to that for the ring-opening C-O hydrogenolysis of tetrahydrofurfuryl alcohol. Piperidine was obtained in high yield (92%), in a water solvent in the presence of 1 eq. of HCl at 473 K. The primary hydrogenolysis product, 5-amino-1-pentanol, is readily cyclized to piperidine by intramolecular alcohol amination. The addition of appropriate amount of HCl to keep the reaction medium slightly acidic is essential to prevent deactivation of Rh-ReOx/SiO2 catalyst, since M-MO’x type catalysts typically lose the C-O hydrogenolysis activity in basic conditions [150]. Excess amount of HCl also much decreased the activity in tetrahydrofurfurylamine hydrogenolysis, probably because the protonation of amine substrate suppresses its adsorption on the catalyst surface. The proposed mechanism for the C-O hydrogenolysis over Rh-ReOx/SiO2 is shown in Fig. 30, which is similar to that of ring-opening hydrogenolysis of tetrahydrofurfuryl alcohol.

Fig. 30
figure 30

Proposed mechanism of hydrogenolysis of tetrahydrofurfurylamine over Rh-ReOx/SiO2 [149]. Reprinted with permission. Copyright (2023) Royal Society of Chemistry

Full size image

Very recently, Qi et al. [151] have reported direct conversion of furfural to piperidine with H2 and NH3 over single-atom-alloy-type Ru1Cox/HAP (x: Co/Ru molar ratio; HAP: hydroxyapatite) catalysts. The formation of single atom alloy (SAA) was confirmed by EXAFS analysis. Two-step reaction consists of low temperature (373 K) and high temperature (453 K) was carried out (Fig. 31). Furfural amination to furfurylamine occurs in the first low temperature step. In the high temperature step, furfurylamine is hydrogenated to tetrahydrofurfurylamine, and then ring-opening hydrogenolysis and intramolecular amination occur to give piperidine like the case of Rh-ReOx/SiO2 catalyst. The overall piperidine yield from furfural reached 91% over the x = 20 catalyst (Ru1Co20/HAP), and further higher yield (97%) was obtained with the x = 160 catalyst. The catalysts with higher Ru/Co ratio, which are not single-atom-alloy, did not show activity in ring-opening hydrogenolysis, and the main product was tetrahydrofurfurylamine. The examples of single-atom-alloy catalyst for C-O hydrogenolysis reactions have been limited [152], and further development is expected.

Fig. 31
figure 31

Reductive amination of furfural to piperidine over Ru1Co20/HAP catalyst. Conditions: catalyst (0.43 wt% Ru and 5 wt% Co) 50 mg, furfural 0.5 mmol, p-xylene 5 g, H2 1 MPa, NH3 0.5 MPa. Reprinted with Creative Commons Attribution 4.0 International License https://creativecommons.org/licenses/by/4.0/. The compound names that are different from the main text were removed from this image

Full size image

Ring-opening of tetrahydrofuran-2-carboxylic acid has been also attempted with M-M’Ox/SiO2 catalysts (M = Rh, Pd, Pt, Ru, Ir; M’=W, Re, Mo). However, all these catalysts did not show good selectivity in ring-opening hydrogenolysis because the active catalysts in ring-opening also have activity in hydrogenation of carboxylic acids [153]. As hydrogenolysis systems of tetrahydrofurancarboxylic acids, Gilkey et al. [154,155,156] developed reduction systems of tetrahydrofuran-2,5-dicarboxylic acid using I− as a key component. Adipic acid was obtained in good yield by successive reduction of OH groups via nucleophilic substation with I− and reduction of organic iodide with H2.

4 Comparison of approaches to synthesis of each product

There are several approaches to synthesis of each target product, especially 1,5-pentanediol. Here, these approaches and other production methods from biomass are compared for each product. The most important product, 1,5-pentanediol, is discussed first. Table 6 summarizes the representative systems of ring-opening hydrogenolysis to 1,5-pentanediol. The reaction rates are comparable between openings of unsaturated ring (furfural and furfuryl alcohol) and saturated ring (tetrahydrofurfuryl alcohol). Noble metal catalysts tend to show higher activity than non-noble metal catalysts. Some non-noble metal catalysts for opening unsaturated ring such as Co-Cu-Al showed good activity, probably because of the large content of active metals in the catalyst. Systems for ring-opening hydrogenolysis of unsaturated ring have lower selectivity to 1,5-pentanediol than those for hydrogenolysis of tetrahydrofurfuryl alcohol, and the selectivity should be improved.

Table 6 Comparison of catalytic systems for synthesis of 1,5-pentanediol (1,5-PeD) from furfural derivatives
Full size table

Other methods to produce 1,5-pentanediol include the hydrogenation of glutaric acid (C5 dicarboxylic acid) and the indirect hydrogenolysis of tetrahydrofurfuryl alcohol via dehydration, hydration and hydrogenation. Hydrogenation of glutaric acid or its ester is the conventional process for 1,5-pentanediol production in petrochemistry [157], and the glutaric acid is obtained as a by-product of adipic acid in oxidation of K-A oil (cyclohexanone-cyclohexanol mixture). Sustainable synthesis of glutaric acid or K-A oil can give renewable 1,5-pentanediol simply replacing the substrates, but such synthesis is not easy. The indirect hydrogenolysis (Fig. 32) was developed over 70 years ago [158], and recently the feasibility has been re-examined [159, 160]. The recent techno-economic analysis [160] showed that the indirect process is superior to the direct hydrogenolysis catalyzed by Rh-ReOx/SiO2, although the indirect process is composed of many steps (Fig. 33). The raw feedstock cost (mainly the cost of tetrahydrofurfuryl alcohol) was estimated to be similar between the two processes because the yield of 1,5-pentanediol from tetrahydrofurfuryl alcohol was assumed to be similar. The price of tetrahydrofurfuryl alcohol was assumed to be the median value (1250 USD/ton) of the recent furfural price (500–2000 USD/ton) with addition of 300 USD/ton as the processing cost of total hydrogenation of furfural to tetrahydrofurfuryl alcohol. For the direct hydrogenolysis, large cost was calculated in catalyst costs, capital costs (depending on catalyst activity and reaction pressure) and utility & electricity costs (mainly depending on solvent type and amount). Higher activity of cheaper catalyst at lower pressure is necessary to commercialize the ring-opening hydrogenolysis system. In order to decrease the utility costs, the solvent amount should be lower, and water may not be the best solvent because of the large thermal capacity and heat of evaporation. The raw feedstock costs are also large in Fig. 33. The use of furfural instead of tetrahydrofurfuryl alcohol can decrease the raw feedstock costs.

Fig. 32
figure 32

Indirect hydrogenolysis of tetrahydrofurfuryl alcohol to 1,5-pentanediol

Full size image
Fig. 33
figure 33

Techno-economic analysis results of indirect hydrogenolysis (dehydration-hydration-hydrogenation (DHH) pathway) and direct hydrogenolysis pathway using Rh-ReOx/SiO2 catalyst [160]. MSP: minimum selling price. Abbreviations of compounds are the same as those in Fig. 32. Reprinted with permission. Copyright (2017) John Wiley and Sons

Full size image

For 1,2-pentanediol, the conventional production method is the dihydroxylation of 1-pentene via epoxidation and hydration. Renewable synthesis of 1-pentene or 1,2-pentanediol with other methods has not been established, but the synthesis by ring-opening hydrogenolysis of furfural or furfuryl alcohol is a promising means. There have been some studies showing the formation of 1,2-pentanediol in xylose hydrogenolysis [161, 162] however, the selectivity was low, and xylose is not a convenient substrate in comparison with furfural.

For ring-opening hydrogenolysis products of furancarboxylic acids, adipic acid is the most important compound in industry. There are many production routes from biomass to adipic acid, such as K-A oil from bio-oil, fermentative synthesis, hydrodeoxygenation of sugar acids, oxidative C-C cleavage reactions of disubstituted cyclohexanes, and so on. The comparison of possible routes to adipic acid and other ring-opening products have been already discussed in other review papers [10, 163]. A recent progress not covered by these review papers is the production of 2-hydroxyadipic acid by oxidative C-C cleave reaction of 2-hydroxycyclohexanone [164].

The feasibility of 1,6-hexanediol from tetrahydrofuran-2,5-dimethanol or isosorbide also depends on the progress of renewable adipic acid because 1,6-hexanediol is manufactured by the hydrogenation of adipic acid or its ester [165]. Nevertheless, adipic acid hydrogenation of 1,6-hexanediol consumes 4 equiv. of H2, and the production of renewable adipic acid also consumes a large amount of reducing agent during any synthesis methods. Meanwhile, synthesis of 1,6-hexanediol by hydrogenolysis of tetrahydrofuran-2,5-dimethanol or isosorbide consumes 5 equiv. of H2 during the overall process from carbohydrate and has a merit of smaller requirement of reducing agent.

For the ring-opening products of 1,4-anhydroerythritol, there are other methods to produce the target compound from biomass such as 1,3-butanediol from acetaldehyde (aldol reaction and hydrogenation) [166], direct fermentative production of 1,3-butanediol [167] and hydrogenative decarboxylation of levulinic acid to 2-butanol [168]. The feasibility much depends on the cost of erythritol. While erythritol has been commercially manufactured as a sweetener, it is not still regarded as an industrial chemical. On the other hand, erythritol and 1,4-anhydroerythritol can be versatile platform compounds for various oxygenated chemicals such as butadiene, 1,4-butanediol and 1,2-butanediol [2, 20]. Development of efficient conversion systems of erythritol to other useful compounds such as butadiene will decrease the cost of erythritol.

5 Utilization of 1,5-pentanediol as monomer of plastics

In the cutting-edge researches mentioned in the previous sections, 1,5-pentanediol is the most readily obtained product in ring-opening C-O hydrogenolysis of biomass-derived substrates. 1,5-Pentanediol has two functional groups on both ends of the carbon chain, and therefore it can be used as a monomer. However, the traditional petroleum chemistry does not use 1,5-pentanediol largely, while α,ω-diols with even carbon number (ethylene glycol, 1,4-butanediol and 1,6-hexanediol) have been major monomers. Here, we summarize the properties of 1,5-pentanediol-based polymers and their recent developments to understand the potential of 1,5-pentanediol production from biomass.

5.1 1,5-Pentanediol as monomer of polyester

Polyesters are the most widely manufactured family of polymers using diols as monomers. The combination of ethylene glycol and terephthalic acid is the most common (poly(ethylene terephthalate); PET). First, the polyester of 1,5-pentanediol and terephthalic acid (poly(pentamethylene terephthalate); P5T) is considered (Fig. 34).

Fig. 34
figure 34

Structures of PET and P5T

Full size image

Synthesis of P5T is typically conducted by the conventional method for polyesters composed of two-step esterification and polycondensation of carboxylic acid or methyl carboxylate with diol using an appropriate catalyst [169,170,171]. Oligomers are formed in the first step and polyester is formed by the successive transesterification in the second step under high temperature and vacuum conditions to remove unreacted monomers and shift the equilibrium toward the product side. By using this method, α,ω-diols such as 1,5-pentanediol give the corresponding polyester with high yield. In contrast, vicinal diols such as 1,2-pentanediol also give acetal, which decreases the polyester yield. Xiao et al. [172] reported the synthesis of PET-co-P(1,2-pentane)T, where a part of ethylene glycol is substituted to 1,2-pentanediol. However, the formed polyester contained a less amount of vicinal diol as expected due to the acetal formation by intermolecular dehydration of two molecules of α,ω-diol.

The properties of P5T are discussed next. P5T has been reported to be a colorless transparent resin, but it gradually became white to light yellow and translucent to opaque at room temperature, suggesting that P5T crystalized slowly at room temperature [173]. P5T is soluble in chloroform but insoluble in tetrahydrofuran, which is different for PET. Typical properties of PET and P5T are listed in Table 7 [173,174,175,176]. Obviously P5T is a flexible polyester compared with PET. The decreases in glass transition temperature (Tg) [177] and Young’s modulus [178] from PET to P5T are generally observed in the increase of methylene number in the repeating units of poly(alkylene terephthalate)s. In contrast, melting temperature (Tm) [178, 179] and crystallization temperature (Tc) [180] shows a zig-zag variation dependent on the methylene number in the repeating unit, which is known as the odd-even effect. P5T has lower Tm and Tc values than poly(tetramethylene terephthalate) (P4T or PBT) and poly(hexamethylene terephthalate) (P6T), which means that crystallization of P5T is more difficult than that of P4T and P6T. Two crystalline phases, named as a and b forms, of the oriented fiber of P5T have been discovered [181]. The unit cells of both a and b phases are triclinic, and the crystallographic repeat along the fiber axis is 24.7 Å and 28.2 Å, respectively, while the chemical repeat of P5T is 14.5 Å [182]. P5T annealed under unstressed tension at room temperature provides the a form with a contracted chain. The b form with nearly fully extended length is obtained by induced by tension, although the a form is always mixed in the relaxed state. Under the influence of stress, the a form is reversibly transformed into the b form. The crystallization of P5T is much slower than that of P4T and P6T, as reflected by the lower Tc [180].

Table 7 Comparison between PET and P5T
Full size table

Another common polyester is poly(alkylene 2,5-furandicarboxylate), the polyester of α,ω-n-alkanediols and FDCA (Fig. 35) [183]. Poly(ethylene 2,5-furandicarboxylate) (PEF) and poly(butylene 2,5-furandicarboxylate) (PBF or P4F) have been studied intensively since these polyesters can be fully derived from renewable resources. It has been reported that these polyesters provide superior mechanical and gas barrier properties in comparison with PET. The thermal properties of poly(alkylene 2,5-furandicarboxylate)s are summarized in Fig. 36 [184].

Fig. 35
figure 35

Structure of poly(alkylene 2,5-furandicarboxylate)

Full size image
Fig. 36
figure 36

Effect of number of methylene groups in poly(alkylene 2,5-furandicarboxylate) on the thermal property [184]. Reprinted with permission. Copyright (2016) Elsevier

Full size image

Polyesters using aliphatic α,ω-diacids have been also well known. The Tm and Tc values of poly(alkylene succinate)s [185], poly(alkylene adipate)s [186] and poly(alkylene dodecanedioate)s [187] show zig-zag variation dependence on the number of methylene groups in diol monomer, and the odd-numbered polyesters including those using 1,5-pentanediol have lower Tm and Tc. The variation is caused by the distortion of the crystalline chain: the dipole of polyesters with even-numbered constitutional units in the crystal lattice is more excellently aligned than that of odd-numbered polyesters. This factor affects the mechanical properties such as elongation at break, where odd-numbered polyesters are more flexible materials than even-numbered ones.

When aliphatic polyesters synthesized from 1,5-pentanediol and various aliphatic α,ω-diacids are compared, an odd-even effect was also observed for Tm and Tc; however, Tm and Tc of “odd-odd” polyesters are higher than those of an adjacent “odd-even” polyester, Such “odd-odd” type odd-even effect seems to be a beneficial supplement to the usual concept of “even-even” type odd-even effect [171].

In addition to thermal and mechanical properties described above, the functionality as a polymer material is much affected by the crystallinity. The gas barrier behavior is well known to be significantly influenced by the crystallinity of the polymer because gas molecules hardly diffuse and permeate in the crystals. Consequently, polymers with higher crystallinity degree are better barrier materials. From this point, amorphous P5F seems to be a worse gas barrier material because of the large fraction of free volume, which permits gas molecules to easily cross the polymer membrane. Contrary to this expectation, it has been reported that the ability of P5F to block gases is significantly higher than that of the most polymers used in food packaging [188]. Although furan-based polyesters generally have more than one phases including crystalline one, P5F does not present a crystalline phase, which leads to the lower amount of interphase domains. One explanation is that helical structures with π-π interactions of the furan ring are formed with the amorphous state and contribute to the gas barrier improvement [189]. Furthermore, it has also been reported that random copolyesters of ethylene glycol and 1,5-pentanediol with FDCA (poly(ethylene-co-1,5-pentylene 2,5-furandicarboxylate)s) showed excellent thermal, mechanical and O2 gas barrier properties [190].

1,5-Pentanediol is often used as a polymerization additive (plasticizer) to improve the flexibility and lower Tm, resulting in improvement of the processability of the polyester. In this point of view, copolymers [169, 191,192,193] and block polymers [194] have been reported, where polyesters were synthesized by changing the ratio of a main diol to 1,5-pentanediol. Among these, some polyesters have reported to be used as coatings and adhesives. In addition to the semi-crystallinity, low Tm and melting enthalpies, 1,5-pentanediol-based coatings exhibited gel content, hardness, flexibility, adhesion strength and solvent resistance similar to those of 1,6-hexanediol-based coatings [170]. Paint testing results showed that 1,5-pentanediol containing fully bio-based polyester for coil coating application and expected as flexible coatings but with better hardness than currently used diols such as 1,6-hexanediol [195].

To change the properties of polymers, blends are another technique. Although P6T and P5T are much different in crystallization rates, polymorphism, and spherulite morphology, the P6T and P5T blend is a miscible crystalline/crystalline blend in the melt state or quenched amorphous phase. The Tc of P6T-P5T apparently increased with increasing P5T contents, which suggests that the chain mobility of P6T in the blends is influenced by its interactions with P5T [196].

5.2 1,5-Pentanediol as monomer of polycarbonate

Polycarbonate usually means the aromatic one based on bisphenol-A. Poly(alkylene carbonate)s have received much attention over the last decade for biomedical application due to its low toxicity, biocompatibility and biodegradability. Typical synthesis method of poly(alkylene carbonate)s including poly(pentamethylene carbonate) (P5C) is similar to that of polyester: two step esterification and polycondensation of organic carbonate with diol using an appropriate catalyst [197, 198]. By using this method, α,ω-diols such as 1,5-pentanediol give the corresponding polycarbonate with high yield. In contrast, vicinal diols cannot give polycarbonates but give five membered cyclic carbonates. Zhang et al. [199] reported the reaction of 1,2-pentanediol and dimethyl carbonate to give 95% yield of the corresponding five-membered cyclic carbonate. Our group reported the direct polymerization of 1,5-pentanediol and CO2 to afford P5C oligomers for the first time, using CeO2 catalyst and 2-cyanopyridine as dehydrating agent (Mn=930, Mw/Mn=1.34) [200]. This method is effective to produce various poly(alkylene carbonate)s, and later we developed improved systems using 2-furonitrile as a dehydrating agent [201] or gas-stripping process without dehydrating agent [202]. These processes produce polycarbonate oligomers without the organic units derived from the dehydrating agent. Although not tested, application of these systems to synthesis of P5C oligomers is promising.

Consistent with the cases of polyester, Tg of aliphatic polycarbonate decreased with increasing the methylene number in the repeating unit. This can be explained by that the higher the methylene number in the repeating unit of the main chain, the lower the concentration of the stiffer carbonate groups, resulting in a more flexible polymer [203]. The reported Tg of P5C is around − 40°C (233 K) in most cases. Td of polycarbonate are slightly increased with increasing the methylene number in the repeating unit, and the reported Td of P5C is around 600 K. These tendencies of Tg and Td of the methylene number in the repeating unit is also observed in fully bio-based poly(diol-co-isosorbide carbonate)s including poly(1,5-pentanediol-co-isosorbide carbonate) [197]. For Tm, the odd-even effect is also observed in polycarbonates. Although Tm of P5C cannot be detected by DSC in some reports, P5C crystallizes much more slowly than C4, C6, C10 polycarbonates. The order of crystallizing ability of polycarbonate is C3 (amorphous) < C5 < C4 < C6 < C10. The trend in crystallization rate almost follows the order of Tm.

Biodegradability is an important character of aliphatic polycarbonates in practical view. The enzymatic degradation of polycarbonates revealed that the degradation rate of P5C is faster than those of poly(1,4-tetramethylene carbonate) (P4C) and poly(1,6-hexamethylene carbonate) (P6C) [198]. The monitoring of weight loss of these films showed that the weight loss of P4C and P6C was 20.7 and 75.5%, respectively, after 200 h of enzymatic degradation, while P5C was completely degraded even in 160 h. It is known that enzymatic degradation rate of a polymer depends mainly on the availability of the active center of the lipase [204]. Therefore, the key factor is the ability of the polymer chain to reach the active center of the lipase: the chain flexibility and crystal structure of the polymer. Among the tested polycarbonates, the enzymatic degradation rate of the most rigid P4C was the slowest. Although the chain flexibility of P5C is lower than that of P6C, the degree of crystallinity of P5C is lower (27.4%) than that of P6C (36.6%). Furthermore, the melting temperature of P5C (37.9°C; 311 K) is close to the enzymatic degradation test (35°C; 308 K). These factors can give the fastest enzymatic degradation rate of P5C.

6 Summary and outlook

One of the promising targets in the biomass-derived chemicals is the oxygen- and/or nitrogen-containing linear compounds functionalized at both α- and ω- positions because these compounds can be very useful as monomers for the production of plastics, resins, and so on. A possible synthesis route of these compounds is proposed to be the ring-opening C-O hydrogenolysis of saturated and unsaturated furanic compounds. The ring-opening C-O hydrogenolysis of furanic compounds functionalized at 2-position generally gives two different types of the products: 1,2- and α,ω-types.

Regarding the selective synthesis of 1,2-pentanediol, unsaturated furanic compounds such as furfural and furfuryl alcohol are more suitable, and the yield was increased by the recent progress on the catalyst development. The keys for further improvement of 1,2-pentanediol selectivity or yield include the decrease of activity in ring hydrogenation and increase of steric hindrance around the active center to suppress the attack at the more crowded O1-C2 bond. Currently highly selective catalysts to 1,2-pentanediol are noble-metal-based ones, and improvement of selectivity of inexpensive catalysts such as Cu-based ones is highly desirable. In contrast, high yield of 1,2-pentanediol from saturated furanic compounds such as tetrahydrofurfuryl alcohol has not been achieved, and this topic has remained challenging even now. The use of saturated substrates has a merit in obtaining long life of catalyst and whole system because unsaturated furanic compounds are relatively easily polymerized to humins, while hydrogenation of furanic compounds is a relatively easy reaction. Increase of steric hindrance may be one of the possible approaches to obtaining 1,2-pentanediol from tetrahydrofurfuryl alcohol.

The selective synthesis of 1,5-pentanediol from tetrahydrofurfuryl alcohol has been achieved previously using Rh-ReOx catalysts, and it has been recently reported that Pt-WOx catalysts and Ni-La(OH)3 catalysts showed similarly high yield of 1,5-pentanediol from tetrahydrofurfuryl alcohol. The high selectivity is derived by the directing effect: the -CH2OH group of bonding to the C2 atom in tetrahydrofurfuryl alcohol is strongly bound on the catalyst, and the O1-C2 bond is attacked with direction by the -CH2OH group. The conversion of furfural and furfuryl alcohol to 1,5-pentanediol was not so selective compared with the production of 1,5-pentanediol from tetrahydrofurfuryl alcohol and that of 1,2-pentanediol from furfural and furfuryl alcohol. The topic on the selective conversion of furfural and furfuryl alcohol to 1,5-pentanediol is also challenging and has been intensively investigated. The largest difficulty is the selectivity control between formations of 1,5-pentanediol and 1,2-pentanediol, and precisely controlled uniform active sites are probably necessary to obtain high degree of such selectivity. Further detailed investigation of the reaction mechanism is also desirable; however, the differences in selectivity and activation energies were small and sometimes affected by solvents. It is difficult for the current level of DFT to discuss the difference. Discussion should consider in-situ characterization results (rather than ex-situ or those for as-prepared catalysts) and kinetics, and the kinetics should be analyzed for the conditions giving high selectivity.

On the basis of the techno-economic analysis report on the direct and indirect (dehydration-hydration-hydrogenation) synthesis of 1,5-pentanediol from tetrahydrofurfuryl alcohol, the present candidate Rh-ReOx catalysts are too expensive. The catalysts with more reasonable price and/or with much higher activity/stability can be required. Another important point is that H2 pressure should be decreased with maintaining high catalytic activity considering the low pressure of H2 supplied from the renewable methods and high energy input for pressurization. Actually, the positive reaction orders with respect to H2 pressure have been frequently reported for the ring-opening C-O hydrogenolysis, and the lower pressure is connected to lower reaction rate. A new strategy for easier activation of H2 to supply more active hydrogen species is required for the drastic decrease in H2 pressure. Another recent progress is that the catalysis of the ring-opening C-O hydrogenolysis of furfural, furfuryl alcohol and tetrahydrofurfuryl alcohol is contributable to the development of the reaction systems for other substrates such as the related carboxylic acids and amines, 1,4-anhydroerythritol, 1,4-anhydroxylitol, isosorbide, and so on, indicating that ring-opening C-O hydrogenolysis can expand the options of the biomass-derived chemicals. In this review, we focused on the utilization of 1,5-pentanediol as a monomer for polyester and polycarbonate, and the potential and wide application of the biomass-derived monomers as mentioned above will be investigated in the future.