The Cascade Transformation of Furfural to Cyclopentanone: A Critical Evaluation Concerning Feasible Process Development

Abstract

Furfural is a fascinating bio-based platform molecule that can be converted into useful cyclic compounds, among others. In this work, the hydrogenative rearrangement-dehydration of furfural towards cyclopentanone using a commercially available Pt/C catalyst was investigated in terms of its reaction performance to assess its feasibility as an industrial process. However, acquiring an acceptable cyclopentanone yield proved very difficult, and the reaction was constrained by unforeseen parameters, such as the relative liquid volume in the reactor and the substrate concentration. Most strikingly, the sacrificial formation of furanoic oligomers that precipitated onto the catalyst’s surface was a troublesome key factor that mediated the product’s selectivity versus the carbon mass balance. By applying a biphasic water–toluene solvent system, the yield of cyclopentanone was somewhat improved to a middling 59%, while tentatively positive distributions of reaction components over these solvent phases were observed, which could be advantageous for anticipated down-stream processing. Overall, the sheer difficulty of controlling this one-pot cascade transformation towards a satisfactory product output under rather unfavorable reaction parameters renders it unsuitable for industrial process development, and a multi-step procedure for this chemical transformation might be considered instead.

1. Introduction

The society of the 21st century has grown particularly aware of and concerned by the imminent dangers of depleting fossil fuels, pollution by micro-plastics and greenhouse gases, observable climate change, and their looming consequences for global food production to sustain the world’s population. Several of these issues were already forecasted in 1972 by the Club of Rome in their famous book ‘The Limits to Growth’ [1], and, in 1987, the Brundtland Commission defined social-international action points to deal with them in their report ‘Our Common Future’ [2]. Fortunately, the growing compliance by various industries in prioritizing sustainability and circularity within their portfolios is increasing the demand for renewable materials and energy and is thereby promoting accelerated scientific research into the development of bio-based and bio-degradable chemicals [3,4,5,6,7]. However, the emergence of furfural (FAL) as a widely available bio-derived commodity chemical is a rare example of a serendipitous discovery that kick-started development towards commercial production instead, without there having been a demand for the product initially [8].

This historical event started in the early 1920s at the milling factory of Quaker Oats Company at Cedar Rapids, IA, USA, where oat hulls formed a troublesome by-product that spoiled them beyond their storage capacity at a staggering 50–60k tonnes per year. Therefore, investigations were undertaken to turn this waste problem into a profitable extended market. While the chemical processing thereof could not upgrade this material into a suitable source for nutrition, the notable loss of solid material upon acid treatment was further investigated and it was found that this process retrieved remarkably high quantities of FAL. Even though the market demand for FAL was very limited at the time (1922), the ease of producing FAL from oat hulls was far superior to other known methodologies that used corn cobs as raw material [9], and the situation at the Quaker Oats Company allowed the immediate deployment of a factory-unit for scale-up trials. The optimized process constituted suspending 5000 pounds of oat hulls in diluted sulfuric acid in a corrosion-resistant digestion reactor, which was sealed, mixed, and heated to generate steam pressure. Subsequently, the steam-carrying volatile reaction products were released directly into a fractionating distillation column to conveniently yield FAL in >98% purity [10]. At first, gaining market attraction for FAL was met with difficulty, despite its known applications in resin manufacturing, tanning, and dyeing processes and as precursor to agricultural pest control agents and other furan derivatives [11]. However, in 1936, the sales of FAL skyrocketed due to its usefulness as an extractive solvent for nearby oil refineries producing butane fuel gases and butadiene for synthetic rubber [12], which admittedly all profited from highly increased demands in the run-up towards World War II. Consequently, the concomitant demand for FAL exceeded the supply of oat hulls to such an extent that other agricultural waste sources, including corn cobs, were reverted to for its production [8].

In 2022, the global production of FAL accounted for 365 metric tonnes, constituting a market value of $595M at a sales price of 1500 $/tonne, and its demand and production continue to grow steadily [13]. Meanwhile, FAL has also become an important raw material for numerous fine chemicals and polymeric materials (Scheme 1) [14,15,16,17,18].

FAL finds application in various thermosetting resins with e.g., urea (1) [19,20], acetone (2) [21,22], melamine (3) [23], or phenolics (4) [24,25], and the present importance of these resins relates to the substitution of highly toxic formaldehyde with FAL. The derivatization of FAL provides a broad plethora of organic building blocks that are useful in the agricultural, pharmaceutical, and nutritional sectors. The oxidation of FAL yields furoic acid (5) [26,27], while its reductive amination yields furfurylamine (6) [28], and, through perhydrogenation to tetrahydrofurfuryl alcohol (THFOL) and subsequent thermal catalytic rearrangement, 3,4-dihydropyran (7) is yielded, which is a very important protecting agent of hydroxyl groups used in the medicinal chemistry [29,30].

On the other hand, the catalytic deoxygenation of FAL facilitates the production of furan (8) [31,32,33,34] and 2-methylfuran (2-MF) [35,36,37,38,39], which can be hydrogenated into the industrially relevant solvents tetrahydrofuran (THF) [40,41,42] and 2-methyltetrahydrofuran (2-MTHF) [42,43,44]. Further conversion of FAL-derived furan also enables the production of butanal (9), n-butanol (10), and butane (11) for utilization in bio-fuels and related additives [35]. Other catalytic pathways lead to the formation of maleic anhydride (12) [45], from which maleic acid (13), fumaric acid (14), and succinic acid (15) can be accessed [16,45,46] to be utilized in bio-based polymer syntheses [47]. Moreover, the catalytic ring-opening of FAL furnishes 4-oxopentanal (16) in situ, which can be hydrogenated to pentane-1,4-diol (17) [48,49,50], and the latter is envisioned as a renewable monomer as well [51,52].

FAL has a very rich chemistry in the field of hydrogenation. Its hydrogenation in organic solvents has long been known to yield furfuryl alcohol (FOL) [53,54,55,56], which can be further reduced to THFOL [57,58,59,60], pentane-1,2-diol (18) [61,62,63,64], and pentane-1,5-diol (19) [65,66,67,68]. FOL can also be hydrolyzed into levulinic acid (LA) [69,70,71], the subsequent hydrogenation of which yields the versatile molecule γ-valerolactone (20) [72,73]. Alternatively, the acid-catalyzed rearrangement of FOL leads to the formation of 4-hydroxy-cyclopent-2-enone (4-HCP) [74,75,76,77,78,79], from which the specialty monomer cyclopentane-1,3-diol (CPdiol) can be obtained by selective hydrogenation [80], and the further transformation thereof via the use of cyclopentadiene (21) as intermediate furnishes JP-10 jet fuel (i.e., exo-tetrahydro-dicyclopentadiene (22)) [81]. However, under very specific reaction conditions, a hydrogenative cascade transformation of FAL is accomplished [82], which unveils the bio-based route towards a platform molecule that is highly valuable to chemical industries and beyond: cyclopentanone (CPON).

CPON is commonly known for its applications in fragrances and pharmaceuticals. While CPON can function as a perfume ingredient by itself [83,84], it is also a common precursor thereof, for example in the synthesis of jasmone [85]. CPON also finds application in the pharmaceutical sector as a solvent [86], and it is a synthon for barbiturates [87] as well as for the specialty chemical dimethyl cubane-1,4-dicarboxylate and its derivatives [88,89]. However, a quickly emerging class of CPON applications is the synthesis of renewable high-density aviation fuels (23), which remain in a liquid state even at typical mid-tropospherical (i.e., at 10 km altitude) temperatures, around −40 °C [90]. The chemistry for forming intermediate products from CPON that aid in the creation of these renewable fuels generally relies on aldol reactions catalyzed by alkaline [91,92,93], MgAl-hydrotalcite [94,95,96,97,98], or acidic catalysts [99,100] (Scheme 2). Both single- and double-aldol condensation products can be obtained from CPON with FAL [91,92,93,94,99,100], with 5-hydroxy-methylfurfural (HMF) [91,92,93], and with butanal [96] to yield 24a, 24b, and 25, respectively. These are subsequently reduced over supported Pd or Ni catalysts to cause hydrodeoxygenation and to furnish 1,3-dialkylated cyclopentane hydrocarbons with linear aliphatic C₅, C₆, and C₄ branches, respectively (i.e., 26a, 26b, and 27, respectively). Similarly, the hydroxyalkylation-alkylation of CPON with two equivalents (eq.) of 2-MF (giving 28), followed by its catalytic hydrodeoxygenation, yields a geminally dialkylated cyclopentane-based fuel (29) [101]. An alternative example by Xie et al. [102] comprises a three-step transformation of cyclic ketones (including CPON) through reductive coupling (giving 30), pinacol rearrangement (giving 31), and finally Wolff–Kishner reduction to generate the spiro-structured hydrocarbon fuel (32). However, the most advanced fuel structures arise from single and double self-aldol condensations of CPON, which yield 33 and 34, respectively, and which open the door to further conversions through the strategic selection of various hydrogenation-, hydrodeoxygenation-, and/or acidic rearrangement-enabling catalysts to synthesize bicyclopentane (35) [91,95,97,99,103,104], ter-cyclopentane (36) [91,97,105], quarter-cyclopentane (39) (via intermediates 37 and 38) [104], decalin (42) (via intermediates 40 and 41) [99,103], and the aromatic trindane (43) [105]. Several of these multi-step synthesis routes towards the final hydrocarbon fuel products can also be established in a plug-flow reactor with multiple consecutive packed catalyst beds to enable a convenient cascade production process [94,97,98,101,104]. Moreover, it is worth noting that the aldol reaction between CPON and aldehydes proceeds much more readily than the self-aldol condensation of CPON, which could be a relevant insight concerning the one-pot cascade synthesis of CPON from FAL (vide infra).

2. Technological Background

As mentioned before, hydrogenation reactions of FAL have been investigated for many decades, and are industrially applied for the manufacture of FOL, THFOL, 2-MF, and 2-MTHF [106,107]. However, it was not until 2012 that Hronec et al. tested the hydrogenation of FAL over several supported metal catalysts (e.g., Pt/C, Pd/C, Ru/C) in aqueous media at 160–175 °C under 80 bar H₂ pressure, which unexpectedly led to the quite selective formation of CPON, and occasional over-hydrogenation to cyclopentanol (CPOL) [82]. This peculiar product formation can be understood from the reaction equation depicted in Scheme 3.

First, FAL undergoes aldehyde reduction towards FOL, which can be achieved selectively and in high yields when using specific catalysts and solvents (see Supplementary Materials Figures SM-1.A and SM-1.B) [53,54,55,56]. In the second step, the aqueous solvent and high temperature are the crucial factors for converting FOL into 4-HCP, which is known as the ‘Piancatelli rearrangement’ [74,75,76,77,78,79]. Next, the 4-HCP is selectively converted to 3-hydroxycyclopentanone (3-HCP) via the olefinic bond reduction of its ketenone functionality [80], after which it readily dehydrates to form cyclopent-2-enone (CPEON). Finally, CPEON also undergoes olefinic bond reduction to yield CPON. The inevitable necessity of employing water as one of the solvents for this cascade transformation to occur was demonstrated in additional experiments by Hronec et al. as well, where the use of n-butanol or tetrahydrofuran as solvents led to the formation of, notably, FOL and 2-MF instead [82].

Moreover, the Piancatelli rearrangement proceeds via an acid-catalyzed reaction mechanism (see Supplementary Materials Scheme SM-2), but is usually accompanied by several side reactions, among which the self-polymerization of FOL is the most notable occurrence (Scheme 4), while LA is another common side product [80]. A typical mitigation strategy to improve the selectivity of this conversion was demonstrated by the groups of Reiser [75] and Fadnavis [76], who applied organic co-solvents—toluene and N-methylpyrrolidone, respectively—to dissolve and also suppress the (formation of) polymers from FOL in their dedicated flow-reactors.

Shortly after their initial publication on selective FAL hydrogenation to CPON, Hronec et al. reviewed several related studies [108,109,110], including the important role of polymers from FOL with respect to the selectivity of CPON formation [109]. They determined with X-ray photoelectron spectroscopy that these water-insoluble polymers had precipitated onto the surface of the supported metal catalysts used in the hydrogenation reaction, and that these polymers typically consist of 5–30 furanoic repeating units. Hydrogenation experiments using recycled catalysts as well as fresh catalysts that were impregnated with isolated furanoic polymer separately and unanimously evidenced the selectivity effect of the furanoic polymer, where FAL was always converted to CPON with appreciable selectivity, while the hydrogenation of FAL using non-treated fresh catalyst exhibited significant over-hydrogenation to CPOL. The mechanism behind the selectivity enhancement by the furanoic polymer is believed to rely on competitive chemisorption onto the active metal sites of the catalyst, which promotes the desorption of CPON before it can be over-hydrogenated to CPOL [109,111].

Meanwhile, the cascade transformation of FAL to CPON (and CPOL) was quickly pursued by many other research groups worldwide, with a dominant focus on testing different catalysts, especially custom composites of a bimetallic nature or with tailored carrier supports [112,113,114]. The corresponding literature publications [82,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150] from the period of 2012–2020 are listed in Table 1.

In 2013, Yang et al. conducted thorough research on the use of Ni-Cu on an SBA-15 (mesoporous silica) catalyst, which yielded 62% CPON within 4 h at a milder 160 °C under 40 bar H₂ [115]. Guo et al. achieved a 60% CPON yield under similar reaction conditions with a calcined CuZnAl catalyst in the next year [116]. However, the period of 2015–2017 witnessed a strongly increasing number of related publications, wherein some outstanding CPON yields (>90%) were reported by Fang et al., who used Ru nanoparticles on an acidic metal–organic framework [117]; Hronec et al., who reappeared in the literature with a carbon-supported Pd-Cu catalyst [118]; Li et al., who reported a miraculous >99% CPON yield using their Au/TiO₂ (anatase) catalyst [119]; Wang et al., who established a thermolyzed Cu-Ni@C matrix catalyst [120]; Liu et al., who used carbon nano-tubes as carrier support for their Ru catalyst [121]; among other reports [122,123,124]. From 2018 onward, the number of publications on the cascade transformation of FAL or FOL to CPON and/or CPOL grew exponentially, and these publications presented interesting catalyst compositions, but they also had highly variable catalytic performances [125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177]. However, the most noteworthy concept was reported by Dohade & Dhepe [128], who applied the first biphasic solvent system to this reaction, which comprised a mixture of toluene and water (3:4 v/v) in combination with Pt-Co/C catalyst, yielding an appreciable 75% CPON yield, which offers beneficial perspectives in terms of down-stream processing (DSP) to isolate CPON. Furthermore, some other interesting catalytic concepts involving a homogeneous iridium complex [178] and a montmorillonite K10 clay catalyst [179] were reported to selectively yield CPEON in the hydrogenation of FAL.

While the reported optimized reaction conditions of these publications generally indicate highly successful performances in terms of CPON yields, Table 1 explicates the numerical reaction parameters of these established chemical systems to crudely evaluate their catalytic efficiencies and overall techno-economical feasibility. The applied catalysts are essentially always supported-metal catalysts, metal–organic frameworks (MOFs), or other multi-metallic composites, with widely diverse support materials and active sites featuring base transition metals as well as platinum-group metals. However, the metal/support ratios and/or catalyst/substrate loadings are often exorbitantly high (≡ >10 wt%), which limits the results to poor turn-over numbers (TONs) and may therefore lead to expensive catalytic concepts. In addition, it is striking that the substrate concentration is typically 5 v% or lower and that the reactor fill usually renders less than 50 v%. As a consequence, the vast majority of the reported reaction systems for the hydrogenative cascade transformation of FAL to CPON score rather poor space-time yields (STYs). Given the complexity of the in situ-generated mixture of intermediate compounds—in particular concerning the incorporated Piancatelli rearrangement of FOL to 4-HCP—the low substrate concentration is deemed a necessity for a selective conversion towards CPON (and CPOL). On the other hand, the use of restrained quantities of reaction mixture to fill the reaction vessel gives a larger overhead space that allows a higher accumulation of pressurized H₂, which is considered to plausibly enforce the arduous solvation of H₂ in the aqueous medium to facilitate adequate hydrogenative conversion [82,180]. As such, the industrialization of the bio-based synthesis of CPON from FAL could remain a tremendous challenge.

Table 1. The literature on the selective catalytic conversion of FAL to CPON in chronological order.

Ref.	Author(s)	Catalyst System (Metal-to-Support wt%)	Cat. Load. (wt%)	T (°C)	PH2 (bar)	Time (h)	Sub. Conc. (v/v%)	Reactor Fill (v%)	CPON Yield (%)	CPOL Yield (%)
[82]	Hronec *	Pt/C (5%)	5	160	80	0.5	5	20	76.5	5
[115]	Yang, Xu *	Ni-Cu/SBA-15 (10%; 5%)	2	160	40	4	5	16.7	62	3
[116]	Guo, Guo *	CuZnAl	41.6	150	40	6	3.3	36	60.3	2.5
[181]	Liu, Xiao *	Ni/zeolite Y (20%)	1.5	150	40	9	5	N/D	86.5	4.9
[182]	Li, Fu *	Cu-Co3O4 (5%)	26	170	10	1	2	50	67	10
[183]	Zhu, Xiao *	Cu-Ni-Al/HT (ratio 1:14:5)	25.9	140	40	8	5	50	95.8	3.0
[117]	Fang, Li *	Ru/MIL-101 (3%)	10	160	40	2.5	10	60	96	1
[118]	Hronec *	Pd-Cu/C (5%; 10%)	1	160	30	1	5	20	92.1	0.4
[119]	Zhang, Cao *	Au/TiO2 (0.10%)	10	160	40	1.2	5	20	>99	0
[120]	Wang, Xiao *	Cu-Ni@C (0.02%; 0.01%)	2	130	50	5	5	50	96.9	1.1
[121]	Liu, Wang *	Ru/CNT (6%)	4.3	160	10	5	5	50	91	4
[122]	Liu, Mu *	Pt/NC-BS-800 (5%)	5	150	10	4	2	N/D	76	9
[123]	Wang, Xiao *	Cu/MgO (20%)	13.8	140	40	8	5	50	85	9
[124]	Zhou, Jiang *	Cu-Zn/CNT (17%; 3%)	2	140	40	10	5	50	85.3	5.8
[125]	Li, Shi *	Ni-Co/TiO2 (10%; 10%)	60	150	40	4	3.3	30	53.3	16.3
[126]	Zhang, Li *	Cu/ZrO2 (39%)	10.4	150	15	4	3.3	N/D	91.3	3.8
[127]	Date, Rode *	Pd/f-SiO2 (4%)	10	165	34.5	5	5	33	89	0
[128]	Dohade, Dhepe *	Pt-Co/C (3.12%; 3.10%)	22.3	180	10	5	5	33	75	2
[129]	Shen, Ying *	Ru/C + Al11.6PO23.7 (0.5%)	2.6	160	40	4	4	52	84	0
[130]	Cherkasov, Rebrov *	Pd-Bi/SiO2 (4.87%; 1.36%)	10.4	150	50	2.3	5	63	54.6	N/D
[131]	Zhou, Huang *	Cu0.4Mg5.6Al2	66.6	180	2	5	5	4	98.1	0
[132]	Li, Deng *	Pd/Fe-MIL-100 (5%)	10	150	40	6	2.5	43	92	N/D
[133]	Pan, Feng *	Cu-Fe3O4 (10%)	50	160	30	4	1	40	91	6
[134]	Deng, Zhang *	Pd/Cu-BTC (5%)	1	150	40	6	2.5	43	93	N/D
[135]	Mironenko *	Pd/CNT (1%)	8.6	150	30	2	4	N/D	37	N/D
[136]	Astuti, Mujiyanti *	Ni-Co/TiO2 (20%; 6.8%)	47.3	170	30	6	5	N/D	27	41
[137]	Wang, Zhang *	Pd-Co@UiO-66 (4.4%; 0.62%)	17.3	120	30	12	1	20	96	N/D
[138]	Ren, Li *	Cu4Zn/Al (film: 40 cm2)	N/D	140	20	2	0.5	40	86.5	4
[139]	Li, Deng *	Pd/FeZn-DMC (5%)	10	150	40	6	2.5	41	87.5	N/D
[140]	Deng, Deng *	Pd/pyrochlore (5%)	10	150	40	6	2.5	50	92	N/D
[141]	Lee, An *	Pd/CMK-3 (0.94%)	1	160	30	5	5	21	41.9	N/D
[142]	Liu, Li *	Ni2Cu1/Al2O3 (44%; 24%)	4	140	10	1	3.3	N/D	89.5	7
[143]	Zhu, He *	Cu0-Zn/(Al)(Zr)O-2 (11%)	5	160	40	2.5	5	N/D	92	2
[144]	Herrera, Escalona *	Ni/CNTox (10%)	14	200	20	1	2.2	N/D	20	7
[145]	Jia, Wang *	NiFe/SBA-15 (5.0%; 1.6%)	20	160	34	5	6	53	90	N/D
[146]	Mironenko *	Pd/CNT (1%)	8.6	200	80	1	4	N/D	79	N/D
[147]	Gao, Hu *	Ni-P/γ-Al2O 3 (15%; 10%)	104	150	30	2	1	40	90.1	0

Abbreviations: SBA = Santa Barbara amorphous; HT = hydrotalcite; MIL-101 = Materieaux de l’Institut Lavoisier n° 101; CNT = carbon nanotubes; NC-BS = heteroatom-doped carbon from biomass; f-SiO₂ = fumed silica; BTC = 1,3,5-benzene tricarboxylate; DMC = double-metal cyanide; UiO-66 = metal–organic framework of [Zr₆O₄(OH)₄(CO₂)₁₂] clusters linked with terephthalate ligands; CMK-3 = ordered meso-porous carbon; CNTox = oxygen-functionalized carbon nanotubes. * Corresponding authors of the literature references; other mentioned authors are always the first author.

Table 1. The literature on the selective catalytic conversion of FAL to CPON in chronological order.

Ref.	Author(s)	Catalyst System (Metal-to-Support wt%)	Cat. Load. (wt%)	T (°C)	PH2 (bar)	Time (h)	Sub. Conc. (v/v%)	Reactor Fill (v%)	CPON Yield (%)	CPOL Yield (%)
[82]	Hronec *	Pt/C (5%)	5	160	80	0.5	5	20	76.5	5
[115]	Yang, Xu *	Ni-Cu/SBA-15 (10%; 5%)	2	160	40	4	5	16.7	62	3
[116]	Guo, Guo *	CuZnAl	41.6	150	40	6	3.3	36	60.3	2.5
[181]	Liu, Xiao *	Ni/zeolite Y (20%)	1.5	150	40	9	5	N/D	86.5	4.9
[182]	Li, Fu *	Cu-Co3O4 (5%)	26	170	10	1	2	50	67	10
[183]	Zhu, Xiao *	Cu-Ni-Al/HT (ratio 1:14:5)	25.9	140	40	8	5	50	95.8	3.0
[117]	Fang, Li *	Ru/MIL-101 (3%)	10	160	40	2.5	10	60	96	1
[118]	Hronec *	Pd-Cu/C (5%; 10%)	1	160	30	1	5	20	92.1	0.4
[119]	Zhang, Cao *	Au/TiO2 (0.10%)	10	160	40	1.2	5	20	>99	0
[120]	Wang, Xiao *	Cu-Ni@C (0.02%; 0.01%)	2	130	50	5	5	50	96.9	1.1
[121]	Liu, Wang *	Ru/CNT (6%)	4.3	160	10	5	5	50	91	4
[122]	Liu, Mu *	Pt/NC-BS-800 (5%)	5	150	10	4	2	N/D	76	9
[123]	Wang, Xiao *	Cu/MgO (20%)	13.8	140	40	8	5	50	85	9
[124]	Zhou, Jiang *	Cu-Zn/CNT (17%; 3%)	2	140	40	10	5	50	85.3	5.8
[125]	Li, Shi *	Ni-Co/TiO2 (10%; 10%)	60	150	40	4	3.3	30	53.3	16.3
[126]	Zhang, Li *	Cu/ZrO2 (39%)	10.4	150	15	4	3.3	N/D	91.3	3.8
[127]	Date, Rode *	Pd/f-SiO2 (4%)	10	165	34.5	5	5	33	89	0
[128]	Dohade, Dhepe *	Pt-Co/C (3.12%; 3.10%)	22.3	180	10	5	5	33	75	2
[129]	Shen, Ying *	Ru/C + Al11.6PO23.7 (0.5%)	2.6	160	40	4	4	52	84	0
[130]	Cherkasov, Rebrov *	Pd-Bi/SiO2 (4.87%; 1.36%)	10.4	150	50	2.3	5	63	54.6	N/D
[131]	Zhou, Huang *	Cu0.4Mg5.6Al2	66.6	180	2	5	5	4	98.1	0
[132]	Li, Deng *	Pd/Fe-MIL-100 (5%)	10	150	40	6	2.5	43	92	N/D
[133]	Pan, Feng *	Cu-Fe3O4 (10%)	50	160	30	4	1	40	91	6
[134]	Deng, Zhang *	Pd/Cu-BTC (5%)	1	150	40	6	2.5	43	93	N/D
[135]	Mironenko *	Pd/CNT (1%)	8.6	150	30	2	4	N/D	37	N/D
[136]	Astuti, Mujiyanti *	Ni-Co/TiO2 (20%; 6.8%)	47.3	170	30	6	5	N/D	27	41
[137]	Wang, Zhang *	Pd-Co@UiO-66 (4.4%; 0.62%)	17.3	120	30	12	1	20	96	N/D
[138]	Ren, Li *	Cu4Zn/Al (film: 40 cm2)	N/D	140	20	2	0.5	40	86.5	4
[139]	Li, Deng *	Pd/FeZn-DMC (5%)	10	150	40	6	2.5	41	87.5	N/D
[140]	Deng, Deng *	Pd/pyrochlore (5%)	10	150	40	6	2.5	50	92	N/D
[141]	Lee, An *	Pd/CMK-3 (0.94%)	1	160	30	5	5	21	41.9	N/D
[142]	Liu, Li *	Ni2Cu1/Al2O3 (44%; 24%)	4	140	10	1	3.3	N/D	89.5	7
[143]	Zhu, He *	Cu0-Zn/(Al)(Zr)O-2 (11%)	5	160	40	2.5	5	N/D	92	2
[144]	Herrera, Escalona *	Ni/CNTox (10%)	14	200	20	1	2.2	N/D	20	7
[145]	Jia, Wang *	NiFe/SBA-15 (5.0%; 1.6%)	20	160	34	5	6	53	90	N/D
[146]	Mironenko *	Pd/CNT (1%)	8.6	200	80	1	4	N/D	79	N/D
[147]	Gao, Hu *	Ni-P/γ-Al2O 3 (15%; 10%)	104	150	30	2	1	40	90.1	0

Abbreviations: SBA = Santa Barbara amorphous; HT = hydrotalcite; MIL-101 = Materieaux de l’Institut Lavoisier n° 101; CNT = carbon nanotubes; NC-BS = heteroatom-doped carbon from biomass; f-SiO₂ = fumed silica; BTC = 1,3,5-benzene tricarboxylate; DMC = double-metal cyanide; UiO-66 = metal–organic framework of [Zr₆O₄(OH)₄(CO₂)₁₂] clusters linked with terephthalate ligands; CMK-3 = ordered meso-porous carbon; CNTox = oxygen-functionalized carbon nanotubes. * Corresponding authors of the literature references; other mentioned authors are always the first author.

In this work, the hydrogenative/dehydrative one-pot cascade transformation from FAL to CPON was first targeted for techno-chemical evaluation in terms of its experimental reproducibility using a 100 mL stainless-steel autoclave fit for high-pressure operations. The results and observations derived from these bench-scale experiments were then envisioned to provide guidelines for a potential development of the process towards the scaled-up production of bio-based CPON in an industrial context. Hereto, a reactor design supporting batch reactions was anticipated, and a commercially available catalyst was strategically selected: Pt/C. A tactical DSP was devised, wherein the heterogeneous catalyst would first be removed by filtration, while the filtrate containing CPON + CPOL products, tentatively some intermediate reagents and (polymeric) side products, and the aqueous solvent would subsequently be separated via fractionated distillation. To facilitate the DSP, it was additionally considered whether to employ a biphasic water–toluene solvent system to facilitate some initial product isolations through phase separation and/or enable the advantageous azeotropic evaporation of the solvent system in the distillation.

3. Materials and Methods

3.1. General Considerations

Unless otherwise stated, all commercial chemicals were used as received without further purification: Ru/C (5 wt%, dry, reduced) was purchased from Strem Chemicals (Newburyport, MS, USA); Pt/C (10 wt%, wet, reduced) was purchased from Johnson Matthey (London, England); Pt/C (5 wt%, dry, reduced), Pt/Al₂O₃ (5 wt%, dry, reduced), ‘Heterogeneous Palladium Catalyst Kit I’ {including Pd/C (10 wt%, dry, reduced), Pd/C (5 wt%, wet, reduced), Pd/Al₂O₃ (5 wt%, dry, reduced), Pd/CaCO₃ (5 wt%, dry, reduced, Pb-poisoned), Pd/BaSO₄ (5 wt%, dry, reduced), Pd(OH)₂/C (5 wt%, wet, reduced)}, Rh/C (5 wt%, dry, reduced), furfural (99% pure), furfuryl alcohol (>98% pure), cyclopent-2-enone (98% pure), tetrahydrofurfurylalcohol (99% pure, Sigma-Aldrich), cyclopentane-1,3-diol (mixture of cis and trans, 95% pure, Sigma-Aldrich), and 2-methylfuran (99% pure) were purchased from Sigma-Aldrich (St. Louis, MO, USA). 3-hydroxycyclopentanone (95% pure (but spontaneously dehydrated)) was purchased from Fluorochem (Hadfield, England). Cyclopentanone (>99% pure), cyclopentanol (99% pure), and Celite^® R566 were purchased from Acros Organics (Geel, Belgium). 2-methyltetrahydrofuran (AR grade), toluene (HPLC grade), and acetonitrile (HPLC grade) were purchased from Biosolve Chimie (Dieuze, France). Naphthalene (≥99% pure) was purchased from Alfa Aesar. Dimethylsulfoxide-d6 (99.7% D) was purchased from Fisher-Scientific (Pittsburgh, USA); deuterium oxide (99.96% D) was purchased from Cambridge Isotope Laboratories (Tewksbury, MS, USA), N₂ (≥99.9% pure, dry) and H₂ (≥99.9% pure, dry) were purchased from Linde Gas (Dublin, Ireland).

Distillations were conducted under an inert N₂ atmosphere using a dual Schlenk manifold (Glasblazerij Janssen; Beek, The Netherlands), which was connected with one line to a vacuum pump (VacuuBrand RE2.5; Wertheim, Germany) that created a vacuum in the order of 10⁻³ mbar, while the other line was connected to a supply of N₂ gas, unless stated otherwise.

Hydrogenation reactions were conducted in a Series 4560 Mini autoclave with a 100 mL reaction vessel (Parr Instrument Company; Moline, IL, USA), equipped with a sampling outlet. This device was connected to a custom-made gas control manifold containing pressure-control valves (Swagelok, Solon, OH, USA). The complete setup was connected to separate supplies of N₂ gas and H₂ gas. The autoclave was also equipped with a by-pass injection vessel to add a (solution of) substrate facilitated by locally increased pressure into the reaction vessel when pre-heated.

All hydrogenation reactions were performed using distilled substrates (i.e., furfural, furfuryl alcohol, 4-hydroxycyclopent-2-enone), which were properly stored at −20 °C. Unless stated otherwise, all reactions were performed in demineralized water using 5 wt% of a commercial supported metal catalyst with respect to the substrate, and the standard substrate concentration in the liquid phase was 5 v%. Reaction mixture volumes ranged from 21 to 63 mL throughout the studied experiments.

NMR spectra were recorded on a Bruker Avance AV 300 Ultra Shield spectrometer (Billerica, MA, USA), operating at 25 °C, with 16 scans for ¹H-NMR measurements at 300 MHz and with 2048 scans for ¹³C-NMR at 75 MHz, using the Topspin 3.5pl7 software. For spectrum processing, the software MestReNova v5.1 was used. All spectra were calibrated on the residual proton signal of the applied NMR-solvent in accordance with the corresponding values in the literature [184], and the recorded chemical shifts for the chemical products were matched with the reported values for 3-HCP [80], CPON [185,186] and for CPOL [187], and CPdiol [80]

FTIR spectra were recorded on a MIRacle 10 spectrometer (Shimadzu; Kyoto, Japan) equipped with a single-reflection ATR crystal, in the range of 400–4000 cm⁻¹, with a resolution of 2 cm⁻¹, with 32 accumulating scans per spectrum, and applying Happ-Genzel apodization. For spectrum processing, the software IR Solution (version 1.6) was used.

GC-FID analyses were performed using a Hewlett Packard 5890 series II gas chromatograph (Palo Alto, CA, USA) equipped with a flame-ionization detector and a CP-Chirasil-Dex CB capillary column (length = 25 m; internal diameter = 0.25 mm; film thickness = 0.25 μm). The heating program was 5 min at 40 °C, a ramp-up of 5 °C/min up to 200 °C, and a final 8 min at 200 °C.

3.2. Synthesis of 4-Hydroxycyclopent-2-Enone (4-HCP)

A total of 11.0 g freshly distilled furfuryl alcohol (FOL) was dissolved in demineralized water to a total of 500 mL. The solution was divided into portions of 20.0 mL over 25 microwave vials with a size of 25 mL. The vials were sealed under aerobic conditions using aluminum crimp caps fitted with a PTFE septum. Each vial was irradiated using a Biotage Initiator+ microwave (Uppsala, Sweden). The applied heating program consisted of a heating ramp-up to reach 200 °C within 100 s, then remained at 200 °C for 10 min, and was finally cooled down to room temperature within 5 min by means of a pressurized air flow. After the heating procedure, the reaction liquids had become orange-brown, and a suspension of a small amount of dark-brown solids had formed. The 25 reaction mixtures were combined and centrifuged for 15 min at 15,000 rpm in order to trap all solids in a pellet. The light-orange liquids were then decanted carefully, while the solid pellets were discarded. The liquid mixture was rotavaporized to give an orange oil (10.92 g crude yield). Finally, this oil was vacuum-distilled at 83 °C and 6.10⁻² mbar (oil bath at 115 °C, vigreux = 8 cm long) to furnish 7.98 g (64.5%) of pure (99%) light-yellow viscous liquid product. A trace amount of LA impurity was observed in NMR.

Analysis [80]: ¹H-NMR (300 MHz), 25 °C, DMSO-d6 (2.50 ppm): δ = 7.64 (dd, J₁ = 5.6; J₂ = 2.3 Hz, 1H, O=C-CH=CH), 6.15 (dd, J₁ = 5.6; J₂ = 1.0 Hz, 1H, O=C-CH=CH), 5.46 (d, J = 3.9 Hz, 1H, OH), 4.83 (s, 1H, CH-OH), 2.63 (dd, J₁ = 18.2; J₂ = 6.0 Hz, 1H, CHH), 2.03 (dd, J₁ = 18.2, J₂ = 2.1 Hz, 1H, CHH) ppm. ¹³C-NMR (75 MHz), 25 °C, DMSO-d6 (39.50 ppm): δ = 207.0 (s, 1C, C=O), 166.0 (s, 1C, C=C-C=O), 133.5 (s, 1C, C=C-C=O), 69.0 (s, 1C, CH-OH), 44.2 (s, 1C, CH₂) ppm. FTIR (neat): ν = 3381 (br, m, OH), 3087 (vw, C=C-H), 2920 (vw, CH₃), 1703 (vs., C=O), 1585 (w, C=C), 1400 (m), 1341 (m), 1265 (w), 1233 (w), 1184 (m), 1151 (m), 1101 (s), 1038 (vs., C-O), 945 (s), 854 (w), 831 (w), 793 (s), 731 (w), 656 (m) cm⁻¹.

3.3. Cascade Hydrogenation-Dehydration Reactions

A typical experiment was performed by loading the reaction vessel of a 100 mL autoclave with a carefully weighed amount of the catalyst and half the amount of the solvent. The substrate was dissolved in the other half of the solvent, which was then transferred into an injection vessel under an outflow of N₂. The autoclave was assembled, sealed, purged under N₂ by applying three vacuum/N₂ cycles, and finally charged with 10 bar cold H₂ pressure. Under mechanical stirring at 1800 rpm, the reaction vessel was pre-heated to 20 degrees above the desired temperature (i.e., 160 °C or 180 °C), upon which the pressure increased to 20–35 bar. Next, the injection vessel was charged with 50 bar H₂, to force the cold substrate solution into the reaction vessel, which caused the resulting reaction mixture to reach the desired reaction temperature within a margin of ±5 °C. Immediately afterwards, the injection vessel was closed, the H₂ pressure inside the reaction vessel was manually increased to 80 bar H₂, and the settings of the autoclave system were adjusted to maintain the desired reaction temperature. Optionally, kinetic samples were collected at select time intervals of (5, 10), 15, 30, 45, 60, 90, 120, 180, 240, 300, and (360, 420) minutes to capture the reaction curves of all measured compounds accurately in the graphs. Afterwards, the heating mantle of the autoclave was removed, and an ice water bath was used for cooling. When a temperature below 40 °C was reached, the remaining H₂ pressure was carefully released. The autoclave was further neutralized by purging it three times with 3 bar N₂, and was finally opened to retrieve the reaction mixture. Aliquots (20.0 μL) of the kinetic samples were dissolved in 1.00 mL stock solution of 0.100 wt% naphthalene as internal standard in acetonitrile.

3.4. Control Hydrogenation of 4-HCP at Room Temperature

Glass vials with a size of 3 mL that were equipped with a magnetic stirring bar were loaded with 2.0 mg supported metal catalyst, 20 mg 4-HCP, and 1.00 mL deuterated water, and a disposable snap-cap was fitted on each vial. Four such reaction vials at a time were mounted inside a Series 4560 Mini Stainless steel 100 mL autoclave from Parr Instrument Company (Moline, IL, USA)—from which the stirring propeller had been removed—and the snap-caps were pierced once with a thick needle to allow gas exchange. The reactor was sealed, purged with 5x 2.5 bar N₂ and 3x 10 bar H₂, and finally charged with 50 bar H₂ pressure. The reaction mixtures were magnetically stirred at 300 rpm using a stirring plate placed underneath the autoclave. After 60 min, the reactions were stopped by purging, where the reactor was purged with 3x 2.5 bar N₂ before opening. The reaction mixtures were separated from the catalyst through a Millipore filter and were subsequentially subjected to NMR analysis for identification and to GC-FID analysis for quantification.

4. Results and Discussion

4.1. Reproducing the Work of Hronec

Despite several reported advancements in the hydrogenative cascade transformation of FAL to CPON in terms of the use of tailored heterogeneous catalysts [115,116,117,181,182,183], I decided, in view of feasible process designing to achieve tentative scale-up activities, to pursue this research using commercially available standard catalysts, and thereby to revert to the primeval concept as described by Hronec et al. in 2012 [82]. By adopting the optimal reaction conditions, the effects of several supported precious-metal catalysts were revisited (

Table 2. Single-end-point screening experiments of FAL hydrogenation over different precious-metal catalysts ^a.

Entry	Catalyst	T (°C)	FAL Conv. (%) b	Product Yields (%) b						Mass Balance (%)
				FOL	THFOL	4-HCP	CPEON	CPON	CPOL
1	Pt/C	25	3.5	2.3	0	0	0	0	0	98.8
2	Pt/C	160	85.2	0.9	0	5.7	4.6	19.0	32.3	77.2
3	Pd/C	25	100	0.9	64.3	0.2	0	0	0	65.4
4	Pd/C	160	100	0.9	46.8	0.6	0	0	11.8	60.1
5	Ru/C	25	16.9	8.0	1.5	0	0	0	0	92.7
6	Ru/C c	160	97.9	0	9.3	0	0	14.4	11.5	37.2
7	Pt/C d	160	59.1	0.8	0	10.1	4.3	7.9	23.4	59.1
8	Pt/Al2O3	180	63.3	5.0	0	0.5	0	0.9	0	43.2
9 e	Pt/C	160	100	0	0.4	N/D	N/D	76.5	4.8	85.2

^a Reaction conditions: 1.00 g FAL and 50 mg catalyst in 20.0 mL water, P = 80 bar H₂, t = 60 min, stirring rate = 1800 rpm; ^b determined by GC-FID; ^c 36.5% cyclopentane-1,3-diols were detected; ^d alternative Pt/C batch, purchased from Johnson Matthey; ^e best result by Hronec et al. [82].

). Catalysts purchased from Sigma-Aldrich were tested first at two different temperatures: Pt/C barely gave conversion at 25 °C, but at 160 °C it rendered the highest product selectivity for the combination of CPON (19.0%) and CPOL (32.3%); Pd/C readily hydrogenates FAL even at room temperature, but yields THFOL (64.3%) as the main product; and Ru/C only showed appreciable FAL conversion at 160 °C, which led to a mixture of THFOL (9%), CPON (14.4%), CPOL (11.5%), and cyclopentane-1,3-diols (CPdiol) (36.5%). Clearly, platinum is the most suitable metal to catalyze this reaction at a high temperature in terms of its {CPON + CPOL} selectivity and carbon mass balance. Therefore, two more supported platinum catalysts were tested, namely a different Pt/C batch with a more hydrophilic carbon support and Pt/Al₂O₃ (Table 2, entries 7 and 8). However, these catalysts rendered inferior conversions and carbon mass balances compared to the standard Pt/C catalyst. Considering the reaction performance and product distribution attained from applying this standard Pt/C catalyst at 160 °C, the main differences in comparison with the result of Hronec et al. [82] (Table 2, entry 9) concern the lesser catalytic activity and the seemingly absent suppression of the undesired over-hydrogenation of CPON to CPOL observed in my experiment (Table 2, entry 2). Meanwhile, vastly different product distributions were obtained in comparison with other results of Hronec et al. [82] for the reactions catalyzed by Pd/C and Ru/C. Overall, these widely diverse results underline that reproducing the selective heterogenously catalyzed hydrogenation of FAL to CPON is not a trivial occurrence.

Using the specific equipment and most optimal Pt/C catalyst available in our lab, further investigation was carried out, wherein the following reaction parameters were defined as the standard conditions: 80 bar H₂ as the desired reaction pressure that was maintained under continuous gas feed, a stirring rate of 1800 rpm, the pressure-aided injection (which defines t = 0) of an aqueous 10 v% FAL solution via a bypass vessel into a pre-heated 100 mL stainless steel autoclave, and a final FAL concentration of 5 v%. For clarity with respect to the graphs representing the reaction progression of these experiments (vide infra), ‘relative concentration’ on the y-axis refers to the initial amount of FAL at t = 0, which is normalized to 100% relative concentration.

The first optimization that was applied to this hydrogenative cascade transformation involved variation of the temperature in the range of 160–200 °C (Figure 1). The observed reaction compositions after exactly 60 min (through extrusion sampling) revealed clear trends of higher reaction conversions and a greatly improved selectivity of CPON over CPOL upon increasing the reaction temperature. However, the retracable mass balances in the GC-FID were dramatically reduced at higher temperatures, and poor CPON yields in the range of 20–30% were always obtained. The CPON/CPOL selectivity trend appears counter-intuitive, as this ratio is increased at higher temperatures, and the overall mass balances appear to be co-dependent consequences of the reaction temperature as the key parameter here. However, as mentioned before, Hronec et al. described how the thermally unstable reaction-intermediate FOL is prone to degrade under the given reaction conditions into polymers that precipitate on the catalyst surface and thereby suppress the hydrogenation of CPON into CPOL [109]. Thus, the observed product distributions in these temperature screening experiments are qualitatively in good accordance with the reported theory.

In order to gain a better understanding of the exact reaction profile over time, I then anticipated kinetic studies, but higher reaction volumes were required for sufficient sampling extrusion. Hence, first, the effect of the liquid phase volume in the 100 mL autoclave had to be investigated under the standard conditions and 160 °C (Figure 2). These experiments revealed a clear trend of lower FAL conversion upon increased liquid phase volume. A plausible reason for this effect is that higher liquid volumes in the same total volume of the autoclave inflict less turbulent mixing of all reaction components, which may cause a diminished H₂ concentration in the liquid phase. Hence, the lesser availability of dissolved H₂ in larger liquid reaction volumes slows the reaction down [188].

In further detail, the experiment with the smallest liquid volume (i.e., 21 mL) also rendered by far the highest CPON/CPOL ratio, while the absolute CPOL yields did not differ much throughout all three experiments. This observation suggests that the build-up of furanoic polymers on the catalyst surface—that typically inhibit the over-hydrogenation of CPON to CPOL—is much more prominent in the smallest liquid volume reaction as well. Herein, the rapid FAL conversion directly after reaction initiation generated a sudden large FOL concentration, which presumably was more prone to degrade into furanoic polymers than in the other two reactions.

Nevertheless, to pursue the kinetic studies next, a liquid phase volume of 63 mL was still selected to enable the possibility of collecting a sufficiently large number of samples over time. An important factor to understand is that the collection of samples reduces the liquid reaction volume in the autoclave over time, which is expected to inflict deviations with respect to the single-end-point experiments discussed before. In addition, a reaction temperature of 160 °C was maintained to moderate the conversion rate for the thorough monitoring of all reaction intermediates.

The cascade transformation of FAL was re-examined as such (Figure 3), and displayed a rapid conversion of FAL, a sharp but brief build-up of FOL in the first 20 min, and a more tempered build-up of the 4-HCP, 3-HCP, and CPEON intermediates during the first 60–90 min. Meanwhile, CPON was produced up to a 32% yield after 120 min, but then slowly degraded away into presumed oligomeric or polymeric adducts. CPOL was also already detected up to a 10% yield within the first 90 min, but, remarkably, its presence decreased after 30 min. Even though an equilibrium between CPON and CPOL seems highly unlikely under the firm hydrogenative conditions, its existence in this catalytic reaction was confirmed by Hronec et al. [82]. However, the observed complete disappearance of CPOL suggests its further conversion. According to some reports in the literature, CPOL can be catalytically dehydrated to cyclopentane in situ [115], but may also engage a variety of other adduct formations (e.g., ethers, acetals) with itself or with the other reaction components, in particular the furanoic compounds [18,112,189,190,191].

In a first effort to increase the CPON yield, the cascade transformation of FAL was conducted at a two-fold dilution to suppress the formation of undesired polymeric side products (Figure 4). This resulted in a similar overall reaction profile, while a commendably higher CPON yield of up to 58% was achieved. However, this strategy demands a significant toll on efficiency of the STY yield in the context of process designing.

Alternatively, the reaction’s performance was considered to benefit from the elimination of FOL and/or FAL from the reaction’s equation, whereas these furanoic compounds play a key role in the formation of undesired polymeric adducts.

The use of FOL as the substrate rendered a build-up of 28% CPON after 60 min, which was subsequently further hydrogenated to CPOL (Figure 5). Although the reaction still suffers from profuse carbon mass loss, the steady CPOL formation shows that the active sites on the Pt/C catalyst are not affected by furanoic polymer in the same way as seen for the reaction with FAL as the substrate (Figure 3 and Figure 4).

The reaction in which 4-HCP was used as the substrate proceeded very rapidly and gave a CPON build-up to 57% after only 20 min, which was subsequently converted completely to CPOL in a 76% yield after 90 min (Figure 6). The notably higher carbon mass balance of this reaction and its steadiness at prolonged exposure to the reaction conditions, compared to the other experiments that use FAL or FOL as a substrate, evidence that furanoic compounds indeed pose a major hurdle to maintaining an acceptable mass balance. Conversely, the absence of polymeric adducts from these furanoic compounds on the surface of the Pt/C catalyst is detrimental to the absolute CPON selectivity, as evidenced by the rapid and complete conversion of CPON to CPOL.

4.2. Control Hydrogenations of 4-HCP

Whereas the hydrogenation of 4-HCP as the substrate (Figure 6) produces CPON and CPOL at a significantly faster rate and higher carbon mass balance than the hydrogenations of FAL and FOL (Figure 3 and Figure 5, respectively), it sparked curiosity about the reaction progression and product distribution of 4-HCP hydrogenation under milder conditions. Reducing the reaction temperature could potentially reduce over-hydrogenation and thus lead to a more selective product distribution in favor of CPON. However, since furanoic compounds and their corresponding polymers are no longer present on the catalyst surface in this reaction, the reactivity and selectivity of the catalyst are now expected to be more impetuous.

Therefore, as a control study, 4-HCP was subjected to a series of mild hydrogenations at room temperature, in which several supported precious-metal catalysts were tested (Scheme 5 and

Table 3. Catalyst screening for the room-temperature hydrogenation of 4-HCP ^a.

Entry	Catalyst	Conversion 4-HCP (%) b	Product Yields (%) b				Mass b
			3-HCP	CPdiol	CPON	CPOL	Balance (%)
1	Ru/C	95.3	49.9	24.8	6.3	15.2	100.9
2	Pd/C	95.6	66.7	0.5	8.9	19.6	100.1
3	Rh/C	75.1	45.9	7.6	7.3	15.2	100.9
4	Pt/C	91.6	55.9	8.6	8.5	18.7	100.1
5 c	Pd/C	95.5	77.2	0.3	6.1	11.8	99.9
6	Pd/Al2O3	68.5	53.1	0.2	5.4	9.9	100.1
7	Pd/CaCO3	34.6	18.6	0	5.5	10.5	100.0
8	Pd/BaSO4	41.8	26.2	0	5.3	10.3	100.0

^a Reaction conditions: 50 mg 4-HCP and 2.0 mg catalyst in 1.00 mL D₂O, P = 50 bar H₂, t = 60 min; ^b determined by GC-FID via calibration curves; ^c Alternative Pd/C batch: Degussa type E101.

). In each reaction, the expected dehydration products CPON and CPOL were detected, wherein the concentration of CPOL was always double that of CPON. All carbon-supported catalysts rendered near-complete 4-HCP conversion, while the catalysts with different supports appeared less reactive, but a virtually complete mass balance was always maintained. Since the reaction duration was only 60 min, it is assumed that prolonged hydrogenation would ultimately afford complete conversion to CPOL. The fact that these conversions proceeded readily at room temperature suggests that the hydrogenation steps from 4-HCP to CPOL are not rate-limiting in the complete cascade transformation of FAL. Moreover, as the reactions were carried out using deuterated water as the solvent to conveniently perform ¹H-NMR analysis on the reaction mixtures after catalyst filtration, the presence of the reaction intermediate 3-HCP and the alternate product CPdiol could be verified (see Supplementary Materials Figures SM-3.A and SM-3.B). CPdiol was dominantly formed by Ru/C, and in lesser quantities by Rh/C and Pt/C, but was barely by supported palladium catalysts. The analytical data revealed very clean conversions to the main intermediates and products in these experiments, which further supports the suspicion of chemical interference by furanoic compounds in the reaction, for example aldol adduct formations between FAL and CPON [91,92,93,98,100,101].

In essence, the hydrogenation of 4-HCP already yielding CPOL at room temperature, but without carbon mass loss, further evidences the crucial role of furanoic polymers in acquiring a high selectivity to CPON instead of CPOL, while it also demonstrates the disturbing effect of furanoic compounds in terms of side reactions and decreased {CPON + CPOL} yield.

4.3. A Biphasic Water-Toluene Solvent System for Improved Down-Stream Processing

In view of the original aim to assess the cascade hydrogenation/dehydration of FAL to CPON in the context of industrial process development, further experimentation was focused again on the one-pot cascade transformation of FAL. Herein, the use of FAL would also enable the product selectivity to be in favor of CPON via the in situ catalyst poisoning from furanoic polymer deposits; however, an acceptable balance with the CPON yield and carbon mass balance is still an important requirement to be established at this point. The precedent literature demonstrates that (parts of) this chemical transformation are compatible with a biphasic solvent mixture, and could even benefit from it in terms of chemical selectivity, H₂ solvation, and DSP [128,145]. Herein, an aqueous part is essential for the Piancatelli rearrangement of FOL to 4-HCP to occur, while toluene was selected as the organic part to enable the facile azeotropic distillation of water in the DSP [75].

To investigate the effect of introducing an organic solvent to the reaction in detail, water/toluene ratios were screened in a broad range from fully aqueous to 75 v% organic. These experiments were at first carried out again in single-end-point reactions at 180 °C. After exactly 60 min, large volumes of the reaction mixture were extruded under vigorous stirring of 1800 rpm to afford homogeneous emulsification. After allowing the different solvent phases to stabilize, aliquots that were proportional to each layer were selected and combined for accurate GC-FID analysis. The introduction of toluene as an immiscibile organic co-solvent clearly inhibits the reaction progression (Figure 7a). Whereas the reaction in pure aqueous phase afforded 93% FAL conversion and 46% {CPON + CPOL} yield, the reactions with 25 v% and 50 v% toluene co-solvent both rendered 45% FAL conversion with combined {CPON + CPOL} yields of about 30%. However, with 75 v% toluene, the FAL conversion dropped below 20%. Moreover, the introduction of toluene appears to stimulate the over-hydrogenation of CPON to CPOL somewhat, which could be rationalized by the inhibited precipitation of furanoic polymers onto the catalyst surface due to enhanced solubility in the biphasic solvent mixture.

Moreover, aliquots of the aqueous phase and organic phase from these reactions were also analyzed separately by GC-FID. The corresponding results reveal that the protic compounds FOL, 4-HCP, and CPOL are present in considerably higher quantities in the aqueous phase, while, notably, FAL resides abundantly in the organic phase (Figure 7b,c). However, the water/toluene ratio appears to have a significant effect on the relative division of each reaction component over the solvent phases (

Table 4. Distribution of reaction components in biphasic water–toluene mixtures with different volume ratios.

Water–Toluene Volume Ratio	[caq/ctol] Per Reaction Component
	FAL	FOL	4-HCP	CPEON	CPON	CPOL
75:25	0.63	N/A	2.99	2.56	0.58	10.51
50:50	0.23	0.97	N/A	0.86	0.20	3.15
25:75	0.07	N/A	0.22	0.25	0.10	0.99

). The 50:50 water/toluene ratio gives, overall, the optimally balanced performance in terms of the conversion, the {CPON + CPOL} yield, and the separation of CPON from the other reaction components into the organic phase. By anticipating the complete conversion of FAL upon an extended reaction time, through which a product mixture predominantly consisting of CPON and CPOL would be afforded, the phase separation of a biphasic reaction mixture with a 50:50 volume ratio of its solvents would conveniently lead to a CPOL-rich aqueous solution and a CPON-rich organic solution. Hence, these initial experiments indicate a promising potential of the biphasic solvent concept for the DSP of the cascade transformation of FAL.

The further reaction progression towards complete FAL conversion in a biphasic solvent system was investigated by means of kinetic studies as well. For these experiments, a 50:50 volume ratio of water–toluene was selected, and the standard reaction parameters were applied, with the exception of the higher reaction temperature of 180 °C, to compensate for the inhibited conversion rate caused by the implementation of toluene solvent. First, a reference kinetic experiment in a fully aqueous solution was conducted under these reaction conditions (Figure 8), which showed a similar reaction profile as for the reaction at 160 °C (Figure 3), but at a faster rate. A notable difference is the slow over-hydrogenation of CPON to CPOL that occurs after 60 min, which is seemingly induced by the elevated reaction temperature.

When the reaction was conducted in the biphasic solvent system, the conversion rate of FAL was notably inhibited, as expected (Figure 9). However, the overall production rate of CPON was less affected, as it reached 46% after 120 min. During the extended reaction time of up to 180 min, the CPON concentration remained appreciably stable, and the formation of CPOL was ultimately negligible in this experiment. Although this observation stands in contrast with the results of the corresponding single-end-point biphasic hydrogenations that are described in Figure 7a, the same kind of discrepancy is also found between the single-end-point experiments and kinetic studies for the fully aqueous cascade transformations of FAL (Figure 1, Figure 2, Figure 3 and Figure 4). Clearly, as already shown in Figure 2 and Figure 7a, variation of the liquid phase volume of the reaction mixture in the fixed total volume of the autoclave has a severe influence on the product distribution, and this effect also applies to kinetic studies due to the collection of samples over time.

Furthermore, the effect upon the kinetic reaction profile and final CPON yield that is achieved by diluting the FAL concentration to 2.5 v% in a biphasic solvent system was investigated as well (Figure 10). Similar to the reaction profile of its fully aqueous analogue experiment (Figure 4), a slightly improved CPON yield of 59% was achieved after 180 min. Contrarily, this CPON yield did not degrade at any monitored time point, and a CPOL yield of 11% was attained in addition, with a substantially higher mass balance being rendered for the diluted FAL conversion in a biphasic solvent system. A strange observation, however, was the exceptionally high FOL build-up of up to 42% during the first 60 min of the reaction. While suppressed FOL conversion is plausible if the FOL favors residence in the organic phase over the aqueous phase as a result of the substrate dilution—since water is essential for its further conversion via the Piancatelli rearrangement—no certain rationale for this anomaly could be formulated.

5. Conclusions

The one-pot cascade transformation of FAL to CPON that proceeds through several intermediates that involve several hydrogenation, rearrangement, and dehydration steps was investigated. The aim of this study was to establish a reproducible and high-yielding protocol, in order to lay the experimental foundation for scale-up in a pilot batch-reactor to achieve the industrial manufacture of CPON as a bio-based bulk product.

However, various attempts to reproduce this chemical transformation in accordance with the leading literature report [82] proved to be extremely difficult. The experiments conducted on bench scale indeed showed that supported platinum is the most suitable catalyst (among the commercially available types) to yield CPON as the main product, but the reactions were always accompanied by significant over-hydrogenation towards CPOL and unacceptably large carbon-mass losses. In closer detail, the mixture of various reaction components that are highly susceptible to forming undesired adducts was an evidenced cause hereof, with the self-polymerization of FOL being the largest contributor [109]. However, a complicating feature regarding the resulting furanoic polymer is its ability to poison the Pt/C catalyst in a remarkable and desirable way, so that over-hydrogenation from CPON to CPOL is strongly suppressed. In other words, to obtain CPON selectively through hydrogenative chemistry, a significant part of its yield seemingly must be sacrificed.

Although strategic variations of the reaction parameters, such as the temperature, substrate concentration, overhead space, and use of aqueous–organic biphasic reaction media, led to moderate improvements in the CPON yield and the carbon-mass balance, most of these variations worked at a severe cost to the STY yield, and CPON was obtained in only a 59% yield at highest. Hence, the cascade transformation of FAL to CPON using some standard commercially supported precious-metal catalysts suffers greatly from intrinsic limitations of a chemical nature and is therefore not suitable for a process development nor scale-up trials in its current form. Whereas the crux of balancing the CPON yield versus the process efficiency is encountered for many other catalytic systems reported in the literature as well, this critical evaluation extends beyond the experimentation described in this work.

For future endeavors, a three-step approach for the conversion of FAL to CPON is recommended, and should comprise the facile catalytic hydrogenation of FAL to FOL, the aqueous non-catalytic Piancatelli rearrangement of FOL to 4-HCP, and finally the selective catalytic hydrogenation of 4-HCP to CPON. Despite the latter step currently still being rather under-explored, the findings in this work warrant conversion under mild reaction conditions, while selective CPON formation therein must still be established.