1 Introduction

Menthol is an organic compound with fresh and cooling sensation being widely used in pharmaceutical, cosmetic and flavoring applications such as ointments, cough drops, toothpaste, mouthwash, chewing gum, candy and perfume [1]. Menthol as a chiral compound can be found with eight possible stereoisomers including: ( ±)-neomenthol, ( ±)-menthol, ( ±)-isomenthol and ( ±)-neoisomenthol. Among them, ( −)-menthol has the most effective cooling sensation [2]. Naturally, menthol was obtained from peppermint essential oil separation, however due to the increasing demand, synthetic menthol was produced by two companies, Haarmann & Reimer and Takasago International Corp. In Haarmann & Reimer process, m-cresol was converted to thymol by propylation followed by hydrogenation to racemic ( ±)-menthols in the presence of Ru/Al2O3 catalyst. Thereafter, ( −)-menthol was obtained from racemic ( ±)-menthols by separation crystallization process [3]. Takasago process was based on conversion of myrcene to ( −)-menthol in the presence of Rh- BINAP catalyst [4, 5]. Citral and its derivatives, especially citronellal have attracted significant attention in one-pot menthol synthesis in the recent year [1, 6,7,8,9,10,11,12,13,14,15,16]. Citral can be extracted as an essential oil from distillation of lemongrass oil possessing ca. 70–80% citral inside [11]. In 2004, BASF established a new continuous process for citral production in Ludwigshafen, Germany with annual capacity of 40,000 metric tons [17]. Citral can be selectively hydrogenated to citronellal using metal-supported catalysts. In addition, BASF has also established a technology for production of ( +)-(R)-citronellal by direct catalytic hydrogenation of neral or geranial [18]. Hence, in the current work, citronellal was used as a model reactant, and its catalytic transformations to menthol in a batch system were assessed.

Menthol synthesis from citronellal (Scheme 1) proceeds via two steps including citronellal cyclization to pulegols over the acid catalyst followed by their hydrogenation to menthols in the presence of an active metal phase. According to the reaction mechanism, four different enantiomeric pairs of isopulegols and menthols can be produced. In addition, occurrence of side reactions such as hydrogenation of citronellal, dimerization and polymerization, and defunctionalization of citronellal and isopulegol has to be minimized [10].

Scheme 1
scheme 1

One-pot synthesis of menthol from citronellal

Full size image

Hence, selection of an appropriate catalyst to control the reaction pathways toward the desired product is of crucial importance. Due to the apparent drawbacks of homogeneous catalysts such as challenging recovery and separation, along with environmental and economic issues [19], heterogeneous catalysts have attracted much more attention. Various heterogeneous catalysts such as ZrO2, MCM-41, SiO2, γ-Al2O3 and zeolites were studied for efficient cyclization of citronellal [10, 12, 16, 20,21,22]. Although majority of the acid catalysts led to high yields of isopulegols, textural and acidic properties of the catalysts were the main factors allowing high stereoselectivity towards (−)-isopulegol. In addition, for the second reaction step, various supported noble and non-noble metals catalysts including Ir, Pd, Pt, Ru, Cu, Ni, Co as an active metal have been used. There are studies in which Beta type zeolites have been used as the catalyst support for the desired reaction [9, 15, 23,24,25]. Beta zeolite, from BEA topology, is formed by corner-sharing TO4 (T = Si or Al) tetrahedra and is a three-dimensional (3D) crystalline material with three sets of vertical channels and 12-membered ring porous structure. Brønsted acid sites of Beta zeolite are generated after an aqueous ion-exchange using NH4Cl solutions with subsequent calcination releasing ammonia [26,27,28].

Plößer et al. performed one-pot transformation of citronellal to menthol using bifunctional metal–acid heterogeneous catalysts, typically Pd, Pt and Ru as the active metals and H-Beta zeolite as the support. Among different catalysts, 1 wt.% Ru/H-Beta-25 was the best one giving the highest menthol selectivity exceeding 93% [15]. In addition, ca. 70% stereoselectivity towards ( ±)-menthol was obtained using Pt- and Ru- supported H-Beta-25 catalysts both in trickle bed and batch reactors [9]. Noble metals have been used as the catalytically active phases in several other studies exhibiting high activity and selectivity [7, 8, 29, 30]. Moreover, low-cost non-noble metals especially Ni were reported to transform both citral and citronellal to menthol with a high selectivity [23, 24, 31]. For instance, either a bifunctional Ni/Zr-Beta or a dual catalytic system of Zr-Beta and Ni/MCM-41 demonstrated 86–97% yield of menthol with 90–94% diastereoselectivity to ( ±)-menthol from citronellal. The maximum yield of isopulegol in the first step of the reaction was obtained under He atmosphere in the presence of the acid sites of Beta zeolite with modified Lewis acidity by exchanged Zr. Thereafter, hydrogen was introduced to the system and hydrogenation of isopulegol to menthol was catalyzed by Ni at 80 °C and 20 bar H2 [23, 24]. A similar strategy was followed in [14] using Ni/γ–Al2O3 as the catalyst. The hydrogenation step was carried out under very mild reaction conditions of 90 °C and 1 bar H2 giving 86% yield of menthols [21]. In addition, one-pot synthesis of menthol from citronellal was catalyzed by sulfated zirconia supported Ni under 14 bar of H2 and at 100 °C, with enantioselectivity of 88% towards ( −)-menthol analyzed with an α-DEX Supelco chiral column [14]. However, sulfated zirconia pillared montmorillonite supported Ni, Ni/ZSM-5 and Ni/natural zeolite exhibited very low activity toward menthol in [6, 16]. One-pot transformation of citral to menthols over different bifunctional catalysts including Ni and Pd as the active phases and H-BEA, SiO2–Al2O3, and Al-MCM-41 as the supports was studied by Trasarti and co-workers [25]. It was suggested that Ni active metal and the solid acid support with mild acidity (Al-MCM-41) could be more active and selective to racemic ( ±)-menthol giving more than 90% yield of menthols (71% stereoselectivity to ( ±)-menthol) at full citral conversion at 70 °C, 20 bar H2, 1 g catalyst, citral/toluene = 2:150 (ml), 5 h.

From the overview given above it is apparently clear that a delicate balance between acidity of a zeolite bearing a pore size favorable for the desired products and the hydrogenation ability of a non- noble cost efficient Ni is required. In the current study beta zeolites with different acidity and Lewis to Brønsted acid sites ratio have been selected as supports for nickel. Loading of latter was varied to adjust the metal dispersion and explore a potential synergy between H-Beta acidity and Ni loading and dispersion to enhance the yield of the main cyclic product ( ±)-menthol at the same time minimizing the undesired side reactions, which also require the acid sites. Such types of neat H-Beta zeolite supported Ni have never been reported as the catalyst of one-pot transformation of citronellal to menthols in the presence of molecular hydrogen. Subsequently an in-depth study on the synergetic effect of Ni active phase and H-Beta acid sites nature, strength and concentration on the catalytic behavior has been performed. In particular, H-Beta zeolites with two different SiO2/Al2O3 molar ratios, 25 and 300 and Ni loading of 5, 10 and 15 wt% were investigated allowing to correlate the catalysts properties with their reaction performance.

2 Experimental

2.1 Chemicals

The following chemicals were used in the current work: cyclohexane (Alfa Aesar, ≥ 99.9 wt%), racemic ( ±)-citronellal (Sigma-Aldrich, ≥ 95.0 wt%), menthol (Aldrich, ≥ 99.0 wt%), geraniol (Lancaster, tech. quality), nerol (Sigma-Aldrich, ≥ 97.0 wt%), iso-menthol (Molekula), neomenthol (Fluka, ≥ 97.0 wt%), citronellol (Fluka, ≥ 99.0 wt%), isopulegol (Fluka, ≥ 99.0 wt%).

2.2 Catalyst Preparation

Two different commercial zeolites namely NH4-Beta-25 (theoretical SiO2/Al2O3 = 25 molar ratio) (CP814E) and H-Beta-300 (theoretical SiO2/Al2O3 = 300 molar ratio) (CP811C) were provided from Zeolyst International. Ammonium form NH4-Beta-25 was transferred to H-beta-25 by stepwise calcination at the final temperature of 400 °C for 4 h. Ni modified H-Beta zeolite supports were prepared by the evaporation impregnation method using nickel (II) nitrate hexahydrate (Ni(NO3)2.6H2O, Merck, 99.999%) as the Ni precursor. Different catalysts with nominal Ni-metal loadings of 5, 10 and 15 wt % on H-Beta-25 and H-Beta-300 were synthesized. First, the Ni precursor was dissolved in 250 mL distilled water in a flask and the desired amount of H-Beta support with the sieved fraction below 63 μm to avoid internal mass transfer limitations was added to the solution. The impregnation was carried out in a rotary evaporator with a rotational speed of 10 rpm at room temperature for 24 h. Thereafter, the aqueous solution was evaporated in the rotavapor using a water jet vacuum pump at 50 °C. The obtained catalyst powder was dried at 100 °C for 12 h and calcined in static air at 400 °C for 3 h. The final catalysts were once more sieved below 63 μm. The catalysts were labeled 5Ni/H-Beta-25 (5 wt.% Ni/H-Beta-25), 10Ni/H-Beta-25 (10 wt.% Ni/H-Beta-25), 15Ni/H-Beta-25 (15 wt.% Ni/H-Beta-25), 15Ni/H-Beta-300 (15 wt.% Ni/H-Beta-300).

2.3 Catalyst Characterization

The effective amount of Ni was analyzed by inductively couple plasma–optical emission spectrometry (ICP-OES, PerkinElmer Optima 5300 DV instrument). About 0.1 g of the catalyst was microwave digested with a mixture containing 9 mL of 37% HCl, 3 mL of 65% HNO3 and 1 mL of 50% HBF4. After digestion, dilution to 100 mL was done using distilled water.

The X-ray diffraction (XRD) characterization was performed using PANalytical Empyrean diffractometer with five axis goniometers. The incident beam optics consisted of Bragg–Brentano HD X-ray mirror, fixed ¼° divergence slit, 10 mm mask, 0.04 rad soller slit and 1° anti scatter slit. The diffracted beam optics consisted of 7.5 mm divergence slit, 0.04 rad soller slit and PIXcel detector array. The used X-ray tube was Empyrean Cu LFF. The X-ray radiation was filtered to include only Cu Kα1 and Cu Kα2 components. The results were analyzed with MAUD (Material Analysis Using Diffraction) analysis program [32]. Instrumental broadening was evaluated with Si standard sample. The results were obtained with θ − 2θ scan range from 5 to 120°.

Scanning Electron Microscopy (Zeiss Leo Gemini 1530) was used to investigate the surface morphology and the shape of the fresh and spent catalysts. 10–15 mg of the catalyst was used for analysis on a thin film coated with activated carbon. SEM images were obtained by using an accelerating voltage of 2.7 kV with a working distance of 2–7 mm.

Transmission Electron Microscopy (TEM, Jeol JEM-14000 plus) was carried out to determine the Ni metal particle size distribution. ImageJ software was used to determine the particle size of Ni from TEM images.

Textural properties were studied by nitrogen physisorption (Micrometrics 3Flex-3500) at − 196 °C. The specific surface area and porosity (pore volume, pore size distribution) were calculated using the Dubinin-Radushkevich (DR) and the density functional theory methods (DFT), respectively, while for the average micropore width and the mesopore diameter the Horvath-Kawazoe and Barrett –Joyner-Halenda methods were used.

Temperature Programed Reduction (TPR) was carried out with a lab-made instrument under 5 vol% H2 in Ar flow, which was heated from 25 to 800 °C with a ramp of 10 °C/min. The effluent gases were analyzed by a Micrometrics TPDTPR 2900 analyzer equipped with a TCD detector (Gow-Mac 24-550 TCD instrument CO, Bethlehem, PA, USA).

Characterization of the Brønsted and Lewis acid sites, their amount and strength were performed using pyridine (Sigma Aldrich, ≥ 99.5%) adsorption–desorption with FTIR method. The measurements were performed with ATI Mattson FTIR using ca. 20 mg of the catalyst pellet. The catalyst was pretreated in the IR cell by heating from room temperature to 450 °C under vacuum (1 × 10−4 Pa). In order to discriminate between weak, medium and strong acid sites, desorption of pyridine was performed at 250 °C, 350 °C and 450 °C, respectively. Pyridine was desorbed for 60 min at each temperature and the spectra were taken at 100 °C. Quantification of Brønsted and Lewis acid sites was done by considering intensity of IR signals at 1545 cm−1 and 1455 cm−1, respectively, using the molar extinction factor given by Emeis [33].

2.4 Catalytic Tests

First, the desired weight of calcined catalyst was charged in an ex-situ glass tube to be reduced under H2 flow of 50 mL/min with the temperature program of 25 °C—ramping 2 °C/min to 400 °C and holding at this temperature for 3 h. After reduction, the system was cooled down to the ambient temperature. To avoid the reduced catalyst oxidation, it was sealed in the reduction tube by adding 10 mL of the reaction solvent, which was cyclohexane in this work.

Citronellal transformations to menthols were performed by loading 400 mg of citronellal, 300 mg of the pre-reduced catalyst with the particle size below < 63 μm and 90 mL of cyclohexane as the solvent into an autoclave. Before the reaction, the reactor was flushed with a mixture of nitrogen (≅ 95%) and argon (≅ 5% Ar) supplied by AGA for 10 min. Thereafter, the reactor temperature was increased to the desired level (50, 60 and 80 °C) under H2 pressure of 20 bar (99.99%, AGA). The reaction, which duration was 3 h, commenced after reaching the desired temperature by initiating stirring with the rate of 900 rpm. The small catalyst particle size and high stirring speed were used to suppress external and internal mass transfer limitations. The liquid samples of the reaction mixture were taken at certain intervals, diluted with cyclohexane and analyzed by an Agilent GC 6890 N. The GC was equipped with an FID detector at 340 °C and DB-1 column (length 30 m, internal diameter 250 μm and film thickness 0.5 μm). The following temperature program was used for GC-analysis: 110 °C—0.4 °C/min—130 °C—13 °C/min—200 °C (held 5 min). The types of products were confirmed with an Agilent GC/MS 6890 N/5973 N using a DB-1 column (length 30 m, internal diameter 250 μm and film thickness 0.5 μm). For catalyst regeneration, the final solution of the reaction after 3 h was filtered and the solid catalyst on the filter was collected and washed several times with cyclohexane. Thereafter, the spent catalyst was dried at 100 °C for 12 h, reduced with a similar procedure as for the fresh one with the aim to reduce NiO formed during the catalyst recovery, and subsequently used in the second reaction run.

The citronellal conversion, the yield and selectivity, the reaction rate and turnover frequency were calculated using the following equations:

$${\text{X }}\left( \% \right) \, = \frac{{C_{0} - C_{i} }}{{C_{0} }}\, \times {1}00$$
(1)

where X is the conversion of citronellal at time t, C0 and Ci correspond to the molar concentration (mol L−1) of citronellal at time zero and t, respectively,

$${\text{S}}_{{\text{p}}} \left( \% \right) \, = \frac{{C_{p} }}{{\sum \left( {C_{a} + C_{b} + C_{c} + \ldots + C_{z} } \right)}} \times {1}00$$
(2)
$$Y_{p} \,\left( \% \right)\, = \,\frac{{C_{p} }}{{C_{o} }}\, \times \,100$$
(3)

where Sp and Yp are the selectivity and yield of the product p at a certain conversion, respectively, Cp and (Ca + Cb + Cc + ⋯ + Cz) correspond to the molar concentrations (mol L−1) of the product p and all products at the same conversion, respectively,

$$r_{0} = \frac{\Delta n}{{\Delta t.m_{cat} }}{\text{mol s}}^{{ - {1}}} {\text{g}}^{{ - {1}}}$$
(4)
$${\text{Initial TOF }} = \frac{\Delta n }{{\Delta t.n_{metals} }}{\text{s}}^{{ - {1}}}$$
(5)

where r0 is the initial reaction rate and initial TOF is the initial turnover frequency, Δn/Δt corresponds to the number of reacted moles per time interval between 1 and15 min, mcat is the mass of catalyst, nmetals is the moles of exposed Ni active metal ((moles of Ni in the catalyst) \(\times\) (dispersion/100)).

3 Results and Discussion

3.1 Catalyst Characterization

The ICP results show that the Ni loading of all catalysts was very close to the nominal value (Table 1). The N2 adsorption–desorption isotherms of the parent H-Beta-300 and H-Beta-25 zeolites and the respective catalysts loaded with Ni are shown in Fig. 1. A combination of I and IV type isotherms can be seen for all samples reflecting the coexistence of micropores (ca.80 vol% for Beta-300 catalysts and ca. 30 vol% for Beta-25 catalysts) and mesopores in all materials similar with the results reported in the literature [34]. In particular, a sharp increase of the uptake displayed at a low P/P0 pressure is assigned to the microporous structure of materials, whereas the hysteresis loops at about 0.5 < P/P0 < 0.9 and 0.6 < P/P0 < 0.9 are related to the inter-crystalline mesopores of H-Beta-300 and H-Beta-25, respectively [35]. After impregnation of Ni on both zeolites, the shape of isotherms of catalysts stayed unchanged suggesting that the crystalline structure and the zeolites textures were maintained. According to the data in Table 1 and pore size distribution of Fig. 1c,d, by impregnation of Ni on both H-Beta-300 and H-Beta-25 zeolites, a decrease of surface area and pore volume occurred as could be anticipated after Ni introduction into zeolites [36]. Moreover, a further decrease of H-Beta-25 surface area by increasing Ni loading was related to a partial blockage of primary micro-mesopores [37]. Analysis of the textural properties of the 15Ni/H-Beta-25 catalyst after the first and second catalytic run (Fig. 1, Table 1), 5Ni/H-Beta-25, 10Ni/H-Beta-25 and 15Ni/H-Beta-300 after first catalytic run (Table 1) allowed to conclude that the fresh and spent 15Ni/H-Beta-25 catalysts exhibited similar isotherms, and subsequently the catalyst texture was maintained during the reaction. However, surface areas and the micropores volume of all spent catalysts decreased, while mesopores volume increased compared to the fresh ones. This could be described by the blockage of some pores of the catalysts with agglomerated Ni particles, polymeric solid byproducts and also by coke formation [9].

Table 1 Ni loading (determined by ICP-OES) and textural properties (obtained from N2 physisorption) for studied catalysts
Full size table
Fig. 1
figure 1

N2 adsorption–desorption isotherms and pore size distribution of a,c H-Beta-300 and 15Ni/H-Beta-300; and b,d H-Beta-25, 5Ni/H-Beta-25, 10Ni/H-Beta-25, 15Ni/H-Beta-25 and 15Ni/H-Beta-25-spent, respectively

Full size image

TPR profiles show a complete reduction of NiO in all catalysts at the temperature below 650 °C (Fig. 2). The shape, area and temperature of the peaks in TPR profiles of different catalysts were related to Ni loading, particles size, dispersion and metal interaction with H-Beta support (Fig. 2, Table 2).

Fig. 2
figure 2

Hydrogen TPR profiles of the catalysts

Full size image
Table 2 Results from hydrogen TPR
Full size table

In all catalysts, the first peak at the temperature lower than 400 °C is related to the reduction of NiO to Ni0 mostly located on the surface of H-Beta support with weaker interactions [39]. 5Ni/H-Beta-25 and 10Ni/H-Beta-25 exhibited a second reduction peak at 480 °C and 450 °C, respectively, while for all catalysts a reduction peak with different intensity was present at the temperature higher than 500 °C. These peaks could be related to the reduction of NiO in the intracrystalline mesopores or isolated Ni2+ inside the pores of the H-Beta support [37, 40]. The location of metal particles inside the zeolite pores could be strongly related to their loading, size and dispersion. It was suggested that during catalyst impregnation with a low Ni loading, some Ni2+ could be ion-exchanged with the protons of hydroxyl groups in the H-Beta channels wall and Ni is thus stabilized close to the zeolite sites. In addition, a part of the intracrystalline NiO during calcination could react with bridging protons (Si–OH–Al) of H-Beta forming Ni(OH)+ [41, 42]. Since the isolated Ni2+ ions, NiO and also Ni(OH)+ located in the zeolite pores exhibited strong interactions with H-Beta structure, their reduction was more difficult and occurred at a higher temperature [43]. For this reason, 5Ni/H-Beta-25 and 10Ni/H-Beta-25 with lower Ni loadings, smaller Ni particle sizes and higher dispersions (Fig. 2 and Table 2) exhibited two reduction peaks appearing at the higher temperatures. In addition, a low Ni loading in 5Ni/H-Beta-25 led to the smallest peak area in H2-TPR and lowest H2 consumption (Table 2).

Relatively more intracrystalline NiO particles might exist in 15Ni/H-Beta-25 for which only one reduction peak was visible in TPR profile at a higher temperature (558 °C). It was reported that a higher Ni loading can lead to a progressive segregation of NiO phase from zeolite surface to the bulk [40]. On the other hand, 15Ni/H-Beta-300 with the highest Ni loading exhibited the largest H2-TPR peak area (Table 2) and a different TPR profile. The intensity of the first reduction peak at 390 °C was very high and the second reduction peak at 550 °C had a very low intensity. It can be suggested that majority of NiO particles had weak interactions with the H-Beta-300 support and also that the ion-exchange was less prominent. H-Beta with a very high SiO2/Al2O3 molar ratio of 102 exhibits a hydrophobic nature and a very low Al content in the structure diminishing NiO interactions with the support [44].

The XRD patterns of 5Ni/H-Beta-25 and 15Ni/H-Beta-25 catalysts are presented in Fig. 3 illustrating that for both catalysts, there are two main reflexes at 2Ө of 7.8° and 22.3° which are typical for Beta zeolites with a well crystalline structure in line with the literature [45, 46]. The reflexes at 2Ө of 44.4°, 51.8°, 76.4° and 93° are ascribed to metallic Ni0. The intensities of the lines related to metallic Ni0 enhanced by increasing the nickel loading. In addition, a small shoulder in the XRD pattern of 15Ni/H-Beta-25 and one of the double-line in XRD pattern of 5Ni/H-Beta-25 located at 2Ө = 43.4° correspond to intracrystalline NiO and demonstrate that the reduction at 400 °C may not completely convert Ni2+ to metallic Ni0 or reoxidation of Ni0 in the air might happen prior to XRD measurement leading to the presence of a small amount of NiO particles in the catalysts [47,48,49]. This result agrees well with TPR results (Fig. 2). Thus, it seems that both Ni0 and NiO (Ni-NiO) were present in the final catalysts being essential for the catalytic reaction. In addition, a comparison between the XRD profile of the support with the catalysts shows that some amorphization occurred in zeolite during the catalyst preparation. It can be assumed that partial amorphization might happen either at the step of the aqueous solution evaporation under vacuum at 50 °C or most probable at the calcination step. In addition, TPR profiles of the catalysts suggested strong interactions of some NiO and Ni2+ ions with the zeolitic structure which might lead to partial amorphization of the zeolite structure at the calcination temperature.

Fig. 3
figure 3

High-angle XRD patterns of the reduced 5Ni/H-Beta-25 and 15Ni/H-Beta-25 catalysts; (■ H-Beta-25, ✰ Ni.0, ★ NiO)

Full size image

The SEM images of H-Beta-25, 5Ni/H-Beta-25 and 15Ni/H-Beta-25 samples in Fig. 4a–c display that H-Beta-25 has a semi-spherical or round shaped particles typical for Beta-zeolite crystal morphology [50]. However, by incorporation of Ni, especially in 15Ni/H-Beta-25 catalyst with a higher Ni loading, the grain surface of H-Beta zeolite became rough due to the amorphization as was also demonstrated by XRD result. A comparison between the SEM image of the 15Ni/H-Beta-25-spent catalyst with that of the fresh one (Fig. 4c,d) shows that the morphology of Beta-zeolite was almost maintained after the reaction. The SiO2/Al2O3 (mol/mol) ratios of H-Beta-25 and H-Beta-300 were measured by EDX analysis in our previous study [38]. The theoretical SiO2/Al2O3 ratio for H-Beta-25 was close to the experimentally measured (SiO2/Al2O3 = 25), while for H-Beta-300 with the theoretical SiO2/Al2O3 ratio of 300, the experimentally measured one was close to 100.

Fig. 4
figure 4

SEM images of a H-Beta-25, b 5Ni/H-Beta-25, c 15Ni/H-Beta-25 and d 15Ni/H-Beta-25-spent. (The magnification of all images is 25.00 K X)

Full size image

Comparison of the TEM images of the four fresh and reduced catalysts with different Ni loadings and supports (Fig. 5) shows clearly that the particle size distribution for the catalysts with lower metal loadings is narrower due to the presence of smaller Ni-NiO particles, which are less agglomerated. In addition, an increase of the Ni loading led to the enhancement of the average particle size and thus lower dispersion (Table 3). As was confirmed by TPR and XRD results (Fig. 2, Fig. 3), some non-reduced exchanged Ni2+, intercrystallite NiO and Ni(OH)+ could be present besides metallic Ni0 in the final catalyst. It can be assumed that larger particles in the TEM images of each catalyst could originate from the intercrystallite NiO as was reported by Wang and co-workers [37]. This trend was also visible in other studies reported in the literature [51,52,53]. In addition, according to TEM for 15Ni/H-Beta-25-spent in Fig. 5c, a slight agglomeration of Ni-NiO nano particles occurred through the reaction increasing the average particle size from 15 nm to 18.4 nm and at the same time particle dispersion decreased from 6.7 to 5.5%. More importantly, the fraction of the Ni-NiO particles bellow 5 nm decreased very much.

Fig. 5
figure 5

TEM images and Ni-NiO particle size distribution histograms of reduced a 15Ni/H-Beta-300-fresh, b 15Ni/H-Beta-25-fresh, c 15Ni/H-Beta-25-spent, d 10Ni/H-Beta-25-fresh, and e 5Ni/H-Beta-25-fresh

Full size image
Table 3 Average Ni-NiO particle size and dispersion determined from TEM images
Full size table

The concentration of weak, medium and strong Brønsted and Lewis acid sites of the supports and catalysts was studied by FTIR spectroscopy in the spectrum range of 1400–1575 cm−1 using pyridine as the probe molecule. In particular, the weak, medium, and strong acidity was estimated from pyridine desorption for 60 min at each 250 °C, 350 °C and 450 °C, respectively, with quantification of Brønsted and Lewis acid sites at each temperature relying on intensity of IR signals at 1545 cm−1 and 1455 cm−1, respectively (Fig. 6). As can be seen in pyridine-FTIR results, the Brønsted acid site concentration decreased and the Lewis acid site concentration increased when Ni was loaded on zeolite (Table 4). The decrease of Brønsted acid site concentration could be due to the exchange of protons of the hydroxyl groups in the zeolite channels wall with Ni cations. Moreover, the enhancement of Lewis acid site concentration might be related to incorporation of non-reduced exchanged Ni2+ and NiO, and creation of new three-coordinated Al in the zeolite structure during the catalyst preparation step leading to an increase of zeolite amorphous phase as was displayed in XRD patterns (Fig. 3) and SEM images (Fig. 4) of the catalysts. This result agrees well with the previous work of the authors [55]. H-Beta-300 support with the SiO2/Al2O3 molar ratio of 102 (determined by EDX) and 15Ni/H-Beta-300 catalyst exhibited lower concentration of both Lewis and Brønsted acid sites compared to H-Beta-25 with SiO2/Al2O3 molar ratio of 25 and the 15Ni/H-Beta-25 catalyst, respectively. Lewis acidity of the zeolite comes from framework Al located on the edges and also extra-framework Al, while the Brønsted acid sites originate from H+ ions of hydroxyl groups bridging aluminum and silicon of the channel walls [56]. Therefore, by increasing the SiO2/Al2O3 ratio and hence decreasing both framework and extra-framework Al, both Lewis and Brønsted acidities decrease as could be anticipated and reported numerous times [57, 58]. In addition, by decreasing the Ni loading from 15 to 10% and further to 5% in H-Beta-25 catalysts, a slight increase of Brønsted acid sites concentrations and considerable increase of Lewis acid sites concentrations occurred. This could be related to better accessibility of H-Beta-25 Brønsted and Lewis acid sites when there are less Ni particles blocking the pores and preventing such access. Since catalyst acidity is important for cyclization of citronellal to pulegols [59, 60], it was expected that catalysts with different acidities would display different performances.

Fig. 6
figure 6

FT-IR spectra of pyridine adsorption/desorption for 15Ni/H-Beta-25

Full size image
Table 4 Concentration of Brønsted acid sites (B) and Lewis acid sites (L) determined by FTIR with pyridine
Full size table

3.2 Catalytic Results

The results for one-pot transformation of citronellal in the presence of 15Ni/H-Beta-300, 15Ni/H-Beta-25, 10Ni/H-Beta-25 and 5Ni/H-Beta-25 catalysts are reported in Table 5 and Figs. 7–12. The relative error of GC analysis was generally less than ± 5%.

Table 5 Catalytic reaction results of one-pot transformation of citronellal to menthol. Conditions: 80 °C, 20 bar H2, solvent: cyclohexane, amount of catalyst: 300 mg, the initial citronellal concentration: 0.029 M. Conversion of citronellal (X) is given at 3 h of reaction time. Yields (Y) of different products, stereoselectivity (SS), are given at 85% conversion out of round bracket and after 3 h of the reaction in the round bracket
Full size table
Fig. 7
figure 7

One pot transformation of citronellal to menthol: a The initial TOF as a function of Ni-NiO average particle size and b conversion vs time for 15Ni/H-Beta-300 (red ■), 15Ni/H-Beta-25 (blue ●), 10Ni/H-Beta-25 (pink ▲), 5Ni/H-Beta-25 (green▼). Conditions: 80 °C, 20 bar H2, solvent: cyclohexane, amount of catalyst: 300 mg, the initial citronellal concentration: 0.029 M

Full size image

A comparison of the initial reaction rates and the initial TOFs for different H-Beta-25 supported Ni catalysts presented in Table 5 shows that the most rapid initial reaction and the highest TOF were observed for 5Ni/H-Beta-25 giving a rate of 1.6.10–6 mol s−1 gcat−1 and TOF of 0.035 s−1. This catalyst exhibited the highest acidity among the studied materials and the smallest Ni-NiO particle size (Tables 3 and 4). On the other hand, a high initial TOF was also obtained over 15Ni/H-Beta-300, which exhibited mild acidity and relatively large Ni-NiO particles, low exposed Ni metal (mole) with a higher Ni loading.

The largest peak area in H2-TPR was observed for 15Ni/H-Beta-300 (Table 2), which, however, exhibited the lowest initial rate. It can be thus stated that the initial transformation rate for citronellal does not only depend on the Ni-NiO particle size, but also on acidity. In particular, citronellal cyclization as well as side reactions, such as etherification, are known to be acid catalyzed [12]. Moreover, as the initial activity of 15Ni/H-Beta-300 is the lowest, the observed behavior is a result of a combined effect of the metal particles and mild acidity. Ni particles were rather large not promoting hydrogenation and under the studied conditions the main reaction in cyclohexane, a hydrophobic solvent, was cyclisation, which occurred slowly due to mild acidity. Both the initial reaction rate and initial TOF decreased by increasing the Ni loading and increasing Ni-NiO particle size (Table 5, Fig. 7a).

Citronellal transformation proceeded in the fastest way over 5Ni/H-Beta-25 with almost 80% of citronellal conversion achieved after 15 min of the reaction. Moreover, this catalyst gave also the highest final conversion of 98% after 3 h, most probably due to the presence of small Ni-NiO particles and high acidity. The conversion increased by decreasing the metal loading (Table 5) and enhancing Lewis acid site concentration (LAS) (Fig. 8a). The same trend was observed in the work of Vajglova et al. by increasing total acid site concentration [1]. All catalysts gave complete mass balance closure of the liquid phase.

Fig. 8
figure 8

Menthol synthesis from citronellal a citronellal conversion as a function of Lewis acid sites concentration (CLAS); at 85% conversion, b yields of menthols (YMEs) as a function of Lewis acid sites concentration (CLAS); c cyclization products yield (YCP) as a function Brønsted acid sites concentration (CBAS); d yield of isopulegols (YIPs) as a function of Brønsted to Lewis acid sites ratio (B/L) e yield of menthols and f acyclic products yield as a function of mole of surface metal-to-mole of acid sites ratio. Legend: 15Ni/H-Beta-300 (red ■), 15Ni-H-Beta-25 (blue ●), 10Ni/H-Beta-25 (pink ▲), 5Ni/H-Beta-25 (green▼)

Full size image

The products formation over different catalysts is shown in Table 5 and Fig. 9a,c,e,g. The highest concentration of ( ±)-isopulegol was already observed after 1 min of reaction, because it was formed during heating in the absence of any stirring. Thereafter, pulegols were reacting further to menthols and menthanes (Fig. 9b, d, f, h). The highest concentration of ( ±)-isopulegol was obtained after 1 min with 5Ni/H-Beta-25 followed by 10Ni/H-Beta-25, 15Ni/H-Beta-25 and 15Ni/H-Beta-300 which corresponds to the order of decreasing acidity and increasing B/L ratio of H-Beta (Fig. 8d) and increasing Ni loading. This is expected, because cyclisation of citronellal is an acid catalyzed reaction [59]. Indeed, YIPs increased with increasing strong Brønsted acid sites concentration in other studies [1, 62].

Fig. 9
figure 9

Menthol synthesis from citronellal for a, b 15Ni/H-Beta-300, c, d 15Ni/H-Beta-25, e, f 10Ni/H-Beta-25 and g, h 5Ni/H-Beta-25: Concentration of citronellal (●), ( ±)-isopulegol (■), (± )-neo-isopulegol (▼), ( ±)-iso-isopulegol (Δ), (±)-menthol (▲), (±)-neomenthol (∇), (±)-neoisomenthol (□), citronellol (), menthanes (), menthenes (✩), 3,7-dimethyloctanol (), 3,7-dimethyloctane (), dimeric ethers (▸). Citronellal concentration presented in all graphs corresponds to right hand side y-axes and color coded: a, b red; c, d blue; e, f pink; g, h) green. Concentration of all other compounds correspond to the left hand side y-axes

Full size image

In the following step hydrogenation of ( ±)-isopulegol to ( ±)-menthol proceeded at a rather low rate over the most acidic 5Ni/H-Beta-25 (Fig. 10) despite of its small Ni-NiO particles. The yield of menthols at 85% conversion level of citronellal decreased with decreasing Ni-NiO particle size and increasing especially Lewis acidity.

Fig. 10
figure 10

Menthol synthesis from citronellal: a Concentration of menthols as a function of time, b stereoselectivity of menthol as a function of conversion. Legend: 15Ni/H-Beta-300 (red ■), 15Ni-H-Beta-25 (blue ●), 10Ni/H-Beta-25 (pink ▲), 5Ni/H-Beta-25 (green▼)

Full size image

This result shows that even rather large Ni-NiO particles of ca. 15 nm facilitated efficient hydrogenation of pulegols to menthols (Fig. 10) and the menthol yield at 85% conversion even increased with increasing Ni-NiO particle size (Fig. 11a).

Fig. 11
figure 11

Menthol synthesis from citronellal, yields of a menthols (YMEs) and b ACP (YACP) at 85% conversion as a function of Ni@NiO particle size. 15Ni-H-Beta-300 (red ■), 15Ni-H-Beta-25 (blue ●), 10Ni/H-Beta-25 (pink ▲), 5Ni/H-Beta-25 (green▼)

Full size image

The maximum yield of menthols, 36% was obtained for 15Ni/H-Beta-25 and 10Ni/H-Beta-25 at 3 h reaction (Table 5). However, neomenthol and neoisomenthol concentrations increased in the first hour of the reaction and reached almost a plateau for all catalysts (Fig. 9). From the concentration profiles for citronellal transformations (Fig. 9), the consecutive nature of the reaction network can be clearly seen and already after 15 min the menthol formation rate increased. Among 5Ni/H-Beta-25, 10Ni/H-Beta-25 and 15Ni/H-Beta-25, the latter exhibited the lowest amount of Brønsted acid sites (Table 4) being superior for menthol formation in comparison to two other catalysts. In the work reported by Murzin and co-workers, YMEs decreased with increasing Brønsted acid sites concentration [9]. The Brønsted-to-Lewis acid sites ratio (B/L) of this catalyst was 0.6 (Table 4) which is in line with the results for Ni/ZrS (nickel sulfated zirconia) catalyst displaying a higher acid cites concentration of 95 μm/g and a B/L ratio of 0.6 [14]. That catalyst was more active in citronellal transformation to menthols compared to nickel on sulfated zirconia possessing the acid sites concentration of 30 μm/g and a higher B/L ratio of 0.9. The highest menthol yield in [63] was 98% at full citronellal conversion over 8 wt% Ni-montmorillonite‑heteropolyacid catalyst at 80 °C, under 50 bar hydrogen and after a longer reaction time of 12 h. This catalyst exhibited the B/L ratio of 0.3, however, the Ni particle size was not reported. It was suggested in [63] that the catalyst with strong Lewis and medium Brønsted acid sites is suitable for the reaction which was also observed in the current study.

Stereoselectivity to ( ±)-menthol was initially the highest for 5Ni/H-Beta-25, decreasing, however, with increasing conversion to the same level, ca. 78%, as obtained for other catalysts (Fig. 10b). When comparing the stereoselectivity to ( ±)-menthol for nickel catalysts reported in the literature [14], it should be mentioned that nickel on sulfated zirconia gave 70% of stereoselectivity to ( ±)-menthol at a full citronellal conversion, at 100 ℃ and under 14 bar H2 after 30 min with the citronellal to catalyst mass ratio of 0.6. Note that the reaction in the current work was carried out under conditions different from [14], namely 80 °C, total pressure of 20 bar, reaction time 3 h and the citronellal to catalyst mass ratio of 1.3.

Formation of the undesired acyclic hydrogenation products (ACP), among which the most prominent product was citronellol, increased by increasing the Ni loading and particle size (Fig. 8, Table 3). In addition to citronellol, also small amounts of 3,7-dimethyloctanol were formed. In particular, by increasing the molar ratio of the exposed metal to the acid sites, the yield of ACP decreased for Ni/H-Beta-25 catalysts (Fig. 8f) similar with the result of ref [10], while 15Ni/H-Beta-300 with a mild acidity (Table 4) produced large amounts of both citronellol and 3,7-dimethyloctanol, and no 3,7-dimethyloctanol was formed over 5Ni/H-Beta-25.

Dimer formation was prominent over all studied catalysts (Table 5, Figs. 9). 5Ni/H-Beta-25 as the most acidic catalyst generated the highest concentration of dimeric products (Fig. 9h). Dimer formation was as expected also independent on the Ni-NiO average particle size. As a comparison, a relatively low menthol yield being less than 10% and high byproducts yield of more than 40% were observed at 24–100% citronellal conversion over natural zeolite supported Ni and Ni/ZSM-5 catalysts, respectively [8]. The reaction conditions were very different from the current work with the citronellal to catalyst mass ratio of 7, 70 °C, 20 bar of H2 and 7 h reaction time [16]. On the other hand, when the reaction was carried out at a higher temperature of 200 °C and 20 bar H2, 50% yield of menthol, 10% yield of isopulegol and 35% yield of byproducts were obtained over Ni/NZB (acidified natural zeolite) catalyst [64].

From the viewpoint of the reaction mechanism, it was interesting to relate formation of acyclic hydrogenation products vs cyclic counterparts. For that purpose, the sum of acyclic hydrogenation products (ACP) versus the sum of the desired cyclisation products (CP = IPs + MEs) was plotted (Fig. 12).

Fig. 12
figure 12

Sum of hydrogenation acyclic products concentrations (ACP) vs the sum of desired cyclization products concentrations (CP = IPs + MEs) at 80 °C, 20 bar total pressure in the presence of hydrogen for 15Ni/H-Beta-300 (red ■), 15Ni-H-Beta-25 (blue ●), 10Ni/H-Beta-25 (pink ▲), 5Ni/H-Beta-25 (green ▼), first point (*, denoted reaction time 15 min)

Full size image

For 15Ni/H-Beta-300 with the highest Ni loading and the mild acidity, both the concentration of cyclic and acyclic products increased after 15 min of the reaction time and more acyclic hydrogenation products were formed with increasing reaction time. Due to a mild acidity, cyclisation was not as efficient as with the other catalysts. On the other hand, for 15Ni/H-Beta-25, only cyclic products were produced in the first 15 min of the reaction, thereafter formation of acyclic hydrogenation products started due to suppression of cyclization. 10Ni/H-Beta-25 and 5Ni/H-Beta-25 generated the maximum concentrations of cyclic products after 120 min and 60 min of the reaction, respectively. The highest amounts of acyclic hydrogenation products were formed over 15Ni/H-Beta-300, which exhibited the highest peak area in H2-TPR (Table 2) and mild acidity (Table 4). On the other hand, the second highest concentration of ACP was obtained over the catalyst with medium acidity and 15 nm Ni-NiO particles (Table 3), while the smallest Ni-NiO particles, 2 nm, present in 5Ni/H-Beta-25 with the highest acidity among the studied Ni/H-Beta catalysts (Table 4) promoted formation of cyclisation products. These results differ slightly from the ones published by Azkaar et al. [9], who reported that highly acidic Ru catalyst promoted ACP formation.

3.3 Regeneration and Reuse of 15 Wt% Ni/H-Beta-25

The characterization results demonstrated stability of the textural and morphological properties of 15Ni-H-Beta-25 catalyst after the reaction. Nevertheless, because minor leaching of Ni could be concluded from ICP-OES measurements, the spent catalyst was regenerated by washing with the reaction solvent (cyclohexane) followed by drying and reduction steps. Thereafter, the regenerated catalyst was tested in the reaction with the results reported in Fig. 13.

Fig. 13
figure 13

a Conversion of citronellal, X (●, ○), b yield of pulegols, YPs (■, □) and yield of menthols, YMEs (▲, Δ), c stereoselectivity to ( ±)-isopulegol (◆, ◊) and ( ±)-menthol (★, ✱) as a function of time in one-pot synthesis of menthol starting from citronellal over 15Ni/H-Beta-25. Symbols: fresh catalyst (filled symbol), regenerated catalyst (open symbol)

Full size image

It can be concluded that very similar conversion, yields of pulegols and menthols, along with stereoselectivity to ( ±)-isopulegol and ( ±)-menthol were obtained in the second run, which is a promising result in terms of catalyst recyclability.

4 Conclusion

Ni/H-Beta catalysts with different silica to alumina ratios and thus acidities (H-Beta-25 and H-Beta-300) and different Ni loadings of 5, 10 and 15 wt.% were prepared and tested in the one-pot transformations of citronellal to menthol in a batch reactor. Different Ni loadings affected the Ni-NiO particle size and dispersion and also accessibility of the H-Beta-25 acid sites. An optimum synergy between H-Beta acidity and Ni loading and dispersion was required to drive the cascade reaction toward the main cyclic product namely ( ±)-menthol isomer limiting undesired side reactions, such as acyclic hydrogenation and dimerization of citronellal and pulegols. Among the tested materials, 15 wt.% Ni on H-Beta-25 exhibited the best synergy of the total acid sites concentration, Lewis to Brønsted acid sites ratio, and the metal loading and dispersion. The reaction over 15 wt% Ni/H-Beta-25 led to 93% citronellal conversion with the highest menthols yield of 36% (77% stereoselectivity to ( ±)-menthol isomer). Regeneration and reusability of the best catalyst was promising as the reused catalyst after simple washing with the solvent and reduction afforded similar activity and selectivity as the fresh catalyst. While the amount of acyclic products was rather low, dimeric ethers of cyclic nature were formed on acid sites with appreciable selectivity, thus further efforts in modulation of acid/metal bifunctionality should be focused on further improving the yields of the desired menthols in the one-pot conversion of citronellal.