Reaction of Glycerol with Trimethyl Orthoformate: Towards the Synthesis of New Glycerol Derivatives

The reactivity of glycerol with trimethyl orthoformate is here described with an emphasis on developing a reliable synthetic approach for glycerol valorization. The glycerol based orthoester 4-(dimethoxymethoxy)methyl)-2-methoxy-1,3-dioxolane (3) was synthesized, under catalytic as well as catalyst-free conditions, by taking advantage of the thermodynamically controlled equilibrium between intermediates. Both Brønsted and Lewis acid catalysts accelerated the attainment of such an equilibrium, particularly Brønsted acidic ionic liquids BSMImHSO4 and BSMImBr were the most effective compounds for this reaction. The kinetic profiles allowed the proposal of a mechanism that accounts for the selectivity of the reaction.


Introduction
Glycerol (Glyc) is one of the most important biomass-derived platform chemicals [1]. Over the last few decades, the overabundance of this bio-diesel by-product stimulated academic and industrial interests towards its conversion into renewable high value-added chemicals [2][3][4][5][6][7]. Some of the most promising processes are summarized in Figure 1.
In this field, our research group has explored different thermal and catalytic synthetic approaches by reacting Glyc with formaldehyde, acetone, dimethyl carbonate, and enol esters to yield acetals, ketals, and other sustainable derivatives [8][9][10][11][12][13]. In a search to further expand this chemistry and to identify new bio-building blocks, we considered orthoesters (OEs) as another family of electrophilic partners for the upgrading of Glyc.
The impressive reactivity of OEs-up to 20,000 times higher than of the above cited ketones and aldehydes [14]-has stimulated significant studies for their applications as protecting groups for carboxylic acids and esters [15] (owing to their stability towards strong nucleophiles and bases), as dehydrating agents [16,17], and as copolymers for the synthesis of polyorthoesters [18,19]. More recently, a fascinating use of OEs in the synthesis of dynamic cryptands has been proposed, employed for the controlled release of metal ion guests [20,21]. Some pioneering research dating back to the '60s and '80s highlighted the potential of the reaction of Glyc with OEs. In 1964, Crank and Eastwood were the first to investigate the reaction between triols and triethylorthoformate to obtain bicyclic orthoesters (BOEs) [22], demonstrating the formation of several BOEs from 1,2,4-1,2,5-and 1,3,5-triols. By using Glyc as reagent, they observed the formation of a mixture of the pentacyclic cis-/trans-2-ethoxy-4-hydroxymethyl-1,3-dioxolane (1) and its hexacyclic isomer cis-/trans-2-ethoxy-5-hydroxy-1,3-dioxane (4) (Figure 2a) in a 9:1 ratio with 67% of yield, which could thermally decomposed to ethanol, CO2 and allyl alcohol [23]. The bicyclic compound 2,6,7-trioxabyciclo [2.2.1] heptane (2) was not observed. A few years ago, Thshibalonza and Monbaliu revisited the reaction by focusing on the formic acid-catalyzed synthesis of allyl alcohol under continuous-flow conditions (Figure 2b) [24].
In this work, the thermal and catalytic reaction between Gly and HC(OMe) 3 is studied with the aim to develop new reliable synthetic protocols for glycerol exploitation. The selective formation (up to 95% of yield) of a new glycerol-based 5-membered ring di-orthoester (the cis-/trans-4-(dimethoxymethoxy)methyl)-2-methoxy-1,3-dioxolane; 3, Figure 2d) was accomplished by using HC(OMe) 3 as model OE via a dynamic and thermodynamically-controlled synthetic strategy. A variety of homogeneous and heterogeneous acids (PPTS, DBUHBr, sulfuric acid, Amberlyst-15, amberlyst-36, FeCl 3 , and AlF 3 ) and ionic liquids (BSMImHSO 4, BSMImBr) were used as Brønsted/Lewis-acidic catalysts. In this context, the ionic liquids (ILs) were tested as they are known to give better yields, turnover numbers, and frequencies, and improved catalysts recyclability [31]. The effect of basic catalysts (such as K 2 CO 3 and trioctylmethylphosphonium methylcarbonate: [P 1888 ]CH 3 OCO 2 − ) was also investigated for comparison. In this work, the thermal and catalytic reaction between Gly and HC(OMe)3 is studied with the aim to develop new reliable synthetic protocols for glycerol exploitation. The selective formation (up to 95% of yield) of a new glycerol-based 5-membered ring di-orthoester (the cis-/trans-4-(dimethoxymethoxy)methyl)-2-methoxy-1,3-dioxolane; 3, Figure 2d) was accomplished by using HC(OMe)3 as model OE via a dynamic and thermodynamically-controlled synthetic strategy. A variety of homogeneous and heterogeneous acids (PPTS, DBUHBr, sulfuric acid, Amberlyst-15, amberlyst-36, FeCl3, and AlF3) and ionic liquids (BSMImHSO4, BSMImBr) were used as Brønsted/Lewis-acidic catalysts. In this context, the ionic liquids (ILs) were tested as they are known to give better yields, turnover numbers, and frequencies, and improved catalysts recyclability [31]. The effect of basic catalysts (such as K2CO3 and trioctylmethylphosphonium methylcarbonate: [P1888]CH3OCO2 − ) was also investigated for comparison.

Results and Discussion
The reactivity between Glyc and HC(OMe)3 was first tested in the absence of any catalyst, in analogy to the reaction conditions previously reported in literature [22,25,26]. The reaction was carried out in the absence of added solvents at 90 °C with a reagent molar ratio Q = 1 (Q = HC(OMe)3:Glyc) for 1 h. The products were isolated and fully characterized by GC-MS, 1 H, 13 C, and 2D-NMR analyses (see Material and Methods or Supplementary Materials for further details) revealing the formation of the five-membered ring OE 1, the bicyclic compound 2,6,7-trioxabyciclo [2.2.1] heptane 2, and the di-orthoester 3 (1 and 3 as mixtures of diastereomers) in 82/11/7 selectivity respectively ( Figure 3). Figure 4 shows the gaschromatogram of the crude reaction mixture.

Results and Discussion
The reactivity between Glyc and HC(OMe) 3 was first tested in the absence of any catalyst, in analogy to the reaction conditions previously reported in literature [22,25,26]. The reaction was carried out in the absence of added solvents at 90 • C with a reagent molar ratio Q = 1 (Q = HC(OMe) 3 :Glyc) for 1 h. The products were isolated and fully characterized by GC-MS, 1 H, 13 C, and 2D-NMR analyses (see Material and Methods or Supplementary Materials for further details) revealing the formation of the five-membered ring OE 1, the bicyclic compound 2,6,7-trioxabyciclo [2.2.1] heptane 2, and the di-orthoester 3 (1 and 3 as mixtures of diastereomers) in 82/11/7 selectivity respectively ( Figure 3). Figure 4 shows the gaschromatogram of the crude reaction mixture.   The six-membered ring isomers 4 and the correspondent 3 derivative were not observed. This was unexpected, since previous results for the reaction between Glyc and triethyl orthoformate reported a five-and six-membered ring isomeric ratio of 3:1 [26] or 9:1 [22] for compounds 1:4. A reasonable explanation is that, under our conditions, the reaction between glycerol and HC(OMe) 3 is kinetically rather than thermodynamically controlled. Indeed, the ring strain energy of five-membered ring 1,3-dioxolane (7.3 kcal/mol) is higher than the six-membered ring 1,3-dioxane analogs (2.9 kcal/mol) [32], while the stabilization energy due to their formation was 4.4 and 13.2 respectively [33]. These trends of energies show that 1,3-dioxane are more stable than 1,3-dioxolane. This explanation is also in   The six-membered ring isomers 4 and the correspondent 3 derivative were not observed. This was unexpected, since previous results for the reaction between Glyc and triethyl orthoformate reported a five-and six-membered ring isomeric ratio of 3:1 [26Error! Bookmark not defined.] or 9:1 [22] for compounds 1:4. A reasonable explanation is that, under our conditions, the reaction between glycerol and HC(OMe)3 is kinetically rather than thermodynamically controlled. Indeed, the ring strain energy of five-membered ring 1,3-dioxolane (7.3 kcal/mol) is higher than the six-membered ring 1,3-dioxane analogs (2.9 kcal/mol) [32], while the stabilization energy due to their formation was 4.4 and 13.2 respectively [33]. These trends of energies show that 1,3-dioxane are more stable than 1,3-dioxolane. This explanation is also in agreement with the results reported by Yokohama et al. for the ring-opening polymerization of 2 that under kinetical control, leads to the almost exclusive formation of the five-membered ring product [26].
Further experiments were then carried out at different temperatures and Glyc:HC(OMe)3 molar ratios by following the product distribution over the time. At 90 °C, under reactive distillation conditions to remove methanol continuously during the reaction, the kinetic profile of the reaction was studied for 24 h at two different molar ratios Q = 1 and Q = 10 ( Figure 5a,b). In both cases, within the first minutes of reaction the initial biphasic mixture turned homogeneous and vigorous distillation of methanol was observed. For both ratios Q = 1 and Q = 10, the Glyc conversion was almost quantitative after 0.5 h (based on GC analysis) towards formation of 1 and 3 respectively, thus proving that the formation of the glycerol-OEs is basically instantaneous, yielding a statistical  Further experiments were then carried out at different temperatures and Glyc:HC(OMe) 3 molar ratios by following the product distribution over the time. At 90 • C, under reactive distillation conditions to remove methanol continuously during the reaction, the kinetic profile of the reaction was studied for 24 h at two different molar ratios Q = 1 and Q = 10 ( Figure 5a,b). In both cases, within the first minutes of reaction the initial biphasic mixture turned homogeneous and vigorous distillation of methanol was observed. For both ratios Q = 1 and Q = 10, the Glyc conversion was almost quantitative after 0.5 h (based on GC analysis) towards formation of 1 and 3 respectively, thus proving that the formation of the glycerol-OEs is basically instantaneous, yielding a statistical product distribution. After 24 h, a relatively slow dynamic covalent exchange reaction yielded a thermodynamically controlled product distribution which is influenced by the reaction conditions. For example, with Q = 1, after 1 h the selectivity was shifted towards 1, but prolonging the reaction time, 1 and 3 slowly decreased in favor of the formation of 2 (39% after 24 h) via an intramolecular reaction favored by the continuous removal of methanol. When using an excess of HC(OMe) 3 (Q = 10) the initial selectivity was shifted towards the formation of 3 (80%, 1 h), as expected; after 24 h the reaction slowly proceeded further favoring conversion of 1 into 3 (97%, 24 h). It is worth noting that, under these conditions, 3 could be isolated by simply evaporating the excess of HC(OMe) 3 without any further purification step (95% isolated yield with 98% purity by GC).
For example, with Q = 1, after 1 h the selectivity was shifted towards 1, but prolonging the reaction time, 1 and 3 slowly decreased in favor of the formation of 2 (39% after 24 h) via an intramolecular reaction favored by the continuous removal of methanol. When using an excess of HC(OMe)3 (Q = 10) the initial selectivity was shifted towards the formation of 3 (80%, 1 h), as expected; after 24 h the reaction slowly proceeded further favoring conversion of 1 into 3 (97%, 24 h). It is worth noting that, under these conditions, 3 could be isolated by simply evaporating the excess of HC(OMe)3 without any further purification step (95% isolated yield with 98% purity by GC). The experimental results indicated that-likewise to the known glycerol-orthoformate reactivity [22][23][24]26]   The experimental results indicated that-likewise to the known glycerol-orthoformate reactivity [22][23][24]26] and the alkoxy-exchange reactivity of OEs [34]-the initial fast reaction of HC(OMe) 3 with Glyc gives 1 which is in equilibrium with 2 and 3, the latter formed by reaction with another molecule of orthoformate. Interconversion of 2 and 3 goes through a dynamic covalent reaction involving 1 as the intermediate. Scheme 1 shows the hypothesized reaction pathways. reaction favored by the continuous removal of methanol. When using an excess of HC(OMe)3 (Q = 10) the initial selectivity was shifted towards the formation of 3 (80%, 1 h), as expected; after 24 h the reaction slowly proceeded further favoring conversion of 1 into 3 (97%, 24 h). It is worth noting that, under these conditions, 3 could be isolated by simply evaporating the excess of HC(OMe)3 without any further purification step (95% isolated yield with 98% purity by GC). The experimental results indicated that-likewise to the known glycerol-orthoformate reactivity [22][23][24]26] and the alkoxy-exchange reactivity of OEs [34]-the initial fast reaction of HC(OMe)3 with Glyc gives 1 which is in equilibrium with 2 and 3, the latter formed by reaction with another molecule of orthoformate. Interconversion of 2 and 3 goes through a dynamic covalent reaction involving 1 as the intermediate. Scheme  The reaction was further investigated by varying Q between 1 and 10, at T = 90 • C and 25 • C, for 24 h. As shown in Table 1, 3 was formed selectively at 90 • C only with a large excess of HC(OMe) 3 (Q = 10) (entry 5, Table 1). At lower Q values instead (entries 1-4, Table 1), the selectivity-determining dynamic equilibrium resulted in a statistical products distribution. Table 1. Q and T effect on products distribution for the catalyst-free reaction of Gly with HC(OMe). Q = 10, the reaction rate increased reaching almost quantitative conversion, but the product distribution was not shifted towards the formation of 3, as was the case for the reaction conducted at 90 °C. Instead, a high (88%) selectivity towards 1 was observed after 1 h, while after 24 h the dynamic equilibrium promoted a 74/5/21 selectivity towards 1/2/3 respectively (entry 7 and Figure S33 for the reaction profile). Overall, the catalyst-free reaction between Gly and HC(OMe)3 lead to the selective synthesis of the thermodynamically-favored product 3-a new orthoester derivative of glycerol-after 24 h and in the presence of an excess of OE.
With the aim to accelerate attainment of the dynamic equilibrium among 1, 2, and 3, and to direct selectivity, different acid catalysts were then tested at Q = 10 and T = 90 °C. The results are listed in Table 2. In the presence of PPTS (10%w/wgly), known to catalyze the reaction of diglycerol with triethyl orthoformate [30], the reaction reached quantitative conversion and 97% selectivity towards 3 in 1 h (entry 3 and 4; Table 2). No further products were observed even after 24 h, indicating that 3 is stable under the reaction conditions ( Figure S34 for the reaction profile). As shown in Table 2, all the other acid catalysts favored the formation of 3 already after 1 h. Only for diazabicycloundecenium bromide (DBUHBr, entry 5-6, Table 2) a slower dynamic equilibrium-in line with the uncatalyzed reaction-was observed (75% and 93% selectivity after 1 and 24 h, respectively) likely due to the lower acidity [35]. Brønsted and Lewis acidic catalysts such as H2SO4, amberlyst-15, amberlyst-36, iron chloride and aluminum fluoride promoted moderate-to-good selectivity (84, 79, 87, 91, 87%, respectively; entries 7, 9, 11, 13, 15, Table 2) towards 3 after 1 h along with small amount of ill-defined products (e.g., aldehydes, ethers, or oligomers). The latter were probably formed by catalytic decomposition of 3 as already described in the literature for OEs [26,36,37]. The reaction became evident after 24 h. In particular, the use of sulfuric acid caused formation of a large quantity (≈41%) of four major undefined products (see Figures S35-S39  A less pronounced exchange equilibrium between the products was observed for the reactions conducted at 25 • C respect to the ones conducted at 90 • C. In fact, at 25 • C and Q = 1 the reaction was very slow and after 24 h, conversion reached only 63% (90% conversion after 48 h), with 79% selectivity towards formation of 1 (entry 6, Table 1 and Figure S32 for the reaction profile). At higher Q = 10, the reaction rate increased reaching almost quantitative conversion, but the product distribution was not shifted towards the formation of 3, as was the case for the reaction conducted at 90 • C. Instead, a high (88%) selectivity towards 1 was observed after 1 h, while after 24 h the dynamic equilibrium promoted a 74/5/21 selectivity towards 1/2/3 respectively (entry 7 and Figure S33 for the reaction profile).
Overall, the catalyst-free reaction between Gly and HC(OMe) 3 lead to the selective synthesis of the thermodynamically-favored product 3-a new orthoester derivative of glycerol-after 24 h and in the presence of an excess of OE.
With the aim to accelerate attainment of the dynamic equilibrium among 1, 2, and 3, and to direct selectivity, different acid catalysts were then tested at Q = 10 and T = 90 • C. The results are listed in Table 2. In the presence of PPTS (10%w/w gly ), known to catalyze the reaction of diglycerol with triethyl orthoformate [30], the reaction reached quantitative conversion and 97% selectivity towards 3 in 1 h (entries 3 and 4; Table 2). No further products were observed even after 24 h, indicating that 3 is stable under the reaction conditions ( Figure S34 for the reaction profile). As shown in Table 2, all the other acid catalysts favored the formation of 3 already after 1 h. Only for diazabicyclo-undecenium bromide (DBUHBr, entries 5 and 6, Table 2) a slower dynamic equilibrium-in line with the uncatalyzed reaction-was observed (75% and 93% selectivity after 1 and 24 h, respectively) likely due to the lower acidity [35]. Brønsted and Lewis acidic catalysts such as H 2 SO 4 , amberlyst-15, amberlyst-36, iron chloride and aluminum fluoride promoted moderate-to-good selectivity (84%, 79%, 87%, 91%, 87%, respectively; entries 7, 9, 11, 13, 15, Table 2) towards 3 after 1 h along with small amount of ill-defined products (e.g., aldehydes, ethers, or oligomers). The latter were probably formed by catalytic decomposition of 3 as already described in the literature for OEs [26,36,37]. The reaction became evident after 24 h. In particular, the use of sulfuric acid caused formation of a large quantity (≈41%) of four major undefined products (see Figures S35-S39 for the gas-chromatography trace and MS spectra) after 24 h at 90 • C. Instead, when using the ionic liquids butylsulfonylmethylimidazolium bromide (BSMImBr) and hydrogensulfate (BSMImHSO 4 ) as catalyst (entries 17-20, Table 2) the fast reaction rates were always accompanied by the quantitative and selective formation of 3 (98%) and by the absence of side reactions (99% also after 24 h). The same acidic ionic liquids allowed to perform the reaction at 25 • C, still with appreciable selectivity towards 3 (81-87%; entries 21-22; Table 2). Finally, no  Table 2). H 0 was chosen as it is one of the most effective ways to express the Brønsted acidity in an organic medium [39,40]. It is readily apparent that stronger acidity (H 0 < 0, entries 7-12) prompts faster reactions but also higher formation of side products. On the other hand, when using weaker acids (H 0 = 1-3, entries 17-22) the reactions are more selective towards the formation of the desired product 3. Table 2. Catalyzed reaction between HC(OMe) 3

and Gly.
Catalysts 2019, 9, x FOR PEER REVIEW 7 of 13 MS spectra) after 24 h at 90 °C. Instead, when using the ionic liquids butylsulfonylmethylimidazolium bromide (BSMImBr) and hydrogensulfate (BSMImHSO4) as catalyst (entries 17-20, Table 2) the fast reaction rates were always accompanied by the quantitative and selective formation of 3 (98%) and by the absence of side reactions (99% also after 24 h). The same acidic ionic liquids allowed to perform the reaction at 25 °C, still with appreciable selectivity towards 3 (81-87%; entries 21-22; Table 2). Finally, no conversion of Glyc was observed when using basic catalysts, such as K2CO3 or [P1888]CH3OCO2 − [38], owing to the stability of OEs towards bases.
To rationalize the catalytic behavior, the Hammett acidity functions (H0) for the different catalysts were compared (see Table 2). H0 was chosen as it is one of the most effective ways to express the Brønsted acidity in an organic medium [39,40]. It is readily apparent that stronger acidity (H0 < 0, entry 7-12) prompts faster reactions but also higher formation of side products. On the other hand, when using weaker acids (H0 = 1-3, entries 17-22) the reactions are more selective towards the formation of the desired product 3.

General Information
All the chemicals were purchased from Aldrich and used as received. Trioctylmethylphosphonium methylcarbonate ([P 1888 ]CH 3 OCO 2 − ) was synthesized according to a literature procedure [38].

Synthesis/Isolation of Reaction Products
3.2.1. Synthesis of (2-methoxy-1,3-dioxolan-4-yl) Methanol (1) Glycerol (5.12 g, 55.6 mmol) was added in a two-neck round bottom flask equipped with a dripping funnel and a heating condenser held at 70 • C. The flask was heated at 90 • C and HC(OMe) 3 (6.1 mL, 55.6 mmol) was added dropwise under stirring and kept for 1 h. In Figure 5, the gaschromatographic trace of the reaction crude is reported, highlighting the presence of diastereomers. The reaction mixture was then distilled over potassium carbonate three times to isolate one pure distereoisomer of 1 in 10% yield, hereafter identified as 1a (the peaks with retention time 15.0 min in Figure 4). The characterization of the other diastereoisomer 1b (retention time 15.8 min) is inferred from the spectra of the mixture. 1  Glycerol (0.526 g, 5.71 mmol) and HC(OMe) 3 (6.3 mL, 57.2 mmol) were added in a round-bottomed flask and the mixture was stirred at 90 • C. During the reaction the methanol was distilled out the mixture by heating the condenser at 70 • C. After 24 h the mixture was cooled down and then concentrated under vacuum (60 • C, 35 mbar). 3 was obtained a mixture of two diastereomers in 95% yield without any further purification (purity 98% by GC). The diastereoisomeric ratio of 3 was 1:1 according to GC and 1 H NMR analyses ( Figure S16). Any attempts to separate the two isomers (hereafter called 3a or 3b) failed. 1

Synthesis of Pyridinium Para-Toluensulfonate (PPTS)
The synthesis was carried out following a literature procedure [46]. 1  1-methylimidazole and 1,4-butane-sultone were charged into a 100 mol round-bottom flask in an equimolar amount (0.1 mol) and stirred at 80 • C for 16 h. The white solid zwitterionic product was washed with ethyl ether (10 × 5 mL) and dried in vacuum (yield = 90%). Then a stoichiometric amount of hydrobromic acid (40% v/v H 2 O) was added and the mixture stirred at 60 • C for 10 h to yield to the ionic liquid BSMImBr. The ionic liquid (IL) was washed again with dichloromethane and ether to remove non-ionic residues and dried under vacuum. The IL was obtained with a 90% yield and used without any further purification. 1

Catalyst-Free Conditions
Glycerol (0.5 g, 5.43 mmol) and cyclopentyl methyl ether (100 µL) as an internal standard were added in a two-neck round-bottomed flask equipped with a dripping funnel and a condenser. The mixture was stirred at the selected temperature (T = 25 or 90 • C) and HC(OMe) 3 was added dropwise by varying the molar ratio Q from 1 to 10. The reaction was allowed to proceed for 24 h under stirring and samples were collected and analyzed by GC at regular intervals. For reactions run at 90 • C, the condenser was heated at 70 • C in order to continuously distill methanol.

Catalyzed Reaction
The above reported procedure was followed with the exception that the selected catalyst (10% w/w Gly ) was added with glycerol and cyclopentyl methyl ether, before the addition of HC(OMe) 3 .

Catalytic Procedure for the Synthesis of 3
Glycerol (0.5 g, 5.43 mmol), HC(OMe) 3 (5.9 mL, 54.3 mmol) and a catalytic amount of the selected catalyst (BSMImHSO 4 , BSMImBr, 3% mol/mol gly ) were added in a round-bottomed flask and the mixture was stirred at 90 • C. During the reaction the methanol was distilled out the mixture by heating the condenser at 70 • C. After 1h, the mixture was cooled down and the solution was passed through a short plug of alumina. The filtered solution was concentrated under vacuum (60 • C, 35 mbar). 3 was obtained as a mixture of two diastereomers in 96% yield without any further purification (purity 98% by GC).

Conclusions
The thermal and catalytic reactivity of glycerol with trimethyl orthoformate was fully explored here for the first time.
Under catalyst-free conditions, with an equimolar mixture of Gly and HC(OMe) 3 (Q = 1), the reaction leads to the unselective formation of a dynamic equilibrium mixture of products 1, 2, and 3 (Scheme 1 and Table 1). The products were isolated and fully characterized by GC-MS, 1 H, 13 C, and 2D-NMR spectroscopy revealing the formation exclusively of the five-membered ring isomers of 1 and 3, obtained as mixtures of two diastereoisomers. The equilibrium of this reaction is remarkably shifted towards the formation of 3 as evinced for longer reaction times, higher temperature and/or using a large excess of orthoformate (Q = 10). 3 could thus be obtained selectively in high yields at 90 • C with 10 equivalents of trimethyl orthoformate after 24 h.
Both the reaction time and temperature can be reduced by using Brønsted and Lewis-acid catalysts. The strongest acidic catalysts (H 2 SO 4 , A-15, A-36, iron chloride, and aluminium fluoride) accelerate attainment of the dynamic equilibrium of the reaction to yield preferentially 3, but also promote ill-defined degradation products. Less acidic catalysts such as the ionic liquids BSMImHSO 4 and BSMImBr also accelerate attainment of the thermodynamically favored product 3 (98-99%) without however further degradations as indicated by the stability of 3 over time.
Overall, the present results shed light on the reaction pathway and on the equilibria that govern conversion of glycerol into its orthoesters 1 and 3 and it allows to selectively synthesize 3 by a proper choice of reaction conditions. The glycerol-based orthoesters represent new building blocks en route to the development of more complex glycerol-OEs architectures.

Conflicts of Interest:
The authors declare no conflict of interest.