Indium-Mediated Acyloxyallylation-Based Synthesis of Galacto-Configured Higher-Carbon Sugar Alcohols as Potential Phase Change Materials

Sugar alcohols fulfilling specific structural requirements are a substance class with great potential as organic phase change materials (PCMs). Within this work, we demonstrate the indium-mediated acyloxyallylation (IMA) as a useful strategy for the synthesis of higher-carbon sugar alcohols of the galacto-family featuring all hydroxyl groups in a 1,3-anti-relationship with three major synthetic achievements: first, the dihydroxylation of the IMA-derived allylic sugar derivates was systematically studied in terms of diastereoselectivity, revealing a high degree of substrate control toward anti-addition. Second, we demonstrated the use of a “double Mitsunobu” reaction, inverting the stereochemistry of terminal diols. Third, the IMA toolbox was expanded to accomplish the synthesis of derivatives with up to 10 carbon atoms from particularly unreactive aldoses. Thermal investigations of all synthesized sugar alcohols, including examples with exclusive 1,3-anti- and suboptimal 1,3-syn-relationships as well as even and odd numbers of carbon atoms, were performed. We observed clear trends in melting points and thermal storage densities and discovered limitations of organic substances in this class with melting points above 240 °C as PCMs in terms of thermal stability. With our study, we provide insights into the dependence of thermal properties on structural features, thus contributing to further understanding of organic PCMs for thermal energy storage applications.


■ INTRODUCTION
In nature, carbohydrates or sugars are abundant polyhydroxylated compounds that play an essential role in various biological processes.−4 Here, sugar alcohols have gained interest since they can act as sugar mimics and chiral synthons and have become especially popular in the design of drugs. 5−7 Later, sugar alcohols became interesting in materials science as well, e.g., in the development of functional materials. 8,9−16 These values are remarkable since the already commercially exploited paraffins typically reach storage densities of 150−250 J/g. 17 Additionally, sugar alcohols allow us to operate in a different temperature range due to their significantly higher melting points as compared to paraffins.Hence, efforts were undertaken to make sugar alcohols practically applicable to TES systems, addressing issues such as cycle stability during repeated melting and crystallization.We became interested in this compound class by a computational study by Inagaki and   Ishida, 18 where extraordinarily high thermal storage densities of up to 450−500 J/g have been predicted for higher-carbon sugar alcohols with specific structural features.These sugar alcohols were designed by the formal elongation of the natural D-mannitol following three structural criteria: a linear carbon backbone, an even number of carbon atoms, and 1,3-anticonfiguration of all hydroxyl groups (Figure 1).Such complex (to synthesize) structures will clearly not represent the next generation of energy storage materials.
However, we took on the challenge to synthesize highercarbon sugar alcohols following the stated criteria to, first, compare the calculated values to experimental ones and second, confirm the importance of the structural features for high thermal storage densities since most of the calculated values could not be supported by experimental data due to the elaborate synthesis 19−23 or inaccessibility of the compounds at that time.The addition of carbon atoms to the anomeric center of reducing sugars is often plagued by poor stereoselectivity, and often only little control is possible.Additionally, most transformations require protection strategies of the sugar species, making the transformations laborious. 24With the indium-mediated acyloxyallylation (IMA), a method was developed and investigated in recent years by us and others, 25−27 which allows the direct elongation of unprotected aldoses by three carbon atoms with good control of diastereoselectivity and therefore allows a more straightforward synthesis of higher-carbon species.In the IMA, a bromopropenyl ester (acetate, benzoate, or pivaloate) is added to the aldehyde functionality of an unprotected sugar.In this transformation, the formation of four different isomers is possible since two new stereocenters are formed.However, it was found that in the case of unprotected aldoses, the isomer with lyxo-configuration is formed with good selectivity (>60:40 dr lyxo:other isomers).Lyxo-configuration corresponds to a syn-orientation of the former α-hydroxyl group and the new hydroxyl group formed from the aldehyde, and an antiorientation of the two new stereocenters.The major factors that influence the success of this transformations are the solubility of the aldose, the reactivity or rather availability of the specific aldose as an aldehyde reflected by its open-chain content (OCC), and the lifetime of the indium organyl under the reaction conditions.Especially, the balance between the latter two was found to be crucial to obtain a clean transformation of the sugar to the enitol species.The IMA was shown to be a useful tool in the synthesis of the bacterial sugar L-glycero-D-manno-heptose, short LD-heptose, allowing for a short sequence at scale (Figure 2a). 26Furthermore, the IMA was applied by us for the synthesis of higher-carbon sugar alcohols of the manno-series that has been proposed and investigated in the mentioned computational study (Figure 2b). 28Investigations of the thermal properties of the synthesized sugar alcohols showed that the calculated thermal storage densities were in consistence with the experimentally observed ones and that sugar alcohols that fulfill the stated criteria indeed possess exceptionally high thermal storage densities, indicated as the latent heat of fusion (ΔH fus ), among thermally stable, organic materials.
Within the current work, our focus was on the utilization of the IMA for the synthesis of higher-carbon sugar alcohols of the galactitol family, the only second series of sugar alcohols that fulfills the three stated criteria in order to obtain high thermal storage densities, in theory.Sugar alcohols of this socalled galacto-series have not been investigated in regard to  The Journal of Organic Chemistry their thermal properties so far, neither computationally nor experimentally.Therefore, we developed a novel strategy toward this class of sugar alcohols with up to 10 carbon atoms (Figure 2c).In the galacto-series, all hydroxyl groups equally are in a 1,3-anti-relationship, and in contrast to the mannoseries, the two hydroxyl groups at the terminal stereocenters are in a syn-relationship.Noteworthily, the parent galactitol possesses an even higher thermal storage density than D- mannitol 10 (Figure 3a).With our research, we contribute to a better understanding of the relationship between structural features of organic molecules and their thermal properties and targeted examples matching all criteria as well as examples not matching all stated criteria (Figure 3b).For consistency, throughout the manuscript, we refer to the longer homologues of the parent galactitol-stereochemistry with the desired all 1,3anti-relative stereochemistry as, e.g., galacto-octitol, addressing the total chain length.Compounds differing from this stereochemical pattern on one of the additional stereocenters are referred to as, e.g., epi-galacto-octitol.

■ RESULTS AND DISCUSSION
Development of the (Retro)synthetic Strategy toward Galacto-Configured Higher Sugar Alcohols.In general, the retrosynthetic analysis of galacto-configured sugar alcohols with an even number of carbon atoms reveals two key steps in the forward synthesis: the selective dihydroxylation of the enitol species toward the sugar alcohol and the IMA of the three-carbon shorter aldose toward this enitol.These two steps can be repetitively performed starting from L-lyxose (8) to obtain the galacto-octitol 4 (C8) and galacto-decitol 6 (C10) (Figure 4); the latter could analogously be extended to the galacto-dodecitol SI-3 (C12), in theory.Initially, focusing on the galacto-octitol 4 (Figure 4a), the synthesis was developed starting from the pentose L-lyxose (8).The sugar alcohol galacto-octitol 4 should be obtained from octenitol 7 via dihydroxylation (DH) with selectivity for the desired configuration of the introduced secondary hydroxyl group.Octenitol 7 with the desired stereochemical composition (all 1,3-anti-relationship) is accessible via indium-mediated acyloxyallylation from the reducing sugar L-lyxose (8) with 3bromopropenyl acetate (9).This reaction is known in the  The Journal of Organic Chemistry literature 25,26 and has also been demonstrated on large scale, facilitated by its superior crystallinity allowing for purification via recrystallization with a high recovery of the targeted main diastereomer.At the dihydroxylation step, we aimed for high selectivity for the syn-2,3-diol to obtain the sugar alcohol with all hydroxyl groups in 1,3-anti-relationship directly from the enitol species 7. Considering classical DH conditions using an osmium-catalyst, we assumed that especially Sharpless asymmetric dihydroxylation (AD) allows us to tune the stereochemical outcome of the reaction.The same principle retrosynthetic steps can be repeated, dissecting the longerchain decitol 6 to the reported LD-heptose 11, which is the oxidation product of the central octenitol 7 (Figure 4b).
Synthesis of Octitols via IMA and Dihydroxylation.For the planned dihydroxylation reaction, protection of the reported octenitol 7 26 was necessary to achieve solubility of the substrate in an organic solvent.Three different protecting groups (PGs) were chosen to investigate their influence on the stereochemical outcome of the DH.Hexaacetyl 12 and hexabenzyl octenitol 13 were obtained in excellent yields using standard reaction conditions.Furthermore, three acetonide protecting groups could be introduced to the six hydroxyl groups present in the octenitol using 2,2-dimethoxypropane, giving compound 14, although with a significantly lower yield of only 19%, mainly due to the formation of an isomeric triacetonide, complicating purification (Scheme 1).
Targeted product 14 was clarified by 2D-NMR and 13 C NMR analysis based on an NMR study from the literature. 29The structure of the second octenitol species with also three isopropylidene groups present could not be unambiguously clarified by NMR analysis (see the SI).
With the three protected octenitols 12−14, different dihydroxylation (DH) conditions were tested to investigate the dependency of the stereochemical outcome on the used protecting group and DH conditions (Table 1).Two different protocols were considered: (1) Upjohn dihydroxylation using K 2 [OsO 2 (OH) 4 ] as the catalyst and stoichiometric amount of N-methylmorpholine-N-oxide (NMO) as a co-oxidant and (2) Sharpless asymmetric dihydroxylation (AD) using the two different mixtures of reagents, namely, AD-mix-α and AD-mixβ, to induce facial selectivity.Early attempts showed that the use of a solvent mixture t-BuOH/H 2 O/dichloromethane (DCM) in a ratio of 1:1:1 resulted in faster reaction progress due to the substrate's better solubility in the reaction mixture compared to the standard solvent system t-BuOH/H 2 O.In the case of Sharpless AD conditions, the use of commercially available AD-mix-α/β resulted in sluggish reaction progress.However, with a three times concentrated AD-mix (for preparation, see the SI) with higher loadings of osmate and ligand, a modification adopted from Kobayashi et al., 30 reaction rates could be accelerated significantly.The catalystenriched AD-mixes are stated as AD-mix-α (×3) and AD-mixβ (×3).
The different pairs of protected diastereomeric octitols were all deprotected toward 4/epi-4 to facilitate comparative analysis of diastereomeric ratios at that stage via NMR analysis.Due to overlapping signals, the quantification could not be deduced from 1 H NMR, but the two diastereomers could easily be distinguished in 13 C NMR spectra (see Figure 5).For a mixture of both sugar alcohols, four signals from the targeted symmetrical 2,3-syn product 4, each representing two carbon atoms, and eight signals from the asymmetrical 2,3-anti product epi-4 were observed in the 13 C NMR spectra.Based on a detailed study by Otte et al., 31 sufficiently similar response factors for the 13 C NMR spectroscopic integration were assumed for diastereomers even with standard 13 C NMR techniques (standard pulse sequences with broad-band decoupling and short D1 values) following two rules: (1) the two different compounds must possess the same number of hydrogen atoms and (2) the ratios of integrals need to be averaged over all different carbon atoms.This technique was used for the estimation of the diastereomeric ratio 4: epi-4.In Table 1, the screening results for all protected octenitol substrates 12−14 and DH conditions are summarized.Detailed reaction conditions and procedures for all dihydroxylations and deprotection steps can be found in the SI.
In all cases, the product with an anti-relationship between the introduced secondary hydroxyl group and the existing alkyl-/acyloxy group was preferably formed independent of the substrate and the DH conditions.Surprisingly, both AD-mixes showed selectivity toward the 2,3-anti product in all cases.Acetate and acetonide protection (Table 1, entries 1−3 and 7− 9) delivered only moderate selectivity; however, outstandingly high selectivity for the anti product was obtained for substrate 13 with benzyl protection under the used DH conditions (Table 1, entries 4−6).
Preparative Synthetic Route toward Galacto-octitol and En Route Synthesis of the C2 Epimer Epi-galactooctitol.Since none of the investigated substrates 12−14 preferably formed the 2,3-syn product under any of the applied dihydroxylation conditions, an adaption of the planned synthetic route to obtain galacto-octitol 4 with all hydroxyl groups in a 1,3-anti-relationship was necessary.The idea was to perform the dihydroxylation with high selectivity for the 2,3anti product and then invert the stereochemistry of the secondary hydroxyl group via a Mitsunobu reaction.Since in our case two free hydroxyl groups are present in the molecule, we planned for a "double Mitsunobu" reaction where both hydroxyl groups are esterified using an excess of reagents.Examples from the literature 32,33 with sterically even more demanding substrates showed that the "double Mitsunobu" of 1,2-diols is in general feasible.Therefore, the DH was carried out with the hexabenzyl octenitol 13 under Sharpless AD conditions using AD-mix-β (×3) since this set of protecting group and conditions gave the highest selectivity with ∼20:1 dr (epi-16/16) (Table 1, entry 6).However, the conditions had The Journal of Organic Chemistry to be adapted to the larger scale in terms of higher catalyst loading to increase the reaction rate, subsequently obtaining the product mixture epi-16/16 in excellent yield.At this stage of the sequence, no efficient preparative separation was achieved and the mixture of diastereomers was submitted to the next step, the Mitsunobu reaction.The reaction was carried out with PPh 3 , diisopropyl azodicarboxylate (DIAD), and 4nitrobenzoic acid, using 5 equiv each relative to the substrate.Expecting superior crystallinity of the targeted galacto-octitol 4, initially, the diastereomeric mixture 18/epi-18 with the 2,3-syn product 18 now as the major component was directly deprotected via transesterification using NaOMe in MeOH followed by reductive hydrogenation using Pd on activated charcoal in MeOH.As expected, a 20:1 mixture (calculated from 13 C NMR, vide supra) of the galacto-octitol 4 with all hydroxyl groups in 1,3-anti-relationship and the epi-galactooctitol epi-4 with one 1,3-syn-relationship was obtained.However, in contrast to the enitol species, where the perfectly configured diastereomer (lyxo-configuration) 7 could be effectively isolated via recrystallization, the minor octitol epi-4 could not be removed completely by recrystallization of the obtained mixture 4/epi-4 from MeOH/H 2 O mixtures or pure H 2 O.
Alternatively, it was found that the transformation into the corresponding octabenzyl species allows separation of the minor diastereomer epi-19 by standard flash column chromatography using a slow gradient.Therefore, the free hydroxyl groups in 16 obtained after cleavage of the ester groups were protected with benzyl groups and diastereomer 19 Table 1.Screening Results for Different Dihydroxylation (DH) Conditions and Octenitol Substrates 12−14 on Analytical Scale a Isolated yield over 2 steps.b Product ratio determined upon deprotection via integration from 13 C NMR (see Figure 5).The Journal of Organic Chemistry was obtained in pure form (>99%) after column chromatography.Galacto-octitol 4 was finally isolated after catalytic hydrogenation, cleaving all present benzyl groups.In contrast to the conditions used earlier (Scheme 2, conditions (d)), the solvent was switched to a MeOH/EtOAc mixture (1:1) since compound 19 was insoluble in pure MeOH.Additionally, the reaction was performed at 50 °C instead of rt for better solubility of the formed intermediates, allowing the cleavage of all benzyl groups at atmospheric pressure within 18 h, giving the galacto-octitol 4. Noteworthily, both perbenzylated 19 and octitol 4 are symmetric compounds, thus proving the stereochemistry to be the annotated one.En route, the C2 epimer epi-4 was synthesized directly from the initial dihydroxylation product epi-16 following the same protocol as it has been used targeting pure galacto-octitol 4, namely, benzylation and furnishing pure epi-19 via chromatographic separation followed by deprotection via catalytic hydrogenation.The epi-galacto-octitol epi-4 was not the main target within this synthesis route but is of interest when it comes to thermal properties since the distribution of the hydroxyl groups is expected to have a high influence on the melting point and melting enthalpy.
Synthesis of Decitols and Nonitol via Elongation of L- Glycero-D-manno-Heptose by IMA.Analogous to the synthetic route toward the octitols, the synthesis of the decitols was started from L-glycero-D-manno-heptose (11) that is accessible via ozonolysis of octenitol 7 as described in the literature. 26Initial attempts for the elongation of heptose 11 toward the decenitol using the standard IMA conditions showed only little conversion to the enitol species.Therefore, we had a deeper look into the IMA of heptose 11, a less reactive aldose, and successfully developed a protocol to achieve clean conversion.
An Improved Protocol for the IMA Allowing the Elongation of Less Reactive Aldoses.In general, fast and complete consumption of the sugar is of high importance when performing IMA.Otherwise, side reactions such as the formation of ethyl glycosides in an acid-catalyzed Fischer glycosylation 25 and Wurtz-type dimerization 34 of the organoindium species upon hydrolysis take over.Furthermore, the purification becomes elaborate since purification via column chromatography must be performed at the peracetylated stage in such cases.Generally, the success of the outcome of this protocol mainly depends on two properties of the used sugar: its open-chain content (OCC), representing the availability of the free aldehyde moiety in solution, and its solubility in the reaction mixture.In some cases, these issues can be overcome by decreasing the concentration of the sugar (<1%) and increasing the reagent amount, which has been shown earlier in our group in the case of D-mannose. 28However, with this protocol, the reaction progress is sluggish and fast addition of indium and efficient stirring become even more crucial.In Table 2, the investigations on the IMA for heptose 11 are summarized.The standard Barbier-type protocol did not proceed with full consumption of the starting material even when more equivalents of reagent were used (4 equiv In, 6 equiv 3-bromopropenyl acetate ( 9)) (Table 2, entry 1).Since heptose 11 is even less soluble in EtOH than D-mannose and was found to exhibit a similarly low open-chain content (∼0.03%, see the SI), first, the solvent was changed to dioxane/H 2 O (8:1) and the reaction was performed with benzoate ester 20 (Table 2, entry 2), following the strategy from Palmelund and Madsen. 25However, this also resulted in

The Journal of Organic Chemistry
low conversion of heptose 11 to the enitols (∼20% based on quant. 1 H NMR with maleic acid as the internal standard) even when more robust pivalate ester 21 27 was used as the elongation reagent (Table 2, entry 3).Next, a two-step Grignard-type protocol was tested (Table 2, entries 4 and 5).This synthetic protocol, developed by Lombardo et al. 34 for simple aldehydes, includes the preformation of the organoindium species in anhydrous THF and addition of the aldehyde species as a solution in acid phthalate buffer (pH 3) (for the detailed protocol, see Scheme 3).Using the benzoate ester 20 (Table 2, entry 4), still no complete consumption of the sugar was observed (∼40% based on quant. 1 H NMR with maleic acid as the internal standard).Only when we combined this approach with the more stable pivalate reagent 21 (Table 2, entry 5), full consumption of heptose 11 was achieved and 1 H NMR analysis after acetylation and global deprotection showed that the desired lyxo-isomer 10 was again the major isomer formed (70:30 dr lyxo:other isomers), with even slightly increased selectivity over the standard Barbier-type protocol.To the best of our knowledge, this Grignard-type protocol has never been implemented in the IMA of reducing sugars and the necessity for the pivalate reagent 21 indicated the importance of matching kinetics of the mutarotation of the equilibrating sugar and the decomposition of the reagent.This protocol, in our experience, represents the current last resort when targeting a particularly unreactive (masked) aldehyde species.
With this redeveloped protocol in hand, the elongation of heptose 11 was carried out on a larger scale (16 mmol) using the two-step Grignard-type protocol (Scheme 3).In the first step, the organoindium species was preformed by the addition of 3-bromopropenyl pivalate (21) to a suspension of indium in anhydrous THF under an argon atmosphere at 0 °C.This was then vigorously stirred at rt for about 40 min until no residual indium powder could be observed.In the second step, sugar 11 was dissolved in acid phthalate buffer (pH 3) (for details, see the Experimental Section) and added to the reagent solution at once.The mixture of decenitol diastereomers with the lyxoisomer 10 as the major component (lyxo/xylo/ribo isomers in ratio 70:18:12, determined via integration of H-3 in 1 H NMR; see the SI) was then isolated upon peracetylation, aqueous workup, and subsequent deprotection.The stereochemical configuration of the diastereomers formed in the IMA is proposed based on the findings of prior research 26−28 that revealed consistent stereochemical patterns across a range of carbohydrates, with respect to the stereochemistry at newly formed centers C-2 and C-3.The lyxo-decenitol 10 could again be obtained in pure form via trituration with MeOH and final recrystallization from H 2 O due to its superior crystallinity compared to the other diastereomers with an overall yield of 34% over 3 steps.Noteworthily, when our new protocol was applied in the attempted elongation of C9-homologue L-lyxo-L- manno-nonose (SI-1) (with equal OCC compared to heptose 11) even with increased amounts of reagents, the lyxododecenitol product SI-3 could only be isolated in a total yield of 11% (see the SI).According to our interpretation, the increasing number of additional hydroxyl groups caused an additional burden for the successful IMA in addition to solubility and open-chain content.

The Journal of Organic Chemistry
The Synthesis of Galacto-decitol, Epi-galacto-decitol, and Galacto-nonitol.With decenitol 10 in hand, the synthesis of galacto-decitol 6 and epi-galacto-decitol epi-6 was performed, repeating the developed synthetic strategy for octitols 4 and epi-4 (Scheme 4).Enitol 10 was protected with benzyl groups, giving the octabenzyl enitol 22 in good yield, and the dihydroxylation reaction was performed using AD-mix-β (3×), giving again the anti-2,3-diol product epi-23 with high selectivity (>10:1 dr).However, in the case of the C10-sugar species, the product epi-23 could be isolated in pure form at this stage via column chromatography using a slow gradient.Targeting the galacto-decitol 6, the "double Mitsunobu" reaction was performed as the next step to invert the stereochemistry of the previously introduced secondary hydroxyl group.This step gave comparably lower yields (51%), but the inversion was equally successful as above.In the next step, the ester groups were cleaved via ester hydrolysis using LiOH in THF, giving 23, which was finally submitted to catalytic hydrogenation to obtain the free sugar alcohol 6.As already discussed for the octitol (vide supra), solubility issues played a major role in the successful cleavage of all benzyl groups.In the case of the partly protected decitol 23, it was necessary to use EtOAc/MeOH in a 4:1 mixture as the solvent and perform the reaction at low concentrations (1%).Furthermore, high Pd loadings of ∼20 mol % were required, and the hydrogenation had to be performed in a high-pressure laboratory autoclave at 50 bar H 2 -pressure to achieve conversion to the fully deprotected sugar alcohol 6, although with long reaction times of up to 7 days.However, with this optimized setup, the targeted sugar alcohol 6 could be obtained in excellent yield and isolated in high purity upon trituration with EtOH and H 2 O. Again, decitol 6 is a symmetric compound, more precisely a meso compound, giving only five signals in the 13 C NMR spectrum for the 10 carbon atoms present, thus again proving the stereochemistry to be the annotated one.The epi-galacto-decitol epi-6 was obtained via catalytic hydrogenation from the dihydroxylation product epi-23, using the same reaction conditions as described for the galacto-decitol 6. En route, the galacto-nonitol 5 was easily accessible using the ozonolysis and reduction protocol that has already been established for the synthesis of higher sugar alcohols by Draskovits et al., 28 starting from the peracetylated decenitol 25.
Investigations on the Thermal Properties of the Synthesized Sugar Alcohols.With the sugar alcohols in hand, we moved on to investigate their thermal properties since the configuration of the galacto-series increased the expectation that similar high heats of fusion can be expected as calculated by Inagaki and Ishida 18 and partly already experimentally confirmed 28 for the manno-series.Among other criteria such as melting point, thermal conductivity, and thermal stability, a high heat of fusion is crucial for a highenergy storage density.For organic PCMs, high storage densities are referred to values >200 J/g.Currently, paraffins are most widely applied, even though their storage densities are at the lower end of the previously defined range, but they display excellent thermal stability. 17For example, paraffin PCMs with melting temperatures of 30−60 °C were found to be useful for solar water heating systems and integration with heat pumps. 35Organic PCMs, however, usually possess melting temperatures for applications below 180 °C, and, in general, fewer PCMs in the middle-to high-temperature range (150−250 °C) have been developed, a temperature range that was found to be covered by the synthesized sugar alcohols within this paper.
For further investigations, all synthesized sugar alcohols were recrystallized from H 2 O or MeOH/H 2 O mixtures since our previous work 28 revealed a major impact of the degree/ quality of crystallinity on the material's behavior during the heating process regarding sharp melting points and high thermal storage densities.However, despite employing slow cooling rates and variations in the solvent system, we were only able to obtain micro-to nanocrystalline material for all sugar alcohols.This observed crystallization behavior has so far The Journal of Organic Chemistry precluded the acquisition and description of X-ray structures within our work.We moved on to investigate their melting behavior using a Kofler-type micro-hot stage microscope.For the sugar alcohols with up to nine carbon atoms, sharp melting points could be observed, matching the data presented in the earlier literature (see the Experimental Section).However, for the galacto-decitol 6, decomposition at a temperature of >290 °C was observed, indicating that the material is not stable at this high temperature anymore.For the epi-galacto-decitol epi-6, melting was observed at 242−243 °C, but at temperatures >260 °C also, decomposition of the material was noticed, giving a dark, "burned" solid on the slide.Nevertheless, the trends of the melting points could be observed in agreement with the proposed criteria of Inagaki and Ishida 18 for the manno-series (Figure 6): first, the perfectly aligned sugar alcohols with all hydroxyl groups in a 1,3-anti-relationship, namely, galactitol (2, 188−189 °C), galacto-octitol 4 (221− 223 °C), and galacto-decitol 6 (>290 °C), have significantly higher melting points compared to their C2-epimers L-altritol (epi-2, 87−88 °C36 ), epi-galacto-octitol epi-4 (165−166 °C), and epi-galacto-decitol epi-6 (242−243 °C).This is expected to be the result of the superior crystallinity of the galactoconfigured sugar alcohols due to the perfect alignment of all hydroxyl groups.Second, with the length of the carbon chain in the molecule's backbone also, the melting point increases.
Next, we moved on to investigate all compounds in simultaneous thermal analysis (STA) that combines differential scanning calorimetry (DSC) with thermogravimetric analysis (TG), allowing the simultaneous measurement of melting enthalpy and weight losses of the sample.In general, weight losses can be attributed either to residual solvent present in the sample or thermal instability of the material leading to decomposition.In Figure 7, a typical STA measurement is depicted where the DSC-curve (values in mW/mg, magenta) and TG-curve (in %, black) are displayed in dependency of the temperature (in °C, x-axis).The melting enthalpy and therefore latent heat of fusion (in J/g) can be determined from the peak area and the melting point (in °C) from the onset of the curve.Detailed information on the measurements can be found in the SI.As stated earlier, crystallinity of the material played a major role in the behavior of the compound during the heating process.Samples that were not recrystallized did not give well-shaped curves.For galactitol 2, the obtained values were in agreement with the literature. 13For galacto-heptitol 3 (330 J/g) and galacto-octitol 4 (321 J/g), melting enthalpies in the same range were observed (Figure 8), the latter quite in contrast to the findings in the mannofamily. 28Some thermal decomposition during the melting process was observed for the high-melting sugar alcohols galacto-nonitol 5 and epi-galacto-decitol epi-6.Still, the measured DSC-curves could be processed to give an approximate value for their melting enthalpies (5: 387 J/g, epi-6: 146 J/g).For galacto-decitol 6, a high degree of degradation was observed, and the DSC-curve did not show a sharp peak for the melting process (see the SI, Figure S10).The measured latent heats of fusion of the sugar alcohols together with the melting points determined from the onset of the peak in the DSC-curve are displayed in Figure 8, and the corresponding curves can be found in the SI.In the case of the epi-4, the measured melting point via DSC was significantly lower than the melting point measured using the Koflertype micro-hot stage microscope.

The Journal of Organic Chemistry
In contrast to the findings for manno-configured sugar alcohols, 18,28 for the galacto-configured octitol 4, a latent heat of fusion of only in the same range as for the "parent" sugar alcohol galactitol 2 (∼300 J/g) was observed.In line with the rules of prediction, for the epi-galacto-octitol epi-4, a significantly lower value of ∼150 J/g, compared to the "perfect" galacto-octitol 4 was measured; however, this compound also showed thermal degradation during the melting process, which also exhibited a broad melting range and an early onset, which contributes to the lower value observed.Surprisingly, galacto-nonitol 5 was found to possess the highest latent heat of fusion of all investigated sugar alcohols with a value of about 390 J/g, but again, decomposition at temperatures above ∼260 °C was observed, indicating a temperature limit as applicable PCMs for structures of the discussed nature.Thermal instability at such high temperatures is not unexpected, especially considering that polyhydroxylated compounds will basically always contain some residual water.At these high temperatures, the equilibrium of H 2 O shifts more to the side of H 3 O + and OH − , which might favor the decomposition of an organic molecule by acid catalysis.This was demonstrated by Yamaguchi et al. 37,38 who investigated the dehydration and further degradation of the C6 sugar alcohols D-mannitol (1) and D-sorbitol, the corresponding alcohol from D-glucose, in high-temperature liquid water (250−300 °C) under inert atmosphere (Ar) without any acid catalysts, showing the thermal instability of sugar alcohols in the presence of water.
Nevertheless, our research contributes to a broader understanding of PCMs and their properties, regarding the relationship between structural features (stereochemistry and chain length) and thermal behavior.Additionally, we shed light on the thermal instability of sugar alcohols with melting points exceeding 240 °C, underscoring the inherent challenges associated with utilizing organic compounds as PCMs when facing this temperature window.Since high thermal stability is one of the most important physical properties for a material considered as a PCM, decomposition at temperatures close to the melting point is an undesirable characteristic for further application, despite efforts in the development of encapsulation techniques for natural sugar alcohols 15,39,40 to overcome thermal stability issues.

■ CONCLUSIONS
In conclusion, a synthetic route for higher-carbon sugar alcohols with high control of stereoselectivity for the introduced hydroxyl groups was developed, allowing the synthesis of galacto-configured sugar alcohols with an even number of carbon atoms in the backbone and their C2epimers.The IMA was shown to be a useful tool in the synthesis of higher-carbon sugar alcohols or sugar species in general, especially with the implementation of a two-step Grignard-type protocol for the elongation of less reactive sugars that allowed the synthesis of the galacto-decitol 6 with 10 carbon atoms.In total, six sugar alcohols were synthesized, two of them fulfilling all of the stated criteria with 8 and 10 carbons in the backbone (galacto-octitol 4 and galacto-decitol 6), their C2-epimers that bear one 1,3-syn-relationship (epigalacto-octitol epi-4 and epi-galacto-octitol epi-6), and additionally two sugar alcohols with all hydroxyl groups in 1,3-antirelationship but with an odd number of carbon atoms (galactoheptitol 3 and galacto-nonitol 5).Further, their thermal properties were investigated regarding their potential use as PCMs.The influence of the distribution of the hydroxyl groups and length of the carbon chain could be shown with the melting points of the synthesized sugar alcohols.For the galacto-configured sugar alcohols, latent heats of fusion above 300 J/g were observed, which is considered as outstandingly high values among organic materials investigated as PCMs.However, especially the sugar alcohols with melting points above 240 °C were found to be thermally unstable in the melting process and are therefore unsuitable in the application as PCMs despite their capability of storing rather high energy amounts in the form of latent heat.
■ EXPERIMENTAL SECTION General.The used reagents and solvents were purchased from commercial sources with a purity of >95%, unless noted differently, and used without further purification.Water-free solvents were available from a PureSolv solvent purification system by Innovative Technology or commercial sources that were stated as water-free and stored in bottles with a septum and over molecular sieves.Dowex-H + resin was washed with the respective solvent before use.
For flash column chromatography, columns were packed with silica gel from Merck with a pore size of 40−63 μm.Purification was done either by hand column or on a Buchi Pure C-850 FlashPrep System.Light petroleum is referred to as LP.
Liquid chromatography-mass spectrometry (LC-MS) analysis was performed on a Nexera X2 UHPLC system (Shimadzu, Kyoto, Japan) comprised of LC-30AD pumps, a SIL-30AC autosampler, a CTO-20AC column oven, and a DGU-20A5/3 degasser module.Detection was accomplished by an SPD-M20A photo diode array and an LC-MS-2020 mass spectrometer.Separations were either performed using a Waters XSelect CSH C18 2.5 μm (3.0 × 50 mm) Column XP at 40 °C, a flow rate of 1.7 mL/min, and with UHPLC grade water and acetonitrile containing 0.1% formic acid as the mobile phase, or a Waters XBridge BEH Amide 2.5 μm (3.0 × 50 mm) Column XP at 40 °C, a flow rate of 1.3 mL/min, and with UPLC grade water (pH 8.5, 2.5 mM NH 4 COOH) and acetonitrile as the mobile phase.
Accurate mass analysis was performed on an Agilent 6230 AJS ESI-TOF mass spectrometer with ESI ionization method or Q Exactive Focus, ESI, FIA injection, mobile phase 18% MeCN with 0.1% formic acid.
1 H NMR and 13 C NMR spectra were recorded at ambient temperature in the solvent indicated using a Bruker Avance Ultra Shield 400 MHz and an Avance III HD 600 MHz spectrometer.Processing of the data was performed with standard software, and all spectra were calibrated to the solvent residual peak.Chemical shifts (δ) are reported in ppm, coupling constants (J) in hertz (Hz), and multiplicities are assigned as s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, etc.All assignments are based on 2D-spectra (COSY, phase-sensitive HSQC, HMBC� depending on the molecule).Within the assignments, PNB is used as the abbreviation for 4-nitrobenzoyl.
Melting points were recorded using a BU ̈CHI Melting Point B 545 with a 40%/90% threshold and a heating rate of 1.0 °C/min or a Kofler-type Leica Galen III micro-hot stage microscope.
Ozone-enriched oxygen was generated using a Triogen LAB2B Ozone generator.
Simultaneous thermal analysis (STA) including differential scanning calorimetry (DSC) and thermogravimetric analysis (TG) The Journal of Organic Chemistry measurements were performed on a Netzsch STA 449 F1 Jupiter under a nitrogen atmosphere with a constant gas flow rate of 40 mL/ min and a heating and cooling rate of 10 °C/min if not stated otherwise.Samples were measured using Al pans (25 μL) with a hole in the lid.Samples were heated to approximately 30−40 °C above their melting points to prevent any decomposition at higher temperature.DSC measurements were performed on a Netzsch DSC 204 F1 Phoenix with the same measuring parameters but using closed Al pans (25 μL).
Step 1: ester cleavage: the diastereomeric mixture after the Mitsunobu reaction 18/epi-18 (164 mg, 0.152 mmol, 1.00 equiv) was taken up in MeOH (HPLC grade, 2.0 mL), and NaOMe (30% in MeOH) was added dropwise under stirring at rt until pH was about 10.After 2 h, LC-MS indicated complete conversion and the reaction mixture was neutralized by the addition of Dowex-H + .After filtration and evaporation of the filtrate, the residue was taken up in DCM (40 mL) and extracted with sat.aq.NaHCO 3 (2 × 10 mL) and brine.After drying over anhydrous Na 2 SO 4 , the solvent was removed in vacuo, giving the crude product (120 mg) as a yellow oil.Step 2: catalytic hydrogenation: the observed material was taken up in MeOH (HPLC grade, 1.0 mL), and Pd/C (10 wt %, 12 mg, 7 mol %) was added under an Ar atmosphere followed by a drop of 2 N HCl.The atmosphere was changed to hydrogen using a balloon, and the reaction mixture was stirred at rt for 14 h, as LC-MS indicated complete conversion.The mixture was diluted with H 2 O, filtered over celite (H 2 O washed), and the filter pad was washed with H 2 O.The filtrate was evaporated to dryness, giving a colorless solid (27 mg, 73% over 2 steps).According to 13 C NMR, the ratio 4/epi-4 was 20:1.Attempts to obtain pure galacto-octitol 4 via recrystallization from MeOH, H 2 O, or MeOH/ H 2 O mixtures failed.

Figure 1 .
Figure 1.Manno-series of sugar alcohols designed by Inagaki and Ishida 18 and investigated in the molecular dynamics (MD) simulations.

Figure 3 .
Figure 3. (a) D-Mannitol (1) vs galactitol (2) regarding their hydroxyl group distribution and thermal properties (determined via STA; see the Supporting Information (SI)).(b) Design of the galacto-series of sugar alcohols starting from galactitol, including examples not matching all stated criteria.

Figure 6 .
Figure 6.Experimental melting points (°C) determined on a Koflertype instrument of the synthesized sugar alcohols and the C6 sugar alcohols galactitol and L-altritol (literature value 36 ) in comparison.The synthesis of heptitol 3 has been reported in the literature.28

Figure 8 .
Figure 8. Experimental latent heats of fusion (J/g) of the sugar alcohols and their melting points obtained from the extrapolated onsets of the DSC-curves, displayed on top of the corresponding bar.In the case of the epi-4, the measured melting point via DSC was significantly lower than the melting point measured using the Koflertype micro-hot stage microscope.
■ ASSOCIATED CONTENT