Thermal conductivity of sugar alcohols

sugar alcohols have been extensively studied for thermal storage purposes. One of the recent of research has been in improving their heat charge and discharge rate by enhancing the thermal conductivity with different types of additives. However, the current literature shows a vast discrepancy in measured values of sugar alcohols. This work presents an experimental study on thermal conductivity of seven sugar alcohols. The aim is to find out the reason for the discrepancy of literature values for erythritol, mannitol and xylitol, and to present new reference data for galacticol, myo-inositol, maltitol and sorbitol. We study the impact of material preparation method, raw material grade and sensor contact on the crystalline structure and the conductivity. The crystalline structure was inspected with optical and scanning electron microscopy and X-ray diffraction, and melting properties with differential scanning calorimetry. We found that different polymorphs, grain structure and crystallite sizes can be obtained by different prep- aration methods. This caused the conductivity of mannitol, galacticol and myo-inositol to vary by tens of per-centages. Crystallization temperatures of xylitol and erythritol were found to affect their grain size but had only a minor effect on the conductivity. Overall, the conductivities of solid phase sugar alcohols were found to be within the upper range of the previous literature; based on the methods of this work, we did not find any evidence for the low and intermediate values for erythritol, xylitol and mannitol. Due to the high amorphous content of maltitol and sorbitol their conductivity was substantially lower than that of the other sugar alcohols. Thermal conductivity of liquid phases was found to accurately follow a linear relationship with the molar mass for sugar alcohols with carbon number between 4 and 6.


Introduction
Natural sugar alcohols appear in some plants and animal tissues. In addition to the medical use, sugar alcohols have been widely used as sugar substitutes due to their sweetness and low caloric content. During the last decade sugar alcohols have been extensively studied for thermal storage purposes as well. Due to their high value of latent heat, they have been regarded as highly promising phase change materials (PCMs). Typical targeted applications consider short-term thermal buffering between cold and warm temperatures [1][2][3][4][5]. Recently, due to the supercooling and cold-crystallization properties, polymer modified erythritol and xylitol have been found promising candidates for long-term storing of heat [6][7][8][9][10] as well.
One of the recent focuses on sugar alcohol and other PCM research has been the improving of thermal conductivity with different types of additives and heat enhancement structures [11][12][13]. However, literature shows large discrepancies in the solid phase data of the most studied sugar alcohols (erythritol, mannitol and xylitol), and with critical details for repeatability and transparency missing [14]. For example, methods and conditions for sample preparation and measurement are not sufficiently reported in a vast majority of the studies. For the other sugar alcohols data is almost completely missing.
The sugar alcohols included in this work are xylitol, mannitol, erythritol, myo-inositol, galacticol (dulcitol), sorbitol and maltitol. Table 1 shows current literature data for conductivity of these sugar alcohols. For example, the reported conductivity values for solid phases of erythritol and mannitol range between 0.196 and 0.89 Wm -1 K -1 and 0.28-1.32 Wm -1 K -1 , respectively. No previous data was found for galacticol, myo-inositol and maltitol.
Based on careful sample preparation, measurements and structure and melting property characterization we aim at determining reliable reference values for solid phases at room temperature and liquid phase sugar alcohols near their melting temperature and analyze the reasons for the discrepancy reported in literature.

Sugar alcohols
Crystal powders of erythritol, xylitol, sorbitol, maltitol, mannitol, galacticol and myo-inositol were used as a raw material for the preparation of samples. Table 2 presents the supplier information of the raw materials. In addition to analytic grade substances, food grade materials of xylitol and erythritol were examined.

Sample preparation
Crystalline powders were melted in a temperature controlled forced convection oven (accuracy ±0.1 • C). Because the melting temperature of myo-inositol exceeds the maximum operation temperature of this oven, myo-inositol samples were exceptionally melted and crystallized in a natural convection oven for which the accuracy of temperature control was lower (±5 • C).
For the preparation of solid phase samples, the melts were placed in silicon rubber molds and mixed to remove air bubbles created during the melting of powders. J-type thermocouples were immersed in 1-2 reference samples of each batch and the samples were cooled towards a prescribed temperature of oven (T oven ). Crystallization was initiated either (i) by seed crystals when the reference sample reached T oven , or, (ii) spontaneously by cooling when T oven was set to a sufficiently low value. The temperature T oven was held constant during the crystallization period in either case. The crystallization initialization temperature (T init ) is estimated for the cooling crystallization from the measured temperature evolution of reference samples and the oven. The accuracy of T init was typically ±3 • C and ±0.1 • C for the cooling crystallization and crystallization by seeding, respectively. The seeding was carried out with 3-5 crystals. Exceptionally, due to a very slow crystallization kinetics of sorbitol, its seeding was carried out with a large number of crystals.
When crystallization begins the temperature of the samples starts to increase. The peak of the temperature (T peak ) during crystallization was recorded from the reference samples. Note that T peak is not a unique material property. It depends not only on the crystallization characteristics and thermal properties of the material, but also on the sample size and heat transfer condition inside the oven. As the flow condition inside the oven and the sample size remained approximately unaltered during the measurements, the T peak can qualitatively be used to compare differences in crystallization kinetics of measured batches.
The radius of the solidified sugar alcohol discs were 38 mm and thickness on average ~12 mm. The discs were carefully smoothed flat and polished with a Struers LaboPol-4 device.
During crystallization, some sugar alcohols, such as xylitol and erythritol, tend to form hollow cavities at the surface and inside the bulk material. Such cavities can easily result in too low conductivity values. For those materials we prepared several additional samples and chose only samples without any visible cavities at the measurement surfaces. Furthermore, after the measurements we broke the samples that gave the lowest conductivity of each batch and verified that no cavities near the measurement surface existed. If cavities existed, the sample was excluded from study and a new sample was measured. Examples of the sample appearances are presented in Figs. 1-4. Approximately half of the spontaneously crystallized galacticol samples resulted in a more fragile structure. Those samples showed low and very inconsistent conductivity values during a single measurement as well as in repeated tests. Except for one reference sample, those are excluded from the data. All galacticol samples that were crystallized by seeds were instead less fragile and resulted in consistent and repeatable conductivity values.
Three liquid phase samples of each sugar alcohol were prepared. Sample size in liquid tests was ~1.5 ml. All measurements were conducted at 1 bar. Solid phase measurements were measured at room temperature (23 ± 2 • C) and liquid phase slightly above the melting temperature.

Thermal properties
Thermal conductivity was measured with CTherm TCi thermal conductivity analyzer, utilizing a modified transient plane source technique (MTPS). The conductivity is primarily determined by a direct measurement of thermal effusivity. Using selected material groups and their calibrated conductivity values, the analyzer converts effusivity to thermal conductivity. The functioning of calibration was verified with standard reference materials: distilled water, LAF 6720-B foam, pyrex, pyroceram and ethylene glycol (in range of 60-170 • C). These tests resulted in deviation from literature data ranging from 0 to 3%. Silicone fluid 50 cS (conductivity 0.146 Wm -1 K -1 ; supplier VWR Chemicals) was used as a contact agent for the sugar alcohol measurements. Using the known values of the reference materials, a calibration curve for silicone oil as a contact agent using pyrex, polystyrene and PVC was prepared. A 500 g weight was placed on the solid phase samples during the  1) value interpreted from a graph. 2) measured in a supercooled state.   After thermal conductivity and grain size analysis, sugar alcohol discs were broken and 10-25 mg samples were detached from the measurement surface for differential scanning calorimetry (DSC) analysis with Netzsch DSC204 F 1 Phoenix. The samples were scanned at 5 K/ min in sealed aluminum crucibles under N 2 atmosphere.

Structure characterization
Some of the sugar alcohol discs showed different visible appearances depending on the crystallization temperature (see Figs. [1][2][3][4]. This implies changes in grain structure, crystal form or degree of crystallinity. The smaller the grain size is, the more grain boundaries the sample has. Because the grain boundaries can disperse the energy carriers, here phonons, the grain size could in principle affect the conductivity. Lowered crystallinity of a sample in turn implies higher fraction of amorphous sections in the sample, which would be manifested as a lower conductivity. Optical microscope (ZEISS Axio Vert. A1) and camera (Canon EOS 1100D with Canon EF-S 18-55 mm objective lens) were used for the grain size analysis of polished solid samples. The grain size was determined using the principle of linear intercept method. Several straight lines were drawn on the sample images in two directions perpendicular to each other. The line lengths and the number of intersections between the lines and grain boundaries were calculated. As a result, this method yields an average diameter of the grains.
Scanning electron microscopy (SEM) was conducted to observe the sample structure by a JEOL JSM-7500FA analytical field emission equipment under vacuum and at 1.5 kV acceleration voltage. To prepare the specimen for SEM, a small piece containing both the inner and surface parts was separated from the sugar alcohol disks. The sample was then fixed on a metal stub with carbon tape and coated with 4 nm of gold palladium alloy using a LECIA EM ACE600 sputter coater.
The disc samples were also studied with X-ray diffraction (XRD) analysis using a Rigaku SmartLab X-ray diffractometer equipped with a rotating anode X-ray source (9 kW, Cu Kα1) and a HyPix-3000 2D detector. The patterns of the powders (raw materials of the disks) were measured as reference. The X-ray patterns were then used to determine the average crystallite size (D) via Origin Curve Fitting using Scherrer equation where K is a constant (0.9), λ is the X-ray wavelength (0.15406 nm), β is the full-width-at-half-maximum minus the instrumental peak width (radians), and θ is the diffraction angle (radians). The degree of crystallinity (X) for the samples with amorphous content was determined as follows: where a c is the area under crystalline peaks and a c + am is the area under the crystalline peaks and the amorphous background.

Solid phase
The results for solid phase conductivity given in Tables 3-5 show the average value ± standard deviation of each batch. Typically, 3-4 solid phase samples were measured from each preparation batch. Samples from a spontaneously crystallized mannitol batch showed higher deviation compared to other batches. Therefore, eight samples were measured from that batch. Due to a possible structure inhomogeneity of samples and possible inhomogeneity of heat pulse provided by the measurement device, each sample was measured from three angular positions, rotating the discs 0 • , 120 • and 240 • on a sensor surface. This means that in overall 9-18 repeated tests for each solid phase batch was done.
The crystallization of several samples of the spontaneously crystallized erythritol batch started near 67 • C. Other sporadic temperatures occurred as well, such as 52 • C reported in Table 4. For some other sugar alcohols only an approximate range for T init can be reported. Xylitol and sorbitol did not crystallize spontaneously between room temperature and melting temperature. Maltitol did not crystallize at all. Instead, it formed an amorphous solid. This finding was confirmed in the calorimetric and XRD studies (see Chapters 3.2 and 3.3.2).
Comparing Tables 3-5 to Table 1 we realize that the measured conductivities in this study are near (xylitol, mannitol) or slightly above (erythritol) the maximum values presented in previous data. No clear effect in the grade of the raw materials was found. The food grade xylitol showed either slightly higher or slightly lower values depending on T init , to those of analytical grade samples. The variation in T init had an effect on conductivity of xylitol; e.g., conductivity of analytical grade had a maximum (1.22 ± 0.04 Wm -1 K -1 ) at T init = 55 • C and a minimum (1.07 ± 0.02 Wm -1 K -1 ) at T init = 89 • C. For erythritol the dependence of conductivity on T init was found to be negligible (within the measurement accuracy).   Table 3 Thermal conductivity of xylitol at 23 • C. All samples are crystallized by seed crystals. T init = crystallization initialization temperature, T peak = maximum observed temperature during crystallization, T oven = temperature of the environment (oven) during crystallization. The conductivity of mannitol and galacticol resulted in 33% and 15% variation between different crystallization methods, respectively. Standard deviation of cooling crystallized mannitol samples was substantially high. Indeed, the conductivity of eight measured samples varied from 0.73 Wm -1 K -1 to 1.01 Wm -1 K -1 . One fragile sample of galacticol was measured for reference. Its conductivity (0.62 Wm -1 K -1 ) was significantly lower to any other samples. However, the measurement of that sample was challenging (see Section 2.2 sample preparation). Therefore, the value represents rather the order of magnitude than an exact value. Conductivity of myo-inositol varied substantially between disks of the same batch. The discs (see Fig. 4) with blotchy visual appearance resulted in 62% higher values to samples with dendritic appearance. T init could not be controlled or clearly detected for myo-inositol. Therefore, it is possible that the difference in visual appearance and conductivity originates from different T init values.

Liquid phase
The thermal conductivity of the liquid phase near the melting point is presented in Table 6. The values for the three isomers (sorbitol, galacticol, mannitol) were found to be equal within the measurement accuracy. Furthermore, the conductivities of erythritol, xylitol, mannitol, sorbitol and galacticol follow highly accurate linear relationship (R 2 = 0.99914) with their molar mass and carbon number (Fig. 5). The carbon number of these compounds varies between four to six and the molecular structure is similar as one hydroxyl group is attached to one carbon (see Table 2). Additionally, we tested glycerol which has the same type of molecular structure with carbon number three. The linear relationship with molar mass does not hold for glycerol accurately but the order of magnitude of conductivity still follows the carbon number and molar mass. Maltitol molecule instead has a ring-structure and its order of magnitude cannot be predicted by the carbon number (=10).
The liquid phase conductivity of erythritol corresponds well to data presented in literature (Table 1). Measured conductivities of xylitol and mannitol differ from some of the previous publications. Liquid phase of myo-inositol was not measured because its melting temperature exceeds the limit of the conductivity device.

Effect of surface contact on conductivity measurements
We found that an incomplete sample preparation that leads to insufficient contact between the sample and sensor can very easily result in low conductivity values for the solid phase sugar alcohols. For instance, if erythritol and xylitol samples were insufficiently polished, they resulted in values ranging from 0.3 to 0.7 Wm -1 K -1 (compare values in Tables 3-4), even though samples appeared sufficiently polished by visual inspection. Careful polishing by hand with sandpaper resulted typically in conductivity values 5-15% lower to those polished by machine. Also, the repeatability of tests was considerably lower for handpolished samples. In the present method the use of liquid contact agent reduced the risk of harmful air layers appearing between the sample and sensor surface. Therefore, measurement techniques that do not use any contact agent (e.g., TPS) the need for proper shaping and polishing the samples is even higher compared to the present study.
In addition, as mentioned earlier, sugar alcohols such as xylitol and erythritol tend to form hollow cavities on the surface and inside the bulk material during crystallization. Such cavities easily result in too low Table 4 Thermal conductivity (k) of erythritol at 23 • C. T init = crystallization initialization temperature, T peak = maximum observed temperature during crystallization, T oven = temperature of the environment (oven) during crystallization. NR = not recorded.  Table 5 Thermal conductivity of galacticol, maltitol, mannitol, myo-inositol and sorbitol at 23 • C. T init = crystallization initialization temperature, T peak = maximum observed temperature during crystallization, T oven = temperature of the environment (oven) during crystallization. NR = not recorded.  conductivity values. Preparation methods such as crystallizing the melt directly on the measurement sensor surface can thus face serious challenges in forming a proper contact between the sample and sensor. In liquid phase measurements the main contact problem occurs due to gas bubbles that may be trapped at or near the sensor surface. However, that can be easily to controlled and recognized by visual inspection.

Melting properties
The average and standard deviation of melting properties, and the conductivities of discs from which the DSC samples are detached are presented in Table 7. Note that these conductivities correspond to the values of a single disc. Tables 3-5 show instead the average of a disk batch.
All results correspond to the samples prepared from analytical grade powders. Melting entropy is determined based on the measured melting temperature T m (in Kelvins) and melting heat Δ H m from ΔS m = ΔHm Tm . Erythritol and xylitol showed relatively consistent calorimetric properties between powder and disks crystallized at different temperatures. This data corresponds well to previous literature data presented e. g. in Ref. [43].
Mannitol, myo-inositol and galacticol showed substantial variance in melting heat within samples extracted from the same disk, within disks prepared at different temperatures, and within disks of different visual appearances. The variance was found to be more prominent for disks crystallized spontaneously by cooling compared to those crystallized with seeds at higher temperature. Interestingly, the melting heat inside a single myo-inositol disc showed distinct lower and higher values. Those values are presented in separate rows in Table 7. For example, the samples taken from a disk with a blotchy appearance (see Fig. 4) resulted in values 271 ± 5 J/g and 347 ± 5 J/g.
The previous literature for calorimetric properties of galacticol, mannitol and myo-inositol also show substantial discrepancy. Based on literature reviews and experiments presented in Refs. [5,43] different studies have found melting temperature and melting heat to vary between 180 and 190 • C and 246-354 J/g for galacticol, 155-169 • C and 238-338 J/g for mannitol, 221-230 • C and 209-266 J/g for myo-inositol, respectively. In addition, a melting heat of 352 J/g was found for myo-inositol in Ref. [44] which corresponds closely to the upper value found in this study.
According to Pitkänen et al. [45] melting temperatures of α, β, γ, δ and κ polymorphs of mannitol are between 164.1 and 167.9 • C. Because the melting temperatures of this study are all within this range, and because the melting heats varied substantially, it is highly probable that the mannitol samples contain different polymorphs. However, discrepancy in measured values appears within those few studies published on mannitol polymorphism. For example, melting temperature of 157 • C for the δ form of mannitol has been reported in Ref. [46]. Similarly, the variation in melting heats of galacticol and myo-inositol imply the presence of polymorphs.
Sorbitol showed two distinct melting peaks in each DSC curve. Some deviation in melting temperature and melting heat appeared between the powder and disks. The lower melting heat of sorbitol disks compared to powder implies lower crystallinity of disks. The melting heat of maltitol was negligible which implies that the samples are fully amorphous.

Optical and scanning electron microscopy
Optical microscopy was applied to examine the influence of crystallization temperature on the grain size. The polished surfaces from which the conductivity was measured are examined. Xylitol and erythritol were analyzed in detail. For the rest of the sugar alcohols visible crystal or grain structure was not detected in the optical microscopic analysis.
Erythritol and xylitol showed grain size (diameter) ranging from 0.20 mm to 17.5 mm and 0.15 mm-3.59 mm, respectively. Because of the large grain size of some samples, only few grains existed for their analysis. This decreased the accuracy of the linear intercept method. Therefore, grain sizes presented indicate an order of magnitude of the grain size.
Nonetheless, xylitol samples showed decrease in grain size with reduced crystallization temperatures, as illustrated in Fig. 6A. Erythritol did not display as evident temperature dependence as xylitol, but large grains (>3 mm) formed at higher temperatures of T init = 85 • C and T init = 114 • C, while grains remained smaller (<3 mm) at lower T init of 55 • C and 68 • C, as shown in Fig. 6B. Nucleation density and crystal growth velocity form a bell-shaped curve in a function of crystallization temperature; as the crystallization temperature is decreased below melting temperature, nucleation density and crystal growth velocity first increase, and then decrease towards zero. However, the shape of nucleation density and growth velocity curves rarely coincide, hence the grain size may show non-linear dependency to the crystallization temperature. This could explain the grain size behavior of xylitol and erythritol. Erythritol also showed large variations in the grain size within the samples of the same batch (same T init ), which is manifested as large error bars in Fig. 6B. Presumably, random variations during crystallization, such as variation in number of formed nuclei, and relatively fast crystallization caused large variance in erythritol's grain size. Fast crystallization can also increase the temperature of the supercooled sample towards melting temperature, because latent heat is released faster than the sample can transfer it to the environment. Consequently, crystallization rate might increase as the crystallization progresses (compare T init and T peak in Tables 3-4).
The quality and different suppliers of erythritol (Luontaistukku, Alfa Aesar and Tokyo Chemicals) and xylitol (Danisco and Alfa Aesar) did not show significant influence on the grain size, as can be seen in Supplement Fig. S1, which shows grain size of all analyzed samples. Fig. 7 depicts grain structure of erythritol and Fig. 8 for xylitol for different grain sizes. Linear intercept analysis showed that the shape of erythritol grains does not depend on the direction, i.e., the grains show mostly regular shape. On the other hand, xylitol samples exhibited some variation in shape, when crystallization temperature was decreased to 25 • C. It appears that low crystallization temperature promotes smaller and elongated fan shaped grains. This is depicted in the leftmost image in Fig. 8, where several grains appear to originate from a nucleation point forming an elongated fan shape. This could be influenced by the interplay of reduced nucleation rate and crystal growth velocity; the speed of crystallization of xylitol has been found to be ~8% at 25 • C of its maximum value near 75 • C [47]. However, detailed analysis of the factors influencing the grain size and shape of sugar alcohols was out of the scope of this study.
We selected one high conductivity sample (HCS) and one low conductivity sample (LCS) of mannitol, galacticol and myo-inositol for the SEM study. All images were taken from a lateral direction to the surface of the disk. We can see (Fig. 9A-D) that mannitol forms long rod-shaped grains. For the HCS (9A-B) grains are oriented mostly parallel to the normal of the surface (i.e., direction of the heat pulse created by the measurement equipment) but for the LCS (9C-D) grains are oriented to the direction of tangent of the surface. This means that the longitudinal rods of HCS operate as better conduction paths in direction of the conductivity measurement (with less grain boundaries) compared to the LCS, which can explain the differences in conductivity values.
SEM images of galacticol (Fig. 10) also show clear differences between the crystalline structure of LCS and HCS. While the LCS has a granular structure, the HCS structure is more uniform. The granular structure may explain why the several LCSs were fragile. SEM image of myo-inositol LCS (Fig. 11B) shows several cavities in the scale of  μm. This may impair the conductivity somewhat.

X-ray diffraction
The X-ray patterns of the disc samples are presented in Supplement  Fig. S2. These patterns were used to determine the average crystallite sizes (D) of the crystalline samples. A crystallite can be defined as a single crystal phase related to the coherent diffraction domains in X-ray diffraction. Table 8 shows the average and range of crystallite size, and the conductivity of the discs from which the XRD samples are detached. The crystallite sizes varied from 34 nm to 185 nm for different sugar alcohols. The conductivity was found to increase with increasing crystallite size of galacticol, erythritol and mannitol. A similar dependency between thermal conductivity and crystallite size was previously reported for BaTiO3 nanoceramics [48]. The sugar alcohol samples were produced either by spontaneously through cooling or seed-aid crystallization at different temperatures, which can explain the differences in crystallite size. For instance, the average crystallite size (and grain size as well) of xylitol samples increased by increasing crystallization temperature which can indicate the thermally activated process of the crystallization, while in the case of galacticol, the average crystallite size decreased at higher crystallization temperature.
It was observed that erythritol, xylitol, mannitol, galacticol and myoinositol samples were highly crystalline. Maltitol sample showed a characteristic amorphous peak at around 2θ = 20 • which is typical for glass-forming materials and indicates some level of molecular orientation and hydrogen bonding [49]. Sorbitol demonstrated semi-crystalline peaks on an amorphous background. The degree of crystallinity for the sorbitol sample was determined at around 65% and the crystallite size was the smallest (14.6 nm) among the measured samples. Compared to the rest of the sugar alcohols both samples showed low thermal conductivity, 0.51 Wm -1 K -1 for maltitol and 0.6 m -1 K -1 for sorbitol, analogous to that of glassy materials.
The X-ray patterns of the discs resemble those of their initial crystal powders (Supplement Fig. S3), suggesting that they were from similar crystalline forms yet with varying degree of crystal orientation in different samples. However, differences in the patterns of the discs (e.g., myo-inositol and mannitol samples) can suggest presence of different polymorphs.

Discussion
The existence of different crystalline structures can create different    types of structural defects. Consequently, that can change the phonon scattering from these defects changing the heat transport. The results show that different polymorphs, grain orientations and crystallite sizes can be obtained by different preparation methods of solid phase mannitol, galacticol and myo-inositol. As a consequence, thermal conductivity of these sugar alcohols can vary by tens of percentages.
There seems to be ambiguity in current literature corresponding to the melting heats and temperatures of different polymorphs of sugar alcohols. A detailed study on the topic is needed for deeper analysis on their effect on conductivity.
Different crystallization temperatures alter the grain and crystallite size of xylitol and erythritol. That has a small but still noticeable effect on conductivity of xylitol but not on erythritol. Also, the melting properties of xylitol and erythritol were found to be independent of preparation methods of the samples. Due to the high amorphous content of maltitol and sorbitol their conductivity was substantially lower than that of the other sugar alcohols.
We can see from Table 1 that the reason for discrepancy in the previously reported conductivities of xylitol, erythritol and mannitol is not based on different measurement techniques; e.g., the most frequently used methods, TPS and LFA, seem to both result in low, medium and high values. The thermal conductivities found in this work support the upper bound of values found in previous literature (see Table 1). No evidence for the previously reported medium or low conductivity values were found here. Such values were detected only for samples of which surface was not sufficiently well polished or which had internal hollow macroscale cavities inside their crystalline structure.

Conclusions
Thermal conductivity of seven sugar alcohols were analyzed. We examined whether different sample preparation methods could alter the crystalline structure sufficiently enough to explain the vast discrepancy   in literature values of the solid phase conductivity of erythritol, xylitol and mannitol. We conclude that different polymorphs, grain shapes and orientations, and grain and crystallite sizes can be obtained by different preparation methods. This can cause the conductivity of mannitol, galacticol and myo-inositol to vary by tens of percentages. For xylitol the variation was found less significant and for erythritol insignificant. For most part these variations cannot explain the discrepancy in previously reported data. However, the conductivities of solid phases found in this work support the upper bound of the previous data. Incomplete sample preparation was found to result in substantial decrease in conductivity values, which could explain at least part of the discrepancy found in the literature. Due to the high amorphous content of maltitol and sorbitol their conductivity was substantially lower than that of the other sugar alcohols.
The liquid phase conductivity was found to linearly depend on the molar mass for all five sugar alcohols with carbon number between four to six.