Differential scanning calorimetric studies on phase transition of glucose and cellulose oligosaccharides Tatsuko Hatakeyama Research Institute for Polymers and Textiles, Sawatari, Kanagawa-ku, Yokohama, Japan
and Hiroshi Yoshida*, Chikage Nagasakit and Hyoe Hatakeyama Industrial Products Research Institute, Shimomaruko, Ote-ku, Tokyo, Japan (Received 29 August 1975; revised 30 January 1976) The phase transitions of D-glucose and an homologousseriesof celluloseoligosaccharidesup to cellotetraose have been studied using a differential scanning calorimeter. A detailed comparison of our experimental results with those reported by other workers has been made in order to derive precise information on melting and glass transitions of carbohydrates. Particular attention has been given to the static values of melting temperature of D-glucose and cellobiose. It has also been found that cellulose oligosaccharides which have more than three anhydroglucose units decompose below melting temperature.
INTRODUCTION In the course of the studies of the phase transition of glucose and an homologous series of cellulose oligosacchardies in solid state, information on precise values of phase transition temperatures such as melting and glass transition temperatures are undoubtedly necessary. Pigman et aL ' reported melting temperatures of D-~lucose in different crystal forms. Wolfrom and Dacons" reviewed the values of transition temperatures of glucose and cellulose oligosaccharides up to celloheptaose. Alfthan et al. 3 tried to estimate glass transition temperatures of several oligosaccharides from cellulose and xylan. However, those values reported seem to contain some errors from the thermodynamic point of view, since the values were not obtained under carefully defined thermal conditions and time factors. Therefore, in this study, we have re-examined phase transitions of D-glucose and cellulose oligosaccharides up to cellotetraose by the use of a differential scanning calorimeter. The isothermal melting of the carbohydrates which have melting points was carried out and attempts were made to obtain the static values of melting temperature. At the same time glass transition temperatures of glassy samples of glucose and cellulose oligosaccharides were evaluated.
EXPERIMENTAL Sample Preparation
D-glucose monohydrate obtained commercially from Merck AG was ground to a fine powder and was dissolved into hot, absolutely dry ethanol. The solution was then placed for one night at room temperature. Anhydrous a-D-glucose separated out as crystals. * Present address: Tokyo Metropolitan University, Faculty of Technology, Setagaya-ku, Tokyo, Japan. t Presentaddress: Tokyo University of Agriculture and Technology, Faculty of Technology, Koganei, Tokyo, Japan.
The D-glucose, above (obtained commercially) was also dissolved into water. The solution was placed for 48 h in a refrigerator. Crystals of a-D-glucose monohydrate separated out. /3-D-glucose was obtained commercially from Tokyo Chemical Industry Co. Ltd and contained 7.7% of a-Dglucose as determined from a gas chromatogram obtained under the following conditions using a Shimadzu GC-4AIT gas chromatograph: the column was a stainless tube packed with 5% silicone SE-30/Celite; temperature, 200°C; carrier gas, He 30 ml/min; TCD detector, 3 mV full scale. Cellobiose was obtained commercially from Merck AG. The crystal forms of the above compounds were examined from X-ray diffractograms4-7. According to the procedure reported by Miller et al. 8, cellulose oligosaccharides containing more than three anhydrous glucose units were prepared as follows. First cellulose was acetylated with a mixture of glacial acetic acid, acetic anyhydride and concentrated sulphuric acid. Then the acetylated oligosaccharides produced were separated into two portions, the anyhydrous methanol soluble portion was deacetylated with sodium methylate. The free oligosaccharides obtained were separated by ethanol-water gradient elution from a chromatographic column composed of stearic acid treated mixtures of charcoal and Celite. Each separated fraction was concentrated with a rotary vacuum evaporator and dried in vacuo. Purification of the samples was carried out by re-chromatographic treatment and the purity examined by thin layer chromatography 9. Characterization of the samples was carried out by measuring the molecular weight with a Hitachi model 115 molecular weight apparatus. Amorphous glucose and cellulose oligosaccharides were prepared by freeze-drying their aqueous solution 1°. First small amounts of the samples were dissolved in water. The solution was then frozen quickly using liquid nitrogen as a refrigerant, and dried in vacuo so as to drive the solvent off. Amorphous materials obtained were dried thoroughly in vacuo for more than two days. Amorphous glucose was also prepared by rapid cooling from the molten state.
POLYMER, 1976, Vol 17, July
D.s.c. studies on glucose and cellulose oligosaccharides: Tatsuko Hatakeyama et aL
A 0 "t0LU
Ternperoture (K) Figure I D.s.c. melting curves of glucose and cellobiose. A, e-DGlucose monohydrate; B, e-D-glucose anhydride; C, 3-D-glucose; D, cellobiose. Heating velocity, l°C/min
Measurem en ts
A Perkin-Elmer differential scanning calorimeter, DSC-II was used throughout this experiment, low temperature equipment being attached to the DSC-II when the glass transition of D-glucose was measured. Temperature was calibrated at all scanning velocity ranges using the melting temperature of indium as a standard. At low temperature regions, melting of spectrograde benzene was used for the calibration. In both cases, the foot temperature of a melting peak observed in a differential scanning calorimetric (d.s.c.) curve was taken as the melting temperature according to the procedure recommended as a common calibration method in the manual for DSC-II H. Samples were sealed into d.s.c, pans for volatile samples as this eliminates the effect of humidity. Sample wei~_Jat used asin this study was from 2 to 3 mg. Heating velocity was varied from 0.63 to 20°C/min. When crystalline samples were measured by dynamic method at the desired heating velocity, peak temperature was adopted as a criterion of melting for a matter of convenience ~z. We mixed some samples with copper powder (1:1 in weight) and pressed into thin, small plates to increase the thermal conductivity, as the samples used were rather bulky. These mixed samples were also tested and compared with unmixed ones. The following isothermal melting was carried out using the DSC-II, in order to measure the equilibrium melting temperatures of D-glucose and cellobiose. At first the sample was heated to the melting temperature which was roughly estimated from the value extrapolated to zero heating velocity from dynamic measurements carried out at different heating velocities. The heat change was recorded while the sample was maintained above the melting temperature. After keeping the sample at the temperature for a certain time, the sample was reheated to ascertain whether any crystals remained. The equilibrium melting temperature was thus determined by successively changing temperatures in the vicinity of the supposed equilibrium temperature. The glass transition temperature was defined as the point when the extension of the base line intersects with a line tangent to the maximum slope of the endothermic peak.
viously2. Melting curves of glucose and cellobiose were broad compared with those of other materials used as calibration standards such as benzene, benzoic acid and indium. The initial melting temperature of the above carbohydrates was difficult to estimate precisely from the foot temperature of a melting peak because of a gradual change in specific heat occurring from a temperature far lower than that of melting. Therefore, peak temperature was adopted in this study as a criterion of melting as described previously. In the case of cellobiose, the specific heat change was observed as a deflection in the base line of a d.s.c, curve before and after melting. Partial decomposition seems to be masked with the melting peak 13. Figure 2 shows the relation between the temperature of a peak and heating velocity for the samples of or-D-glucose anhydride and 3-D-glucose. The mleitng temperature increased with increasing heating velocity. The change of peak temperatures was remarkable compared to known values of organic compounds. Similar results were also obtained in the case of a-D-glucose monohydrate and cellobiose. Hellmuth et aL x4 suggested that super heating was taking place after considering the special character of the structure of these organic compounds. On the other hand, lchihara is suggested that the heat conductivity of an organic compound could not be ignored when the heating velocity dependency was discussed even if the sample weight was very small. Iguchi 16 and Hatakeyama 17 obtained good agreement with the results of Ichihara ~s by using extended chain-type polyoxymethylene crystal and high molecular
RESULTS AND DISCUSSION Figure i shows the d.s.c, curves of or-D-glucosehaving differ-
ent crystal forms, anhydride and monohydrate; 3-D-glucose and cellobiose respectively. Melting temperatures of these compounds measured at the heating velocity of l°C/min are not very different from the temperature reported pre-
POLYMER, 1976, Vol 17, July
i IO Hcatin9 velocity (oC/rnin)
Figure 2 Heating velocity dependency of melting peak temperature for e-D-glucose anhydride and ~-D-glucose. A, a-D-Glucose anhydride; B, 3-D-glucose
D.s,c. studies on glucose and cellulose oligosaccharides: Tatsuko Hatakeyama et al.
ated is shown in Figure 6 for the case of a-D-glucose monohydrate. As described above, the half-time of melting shown in Figure 6 was calculated from the isotherm of a sample which remained isothermal until it melted completely. In this case, however, if the temperature at which a sample is kept isothermally approaches an ideal equilibrium melting temperature, the time interval for melting is also expected to approach infinity, which cannot be attained in the experimental time interval. Moreover, the sensitivity of the d.s.c, is not sufficient to follow such a small amount of heat in unit time, since the transition occurs over a long period. However, the difference between the maximum and minimum temperatures where isothermal melt could occur in a sample converged within -+0.5°C by holding a
"6 330 I
20 3O 40 Time (mini Figure 4 Isothermal melting curves of s-D-glucose monohydrate. A, 336K; B, 334K; C, 333K
Heating velocity (°C/mini
Heating velocity dependency of melting peak temperature for e-D-glucose monohydrate. A, a-D-Glucose monohydrate; B, ~-D-glucose monohydrate with Cu powder
weight polyethylene as samples. In order to clarify the effect of super heating and/or heat conductivity on the melting of carbohydrates, the melting temperature of a-D-glucose monohydrate was compared with that mixed with copper powder, as described earlier. As shown in Figure 3, the melting temperature of a-D-glucose monohydrate was affected seriously by the conditions of the sample in the high heating velocity region. This fact suggests that the heat conductivity of a sample played an important part in heating velocity dependency. It was difficult to evaluate the melting temperature of the compounds from the data obtained by dynamic measurements, since the melting peak temperatures of glucose and cellobiose depended on heating velocity. Therefore, we tried to measure the equilibrium melting temperatures from melting curves obtained by the procedure described earlier. Figure 4 shows the isothermal melting curves of a-Dglucose monohydrate. From this Figure, it is shown that the melting of the sample takes place over a very long time interval. The same melting behaviour was observed in the case of cellobiose. The large heating velocity dependency of glucose and cellobiose seemed due to the above features of the compounds. Isothermal melting was carried out for all samples having melting temperatures. The cellulose oligosaccharides having more than three anhydroglucose units decomposed without melting. It is possible to redraw the endotherms shown in Figure 4 into accumlation type melting curves as a function of time as shown in Figure 5. From these curves, we could estimate the half-time of melting at various temperatures. The half-time thus evalu-
I0 40 I00 Time (mini Figure 5 Accumlation type of melting curves for e-D-glucose monohydrate; where H t is the heat of fusion at time t and Hoo is the final heat of fusion. A, 336K; B, 335K; C, 334K; D, 333K
E E ,.J
334 336 Temperature (K) Figure 6 Relationship between half-time of melting and the temperature for e-D-glucose monohydrate kept in an isothermal state
P O L Y M E R , 1976, V o l 17, J u l y
D.s.c. studies on glucose and cellulose oligosaccharides: Tatsuko Hatakeyama et aL Table 1 Melting (Tm) and glass transition (Tg) temperatures of cellulose oligosaccharides
Tm (K) Sample
e-D-Glucose monohydrate e-D-Glucose anhydride #-D-Glucose Cellobiose Cellotriose Cellotetraose
331 416 419 505 ---
356 419 421--428 521
Tg (K) This work*
335 364 355
* In this work, Tm was determined by isothermal melting and Tg was estimated from a d.s.c, curve at the heating velocity of 10°C/rain
Tempertur¢ ( K ) D.s.c. curves of amorphous glucose, cellobiose, cellotriose and cellotetraose near the glass transition temperature. A, DGlucose; B, cellobiose; C, cellotriose; D, cellotetraose. Heating velocity, 10°C/min Figure 7
samples. The glass transition temperatures were determined from the curves in Figure 7. The temperature errors of glass transition were within -+1°C among the runs made using different samples. No serious effects of temperature cycling was observed among the runs carried out in the temperature range from 270K to 400K. The melting and glass transition temperatures estimated above are all listed in Table i together with the data obtained previously. From Table i it can be seen that the melting temperatures estimated in the present study are rather low as a whole compared with data reported previously. This is because the effect of time factor on phase transition process is taken into consideration. As seen in Table I the glass transition temperatures of oligosacchardies increase with increasing degree of polymerization except for cellotetraose which might be contaminated by small amounts of impurities. The glass transition temperatures measured in the present study are quite different from those obtained by Alfthan et aL 3, even after considering the different techniques used. They estimated the glass transition by torsional brade analysis and the maximum temperature of damping was adopted as the criterion of the glass transition temperature. As is well known, dynamic loss in viscoelasticity does not correspond only to the main motion but also to other phase transition, such as crystallization and local mode relaxations. On the other hand, by the thermal analysis, the glass transition is observed as the specific heat change, which is clearly distinguishable from the other first order transitions. Judging from the results obtained in this study, the glass transition temperatures reported by Alfthan et ~/, 3 seem to coincide rather well with the cold crystallization temperatures which were detected in d.s.c, curves when the amorphous samples contained small amounts of water. REFERENCES 1 2 3 4
sample successively at different temperatures. The isothermal melting temperature obtained by the above method coincided well with that obtained by extrapolating values from the dynamic measurements shown in Figures 2 and 3. Table 1 shows phase transition temperatures obtained in this experiment and those reported in the literatures. Amorphous samples of D-glucose and cellulose oligosaccharides were prepared following the method described earlier. X-ray diffractograms of the samples showed the typical amorphous pattern. It was possible for crystallization to take place by conditioning amorphous samples under an atmosphere with a very small amount of water at a suitable crystallization temperature. Crystallization of amorphous samples has been discussed in another report 18. Crystallization was never observed by d.s.c, or X-ray diffraction measurement, if a sample was completely dried and handled in water free conditions. Figure 7 shows the d.s.c. heating curves of amorphous samples measured at the heating velocity of 10°C/min near the glass transition temperatures. The specific heat change was seen clearly for all
P O L Y M E R , 1976, V o l 17, J u l y
5 6 7 8 9 10 11 12 13 14 15 16 17 18
Pigman,W. W. and Coepp Jr., R. M. 'Chemistry of the Carbohydrates, Academic Press, New York, 1948 Wolfrom, M. L. and Dacons, J. C. J. Am. Chem. Soc. 1952, 74, 5331 Alfthan, E., Ruvo, A. and Brown, W. Polymer 1973, 14, 329 McDonald,T. R. R. and Beevers, C. A. Acta. Crystallogr. 1952, 5,654 Kiliean,R. C. G., Ferrier, W. G. and Young, D. W. Acta. Crystallogr. 1962, 15, 911 Ferrier, W. G. Acta. Crystallogr. 1963, 16, 1023 Jacobsson, R. A. Acta. Crystallogr. 1961, 14, 598 Miller,G. L., Dean, J. and Blum, R. Arch. Biochem. Biophys. 1960, 91, 21 Becker, E. S., Hamilton, J. K. and Lucke, W. E. Tappi 1965, 48, 60 Mann,J. Proc. Wood Chem. Syrup. Montreal. Canada 1961 p91 Perkin Elmer Manual for DSC-II. Seki,S. et al. 'Jikken Kagaku Koza' (Eds M. Otake et al. ), Maruzen, Tokyo, 1958, Vol 5, pp 324 Shafizadeh, F. and Lai, Y. Z. Carbohydr. Res. 1973, 31, 57 Hellmuth, E. and Wunderlich, B. Z Appl. Phys. 1956, 36, 309; Hellmuth, E., Wunderlich, B. and Rankin Jr., J. M. Appl. Polym. Symp. 1966, 2, 101 Ichihara, S. Proc. 5th Semin. Therm. Analy. Japan, Tokyo, Japan 1975; Ichihara, S. Proc. 23rd Meet. Polym. Sci. Japan. Tokyo, Japan 1974 p 515 Iguchi, M. Makromol. Chem. in press Hatakeyama, T. Sen'i Gakkaishi 1965, 31,289 Hatakeyama, H., Yoshida, H. and Nakano, J. Proc. 25th Meet. Wood Chem. Japan, Fukuoka, Japan 1975 p 116