J. Phys. Chem. 1996, 100, 16457-16459
16457
Calorimetric Investigation of a New Solid Phase in Triphenylphosphite Kees van Miltenburg* and Koos Blok Department of Interfaces and Thermodynamics, Debye Institute, Utrecht UniVersity, Padualaan 8, 3584 CH Utrecht, The Netherlands ReceiVed: April 30, 1996; In Final Form: July 17, 1996X
Adiabatic heat capacity measurements were made on triphenylphosphite, molecular formula C18H15O3P, between 140 and 335 K. The results confirm the existence of a phase which is intermediate in stability between the undercooled liquid and the crystal phase. This phase, called the glacial phase, was recently reported for the first time. At 227 K this phase disappears as the crystallization starts. The melting temperature was found to be 297.7 ( 0.2 K, the melting enthalpy 25090 ( 100 J mol-1. The glass transition temperature was 201.8 K, the ratio between the heat capacity of the undercooled liquid and the glass at this temperature being 1.68. T0, the temperature at which the entropy of the undercooled liquid equals that of the crystalline phase was calculated to be 169 K. A discontinuity in the specific heat of the liquid phase was found close to the melting point. This is related to the existence of a narrowly avoided thermodynamic transition point.
Introduction Recently1
the discovery of a new phase in triphenylphosphite (denoted as TPP) was reported. This phase is an amorphous solid phase, formed in a narrow temperature range just above the glass transition temperature. The authors denoted this phase as a “glacial phase”, thereby stressing the amorphous character and distinguishing it from the vitrous phase. In an earlier publication Kivelson2 discussed a new thermodynamic theory of supercooled liquids. In this theory a narrowly avoided transition point (T*) close to the melting transition plays an important role. The occurrence of such a point may give rise to (very) small discontinuities in physical properties, as for instance in the heat capacity of the liquid. We3 report here on adiabatic calorimetry measurements of TPP in the temperature range 140-330 K. This range includes the “normal” glass transition, the formation of the glacial phase from the undercooled liquid, the crystallization and the melting of the compound. By cooling only to about 245 K, direct crystallization from the undercooled liquid was followed. Adiabatic calorimetry is the better method (compared to DSC) to study thermal effects in compounds with slow relaxation phenomena, as the measurement can be performed with high precision and with very slow heating rates. Thermal equilibration can be performed, if necessary for weeks. Furthermore the method is an absolute one. Different measurement series can be tied together when they have been performed up to an equilibrium state. In this case the measurements were all made to above the melting point. We assumed that the enthalpy of the liquid phase above the melting point is a state function and the enthalpies calculated in the different series were shifted to overlap. Experimental Section The calorimeter used has been described before.4 The calorimeter vessel (volume 11 mL, mass 19 g) is suspended in an evacuated space and surrounded by temperature-regulated shields. The temperature of the inner shield is always kept very close to the temperature of the vessel. The wires needed for a Corresponding author. Phone: +31 30 2532386. Fax: +31 302533946. E-mail:
[email protected]. X Abstract published in AdVance ACS Abstracts, September 15, 1996.
S0022-3654(96)01223-3 CCC: $12.00
temperature measurement and electrical heating are very thin (0.08 mm diameter) and pass a wire-heater which is also regulated. The measuring method is an alternating one. Successive periods of stabilization (ts) and heating (ti) are applied. During the stabilization period the temperature is followed as a function of time. The second half of the period is used to analyze the temperature drift by a linear fit. The standard deviations of these fits generally did not surpass 3040 µK. The precision of the apparatus is on the order of 0.01% or better (depending also on the amount of compound); the accuracy checked with standard compounds like sapphire and n-heptane is on the order of 0.1-0.2%. The temperature drifts in the stabilization period give the possibility of calculating the energy exchanged with the surroundings as a function of temperature and time:
qexchange ) C dT/dt in which C ) total heat capacity of sample and vessel and dT/ dt the temperature drift mentioned. Deviations of qexchange from “normal” values as found in an empty vessel experiment or in a temperature region where the compound is in a thermodynamic equilibrium give a very sensitive indication of enthalpy relaxation in the compound. Within one series of successive measurements qexchange is usually determined to within 5 µW; between different series the difference can amount to 20 µW. The calculation of heat capacity data and the enthalpy increments are normally performed with the observed exchanged energy. In the region where relaxation processes take place, a polynomial fit of the heat exchange with the surroundings as a function of temperature is used, this function is calculated with the data in the temperature regions where the compound is in thermodynamic equilibrium. TPP was purchased from Acros Chimica, purity 99%, and used as received. Warned by ref 1 in which the breaking of glass ampules due to shrinking and expanding of the compound in the subsequent phase transitions was reported, we used only 1.286 g, corresponding to about 1 mL. We expected that when the compound was distributed in a thin layer on the bottom of the vessel, these changes would not damage it. After the measurements however, a faint rim was visible in the wall of the vessel. Five series of measurements were made. When measuring from low temperatures up to the stable liquid phase the same © 1996 American Chemical Society
16458 J. Phys. Chem., Vol. 100, No. 41, 1996
Figure 1. Schematic view of the different phases and the thermal history of the five heat capacity experiments. The existence regions of the vitrous, the glacial, and the crystalline phase are given as shaded surfaces. The experiments, indicated by the paths numbered 1-5, start at room temperature at the left side of the figure. Measurements were made in the heating parts of the paths.
settings were used. These were ts ) 600 s, ti ) 758 s, so all measuring points were 1358 s apart. Cooling from the liquid state was done by breaking the vacuum in the calorimeter compartment with helium gas. A mean cooling speed of 4 K min-1 was obtained, sufficient to avoid crystallization. Measurements. The existence regions of the different phases are given in Figure 1. The three phases at low temperature, vitrous, glacial, and crystalline, increase in stability from left to right. Five series were measured. The thermal history of these series is given schematically in Figure 1 by the paths numbered 1-5. In the left side of Figure 1 the experiments start with cooling from room temperature. Horizontal parts of the paths indicate waiting periods under adiabatic conditions. This does not imply that the temperature of the sample stayed constant as relaxation to a more stable state could take place. This was the case when the horizontal line crosses the vertical domain boundary of two phases. Heat capacity data were collected in the rising parts of the paths. Experiment 1: Starting in the Vitrous Phase. The vessel was cooled to 148 K. No crystallization was observed during cooling. From this temperature a continuous series of heat capacity measurements was made up to 320 K. The heat capacity data are given in Figure 2 (open triangles) and in Figure 3 the enthalpy increment is given with the same symbols. A glass transition was found at 201.8 K. The undercooled liquid formed after the glass transition started to transform to the glacial phase at 217 K with a large exothermic effect. A second exothermic effect was found starting at 230 K where the glacial phase transformed into the crystalline phase. This crystallization was so slow that the heat capacity data between 230 and 280 K are nonequilibrium data. The energy released by the crystallization gives rise to a temperature increment, which results in low heat capacity data. The enthalpy path followed can however be calculated as discussed in the experimental part. The melting point was found at 296.7 K. Experiment 2: Heat Capacity of the Glacial Phase. From experiment 1 we knew the temperature at which the undercooled liquid phase started to transform into the glacial phase (217 K). After cooling to 200 K, we heated the vessel to 217 K and followed the temperature evolution under adiabatic conditions. A spontaneous temperature rise occurred to 220 K. The main part of this effect was completed in 2 h, but after 20 h still a small exothermic effect persisted. We assumed that the
Letters
Figure 2. Experimental molar heat capacities. (4) Series 1: From vitrous to undercooled liquid to glacial to crystalline to liquid. (]) Series 2: From glacial to crystalline to liquid. (O) Series 3: Crystalline phase with still some metastability. (2) Series 4. The crystalline phase and the melting. (0) Series 5. The undercooled liquid phase.
Figure 3. Enthalpy values for the different phases. The relative enthalpies have been shifted to overlap at 305 K. The symbols used for the different series are the same as in Figure 2.
formation of the glacial phase was almost complete. After recooling to 147 K, series 2 (open diamonds) was started. The heat capacity rose sharply at the end of the existence range of the glacial phase; at 227 K crystallization started. The kinetics of the crystallization process where again so slow that the heat capacity data up to 280 K are nonequilibrium data. Experiment 3: Heat Capacity of the Crystalline Phase. To obtain the stable crystalline phase, we repeated the thermal history of experiment 2, but this time, after forming the glacial phase, the sample was heated till 226.6 K and stabilized for 25 h. The temperature rose from 226.6 to 229.0 K. After recooling, series 3 was made (open circles). It is clear from Figure 2 that some metastability still persisted. Experiment 4. This time we cooled only to 245 K. At this temperature direct crystallization from the liquid could be expected according to ref 1. Crystallization started within minutes after restoring the vacuum. The crystallization rate was constant over a large part of the process. After 2 h an exothermic effect could still be seen. So we heated the sample to 270 K and stabilized it overnight. After recooling to 160 K, series 4 was started (filled triangles). Judging from the heat capacity data alone the stable crystalline phase was formed. The more sensitive drift function revealed a small exothermic effect around 285 K. Experiment 5: Heat Capacity of the Liquid Phase around the Melting Point. As the heat capacity of the liquid around the melting point is of theoretical interest, we undercooled the sample to 270 K. No crystallization took place, and we made
Letters
J. Phys. Chem., Vol. 100, No. 41, 1996 16459
TABLE 1: Coefficients of Polynomial Fits of the Enthalpy of the Different Phases and the Appropriate Temperature Ranges: H ) (A0 +A1T + A2T2 + A3T3) J mol-1 phase and range
A0
A1
A2
A3
glass phase 140-200 K glacial phase 150-225 KÅ liquid phase 206-335 K crystalline phase (experiment 4) 175-275 K
-18 393 -70 370 -59 620 -24 235
50.4680 824.2940 298.471 -16.1281
0.489 075 -3.922 68 0.273 70 0.644 39
0 0.008 323 0 0
Figure 4. Molar heat capacities of the liquid phase around the melting point. The different symbols refer to the different experiments as discussed in the text and in Figure 2.
a series of heat capacity measurements between 270 and 335 K (open squares). These data, together with the heat capacity data of the liquid above the melting point collected in the other series are given in Figure 4. Summary of the Results The enthalpy curves given in Figure 2 were fitted to polynomial functions. The coefficients are given in Table 1. Used within the specified range, these functions do reproduce the data of Figure 3 faithfully. The enthalpy of melting was calculated from experiment 4 to be 25 090 ( 100 J mol-1, the melting temperature is 297.7 ( 0.2 K. The enthalpy difference between the glass phase and the glacial phase can be calculated as a function of temperature from the polynomial functions. At 180 K the difference is 7084 ( 50 J mol-1. Between the glacial phase and the stable crystalline phase the difference at 180 K is 5744 ( 50 J mol-1. The heat capacity of the three phases (vitrous, glacial, and crystalline solid) at low temperature diminishes in this order. At 150 K we found respectively Cp(vitrous) ) 197.1 J mol-1 K-1; Cp(glacial) ) 192.8 J mol-1 K-1 and Cp(crystalline) ) 189.8 J mol-1 K-1. Using secondorder polynomial fits for the heat capacity of the undercooled liquid and the stable crystalline phase, the Kauzmann temperature5 was calculated to be 169 K. The ratio of the heat capacities of the undercooled liquid phase and the vitrous phase is 1.69 at 201.8 K. Discussion The existence of a new “glacial” phase cannot be proven by adiabatic calorimetry. The observed phase, intermediate between the liquid phase and the crystal phase can, from
calorimetric data alone, also be explained as the formation of a metastable crystalline phase. We recently did some experiments in which this was the case.6.7 The relaxation processes observed in those cases however showed a different behavior. Crystallization followed an Avrami-Kolmogorov like function, meaning that it started slowly and as more crystalline material was formed the process accelerated. Proof of the existence of the glacial phase comes from X-ray data,1 so we have to interpret the rise of the heat capacity of the glacial phase just before 227 K as the onset of the return to the liquid state. This implies a new glass or glacial transition, which is interrupted by the crystallization. Direct crystallization from the undercooled liquid (at 245 K) is clearly the best method to prepare the stable crystalline phase. The existence of a discontinuity in the heat capacity of the liquid phase is of theoretical interest. Our data do show a discontinuity in the slope of the heat capacity just above the melting point. This effect is not far from to the estimated2 value of T* of 312 K, which was derived from fitting viscosity data to a model. In the measurements of the empty vessel no such effect was observed. This excludes trivial explications such as errors produced by thermometer calibration or the use of materials with phase transitions in the construction of the vessel. Still we find that as the observed effect hardly rises above the “noise” of the measurements, more measurements from different laboratories are needed. Supporting Information Available: All experimental heat capacity data are available in the supporting information. The file TPP.asc contains these data in the form: Temperaturemolar heat capacity-relative enthalpy. An additional measurement of the “normal” glass transition with longer stabilization time and smaller heat inputs is also given in this file (3 pages). Ordering information is given on any current masthead page. References and Notes (1) Ha, A.; Cohen, I.; Zhao, X.; Lee, M.; Kivelson, D. J. Phys.Chem. 1996, 100, 1. (2) Kivelson, D.; Kivelson, S. A.; Zhao, X.; Nussinov, Z.; Tarjus, G. Physica A 1995, 219, 27. (3) A description of our group is to be found on our home page of the world-wide web: the address is http://www.chem.ruu.nl/gt/www/gt.html. (4) Miltenburg, J. C. van; Berg, G. J. K. van den; Bommel, M. J. van. J. Chem. Thermodyn. 1987, 19, 1129. (5) Kauzmann, W. Chem. ReV. 1948, 43, 219. At the Kauzmann temperature the entropy of the undercooled liquid equals that of the crystalline phase. The polynomial functions used to calculate this point are as follows: Liquid: Cp ) 410.873 - 0.284483.T + 0.0015131T2. Crystalline phase: Cp ) 41.4543 + 0.861315T + 0.000744T2, the entropy of fusion being 84.59 J K-1 mol-1. (6) Miltenburg, J. C. van; Lebrun, N. ;Eerden, J. P. van den; Foulon, M. J. Cryst. Growth 1996, 160, 141. (7) Miltenburg, J. C. van; Eerden, J. P. van den; Oonk,H. A. J.; Gallis, H. E. Thermochim. Acta 1995, 259, 103.
JP9612238