Precise Phase Transition Measurements of Organic Materia Is by DifferentiaI Therma I Ana lysis D. A. VASSALLO' and J. C. HARDEN Polychemicals Department, E. 1. du Ponf de Nemours &
b A differential thermal analysis method is presented for the precise, semiautomatic determination of phase transition temperatures. Several DTA methods are compared for estimating transition temperatures. The effects are given of variation in thermocouple position and gage, and in heating rate. A recommended procedure gives a precision of *0.3" C. for 2- to 5-mg. samples over the temperature range from -150" to +450° C. Liquid or solid samples are tested directly; gases, after condensation into cooled tubes within the apparatus. Conventional melting point capillary tubes are used in the simultaneous determination of melting, boiling, and inversion temperatures of organic materials.
D
thermograms of organic materials taken over a wide temperature range have been proposed for use as ['fingerprints" of the niaterials studied, because of the unique nature of the complex scans (6). This use has been limited generally to intralaboratory studies, since such thermograms are strongly dependent on operational variables. The situation is compounded by the lack of generally accepted or readily available standardized , proliferation of applicaequipment. 4 tions and a paucity of definitive work have promoted duplication of effort, so that experience with the equipment a t hand remains the primary requirement for meaningful results. This is evident even in the determination of well known phase transition temperatures, T t , where precision has been usually reported as 1 3 " to 5" C. (6, 9, 10). Other merits of the differential thermal analysis (DTA) technique are likely to be discounted if this supposedly objective instrumental method is not shown capable of better precision and accuracy. This relatively poor precision is adequate for characterizing most inorganic systems, where transitions usually are well separated. For studies of organic IFFERENTIAL
Present address, Sabine River Works, Orange, Tex. 1
132
ANALYTICAL CHEMISTRY
Co., Inc.,
Wilmington, Del.
substances most workers have used the same techniques. Because of the complexity of thermograms of organic compositions, however, greatly improved precision is essential, since overlapping or closely occurring bands are the rule rather than the exception. Precision is affected by instrumentation and by sample dilution. The most common sampling condition for inorganic compounds employs addition of reference material to the sample compartment, to equate heat capacities as much as possible. Often samples are diluted 60 to 90% with reference material, resulting in broad thermal peaks. For this reason transition temperatures have been estimated from different parts of the peak. For example, recommendations have included the first deviation from base line (8), the inflection point (7), or intersection with the base line of the low temperature side of the peak ( 5 ) ,and the peak temperature (8). Although DTA is a dynamic method, melting points, T,, and boiling points, Tal occur a t specific temperatures. Therefore, heating rate and sample size may have a marked effect on the observed temperature, since a finite time is required to undergo transition. If the heating rate is high or the sample is large, an error can be expected as in manual methods for measuring melting point. Just as the response of a thermometer affects manual melting point determinations, the size and response
9 P R E AMP
P R E AMP
1 ~
t
r-----l
1
TEMP.
PROGRAMMER
Figure 1. Block diagram of differential thermal analysis components
of the temperature measurement device in DTA can be limiting. With the foregoing in mind, several DTA methods for measuring T, were studied to minimize the effects, if any, of heating rate, sample size, and detector response. An improved DTA technique has been developed which provides rapid, precise, and accurate T r measurements based on the use of small undiluted samples and continuous temperature measurement within the sample. A precision of *0.3" C. is easily obtained over a wide range of heating rates. Liquid or solid samples are conveniently handled in melting point capillary tubes into which the thermocouples are inserted. To extend the application of the technique, a cell was devised to cover the temperature range from -150" to $450' C. This extended range required blanketing the sample compartment with flowing dry inert gas to prevent condensation a t low temperature and oxidation or hydrolysis a t high temperature. DESIGN
OF
APPARATUS
With the exception of the temperature programmer ( I ) , the electronic components of the DTA apparatus are commercial products for sensitive differential temperature measurement, accurate temperature measurement, and X - Y potentiometric recording. The block diagram in Figure 1 illustrates the flow of thermocouple signals and the general plan of the apparatus. The differential thermocouple e.m.f. is preamplified by a Leeds & Northrup low level d.c. preamplifier KO.98358 and fed to the Y axis of a Moseley Model 4s X - Y potential recorder (15 x 10 inch recording surface). The temperature e.m.f. is similarly fed to the X axis of the recorder with or without preamplification, depending on the sensitivity required. -4 preamplifier has been included in the temperature-measuring circuit when accuracy better than 10.5" C. (the reading error of the recorder at its narrowest range) is desired, but the basic recorder sensitivity has proved sufficient for most purposes. The preamplifier was used to test the various methods without limitation from the
PROGRAMMER
t
TEMP ‘C.
t
COOLANT
readability of the recordings. When the temperature preamplifier is used, a “bucking potential” from a voltage divider is placed before the amplifier in series with the thermocouple signal, to increase the range of temperature measurement. Both temperature and differential temperature e.m.f. can be attenuated a t the recorder. Since the maximum gain of the preamplifier is 200X, the narrowest Y range is 2.5 pv. or 0.04’ C. on the 0.5-mv.-per-inch range of the recorder. Kormal operational sensitivy is 25 pv. per inch for differential temperature measurement. The temperature programming thermocouple is in a “feedback” circuit with a proportionally controlled linear temperature programmer (1). Temperature programming from any selected temperature is achieved by a “bucking” reference voltage, usually a thermocouple placed in a bath a t the selected temperature. The cell, shown in cross section in Figure 2, was designed to accommodate small samples contained in 1.5 to 2.0 X 30 mm. melting point capillary tubes (KIMBX 34505). It is contained in a standard 1-quart vacuum flask. The cartridge heater (30 watts 1 X 0.25 inch, Hot Watt, Danvers, Mass.) is in the center of the 1 x 1 inch aluminum block with sample and reference holes placed symmetrically about the heater and a hole to accommodate the temperature programming thermocouple close to the heater. A double coil of copper tubing serves as a cooling line. One coil is immersed in coolant in the lower part of the vacuum flask, while the second coil surrounds the block. The block is cooled by a flow of air or nitrogen through the lower coil, thence to the coil surrounding the block. The outer sheath is stainless steel of lower thermal conductivity. All components except the heatercooler cell are housed in two interlocking Emcor FR-14A cabinets.
-
The cell and DTA system has several features which provide speed and simplicity of operation with versatility. Rapid controlled heating and cooling are achieved by placing both the heater and cooler in contact with the blockthe heater a t the center and the cooler at the pcripherp. Heating rates of
T.C.
BLOCK
t
I,
TIC.
TEMP. REF. TEMP T.C.
Figure 2. Cross section of heater-cooler differential thermal analysis block
AT -
7
1
SAMPLE T.C.
REF
SAMPLE BLOCK
Figure 3. Thermocouple junctions for temperature programming, temperature, and differential temperature measurements
2.5’, 5’, lo’, 20’, 25’, and 30’ C. per minute can be selected from the programmer. It is thus possible to select a heating rate commensurate with the resolution and thermal effect being studied. The cooling may be controlled by adjusting the rate of flow of inert gas or liquid coolant through the cooling coils. This feature gives a temperature range of -150’ to $450’
C.
Samples are contained in disposable capillary tubes, which are easily filled and packed. All thermocouple junctions are arranged on a Jones terminal block (Figure 3), which is set into the Marinite cover of the cell. Unusable or broken thermocouples may be quickly replaced. The two methods of connecting the thermocouples are shown in Figure 3. Thermograms recorded with respect to block temperature are obtained by connections shown in the solid line. An ice reference junction was used for all temperature measurements. When recording with respect to sample temperature, the temperature measurement circuit is modified, as shown by the dashed lines, and the preamplifier is grounded. By use of extension leads, the cell may be placed a t a distance from the electronic components. Thus, if a noxious atmosphere or samples which degrade to form noxious gases are used, only the cell need to be moved to a hood. The method may be applied objectively when little sample is availablee.g., a gas chromatographic fraction or an expensive reagent.
a heat capacity curve. The sample and reference are packed in 1.5- to 2.0-mm. melting point capillary tubes cut t o a length of 30 to 40 mm. Between 2 and 5 mg. of sample are used. The heat capacities are adjusted by adding reference substance, 100- to 200-mesh porcelain, to the reference tube. Porcelain is thermally inert over B wide temperature range and is not hygroscopic. Silicone oil also performed well as reference material, but has temperature limitations. The differential thermocouples are centrally positioned in the sample and reference, and the tubes inserted into the heater block. Other variables are adjusted. The proper blanketing atmosphere is selected, with the proper heating rate, starting temperature, and sensitivities on the recorder, The temperature programmer is started and the recorder pen actuated. If the variables have been selected properly, the instrument requires no attention until the run is completed, (A limit switch can be included to prevent heater burnout.) In scouting runs, a faster heating rate is used and the large “peaks” are attenuated during the run. Proper variables can then be estimated from this scan. While the apparatus is cooling, another sample can be prepared and positioned in the capillary. The thermocouples are cleaned by burning off organic residues. An unusable thermocouple can be replsced in a matter of seconds a t the termmal block connection.
PROCEDURE
RESULTS AND DISCUSSION
After connecting suitable thermocouples -e.g., gold-constantan or copper-constantan for low temperature (-150’ to +150’ C.), or Chromelalumel for high temperature (-50’ to +450° C.) -and their transmission wires, the circuits are checked for continuity. Since most of our work has involved endothermic phenomena, the differential thermocouples are adjusted to give a positive deflection for a temperature lag in the sample compartment. This results in a scan which resembles
Comparison of Methods of Measuring Transition Temperature. To
compare various methods for nieasuring T t , the above apparatus \vas used with constant cell geometry. Chrome1 - Alumel thermocouple outputs were taken from different sources and the temperature n as estimated from different portions of the DTA peak. The tn-o major conditions were temperature measurement in the reference material (equivalent to block VOL. 34, NO. 1, JANUARY 1962
133
l l
I
ABC
D
-
TEMP. INCREASING
Figure 4. Transition temperature estimation methods
temperature) and temperature measurement in the active sample. The temperatures estimated for melting point, T,, or boiling point, Tb, were selected from the most often recommended portions of the DTA peak, as illustrated in Figure 4. Point A is the intersection of the extrapolated straight-line portion of the low temperature side of the peak with the base line, and point B is the inflection point of the low temperature side. This inflection point has been obtained electronically by derivative techniques ( 3 ) . Point C is extrapolated peak temperature and point D the extrapolated return to base line. Table I gives the sample or reference temperatures a t which the various points occurred during melting of benzoic acid and boiling of toluene a t a heating rate of 10" C. per minute, using 40gage Chromel-Alumel thermocouples.
Table I.
With the exception of point D in Table I, T I estimates using sample temperature are generally closer to the true values. The closest estimates of T t were achieved by use of sample temperature a t point C or reference temperature a t point A . Similar data resulted when thermocouple gage and heating rate were varied. The difference between the use of sample temperature a t point A or point C was usually less than 1" C., with the exception of the case in which 40-gage thermocouples were used a t high heating rates to estimate boiling point. Tbestimated for toluene a t points A , B, and D are higher than a t point C. This is probably due to superheating which is detected by the very responsive 40-gage thermocouples. Even a t a 40" C. per minute heating rate, 28-gage thermocouples show only an abrupt stop in temperature rise. Some of these effects are shown in Figure 5, where the character of boiling endotherms under various conditions is illustrated. The relatively poor results obtained for boiling points compared to melting points by measuring reference temperature could be related to superheating also. The effect of heating rate and thermocouple gage is shoEn in Table I1 for the two best methods (A and C). Point A , representing the best measure using reference temperature, compares with the worst measure using sample temperature, point A , but contrasts poorly with the best measure from sample temperature, point C. The precision, a, of any measurement at 10" C. per minute with 28-gage thermocouples was 0.5" C. (m.p.), 0.7" C.
Comparison of Various Methods for Transition Temperature Measurement"
Benzoic Acid,&Melting Point, "C. Temperature Measured Reference Sample
a
b e
A 121.5 120.3
Method Reference Sample See Figure 4. NBS, m.p.
Merck, b.p.
=
A 112.5 114.6
B
C
D
125.0 121.o
128.0 121.7
131.0 131.9
Toluenec Boiling Point, "C. C B 114.2 113.2
115.9 111.2
1
I SAMPLE TEMP.'
I
REF. TEMP.
t
"i ____)
TEMP. INCREASING
Figure 5. Character of boiling erldotherms using sample and reference temperature measurement Rh l o o c
(b.p.) for point A reference and 0.3" C. (m.p.), 0.3" C. (b.p.) for point C sample. When various heating rates and thermocouple gages are included as in Table 11, the precision using sample temperature remains constant (a = 0.3 m.p.; 0.3 b.p.), whereas precision is decreased for reference temperature measurement (a = 0.7 m.p.; 1.9 b.p.). Since outstanding precision and accuracy were indicated for both melting and boiling points by measuring sample temperature a t point C, peak temperature, with wide variations in heating rate and thermocouple size, this method was selected for other studies. Variation of T t with Heating Rate. Since there was little difference in accuracy between 40- and 28-gage thermocouples, the 28-gage thermocouple )vas selected as a standard because of its ruggedness, although the peak heights developed were about 15% less. Sensitivity was not a problem, for even a t 5' C. per minute Rh, a peak height of 0.8" C. (50 p v . ) was obtained on 4 mg. of benzoic acid ( A H I = 33.9 cal. per gram), while a differential of 0.05" C. was easily dis-
D TRlSTEARl N
116.2 118.8
121.8".
= 1.0"
boiling range including 110.6" C. MARLEX 50 50 POLYETHYLENE MZLEX
l
Effect of Varying Heating Rate and Thermocouple Size on T , Estimation Benzoic Acid, M.P., "C. Toluene, B.P., "C. T.C. Gage Rh, 'C./hfin. Ami Csample Aref Csarnple 109.1 111.1 120.8 121.8 28 10 111.3 121.7 112.5 121.5 40 10 111.1 121.8 109.5 28 40 120.6 121.9 116.6 111.3 40 40 122.0 U 0.7 0.27 1.9 0.27
I\
Table 11.
134
0
ANALYTICAL CHEMISTRY
I
,
,
25
50
A M Y N I U M , NITRATE
75 100 TEMPERATURE ,OC.
125
, 150
Figure 6. DTA scans of representative materials
tinguished with this system. Table 111 shows the variation of T , measurement over a wider range of heating rates. For comparison, both reference temperature (point A ) and sample temperature (point C) were studied. Again, sample temperature is superiorT , is constant to k 0 . 2 " C. even with the partially crystalline Marlex 50 polyethylene. T , is constant when sample temperaturr is measured, whereas the T , through reference measurement increases with heating rate. Representative thermograms of ammonium nitrate, Marlex 50 polyethylene, and tristearin are shown in Figure 6. These materials represent three types of morphology. Ammonium nitrate undergoes three crystalline transitions between room temperature and its melting point at 170" C. Marlex 50 polyethylene has a broadened melting endotherm, normal to partially crystalline high polymers. Further examples of this behavior can be seen in the work of Ke (4). Tristearin is monotropicthe lowest melting crystalline form of this polymorphic material is obtained by crystallization from the melt. As tristearin is heated, the various polymorphic forms melt and sequential recrystallization and melting occur.
Table 111.
(Tf, "C., and method of measurement) Benzoic Acid
6
Trn AT
i
e xo
-135'C.
Aref 121.5 121.6 122.1 124.0 125.5
Rh
5 10 15 25 40 80
Table IV.
n-Butane n-Pentane n-Hexane n-Heptane n-Octane n-Nonane n-Decane n-Dodecane Benzoic acid Water Toliiene - -. - ...
Marlex 50 .4mf 136.0 138.0 138.7 139,5
CWlWle
121.8 121.7 121.9 121.9 121.9 121.8
Benzene Acetic acid Alathon 10 polyethylene resin Marlex 50 polyethylene resin Teflon TFE fluorocarbon resin Teflon FEP fluorocarbon resin Delrin acetal resin
Melting Point, T oC. Found Reported -135.5 -129.7 -95.3 -90.6 -56.8
-135.0 -129.5 -94.5 -90.3 -57.0
...
...
iii .8
iii.8
0.0
0.0 ...
-50 -25
0
t25
TEMPERATURE,'C.
Figure 7. Low temperature DTA scan of n-butane
Crystallization of tristearin from solution gives only the highest melting form. A low temperature, simultaneous determination of T , and Ta for a normally gaseous material is shown in Figure 7 . A small amount of nbutane was condensed into a capillary tube and run from -150" to $10" C.
134.2 134.4 134.2 134.2 134.2 134.4
Boiling Point, 7'" C. Found Reported -0.5 36.2 69.0 98.2 125.5 150.2 173.0 215.5
-0.55 i36.0 68.8 98.4 125.6 150.7 174.0 2L6.0
100: 0 111 1
100:0 110 6 80.1 118.1
~~~
5.2 16.5 110.5 134.2 327.5 272.0 170.5
5.5 16.6 110.5" 134.5" 327. O4 272.5O 171.on
~
80.5 118.4
... ... ... ... ...
... ...
... .. ...
Melting points taken with Kofler hot stage microscope, or by x-ray techniques.
The Tt's for hydrocarbons are particularly sharp. The transition temperatures of several other materials are shown in Table IV, using a 28-gage thermocouple and a 15" C. per minute heating rate. T t measurements below 100' C. were taken with a calibrated gold-constantan thermocouple, those above 0" C. with a Chromel-Alumel thermocouple, and the 0" to 100" C. range mas used as a check. Bverage error calculated as the average deviation of the measured values from reported values was 0.3 for T , and 0.45 for Tb. This average error thus approaches the standard deviation of ten determinations of benzoic acid T , ( u = 0.2" C,). It is probable that even more precise data can be obtained by DTA with strict control of variables such as rigid standardization of thermocouples and greater amplification of the temperature signal t o reduce systematic errors.
LITERATURE CITED
( 1 ) Dal Sogare, S., Harden, J. C., ANAL. CHEM.31, 1859 (1959). ( 2 ) Frederickson, A. F., Am. &!ineralogist 39, 1023 (1954). 13) Camobell. C.. A N A L . , _ Gordon. S.. CHEM.31, 1188 (1959). (4) Ke, B., in "Organic Analysis," Vol. IV, p. 361, Interscience, New YorkLondon, 1960. (5) Keavney, J. J., Eberlin, E. C., J. Appl. Polymer Sci. 3, 47 (1960). f6') ~, Morita. H.. Rice. H. M., AXAL.CHEW 27, 336 (1955). ' (7) Partridge, E. P., Hicks, V., Smith, G. W.,J . Am. Chem. SOC.63, 454 (1941). (8) Smyth, H. T., J . Am. Ceram. SOC. 34, 221 (1951). (9) Stross, F. H., Abrams, S. T., J . Am. Chem. SOC.73, 2825 (1951). (10) Varma, M. C. P., J . Appl. Chem. 8 , 117 (1958).
I , : -150 -125 -100 -75
Csample
DTA Transition Temperatures of Various Compounds
Compound
a
Tb -0.5'C
Variation of Melting Point with Heating Rate
I
,
RECEIVEDfor review June 30, 1961. Accepted October 9, 1961. Delaware Science Symposium, Wilmington, Del., February 15, 1961.
VOL. 34, NO. 1, JANUARY 1962
135