fore, a theoretical injection time was established from the CI peak by comparison with C, injections. Retention times were then calcu1:tted relative to the C, peak. Reproducibility of Pyrolysis. Early in the investigation the reproducibility of the cracking procedure was determined. Separations were made on a 1-meter dinonyl phthalate column. Eight consecutive runs were made on a copolymer containing 15 wt. methyl acrylate. Peak areas were determined relative t o the CT peak. The results are shown in Table 11. Separation Procedure. The precut column was sized t o pass pyrolyzates np t o Clo hydrocarbons in 4 minutes. The O D P N column then selectively retained the polar tcomponents such that the Clo peak eluted before methanol. Since all the materials boiling above Clo were retained by the precut column and back-flushed, the alcohol and estrr peaks were? free from inter-
ference by higher fragments. The 20inch section of Carbowax 1000 served two purposes. It functioned as a "getter," preventing ODPN from bleeding into the detector, and if modified the ODPN selectivity sufficiently t o effect separation of the methyl acetate from methanol. Initially LAC 296 was used as a "getter" but was replaced because it failed to effect a complete separation, LITERATURE CITED
Table II.
Reproducibility of Pyrolysis
Component
+ methanol Cs+ methyl acetate
Cg
Av. relative area 0.931
1.617 0.245 C, 1.000 Methyl methacrylate 0.235 C8 0.697 Ca 0.465 ClO 0.555 Methyl acrylate
Std. dev. f0.084 =to.203 f0.023
...
*0.020
f0.063 f0.071 f0.095
(1) Burns, C., Brauer, G. M., Forziati, A. F., Abstracts, p. 118, 35th Meeting,
Intern. Assoc. Dental Research, Atlantic City, N. J., March 1957. (2) . , Clamp$, B. H., ANAL.CHEM.35, 577 (1963). (3) Davison, W. H. T., Slaney, S., Wragg, A. L., Chem. Ind. London 1954, 12.56
(4) Haslam, J., Hamilton, J. B., Jeffs, A. R., Analyst 83,66 (1958). (5) Haalam, J., Jeffs, A. It., J . A p p l . Chem. 7 . 24 11957). (6) Radell; E. 'A., Strutz, H. C., ANAL. CHEM.31, 1890 (1959).
(7),Strassburger, John, Brauer, G. M., E orsiati, A. F., Abstracts, 36th Meeting, Intern. Assoc. for Dental Research; J . Dental Res. 37, 86 (1958). (8) Strassbiirger, J., Brauer, G. M., Tryon, M., Foraiati, A. F., ANAL. CHEM.32, 454 (1960). RECEIVEDfor review June 5, 1963. Accepted July 10, 1963. Division of Analytical Chemistry, 144th Meeting, ACS, Los Angelcs, Calif., April 1963.
Dif f e renti CII The r ma I An a lysis A ppa ra tus EDWARD M. BARRALL I!, JAY
F. GERNERT, ROGER S. PORTER,
and JULIAN
F. JOHNSON
California Research Corp., Richmond, Calif.
b An apparatus far the differential thermal analysis of organic compounds has been designed crnd constructed to operate in the temperature range -100' to 500' C. with maximum flexibility and sensitivity in operation. Thermograms are recorded on an x-y recorder. Temperciture differences between sample and reference as small as 0.10' C. can be amplified to occupy the whole y-axis of the chart. A bucking potenticmeter and temperature axis amplifier are arranged so that any range from 300' to 1 ' C. may be centered on the recorder and the temperatures nieasured. Linear programming and a recorded chart of the heating rate error are provided. A thermogram of anisaldazine is shown with fusion cind liquid crystal peaks located to f0.05' C.
D
IFFERENTIAL TI3ERMAL ANALYSIS
(DTA) was first applied in the fields of geology :md geochemistry for the measuremeni; of phase transitions and chemical reactions in thermosensitive clays and minerals (6). The apparatus designed for this purpose has, by reason of the wide thermal range covered and Imge thermal processes studied, been rather crude and imensitive. Normal tolerances stated
for many machines have been a reproducibility of peak location to f 5 " C. and a temperature differential of 2' to 10" C. for full-scale deflection on the AT axis (6). There has been an increased interest in DTA devices capable of measuring differential temperatures equivalent to 0.5' C. between sample and reference materials in the temperature range -100' to 500" C. This interest has been due largely to the pioneering efforts of Bacon Ke (5) and others (2, 9, IO) in the field of organic DTA. The transitions of interest in organic and polymer thermal analysis (glass transitions, melting points, crystalline struetural changes, and decomposition temperatures) involve relatively small amounts of energy compared to phase and chemical transitions in inorganic DTA. For this reason and others, most commercial DTA machines suitable for mineral and inorganic DTA have limitations which do not permit their direct application to organic and polymer DTA. Three commercial machines have recently been introduced which reportedly have the necewary sensitivity for organic DTA (7, 5, 11). In all cases the machines are limited either by the low power output of thr temperature programmer or to :t single
block design because of the unique measuring elements employed. The apparabus described here takes into account the requirements of sensitivity and temperature range required for the precise DTA of polymers and other organic materials in the temperature range -100" to 500" C. The design of the DTA furnace block or chamber is not limited by the power output of the programmer. The measuring elements employed, thermocouples, can be connected in any way necessary to carry out any previously described technique of DTA. Although many of the general features have been described previously (1, 4 , 6, 9, IO), the capabilities of this machine appear to he unique in the field of DTA. APPARATUS
The apparatus consists of two h i c sections-a temperature controlling seetion and a recording section as shown in Figure 1. In the temperature controlling section are the programmer, hrating error recorder, cycle control, power amplifier, and sample furnace block. The recording section contains the sample and reference thermocouple system, the differential ( A T ) and system temperature ( T ) amplifiers, zero sunpre4ori potentiometer, and an x-y recorder. VOL. 35, NO. 12, NOVEMBER 1963
e
1837
RECORDING SECTION
=.'
7
:rential thermal analysis apparatus ate constantan leads
J
~
L
~
Y
U ~ LL U ~ ~ L L " L U1-
. ..
.
~ r ~ r l a u r ;VuJ
.
.
a Leeds and lsorthrup (L and N ) 10170 program control unit Connected to a L and N 0- to 0.5-mv. center zero Model H recorder and a L and N Series 60 current adjust controller equipped with a rate and reset feedhack. The Model H recorder has been modified so that the error signal arisiue between the furnace sensine thermkouple and the program voltage from the 10170 control unit is displayed about the chart center zero. I n this way, deviations from the linear temperature program as small as 0.1' C. can he noted. This furnishes a permanent record of the heating rate for a thermogram. The control cams of the 10170 control unit are cut according to the temperature-voltage curve of the sensing thermocouple with the cam slope proportional to the desired heating rate. The temperature-voltage relationship was established by calibrating the thermocouple with an NBS platinum resistance thermometer. By the choice of the proper cam, motor, and gear train, any heating or cooling rate b e tween 0.1' C. and 30' C. per minute in the temperature range -looo to 500" C. may be obtained. The 16gauge copper-constantan thermocouple is embedded in the furnace block a short distance from the heating e l e ments. An ice junction is used as the reference. The output of the program unit is used to control a two-stage, single phase, 2-kw., self-saturating, saturable core reactor. Twenty amperes of current is available. The furnace, shown in Figure 2, is heated by two rod heaters wrapped around the stainless steel furnace block. The heaters are of the coaxial type with Incoloy outer sheath, magnesium oxide packing, and nickelchrome center element. The bars are connected so that the magnetic flux will be a minimum. The spaces h e tween heater coils and block are filled with Thermon, a heat-conducting cement. The furnace block has eight holes drilled longitudinally around the perimeter. These holes conduct liquid carbon dioxide which is used as coolant in the downward program mode (programmed cooling).
1838
ANALYTICAL CHEMISTRY
Recording Section. The differential and. sample temperature measurmg probes oi the D T A consist of four copper-constantan thermocouples -one in the sample, one in the reference, and two in the ice bath as shown in Figure 1. Both the absolute temperature of the sample and the differential temperature of the sample and reference are obtained with a single thermocouple. The thermocouples were made by silversoldering Leeds and Northrup No. 30-gauge copper-constantan, glass-insulated, duplex thermocouple wire. The silver-soldered couples, when carefully cleaned of flux and with the b a d size made less than 1mm., were identical to fluxless welded couples which had been recommended previously ( I ) . The thermal lag of an average couple experiencing a 10' C. instantaneous change was 50 miliiseconds from 100' to 110" C. The mierovolt amplifiem for differential and sample temperature measure ment are modified Leeds and Northrup Model 9834-A d.c. amplifiers. The modiikations, expansion of the voltage dividers by the addition of extra r e sistances, were used to obtain extra ranges of attenuation. The amplifier output at full-scale input is 10 mv. Ten microvolts will drive the modified ampliers to full scale at maximum sensitivity. Amplifier linearity has been maintained under experimental conditions when the input was SUEcient to drive the amplifiers 80 times their rated full-scale range. In series with the sample temperabure thermocouple is a zero suppression unit which will suppress up to 22 mv. of thermocouple signal with a resolution of 1 MY. In order to maintain circuit stability in the divider resistances and soldered joints, the unit is thermostated at 100" =t0.05" F. This highly stable potentiometer in combination with the Leeds and Northrup amplifier permits kemperature range centering and amplification. Thermograms are recorded on a Moseley 15- by IO-mv., I-mv.-per-inch x-y recorder. AT is recorded on the 10-mv. full-scale Y-axis, and 2' is recorded on the 15-mv. full-scale X-axis. A I-minute marker is added to the Taxis with a second pen to furnish an additional time axis.
Figure 2. sembly A. B. C.
D. E.
F.
Furnace-sample block as-
Glass bell jar Heater Refrigerant entrance Refrigerant exit Thermocouple leads Top cap of romple wells
The sample core design has been reported previously (1). The design of the sample cells is shown in Figure 3. This cell can accommodate diluted samples on 500-mesh carborundum, solid polymer samples, and solid samples fused directly to the thermocouple bead. Gas-tight seal is effected by screwing the second cap into the sample well. The desirability of measuring the system temperatuie, T,in the ceder of the active sample is well known (1, 7). This necessitates the measurement of the differential and system temperature with the same thermocouple. MacK e n i e (6) has reviewed a system where the sample thermocouple is connected on the constantan side to both the reference and ice bath thermocouple. Because of the input impedance of the Leeds and Northrup amplifiers and the unequal I R drop introduced by them across the sample and reference thermocouples, a back e m f . appears at the AT leads which amounts to several millivolts when the system is operated above 50' C. This introduces on the thermogram a pronounced drift in the exothermal direction. The introduction of a dummy ice hath thermocouple on the reference side loaded with a 12,000-ohm resistor removes this drift. The thermocouple circuit is shown in Figure 1. The absence of any drift arising from extraneous thermocouple effectsis shown in Figure 4. The instrument is not limited to any specific sample core configuration and can be rapidly adapted to any furnacesample core design which appears to
AN ISAL DA ZIN E
I
0.lS.C
SCAN A
'
I
'
I
'
I
I
'
100
50
0
' '
I ' 150
I;,
,
200
, ,
I 250
, ,
SAMPLE TEMPERATURE,'C
1
SCAN B
SAMPLE TEMPERATURE,'C Figure 4. 1.
Figure 3. Sample cell A 8.
C
Differential thermogram of anisaldazine
Fwion point
2. Liquid m p t o l tramition point 0.0150-gram sample containing 15% aniioldozine on 500-mesh corborundum. 4.6" C. per minute heating rate Scan A and 8, AT 0.05' C. = 1 inch Scan A, T = 1 mv. per inch Scan 8, T = 1 mv. per 8 insher
Thermocouple Sompie cuvette Top cop
he useful for a given DTA study. The assembled furnace and core with bell jar for inert atmosphere is shown in Figure 2. DISCUSSION
Stable, linear heating rates with only small fluctuations from linearity are necessary for high-sensitivity DTA. Hay (4) has shown that a loC . sudden departure from linearity can appear as a "glass transition point" on the resultant thermogram. DTA represents a dynamically balanced system which can he easily disturbed by heating rate nonlinearity. The Leeds and Northrup control arrangement when connected as described to the furnace block shown in Figure 2 provides a heating rate which departs from linearity by only 5' C. over the range 0" to 200' C. No local cycling effects were noted in blank rims at highest sensitivity-0.01' C per inch on the AT axis. The x-y plotting technique is employed as a convenience in the espression of the thermograms. The 1minnte pulse provided hy the second
pen permits translation of the thermograms to the time base system if desired. The I-minute pulse provides an additional secondary check on the heating rate separate from the error record of the controller. The total scan of most x-y recorders is limited in both the x and y directions. The Moseley x-y recorder has a 15-mv. x-axis which is equivalent to a 0" to 300" C. scan with a copper-constantan thermocouple system. For routine survey work this scan is adequate; hut for precise determination of peak temperatures, as we11 ES obtaining a sufficiently peak wide to integrate, some s-scale expansion is necessary. When a thermogram is recorded on a time-base strip chart system, the necessary expansion is generally obtained by speeding up the chart drive and/or amplifying the temperature signal which is recorded separately. Obviously, sneh a solution is impossible with an x-y system. Using the Leeds and Northrup amplifier and a snitable bucking voltage, any range of 300" to 1' C. may be spread over the full x-axis. In Figure 4, Scan A shows an actual 0" to 280" C. scan of anisaldazine. The peaks are too sharp to integrate, and the tem-
perahre axis is compressed so that it is irnpossihle to measure temperature more accurately than &lo C . Scan B shows an expanded scale scan with large peak areas and temperatures measurable to =!=0.05° C . This scale was obtained by bucking out 6.000 mv. of the system temperature thermocouple signal (equivalent to 135.80" C.) before amplification. The remaining T signal was amplified eight times over that of Scan A. Wit,h this arrangement the recorder pen will not cross recorder zero until the T signd becomes 6.000 mv. The recorder x-axis in B i? 1 mv. per 8 inches as opposed to 1 mv. per 1 inch. Amplification of the difference signal up to 1000 times is possible with this apparatus. Any interval from 300' to 1' C. from -100" to 500" C. mav he snread over the 15-inch chart range. The instrument is not limited to use as a differential thermograph. The temperature programmer and recording section can be used with the torsion spring microbalance system of the type described by Fujii, Carpenter, and Meussner (3). This instrument has also been operated with a thermal erpansion nppa,rat,us of our design using VOL. 35, NO. 12, NOVEMBER 1963
1839
a movable core transformer which will be discussed in a later publication. With alteration of the programmer slide wire voltage and with a suitable block and thermocouple system, thc temperature range can be extended to 1200O c. LITERATURE CITED
(1) Barrall, E. M., 11,Rogers, L. B., ANAL. CHEM.34, 1101 (1962).
(2) Borchardt, H. J., Daniels, F., J . Am. C h a . SOC.79, 41 (1957). (3) Fujii, C. T., Carpenter, C. D., Meuesner. R. A.. Rev. Sci. lnstr. 33.362 (1962). ( 4 ) Hay, A. W., 144th National Meeting, ACS, Lo8 Angeles, Calif., April 1963. (5) Ke, Bacon, “Organic Analysis,” Vol. I V , J. Mitchell, Jr., Ed., Interscience, New York, 1960. (6) MacKenzie, R. C., “The Differential Thermal Investigation of Clays,” Mineralogical Society, London, 1957. (7) hfuller, Ralph H., ANAL. CHEM.35,
No. 4, 103A (1963). (8) Stone, R. L., Zbid., 32, 1582 (1960). (9) Vassallo, D. A., Harden, J. C., Zbid., 34, 132 (1962). (10)Vold, M. J., Zbid., 21, 683 (1949). (11) Watson, E. S., O’Neill, M. J.,
Justin, J., Brenner, N., Pittsburgh Conference for Analytical Chemiat,ry and Applied Spectroecopy, March 19b3.
RECEIVEDfor review June 25, 1963. Accepted September 16, 1963. Division of Analytical Chemistry, 144th Meeting, ACS, Los Angeles, Calif., April 1963.
Simu I tu neous Differentia I The rma I Ana lysis a nd Thermogravimetric Analysis Using the Open-Pan Type of Sample Holder H. G. McADlE Ontario Research Foundation, 43 Queen’s Park, Toronto 5, Ontario, Canada
b The correlation of information obtained from differential thermal and thermogravimetric analyses is facilitated by performing both operations simultaneously. Tra ditionaI sample holder designs have complicated results by preventing adequate control of the sample environment, and thermocouple positioning has often led to erroneous sample temperatures and uncertain kinetic data. A new instrument is described capable of the simultaneous differential thermal and thermogravirnetric analysis of samples spread in a thin layer on an open pan, the temperature of which can be closely measured. A controlled atmosphere furnace is used to esfublish the sample environment and details of the sample holder design are described. A close correlation of enthalpy and weight changes has been found in a variety of inorganic and organic systems in different atmospheres. The abbreviation DATA is suggested for this technique: differential and thermogravimetric analysis.
T
HE TECHNIQUES of
differential thermal analysis (DTA) and thermogravimetric analysis (TGA) are proving to have rapidly increasing application, both individually and in combination with each other and with other modern analytical methods such as gas chromatography. Several methods for combining DTA and TGA have been described in the literature (6, 8, 9) but nearly all suffer from an inherent weakness in the sample holder geometry in that, traditionally, samples have been packed in cylinders or placed as com1840
ANALYTKAL CHEMISTRY
paratively thick layers in open crucibles. Under both these geometrical conditions, there is considerable impedance to the changing ambient atmosphere within the sample, hence the sample environment cannot be known with any degree of certainty. Lack of an appreciation of the sample environment has been one of the niajor factors complicating the correlation of enthalpy change with weight change. One obvious way to improve this correlation is to perform both techniques simultaneously, so that all the data are obtained from the same sample heated a t one rate under one condition of sample geometry, packing, and ambient atmosphere. Perhaps the most suitable sample geometry for establishing rapid exchange with the environment is a thin layer spread on a flat pan. This is also a very flexible arrangement in that no special packing techniques are required and sample fusion need not be a problem. In the present work this geometry has been adapted to permit study of thermal decompositions simultaneously by DTA and TGA. A further problem in existing a p paratus for thermogravimetry is that sample temperatures are usually estimated by a thermocouple located at some point beside, beneath, or above the sample holder. Rarely does this produce an accurate record of the sample temperature, particularly during transitions. For instance, in Figure 1 is shown the temperature of a flat-pan sample holder with sample, programmed a t 1’ C. per minute, showing the deviations in actual sample heating rate during two thermal transitions. The furnace, meanwhile, has continued to increase a t :t constant rstr, indicated by the broken
line. This deviation depends upon the masses of the sample and sample holder, the enthalpy change of the process, and the relative rates of heating and transition, among other factors. Since the actual heating rate changes during a thermal transition, it is desirable to report DTA and TGA results in terms of the wtual sample temperature rather than in terms of furnace temperature. Otherwise, depending upon the time interval associated with the transition and the heating rate of the furnace, the furnace temperature prevailing a t a particular instant during the transition may bear little relation to the actual sample temperature a t that instant. EXPERIMENTAL
The deviations from linearity constitute endotherms, and it remained only to devise a means for comparing sample holder temperature with an equal mass and shape of reference material under as nearly identical conditions as possible. The manual apparatus adopted for this purpose is shown in Figure 2. The sample holder is the lower of two No. 304 polished stainless steel pans with raised edges, weighing 10.300 f 0.002 grams and machined to identical dimensions, each pan having two square-ended wells 0.036 inch in diameter and 0,050inch deep. The pans are separated 11/2 inches on a 4-hole refractory thermocouple insulating tube, and are held in place by small spring clips seated in shallow grooves scored in the circumference of the tube. The holes through the pans are machined to give as close to a sliding fit as the refractory dimensions permit. The top end of the refractory tube is held by a pair of O-rings pressed into a well in a Lucite terminal block