give a value of E* of 68 kcal. per mole. Jelliiek (8) shows that E* depends on the molecular weight. The molecular weight used in this study would have an E* of about 60 kcal. per mole based on that correlation. The Oakes and Richards (11) range of 60 to 70 kcal. per mole also agrees well with all these estimates.
APPENDIX
3
t+.,
I
-
--
LITERATURE CITED
CONCLUSIONS
The information presented here and the experience of our laboratory indicate that a worthwhile preliminary analysis of TGA curves is to plot the double logarithm of the reciprocal of the weight fraction of the reactant us. the temperature. If this plot is a straight line, then an activation energy and a pre-exponential factor may be calculated. These may serve as a convenient means of comparing the relative thermal stabilities of a series of substances. Perhaps clues as to differences in the causes of the instabilities may be obtained. I n general, because of the assumptions made in the derivation, this method is best applied where the temperature range is narrow (say 10% of the absolute temperature), where the products of the
that the above conditions are often met. In general this method can be considered as a complement to the more detailed slope plotting methods, particularly for quick analyses, and especially for pyrolyses with high activation energies (steep slopes).
- 2l t
W -
wo
Figure A. Plot of In In W,/W vs. W/ W, where W/W,variesfrom 0 to 1
reaction are gases which escape immediately without intermediate reactions with the starting materials, and where the nature of the reaction (order, energy of activation, or frequency factor) does not change over the temperature range studied. That the function fits many actual traces is evidence
(1) Anderson, D. A,, Freeman, E. S., J . Polymer Sci. 54,253 (1961). (2) Anderson, H. C., SPE Trans. 1, 202-6 (1962). (3) . , Anderson. H. D.. Nature 191. 1088 (1961). ' (4) Doyle, C. D., J. A p p l . Polymer Sci. 5,285 (1961). (5) Doyle, C. D., ANAL. CHEM.33, 77 (1961). (6) Foley, R. T., Guare, C. J., J. Electrochem. Soe. 106,936 (1959). (7) Freeman, E. S., Carroll, Benjamin, J. Phys. Chem. 62,394 (1958). ( 8 ) Jellinek, H. H. G., J . Polymer Sci. 3,850 (1948); 4 , 1 , 1 3 (1949). (9) Kofstad, P., Nature 179, 1362 (1957). (10) Madorsky, S. L., J. Polymer Sci. 9, 133 (1952). (11) Oakes, W. G., Richards, R. B., J. Chem. SOC.1949,2931. (12) Van Krevelen, D. W., Van Heerden, C., Huntjens, F. J., Fuel 30, 253 (1951).
RECEIVEDfor review October 25, 1962. Accepted June 28, 1963.
Controlled -Pressure, Controlled -Atmosphere Thermogravimetry with a Conventional Null-Type Recording Balance B. G. HURD Field Research laboratory, Socony Mobil Oil Co., Inc., Dallas, Texas
b Thermogravimetric measurements in controlled gaseous atmospheres and at controlled pressures have been made with a conventional null-type thermobalance b y using a small-diameter quartz tube for a sample holder. Pressure or composition of the gaseous atmosphere in the tube i s controlled externally via a length of flexible rubber tubing connected to the lower end of the sample tube extending outside the furnace. The upper end of the tube may be sealed to maintain a constant, regulated pressure within, or left open to permit continuous passage of a dynamic carrier gas through a powdered sample to control the gas composition. Curves showing the effect of total pressure of carbon dioxide on the thermal decomposition of dolomite and rhodochrosite illustrate the utility of t k static pressure control technique. Measurements at total absolute pressures in the range 1468 *
ANALYTICAL CHEMISTRY
of 5.8 to 417 cm. of mercury have been made. Curves showing the thermal decomposition of dolomite in carbon dioxide, nitrogen, oxygen, and hydrogen illustrate the use of a dynamic carrier gas to control the composition of gases in the immediate vicinity of the reacting sample. The advantages and limitations of the techniques are discussed.
D
in thermogravimetric data reported in the technical literature by different workers have generally been ascribed to inadequate control over experimental variables such as heating rate, thermometry, and gas composition and pressure in the sample vicinity. Garn and Kessler (1, 2 ) have emphasized the need for controlling the composition, pressure, and rate of exchange of gases in the immediate vicinity of the sample, and have pointed out ISCREPANCIES
the inadequacy of the conventional, impervious sample crucible for making precise thermogravimetric measurements. They have suggested use of a special sample holder to restrict diffusion of gases to and away from the sample to provide decomposition in a selfgenerated atmosphere (1) or, alternatively, use of a shallow pan sample container, which allows rapid diffusion of gases to and away from the sample, when decomposition in an atmosphere other than that of the decomposition product gases is desired ( 2 ) . The degree of control achieved by these special containers is, of course, limited; they were recommended for studies where more elaborate vacuum and controlledpressure thermobalances were unavailable. Significantly, however, the suggestions by Garn and Kessler directed attention to the possibility of controlling pressure and composition of gases in the immediate vicinity of the sample with-
MAUER N U L L - T Y P E RECORDING B A - L A N C E
PLATINUM B A I L
GLASS IiOOKS
-_
PLATINUM W I R E
O U A R T Z WOOL PLUGS
1 -I
TO PRESSURE ATMOSPHERE
SAMPLE
TUBULAR FURNACE
A
SWEEP $ A S
1
S A N P L I I H O L D E R FOR DYNPMIC-GAS TGA
I
_
~
-
RING STAND AND CLAMP
Figure 1. Sample holders and apparatus for controlled-pressure, controlled-atmosphere thermogravimetry
out controlling the entire gaseous environment of the furnace. The purpose of this paper is to describe simple experimental techniques for accomplishing this type of gaseous atmosphere and pressure control in thermogravimetry. The essential feature of the proposed techniques is the use of a small-diameter quartz tube as a sample container. The tube is suspended from a null-type recording balance, so that the lower end extends outside the furnance. This end is connected to an external pressureatmosphere control system via a length of flexible rubber tubing. Except for a small sacrifice in absolute accuracy and sensitivity, the opera1ion of the balance is not impaired by this attachment. The upper end of the tube, which contains the sample, may be sealed to permit precise regulation of the internal pressure; alternative y, it may be left open to allow continuous passage of a dynamic carrier gas through a powdered sample. The dynamic gas is used when it is desired to control the composition of the gases in contact with the reacting sample. In general, precise control over both pressure a i d composition is not achieved simultaneously; if the tube is sealed to control pressure, a gaseous atmosphere composed essentially of the decomposition product gases and the gas-regulator supply gas must be accepted. On the other hand, if the dynamic carrier gas is utilized, the control over the internal prcasure is severely limited. The quartz tube affords a much better control over pressure and gas composition than the holder,., previously sug-
gested. However, because of the attendant reduction in accuracy and sensitivity, in addition to several other limitations discussed below, the techniques are not recommended for routine data gathering. They are suggested, rather, as a simple, inexpensive method of obtaining controlled-pressure, controlled-atmosphere thermogravimetric data for special limited studies. APPARATUS AND PROCEDURE
A schematic diagram of the experimental arrangement is shown in Figure 1. A powdered sample is supported in the upper end of the tube by a loosely packed quartz wool plug. For controlled-pressure experiments, the upper eud of the tube is sealed before the sample is weighed into the tube. When the dynamic sweep gas is to be used, the upper end of the tube is left open and a second quartz wool plug is placed above the sample, as shown in the inset of Figure 1. The tubes employed were 3 mm. in i.d. with 1-mm. walls; the length is selected so that the lower end extends about 6 to 10 cm. outside the furnace when the sample is positioned in the centrally heated portion, about 1/4 inch away from the furnace temperature control thermocouple. The sample tube is connected to the pressure-atmosphere control system by a length of flexible rubber tubing which is U-looped and connected to a short length of glass tubing mounted in a rigid support. The use of a ring stand and clamp for this support facilitates positioning of the tube in the center of the 1-cm. hole in the lower end plug (not shown) of the furnace. Pure gum microsurgical tubing, 1.5-mm. i.d. with 0.75-mm. walls, has proved most satis-
factory for this connection. It will withstand pressures up to 30 p.s.i.g., which permits controlled pressures within the tube up to about 3 atm. For higher pressures (up to nearly 6 atm.), heavy-walled microbore tubing, 1/16-inch i d . with :/8-inch walls, has been u s d ; it is lcss flexible than the surgical tubing, and causes more “noise” and zero drift in the weight curve, but still yields useful, although less accurate, weight data. Depending on the type of tubing emp!oyed, the entire assembly contributes about 12 or 33 grams to the load on the balance. The recording balance used in these experinients wa5 built according to the design by lCIauer (4). I t has a capacity of 200 gram3 and an accuracy to about 0.2 mg. Any null-type thermobalance with moderate capacity could be used; the sample tube and U-looped tubing exert the same general type of influence on the balance as does the chain on the standard Chain-0-matic balance. If the tube and tubing remain in a fixed position, as they essentially do with the null-type balance, they apply a constant load to the beam. Changes in weight of the sample are recorded as usual if the reaction product gases can escape without condensing anywhere in the system. Escape of product gases is no problem with the dynamic gas technique; they are swept continuously out of the sample tube by the sweep gas. However, when the controlled-pressure technique is used! product vapors must traverse the entire length of tube and tubing to be exhausted a t some point in the pressure-control system, normally a t a gas-regulator bleed. If water vapor or other gases condensable a t ambient temperatures are evolved from the sample, they mill normally condmse in the lower, cool portion of the sample tube. The weight loss in this case is incompletely recorded, or is delayed by the amount of time required to evaporate the condensate under the conditions of the experiment. Thus, the static control pressure technique cannot be used wlim the sample will evolve a condensable gas. A degree of pressure control for such samples can be achieved by using the dynamic gas technique and blending gases in suitable proportions to control the partial pressure of the desired gaseous constituent. The total pressure above atmospheric could be regulated within limits by iise of a constriction or a low permeability packing above the sample. Any convenient gas-composition, pressure-control system may be connected to the glass tubing held by the rigid support. For the controlled-pressure carbonate decomposition studies reported herein, the sample tube was evacuated twice with a vacuum pump and subsequently fi!led with dry COz. Thereafter, a regulated COz preFsure was sct with commercial gas pressure regulators. VOL. 35, NO. 10, SEPTEMBER 1963
* 1469
bW(
)"-'
-
LIGHT TUBING
TUBING
u
6%
200
600 860 TEMPERATURE.*C.
403
Id03
'
I&
Figure 2. Sample curves showing zero drift with light and heavy flexible tubing
For controlled pressures less than atmospheric, a subatmospheric pressure regulator backed by a floor-vacuum pump was used. A manometer and Bourdon gauge teed into the line monitored the regulated pressure. Carbon dioxide produced by the carbonate decompositions escaped through the gas regulator bleed. For the dynamic gas studies, a regulated gas pressure was set to provide a flow of the desired sweep gas of approximately 5 cc. per minute as measured by a, series rotameter in the line, When hydrogen was used as the dynamic gas, the furnace was flushed continuously with nitrogen to prevent accumulation of an explosive mixture of gases within the furnace. The samples for the dynamic gas studies were particle-sized to pass 70mesh and be retained on a 100-mesh sieve. Screening out of the fine particles prevents drastic changes in gas permeability of the sample- with consequent sudden increases in flow rate of the dynamic gas-upon decomposition. For samples of smaller particle size, a packing of finely ground, thermally inert material, such as aluminum oxide, placed above the upper quartz wool plug prevents these drastic changes in gas permeability of the entire sample and packing assembly when the sample decomposes. Zero drift due to thermal convection currents and decreased air buoyancy is appreciably greater with the quartz tube sample holder than with the usual small sample crucible. This is to be expected because of the greater surface
area and volume of the quartz tube, and because both ends of the furnace must have an air vent to accommodate suspension wire and sample tube. The upward flow of air through the furnace tube was minimized by making the effective size of the hole in the upper end plug, through which only the platinum suspension wire extended, as small as practical-about 2 mm. in diameter. The zero drift for this arrangement is almost linearly related to temperature, and compensating corrections are easily made. The type of zero drift is illustrated in Figure 2 by two curves which represent extreme rather than typical cases. The greater drift experienced when the heavy tubing is used is attributed in part to the greater surface area of tubing past which air flows upward into the furnace. Perhaps also responsible is the slight tilt of the sample tube which sometimes results when the less flexible tubing is used. Tilting of the tube presents a greater surface area t o the direct effects of air and thermal convection currents. Vertical alignment and horizontal positioning of the sample tube in the furnace have been found to be an extremely tedious operation. Corrections for zero drift can be made by extension of the drift curve past the range over which the weight loss occurs. This extension can be made by extrapolation, assuming linear drift, as shown in Figure 2, or by determining the total drift over the entire temperature range as the difference between the apparent and the actual weight losses. $ctual weight losses may be determined by weighing the tube and its contents after the experiment. The true weight loss a t any temperature is taken as the weight interval between the recorded curve and the corresponding point on the drift curve. In addition to the drift described, the zero position of the curve is also displaced by a change in the regulated pressure within the sample tube and
tubing. The U-looped tubing behaves like a Bourdon tube, and a change in pressure causes a change in torque whose vertical component is applied to the balance. The magnitude of the shift depends upon the type and length of the tubing used; for a 1-meter length of surgical tubing, the shift amounted to about 0.2 mg. per p.s.i. change in pressure. The shift can be minimized, but not eliminated, by making the U-loop out of small glass tubing and connecting it into the system with short lengths of the flexible rubber tubing. This zero shift does not prevent changing of the regulated pressure during the course of an experiment, but does require that a nrw zero position he established in this event. In addition to zero drift, another problem accentuated by use of the quartz tube sample holder is that of "noise" in the weight curve caused by erratic air currents. Whereas the curve rarely departs from its true position by more than 1 0 . 2 mg. with the normal arrangement-Le., platinum crucible holder and one small air vent in the furnace-sudden excursions up to It 1 mg. away from the mean locus sometimes occur with the quartz tube and tubing attached. Normally, however, the curves keep within a well-defined envelope that is about 0.4 mg. wide initially-i.e., & 0.2 mg. about a mean locus-which broadens to about 2 mg. at 1200' C. The noise is minimized by keeping the holes in the furnace as small as possible and by shielding exposed tube and tubing from stray laboratory air currents. RESULTS AND DISCUSSION
Thermal decomposition curves at regulated C 0 2pressures of the carbonate minerals, rhodochrosite (NnCOa) and dolomite [CaMg(CO&l, are shown in Figures 3 and 4. The curves are corrected for zero drift as described earlier.
e
9
10-
v)
U
0
5
20-
W
v;
30-
HEATING RATE
- IO*C./mln.
4
MOO
c 5.0 w
* ---/-----!.-.
L---J.-350
400
3
Figure 3.
1470
450
500 350 600 T E M P E R A T U R E , 'C.
650
7M)
Thermal decomposition of rhodochrosite
ANALYTICAL CHEMISTRY
750
Figure 4.
Thermal decomposition of dolomite
Procedural decomposi1;ion temper atures, taken as the temperature of departure from the base line, are indicated by the arrows. Because the zero drift is not exactly linear, the departure temperatures are most accurately determined by superposition of two curves obtained at different pressures on ilight table. The point a t which the curves separate can generally be precisely determined; however, for reactions which proceed very slowly above the equilibrium decomposition temperature, there may be a 5' to 15' uncertainty in selecting the procedural temperature. The curves indicate a good control over the pressure of COSin contact with the reacting samples. From Figure 1 it may be seen that the total pressure a t the sample will be greater than the regulated pressure by the amount of pressure drop due to flow of gas from the decomposing sample through the tube and tubing system to the exit bleed. This pressure drop will depend upon the system used, primarily upon the length and inside diameter of tube and tubing, and upon the rate a t which gas is evolved from the sample. For the decomposition reactions reported here, it has been estimated from Poiseuille's law and t i e maximum rate of reaction that the flowing pressure drop through tube and tubing will not exceed 0.15 cm. of mercury. The pressure drop across the quartz wool plug is variable, in that the packing cannot be precisely reproduced, but in most practical cases it will not exceed 1 mm. These pressure drops, are, of course, insignificant in the pressure ranges investigated here. A further indication that the pressure control is adequate for many investigations of interest is the Clausius-Clapeyron plot of log p us. 1/ T for the calcium carbonate decomposition, shown in Figure 5. The data plotted represent
30 U
w
vi
40
0
c 50 c? 1
100
200
?\-
L
0
the second stage of decomposition of dolomite. The linearity and close approach of the curve to the equilibrium dissociation data (dashed curve) reported by Hill and Winter (3) indicate that the temperature lag between onset of this reaction and the initial observation of it is small. A similar plot for the MnC03 decomposition (not shown) was also linear, but greatly displaced toward higher temperatures from accepted equilibrium dissociation data. Data illustrating the use of the dynamic gas sweep to control the composition of the gaseous atmosphere in contact with a reacting sample are shown in Figure 6. The curves represent the thermal decomposition of dolomite in the different gases, each at approximately 1-atm. pressure. Using the sweep gas, the pressure a t the sample is less than the upstream regulated pressure by the amount of the pressure drop due to the flow of the sweep gas through the tube and tubing. This flowing pressure drop is always insignificant, since it will be on the order of 1 or 2 mm., whereas the total regulated pressure ill be slightly greater than atmospheric.
w E
-
I
800
900
1000
One advantage of using the sweep gas rather than controlling the entire furnace environment is that it eliminates the total dependence on diffusion as a method of exchange of gases a t the exterior surface of a reacting sample. Where the particle size of the sample is appreciable, as in these experiments, diffusion is probably still active as B mechanism of gaseous exchange as the reaction proceeds from the exterior to the interior of the individual particles. Even so, if the concentration of the product gases at the exterior surfaces is kept very low by the passage of the sweep gases, the diffusion process is speeded. Detailed interpretation of the controlled-pressure, controlled-atmosphere thermogravimetric data reported herein is considered to be beyond the scope of the present paper. These results are reported merely to illustrate the type of data which may be obtained with a conventional null-type recording balance using the suggested procedures. Because of the rather severe limitations already discussed, the techniques are not recommended for extensive data gathering. For limited investigations, however, these procedures can be used to obtain very worthwhile controlledpressure, controlled-atmosphere thermogravimetric data.
100-
-
PROCEDURAL CURVE
_
LITERATURE CITED
N
0
700
Credit is due H. G. Bufford and 0. T. Ward for collecting the experimental data reported herein. The paper is published by permission of the Socony Mobil Oil Co., Inc.
n I
600
ACKNOWLEDGMENT EQUILIBRIUM C U R V E
W
V
500
Figure 6. Thermal decomposition of dolomite in different gaseous atmospheres
$
5I
400
TEMPERATURE, 'C.
'Oo0l
I"
300
-
(1) Garn, P. D., Keader, J. E., ANAL. CHEM.32, 1563-5 (1960).
1
T'C 1000
950
900
77
78
79
80
81
02
E3
84
85
8 50
-4-Ti-,
--
-rL-T--T-+-
86
U7
88
89
90
91
IOOOlT'K
Figure !j. Thermal decomposition curves for calcium carbonate
(2)Ibid. p. 1900-1. (3) Hill,'l!. J., Winter, E.R. S.,J. Phys. Chcn. 60, 1361-2 (1956). (4) Mauer, F. L., Rev. Sci. Inatr. 25, 59% 602 (1954). RECEIVED for review November 20, 1962. Accepted May 17, 1963. Southweet Regional Meeting, ACS, Dallas, Texss, December 1962. VOL. 35,
NO. 10 SEPTEMBER 1963
1471
