FEW SPECIMENS OF CLAY ARE FOUND in amateur mineral collections, chiefly because they are so difficult to identify. None of the field guides compiled for amateurs even list such interesting clays as anauxite, allophane or montmorillonite The two most comprehensive guides carry short descriptions of kaolinite, a basic material of the ceramics industry, but discourage its collection by pointing out that clay minerals can be identified only by an intricate combination of X-ray, electron-microscope and chemical tcchniques that are beyond the capabilities of most amateurs. This was certainly true prior to the last decade. But in recent years the much simpler technique of differential thermal analysis has been successfully applied to the classification of clays, as well as of shales and some uranium ores, by Ralph J. Holmes, Paul F. Kerr and J. Laurence Kulp of Columbia University.

The method is based on the fact that substances absorb heat at characteristic rates. For example, when a bar of iron is heated, within certain limits it gets hotter in direct proportion to the amount of time it spends in the fire. In contrast, when a container of ice is heated, it stays at 32 degrees Fahrenheit until the ice melts. The temperature of the water then climbs steadily to 212 degrees, where it again holds constant until the water boils away. This difference in the heating rate of iron and water constitutes a means, although a trivial one, for distinguishing between the two. In the case of clay minerals, shales and some uranium ores, the temperature fluctuations are more numerous, subtle and definitive.

Part of the apparatus used for making differential thermal analyses at Columbia was designed, constructed and tested experimentally by David Smith, a former student of mineralogy who is now a technical specialist in public relations in New York City. "In essence," he writes "the method oú identifying clay minerals by differential thermal analysis consists in placing the unknown specimen in a furnace along with a relatively inert substance like alumina, and gradually increasing the furnace temperature to about 1,000 degrees centigrade while recording the temperatures of the specimen and the inert substance. A graph made by plotting the temperature fluctuations of the specimen against the constantly rising temperature of the inert substance is uniquely characteristic of the specimen and is roughly analogous to a spectrogram.

"The furnace we used at Columbia was heated by resistance wire of the kind used in electric toasters. With the exception of the furnace liner, a tube of alundum, the furnace can be improvised out of commonly available materials. It should be possible to build an adequate one from a five-gallon paint bucket by replacing the ends with a heat-insulating composition board such as Transite. A tube of alundum some two inches in diameter and as long as the paint bucket can be wrapped with a helical coil of Nichrome wire held in place by a coat of alumina cement (powdered alumina mixed with water glass). This assembly can be supported by holes cut in the centers of the Transite ends to make a snug fit with the tube. The space between the tube and the wall of the bucket should be packed with heat-insulating material such as calcined diatomite or bubble alumina. Leads for connecting the heating coil to the power line can be brought out of the furnace through porcelain tubes of the kind used for house wiring. As a convenience, the furnace should be mounted on a metal frame so it can be raised and lowered while hot [see illustration above left].

"Our specimen holder was machined from a two-inch length of cylindrical stainless steel three inches in diameter. Twelve 34-inch holes for the specimens were drilled into one face of the block to a depth of one inch. Each hole was then drilled the rest of the way through the block with a 1/8-inch drill as shown at the upper right in the drawing on the left. An additional 1/8-inch hole was drilled through the center of the block and a second one, about inch off center, was drilled into it to a depth of one inch from the opposite face, as shown. The small holes admitted thermocouples for measuring the temperature of the specimens, furnace and specimen holder respectively. The specimen holder was supported in the center of the furnace by a hollow cylinder of alundum.

"Thermocouples are easy to make. Twist the ends of two wires of dissimilar metal around each other two or three times. Snip off the excess and clip the free ends to one side of a 110-volt source. Then dip the twisted ends momentarily into a metal bowl of mercury connected to the other side of the circuit. The resulting arc will melt the wire ends into a bead that constitutes a butt weld. The diameter of the bead should be about twice the diameter of the wire. Any two dissimilar metals will exhibit thermocouple action; the most satisfactory are those that develop a relatively high voltage that increases in direct proportion to differences of temperature between the welded junction and the free ends.

"The analysis of clay requires thermocouples that can withstand repeated exposure in air to temperatures on the order of 1,000 degrees. Ours were made of two special alloys: Chromel-P and Alumel. Chromel-P is 90 per cent nickel and 10 per cent chromium. Alumel is 94 per cent nickel, 2 per cent aluminum,

per cent magnesium and 1 per cent silicon. The alloys are produced by the Hoskins Manufacturing Company in Detroit and are available in the form of No. 22 gauge wire from the Edmund Scientific Corporation in Barrington, N.J., in minimum quantity for $1. Chromel-Alumel thermocouples develop a potential difference of approximately one millivolt for each 28.2 degrees C. of temperature difference between the junction and free ends, as shown by the top graph on the right. Electrons flow from Chromel to Alumel inside the junction.

"Thermocouples can be connected in pairs to indicate the difference in temperature between two objects by splicing one set of similar leads together, such as the Alumel pair, as shown at the lower right in the illustration in Figure 2. When the junctions of two thermocouples so connected are at the same temperature, the electrical outputs cancel and no potential difference appears across the Chromel leads. When the junctions are at different temperatures, the thermocouple action of the hotter unit exceeds that of the cooler unit and a voltage appears across the Chromel leads that varies in direct proportion to the temperature difference. Adjacent holes of our specimen holder were equipped with pairs of thermocouples connected for measuring differential temperature. A single thermocouple was installed in the holder for measuring the temperature of the stainless steel block and a second one was extended above the holder for sensing furnace temperature. Care was taken to center the units precisely in the holes so that the junction would not heat unevenly and generate false temperature indications. The Chromel and Alumel leads were insulated by small tubes of alundum and were fastened in the specimen holder by alumina cement.

"The output voltage can be measured either by a voltmeter sensitive to a billionth of a watt and calibrated to 50 millivolts or by pen recorders equipped with appropriate amplifiers, depending on the experimenter's budget. We used a battery of pen recorders and synchronized the speed of the charts with the rate at which the furnace heated by means of an electronic program controller. The controller operated automatically and could be adjusted for any desired rate of heating up to 50 degrees per minute or for holding the furnace at any fixed temperature. A heating rate of 12 degrees per minute resulted in the most clearly defined peaks. Ample information on all specimens appeared in the temperature range from 100 to 1,050 degrees C. Higher temperatures added little of interest to the graphs and reduced the life of the coil and the thermocouples. "The operating procedure is simple. Specimens are prepared by dry grinding with a mortar and pestle. The specimen holders are filled with ground material that passes through a sieve consisting of 50 crossed wires per inch-a standard 50-mesh screen. The powder is poured into the holes loosely; tamping appears to have little effect on the results, and tightly packed specimens are difficult t o remove at the conclusion of a run. The companion hole associated with each specimen is charged with inert comparison material. We used powdered alumina that passed a 60-mesh sieve. Once the: samples are inserted and the thermocouples are connected, the furnace can be started.

"We adopted a standard procedure for plotting and interpreting the curves. Temperature fluctuations are plotted upward when the specimen emits heat and gets hotter than the comparison material, as a consequence of exothermic (heat-emitting) reactions, and are plotted downward when it absorbs more heat than the comparison material, during endothermic (heat-absorbing) reactions.

"Reactions of one type or the other may occur for a variety of reasons, including loss of absorbed water and water of crystallization, release of the hydroxyl radical (OH) from the crystal lattice, transitions in the structure of the crystal lattice, the loss of carbon dioxide from carbonate compounds, the devolatilization of coals and oil shales, and the oxidation of sulfide compounds and uranium-bearing shales. The temperature of the furnace is plotted horizontally on the graph; fluctuations of specimen temperature are plotted vertically. The area enclosed by fluctuations or peaks in the graphs varies in proportion to the amount of thermally active mineral involved in the reaction. Mixed specimens can be analyzed by measuring the area enclosed by a selected peak and dividing it into the area of an equivalent peak of a pure specimen that has been measured with the same apparatus. The ratio represents the amount of the known mineral that is present in the mixed specimen.

"At least five basic types of temperature fluctuation occur, each of which produces a typical curve [see second graph from top at right]. Broad, low endothermic variations between 100 and 300 degrees C. (1) indicate loss of absorbed water. Broad exothermic curves (2) signify the oxidation of organic matter in clays between 200 and 600 degrees. The loss of lattice water between 450 and 700 degrees yields curves (3) with a broad base. Carbonate reactions show large asymmetrical deflections (4) between 450 and 900 degrees. The comparable narrow, high-amplitude exothermic peaks that correspond to transformations in the structure of the aluminum-based part of the molecule occur near 1,000 degrees (5).

"A number of the gangue minerals (waste materials in ore deposits) also exhibit interesting reactions [see third graph from top at right]. The crystalline structure of quartz, for example, undergoes an abrupt endothermic transformation at 575 degrees C. that appears as a dip in the differential curve so sharp and reproducible that it can be used for calibrating thermocouples. (Fairly pure quartz sand for calibrating thermocouples can be taken from the little sandglasses sold as egg timers by novelty stores. Colored sand may also contain quartz, but the impurities often introduce reactions that mask the endothermic peak at 575 degrees. Calcite, also a gangue, shows an endothermic peak between 800 and 900 degrees when carbon dioxide is evolved. Two peaks characterize dolomite, one indicating the release of carbon dioxide from the magnesium ion in the vicinity of 750 degrees and another from the calcium ion at 860 degrees.

"The clay minerals all undergo a typical exothermic reaction at 980 degrees, the point at which amorphous alumina adjusts to what is known as its alpha-alumina form. In addition, each member of the clay family is uniquely characterized by one or more endothermic reactions and at least one member by an exothermic reaction. Adjacent curves of the third graph from the top of Figure 5 show the characteristics of three kaolin minerals: nacrite, dickite and kaolinite. The graph for anauxite resembles kaolinite except for a smaller area under the endothermic peak that appears during the decomposition of two crystal types into amorphous silica and alumina. Halloysite can be distinguished from other kaolin minerals by two extra low-temperature peaks, one at 150 degrees and a substantially smaller one at 325 degrees. Allophane is so fine-grained that it can almost be considered amorphous. One reaction that has a peak at 180 degrees extends across nearly 100 degrees, and another, at 510 degrees, spans 200 degrees. Neither is clearly defined. Montmorillonite minerals are characterized by two endothermic peaks.

"The graph of talc, which is not included in the illustration, shows a pronounced endothermic peak at 990 degrees, indicative of the strength of the hydroxyl bonds in the talc crystal lattice.

"The temperature of some shales fluctuates almost continuously through a span of 1,000 degrees. Two of them, arbitrarily designated A and B, are graphed at the bottom of Figure 6. Most of the carbonates in shale A were in the form of dolomite. Those in shale B were mostly limestone. This composition is indicated by the double endothermic reactions between 700 and 900 degrees in the case of shale A and the one large peak between 800 and 900 degrees in B. The exothermic reaction between 400 and 500 degrees in both shales arises from the combustion of carbonaceous material and the oxidation of marcasite. It is tempting to conclude that the exo-endothermic peaks between 100 and 150 degrees are a consequence of the simple drying of the shale. But dehydration of mechanically held water never occurs exothermically in clays, and the reaction actually signifies the transformation of tridymite quartz. The reactions are typical of illite shales that contain less than 50 per cent illite.

"Characteristic differential thermal reactions of these mineral classes are listed in recently published reference texts [see "Bibliography," below]. Still others await analysis, and amateurs can help to fill in the gaps. Problems have a way of arising, however, when the technique is applied to unfamiliar materials. Oxides, sulfides and arsenides react with thermocouples, for example, and can ruin them during a single run. Organic impurities tend to mask other reactions when they oxidize. This is particularly true in the case of uranium-bearing shales. Sometimes thermocouples can be protected by a ceramic covering. If the metal specimen holder is attacked, a ceramic cup can be substituted to hold the sample. It is possible to prevent the oxidation of organic materials by running specimens in an atmosphere of an inert gas, such as helium, a procedure that requires the furnace to be enclosed in an airtight bomb.

"At the conclusion of a run the powdered specimens should immediately be blown from the specimen holder by a blast of compressed air. Some specimens tend to cake if permitted to cool in the containers; they must then be dug out and the thermocouples are ruined."

This department shares Smith's enthusiasm for differential thermal analysis. It is a fascinating experiment, both to set up and to run. Not all of us have access to electronic programmers and expensive pen recording systems, however, and even such prosaic materials as three-inch cylinders of stainless steel are not easy to pick up. Nonetheless, with a few tips from Smith I have just built a primitive model of his apparatus. It cost only $20 and it worked very nicely. The principal outlay went into a piece of two-inch alundum tubing for the furnace liner (obtained from the Fisher Scientific Company, 638 Greenwich Street, New York 14, N.Y.) and thermocouple wire.

All Smith's suggestions for constructing the furnace were adopted except one: the paint bucket was not equipped with Transite ends. This was a mistake. Heat buckles the tin and pulls it away from the muffle, and the ends are now being replaced by Transite. The alundum tubing was wound with Nichrome wire removed from a pair of 660-watt radiant-heater units bought at a local electrical-supply store. Each element has a 1/8-inch helical spiral winding. When the spiral is removed from the element, the wire can be straightened by slipping one end of the helix over the point of a shingle nail (held in a vise) and pulling the free end away from the nail. The straightened wires from two radiant elements are twisted to form a pair, then wound on the alundum tube. The ends of the wound coil are clamped to the alundum tube by narrow metal bands as shown in the illustration in Figure 7.

I did butt-weld one thermocouple with a mercury arc as Smith suggests. In making such welds a ballast resistor must be connected in series with one lead to limit the current; I used a 660-watt radiant-heater unit, as shown in the drawing below. But when a mercury arc is struck, a small, yellowish cloud of highly toxic mercury vapor rises, a whiff of which could land the experimenter in the hospital. It is better to make the welds with a carbon arc, a more difficult procedure but one that can be mastered with practice. Remove the carbon rod from a size D flashlight cell and wash it thoroughly in strong soap and water. Since an arc will not be sustained between the wires and solid carbon at five amperes (the current passed by the ballast resistor), first drill an axial hole slightly smaller than the diameter of the rod into one end to a depth of about 34 inch. (The rod is so soft that you can make the hole by twisting the bit with your fingers.) Save the drillings and tamp them into the hole. Then strike the arc between the thermocouple wires and the solidly packed carbon dust. The small cloud of carbon dust kicked up by the arc seems to conduct just as well as mercury vapor. The job is made still easier, and better welds result, when a flux of borax is applied to the wires before striking the arc. Just moisten the twisted ends and dip them into powdered borax of the 20 Mule Team variety that is found in most kitchens. Try to control the arc so that the twisted ends melt into a bead about the size of a pinhead. Make a few practice welds with Nichrome wire from a heater unit until you acquire the knack.

The potential difference across Chromel-Alumel thermocouples varies from approximately zero volts at room temperature to 45 millivolts at 1,050 degrees C., as shown by the graph in Figure 4. The short-circuit current of the thermocouple reaches a maximum of about 30 microamperes at 1,050 degrees. The maximum output power of the device at this temperature is on the order of one microwatt. To hold costs down I measured the output voltage directly and did not attempt to amplify the current. The measurements were made with the potentiometer circuit shown above. The potentiometer was of the large, wire-wound precision type currently available on the surplus market for $1. The fixed resistor is of the carbon type and rated at 1/2 watt. The circuit is energized by a single flashlight cell. The voltage of the cell should be determined while under load by an accurate voltmeter. The dial of my potentiometer turns through 270 degrees and is calibrated in 20-minute intervals of arc. A tenth of the cell voltage (.152 volt) appears across the potentiometer. So each scale division equals .00018 volt.

The thermocouples are connected through a double-pole, double-throw switch to the arm of the potentiometer and the positive terminal of the battery so that the thermal current opposes that of the battery. A 0-20 microammeter of the flush-mounted type is included in the common circuit to indicate the direction of current flow. It was converted to function as a sensitive galvanometer by cementing to the pointer a small mirror that reflected a beam of light to a screen 10 yards away. To make the conversion, remove the movement from the case and carefully bend the pointer back over the pivot bearing in the form of a hairpin loop. Center a 1/8-inch square of glass broken from a silvered microscope cover slip above the pivot and cement it there. The movement is then mounted for protection in a box fitted with a window of reasonably flat glass, such as is used for photographic plates. I used a 85-millimeter slide projector as a light source.

With the potentiometer set at zero, the switch is thrown to connect any cold thermocouple. The position of the beam on the screen is then marked "zero." When heat is applied to the thermocouple, the beam will drift to one side. It is returned to zero by rotating the potentiometer arm. The temperature of the thermocouple is then computed by multiplying the voltage, as indicated by the position of the potentiometer arm, by 23,200. The potentiometer scale can be calibrated directly in degrees centigrade if desired.

I could not locate stainless steel cylinders of the size specified by Smith. So I improvised a one-specimen sample holder out of two stainless steel tablespoons, hammering the bowls into small cups by heating the spoons with a torch and gradually driving the metal into the hole of a 5/8-inch nut with the rounded end of a 3/8-inch bolt. Wrinkles were prevented by alternately hand-forming the cup on an anvil and then driving it a little farther into the nut. The bottoms of the finished cups were drilled for the thermocouples and assembled with alundum cement to an inverted alundum crucible, which had previously been drilled for the thermocouples. Alundum insulating tubes could not be located, so the thermocouple leads were insulated with coatings of alumina cement. The heating rate of the furnace was controlled by a continuously variable auto transformer. A simple rheostat would work as well, although it would run up the light bill a bit faster. Experimenters who duplicate this version of the apparatus will find that cranking the transformer and potentiometer while simultaneously jotting down temperature readings will keep them hopping during the hour of the run. The results, however, will more than justify the effort.

Bibliography

DIFFERENTIAL THERMAL ANALYSIS: THEORY AND PRACTICE W. J. Smothers and Yao Chiang. Chemical Publishing Co., Inc., 1958.

MULTIPLE DIFFERENTIAL THERMAL ANALYSIS. Paul F. Kerr and J. L. Kulp in The American Mineralogist, Vol. 33, Nos. 7 and 8, pages 387-419; July-August, 1948.

The Society for Amateur Scientists (SAS) is a nonprofit research and educational organization dedicated to helping people enrich their lives by following their passion to take part in scientific adventures of all kinds.