Modified Molecular Sieve Reflux Extractor for Efficient Dust-Free

Heating of the sample is carried out with a hot plate con- nected to a regulated ... ture of the product; and the use of a boiling stick, a sintered g...
0 downloads 0 Views 250KB Size
Heating of the sample is carried out with a hot plate connected to a regulated voltage supply. The bottom of the condenser is so constructed that the cold condensate does not drip on the sheathed thermocouple. The temperature of the boiling fluid is measured rather than that of the vapor because the vapor pressure of the bulk fluid is to be correlated with the fluid temperature. The measurement of the liquid temperature is complicated by superheating phenomena. This is of much greater magnitude for a mixture than for a pure component. As a result, the temperature slowly increases until the hot liquid bumps, momentarily dropping the temperature. The cycle is then repeated. This action obscures the “true” boiling temperature of the product; and the use of a boiling stick, a sintered glass bottom on the boiling chamber, or a glass-covered magnetic stirrer is necessary to reduce this phenomenon. Another manifestation of superheating is the dependence of the boiling temperature upon the rate of heating, particularly at low pressures. If the heat input to the oil sample is increased over the minimum required to cause ebullition, the boiling temperature will also increase. No attempt was made ( 5 ) F. D. Rossini, “Selected Values of Physical Constants and

Thermodynamic Properties of Hydrocarbons and Related Compounds,” Carnegie Press, Pittsburgh, Pa., 1957.

to determine the maximum effect of rate of heating upon the boiling point, but increases of 3-5” F have been observed. A medium rate of heating which causes rapid but steady boiling is desired in all cases. Because of these factors and because of the type of products involved, results are only accurate to within a few degrees. This suffices for most purposes. RESULTS

Figure 2 shows vapor pressure data obtained with this apparatus. The octadecane line has been constructed from published data (5). The circles on this line represent the measured points obtained in this study. The bottom line has been obtained on a 2-ethylbutoxypolysiloxane-based hydraulic fluid. The center line has been constructed through the points obtained on a petroleum-derived hydraulic fluid. The two sets of data for this line were obtained with an interval of more than one year between the two series of measurements. The excellent correlations in all these cases attest to the reproducibility of this method. RECEIVED for review January 29, 1968. Accepted February 14, 1968.

Modified Molecular Sieve Reflux Extractor for Efficient Dust-Free Dehydrations David S. Rulison, Paul Arthur,’ and K. Darrell Berlin* Department of Chemistry, Oklahoma State Unisersity, Stillwater, Okla. 76074

REMOVAL OF TRACES of moisture from organic solvents and solutions can be of prime importance as, for example, in studying reaction kinetics, performing nonaqueous amperometric titrations, and in the trace analysis of complex ions in which water may compete for a coordination site. Many other ramifications of an efficient dehydration technique are evident. In a previous report ( I ) a general method of dehydration via an approach employing a Soxhlet extractor was described. However, it has been discovered that traces of dust from the molecular sieve may catalyze certain reactions of organic molecules during the dehydration process. To illustrate, methyl ethyl ketone acquires a light yellow color (presumably a result of an aldol condensation) when attempts were made to prepare an anhydrous sample. The apparatus illustrated in Figure 1 provides a continuous recycling system which circumvents the difficulty of contamination of the anhydrous solvent by products arising from a catalytic influence of dust from the molecular sieve. Checks were made on solvent purity (after dehydration) by gas-liquid chromatographic (GLC) analysis employing a hydrogen flame detector. EXPERIMENTAL

Apparatus. Figure 1 shows the modified apparatus; the left-hand portion of the diagram is similar to the apparatus already described ( I ) . In the present modification the liquid siphons and empties into flask B, instead of returning to the original flask. Thus any molecular sieve (Molecular Sieve

Deceased.

* To whom inquiries should be addressed. (1) P. Arthur, W. M. Haynes, and L. P. Varga, ANAL.CHEM., 38, 1630 (1966).

4A-Linde Co., Division of Union Carbide Corp.) dust and other impurities, do not contaminate the material in flask A . Parts C and D constitute a device for regulating the heating rate of flask B. A platinum band is mechanically held around the shaft of part D,which is a float. The position at which the platinum band is to be placed must be determined experimentally. In part C, two platinum wires (22 gauge) are attached so that their lower ends bend under the edges of the tube which fits around the shaft of part C. Part E is a one-way glass valve which allows flow only from the condenser F to flask A . The upper ends of the platinum wires of part C are connected to a relay which controls the heating of flask B. Two types of relays have been used successfully: a Supersensitive Relay No. 4-5300 available from American Instrument Co., Inc., Silver Spring, Md., and a Thermocap Industrial Control Relay available from Niagara Electron Laboratories, Andover, N. Y . Operation. The apparatus is normally filled through port G. Solvent is then distilled through condenser F into flask A until the latter is about two thirds full. Flask B is then filled to approximately the same level. The apparatus is ready for operation. If it is desired to dry a solution, the solute must be added to flask A before the drying operation is begun. Flask A is heated by means of a heating mantle until the solvent boils. Vapors from the solvent rise and condense so that solvent drops onto the molecular sieve bed. After passing through the bed, the solvent siphons into flask B when a sufficiently high level is attained in the Soxhlet extractor. This process continues until the float D is elevated to the required level to cause the platinum band to make contact with the two platinum wires of part C, thus actuating the relay and turning on the heating mantle under flask B. The vapors from the boiling solvent in flask B enter condenser F, after which the condensed liquid passes through the one-way valve into flask A which is heated continuously. VOL 40, NO. 6, MAY 1968

1015

This process proceeds until the float D is lowered to the extent the relay contact is broken, thus turning off the heating mantle under flask B. As flask B cools, a partial vacuum develops. This closes the one-way valve E and pulls more solvent over from the molecular sieve bed. This in turn actuates the heating mantle under flask B, and the process continues. It is important that the heating mantle under flask B is operated at a higher voltage than that under flask A . This differential prevents flooding of flask B and condenser F because the rate of transfer of solvent from flask B to flask A is greater than the rate of filling of flask B from the molecular sieve tower. Experience has shown that apparatus performs well with the heating mantle under flask A operated at about 70 V from a variable transformer with flask B at about 75 V.

24/40

URANIUM

*s

dJ RELAY

24/40

RESULTS AND DISCUSSION The modified apparatus has been found to dehydrate efficiently both pure solvents and solutions. Water levels have been reduced from about 0.1 to 0.001% in reagent grade isopropyl alcohol and a 0.1M solution of lithium chloride over a 30-hour period of reflux. The water content was determined by means of a modified Karl Fischer titration (2). 2-Butanone has been successfully dried with this apparatus, but the process must be performed under an inert atmosphere such as dry nitrogen. Without the inert atmosphere, the solvent in flask B acquires a bright yellow color after approximately 12 hr of reflux. In addition, the solvent in flask A becomes a much lighter, but noticeably, yellow color. The water content was found to have dropped from 0.8% to approximately 0.002% as assayed by modified Karl Fischer titration ( 2 ) . However, GLC analysis of the solvent detected a large number of peaks probably due to aldol condensation of the ketone on the basic surface of the drying agent. When a new batch of solvent was dried under a much slower rate of reflux and under an inert atmosphere of dry nitrogen (which was slowly admitted through the filling port C), the solvent in flask B became only light yellow in color and that in flask A was colorless. GLC analysis showed no extra peaks. The water content in the solvent was found to be 0.001% after 48 hours. Reagent grade acetonitrile has also been successfully dried. After 36 hours of reflux, the water content dropped from 0.09% to less than 0.0005%. No decomposition of the solvent was detected. This apparatus has been successfully used to dehydrate solutions of certain phosphorus esters which are hygroscopic. For example, when a 0.01M solution of diethyl benzoylphosphonate (3)in acetronitrile was dehydrated using an apparatus identical to that reported in (I),polarographic analysis showed that extensive hydrolytic decomposition had occurred. The hydrolysis is evidently catalyzed on the basic surface of the molecular sieve dust which is washed down into flask B. However, when a similar dehydration was performed using the present apparatus, no hydrolytic decomposition was de(2)

w.

M. Haynes, Ph.D. Thesis, Oklahoma State University,

May 1966. (3) K. D. Berlin and H. A. Taylor, J . Am. Chem. SOC.,86, 3862 (1964).

1016

ANALYTICAL CHEMISTRY

-28

MM O . D .

C

A B Figure 1. Modified dehydration apparatus

PART

tected evidently because the diethyl benzoylphosphonate itself did not come in contact with any molecular sieve dust. Conclusions. A modified dehydration apparatus has been found effective in drying several organic solvents and solutions. One of the major advantages of this setup is that the drying process may be accomplished without introducing impurities due to reaction of the solvent with the drying agent. In addition, solutions may be dried without the solute coming in contact with molecular sieve dust, thus avoiding undesirable side reactions. Although this has not yet been attempted, it should be possible to volumetrically calibrate the distilling flask A in order to dehydrate quantitative organic solutions containing relatively nonvolatile solutes. This could be done by turning off the flask A after sufficient heating time and distilling solvent from the flask B to flask A until the solvent level reached the quantitative mark. ACKNOWLEDGMENT

We express our gratitude for helpful discussions with M. Wayne Adkins concerning the equipment described. RECEIVED for review January 25, 1968. Accepted February 23, 1968. Research was supported in part by the Research Foundation, Oklahoma State University. One of the authors @. S. R.)is a Du Pont Teaching Fellow, 1967-68.