Electroanalytical Determination of Molecular Fluorine in the Atmosphere Samuel Kaye and Michael Griggs General Dynamics Conuair, San Diego, Calg. FLUQRINE is among the possible energetic oxidizers which may be used for propellant systems. A serious disadvantage is its toxicity, the threshold limit for which is 0.1 ppm by volume in air (I). A study of the atmospheric diffusion of fluorine from spills of liquid fluorine-oxygen mixtures was recently made (2). Instruments were required in this program to monitor the fluorine vapor concentration in air. Four types of instruments and methods were selected to monitor fluorine and hydrogen fluoride. Absorption and titration by wet chemical methods, conductometric methods (Davis H F indicator-recorder Model 11-7010-RP Special), Kr8j quinolclathrate (Tracerlab Model FM-2 fluorine indicator-recorder), and electrochemical techniques were used. This paper describes the electrochemical fluorine detector-indicator instrument. No suitable developed instrument was available to monitor fluorine specifically. We chose, therefore, to adapt the principle of the detector used by one of us (3) for balloon-borne measurements of ozone. In this instrument, air containing oxidant (Os) oxidizes the I- ion in a buffered KI solution in a cell containing platinum and silver electrodes. The iodine formed is reduced at the silver cathode. Insoluble AgI is produced on the cathode so that the iodide is removed from the reaction. This system was made specific for fluorine by substituting LiCl for the KI solution. Atmospheric oxidants do not affect the salt. Fluorine, however, oxidizes the chloride ion which undergoes the following electrode reactions :
Anode 2C1Cathode Clz
- 2e -+
+ 2e
+
Clz 2C1-
For every molecule of fluorine, two electrons flow through the coulometric circuit and two atoms of chlorine are transferred from solution to cathode. Hence, the current generated may be calculated to be 0.14 pA times the flow in ml/min/ppm FBin the atmosphere. This current is easily measured when the flow rates are 100 to 300 ml/min. Theoretical details of coulometric halide determinations have been described (4). There are other methods and instruments available, but most do not distinguish the fluoride ion from the fluorine molecule concentration (5-10). EXPERIMENTAL Apparatus. A schematic diagram of the apparatus is shown in Figure 1. It consists of a bubbler fabricated from a lOjl8 T joint. The inner joint bears a tube reaching to within a few millimeters of a 20-gage silver wire electrode which is fused into the bottom of the outer joint. The unit is then sealed so as to form a reaction cell about 13 to 15 cm long. The sensing element is thus a silver-silver chloridechlorine galvanic cell. The second electrode is a 1-in. square of 80-mesh platinum gauze formed into a cylinder encircling a 4-mm dip tube. This acts as a bubbler. A Mast Model 2 AP-X piston pump and motor assembly (Mast Development Co., Davenport, Iowa) is attached to a side arm on the reaction tube by rubber tubing. This is used to suck air through the sell. A Teflon tube is attached to the inlet of the dip tube to bring in outside air for sampling. About 6 ml of un-
--CCVCL
L
MI CROAM M ETE R
Figure 1. Schematic diagram of fluorine analyzer buffered LiCl solution is sufficient to bring the liquid level above the platinum anode. The cell has a spontaneous EMF of about 0.25 V. This is balanced by an external EMF of the same magnitude so that no current flows in the absence of fluorine. The compensating EMF is selected by a voltage divider across a mercury cell. The current generated by the cell when fluorine passes through it is read on a 0 to 20 or 0 to 50 pA Rustrak Miniature Model 88 recorder or on a microammeter. For calibration purposes fluorine standards of air containing 5 to 50 ppm by volume of fluorine were prepared by diluting pure fluorine (approximately 98.0% FJ with compressed air in clean passivated stainless steel tanks. The fluorine concentration was determined by absorbing a metered volume of the fluorine standard in 1% potassium iodide solution. The potassium iodide solution was then analyzed for fluoride by colorimetric analysis with thorium nitrate. Between calibrations no more than 2 5 Z of the mixture was used; this minimized the effects of fluorine release and adsorption from the tank walls. The actual measured output of the instrument agreed with the theoretical output, and the electroanalytical principle was as accurate as other methods of analysis of fluorine at these low concentrations. Further calibrations were therefore limited to measurements of the pump flow rate. (1) “Threshold Limit Values for 1962,” American Conference of Governmental Industrial Hygienists, 1014 Broadway, Cincinnati 2, Ohio. (2) “Atmospheric Diffusion of Fluorine from Spills of FluorineOxygen Mixtures,” General Dynamics Convair Rept GDCDDB66-001,May 1966. (3) M. Griggs, Ph.D. thesis, University of Oxford, Clarendon Laboratory, 1961. (4) J. J. Lingane, “Electroanalytical Chemistry,” 2nd ed., Interscience Publishers, New York, N.Y., 1958. (5) Philip Diamond, J. Arne?. 2nd. Hyg. Ass., 24, 399 (1963). (6) 0. H. Howard and C. W. Weber, ibid., 23,48 (1962). (7) B. B. Baker and J. D. Morrison (to Southern Research Institute), U.S. Patent 2,870,067 (Jan. 20, 1959). (8) M. G. Jacobson (to Mine Safety Appliances Co.), U. S. Patents 3,039,053 (Jan. 12, 1962), 2,861,926 (Nov. 25, 1958), 2,464,087 (March 8, 1949). (9) G. H. Farrah (to Aluminum Co. of America), U. S. Patent 3,058,901 (Oct. 16, 1962). (10) W. A. Darrah, U.S.Patent 2,278,248(March 31, 1942). VOL. 40, NO. 14, DECEMBER 1968
e
2217
5
0.6
0.2 .d4
5.0 CELL DEPTH
6.0
(CM )
Figure 2. Effect o f depth of electrolyte 030 current output and response time
The pump operated at a constant speed of about 220 ml/ min. This volume was calibrated and checked for each pump as necessary. For monitoring low concentrations of fluorine-Le., when the instrument was over 1000 feet from a spill source-the recorder was converted to a maximum range of 10 pA for greater sensitivity. DPSCUSSIBN
Several considerations are important for an optimum design for the instrument. Lingane (4) has discussed the factors which afford good time response. For a minimum value of response time the volume of solution and the diffusion layer ess must be small; the electrode area and diffusion comust be large. iffusion coefficient of the chloride ion is larger than any other halide ion which might be used. The use of chloride rather than another halogen ion also achieves a specificity for the detection of fluorine. A high temperature also raises the coefficient for diffusion and makes the diKusion layer thinner, but no attempt was made to control these factors by temperature regulation except to avoid Barge changes in temperature. The design and dimensions of the dip tube cell and platinum gauze provide vigorous stirring which keeps the diffusion layer thin. The bubbling is necessary to bring the chlorine into contact with the cathode so current can pass. The cyhdrical fine mesh electrode permits continuous and Bee circulation of the Xiquid and affords a large reaction area and repeated contacts between solution and electrode so that CQ chlorine i s lost. The size of the bubbles is such as to cause the solution close to the walls to pass into and out of the platinurn cylinder more than 20 times per sec (3). The optimum stirring design was produced by a bubble tube 4 mm in diameter which reached to B to 3 mm from the bottom of the gauze cylinder. A 0.1M KI solution buffered to pH 4 was first used for the electrolyte. This solution is affected by oxidants such as O x and NOz in air. ICCl was therefore used to replace the KI solution. Ozone and NOz did not oxidize this solution. However, LiCl performed better than KCI and an unbuffered ion was more stable than a buffered solution. rformance. The effect of the various parameters discussed above was determined experimentally. The operating
0
ANALYTICAL CHEMISTRY
.d6 .d8 ,Ib .I: FLUORINE (PERCENT)
.Ib
.
Figure 3. Effect of fluorine concentration on current output
conditions were then optimized. There is considerable leeway in design and in the mode of operation such as pumping rate, compensating voltage, and solution concentrations. The compensating voltage may vary from -0.10 to about -0.35 without greatly affecting the current output. The best value is about -0.25 V because the current caused by fluorine is affected least at that value. The polarization of the electrodes by too low or too high a compensating voltage does affect the response time, however. A salt concentration of about 0.2M furnishes uniform response time and fairly uniform current output over a considerable range, but a range of 0.2 to 0.6M is acceptable. Figure 2 shows how current output and response time vary with the depth of solution in the cell. The greater the amount of solution, the longer the response time. The current output is constant even when the depth of solution is less than 1 cm. However, to maintain stability, the depth of solution used was always sufficient to immerse completely the platinum ebctrode. The capacity of the cell for fluorine was also increased thereby. The relation between the flow rate or fluorine amount and the current output is a linear function. Figure 3 shows that the output is linear for fluorine concentrations to more than 1000 ppm. Very high or very low fluorine concentrations could be accommodated, if known ahead of time, by calibrating the constant speed pump for much lower or higher constant speed, respectively. The ammeter must be selected to encompass the appropriate range. The instrument was satisfactory for measuring and recording fluorine concentrations over a wide range of values. The accuracy, reliability, operational life, simplicity, and servicing were all well within the requirements of an instrument for toxic-gas monitoring. It is sensitive only to fluorine of all atmospheric oxidants and needs no basic calibrations once complete fluorine absorption is assumed. At a constant pump speed the relation between current output and fluorine concentration is absolute because the instrument essentially counts fluorine molecules. RECEIVED for review July 1, 1968. Accepted September 3, 11968.
