AN INVESTIGATION OF CERTAIN PHYSICAL PROPERTIES OF HALOGEN FLUORIDES By Herbert Bradford Thompson, Jr. A THESIS Submitted to the School of Graduate Studies of Michigan State College of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1953 ProQuest Number: 10008440 All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete m anuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. uest ProQuest 10008440 Published by ProQuest LLC (2016). Copyright o f the Dissertation is held by the Author. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code Microform Edition © ProQuest LLC. ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106 - 1346 ACKNOWLEDGMENT The author wishes to express his sincere aicpreciation to Professor 1-1. T. Rogers for his guidance and assistance throughout the- course of this work, to Dr. J. L. Speirs for his help in the preparation of materials and equipment, and to the Atomic Energy Commission for a grant subsidizing this research. 049665 *'TABLE OF CONTENTS Page I. INTRODUCTION................ . .................. 1 II. HISTORICAL SUMMARY ............................... 2 Chemical Properties 2 ............................ The Physical Properties of Chlorine Lionofluoride . III. 7 The Physical Properties of Chlorine Trifluoride . 9 The Physical Properties of Bromine Konofluoride . 13 The Physical Properties of Bromine Trifluoride . . 14 The Physical Properties of Bromine Pentafluoride . 16 The Physical Properties of Iodine Konofluoride . . 17 The Physical Properties of Iodine Pentafluoride • 17 The Physical Properties of Iodine Heptafluoride . 19 PREPARATION AND HANDLING- OF M A T E R I A L S ..........21 The Gas-handling System . . ................... 21 Specific Procedures for Iodine Pentafluoride . .. The Auxiliary Solution-mixing System Purification and Handling of materials 25 .... 25 ... 52 Specific Procedures for Chlorine Trifluoride . .. 34 Filling the C e l l ......................... 54 The Hydrogen Fluoride Solution iii ............ 36 Page IV. FREEZING--POINT M E A S U R E M E N T S .......................37 Theoretical Aspects ............................ 37 The Nature of the Measurement................ 37 Temperature Measurement Using Thermocouples . 4-1 Experimental M e t h o d s ............................... 4-2 Temperature Measurement .....................4-2 The Cell and Cooling B a t h .....................54Preparation of S a m p l e s .......................59 R e s u l t s ........................ 59 Data and Calculations.........................59 Discussion of the R e s u l t s .....................65 V. CONDUCTANCE MEASUREMENTS ........................ 68 Theoretical Aspects 68 ............................ The Nature of the Measurement.................68 Conductance Units and Gell Constants The Instrument .... 70 ........................ 74- Experirnental M e t h o d s ...............................77 The I n s t r u m e n t ...............................77 The Conductance C e l l ........................ 96 Preparation of S a m p l e s ......................104Measurements vrith Iodine Pentaf luoride . , . Ill R e s u l t s .......................................... 113 Data and Calculations........................113 Discussion of the R e s u l t s ....................113 Iv Page VI.DIELECTRIC-CONS TANT MEASUREMENTS.................... 127 Theoretical Aspects . .......................... 127 The Nature of the Measurement............127 The Instrument ...................... 129 The Gell and theUnknownL i q u i d ................ 133 Experimental Methods ............................ 135 The I n s t r u m e n t .............................. 135 Measurements with Chlorine Trifluorido . . . 139 Measurements with Iodine Pentafluoride . . . 148 R e s u l t s .......................................... 159 The Dielectric Constant of Chlorine Trifluoride .............. . . . . . . . . . 159 The Dielectric Constant of Iodine Pentafluoride .............................. 166 Discussion of theR e s u l t s ..................... 174 VII. List S U M M A R Y .................... 179 of Symbols U s e d ................................ 181 List of Abbreviations U s e d ............................. 184 LITERATURE C I T E D ........... •............................ 186 v LIST OF TABLES Table Page I. Ranges Available on Amplifier-recorder C o m b i n a t i o n ..................................... 44- II. Data for Calibration of Thermo c o u p l e s ........... 47 III. IV. V. VI. VII. VIII. IX. Ioe-point Potentials for Thermocouples ......... 50 Freezing-point Data for Iodine Pentafluoride Samples ................................. 64 Audio Oscillator Capacitances and Frequencies . Vertical Amplifier Frequency Response on 400-cycle Setting ............................ 87 90 Solutions Prepared to Standardize the Glass Conductance C e l l .................................93 Calibration Data for the (Glass Conductance Cell 94 Data, on the Composition of Several Potassium Cnloride Solutions ............................ 97 X. Conductance Data on Several Potassium Chloride S o l u t i o n s .......................................98 XI. Residual Conductance of the Cell Immersed In the Oil B a t h ........................................ 105 XII. XIII. XIV. XV. XVI. Calibration Data for the Teflon Conductance Cell 106 Calculation of the Cell Constant for the Teflon Conductance C e l l ................................ 108 Data on Preparation of hydrogen Fluoride Solutions in Iodine Pentafluoride ........ . . 110 Data on Preparation of Various Solutions in Iodine Pentafluoride .......................... Conductance Data for Iodine Peneafluoride vi . 112 . . 114 Table XVII. XVIII. XIX. XX. Page Conductance Data for Solutions of Hydrogen Fluoride in Iodine Pentafluoride .............. 116 Conductance Data for Supposed Solutions of Solids in Iodine Pentafluoride ................ 119 Conductances of Iodine Pentafluoride and of Various Solutions ...................... . . . 121 Measured Resistance a,s a Function of Time . . . 122 XXI. Data and Initial Calculations for Calibration of Dielectric Constant Cell A .................. 14-5 XXII. Calculation of Cell Parameters for Dielectric Constant Cell A ............................ 14-7 XXIII. Dielectric-constant Data from national Bureau of Standards Circular 5 1 4 ...................... 147 XXIV. Thermistor-calibration Data ................... 151 XXV. Data and Initial Calculations for Calibration of Dielectric Constant Cell B .................. 154 XXVI. Calculation of Cell Parameters for Dielectric Constant Cell B ............................ 1 5 6 XXVII. Calibration Data for the Added Conductance Scale of the Twin-T B r i d g e ........................ . . 1 5 8 XXVIII. Data and Initial Calculations for Chlorine T r i f l u o r i d e .................................160 XXIX. Dielectric Constant Data for Chlorine Tri f l u o r i d e ................................ 161 XXX. XXXI. XXXII. The Temperature Dependence of the Dielectric Constant of Chlorine Trifluoride .............. I63 Data and Initial Calculations for Iodine Pentaf l u o r i d e ................ 167 Dielectric Constant Data for Iodine Pentaf l u o r i d e .......................... 168 vii Table XXXIII. Page The Temperature Dependence of the Dielectric Constant of Iodine Pentafluoride .............. 172 XXXIV. High-frequency Conductance Data arid Calculations for Iodine Pentaf l u o r i d e ............ .. 172 XXXV. Calculation of Apparent Dxpole Moments in Liquid Chlorine Trifluoride and Iodine Pentafluoride . 177 viii LIST OF FIGURES Figure 1. Schematic diagram of the still and gas-handling system .......................................... 2. The gas-handling system 3« The solution-mixing system ...................... 4. The solution-mixing system ...................... 5* The cell-filling apparatus for chlorine trifluoride...................................... 6. A cooling curve in the region of the freezing p o i n t .............. . .......................... Y* Temperature-measuring circuit 8. Ice-point tube in place within a Dewar flash . . . 9. The cooling bath ................................ 10. Thermistor thermoregulator circuit .............. 11. The freezing-point cell 12. Cooling curve for the first sample of iodine pentafluoride .................................. Ip. Cooling curve for the second sample of iodine pentafluoride .................................. 14. Cooling curve for the solution of hydrogen fluoride in iodine pentafluoride ................ Ip. Phase relationships between an applied potential and the components of the resultant current . . . 16. The relationship between the total impedance of the conductance cell and the various constituent impedances ...................................... ........................ .................. ........................ Figure Page 17* Graph of Rx versus l/-,/f".......................... 72 18* Simplified circuit of the conductance bridge .. . 19* Relationship of the components of the conductancemeasuring instrument . . ........ .................. 7 8 20. The ratio b o x ................. 21. The conductance-measuring a p p a r a t u s .............. 79 22. The ratio resistors in place within a double shield . . . . . . . . . ............... . . . . . 81 The ratio box and standard capacitance and resistance decodes with shielding removed . . . . 83 The ratio box and standard capacitance and resistance decades with back cover removed and shield in p l a c e ........................ 84 23* 24. 75 78 ' 25. Circuit of the audio-frequency generator.......... 85 26. Circuit of the oscilloscope detector ............ 89 27* Frequency response curve for the vertical amplifier on the 400-cycle setting .............. 91 28. Parallel-T filter c i r c u i t ........................ 91 29. The conductance bath and thermistor regulator . . 95 30. TV versus Vc~ for a series of potassium chloride s o l u t i o n s ........... 96 31. Section through the upper and center portions of the conductance c e l l .......................... .. 1 0 0 52. The conductance cell 33. Circuit for determination of "che residual conductance of the conductance c e l l ............. 1 0 3 34. Rx versus 1/-/F" for solutions used to calibrate the Teflon conductance c e l l ......................107 ................... 101 x Figure 35• Page Rx versus 1/Vf~ for several samples of iodine pentaf luoride 36. .................................. 115 Rx versus 1/Vf for several solutions of hydrogen fluoride in iodine pentaf l u o r i d e ..................117 37- 38. Rx versus 1/VF for two solutions of hydrogen fluoride in iodine pentafluoride ................ 118 Rx versus 1/VF for supposed solutions of solids in iodine pentafluoride ........................ 120 39* Change of Rx with t i m e ........................... 123 40. Conductance as a function of concentration of hydrogen fluoride in iodine pentaf luoride......... 125 41. Circuit of the Twin-T b r i d g e ..................... 130 42. Equivalent circuit for Cvnd leads to she unimown at high f r e q u e n c i e s ..............................132 43* Equivalent circuit for the unknown at high f r e q u e n c i e s ................................... . 132 44. The Twin-T bridge with generator and. detector 45* Power supply for the frequency meter ............. 140 46. Section borough the cylindrical cell A ............ 14-1 47. Section through the cylindrical cell B ............ 149 48. Calibration curve for the thermistor used with dielectric-constant cell B ........................152 49. l/Cx versus f2 for iodine pentafluoride in cell 3 169 50. Temperature dependence of rhe dielectric constant of iodine pentafluoride ........................ 171 . . 13o I. INTRODUCTION Stimulated by the rapid growth of the field of fluorine chemistry, interest in the halogen fluorides as fluorinating agents has increased greatly during the last five years, and a knowledge of their physical properties has become increasingly important. No dielectric constant measurements for these compounds have been, reported aiad the few conduc­ tance data available are not of high precision, nor were they made on materials of established purity. Accordingly, the construction of equipment for purifying halogen fluorides and for measuring the liquid dielectric constants and conductivities of these materials was under­ taken, and the methods devised were tested by actual measurements. Freezing-point measurements were also carried out, as these provided evidence as to the purity of the materials used. The high reactivity of the halogen fluorides necessitated the development of special procedures and of equipment made from highly resistant materials for all measuring and handling processes. II. HISTORICAL SUMMARY Chemical Properties Our knowledge of the properties of the halogen fluorides has grown rapidly during the past five years. The compounds had been known for some time before this, however; Gore (l) prepared iodine pentafluoride in 1870 and Moissan (2), Prideaux (3, 4), and Lebeau (5, 6, 7) made bromine tri­ fluoride before 1906. Otto Ruff and his eoworkers, who did much of the early work on interhalogen compounds, announced the preparation of the remaining five halogen fluorides— chlorine monofluoride, chlorine trifluoride, bromine mono­ fluoride, bromine pentafluoride, and iodine heptafluoride— between 1930 and 1933 (8, 9, 10, 11, 12, 13). An eighth compound, iodine monofluoride, has only been observed spectroscopically (14). Many phases of the chemistry of these materials have been investigated. The recent book Fluorine Chemistry (15a) contains a chapter by H. S. Booth and J. T. Pinkston, Jr., entitled "The Halogen. Fluorides,” and several review articles on the chemistry of these materials have appeared (16, 17, 18, 19). In the present summary only an outline of halogen fluoride chemistry in general will be presented; - 2 - the following sections will concentrate on the work to date on the physical properties of these seven compounds. The most striking single characteristic of the halogen fluorides is their extreme reactivity. All are powerful fluorinating agents; the reactivity of chlorine trifluoride has been compared with that of fluorine itself. In fact, chlorine trifluoride has been considered as a rocket fuel (2 0 ). Booth and Pinkston give the approximate order of reactivity: C1F3 ) BrF 5 ) IFr > GIF ) BrF3 ) IF5 > BrF. Inorganic elements in general are vigorously fluorinated by these compounds. However, certain metals are quite resist­ ant to attack by halogen fluorides; steel has proved suitable for shipping containers, and both steel (2 1 ) and copper (1 3 ) have been used for reactors in which to produce the materials. Copper is reported to resist reaction with chlorine tri­ fluoride and bromine trifluoride up to 4-00°G, though it shows visible evidence of corrosion after prolonged exposure. Nickel, Konel metel, and aluminum stand up well. reported to be usable to 750° G (1 5 a). Nickel is Platinum resists attack and has been used as an electrode material (2 3 ), though bromine trifluoride has been reported to corrode it at 100-150°C (22). Gutmann (24) makes the reasonable pro­ posal that metals resisting corrosion by halogen fluorides do so by building a protective fluoride coating on their surfaces. This hypothesis is supported by the discovery that the fluorides of these metals appear to he insoluble in bromine trifluoride. Molybdenum and tungsten, possessing volatile hexafluorides, are observed to react vigorously (1 5 a). In addition, silver, gold, niobium, and tantalum, known to form fluorides soluble in bromine trifluoride, are themselves easily dissolved (25, 26). Halogen fluorides do not react with atmospheric oxygen or nitrogen, but fluorides are readily, often violently, produced from other nonmetals, including sulfur, selenium, tellurium, phosphorus, arsenic, antimony, silicon, and boron (27). Hydrogen burns in halogen.fluoride vapors. Inorganic salts other than fluorides are, in general, attacked. Silver(Il) fluoride, the most powerful solid fluo- rinating agent yet prepared, can be made by oxidation of silver(l) chloride or of silver(I) fluoride (28). Water re­ acts violently (24), and copper(Il) oxide, boron(IIl) oxide, germanium(IV) oxide, titanium(IV) oxide, ioaine(V) oxide, and other oxides have been completely converted to the fluo­ rides by bromine trifluoride. Potassium chloride and barium chloride are also readily dissolved by bromine trifluoride, and again it is suggested that the oxides and halides which resist attack do so by forming a fluoride surface layer (28, 2 9 , 3 0 ). Most elements, oxides, and chlorides which undergo attach bv halo? ;en fluorides are converted to the hi .best - 4 - fluoride of the element involved. However, cobalt(Il) chloride is only partially converted to cobalt(III) fluoride hy bromine trifluoride, iodine pentafluoride, or iodine heptafluoride (51), half or more of the metal ending up as cobalt(ll) fluoride. Chlorine trifluoride, on the other hand, gives only cobalt(III) fluoride. The violence with which the halogen fluorides react with many materials greatly complicates their study. glass containers are readily attacked. In general, Quartz fares somewhat better (24) and has been used for conductivity cells for chlorine trifluoride, bromine trifluoride, and iodine pentafluoride (25)- However, while using a quartz vapor pressure apparatus with iodine pentafluoride, Ruff and Braida (11) found it necessary to pump the silicon tetrafluoride from the device between measurements. Reactions of halogen fluorides with organic materials have in general proceeded violently and led to mixtures of organic degradation products and charred material. Ruff and Keim (32) thus isolated a small quantity of fluorobenzene from the iodobenzene and carbonized matter produced by action of iodine pentafluoride on benzene. Attempts have been made to control the violence of these reactions. In general, success has been achieved only with bromine trifluoride and iodine pentafluoride, though Ellis and Kusgrave (33) recently employed chlorine trifluoride to fluorinate benzene in carbon ■tetrachloride. The use of lialogenated solvents to slow these reactions has proved successful in other cases; McBee, Liggett, and Lindgren (34, 35, 36, 37, 38) have patented processes for halogenation of benzene and for simultaneous addition of fluorine and bromine using bromine trifluoride in carbon tetrachloride or pentachloro-o^o^ot-trifluorotoluene. Though useful as a solvent for fluorinations, carbon tetrachloride is not entirely Inert toward interhalogens. Banks, Emeleus, Haszeldine, and Kerrigan (35) produced a mixture of fluorinated products including dichlorodifluoromethane and trichlorofluoromethane from carbon tetrachloride and bromine trifluoride. Iodine pentafluoride apparently acts quite slowly on some organic substances, as Bcott and Burnett (40) were able to observe the formation of a. dioxanate, IF5 .C4H 802 , upon adding this fluoride to dioxane. Cells for measurement of certain thermal and electrical properties require materials that are good insulators. For work with halogen fluorides these requirements are met by two recently-developed plastics, Teflon (polytetrafluoroethylene), and Fluorothene or Kel-F (polychlorotrifluoroethylene). Both materials are Inert toward halogen fluorides and. are poor heat conductors. Both are extremely poor electrical conductors at low frequencies. Teflon, moreover, is highly recommended as an insulator up to 100 megacycles. The Physical Properties of Chlorine Monofluoride Chlorine monofluoride was first prepared by Ruff and Ascher (8) by direct union of the elements. According to Domange and Neudorffer (41) this reaction proceeds readily at 220-230°C, starting with a 1:1 ratio of fluorine to chlorine. Schmitz and Schumacher (42) reported that chlorine mono­ fluoride may also be prepared by the reaction C1F3 + Cl2 = 3OIF. The heat of formation of chlorine monofluoride from the gaseous elements has been reported as -11.6 + 0.4 kcal. mole"-*by Wicke (43) and as -27*4 kcal. mole"-*- by Ruff and Menzel (9)* On the other hand, Schmitz and Schumacher (42) studied the reactions of fluorine and chlorine monofluoride with sodium chloride and found NaCl + MaCl + that GIF “ NaF + Cls } 1/2Fa = NaF A H = -24.5 + kcal. mole"-* 1/2G1s ;a H = -39.5kcal.mole-1 which leads to 1/2FS + 1/2012= C1F; a Hf= -15.0 kcal. mole-1. Still another value was obtained by Clockler (15b), using the heats of dissociation of chlorine monofluoride, chlorine, and fluorine into the constituent atoms: l/201g = Cl; a H = +28.6 kcal. mole-**- 1/2 Fs = A H = +31*7 kcal. mole--*- F + F; 01 = GIF; 1/2 Fs + l/2 Cl: 2 = GIF; A R ~ A Hj-*= -60.51 kcal.mole-1 0.0 kcal. mole"-1-. G-lockler pointed out that the formation of chlorine mono­ fluoride is observed to be exothermic, indieating that the heats of dissociation employed in the calculation must be in error. His suspicion of the accepted dissociation energy of fluorine on this basis is borne out by recent work by Wicke (44, 45) leading to a value of 40.2 kcal. mole”-, and by Doescher (46), giving about 5 8 kcal. mole"***. Substituted above, these would give -13 and -12 kcal. mole"-*- respectively for AHj*. Wahrhaftig 1a value (47) for the heat of dissociation of chlorine monofluoride into unexcited atoms, used by G-lockler in the above calculation, is in essential agreement with that determined by Schmitz and Schumacher (48) from band spectra: 60 .3 or 58.9 kcal. mole”-*-. The uncertainty arises because it is not known whether the dissociation producing the observed band gives a normal fluorine and an excited chlorine atom, or vice versa. The authors favored the higher value (49)- Chlorine monofluoride is by far the most difficult of the halogen fluorides to liquefy. The colorless gas condenses to a yellowish liquid at about -100°C. Ruff and Laass (50) gave the vapor pressure of the liquid as IOS1 0 P HS) = 15.758 + 3109/T + 1.538 x 105/T2 where T is the temperature in degrees Kelvin. From this curve a slight extrapolation gives a boiling point of - 8 - (!) -100.1°C. These authors reported the melting; point of the white solid as -1 5 5 .6 °G. Electron diffraction studies (51) give 1.63 ± 0.01A as the chlorine monofluoride bond length, in agreement with the infrared value (47) of 1.625& and the microwave distance (53) of 1.6281&. The microwave data give a dipole moment of 0.88 + 0.02D and indicate 20 per cent ionic bond character. The Physical Properties of Chlorine Trifluoride Ruff and Krug (10) in 1930 prepared chlorine trifluoride by direct union of the elements at- -170°G. Booth and Pinkston (15a) reported that the,reaction is carried out at 200°G in modern commercial preparation (54). This compound is also produced at 270-280°G from a 1:1 mixture of fluorine and chlorine monofluoride, but this reaction has little practical importance as a preparation (4l). Schmitz and Schumacher (55) reported the equilibrium constant for this reaction C1FS = C1F + Fa; K = ,2 j p C1F3 as 2.98 x 10”^ atm. at 205°C, 2.4 x 10**^ atm. at 300°C, and 1.43 x 10"2 atm. at 3 5 0 °C. Von Wartenberg and Riteris (5 6 ) reported the heat of formation as -28.4 kcal. mole-1. - 9 - However, Schmitz and Schumacher (42) gave 1/2FS + 1/2Clg = GIF ; A H f = -15.0 kcal. mole"1 GIF = G1F3 } F3 + RH = -25.0 kcal. mole*"1 which leads to 3/2Fs + 1/2G1S = C1F3 J / Hf - -40.0 kcal. mole"1 . Schmitz and Schumacher (57) discovered that the behavior of the gas could not easily be attributed to deviations from the ideal gas lav/s, and proposed the existence of a dimer in the gas phase. Their values for the dissociation constant of the dimer K = (5) “" V ) s are 26.9 atm. at 9.5°C, 52.1 atm. at 20°C, and 35*4 atm. at 24.2°G. Ruff and Krug (10) observed that the colorless gas con­ denses at 1 2 .0 °C to a pale green liquid and freezes to a white solid at -S2 .6 °G. reported the triple as 11.75°0. G-risard, Bernhardt, and Oliver (67) point as -76.32°C and the boiling point These authors measured the heat capacity and dis­ covered a transition occurring in the solid state The liquid density is at-8 2 .7°C. given by Ruff and Krug -as d* = 1.3853 - 2.942 x lCT^t - 3-79 x 10~6 t2 where t is the temperature in degrees centigrade, while Gwinehart, as quoted by Booth and Pinkston (15a) from a - 10 - private comiminication, ^ives d!? = 2.723 - 0.00307 T (4) which, converting from Kelvin to centigrade scales, becomes a£ = 2.890 - c.00307 t. (5) Raff and Krug gave the vapor pressure of the liquid as l°s10p ( ™ - Hs) = 7.42 - 12.92/T. (6) Schafer and Wicke (58) studied the Raman spectrum of liquid chlorine trifluoride and proposed a -symmetrical pyramid structure. A line at 100 cm"^, not predicted for this struc­ ture, was attributed to dimer formation in the liquid. On the basis of this structure, these authors calculated entro­ pies and heat capacities from spectral data.. Recent work, however, has failed to confirm the pyramidal configuration. X-ray (59), microwave (60), and electric dipole moment (6 l) studies favor a planar structure: Cl F F The various parameters are summarized in Table II-l. The bond, angle 0 is 86.59^ according to x-ray data at - 11 - -120°C, while distance a is 1.7161 and b is 1.62li. Micro­ wave values for the parameters a, b, and 0 are 1. 69 8i, 1.59si, and 87.29° respectively. Jones, Parkinson, and Murray (62) studied both Raman and infrared spectra and concluded that association in the liquid is indicated for the planar as well as the pyramidal struc­ ture. Magnusson (61) reported the electric dipole moment as O.5 5 4 D after correcting for dimer formation in the gas sample employed. The dimer was assumed to have no permanent moment. Recent analyses of infrared and Raman spectra (62, 67) also indicate a planar structure. However, nuclear magnetic resonaioce absorption studies (65) have not shown the "chemical splitting" to be expected for a structure where one fluorine atom is in a different electronic environment than the other two.(64). On the basis of rotational moments based on the new structure, Scheer (66) undertook the recalculation of the thermodynamic properties of chlorine trifluoride from the spectral data. He obtained an entropy value at the boiling point of 6 6 .9 0 e.u. in excellent agreement with a third law value by Orisard, Bernhardt, and Oliver (6 7 ), 66.87 + O.lOe.u. This agreement, however, must be fortuitous according to Weber and Ferigle (68), since the vibrational frequencies as well as the rotational moments must be reassigned to agree with the newer structure. These workers obtained 66.64e.u. as the entropy at 284.91°G, taken as the boiling point. - 12 - Schmitz and Schumacher (69, 7 0 ) reported that in the visible and ultraviolet regions chlorine trifluoride is trans­ parent above 4700A, increases in absorption gradually to 2200&, then rapidly to below 200Q&. The electrical conduc­ tivity of liquid chlorine trifluoride is reported to be less than 1 0 ~ 6 ohm“l cn"^ at 0 °G (23)* Greenwood (17) has quoted an unpublished value by Banks of 3 x 10”9 ohm"-1- cm"-1-. The Physical Properties of Bromine Monofluoride Bromine monofluoride is extremely unstable, and has only been prepared in mixtures with bromine trifluoride. Ruff and Braida (11) exposed bromine vapor to fluorine in nitrogen at 1 0 °C, then fractionally condensed the product at successively lower temperatures. The fourth fraction, at -120°C, was esti­ mated to be about half bromine monofluoride, half bromine tri­ fluoride. Bromine monofluoride disproportionates completely into bromine trifluoride and bromine at 50°G. Ruff and Braida estimated that the pure liquid would freeze at about -33°C and boil at 2 0 °C. Durie and G-aydon (14) have analyzed the infrared emission “I spectrum and obtained 49*7 ± 0 . 0 3 kcal. mole“x as the energy of dissociation. However, Borderson and Schumacher (71, 72) examined certain absorption bands attributed to gaseous bromine monofluoride and gave the heat of dissociation into normal atoms at 5 9 * 6 kcal. mole"-1-, and the bond distance as - 13 - 1.74&. Smith, Tidwell, and Williams (73) examined the micro­ wave spectrum and obtained a bond distance of 1.759& and a dipole moment of 1.29&. The Physical Properties of Bromine Trifluoride Moissan (2), Lebeau (5, 6, 7), and Prideaux (3, 4) are credited with the discovery of bromine trifluoride, Ruff and Braida (11, 13) described it as a pale yellow-green liquid freezing at 8.8°C to a solid of the same color and boiling at 127•6 + 1°C to give a colorless vapor. The latter workers gave for the density and vapor px°essure of the liquid: <3.4 = 2.867 - 2.77 X 10“3t (7) los10p = 8.41954 - 2220.2/T. (8) A recent paper (74) by Oliver and G-risard gave the melting point as 8.77°C and the heat of fusion as 2.8746 + 0.003 kcal. mole“3-. data. These authors also gave heat capacity and entropy Stein and Vogel (8 5 ) measured the refractive indices of mixtures of bromine trifluoride and bromine pentafluoride and gave the equation - 1.4536 - 0.1237N + 0.00968N2 .+ 0.0137N5 (9 ) 'where N is the mole fraction of bromine pentaf luoride. Electron diffraction measurements (51) have been inter­ preted as indicating a trigonal pyramid structure for bromine trifluoride, with the bromine occupying the apex and the three fluorine atoms taking up the corners of the base. - 14 - The Br-F distance is 1*785. and the F-Br-F angle is 86°. This is in agreement with nuclear magnetic resonance absorption exami­ nations (6 5 ) in which no chemical splitting was observed, indicating that the three fluorines are in equivalent positions * Interest in bromine trifluoride has been greatly aug­ mented by the discovery (76) that it is an ionizing solvent: 2BrF3 = BrFs+ + BrF4” . The electrical conductivity of liquid bromine trifluoride at 25°C and 1000 cycles is given by Banks, Emeleus, and Woolf (2j5) as 8.0 x 10“3 ohm"^ cm”^-. A negative thermal coefficient of conductance would appear to indicate that the stability of the ions formed decreases with temperature. In this system the BrFs+ ion is an electron-pair acceptor or Lewis acid; thus fluorides dissolving to produce this ion may be considered as acids in bromine trifluoride. Emeleus, Sharpe, and coworkers (76, 25) have discovered several of these, among them niobium(V) fluoride and tin(IV) fluoride: NbFs + SnF* + BrF3 = NbF6" + BrFs+ 23rF3 = SnFs“ + 2BrFs+. Similarly, bases have been prepared from potassium and barium fluorides (76, 29): KF + BaF3 + BrF3 = K+ + BrF4” 2BrF3 = Ba++ + 2BrF4” . Neutralization-type reactions have proved possible, and - 15 - several hitherto unknown complex fluorides have been prepared in this manner (2 5 , 77, 78, 5 0 , 26): BrFsAuF4 + AgBrF* = AgAuF* + 2BrFs BrF sNbF 6 + KBrF4 = KNbFs + 2BrF3 . Furthermore, these acid and base solutions in bromine trifluoride exhibit electrical conductances greater than that of the solvent, and it has proved possible to carry out con­ ductance titrations (76). Thus bromine trifluoride as a solvent possesses a complete acid-base system; this phase of thechemistry of bromine trifluoride detail in a recent is discussed in some review article by Gutmann (24-) • The Physical Properties of Bromine Pentafluoride Though bromine pentafluoride may be purchased commer­ cially in the United States today, very little has been pub­ lished about its physical and chemical properties. Greenwood (17) observed in 1951 that we were still almost entirely dependent on the original paper of the discoverers, Ruff and Menzel (12), for data on this substance. Bromine pentafluoride is formed by the reaction of fluorine and bromine trifluoride and is a colorless liquid solidifying at -6l.3°C and boiling at 40.5 +0.5°G. The vapor pressure between -60° and +24°G is log]_0P (mm. Hg) = 8.0716 - 1627/T. The density throughout the <3.4 = 2.551 - (10) liquid range is given by 3.46 - 16 - X 10-h. (11) At the melting po-int the density of the solid is 3 - 0 9 and that of the liquid is 2.763* Bromine appears to be insoluble in bromine pentafluoride. A recent study of the Raman spectrum (79) seems to indi­ cate a tetragonal pyramid structure for bromine pentafluoride, with fluorine atoms at the apex and the corners of the base and the bromine atom within the body of the pyramid. Nuclear magnetic resonance absorption (6 5 ) shows the chemical split­ ting to be expected for such a structure, in i^hich four fluorines have identical symmetry while the fifth differs. The refractive index (8 5 ) of the liquid at 25°C is 1.341. The Physical Properties of Iodine Monofluoride This species has only been observed in emission spectra from the combination of iodine and fluorine. Durie and G-aydon (14) analyzed the data and calculcated the energy of dissociation to normal atoms as 45*6 + 1 kcal. mole”^. The Physical Pronerties of Iodine Pentafluoride Iodine pentafluoride is the only halogen fluoride not originally prepared from direct union of the elements or by oxidation of lower fluorides with fluorine (17). G-ore (l) first made iodine pentafluoride in 1870 by the reaction 5AgF + 3Is - 5AgI + IFb . Moissan (2) prepared it from the elements in 1891* - 17 - Ruff and Braida (11) gave 9 *6 °G as the melting; point and estimated the boiling; point from the vapor pressure data as 9 8 + l.ghC. Ioaine pentafluoride is a colorless liquid of density dj = in the range 0-35°0. 3*754-- 3.29 - 0.004t (1 2 ) The density of the solid at 0°G is The vapor pressure of solid and liquid, respectively, are given by los10p - 1°310P = 11.765 - 3035/T and 8 -2b (1 3 ) - 2205/T. The heat of fusion is +3.80 kcal. mole”^- (80). (14) Above 5G0°G iodine pentafluoride disproportionates to iodine and iodine heptafluoride. Infrared and Raman spectra (81) have been interpreted as indicating a tetragonal pyramid structure, while electron diffraction studies by Braune and Pinnow (82) gave a trigonal F F bipyramid. tore recent electron diffraction work appeared to disprove the latter structure,, and Rogers, Warthaftig, and - 18 - Schomaker (51) suggested a distorted octahedral structure* However, G-lauber and Schomaker (84) have concluded that the methods employed to analyze the electron diffraction data are not valid for molecules containing heavy atoms bonded to light ones. The electrical conductivity of iodine pentafluoride is reported to be 2 - 3 x 1 0 ~ 5 ohm-'*' cm"**" at 2 5 °C and 1 0 0 0 cycles. Banks, Emeleus, and Woolf (23) found that direct current measurements showed deviations from Ohmfs Law, indicating an electrode polarization* From this they concluded that no lower fluoride of iodine exists under these conditions* The Physical Properties of Iodine Hentafluoride Iodine heptafluoride” is the only’known molecule of the type ABy (17). It was first isolated by Ruff and Keim (83), who prepared it by reaction of fluorine and iodine penta­ fluoride at 27°0, Iodine heptafluoride is a white solid below 4*5°G, at which temperature it sublimes to produce a colorless gas. At two atmospheres pressure a colorless liquid freezing at 6 °C may be observed. The vapor pressure of the solid from -6 3 ° to 0 °G is given by 10S10P = 8.6604 - 1602.6/T. - 19 - (15) The infrared and Raman spectra have been investipated (81). Thes e indi cate a p entapo n a1 -b ip yram ida 1 stru ctu re . F F F Nuclear naaretic resonance absorption studies (6 5 ) Nave failed to show chemical splitting due to the two different types of position occupied by fluorines in this structure. - 20 - III. PREPARATION AND HANDLING OF MATERIALS The G-as-handling System As noted in a preceding section, the study of properties of halogen fluorides is greatly hampered hy their high reactivity. Materials for storage containers, cells, and purification equipment must he chosen to avoid contamination with corrosion products, and the high affinity of the halogen fluorides for water makes it Imperative that atmospheric moisture be excluded during the handling processes. To meet these problems, the gas-handling system described in the present section was designed and constructed. This system was housed in a specially constructed hood, provided with sliding doors lined on the hood side with asbestos board. The system is shown schematically in Figure 1, while a photo­ graphic view is provided in Figure 2. Dry air could be admitted into the system through a sulfuric-acid bath and a magnesium perchlorate tower at A. Hydrogen fluoride, chlorine trifluoride, or bromine penta­ fluoride could be drawn from the tanks at B. As these vapors are corrosive to valve materials, each branch line was pro­ vided with a valve in addition to the one on the tank. The pressure in this portion of the line could be read on the - 21 - & & <— <8> To Pressure Regulator To Back Hood HF C1FS BrFs * To Pump To Container Figure 1. Schematic diagram of the still and gas-handling system for halogen fluorides: A, air-drying system with sulfuric-acid bubbler and magriesium-perchlorate drying tower; B, tanks containing volatile materials, hydrogen fluoride, chlorine trifluoride, and bromine pentafluoride; C, aluminum traps for handling and storing materials; D and E, connections for cells, storage containers, a.nd mixing equipment; F, Helicoid pressure gauge, 0-1500 mm. Hg absolute; G-, copper tank for storage of gases; H, Helicoid pressure gauge, 0-400 mm. Hg absolute; J, aluminum trap; K, Fluorothene beaker for prelimi­ nary treatment of iodine pentafluoride; L, Monel still pot: M, fractionating column with nickel packing (jacket not shewn); N, still head and cold finger (water connections not shown); P, small hold-back reservoir; q>£), large valves with phosphor bronze bellows; &, small valves with bra-ss bellows; Q , valves supplied on gas cylinders. - 22 - Helicoid Gauge K, and a gas could be stored in the copijer tank G- if desired* Materials could be condensed in either of the traps at Cj these traps were constructed of aluminum to provide high heat conductivity, and could be cooled with, a Dry Ice-isopropanol mixture or with liquid air. The pres­ sure in the left-hand portion of the line could be read on Helicoid gauge F. This portion of the system had two connec­ tions, D and E, for storage containers, mixing devices, and cells to be filled. The trap J between the line and the pump connection could be cooled to remove condensable vapors. The still was provided with a Fluorothene beaker K for preliminary treatment of liquids, a, Monel still pot L, a *i nickel fractionating column M with nickel packing,-1- and a Fluorothene head with cold finger. The heated jacket around the column, the connections for circulating water through the cold finger, and the thermocouples for observing the tempera­ tures of jacket, column, and head are not shown in Figure 1. Between the still head the the container was a small reser­ voir P, in which the distillate could be caught during the time required to change receptacles. This portion of the system was constructed of Fluorothene, so that the liquid *** Nickel distillation packing, nl6x..l6 ,n made by the Scientific Development Company, State College, Pa., was used. - 24 - distillate would not be contaminated by corrosion products from metals. The valves were so arranged that only vapors would be in contact with the bellows and other metal portions. The pumps and absorption towers and the pressureregulating mechanism for the still are not shown in Figure 1. The pumps were protected by absorption towers filled with sodium fluoride and soda lime to remove halogen fluorides, and anhydrous calcium sulfate to remove water vapor. The pressure at N and P was controlled during distillations by means of a Cartesian manostat and was observed on a Fluorolube-oil manometer. Nickel tubing was used throughout most of this system.; Fluorothene tubing was employed near the still head and liquid-take-off system. The large valves were nickel with phosphor bronze bellows between the valve chamber and the atmosphere; on the still take-off system Fluorothene valve bodies were constructed to replace the nickel. The small valves contained brass bellows. Specific Procedures for Iodine Pentafluoride The Auxiliary Solution-mixing System Figures 3 and 4 illustrate the equipment used to prepare solutions of various materials in iodine pentafluoride. using this apparatus, care was taken: i. to clean and dry all containers before use, - 25 - In w * Figure 3* The solution-mixing system: R, Fluorothene storage beaker; S, small Fluorothene test tube; T, large Fluorothene test tube for mixing solutions; U, experimental cell; V, W, connections to the gas-handling system; X, length of Fluorothene tubing; -- 1-- , flare unions at which system could be taken apart. - 26 - ±i. "to store or keep liquid iodine pentafluoride in contact with Fluorothene rather than nickel, iii. to remove all residual liquid from nickel tubes and from valves immediately after any transferring process. The parts of this system could be disassembled at the flare union s shown. The Fluorothene beaker R could be filled from the still and used as a storage container for the distillate. All parts of this container that 'remained in contact with the liquid, including the outlet tube which extended nearly to the bottom / of the beaker, were constructed of Fluorothene. After removal of the beaker from the still, a drying tube containing anhy­ drous calcium sulfate (Indicating Drier!te) was attached to the inlet tube. Solutions were mixed in test tube T. Fluorothene, with a nickel plug. This tube was of The outlet tube, which extended to the bottom of the test tube, consisted of a length of Fluorothene tubing waxed at the top to a nickel tube lead­ ing through the plug to valve 6 . Before the mixing process the individual parts of this system were cleaned and rinsed with a good grade of chloroform or carbon tetrachloride. Most of the solvent was removed with an aspirator and the parts were assembled as shown in Figures p and 4. Connections V and W were then attached to D and E on the gashandling system and valves 3 , 4, 5 , 6 , and 7 were opened; valves 1 and 2 remained closed* The system was evacuated to remove the last of the rinsing solvent, then closed off from the pump for several minutes to test for leaks* If no leak was indicated by rising pressure as shown on gauge F, the system was then slowly filled with chlorine trifluoride vapor to approximately atmospheric pressure. Often the greenish color of chlorine gas in the Fluorothene tubes S and T would indicate that the chlorine trifluoride had reacted with residues of solvent or other materials in the tubes or valves. After this treatment the system was again evacuated and the mixing process could begin. Iodine pentafluoride was introduced into test tube T by evacuating T, closing valves 3 and 6 , and slowly opening valve 2. If there was insufficient air pressure over the liquid in the beaker R to drive the liquid into T, the pres­ sure could be increased by opening valve 1 slightly. When sufficient iodine pentafluoride had been delivered to tube T, valve 2 was closed and the nickel tubing from this valve to valve 7 was heated slightly with a Bunsen burner to drive any liquid into T. Valve 7 was then closed and valve 3 opened, and the transfer tube was evacuated to remove any residual liquid. - 29 - At this point the test tube T could be removed for weigh­ ing if desired. This could be accomplished by introducing dry air through the gas-handling system up to valves 6 and 7 9 and then disconnecting T from the system at the flare unions shown in Figure 3»^ After reconnecting T to the system, the tubing could -again be evacuated and treated with chlorine trifluoride to remove any atmospheric moisture. G-aseous or volatile solutes could be added to this sample from the Fluorothene test tube S. This tube had a nickel stopper connected to valve 4, and could be removed from the system for weighing, if desired, by disconnecting the flare union just above the valve* As with test tube T, the lines were filled with dry air before removing S, and evacuated and treated with chlorine trifluoride after replacing it. Samples in test tube S v;ere transferred to T by cooling the latter with Dry Ice-isopropanol mixture or liquid air, pumping any air out of T and the intermediate tubing, and opening the valves between S and T while keeping valves 2 and 6 and the valve above W in the gas-handling system closed. The transfer could be speeded somewhat by heating S with warm water or air. Sma.il amounts of hydrogen fluoride were added to the sample in T by filling test tube S with gas at a pressure These weighings were made 011 a. torsion balance having a capacity of 4.5 kilograms. Weighings seemed reproducible to about 0 . 3 grams. - 30 - read from gauge F. The tubing between test tubes S and T was then evacuated, after which the valve above W was closed, valves 4 and 7 were opened, and the vapor was condensed into test tube T using either Dry Ice-isopropanol mixture or liquid air. The pressure after the transfer could be read on gauge F if desired. The volume of test tube S was determined by weighing it empty and filled with water up to valve 4.^ The experimental cell U might be any one of the cells described in Sections V and VI, in which the actual measure­ ments were made. To fill the cell, both it and the adjacent line were evacuated through connection V and the gas-handling system. If not already so treated, the cell was exposed to chlorine trifluoride for several minutes and again evacuated. The valve above V was then closed, and the liquid was forced into the cell U by applying air pressure through valve 7 from the dry-air outlet A of the gas-handling system. The rate of flow of the sample into the cell was controlled at valve 6 , and as soon as liquid rose to the Fluorotnene tube, this valve was closed. The air was then evacuated from T and the liquid in the line to the cell drawn back into T while the latter was immersed in a Dry Ice-isopropanol mixture. The tubing between valves 5 and 6 was warmed slightly to hasten *i These and all other weighings of test tube S were carried out on a small torsion balance. Weighings on this balance were reproducible to about 0.02 grams. - 31 - *bliis process* Wien the line to the cell appeared to "be free of liquid, any remaining vapor was evacuated through connection V. Dry air was then admitted through V and valve 5 was opened slightly and closed to reduce the volume of any gas bubbles within the cell. The cell could then be removed from the line for measurement purposes. However, most of the conductance measurements were made with the cell in place on the line, as will be discussed in Section V. Following several of the conductance measurements, the samples of iodine pentafluoride were returned to test tube T by the same procedure as described above for removing liquid from the line leading to the cell. The presence and inten­ sity of brown color in a sample indicated whether the liquid had been badly contaminated through reaction during the trans­ fer processes. Solid sample's were introduced into test tube T before assembly of the system, and were exposed to chlorine tri­ fluoride when the system was treated v-/ith this gas during the preparation process. The test tube v/as shaken thoroughly at the time when it was removed from the system for weighing. Purification and Handling of Materials Freezing-point, conductivity, and dielectric-constant measurements were made on pure iodine pentafluoride, and freezing-point and conductivity determinations were carried - 32 - out, with solutions of hydrogen fluoride in iodine pentafluoride. In addition, attempts were made to dissolve sodium fluoride, lithium fluoride, and potassium hifluoride in this liquid. The purification of materials and the preparation of these samples is described below. Purification of iodine pentafluoride. The commercial product-^- was poured into beaker K of the gas-handling system shown in Figure 1. This material had been shipped and stored in iron containers, and was very dark in color. This was attributed to iodine formed as the pentaf luoride v/as reduced by the metal of the container or by other materials. Chlorine trifluoride gas was bubbled through the liquid^ when the liquid became clear the bubbling v/as discontinued. The iodine pentafluoride v/as then drawn into the still pot and distilled at approximately 34°C and 40 mm. I-Ig. Freezing- point sample A v/as collected In the fx^eezing-point tube directly from the stillj material for all other measurements involving iodine pentafluorid© v/as collected in the beaker R. Purification of hydrogen fluoride. G-as from the cylinder B in Figure 1 was condensed into the nearer aluminum trap at C, using a Dry Ice-isopropanol mixture. Some ■*" The iodine pentaf luoride used in this research v/as pur­ chased from the G-eneral Chemical Company, New York, N. Y. Chlorine trifluoride a/nd hydrogen fluoride were obtained from the Harshaw Chemical Company, Cleveland, Ohio. - 33 - non-condensable gas came from the cylinder with the hydrogen fluoridej this v/as pumped off at intervals during this first transfer. Next the sample v/as allowed to warm, and condensed into the other aluminum trap; the valve between the two traps v/as closed occasionally, and the pressure in the first trap v/as read on gauge Ii. When this pressure dropped below 100 mm. Hg, the second trap was closed off from the system and the remaining material in the first trap was pumped off. The hydrogen fluoride was then transferred back to the first alu­ minum trap in the same manner, where it was stored for use. Specific Procedures for Chlorine Trifluoride Filling the Cell Figure 5 is a schematic representation of the arrange­ ment employed to fill dielectric-constant cell A with chlorine trifluoride; V 1 is the connection to the gas-handling system discussed above. The system was evacuated, then closed off at a valve above V r. With both valves on the cell U 1 closed, chlorine trifluoride was distilled under its own vapor pres­ sure into the cell-filler T r, which could be cooled with tap water or with a Dry Ice-isopropanol mixture. The cooling jacket around T 1 was equipped with a drain through which excess coolant could be removed* When T* was full, the tank R 1 was closed and the liquid was introduced into U 1 - 34 - Figure 5* The cell-filling apparatus for chlorine trifluoride: R' steel storage cylinder} T' cell-filler with jacket for coolant; U' dielectricconstant cell k} V' connection to the gas-handling system* - yd - through the bottom valve* Through the dry-air intake of the gas-handling system, inert gas at a pressure of about 1 5 0 0 mm. Iig was applied to force the chlorine trifluoride from the cell-filler into the cell. The valve above T 1 could then be closed, and the cell and cell-filler could be removed and placed on the measuring instrument. The Hydrogen Fluoride Solution One solution of hydrogen fluoride in chlorine trifluoride was prepared by filling the gas-storage tank (G- in Figure l) with hydrogen fluoride at a pressure of 1000 mm. Hg and the condensing this gas into the cell-filler, using a Dry Ice-isopropanol mixture. The solution formed was intro­ duced into the cell in the usual manner. The concentration of hydrogen fluoride in this sample was very■uncertain for several reasons: the volume of tank G- was not definitely known, the quantity of chlorine trifluoride in T r could not be accurately measured, and the vapor' density of hydrogen fluoride, which varies considerably from ideal-gas behavior, is not well known in the pressure range employed. In addition, some non-condensable gas appeared to be present in the hydrogen fluoride sample. However, data on this mix­ ture were of some value in spite of the lack of information as to its composition, as will be seen in Section VI. - 36 - IV. FREEZING-POINT MEASUREMENTS Theoretical Aspects The Nature of the Measurement The vast majority of pure substances undergo the transi­ tion from the liquid to the solid state sharply at a definite temperature called the freezing point. This temperature is not entirely independent of pressure; the freezing point of a liquid under its own vapor pressure is called the triple point. The more commonly discussed freezing point of a liquid is that under approximately one atmosphere pressure; this latter point is the one measured in the present research. In addition, the freezing point differs somewhat with the amount of dissolved atmospheric gas in the liquid. The measurements In the present research were conducted with liquid which had been in prolonged contact with dry air, and which was presumably saturated with atmospheric gases. As dissolved impurities alter the freezing point of a liquid, and as no sample will be absolutely pure, the effect of a solute upon the freezing point must be considered. depression of the freezing point - 37 - The is related to the mole fraction Ns of the solute by the equation-^- A Tf =™ o 2 (16) I* where R is the gas constant, Tq is the freezing point of the pure solvent, and Lf is the molar heat of fusion of that solvent. If the mole fraction Ns is very small, as in the case of a small impurity, it may be shown that A Tf = Kf m (17) where m is the molal concentration of solute and Kf is defined by the equation K- = *frp2 (18) Lf x (lOOO/Mg) where Mq is the molecular weight of the solvent. As the solvent solidifies during the course of a freezing-point measurement, the concentration of solute increases, so that the freezing-point depression increases and the equilibrium temperature decreases. Figure 6 illus­ trates the shape of the cooling curveobtained. In the region B the sample has undercooledbeforesolidification begins, while along curve CDE the solvent is freezing and the solute is becoming more concentrotted. At approximately point E the solvent is entirely frozen. A derivation of Equations 16, 17, and 18 may be found in G-lasstone*s Textbook of Physical Chemistry (8 6 ; and in many elementary texts in this field. _ 38 - LU QC < cc LU Q. TIME Figure 6. A cooling curve in the region of the freezing point: Tf, freezing point of the samplej i11 , temperature at which half the original solvent has solidified. - 39 - The time x represents the period over which the heat of fusion was withdrawn from the sample. Along the line AB the heat leaving the sample did not directly correspond to the freezing process \ however, when the liquid returned to the equilibrium temperature along BG, the heat necessary to warm the sample again was produced by the freezing of a part of the solvent. Thus an extrapolation of the line CDS to A establishes the time and temperature at which fusion \\rould have begun in the absence of undercooling.-1* If the temperature of the surrounding bath is quite far removed from the sample temperature so that the temperature differential between bath and cell does not vary greatly, the rate of heat loss from the sample will be approximately constant. As almost all of this heat is produced by solidi­ fication of the solvent, the freezing rate will be quite constant during the time interval shown by x in Figure 6 . At time x/2 the material will be half solidified, so that the The actual rate of heat transfer from the sample to the surrounding bath is not exactly equal to the theoretical rate for the extrapolated curve, since the temperature differences between sample and bath are not identical for the two cases. However, if undercooling is slight, the error thus introduced will be small. A more elaborate method of determining points from slightly impure samples has been developed at the National Bureau of Standards (87)? the latter method is, however, difficult to apply to very small freezing-point depressions. - 40 - solute concentration at point B will be twice that at the point A. From Equation 17, the freezing-point depression at Dwill also be twice that at A, so that T0 _ Tf where Tj_/2 = 2(T0 _ T1/2) (19) ."the temperature corresponding to point D and Tf is that corresponding to point A. To *" Tf = Thus Tf - Tqy'p (20) and the true melting point of the pure solute may be deter­ mined directly from the cooling curve of a sample containing a small impurity. Temperature Measurement Using Thermocouples Many means of temperature measurement are available to the researcher; for the present research the use of coppercon stantan thermocouples was chosen since it appeared that these materials are resistant to attack by halogen fluorides, and thus could be inserted directly into the sample with no protective covering. The chance of temperature differential existing between the sample and the measuring device was thus greatly reduced. The use of a thermocouple for temperature measurement requires a well established reference temperature, since the electrica.l potential developed by the thermocouple depends equally upon the temperatures of the two junctions. The ice point, that is, the freezing temperature of air-saturated - 41 - water at atmospheric pressure, was used for this purpose in the present measurement. Every junction of dissimilar metals develops an elec­ trical potential dependent upon its temperature. Since only the potentials developed by the junctions in the sample and in the reference bath are of interest, it is important that the entire thermocouple circuit contain only these two junc­ tions ;of different metals. In the present study this was accomplished by using copper throughout the wiring system. Experimental Methods Temperature Measurement The measuring circuit. Thermocouple potentials were followed using both a precision potentiometer^ and an elec­ tronic recording potentiometer with a direct current a m p l i f i e r T w o thermocouples were used, one immersed in the sample and one in the bath surrounding the sample tube. The circuit shown in Figure 7 was so arranged that the output of either thermocouple could be connected to either measuring device. T A type K-2 potentiometer made by the Leeds and Northrup Company, Philadelphia, Pa., was employed. 2 A Leeds and Northrup type 9835B direct-current amplifier was used in conjunction with a Brora Electronic recorder made by the Minneapolis Honeywell Company, Minneapolis, Minn. - 42 - AMPLIFIER =N< RECORDER R.S POTENTI­ OMETER Figure 7. Temperature-measuring circuit: thermocouple inputsj R.S., switch Tor reversing recorder input. - 43 - T.I, the TABLE I RANGES AVAILABLE ON AMPLIFIER-REGORDER 0CABI:'ATION Amplifi er dial setting Voltage amplificationa r--- Recorder 10^::< volts inrl scalepn °C i n r ^ x 1 200 5 x 2 100 10 0.25 x 4 50 20 0.5 xlO 20 50 1.25 x20 10 100 2.5 x40 5 200 5 0.125 a The amplification given is that of the direct-current amp1i fi er only. These values are approximate and are calculated assuming that the thermocouple used produces 40 x 10"° volts per centigrade degree. _ Ad — The direct current amplifier had several amplification ranges, so that the 1 2 -millivolt scale of the recording potentiometer could be made to correspond to different temperature ranges. The available settings and temperature spans are summarized in Table I. Thermocouple construction. All thermocouples used were prepared from a single spool each of number 24 copper wire and number 24 constantan wire.-*- The junctions on thermo­ couples exposed to halogen fluorides were made by fusing together the ends of the wire. Other junctions were silver- soldered. Thermocouple calibration. Five thermocouples made from the same two spools of wire were compared with a platinum resistance thermometer. One of these thermocouples was calibrated in place on the plug assembly of the freezingpoint cell (see below) and had been exposed to iodine pentafluoride and bromine trifluoride during preliminary tests of the equipment. The calibration data for these couples are presented in Table II. The thermocouple wires used were purchased from the Leeds and Northrup Company. - 45 - The platinum resistance thermometer^- had been calibrated by the National Bureau of Standards in terms of the equation (2 1 ) where is the resistance of the thermometer at t°C. The values for the constants u and v were given as 0,00392604 and 1.4919 respectively. The Bureau advised that the resistance ^o at the ice point should be determined before each set of measurements\ this was done using the ice-point device des­ cribed below, and R0 was found to be 23*5031 ohms. The Mueller bridge employed to measure the resistance of the platinum resistance thermometer was also calibrated by the National Bureau of Standards. The calibration certificate stated that if the ten-ohm and one-ohm decades of the bridge \tfere not altered, resistance differences could be measured accurately to within + 0.02 per cent or 0.00005 ohms. Since the resistance change with temperature was approximately 0.1 ohm per centigrade degree, this limit of accuracy applied throughout the temperature range 0-10°G. The principal factor in limiting the accuracy of the calibration was the scale of the precision potentiometer, which could be read to approximately + 0.1 x 10”^ volts. For ^ Leeds and Northrup serial number 1016073 platinum resist­ ance thermometer was used with serial number 1011419 Mueller bridge, type G-l. - 46 - TABLE II DATA FOR CALIBRATION OF THERMOCOUPLES Rt x lo3, ohms °C Et x 106 Leas., Corr., volts volts Calc., volts A E X 10 6 , volts Thermocouple U 68.1 0.640 25.3 24.6 24.6 0.0 116.1 1.112 43.2 42.5 42.8 -0.1 817.7 8.016 311.3 310.6 310.4 +0.2 833.0 9.163 355.9 355.2 355 «3 -0.1 Thermocouple A 22.7 0.193 7.5 7.5 7.5 44.5 0.408 15.9 15*9 15.8 +0.1 117.1 1.122 43.3 43.3 43.2 +0.1 232.2 2.255 86.6 86.6 86.9 -0.3 396.9 3.878 149.3 149.3 149-5 -0.2 564.7 5-530 213.1 213.1 213.4 -0.3 701.8 6 .883 265.6 265 •6 265.8 -0.2 817.8 8.017 310.6 310.6 310.4 +0.2 932.0 9.153 355.1 355.1 354.9 +0.2 19.52 765.1 765.1 765.0 +0.1 32.84 1285.0 1285.0 1234.7 +0.3 - 47 - 'TABLE II 9 continued Rt x 103, t y, ohms 0G Et x Leas., volts 106 C o r r ., volts ■| Calc volt; Z* E x 10°, volts Thermocouple B 32.6 0.290 12.6 11.6 11.5 + 0.1 67.6 0.635 25.1 24.1 24.4 -0.3 124.1 1.191 47.1 46.1 45.8 +0.3 227.0 2.204 85-9 84.9 85.1 - 0.2 403.0 3.938 152.8 151.8 152.0 - 0.2 557.6 5.4-59 212.1 211.1 211.1 0.0 707.2 6.936 269 .1 268.1 268.4 -0.3 813.6 7.985 310.6 309 .6 309 .4 + 0.2 936.1 9.193 357.4 356.4 356.6 - 0.2 19.52 766.2 765-2 765-0 + 0.2 32.84 1285.4 1284.4 1284.0 +0. 4 0.0 Thermocouple C H. ON CO 0.846 34.1 32.6 32.6 118.3 1.133 45.2 43.7 43 *5 828.7 8.134 316.3 314.8 314.9 - 840.7 8.252 321.2 319.7 319.7 +0.0 + 0.2 0.1 Thermocouple D 75.5 0.713 28.1 27.2 27.4 123 •6 1.186 46.5 45.6 45.6 0.0 824.3 8.090 3 14 •1 315.2 313 .2 0.0 837-1 8.217 319.1 318.2 318.1 - 48 - - 0.2 + 0.1 the ‘thermocouples employed this created an uncertainty of about + 0.0025°C. The data in Table I were fitted to equations of the form Et = E0 + c t + d t2 (22) where E-^ was the potential of the thermocouple with the measuring junction at the temperature t, E0 was the potential of that thermocouple at 0°C, and c and d were constants for the entire set of thermocouples prepared from the same supply of wire. As the V3,lue of d was relatively small, a value was taken which would correspond to the data in published thermo­ couple tables (88, 89)- Values of E0 were determined by placing the two junctions of the thermocouple close together in a tube immersed in an ice bath. The value of the constant was chosen to best correspond with the calibration data In Table II. Substituting the assigned values for c and d, Et = E0 + 58.39 t + 0.042 t2, (23) the values of EQ for the various couples are given in Table III. The fourth column of Table II contains the value of each experimental potential corrected for the ice-point potential E0, while the fifth column contains the potential calculated for each tempera1 ture, assuming E0 to be aero. The deviation of the experimental value from Equation 2p is listed in the final column; the root-mean-square deviation is 0.21 x 10-6 volts, which corresponds to approximately 0.005°C. - 49 - TABLE III ICE-POINT POTENTIALS FOR THERMOCOUPLES Thermo coup1e E0 x 106, The ice-point potential E0 was found to consist of two portions, one a characteristic of the thermocouple itself and one arising from the leads and junctions from the couple to the measuring equipment. As the latter effect could not be completely eliminated or accurately predicted, it was neces­ sary to determine E0 for each set of measurements conducted with this equipment. Although the procedure described above has been called a calibration of the thermocouple, it actually constituted a calibration of the entix°e temperature-measuring apparatus, since the same precision potentiometer, galvanometer, and standard cell were employed throughout all the calibration and tempera'ture measurement described in the present section. Thermocouples A and B were calibrated, also, at two higher temperatures, 19.52 and 3>2.84°C. These data were obtained using the transitions NaCr04 •10Hg0 = NaCr04 -7H80 + 3HS0 NaS04 *10HS0 = NaS04 -7H80 + 3HS0. and The thermocouple was placed in a Dewar flask with a quantity of solid decahydrate and a saturated solution at a tempera­ ture slightly below the transition tempera.ture. To this was added a quantity of solid heptahydrate in equilibrium with a saturated solution slightly above the transition temperature. The two hydrates came to equilibrium immediately, and no - 51 - change in ■thermocouple potentia.l was observed over a period of an hour• The data thus obtained on these thermocouples are included in Table II. Calibration of these thermocouples above 10°C was not necessary for the purposes of the present section but was desirable for work to be described later. Equation 23 appears to be reliable to 32°C within + C.01°C. The ice-point reference. Figure 8 illustrates the ice- point tube employed to keep the reference junction of the thermocouple at 0°C; this is a modification of a device designed by White (90)* The thermocouple was sealed into the center tube, as shown, with paraffin wax. The device was then filled with finely crushed ice and placed in the Dewar flask, which was in. turn filled with crushed ice. Purified water-1-, chilled- to approximately 0°C, was then admitted to the tube and to the Dewar, and the device was allowed to stand for at least one hour. During this period, any impurity adsorbed, on or occluded in the surface of the ice particles will cause the Ice In that immediate region to melt, since the slight Impurity will lower the equilibrium temperature between solid and liquid phases. The water was then removed from the 1 Water from the department distilled water supply was run through a mixed-bed ion-exchange column. An indication of the purity of this water is given by its conductivity, 1.5 x 10-8 ohm”1 cm”1, determined as described in Section V below. - 52 - Figure 8. Ice-point tube in place within a Dewar flask: The thermocouple is sealed into the center tube with paraffin. center chamber by suction and replaced by a clean sample of chilled purified water. White claimed that this device would maintain a tempera­ ture of 0.0000 + 0.0001°C for up to 24 hours, even if a com­ mercial grade of ice was employed. In the present research, distilled water was frozen to produce the ice used. A comparison of two such devices using a thermocouple showed that the temperature differential did not change detectably (+ 0 .0 0 2 5 °C) over a period of fifteen hours. The Cell and Cooling Bath General arrangement. Figure 9 is a schema.tic represen­ tation of the relative positions of the cell and the various elements of the cooling bath. The large outer bath had in­ sulated walls and was cooled by means of a refrigerating machine, the coils of which are shown at the left in the figure. This bath was filled with 95 P©f cent ethanol. The machine was controlled by means of a bimetal thermoregulator while circulation in the outer bath was maintained by a stirrer with a five-inch propeller and by a pump. During the control cycle the temperature in this bath varied over a range of about 0.2°C. The inner bath, which was the actual cooling bath for the freezing-point cell, consisted of a four-liter beaker filled with 95 per cent ethanol. - 54 - The outer surface of the Figure 9* The cooling bath: left, cooling coils of the refrigerating machine; center, inner bath containing the cell, thermistor probe, and heater; right, bimetallic thermo­ regulator controlling the refrigerating machine, A pump employed to maintain circulation in the inner bath and a pump and a propeller-type stirrer used in the outer bath are not shown. - 55 - beaker was wrapped with, tape to reduce the rate of heat transfer to the outer bath. Within the inner bath were the cell., a heater, and a thermistor probe which was the temperature-sensitive element of the thermoregulator system for this bath. Circulation in this inner bath was maintained by means of a pump mounted outside the bath. .The thermoregulator. The circuit shown in Figure 10 is an adaptation of a, thermoregulator designed by Burwell, Peterson, and Kathmann (91)* The temperature-sensitive element was a thermistor^- in a Wheatstone bridge circuit, while the heater current was controlled by means of a satu­ rable reactor in series with the knife heater. With this regulator the temperature of the inner bath could be con­ trolled to within the limit of measurement by means of a copper-constantan thermocouple and the precision potentiometer, which was about + 0 .0 0 2 5 °C. The freezing-point cell. the freezing-point cell. Figure 11 is a photograph of The cell body was a Fluorothene test tube; the stirrer was of Fluorothene while the plug assembly was constructed of Teflon. The thermocouple wires entered through the plug and were wound about the tube which 1 A type 14B thermistor, made by the Western Electric Company, New York, N. Y., was used. - 56 - 10 90V .0 5 0 .0 5 6SJ7 10K AAAo °.5< V T^ < o.0 5 < 6SJV ji■ ^ 0.25 0.25 5K TLO Fil. 6AG 7 80 5K AG ; To Heater Figure 10. Thermistor thermoremulator circuit: T, thermistor; S.R., saturable reactor. The saturable reactor circuit shown within the box was placed in a sepa­ rate metal cabinet. Sizes of capacitors are giyen in microfaradsj resistors are given in megohms (10° ohms) unless otherwise indicated^ K indicates thousands of ohms. - 57 - surrounded the stirrer shaft} grooves for the wires were inscribed in the tube. Preparation of Samples The first sample of iodine pentafluoride employed was distilled directly into the test tube in which measurements were made. The second was stored in the storage beaker (Section III above) and transferred to the test tube by means of the solution-mixing system. tained hydrogen fluoride as a solute. The third sample con­ Test tube S of the mixing system (Section III) was filled with hydrogen fluoride gas at 730 m i * Hg and 24.2°C. This was condensed into the freezing point tube, in which had been pls.ced 201.1 grams of iodine pentafluoride. The hydrogen fluoride sample contained 0.00349 moles so that the resulting solution was 0.0174 molal in this solute. Results Data and Calculations The iodine -pentafluoride samples. The cooling curves for the two iodine pentafluoride samples are shown in 1 The calculation of the quantity of hydrogen fluoride used is discussed in Section V below, where the preparation of a number of such samples is described. - 59 - Figures 12 and 13* The readings of the precision potenti­ ometer were recorded directly on the recorder chart. The potentials corresponding to the points A and D in Figure 6 were estimated from these figures and are listed in Table IV. The potential E-f* at the freezing point of the pure liquid was obtained using Equation 23? the values of E q were determined as described above (Experimental Methods), and the freezing temperatures of the pure liquid calculated using Equation 23• These data are summarized in Table IV. The hydrogen fluoride solution. The coolfng curve for the sample containing hydrogen fluoride in iodine penta­ fluoride is reproduced in Figure 14. In this ca.se, the locs/tion of the point corresponding to D in Figure I was difficult to estimate. However, the potential at the point A could be determined, and the freezing point of the solution calculated; these data are included in Table IV. Errors. The error resulting from the calibration curve for the thermocouples, Equation 23, has been estimated above as 0.005°C. This, then, is the uncertainty of any individual point on the figures in the present section. Three such points, A, D, and the tempera.ture corresponding to E0 , entered into the calculation of the freezing point of the pure liquid. There is a slight additional error due to the uncertainty in locating the point D; this uncertainty would - 60 - 370 S110A Aoi * 360 3 TIME Figure 12. pentafluoride. Cooling curve for the first sample of iodine - 61 - 370 Ex 107 VOLTS x/2 360 TIME Figure 13. pentafluoride. Cooling curve for the second sample of iodine - 62 - E x 10° VOLTS 360 350 TIME Figure 14-. Cooling curve for the solution of hydrogen fluoride in iodine pentafluoride. - 63 - TABLE IV ED x 106 ea x 106 EQ x 106 ! j I /7"Tf °G °G °G volts volts IFe 362.2 363.1 1.3 9o98 9*420 0.022 IF 5 362.1 365.0 9 .408 9 .430 0.022 HF In IFb 361.5 r~\ volts CO • • 0 1 Tf 0 I l FREEZING-POINT DATA FOR IODINE PENTAFLUORIDE SAMPLES - 64 - 9ol5 0.11 appear, from the figures, "to correspond to less “than 0.2 x 10 ^ volts or 0.005°C, The root-mean-square e: -mean-square error-*- from these four sources is 0.01°C. This value was therefore taken as the uncertainty in the estimated freezing points} the values for the two samples agree to within this limit. The temperature corresponding to the point A, which is the freezing point of the sample employed, depends only upon E0 and the potential at A, so that its root-mean-square error is 0.007°C. The freezing-point depression of the sample is estimated from points A and D and contains three sources of uncertainty, so that the root-mean-square error for this value is 0.008°C. Discussion of the Results The freezing point of iodine pentafluoride as deter­ mined in the present study, 9*425 ± 0.01°C, is somewhat lower than Rufffs value (11), 9-6°C. As the early studies with halogen fluorides were complicated by lack of methods **■ This was calculated using; the equation E -where E is the over-all relative error and Eq, E2 , etc., are the various relative uncertainties. This and all other statistical equations used in treatment of data in this thesis are from Treatment of Experimental Data, by Worthing and G-effner (1007« - 65 - and materials for purifying and handling these compounds, this discrepancy is perhaps to be expected# The molal concentration of dissolved materials in these samples may be estimated using Equations 17 and 18. The molal cryoscopic coefficient Kf, calculated using 283° C for Tq and 3*80 kcal. mole"”-*- for the heat of fusion is 9*2°C, so that the freezing-point lowering observed for the purer samples, 0,022 + 008°C, corresponds to a molal concentration of impurities of approximately 0 ,0 0 2 5 * As these samples were taken from the same supply as those employed in the following sections, and as the methods of handling the samples were quite similar, the above information as to the purity of the iodine pentafluoride is of considerable value in considering the reliability of results in the later sections of this thesis# The freezing-point depression observed for the sample to which hydrogen fluoride had been added intentionally was 0.11°G# Thus the additions,! depression due to the hydrogen fluoride was 0.09°G, corresponding to an 0.010 molal concentration of this solute. tion data this sample was 0.0174- molal. From the prepara­ A possible explanation of this discrepancy is that the highly polar hydrogen fluoride molecules tend to associate with one another when dissolved in iodine pentafluoride, so that the - 66 - effective number of solute molecules is lower than expected, A similar association of hydrogen fluoride in water is quite well established (101). - 67 - V. CONDUCTANCE MEASUREMENTS Theoretical Aspects The Nature of the Measurement The current arising when an alternating electric field is applied to a liquid mayhe considered as a sum of two components, one due to the migration of charged particles or ions and one due to the orientation or relocation of charge distribution in neutral molecules. The ionic migration cur­ rent is greatest when the applied potential is at its peakj this current is therefore conductive^ that is, it is in phase with the applied potential, as shown in Figure 15* The con­ ductance of the liquid andits low-frequency measurement are the subject of the present section. The conductanceis of interest because it gives information concerning the pres­ ence and nature of ionic species in a liquid or solution. The second current component, due to the orientation of molecular dipoles and polarization or induced charge separa­ tion in molecules, is not conductive at audio or normal radio frequencies. No net long-range migration can occur, and the redistribution of charge on the molecular scale can easily take place during a very small part of a cycle. - 68 - The amount a Figure 15* Phase relationships betv/een an applied potential and the components of the resultant current: a, applied potential\ b, — conductive current, — — — susceptive current. - 69 - of* charge displacement will depend upon the applied potential, and the maximum displacement will occur at the time of the potential peak. The maximum change in charge location, and thus the maximum current, will take place at the time of maximum change in applied potential. This susceptive current manifests itself in the dielectric constant, and is more important and more easily measured at higher frequencies; it will he considered in Section VI. Conductance Units and Cell Constants In a uniform electric field the conductive current i varies directly with the potential gradient E/y and with the cross-sectional area A of the conductor: i = . (24) The conductance G of the solution is then given by q* — — E — j ~ — a. . Kc (25) The constant k is the specific conductance of the solution and is commonly expressed in ohm"’*- cm"*1*. ing apparatus the value stant. For a given measur­ i/A ~ K c is known as the cell con­ It is generally determined by measuring G after filling the cell with a sample of known specific conductance. Electrode effects and their treatment At frequencies of the general magnitude of 1000 cycles a portion of the measured impedance of a conductance cell does - 70 - not represent the behavior of the solution itself, but results from a process occurring at the electrodes. This additional impedance has both a resistive and a reactive component (92), and the latter does not result from the dielectric constant of the sample. In Figure 16, 1/G corres­ ponds to the resistance of the solution proper while R@ and X e represent the electrode impedance. Z is then the total mea sur ed imp edan ce . In most present-day measurements these effects are minimized by using platinized platinum electrodes. The large surface area of the finely divided metal reduces the elec­ trode reactances to negligible magnitude. This alternative was not available in the present work, however, as the plati­ nized surfaces would be subject to attack by the liquids involved. Two other methods for treating electrode impedances have been developed (92). First, the electrode separation may be varied so that l/G is changed while Re and X e remain constant. This involved complications in cell design and construction, especially considering the nature of the samples to be examined. Fortunately, Re and Xe have been found to vary approximately Inversely as the square root of the frequency. Thus by measuring 1/G + Re at a series of frequencies and plotting this value against l//f one may extrapolate to - 71 - '/G R. Figure 16. The relationship between the total impedance of the conductance cell and the various con­ stituent impedances: Z , total measured Impedance} 1/G-, resistance due to the sample} R( and Xe, components of the electrode impedance. x cr Figure 17. Graph of - 72 - versus l//f. infinite frequency and obtain 1/G-. This procedure was employed in the present research. This extrapolation introduces some error in the final value, since the inverse square root relationship is only an approximation (92). uncertainty: Two procedures minimize this cause of the use of higher frequencies and the use of cells having high constants. A limit upon frequencies is established by the bridge employed and by the nature of the leads to the cell} at some point lead inductances and dis­ tributed capacitances begin to affect the results obtained. With the apparatus and leads used in this research, 4000 cycles appeared to be the limit at which accuracy could be obtained. The second alternative, using cells having high constants so that 1/Gr is large compared to Re, also has its limitations. There is always some conductance across a cell, even when empty, due to the nature of the leads, the cell body, and the temperature bath.. As shown in the following section, this conductance in the apparatus used was several hundredths of a micromho, causing errors of 0.1 per cent or so in measure­ ments of unknown, conductances of the order 10"^ ohm"-3-. Measuring conductances of this magnitude involves other problems. High-impedance circuits tend to pick up stray radiation such as the ever-present 60 cycle power-line signal. Capacitances of lead wires also become more emportant. When possible, therefore, the unknown conductance was kept between - 73 - 10“3 and 10“5 ohm"-*- during this research. For this purpose three cells having different constants were constructed for use with halogen fluorides. The Instrument The Campbell-Shackelton ratio box with resistance and capacitance standards employed in this measurement was con­ structed at Michigan State College. For the purpose of the present section this bridge may be represented as in Figure 18. This bridge is at balance when the series unknown, resistance and capacitance and Cx are equivalent to the parallel combination of Rs and Cs. In complex circuit notation (93) this requires that ** + where lj JCJ0X 1 s + jCjCg (26) is the angular frequency 2 nf, and j represents the square root of minus one. o j ^x~r)C From this we obtain — Rs ~ p ? p ^ ^ 1 + GJ Og Rg (27) and separating real and Imaginary components: ^ ~ 1 r| (28) + cjOg. (29) / u C x = — _ These are the two balance conditions for this bridge* - 74 - Figure 18* Simplified circuit of the conductance bridge: R-j_ and Rp, isl ratio resistors (R^ = 'Rp); R Q and Gg, resistance ana capacitance standard decades, ^and Cx , components of the unknown impedance; ? detec tor } /-\_j t signal generator. - 75 - From Equation 28 we see that Rx cannot be read directly from the standard resistor Rs u n l e s s small. Further, it is necessary to know the frequency in order to calculate Rx . Thus the signal generator must produce several stable frequencies. The presence of cj in both balance equations is of some practical importance in the design of the apparatus. A Wheatstone bridge is a null instrument in that, at balance, the detector receives no signal. However, in the above bridge, balance for a given frequency is not balance for other frequencies that may be introduced into the bridge circuit. Most signal generators produce, in addition to the nominal output frequency, small signals at harmonics (multiples) of that frequency, and these will not be balanced out with the fundamental. These harmonic signals can easily obscure the null point so that the operator cannot determine exact balance. This difficulty may be met by keeping the bridge-input signal as clean as possible and by using a detection device that is more sensitive to the frequency employed than to its harmonics. Both methods were used in this work. employed will be discussed in the next section - 76 - The circuits Experimental Methods The Instrument Over-all circuit. The conductance-measuring instrument employed in this research consisted of an audio-frequency generator, a selective detector and oscilloscope, and the bridge proper, including ratio box and standard resistors and capacitors. The general relationship between these components is shown in Figure 19. As the balance was detected on the oscilloscope by means of Lissajous figures (94), a reference signal was needed for the horizontal deflection. By use of these figures, both the magnitude and the relative phase of the bridge output could be observed, so that one could tell whether a change in Rg or in Cs was needed. Generator, detector, and bridge proper were mounted on standard relayrack panels in an enclosed cabinet, as shown in Figure 21. To avoid the 60 cycle signal at the bridge output it was found necessary to place the power supplies for generator and detector outside the cabinet. The ratio box. The nucleus of a conventional conductance box is the ratio box, containing all components of the bridge proper except the unknown and the standard resistors and capacitors with which it is compared. Behr and Williams (95) have analyzed the problems involved in ratio-box construction and have discussed thoroughly the manner in which these - 77 - ObCIhLObGGPE AUDIO RIDCK GENERATOR Fipure 19. Relationship of the components of the oonductance-ineasuring Instrumen t . R U T tf X AA/^— —a a a a L— a a Aa R4 P Fisure 20. The ratio box: Rp and Rp, ratio resis­ tors } Rp, R^, and R-, auxiliary resistors: Cp and Cp, auxiliary capacitors; S ? connections for standard resis­ tors and capacitors; X, connections for unknown. - 78 - 3 '.WKf problems are met in the design of the Campbell-Shackelton shielded ratio box. The ratio arms (R-^ and R2 in Figures 18 and 20) must be made as nearly alike as possible. Since capacitances between the parts of these resistors and the surroundings cannot be entirely avoided, these resistors must be placed in as nearly identical environments as possible and the distributed capacitances must be."tied down” to a definite location. In the Campbell-Shackelton circuit this is accomplished by plac­ ing the two resistors in similar locations within a double shield. The arrangement of these components In the box employed in this research is shown in Figure 22. The arms containing the unknown and the standard resis­ tors and capacitors cannot be made identical in physical environment, and thus some means of treating the various lead capacitances must be adopted. This may be accomplished by adding auxiliary variable capacitances and conductances to these arms of the bridge. The unknown and the standard com­ ponents may then be removed, leaving all leads and connections in place. The bridge may be balanced by varying the auxiliary components, thus compensating for any lead imped­ ances. The unknown and the standard components may then be compared by connecting them into the bridge circuit and obtaining a balance by means of the standard decade caxoacitors and resistors. - 80 - Figure 22* The ratio resis­ tors in place within a double shield. The covers of both parts of the shield have been removed# The resistance and capacitance standards. The decade resistors and capacitors represented by u to and C„s in Figures 18 and 20 were mounted on the same panel as the ratio box. Figures 23 and 24 are photographs of the back of this panel, showing the relative positions of the. various compo­ nents and the shields separating them. The six resistance , decades*1- were connected in series and had nominal values of 0.1, 1, 10, 100, 1000, and 10,000 ohms per step. The stated accuracy of these resistors was + 0 . 0 5 per cent of the nominal value. The three capacitance decades were connected in parallel and had nominal values of 0.1, 0.01, and 0.001 microfarads per step; their stated accuracy was + 0.1 per cent. Sip;nal venerator. This audio-frequency source was con­ structed from a circuit developed by the Heath Company^, and employed a Wien-bridge oscillator. Several changes from the original circuit were introduced, as shown in the circuit for the new generator, Figure 25. The Heath instrument employed a variable capacitor in the Wien-bridge oscillator; this was replaced by a switch and eight sets of fixed paper or mica The resistance decades used were "type 510, made by the General Radio Company, Cambridge, Mass. The capacitance decades were General Radio type 380. 2 This circuit was employed in the Heathkit audio generator model AG 7, made by the Heath Company, Benton Harbor, Michigan, and is described in the instruction manual for that kit. Two more recent instruments have replaced this generator; neither employs this circuit. - 82 - § .h 4.5K B+ < 1 Power Supply Fil. Fi.rpj.re 25* Circuit of the audio-frequency sipnal generator; The sizes of capacitors are h v e n in microfarad! resistors are given in megohms (10 ° oliras) unless otherwise indicated} K indicates thousands of oliras • capacitors, thus providing definite fixed frequencies. The potentiometer in the grid circuit of the first 6SN7 stage was the main volume control in the original instrument; this was placed on the hack of the chassis in the present circuit, since its use influenced the oscillator frequency slightly. The 6J6 and 6SL7 stages were not in the original circuit. The 6J6 cathode-follower output allowed further isolation of the oscillator from the bridge and provided a low output impedance, needed for the bridge input transformer. The grid of the second 6SN7 stage, originally grounded, was employed in a feedback loop from the cathode of the output stage, thus lowering the distortion produced in the power-delivering cir­ cuit. Volume was controlled by varying the amount of feedback by means of a potentiometer. The 6SL7 dual triode provided one stage of voltage amplification and a phase-shifting network for the reference signal, to be fed directly to the detector. The nominal sizes of the capacitors used in the oscil­ lator circuit and the frequencies obtained are given in Table V. The frequencies were determined by comparison with a continuously variable audio-frequency generator by means of Lissajous figures (94) on the oscilloscope of the detector described below. The scale of the reference generator was compared in the same way with standard audio frequencies broadcast by radio station WOT. - 86 - TABLE V AUDIO OSCILLATOR CAPACITANCES AND FREQUENCIES G, yW/C'-f f, cycles .0004 8550 .001 3820 .0025 1950 .005 980 .01 42 0 .025 200 - O Q' l~ 7 ' - Detector* The detector for this bridge was a three-inch oscilloscope with one stage of vertical and three stages of horizontal amplification. The oscilloscope had the usual controls for intensity, focus, and horizontal and vertical centering. The horizontal amplifier was a simple 6SJ7 stage with a potentiometer in the grid circuit providing a vertical amplifier control. The detector circuit is given in Figure 26. The ..vertical detector: it was amplifier is the unusual feature of this made frequency-selective by placing a parallel-T filter circuit between the cathode of the last 6SJ7 stage and the grid of the second. This provided an inverse-feedback loop, and since the parallel-T filter trans­ mitted all frequencies but the one to which it was tuned, only the selected frequency was fully amplified by the second stage. The frequency response of this amplifier on the 400 cycle frequency setting is indicated in Table VI and Figure 27. The filter circuit is shown in Figure 28. The frequency was changed by changing condensers} a fine frequency adjustment was provided by the small variable components in the resistance arms. At each frequency the values of the resistance R and the capacitance G in the filter were the same as the sizes of the components in the oscillator. Volume control was provided by a step attenuator onthe input of stage one, with fine adjustment by a potentiometer in the oscilloscope input circuit. - 88 - The cathode bypass lOhy 20 -OB 0• 0 0 9 o.l 63 J7 6SJ' 0.33 es^ T-T 2 . 21 560 < 0.22 0 '1 680 0.15 0.27> ,OTPT >0.05 L_ II * IK 28K 8.2 2X2 Figure 26. Circuit of the oscilloscope detector: The sizes of capacitors are giver in microfarads; resistors are giver in megohms (10° ohms) unless otherwise indicatedJ K indicates thousands of ohms. TABLE VI VERTICAL AMPLIFIER FREQUENCY Kn,oP0hSE Oh 400-CYCLE LET Til!G Output, db . cycles -22.7 200 1 250 -15.0 300 - 7.6 350 - 1.6 375 0.0 400 0.0 0 • 150 0 -23 .2 - 1.1 500 600 -14.2 700 800 -19.8 1000 -21.4 1250 -22.3 1500 -22.5 0 cn • 450 1 CM 1 O 425 1 H A1 • \Jl CM 125 *v The output in decibels was related to a 50volt signal at the oscilloscope deflection plates at 400 cycles. - 90 - DB. OUTPUT, -10 -20 400 200 f, 600 1000 CYCLES Figure 2 7 • Frequency response curve for the vertical amplifier on the 400-cycle setting: see Table VI. 2nf = Figure 28. Parallel-! filter circuit. - 91 - condenser on stage one was removable by means of a switch., so that the signal might be improved by cathode degeneration in this stage unless very high gain was desired. Trial of the bridge. To try out the bridge and. asso­ ciated equipment, a series of conductance measurements were carried out using potassium chloride solutions and a glass cell with platinized platinum electrodes. Three of the potassium chloride solutions were ma.de up gravimetrically according to the instructions of Jones and Bradshaw (96) who accurately measured the conductances of solutions of these concentrations. Conductances were run at two fre­ quencies, so that any electrode polarization effects would be detected. The data on the mixing of these solutions are given in Table VII and the conductance data and cell-constant calculations are listed in Table VIII. During measurements the cell was immersed in a bath of transformer oil, the temperature of which was controlled by the saturable-reactor regulator described above (Section IV). Bath and regulator are shown in Figure 29• Temperatures were measured using copper-constantan thermocouple B, described in Section III; corrections for deviation from the standard temperature, 2 9 .00°G, were made assuming a temperature co­ efficient of conductivity of 1.93 por cent per degree centi­ grade. Corrections were also applied for small errors in making up the solutions and for the conductivity of the water used. - 92 - TABLE VII SOLUTIONS PREPARED TO STANDARDIZE THE GLASS CONDUCTANCE CELL Gone (nominal) Df Weights Hs°> KOI, grams grams i ----- Concentration-----—i uncorr. corr.2 theor.1 exact,1 [grams KC1 per 1000 grams) D! 2 0 0 .1 0 1.4827 7.4098 7.406 7.41913 0.01(A) 199.67 0.1491 O .7 4 6 5 0.7462 0 .7 4 5 2 6 3 0.010014 0.01(B) 201.27 O .1 5 1 3 0.7516 0.7513 0.745263 0.010080 6 ,1 0 .0 9 9 8 1 ^ Nominal concentrations are given in moles of solute per cubic decimeter of solution* This "denial” unit is the definition used by Jones and Bradshaw. However, solutions were made up according to specific directions by the above authors, rather than according to- the definition. 2 Concentra-tions were corrected for the buoyancy of air during weighings using the eguation Wt.t ILn QTJ,O = . -LIT cllx^ - 93 - (1 + KC1 H3 O )• TABLE VIII CALIBRATION DATA FOR THE BLABS CONDUCTANCE CELL Solution Cone., D* Et x 1 0 6, volts rl9 5 0 cycles 1 ^s> Gs> ohms f t, °c r~ 9 8 0 cycles-y Rs’ °s> ohms f 0.09981 988.4 25*05 69-2 4 69 .2 4 0.0100l4a 988.7 25 .0 6 627.2 3 627.2 2 988.8 25*06 627.1 3 627.1 2 989.5 2 5 *08 624.0 3 624.0 2 987-8 25.04 624.1 3 624.1 2 988.7 25*06 670,000 G x 106 Df ohm--'- 0.010014 0.010080 14,450 2 cm. -0.10 0.888 r}f +0 . 1 9 14,473 12,856 K c 1,586.9 1,408.8 0.8877 1,587-3 1,408.8 0.8875 1,585-9 1,408.8 0.8884 1,586.8 1,408.8 0.8881 -0.14 1,592.8 -0.11 -0.14 1,602.6 -0 . 1 5 1,602.4 -0 . 0 7 1,592.5 680,000 rCorrections —1 0 r a corr.c kx 106 , for t, for D, x 10° ohrn”^of "b ohmcrn.”-*• Jo /O H 0.09981 3 CO 0 Solution Cone., 1 0 • \ Water 1 0 • 0.010080a 0 * 1 CO 0 liea n , last four values 0.0006 a Each of these solutions was measured twice } the cell was emptied and refilled between runs. D Assuming a temperature coefficient of conductance of 1*93 P©** cent per centigrade degree. 0 Conductances were corrected for temperature and concentra­ tion, and for the conductivity of the water used, 1.5 x 10"° ohm*“-k ^ k is given as determined by Jones and Bradshaw (see text). - 94 - hath Figure 29* The conductance and thermistor thermoregulator In addition to the above, two other solutions of potas­ sium chloride were prepared£ these are described in Tables IX and X and their molar conductances, along with those of the solution described above, were plotted against the square root of the concentration. Such a plot should give a straight line (97) and its intercept should be the conduct­ ance at infinite dilution. The value obtained from Figure 30, 149*9 ohm"-*- mole"-1-, may be compared with Shedlovskyfs value, 149*82 ohm”-1- mole"^- (98). The Conductance Cell Cell desiren * The conductance cell for use with halogen fluorides is illustrated in Figure 31* The electrodes were tapered nickel plugs which fit into a cylindrical Teflon liner inside an aluminum cell body. Each plug was held in place by an aluminum end piece which screws into the cell body. One of the plugs was insulated from the cell body by a Teflon washer* This plug was connected by the center con­ ductor* of a coaxial line to one "unknown” terminal on the bridge. The other plug was In electrical contact with the cell body, which acted as an electrical shield for the cell. This plug was connected by the outer conductor of the coaxial line to the ground "unknown” terminal. The Teflon cell liner was removable and could be replaced by one of different dimensions. - 96 - Three liners were constructed TABLE IX DATA ON THE COMPOSITION OF SEVERAL POTASSIUM CHLORIDE SOLUTIONS li20, grams KC1, grains ------ Concentration---- 1 | - i — Weights,—| Grams KC1 per 1000 g* H20 N x 10"^ Uncorrected Corrected 10 00.5 0.3217 0.3 21 5 0.3214 4.297 1001.2 0.1518 0.1517 0.1516 2.026 7-406 99-34 0.7462 10.01 Solutions listed In Table VIII - 97 - TABLE X CONDUCTANCE DATA ON SEVERAL POTASSIUM CHLORIDE SOLUTIONS Et x 1 0 6, Temp. corr. °c % 987.5 2 5 .0 5 2.026 987-5 2 5 .0 5 99-34 987.4 25 .0 5 10.01 987.7. 2 5 .0 6 N x 10-3 /¥ R 2 c 2 °S) ohms ohms 1429 5 -0.10 3788 2 69.2 4 ' -0 . 1 1 627.2 5 • I 0 H • H O 4.297 1 0 volts t, 0 N x 10-3 G x l O 6, 0 ucorr. 3 ohm-^ x 106 , ohm"*** ohm"-*-cm"-*- K x 106 4.297 0.0655 700 6 9 8 .0 619.5 2.026 0.0450 264.0 262.2 295.0 99.34 0.515 10.0. 0.1000 A ohm~lcm~l 145.8 14,450 14,455 12,800 1 2 9 .0 1,592 1,589 1,410 140.9 See note 2, Table VIII. 2 R o and C0 S were measured at 1950 cycles. 5 Conductances v/ere corrected for temperature and for the conductivity of the water used, 1.5 x 1 0 "6 ohms-1. (See Table VIII. - 98 - 150 130 0.00 0.2 0.1 /c " Figure 3 0 . A versus Ac chloride solutions. for a series of potassiiim - 99 - 0-1 I— 2— 3 —1 SCALE, IN. Fipure 31* Section t h r O U p h the upper and center portions of the conductance cell: A, a lumi rum on d ;iocej B, Teflon v;usher; C, aluminum cell body; D, Teflon cell liner; E, nichel plup connected to valve (not shown). The lower end and plus assembly was identical with the one snown above excepo onat washer B was metal to px^ovide eleccrical connect between tne cell body and the plus. - 100 - for work with, halogen fluorides, so that three different cell constants were available. Only one of these was used in the course of the present study. The liquid samples were introduced by means of tubes through the middle of the plugs, so as to minimize the number of Teflon-to-metal seals required* tubes as shown. Valves were placed on the The technique of filling this cell with halogen fluoride samples is described in Section III above. Cell calibration. Two potassium chloride solutions were prepared for the purpose of calibrating this cell. The con­ ductance of each solution was measured twice using this cell and. twice using the cell described in the preceding section; from these data and the cell constant of the glass cell, the constant of the new cell could be determined. The oil of the bath in which the Teflon cell was im­ mersed was apparently slightly conducting; a slight decrease in the measured conductance occurred immediately upon lifting the cell from the solution. Unfortunately, this conductance was too small for measurement on the bridge. The empty cell was, therefore, immersed in the bath and placed in the cir­ cuit shown in Figure 33• A signal from the generator was applied to the cell and the comparison resistor R, and the voltages at points a and b were determined with an alternatingcurrent vacuum-tube voltmeter. - 102 - The results of two separate GEN VTVM Figure 33* Circuit for determination of the residual conductance of the conductance cell: left, audio generator} center, cell in series with a known resistor Rj right, vacuum tube voltmeter with probe to be placed in contact with point a or b. - 103 - determinations are recorded in Table XI. The conductance of the cell can be determined using the equation i = Ea Rr + R — Eb R (30) where Rr is the resistance corresponding to the small residual conductance of the cell* From Table XI, R^^ is approximately 25 megohms or 25 x 10® ohms, and G-r is 4.0 x 10“® ohm“-*-. This value represented a conductance too small to cause a significant error in any of the measured conductances described in this section. The results of the measurements for the two potassium chloride solutions in the Teflon cell are summarized in Table XII, and the graphs of Rx versus 1/V? (see 'Theoretical Aspects above) are shown in Figure 34. The resistances in this cell and in the glass cell are given in Table XIII. Since each solution has the same specific conductance in either cell, from Equation 25 k rr a Kc = G-' K c ' (3 1 ) where G and G* are the conductances measured in cells having cell constants Kc and Kc 1, respectively. Preparation of Samples Procedure . The' techniques employed in preparing the samples of iodine pentafluoride and the solutions of other materials in iodine pentafluoride have been described above - 104 - TABLE XI RESIDUAL CONDUCTANCE OF THE CELL IMMERSED IN THE OIL BATH Va , Vb xlo3, I^xlO"6 volts volts ohms f rr 1000 cycles, R rr 1000 ohms 27.5 1.1 25 52.4 2.1 25 73.6 3-1 25 f ~ 400 cycles, R = 10,000 ohms 2 5 .2 9 .6 26 5 0 .8 1 9 .7 26 75-0 29.7 25 - 105 - TABLE XII CALIBRATION DATA FOR THE TEFLON CONDUCTANCE GELL :ycles Bridf;e readings GS> ohms am f Rx 5 ohms Solution A, first sample 980 518 3500 518 1950 512 1500 512 3820 509 1100 509 3600 514 1930 1500 508 3820 505 1100 505 C\ 0 CO 514 Ul 0 CO Solution A, second sample Solution 8, first sample 490 9820 1350 9810 980 956 O 689 9550 1950 9360 343 935° 3320 9270 158 9260 Solution B, second sample 490 9740 1370 9730 980 9480 693 9470 1930 9290 346 9280 3820 9170 161 9160 10,000 9,500 OHMS 9 ,0 0 0 8 ,5 0 0 520 x a: 510 500 0 0.01 0.02 0.03 0.04 i/vF Figure 34. versus 1/ -Jt for solutions used to cali­ brate the Teflon conductance cell. - 107 - TABLE XIII CALCULATION OF THE CELL CONSTANT FOR THE TEFLON G OKDU GTANCS CELL __ ______ ___________ I/O V ® 9 Teflon cell 1 -*• Glass cell^ ohms ohms Kc Teflon < cm Solution A First sample 499 825-5 Second sample 495 825.2 Average 497 825-2 0.557 Solution B First sample 3950 14,600 Second sample 8850 14,580 Average 8880 14,590 0.541 ^ The correction for residual conductance (see text) would he less than 0,1 per cent in all cases, ^ The capacitance Cs was j^u/af or> l©ss ir all cases, so that Rs = Rx — 1/G-. - 108 - (Section III). Seven samples of iodine pentafluoride were employed; four of these were returned to the mixing tube after measurement and used to prepare solutions of hydrogen fluoride in iodine pentafluoride. Two other hydrogen fluo­ ride solutions were prepared, and attempts were made to prepare solutions of three solid materials in iodine penta­ fluoride; the data on the concentrations of these solutions are presented below. Hydrogen,fluoride solutions. Five of the six solutions of hydrogen fluoride in iodine pentafluoride were prepared by isolating samples of hydrogen fluoride gas at a definite temperature and pressure in the small Fluorothene test tube (see Figure 3) and then condensing this vapor into the sample contained in the large test tube. In order to calculate the concentrations of these solutions, it was necessary to know the volume of the sma.ll test tube and the vapor density of hydrogen fluoride under the conditions employed. Fredenhagen*s vapor density data (99) were used. The test tube was found to contain 27*4-0 grams of water at 23*4°C when filled to the valve; at this temperature the density of water n —1 is 0.9974 grams ml.-*1* and that of air is 0.0012 grams ml. , so that the volume of the small test tube was 27*47 ± 0.02 ml. Data on the quantities of hydrogen fluoride used and the cal­ culated concentrations of these samples are given in Table XIV. - 109 - TABLE XIV Da TA Oh PREPARATION OF HYDROGEi'^ FLUORIDE SOLUTIONS IN IODINE PENTAFLUORIDE Weight ILFXr5 9 Moles IF5 x 103 grams PhA ii.f mm • Hg Moles HF x 105 Mole cHF °G 24.6 9 •49 1.78 202 730 24.7 3.49 1 .7 0 44.6 201 730 24.5 3.49 1.71 51.0* 230 730 24.3 732 24.3 CD vn • -h 730 'I* 193 CO• 4 3 . 0* t, -3n.49, . 6.97 45.6* 206 732 24.1 3-49 732 24.1 3*49 730 24.2 3 *48 730 24.2 3.48 730 24.2 3.48 2.95 17.42 These four solutions were prepared using iodine_pentafluoride recovered from measurements of the conductance of the pure liquid (see text). - 110 - The temperature of the small test tube was controlled by placing a beaker of water around it for several minutes* One hydrogen fluoride solution was prepared by weighing a quantity of hydrogen fluoride which had been condensed Into the small test tubej the concentration of this solution is calculated in Table XV in the following section* Solutions of solid 'materials. Table XV contains data on the quantities of solid materials and of liquid hydrogen fluoride employed in attempts to prepare solutions of these materials. As will be seen below (Results), there is some doubt as to whether an appreciable amount of any of the solid materials dissolved. Measurements with Iodine Pentafluoride The measurement procedures with these iodine penta­ fluoride samples were described above, except that in many cases the cell was not removed from the solution-mixing system and inserted in the oil bath. In these cases the cell was maintained at a uniform temperature by placing it in a beaker of water equipped with a stirring pump and a ther­ mometer. The reasons for this procedure will be discussed below (Results). - Ill - TABLE XV DATA ON PREPARATION OF VARIOUS SOLUTIONS IN IODINE PENTAFLUORIDE Solute Weight IF5 gram s holes _ IF5 x 10-^ Weight solute grams Moles solute x 10^ 88 Mole % solute 27.4 HF 51*7 253 1.67 LiF 5 2 .1 235 0.1133 4.38 1.84 NaF 60.2 272 0.1184 2.84 1.0 3 NaF 47.8 216 0.0375 0.393 0.41 XHFs 46.8 211 0.2531 3 .24 1.5 2 - 112 - Data and Calculations Conductance data for the various samples described above (Experimental Methods) are given in Tables XVI, XVII, and XVIII. Graphs of Rx versus 1/VF in figures 35 9 36, 37, and 38. for these samples are shown The resultant data for the specific conductances of the samples are listed in Table XIX. Early in the study of these conductances, It was observed that the resistance of a given sample at a given temperature and frequency varied with time. The measured resistance of one sample as a function of time is given in Table XIX and Figure 39* As the change was quite slow at first, increasing more rapidly as time passed, it was considered important that measurements be made as soon as possible after filling the cell. For this reason the procedure described above (Experimental Methods), involving measurement of the conduct­ ance without removing the cell from the filling apparatus, was initiated. All samples were measured in this way except those for which measurement in the oil bath is specified In the tables. Discussion of the Results The specific conductance of iodine pentafluoride as measured in this research is slightly lower than that TABLE XVI CONLUGTANGE DATA FOR IODINE PENTAFLUORIDE t, °G f, cycles RgXlO"5, chins Cs, /'/*1 R x X l O “5 , ohms First sample y 2 5 -0 ^ 930 1950 3820 27.80 27.71 27.61 23 13 7 27.30 27.71 27.61 23 13 r 28.04 27.93 27.85 83 12 8 28.54 28.44 28.34 23 1° 6 28.64 28.54 28.46 22 11 6 28.58 26.47 28.39 23 12 7 23.19 28.08 28.01 Second sample 2p.O* 980 1950 3S20 28.04 27.93 27.35 Third sample 24.3 380 1950 3820 28.54 28.44 28.34 Fourth sample 24.2 980 1950 3820 28.64 28.54 28.46 Fifth sample 24.2 980 1950 28.58 26.47 28.39 Sixth sample 24.2 980 1950 3820 28.19 28.08 28.01 Seventh sample 24.3 980 1950 3820 23.58 23 28.58 28.47 11 28.47 28.40_________ 6_______ 28_po___ _ These tvjo saruoles were measured in the oil bath (see text) 114 - Fourth sample Fifth and seventh samples. O of Sixth sample Second sample So, First sample Uo^ a & Ve^ u a " ^5 _ TAB LE X V I I CONDUCTANCE DATA FOR SOLUTIONS OF HYDROGEN FLUORIDE IN IODINE PENTaFLUORIDE IiF mole % t, °c 1.7 0 x 10“3 ohms cycles Rs x 10-3, ohms 24.1 980 1950 5820 27.05 26.95 26.86 25 12 7 27.05 26.95 26.86 1.71 24.2 980 1950 5820 27.51 27.25 27.16 25 15 7 27.51 27.25 27.16 1.78 25.0* 980 1950 5320 26.92 2 6 .85 26.75 26 15 8 26.92 26.85 26.75 2.95 24.3 980 1950 5320 20.89 20.79 20.75 50 17 9 20.89 20.79 20.75 r: y•O 24.3 930 1950 5820 9 .225 9.099 9.022 75 58 21 9 .224 9.098 9.021 27.4 24.3 980 1950 5320 1.984 1.908 1.844 520 1.984 1.908 1.844 — — ----- — ------ — See note, Table XVI. - 116 - .U/A f 15 0 100 27,200 OHMS 27,000 26,800 cc 20,800 20,600 0.01 0.02 i/v f Figure 3 6 . versus 1 /Vf for several solutions of hydrogen fluoride in Iodine pentafluoride. - 117 - OHMS 9000 2000 x cc 1900 1800 0.01 0.02 0.03 i/Vf Figure 37. Rx versus l/-/f" for two solutions of hydrogen fluoride in iodine pentafluoride. t^ble a v i i i CON DU CTAN GE DATA FOR SUPPOSED SOLUTIONS OF SOLIDS IN IODINE PENTAFLUORIDE Solute rT_ L1F Mole % .--— — . T 1.84 °G 24.3 Rs x 10~3, *4 ohms cycles 980 1950 I .03 24.2 khf2 0.42 1.5 2 24.2 24.3 ohms 25 13 7 28.45 2 8 . 34 28.49 28.39 28.31 25 14 7 28.49 28.39 28.31 28.17 28.07 23 1950 12 28.17 28.07 3820 28.00 7 26.00 980 1950 28.68 28.60 24 14 2 a • 68 2 6. 0 0 3820 28.51 6 23.51 930 1950 3820 NaF ■• 'f Rr X 10“ 28.45 2 c •34 23.25 3820 NaF > 980 - 119 - 2 8 .25 •Pi", K ne‘ :oen ? ei s v «. 1 3uU fr,±//f Uo^ e; rot 5llplPO; eU o s°hii< °'a, Of TABLE XIX CONDUCTANCES OF IODINE PENTAFLUORIDE AND OF VARIOUS SOLUTIONS Solute Solute Mole % none HF LiF NaF" KHF2 I/O- x 10“5 , ohms O x l O D, ohms"- V x 10° ohma"* cm 27.43 27-65 28.13 36.5 36.2 35 •6 28.28 28.19 27.82 28.19 3 5.4 35-5 35-9 35.5 7.8 26.69 26.99 26.58 20.57 8.82 27*4 1.708 37-5 37.0 37.6 48.6 113o 585 1.84 28.05 35*6 19.3 0.42 1.70 1.71 1.78 2.95 19*7 19.5 19.3 19.1 19.2 19-4 19.2 20.2 19.9 20.3 26.2 61.0 315 27.32 35*9 1.03 28.12 35*6 19-4 19-3 I .52 28.32 35*3 19.0 - 121 - TABLE XX MEASURED RESISTANCE AS A FUNCTION OF TIME* Rs x 10^, Time, ohms minutes 27.93 3 27.92 5 27.92 7 27.91 8 27.88 10 27.84 12 27.79 15 27.63 18 27.38 20 The data given is for the second sample of pure iodine pentafluoride at 25-0°C and 1950 cycles. - 122 - Rx , OHMS 27,800 27,600 10 5 T IM E , Figure 39. Change of MINUTES v/ith time. - 123 - 15 20 reported by Banks, Emeleus, and Woolf (2 to 3 x 10-5 ohm"1 cm. *^) (2 3 ), and the spread of the values obtained is much smaller than that reported by these authors. The high con­ ductances for the first and second samples might be attrib­ uted to contamination during the period between filling of the cell and measurement of the conductance, since these two samples were removed from the filling apparatus and carried to the oil bath. The mean of the other three conductance values is 1.92 x 10“5 ohm”1 cm."1 . The hydrogen fluoride solutions have conductances definitely greater than that of the solvent. However, these conductances are very small compared to the conductances of comparable solutions of hydrogen fluoride in water. For example, a solution containing 1.7 mole per cent hydrogen fluoride in iodine pentafluoride has 24 grams of solute per liter of solution if the density is approximately 3*2 thatof the pure solvent. normal solution* a-1.3 This is a 1,3 normal aqueous solution of hydrogen fluoride has a specific —p conductance of approximately — p x 10“^ ohm —"l cm." —1. hydrogen Thus fluoride, considered a weak electrolyte in water, is apparently a much weaker one in iodine pentafluoride. However, a graph of conductance versus concentration, as shown in Figure 40, is not of the form expected for a weak electrolyte. Furthermore, the conductance of the most concentrated solution investigated is less than that of some pure liquids, - 124 - 60 I co 40 s: o o X X 20 10 0 2 MOLE 4 6 8 %HF Figure 40. Conductance as a function of concentration of hydrogen fluoride in iodine pentafluoride: expected, behavior for a ueak electrolyte is seoan by the broken line - 125 - such as bromine trifluoride. Thus it appears that hydrogen fluoride does not behave as a conventional electrolyte in iodine pentafluoride. Apparently, the attempts to dissolve solid electrolytes in iodine pentafluoride were unsuccessful* The measured conductances of the resulting samples do not differ signifi­ cantly from that of the pure solvent, and it seems highly improbable that the salts chosen would dissolve without ionizing and producing a. measurable change in conductance. - 126 - VI. DIELECTRIC-CONSTANT MEASUREMENTS Theoretical Aspects The Nature of the Measurement As discussed in Section V, the admittance of a liquid may be discussed In terms of a conductive component due to ionic migrations and a susceptive component due to induced charge separations and dipole orientations of neutral molecules. At audio frequencies, that is, 20,000 cycles or less, only the conductive effects are generally impor­ tant in moderately conducting liquids. At frequencies of one megacycle (10^ sec."^) or higher, however, both effects may be significant. The localized charge-reorientation, effects that make up the susceptive component of the admittance increase the capacitance between the electrodes employed in a measure­ ment. The ratio betvreen the capacitance when this space is filled with a given substance and the capacitance between the same electrodes in vacuo is called the dielectric constant of that substance. Just as the actual conductance of a given sample in a cell is related to the specific conductance of the liquid by - 127 - a conductive cell constant, the capacitance of such a sample is the product of a capacitive cell constant and the dielectric constant € of the liquid* c '= c -Cr (32) This capacitive cell constant, Cr , is called the replaceable capacitance of the cell# As specific conductance and di­ electric constant are both parts of a specific admittance characteristic cell of the liquid at a givenfrequency,these constants are closely related. If Cr is given two in micromicrofarads and K c in Equation 25 is in cm.”**-, K cCr = 0.8854. (33) Accurate measurement of either the conductive or the susceptive component of an electrical admittance is feasible' only if this component is reasonably large compared with the other. In capacitance measurements, this requires that the susceptance due to the capacitor should be large compared to any parallel conductance. The susceptance of a capacitor is given by B = cj C = 2'-tC, (34) from Which, substituting from Equations 32, 33, and 34, we have | = 0.5563 x 10~12 f i (35) Thus capacitance measurements on somewhat conducting liquids will be more feasible cl*t higher frequencies. - 128 - For this reason, most dielectric constant measurements are made at radio frequencies. The lower frequency limit f 1 for convenient capacitance measurement is given approximately by 3 = G, so that f' = 1.8 x 1012 t C Thus the specific conductance of a liquid will greatly (36) influence the choice of experimental methods. The Instrument Dielectric constants were measured with a Twin-T impedance bridge,^ designed for the frequency range 0.5 -40 megacycles and Figure 4l. employing the circuit shown The balance equations for thisbridge Gx + Ga - R gJ2C'C"(.1 + ^r,)= + CB + C'C"1 (jjj, + in are 0 ) - g2L (37) = 0. (38) When employing this bridge, a balance is first obtained with the unknown removed from the circuit. The unknown impedance is then inserted at X and a second balance is obtained. The effective conductance and capacitance of the unknown are then 1 The General Radio 321A Twin-T bridge, manufactured by the General Radio Company, Cambridge, Mass., was employed. - 129 ~ c' c GEN. DET. IM b 1^Lir e 41 • Cii'cuit of the rvjI r:-T - 150 - bridge* given by G-x - - G^) Gx - (39) CB ! " CBS • (40) ^Gs » ^Bi? an<^ ^Bs are Jc,^-e values of calibrated variable capacitors Cg and Cq. before and after inserting the unknown. For determination of dielectric constants, Equation 40 is of primary interest. At frequencies above one megacycle, certain additional impedances in series with the unknown and the precision con­ denser Cg become important. The manufacturer (102) gives the circuit in Figure 42 as equivalent at higher frequencies to CB plus the leads to the unknown. In the present work, the cell used for the unknown contained a lead inductance in series with L*. It was, therefore, unnecessary to apply the manufacturer’s corrections for L 1, as this inductance could be considered with the cell inductance. Thus simplified, the recommended equation for the unknown capacitance is Or &r Cx = §A-- --------Sa----- • 1 - to IiGCg 1 - a;£"bcCB * (41) The calculations involved are simplified by use of graphs of the function 1 - cj 2I»cCg as a function of uu and Cg. graphs are available from the manufacturer (1 0 3 ). - 151 - These 7 B J Figure 42. Equivalent circuit for Cq and leads to the unknovm at hi gh frequencies. Gu Figure 43frequencies. Equivalent circuit for tie unhnovm at higl - 132 - The Cell and the Unknown Liquid The dielectric constant is defined as the ratio of the capacitance of a given condenser filled with the dielectric in question to its capacitance in vacuo. In practice the dielectric constant of air is so nearly unity that a measurement in air may replace the one in vacuo. The dielectric constant £ is then given by € = Cl_ ^cl (42) where Cl is the capacitance of the condenser filled with the unknown liquid and Ca is Its capacitance in air. Any actual cell Involves inductances associated with the leads from the bridge to the unknown* These become important above about one megacycle, so that the equivalent circuit of the unknown is that given in Figure 43* ductance while capacitance. and is the lead in­ are the cell conductance and Only the latter two will change with the medium filling the cell. The impedance of the circuit in Figure 43 in complex circuit notation (93) is Zx = JojIm + - q. u (43) so the imaginary or susceptive portion of the admittance is B = x 1(3 = u- X ^ c u ^ - _ tiJ2Lucu )2 - 133 - - ...inAi2 u ,2 i ^ 2 q u 2 + ( 44) In a later section we will consider the actual values of the last terms in the numerator and denominator of this fraction. In experimental situations they may turn out to he negli­ gible, and the above equation may be reduced to -1_ = _1_ - (4-5) If Cx is now determined at a number of frequencies and its reciprocal is plotted against cj^, a straight line of slope and intercept 1/C^ should result. ’ This method of treat­ ment was applied to obtain the desired cell capacitance in this research. A straight-line plot of proper slope indi­ cates, also, that Cu does not vary with frequency within the experimental range. Unfortunately, Cu for any actual cell does not corre­ spond to the capacitance £Cr since there Is some capacitance in portions of the cell that cannot be filled with liquid. Thus Cu — + ^i where Ci is the residual or irreplaceable capacitance. (4-6) In order to determine Cr it is thus necessary to measure Cu for two different values of ^LcCB /y f yO u l 7w T Air at 2 5 .1 °C 1 10 out in 913.8 758.2 914.1 758.6 155-5 out in 1100.2 948.6 1128.2 970.4 157-8 TO 1 10 enzene at 24.0°C out in 913.4 575-1 913.7 575-0 out in 929 -0 574.8 949 -8 58 8.8 338-7 5 6 1 .0 Diethyl ether at 24.0°C 1 10 m out 913-4 294.0 913-7 294.0 619-7 out in 929-1 254.0 949-9 255.5 694.4 Chlorobenzene at 25«1°0 1 10 out in out 9 1 3 .8 9 6.3 1100.2 170.4 913 *5 96.5 817*2 1128.2 171.5 956.9 the third and fourth columns of Table XXII. The second column gives the dielectric constant data employed in the calculation, while the replaceable capacitance values ob­ tained using Equation 47 are given in the fifth column. From the data for 1^ in Table XXII, the use of Equation 45 rather than Equation 44 can be justified. is approximately 4.5 x 1 0 “® henries, so that for values of &u below 10~3 ohmr^-, L^.G-^ will be less than 4.5 x 10”*^. can never be less than 1 . 5 x 1 0 “^ , Cu so that the error due to neglecting the last term in the numerator of Equation 44 will be under 0.03 per cent. Further, if no frequencies above 3 0 megacycles are employed, cj ^Lu^G-u 2 will be less than 8 x 1 0 “3 9 and this term may be dropped from the denominator. Upon removal of these two terms, Equation 44 reduces to Equation 45For reasons discussed above, the Cr value obtained with diethyl ether was discarded. Cr was taken as 143 * 5 ± 0 . 2 micromicrofarads and 47^ 1^ as (I . 8 5 ± 0 .1 ) x 1 0 “® henries. Readings on the capacitor Cq. are given in the seventh column of Table XXI. As discussed above (Theoretical Aspects), the accurate measurement of a conductance which is a very small portion of the total admittance is extremely difficult. Therefore these values may reflect small lead resistances and flaws in the bridge circuit, as well as the actual conductances of the liquid samples. - 146 - TABLE XXII CALCULATION OF CELL PARAMETERS FOR DIELECTRIC CONSTANT CELL A e 4it2 lu x 10 6 henries rj C U > aj/ a ! Cp5 /AJLA f Air at 25.1°C 1 .0 0 1.9 155-5 Benzene at 24 ,0°C 2.276 1.84 338.5 143.4+0.4 Diethyl ether at 24.7°C 4.25 1.91 6 1 9 .0 142.6+ 0.3 Chlorobenzene at 25.1°C 5 .6 1 8 1.79 817.1 143*6+ 0.2 TABLE XXIII DIELECTRIC-CONSTANT DATA FROM NATIONAL BUREAU OF STANDARDS CIRCULAR 514 25°C ■ 2 0 °C Benzene 2.274 2.284 Chlorobenzene 5 .6 21 5.708 r ................ 1 Diethyl ether € 4.535 Temperature Coefficient CX. a 0 .0 0 2 0 0.00133 0 .0 2 0 Nitrobenzene 54.82 35*74 0.00225 Water 78.54 80.57 0.0 02 00 * £ t = ^ 0 ° C " ' a’ W l°S10^t “ lo^1 0 ^0 °C - 147 - M.easurements_ with chlorine trifluoride. The measure­ ments made with cell A on chlorine trifluoride are discussed below (Results). The filling of the cell is discussed in Section III, and the methods of measurement and calculation have been described in the present section. Measurements with Iodine Pentafluoride Cell construction. Cell B was designed to have a replaceable capacitance of about 10 micrornicrofarads, thus allowing measurement of dielectric constants up to about 100. This cell is shown in Figure 4^} unlike cell A it has no jacket or provision for temperature control. Cell B is shown with the cell-filling system in Figure 4 and with the Twin-T bridge In Figure 44. Dielectric-constant standards. Chlorobenzene, nitro­ benzene, and water were used to calibrate cell B. These liquids are recommended in the National Bureau of Standards circular cited above, and dielectric-constant data are listed in Table XXIII. Chlorobenzene for this standardization was steamdistilled, removed from the water layer, redistilled at 130-131°C under a pressure of 736 mm. Hg, and dried over barium oxide. - 148 - On I- 6 2J SCALE, IN. Figure 47. Section through the cylindrical cell B: A., B, outer cylinder (two parts); 0, ring covering silversoldered joint; D, inner cylinder; E, F, plugs (to oridge); G, H, Teflon insulating spacers; I, J, inlet tubes for samples (valves not shown). - 149 - Nitrobenzene was steam distilled, removed from the water layer, maintained at 200°C for an hour to remove the last of the water phase, and dried over barium oxide. The water used was taken from the department distilledwater line and put through a mixed-bed ion-exchange water purifier. The conductivity of the resulting sample was less than 2 x 10"^ ohm"-1- cm."*1 . Temperature measurement and regulation.. The teraperature of this cell was measured by a thermistor taped to the cell body. A layer of asbestos tape was wound over the cell body and thermistor. The calibration measurements were made with the' cell at room temperature. Measurements on iodine pentafluoride were made by warming or cooling the cell in a water bath for thirty minutes to assure that cell and contents were at a uniform temperature, removing the cell from the bath,drying it, and applying the thermistor and asbestos tape. The temperature and capacitance were then followed as the cell returned to room temperature. The thermistor1 was calibrated against thermocouple 3; the calibration data are given in Table XXIV and Figure 48. 1 The thermistor used was type 14b , made by the Western Electric Co., New York, N. Y. - 150 - T A B ljE X X IV THERM Io TOR-■CALIBRATION DATA E^ x 10^, volts °C *t, ohms 447.7 11.52 3501 529.7 13.60 3181 606 *9 15.56 2899 706.1 18.00 2630 841.8 21 .43 2304 955.8 24.26 2054 1061.5 26.87 1834 1160 . 9 29 .31 1642 1564.1 34.26 1260 - 151 - OHMS 4-> 2000 1000 25°C 15°C t, °C Figure 48. Calibration curve for the thermistor used with dielectric-constant cell B. - 152 - CalibratioqL data and. calculations. Water and chloro­ benzene were examined at 1 , 3 , and 1 0 megacycles while nitrobenzene was measured at 1 , 3 , 5 , 1 0 , 1 5 , and 2 0 mega­ cycles for calibration purposes. The calibration data and calculations are given in. Tables XXV and XXVI. the capacitances The values of in Table XXVI were obtained by plotting against to 2 as described in Section VI in connection Gx with Equation 45* As for cell A, the values for cell B allow an esti­ mation of the magnitude of the terms containing G-u in Equation 44, and thus provide Information on the conditions under which the use of Equation 45 is justified. about 2 x 1 0 “® henries, As 1^ is 2 will be less than 7 - * one thousandth the size of Cu if G-u is less than 1.7 x 10“-^ ohm”-*-. This Is/tter restriction on Gru is obeyed for all the calibration data, as shown in the seventh column of Table XXV. However, it will be seen in the following section that iodine pentafluoride has a somewhat higher conductivity. For this compound, Cq will be considerably larger than 55 >y*f result of its higher dielectric constant. a For a Cu value of 3 5 O, G-u might be as large as 4 x 10“^. It was shown above (Theoretical Aspects) that the susceptance of a conducting liquid must be studied above some given - 153 " TABLE XXV BATA AND INITIAL CALCULATIONS FOR CALIBRATION OF DIELECTRIC CONSTANT CELL 3 Cell Position me . itydiAT CX, . TCB 1-G->2LcjCb ///y f 11 '<■J? y ■/> r Air at 2 3 .5 °C 1 3 10 out in 616.0 6 1 6 .1 6 1 0 .5 600.6 out in 399.3 594.0 6 0 0 .3 out in 094.1 5 8 8 .6 15.5 64.5 584.8 15.5 64.5 602.4 586.9 15.5 64.5 Chlorobenzene at 2 3 .4°C out in 616.7 553*2 616.8 553*3 63 .5 15.75 3 out in 599.5 536.0 600 .3 536.7 63 *6 15*72 10 out 594.0 531*7 602.3 538.3 64.0 15.62 1 Water at 2 3 .4°C 1 3 10 out in 1047.8 226.0 1048.0 226.0 822 1.217 out in 1028.6 2 0 3 .5 1 0 3 1 .0 20 3 .6 827 1.209 out 1087*0 348.0 1241.2 351.1 680 1.136 - 154 “ Cell Position me. C-Bo> O,f CB> 1- c,j ^L q Cg X f, i ! TABLE XXV 3 CONTINUED // .af i x 10^ GX // Nitrobenzene at 23. 5° C 1 out in 616.0 252.0 25 2.0 364.1 2 .745 5 out in 596.9 232.8 597.7 232.9 364.8 2.740 5 out in 600.2 235-1 602 .4 235.5 366.9 2.724 out in 608.5 240.8 617.5 242.2 375 0 2.664 15 out in 6 00.3 22 5 .6 618.5 228.2 390.3 2.562 20 out in 600.0 217.9 635.2 222.5 412 .7 2.423 10 616 •1 - 155 - TABLE XXVI CALCULATION OF CELL PARAMETERS FOR DIELECTRIC CONSTANT CELL B ^r/^xlO6 cu , henries Air at 23.4°C 1.00 Chlorobenzene at 23.4°C 5-648 Water at 23.4°G Nitrobenzene at 23.5°0 79-03 35-09 - 156 cr , /y- G x 106 !!/J■£ ':•-f ohm -1 - me. ■'Uf °C First sample 1 10 1 10 1 10 out in 914.5 228.7 914.7 228.7 out in 930.0 174.4 950.4 175-1 775-3 out in 915-0 265*4 915-2 265-4 649-8 out in 930.2 215-3 950-7 216.4 out in 915-7 293-4 915-9 293-4 622.5 out in 913-7 229-5 913-9 229-5 684.4 0.7 686 •0 10.6 0.7 6.5 14.2 14.8 734.3 25-4 0.8 Second sample 1 10 out in 915-9 224.0 916.1 224.0 692.1 0.7 out in 915-2 255-3 915-4 255-3 660.1 7-7 12.7 Third sample 1 916.6 227-3 out in 916.8 227-3 0 •6 689-5 9-5 Hydrogen fl-uoride in chlorine trifluoride 10 out in 914.5 128.4 1 °°| iNM out in O H O H 1 H 1 1 914.7 128.4 0.7 786.3 135 0.7 1030.0 117-0 - 160 - 913-0 91-0 TABLE XXIX DIELECTRIC CONSTANT DATA FOR CHLORINE TRIFLUORIDE t, °c Gu> fAM f me. f First sample 0.7 1 6 8 6.0 6 8 5.2 0.7 10 775*3 688 14.2 1 649.8 64S -0 14.8 10 734.3 646 25.4 1 622.5 6 2 1 .8 4.277 0 .8 1 684.4 683 •6 4.708 4.719 +3 4.74 4.467 +3 4.45 Second sample 0.7 1 692.1 691.3 4.762 12.7 1 660.1 659.3 4.539 688.7 4.744 Third sample 0 .6 1 689-5 Hydrogen fluoride in chlorine trifluoride 0 .7 1 10 786.3 785.2 913 781 - 161 - 5.416 +4 5-39 Table XXX with the equation ^t = 4.754 - 0.0187 t. (49) The data talien at one megacycle fit this equation to within 0.031, the standard deviation being 0.008.*^ The values obtained from data taken at ten megacycles were not used in the above treatment because of their greater dexsendence upon the size of the inductance L^. As shown in Table XXII, this value is relatively uncertain. The sample containing hydrogen fluoride was of uncertain composition, as discussed in Section III. However, the rela­ tive changes In conductance and dielectric constant upon adding hydrogen fluoride were of use in considering the pos­ sible error in the result due to hydrogen fluoride impurity. The seventh column of Table XXVIII contains readings observed on the conductivity scale of the bridge. From Equation 33 the conductive cell constant of cell A is 6 .2 x 1 0 ~ 4 cmglj so that the observed conductances, 6 .5 to 1 0 . 6 x 1 0 ”^ ohm"**', correspond to specific conductances of 4.0 to 6 . 6 x 1 0 “ 9 ohm"*-1' cm."*1-. These may be compared to Banks* value, 3 x 1 0 " 9 ohm"-** cm."-1-, quoted in Section II; however Banks 1 measurements were presumably at audio 1 The standard deviation was calculated using the equation where d represents an individual deviation and n is the number of observations made. (See footnote, page 65*) - 162 - TABLE XXX THE TEMPERATURE DEPENDENCE OF THE DIELECTRIC CONSTANT OF CHLORINE TRIFLUORIDE t, Sample Number °c £ -- Exp *1 Calc fd i Dev. 0.6 3 4.744 4.743 + .001 0.7 1 4.719 4.741 -.022 0.7 2 4.762 4.741 + .021 0.8 1 4.708 4.739 -.031 12.7 2 4.539 4.517 + .022 14. 2 1 4.467 4.489 - .021 25*4 1 4.277 4.283 + .006 - 163 “ frequencies• In addition, as discussed in connection with, conductances observed during the calibration process, the exact significance of small conductances obtained in the present measurement is not certain, SourcegL _of _erro r an d un cer tain tv. Errors in the scale of the capacitor of the Twin-T bridge are perhaps the largest source of uncertainty from the instrument itself. The manufacturer states (1 0 2 ) that scale readings are reliable to + 1 micromicrofarad, which corresponds to one part in 600 or more in the present measurement. Error from this source would, therefore, be about 0.17 per cent. The uncertainty in the value 47r2 La will produce errors in the final result } the magnitude of these errors will depend upon the size of the correction for this inductance. Thus at one megacycle, where the capacitance Cu differs from the measured value Cx by only 0 . 8 micromicrofarads or so, this source will contribute an error of less than 0 . 0 1 per cent. However, at 10 megacycles the inductance correction is about 9 0 micromicrofarads, so that the error may be 0 . 6 per cent. ■The uncertainty in the cell constant will, of course, produce an uncertainty in the derived values for dielectric constants. The cell-constant determinations suffered from the same errors in scale readings and lead inductance - 164 - corrections as do the measurements discussed above. As the Gy_ data in Table XXII depended principally upon the readings at one megacycle, scale readings were the principal source of* difficulty; the errors corresponding to one-micromicrofarad errors in scale values are given in the fifth column of that table. 0 .3 From these data, the cell-constant error is about per cent. Errors due to impurities in the samples employed are considerably more difficult to evaluate. While many sub­ stances present as minor impurities would not greatly affect the dielectric constant, a small amount of a very polar sub­ stance such as hydrogen fluoride could cause large errors. Hydrogen fluoride would not easily be separated from chlorine trifluoride by simple distillation, and might be formed by reaction of the sample with small quantities of moisture in the apparatus; thus the presence of such an impurity was a definite danger. However, hydrogen fluoride appears to increase conductance markedly when present, as shown by the data for the sample to which this impurity was purposely added. From the data on this sample in Tables XXVIII and XXIX, the quantity of hydrogen fluoride that produced a con­ ductance of 1 3 5 x 1 0 ~ 6 ohJtrT1 at one megacycle also produced a change of 0.66 in the dielectric constant. If tnese two effects are approximately proportional to hydrogen fluoride - 165 - concentration, a conductance of 6 . 5 x 1 0 “ 6 ohnr1, the lowest conductance found for chlorine trifluoride, would correspond to a dielectric-constant error of 0*03* Since a portion of the measured conductance must have resulted from the actual conductance of the liquid, any error in the dielectric con­ stant caused by the presence of hydrogen fluoride must be less than 0 . 0 3 or 0 . 7 per cent. The root-mean-square error^- obtained from the individual sources discussed in the preceding paragraphs is 1 . 0 per cent; since the over-all spread of values at 0 . 6 to 0 .7 °G was about 0.9 per cent, an uncertainty of + 0.045 was assigned to the dielectric constant data in Table XXIX. As shown In Table XXX, the experimental data fit Equation 49 within this limit. The Dielectric Constant of Iodine Pentafluoride Data and calculations. Two samples of iodine penta­ fluoride were examined in cell B. Data and initial calcula­ tions for these samples are given in Table XXXI, while final calculations and results are listed in Table XXXII. Values of 1/CX are plotted against f2 in Figure 49; the intercept corresponds to a ^ value of 384.0 and a dielectric constant See footnote, page 65* - 166 - TABLE XXXI BATA AND INITIAL CALCULATIONS FOR IODINE PENTAFLUORIDE Cell Position CB» /AjU f me. cx> °B Cb L q jUyUf Rt, t, ohms °C 2046 24.3 2063 24.2 2049 24.3 2052 24.3 2062 24.2 2043 24.3 3304 12.6 2721 17.1 2128 22.3 2078 24.0 2008 24.9 1756 28.4 1052 31.2 First sample out in 600,0 215.4 600.6 215.4 5 out in 597.9 212 •2 600.3 212.2 388.1 10 out in 599.8 210.3 608.4 211.3* 397.1 15 out in 599.6 200.5 620.1 202.5 417.6 20 out in 614.9 199.8 653 .5 203.8 out in 639.7 204.7 707.5 211.2 3 25 385*2 449.7 496.3 Second sample 3 out in 600.0 166.7 600.6 166.7 433.9 3 out in 599.9 186.2 600.5 186.2 414.3 3 out in 599.5 206.4 600.1 206.4 393.7 3 out in 599-5 212.4 600.1 212.4 387-7 5 ,. out in 599.3 215.9 599.9 215.9 384.0 3 out in 599.3 230.7 599.9 230.7 3 out in 599.3 599.9 241.6 369.2 . 358.3.... - 167 - TABLE XXXII DIELECTRIC CONSTANT DATA FOR IODINE PENTAFLUORIDE CX» °C me. y u / A f V /*/-■■ f First sample 24.3 3 385.2 384.0 36.73 24.2 3 3 8 8 .1 385.1 36.78 24.3 10 397.1 384.6 36.74 24.3 15 417.6 387.6 37-02 24.2 20 449.7 391.4 37.45 24.3 25 49 6.3 385.6 36 .8 2 Second sample 1 2 .6 3 433.9 432.7 41.38 17.1 3 414.3 413.1 39.49 22.3 3 393.7 392.5 37.49 24.0 3 387.7 LO • vO 00 36.91 24.9 3 384.0 38 2 .8 36.55 28.4 3 369 . 2 3 6 8.1 35-17 31 .2 3 358.3 357-2 34.20 - 168 - 0.0026 0.0024 X o 0.0022 0.0021 100 300 200 400 600 f2, MC. Figure 49cell B. 1/Cyr_ versus - 169 for iodine pent-afluoride in - °^* 36.73• The slope of t-lie line drawn through these points corresponds to a value of 0 . 8 6 x 1 0 "^ henries for 4 /i^L^j this agrees reasonably well with the values obtained using water and nitrobenzene* The temperature dependence of the dielectric constant is shown in Figure 50» In Table XXXIII the experimental values of £ are compared with those obtained using the equation 6 = 46.22 - 0 . 3 8 8 t. (5 0 ) Conductivity data were taken at three frequencies at 24*3°C for the first sample and once at 24.0°C for the second. The conductive cell constant, obtained using Equation 33, is 8 .6 x 10"^ c m T h e conductances were obtained from scale readings by using the factor 46 x 1 0 ~ 6 f ohm*"'1', obtained from the data in Table XXVII. The dial readings and corresponding conductivities are listed in Table XXXIV. These values are somewhat higher than the audio-frequency conductivities of Section V; this difference might be due to the frequency change, or might correspond to lead resistances or other errors in the bridge measurements. A more careful study of conductance effects would require a thorough calibration of the auxiliary conductance capacitor and an analysis of the corrections necessary due to the presence of series imped­ ances in the cell and bridge. - 170 - 40 35 15 20 25 30 t , °c Figure 50. Temperature dependence of the dielectric constant of iodine pentafluoride. - 171 - TABLE XXXIII THE TEMPERATURE DEPENDENCE OF THE DIELECTRIC CONSTANT OF IODINE PENTAFLUORIDE t Experimental °c Calculated Deviation 12.6 41.38 41.32 +0.02 17.1 39.49 39.60 -0.11 22.3 37.49 37.55 -0.06 24.0 36.91 36.89 +0.02 24.9 36.55 36.52 +0.03 28.4 35.17 35.18 -0.01 31.2 34.20 34.10 +0.10 TABLE XXXIV HIGH-FREQUENCY CONDUCTANCE DATA AND CALCULATIONS FOR IODINE PENTAFLUORIDE f, Scale factor Scale readings Change Gux 103 ohm-1 me. K x 106 ohrn^^cm”^- First sample 3 138 29 11 18 2.5 21 5 230 21 11 10 2.3 20 10 460 16 10 6 2.8 24 19 2.6 22 Second sample 3 138 30 11 - 172 - ..of— error and uncertainty. The expected error of one micromicrofarad in the scale readings on the Twin—T bridge corresponds to about one part in 3 8 0 in the value of Cx j the uncertainty from this source is thus 0.26 per cent. In the present measurements an error in the value 4 ^ L u would not effect the value obtained for o for the first sample since the value of to zero frequency. was obtained by an extrapolation Ho^^ever, the dielectric constants at various tempera-tures for the second sample were obtained using the value (0.82 + 0.03) x 10“ 6 henries for 4tf2 Lu . As . the total correction in Cu was about 12 parts in 4000, the uncertainty from this source would amount to + 0 . 0 1 per cent. The uncertainty in the cell constant for cell 3 will lead to an uncertainty in the calculated dielectric constants comparable to that discussed above in connection with cell A. From Table XXV, this uncertainty appears to be about 0 .1 per cent. Errors due to impurities in the samples employed might be less in the present case than in. the study of chlorine trifluoride, since the iodine pentafluoride underwent a much more careful purification. As before, the estimation of the magnitude of errors from this source is difficult} however, the freezing-point data in Section IV on the purified iodine pentafluoride give some indication of tne purioy of unis material• - 173 - The root-mean-square error from the above sources is O . 3 P©i* cent; this uncertainty was therefore assigned to the dielectric-constant value for iodine pentafluoride• The experimental data fit the temperature-dependence equation (Equation 5 0 ) to within this limit, as shown in Table XXXIII. Discussion of the Results The permanent dipole moment of a molecule in an unassociated liquid Is related to the dielectric constant by the Onsager equation (7 1 ) where n is the refractive index, h is the Boltzman constant T is the absolute temperature, and N /ls the number of rnolecules per cc. of liquid# For many liquids, moments calcu­ lated by means of this equation (1 0 5 ) agree within a few per cent with moments calculated from vapor or dilute-solution data« The refractive indices of chlorine trifluoride and iodine pentafluoride have not been determined experimentally. Moreover, calculation of these values from molar refractions obtained by summing the atomic refractions of the constituent atoms is usually unsatisfactory for fluorine compounds, as the atomic refraction assigned to fluorine varies with the - 174 - ‘typG of compound in which it. is found. The procedure adopted in tnis research was to estimate the molar refrac­ tions of cnlopine tpifluo'pide and iodine pentafluopide fpom the published refractive indices of bromine trifluoride and bromine pentafluoride, respectively. The Lorenz-Lorentz expression for the molar refraction is [Rj = M . a where M is liquid. . (52) +2 the molecular weight and d is the density of the Theoretically, n in Equations 51 and 52 should represent the refractive index at infinite frequency. However, in most practical work, a definite value such as that for sodium D light is employed. The refractive indices of bromine trifluoride and bromine pentafluoride at 25°C using sodium D light are, respectively, 1.4536 and 1*5410 (8 5 )* The densities of these two liquids at 25°0, calcu­ lated from equations in Section II, are 2*798 that the molar and 2*465, so refractions are, respectively, 15.2 and 14.9 The atomic refraction of chlorine is 5*967, that of bromine is 8 .8 6 5 , and that of iodine is 15*900 (10b). Thus the sub­ stitution of a chlorine atom for a bromine in bromine trifluoride should decrease the molar refraction by 2.9, giving a value of 10.5 for chlorine trifluoride. - 175 - The density of the chlorine compound at 25°0 is 1.81 gm. ml-1, so that, aga-in using Equation 52, n^ for chlorine trifluoride is 1.76. By a similar calculation, using a density at 25°C of 3*19 6 m. mlr1 , n2 for iodine pentafluoride is 2.20. Though obtained from data at 25°G, these values will be used at other temperatures, as the refractive index is not highly temperature dependent. The value N in Equation. 51 may be given by N' = where % • a (53) is Avagadrofs number, so that ,, 2 = _2ih_ 4 tjNa M d £. - ,±_ngj e (n2 + 2)2 , } This equation was used to estimate the dipole moment of chlorine trifluoride at 0 and 25° C and that of iodine penta­ fluoride at 12°G, 25°G, and 352°C. These calculations and the data employed are summarized in Table XXXV*. The moment obtained for chlorine trifluoride is much greater than the value found by Magnusson (6l) in the vapor, 0.544D. A possible explanation is that the chlorine trifluoride is an associated liquid, so that Equation 51 may not be applied. No data on the dipole moment of iodine pentafluoride are available for comparison with that calculated in Table XXXV. Kov/ever, the difference in electronegativity on. Pauling fs - 176 - TABLE XXXV CALCULATION OF APPARENT DIPOLE MOMENTS IN LIQUID CHLORINE TRIFLUORIDE AND IODINE PENTAFLUORIDE ■t, rfi d c M D g. ml”1 °c //9 / Ctilorine trifluoride o• o 2 5 .0 1.7 6 4.75 1.89 92.5 1 .0 0 1.7 6 4.29 1.81 92.5 1 .03 Iodine Pentafluoride 1 2 .0 2 .2 0 41.5 3.25 221.9 3-75 2 5 .0 2 .2 0 36.5 5.19 221.9 3 •60 3 2 .0 2 .2 0 33.4 3-16 221.9 3.52 _ 177 - scale (107) Is 1*5, so that a single I-F bond would have a moment of about 1.5b. Any of the proposed structures for iodine pentafluoride involves a somewhat symmetrical arrangement of the fluorines about the iodine atom, so that these moments would tend to cancel£ a resultant molecular moment equal to two unopposed I-F bond moments would be very unexpected. Thus there is reason to believe that Equation 51 cannot be applied to iodine pentafluoride, and that the latter is an associated liquid. The variation of the calcu­ lated moment with temperature reinforces this conclusion, as the degree of association of the liquid might be expected to be temperature dependent. - 178 - VII. SUMMARY Procedures and equipment for handling halogen fluorides and for measurement of conductivities, dielectric constants, and freezing points of these compounds have been developed. Freezing point, conductivity, and dielectric constant data have been obtained for iodine pentafluoride, and the dielec­ tric constant of chlorine trifluoride has been determined. The freezing point of iodine pentafluoride is 9*425 ± 0.01°C, somewhat lower than the previously published value, 9 *6°G. From the shape of the freezing curve, the solute (impurity) content of the purified iodine penta­ fluoride was estimated at 0.0025 molal. The freezing-point depression for an 0.0147 molal solution of hydrogen fluoride in iodine pentafluoride was less than that expected on the basis of theory; this might be explained on the basis of anassociation of hydrogen fluoride molecules In this solvent. The conductivity of iodine pentafluoride was found to be 1.92 + 0.02 x 10~5 ohms"*1 cm.”"1 . Conductivities of severa,l solutions of hydrogen fluoride in iodine penta­ fluoride were determined; while these solutions were somewhat more conducting than the pure solvent, it appeared that - 179 - hydrogen fluoride cannot be treated an an electrolyte in iodine pentafluoride. The dielectric constant of liquid chlorine trifluoride in the temperature range C to 2 5 °C is given by €t = 4.754 - 0.0187 t and that of iodine pentafluoride from 12 to 52°C is given by = 46.22 - 0.388 t. For both compounds, the molecular dipole moments calculated from these data were significantly higher than moments from other sources. There is reason to believe that these discrepancies result from intermolecular association in these two liquids. - 180 - List of Symbols Used The following symbols were used in. the text, tables, and equations of this thesis. The meanings of the various symbols are listed, and where subscripts have been used to give particular meanings to a more general symbol, the page upon which each symbol with subscript was first used is given. Letters of the alphabet have been used, in addition, to designate parts or portions of figures and to distinguish between different thermocouples or dielectric-constant cells. A Cross sectional area B Electrical susceptance Bx , page 129 C Electrical capacitance Ca , 153; 0B , 129; 0G) 129; 0lt 128; c1, 133; Cr , 128; cs, 74; Cu , 133; Cx , 74 and 131 E Electrical potential E-fc, thermocouple potential Gr Electrical conductance 133; Gx , 131 K0 Conductive cell constant Kf Cryoscopic coefficient L Electrical inductance hf Molar heat of fusion - 181 - M Molecular v/eight N Mole fraction N^ A v agadro1s number Nr Number of molecules per cc. R' G-as constant R Electrical resistance Re, 71; Rs, 74; Rt, 151; Rx , 74 [__RJ Molar refraction T Temperature, degrees Kelvin X Electrical reac tan ce Xe, 71 Y Electrical impedance e. Temperature coefficient of the dielectric constant c Empirical constant (see page 49) d Empirical constant (see page 49) + d4 Density f Frequency in cycles (sec. ~ i Electrical current j V^T k Specific conductance k1 The Boltzmann constant 1 Length or distance m Molal concentration n Refractive index p Pressure, vapor pressure - 182 - ) t Temperature, decrees centigrade x Time £ Dielectric constant CJ Frequency in radians (sec.~“ )j /U Dipole moment = 2V f Temperature coefficient of log-^Q TT A mathematical constant, approximately 3*14-16 A Equivalent conductance - 183 - List of Abbreviations Used The following abbreviations were used in the text, tables, and figures of this thesis. Abbreviations marked were employed only in tables and figures. A Angstrom units co. Cubic centimeters cm. Centimeters D Debye units D 1 -« Demal concentration units &b.* Decibels Hy * Henries (circuit 'diagrams only) in.'::' Inches K * Thousands of ohms (circuit diagrams only) Kcal. Kilocalories me. Megacycles ml* milliliters mm. Hg Millimeters of mercury (pressure) sec. Seconds °G Degrees Centigrade °K Degrees Kelvin i0 * Per cent - 184 - Micromicrofarads /.( f Microfarads (circuit diagrams only) * Ohms (circuit diagrams only) - 185 - LITERATURE CITED 1. 2* 3* ;4. d. Gore, Proc. Roy. Soc., 12, 2 5 5 (1870). FI. Moissan, Le Fluor et ses Composes, Steinheil, Paris (1900 ). E. B. R. Prideaux, Proc. Chem. Soc., 2£, 19 (1906). E. B. R. Prideaux, J. Chem. Soc., 82, 5 1 6 (1906). 5* P. Lebeau, Compt. rend., 141. 1018 (I9 0 5 ). 6. P. Lebeau, Ann. Chim. Phys., (8 ), 2> 241 (1906). 7. P. Lebeau, Bull. soc. chim. France (2.)? 55, 148 (1906). 8. 0. Ruff and E. Ascher, Z. anorg. allgem. Chem., 176. 258 (1928 ). 9* 0. Ruff and W. Men.zel, Z. anorg. allgem. Chem., 19 8 , 575 (1931). 10. 0. Ruff and FI. Krug, Z. anorg. allgem. Chem., 190. 270 (1930). 11. 0. F*uff and A. Braida, Z. anorg. allgem. Chem., 214, 91 (1933). 12. 0. Ruff and W. Menzel, Z. anorg. allgem. Chem., 202, 49 (1931). 15. 0. Ruff and A. Braida, Z. anorg. allgem. Chem., 206, 63 (1932) 14. R. A. Durie and A. G. G-aydon, paper presented before the Division of Physical and Inorganic Chemistry of the American. Chemical Society, New York, N. Y., September, 1951. - 186 - 15• J. H. Simons, Ed., Fluorine Chemistry. Academic Press. Inc., New York (1950), a. H. S. Booth and J. T. Pinkston, Jr., The Halogen Fluorides, Chap. 4. h. G. Glocker, The Theoretical Aspects of Fluorine Chemistry, Chap. 1 0 . 16. E. B. Maxted, Recent Advances in Inorganic Chemistry, Clarendon Press, Oxford (1947). 17* N. N. Greenwood, Revs.Pure and Appl. Chem. (Australia), 1 , 84 (1951). 18. Sharpe, A. G., Quart. Revs., 4, 115 (1950). 19* N. V. Sidgwick, Ann. Rep. on Progr. Chem. (Chem. Soc. London), 20, 128 (1933). 20. N. Bowman and W. Proell, J. Space Flight, 2, No. 1, 6 - 9 (1 9 5 0 ); C.A., 44, 4224 (1950). 21. W. Kwasnik, Fiat Rev. of German Sci., 1939-46, Inorg. Chem. I, p. 168. 22. A. I. Popov and G. Glockler, J. Am. Chem. Soc., Z4, 1357 (1952). 23. A. A. Banks, H. J. Emeleus, and A. A. Woolf, J. Chem. Soc., 124£, 2861-5. 24. V. Gutmann, Angew.Chem., 62, 312 (1950). 25. V. Gutmann and H. J. Emeleus, J. Chem. Soc., 19.52? 1046. 26. A, G. Sharpe, J. Chem. Soc., 19.4.2, 2901. 27. H. S. Booth and J. T. Pinkston, Chem. Revs., 41, 421 (1947)• 28. H. J. Emeleus and A. A. Woolf, J. Chem. Soc., 125-0, 164. 29* A. G. Sharpe and H. J. Emeleus, J. Chem. Soc., 1248, 2135- 50. A. G. Sharpe, J. Chem. Soc., 1252, 3444. - 187 ~ 31. J* F. Grail, H. C. Miller, L. S. Verde Hi, and F. D. Loomis, paper presented before the Division of Physical and Inorganic Chemistry of the American Chemical Society, New York, N. Y., September, 1947. 32. 0. Ruff and R. Keim, Z. anorg. allgem. Chert:., 201, 245 (1931). — 33* J- F. Ellis and W. K. R. Musgrave, J. Chem. Soc., 12 5 0 , 3608-12. 34* E. T. McBee, V. V. Lindgren, and W. B. Liggett, Ind. Eng. Chem., 3£, 359 (1947). 35- W. B. Liggett, E. T. McBee,/ and V. V. Lindgren, U. S< Patent 2,432,997, Dec. 23, 1947; C.A., 42, 2618 (1948). 36. E. T. McBee, V. V. Lindgren, and W. B. Liggett, U. S. Patent 2,471,831, May 3 1 , 1949; C.A., 44, 3523 (1950). 37. E. T. McBee, V. V. Lindgren, and W. B. Liggett U. S. Patent 2,488,216, Nov. 15, 1949; C.A., 44, (1950). 1629 38. W. B. Liggett, E. T. McBee, and V. V. Lindgren U. S. Patent 2,480,080, Aug. 23, 1949; C.A., 44, 2020 (1950). 39. A. A. Banks, H. J. Emeleus, R. N. Iiaszeldine, and V. Kerrigan, J. Chem. Soc., 1948, 2188-90. 40. A. F. Scott and J. F. Bunnett, J. Am. Chem. Soc., cS4, 2727 (1942) 41. L. Domange and J. Neudorffer, Compt. rend., 226, 920 (1948). 42. H. Schmitz and H* J. Schumacher, Z. Naturforscn. 2a, 362 (1947). 4"5 • E. Wicke, Nachr. G-es . Wiss. G-ottingen, Matn.-physik Klasse, 1946. 89-90; C.A., 43., 6 5 OO (1949). 44. A. Eucken and S. Wicke, Naturuissenschaften, 2IL9 233 (1950) - 188 - 45* E. Wicke, paper presented before the Division of Physical and Inorganic Chemistry of the American Chemical Society, New York, N. Y., September, 1951. 46. R. N. Doescher, J. Cherru Phys., 1£, 1 0 7 0 (1951). 47. A. L. Wahrhaftig, 48. H* Schmitz and H. J. Schumacher, Z. Naturforsch., 2 a, 359 (1947). - 49- H. J. Schumacher, H. Schmitz, and P. H. Brodersen, Anales asoc. quim. argentina, 2S 98 (1 9 5 0 ); C •A., 45,.2309 (1951). 50* 0. Ruff and F. Laass, Z. anorg. allgem Chem., 185. 214 (1929). 51. M. T. Rogers, A. L. Wahrhaftig, and V. Schomaker, paper presented before the Division of Physical and Inorganic Chemistry of the American Chemistry Society, Atlantic City, N. J., April, 1947* 52. H. J. Emeleus, 53. D. A. Gilbert, A. Roberts, and P. A. Griswold, Phys. Rev., Z£, 1723 (1949). 54. F. R. Lowdermilk, R- G. Danehower, and H. C. Miller, J. Chem. Educ., 28, 246 (1951)* 55. H. Schmitz and H. J. Schumacher, Z. Naturforsch.,2a, 3 6 2 (1947). 56. H. von. Wartenberg and G. Riteris, 258. 356 (1949). 57. H. Schmitz and H. J. Schumacher, Z. Naturforsch,,£a, 363 (1947). 58. K. Schafer and E. Wicke, Z. Elektrochem.,5g, 2 0 5 - 9 (1948). 59. R. D. Burbank and F. F. Bensey, The Structures of the Interhalogen Compounds, Carbide and Carbon Chemicals Co., Nev? York (1952); Nuc. Sci. Abstracts, 6, 735 (1952). J. Chem. Phys., 10, 248 (1942). J. Chem. Soc., 1949. 2979- - 189 - Z. anorg. Chem., 60. D. F. Smith., The Microwave Spectrum and Structure of Chlorine Trifluoride, Carbide and Carbon Chemicals Co., New York (1952); Nuc. Sci. Abstracts, Z, 65 (1953). 61. D. W. Magnusson, J. Chem. Phys., 2 0 , 2 2 9 62 . (1952). E. A. Jones, T. F. Parkinson, and R. B. Murray, J. Chem. Phys., 1Z, 501 (1949). 63. R. L. Potter, J. Chem. Phys., 1Z, 957 (1949). 64. H. S. G-utowsky, D. W. McCall, and C. P. Slighter, J. Chem. Phys., 21, 279 (1953). 65. H. S. G-utowsky and C. J. Hoffman, J. Chem. Phys., 19. 1259 (1951). 66. M. D. Scheer, J. Chem. Phys., 20, 924 (1952). 67. J. N. Grisard, H. A. Bernhardt, and G. D. Oliver, J. Am. Chem. Soc., ££, 5725 (1951)- 68. A. Weber and S. M. Ferigle, J. Chem. Phys., 20, 1497 (1952). 69. H. Schmitz and H. J. Schumacher, Z. Naturforsch., 2a, 363 (1947). 70. H. Schmitz and H. J. Schumacher, Anales asoc. quim. argentina 32, 61 (1950); C.A., 44, 8 2 3 6 (1950). 71. P. H. Brodersen and H. J. Schumacher, Z. Naturforsch., 2 a, 358 (1947)* 72. P. H. Brodersen and H. J. Schumacher, Anales asoc. quim. argentina, 2J3, 52 (1950); C.A., 44, 8235 (1950). 73. D. F. Smith, M. Tidwell, and D'. V. P. Williams, Phys. Rev., ZZt 420-1 (1950). 74. G. D. Oliver and J. W. Grisard, J. Am. Chem. Soc., Z4, 2705 (1952). 75. A. G. Sharpe, J. Chem. Soc., 76. A. A. Woolf and H. J. Emeleus, J. Chem.Soc.,19.42, - 190 - 1221*798. 2865- 77. A. G. Sharpe, J. Cnem'. Soc., 1250, 2907. 78. A. A. Woolf and H. J. Emeleus, J. Chem. Soc., 1950. 1050. 79• C. V. Stephenson and E. A. Jones, J. Chem. Phys., 20, 1830 (1952). — —» 80. 0. Ruff and A. Braida, Z. anorg. allgem Chem., 220, 43 (1934). ° ’ 81. R. C. Lord, J. Am. Chem. Soc., 72, 522 (1 9 5 0 ). 82. H. Braune and P. Pinnow, Z. physik Chem., B55, 239 (1937). 83. 0. Ruff and R. Keim, Z. anorg,allgem.Chem., 123, 176 (1930). 84. K. Glauber and V. F. H* Schomaker, Phys. Rev., 82, 667 (1953). 85. L. Stein and R. C. Vogel, The Refractive Indices of Bromine Trifluoride-Bromine Pentafluoride Mixture, ANL-4941, Argonne National Laboratories, Chicago (1953)? Nuc. Sci. Abstracts, 7, 238 (1953)- 86. S. Glasstone, Textbook of Physical Chemistry, D. Van Nostrand Company, Inc., New York (1946). 87* B. J. Mair, A. R. Glasgow, Jr., and F. D. Rossini, J. Research Natl. Bur. Standards, 26, 591 (1941). 88. International Critical Tables, 1, 58, The McGraw-Hill Book Co., New York (1926). 89. W. F. Rooser and A. I, Dahl, J. Research Natl. Bur. Standards, 20, 337 (1938). 90. W. P. White, J. Am. Chem. Soc., 5 6 , 20 (1934). 91. R. L. Burwe11, Jr., A. H. Peterson, and G. B. Rathmann, Rev. Sci. Instr ., 12, 608 (1948). 92. G. Jones and S. M. Christian, J. Am. Chem. Soc., 5 7 , 272 (1935). 93. L. Page and N. I. Adams, Principles of Electricity, D. Van Nostrand Company, Inc., New York (1949)* - 191 - 94. F. R. Harris, Electrical Measurements, John Wiley and Sons, New York (1952). 95- L. Behr and A. J. Williams, Jr., Proc. Inst, of Radio Engineers, 20, 9 6 9 (1932). 96. G-. Jones and B. C. Bradshaw, J. Am. Ghem. Soc., 55? 1780 (1935). 97. L. Onsager, Physik Z., 28, 277 (1927). 98. T. Shedlovsky, J. Am. Ghem. Soc., 99. K. Fredenhagen, Z. anorg. allgem. Ghem., 210. 213 (1933)- 1411 (1932). 1 0 0 . A. G. Worthing and J. G-effner, Treatment of Experimental Data, John Wiley and Sons, New York (1946). 101. C. B. Wooster, J. Am. Ghem. Soc., 60, 1609 (1938). 102 . Opera.ting Instructions for Type 821-A Twin-T Impedance Measuring Circuit, Form 5 6 6 -C, General Radio Company, Cambridge, Mass. 103. Corrections for Residual Impedances in the Twin-T, General Radio Company, Cambridge, Mass. 104. A. A. Maryott and E. R. Smith, Table of Dielectric Constants of Pure Liquids, Circular 514, National Bureau of Standards, U. S. Govt. Printing Office, Washington, D. C. 105. C. F. J. Bottcher, Theory of Electric Polarisation, Elsevier Publishing Co., Amsterdam (1952). 106. F. Eisenlohr, Z. physik Chem., 75? 585 (1910). 107* L. Pauling, Nature of the Chemical Bond, Cornell University Press, Ithaca (1939)- - 192 -